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} -fgnu89-inline @gol
168 -aux-info @var{filename} @gol
169 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
170 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
171 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
172 -fallow-single-precision -fcond-mismatch @gol
173 -fsigned-bitfields -fsigned-char @gol
174 -funsigned-bitfields -funsigned-char}
176 @item C++ Language Options
177 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
178 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
179 -fconserve-space -ffriend-injection @gol
180 -fno-elide-constructors @gol
181 -fno-enforce-eh-specs @gol
182 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
183 -fno-implicit-templates @gol
184 -fno-implicit-inline-templates @gol
185 -fno-implement-inlines -fms-extensions @gol
186 -fno-nonansi-builtins -fno-operator-names @gol
187 -fno-optional-diags -fpermissive @gol
188 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
189 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
190 -fno-default-inline -fvisibility-inlines-hidden @gol
191 -Wabi -Wctor-dtor-privacy @gol
192 -Wnon-virtual-dtor -Wreorder @gol
193 -Weffc++ -Wno-deprecated -Wstrict-null-sentinel @gol
194 -Wno-non-template-friend -Wold-style-cast @gol
195 -Woverloaded-virtual -Wno-pmf-conversions @gol
198 @item Objective-C and Objective-C++ Language Options
199 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
200 Objective-C and Objective-C++ Dialects}.
201 @gccoptlist{-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-option}
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 -Waddress -Waggregate-return -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 -Wstrict-overflow -Wstrict-overflow=@var{n} @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 -fstrict-overflow -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
473 -mkernel -mone-byte-bool}
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{-meb -mel @gol
702 -mscore5 -mscore5u -mscore7 -mscore7d}
705 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
706 -m4-nofpu -m4-single-only -m4-single -m4 @gol
707 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
708 -m5-64media -m5-64media-nofpu @gol
709 -m5-32media -m5-32media-nofpu @gol
710 -m5-compact -m5-compact-nofpu @gol
711 -mb -ml -mdalign -mrelax @gol
712 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
713 -mieee -misize -mpadstruct -mspace @gol
714 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
715 -mdivsi3_libfunc=@var{name} @gol
716 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
720 @gccoptlist{-mcpu=@var{cpu-type} @gol
721 -mtune=@var{cpu-type} @gol
722 -mcmodel=@var{code-model} @gol
723 -m32 -m64 -mapp-regs -mno-app-regs @gol
724 -mfaster-structs -mno-faster-structs @gol
725 -mfpu -mno-fpu -mhard-float -msoft-float @gol
726 -mhard-quad-float -msoft-quad-float @gol
727 -mimpure-text -mno-impure-text -mlittle-endian @gol
728 -mstack-bias -mno-stack-bias @gol
729 -munaligned-doubles -mno-unaligned-doubles @gol
730 -mv8plus -mno-v8plus -mvis -mno-vis
731 -threads -pthreads -pthread}
733 @emph{System V Options}
734 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
736 @emph{TMS320C3x/C4x Options}
737 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
738 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
739 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
740 -mparallel-insns -mparallel-mpy -mpreserve-float}
743 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
744 -mprolog-function -mno-prolog-function -mspace @gol
745 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
746 -mapp-regs -mno-app-regs @gol
747 -mdisable-callt -mno-disable-callt @gol
753 @gccoptlist{-mg -mgnu -munix}
755 @emph{x86-64 Options}
756 See i386 and x86-64 Options.
758 @emph{Xstormy16 Options}
761 @emph{Xtensa Options}
762 @gccoptlist{-mconst16 -mno-const16 @gol
763 -mfused-madd -mno-fused-madd @gol
764 -mtext-section-literals -mno-text-section-literals @gol
765 -mtarget-align -mno-target-align @gol
766 -mlongcalls -mno-longcalls}
768 @emph{zSeries Options}
769 See S/390 and zSeries Options.
771 @item Code Generation Options
772 @xref{Code Gen Options,,Options for Code Generation Conventions}.
773 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
774 -ffixed-@var{reg} -fexceptions @gol
775 -fnon-call-exceptions -funwind-tables @gol
776 -fasynchronous-unwind-tables @gol
777 -finhibit-size-directive -finstrument-functions @gol
778 -fno-common -fno-ident @gol
779 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
780 -fno-jump-tables @gol
781 -freg-struct-return -fshort-enums @gol
782 -fshort-double -fshort-wchar @gol
783 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
784 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
785 -fargument-alias -fargument-noalias @gol
786 -fargument-noalias-global -fargument-noalias-anything
787 -fleading-underscore -ftls-model=@var{model} @gol
788 -ftrapv -fwrapv -fbounds-check @gol
793 * Overall Options:: Controlling the kind of output:
794 an executable, object files, assembler files,
795 or preprocessed source.
796 * C Dialect Options:: Controlling the variant of C language compiled.
797 * C++ Dialect Options:: Variations on C++.
798 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
800 * Language Independent Options:: Controlling how diagnostics should be
802 * Warning Options:: How picky should the compiler be?
803 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
804 * Optimize Options:: How much optimization?
805 * Preprocessor Options:: Controlling header files and macro definitions.
806 Also, getting dependency information for Make.
807 * Assembler Options:: Passing options to the assembler.
808 * Link Options:: Specifying libraries and so on.
809 * Directory Options:: Where to find header files and libraries.
810 Where to find the compiler executable files.
811 * Spec Files:: How to pass switches to sub-processes.
812 * Target Options:: Running a cross-compiler, or an old version of GCC.
815 @node Overall Options
816 @section Options Controlling the Kind of Output
818 Compilation can involve up to four stages: preprocessing, compilation
819 proper, assembly and linking, always in that order. GCC is capable of
820 preprocessing and compiling several files either into several
821 assembler input files, or into one assembler input file; then each
822 assembler input file produces an object file, and linking combines all
823 the object files (those newly compiled, and those specified as input)
824 into an executable file.
826 @cindex file name suffix
827 For any given input file, the file name suffix determines what kind of
832 C source code which must be preprocessed.
835 C source code which should not be preprocessed.
838 C++ source code which should not be preprocessed.
841 Objective-C source code. Note that you must link with the @file{libobjc}
842 library to make an Objective-C program work.
845 Objective-C source code which should not be preprocessed.
849 Objective-C++ source code. Note that you must link with the @file{libobjc}
850 library to make an Objective-C++ program work. Note that @samp{.M} refers
851 to a literal capital M@.
854 Objective-C++ source code which should not be preprocessed.
857 C, C++, Objective-C or Objective-C++ header file to be turned into a
862 @itemx @var{file}.cxx
863 @itemx @var{file}.cpp
864 @itemx @var{file}.CPP
865 @itemx @var{file}.c++
867 C++ source code which must be preprocessed. Note that in @samp{.cxx},
868 the last two letters must both be literally @samp{x}. Likewise,
869 @samp{.C} refers to a literal capital C@.
873 Objective-C++ source code which must be preprocessed.
876 Objective-C++ source code which should not be preprocessed.
880 C++ header file to be turned into a precompiled header.
883 @itemx @var{file}.for
884 @itemx @var{file}.FOR
885 Fixed form Fortran source code which should not be preprocessed.
888 @itemx @var{file}.fpp
889 @itemx @var{file}.FPP
890 Fixed form Fortran source code which must be preprocessed (with the traditional
894 @itemx @var{file}.f95
895 Free form Fortran source code which should not be preprocessed.
898 @itemx @var{file}.F95
899 Free form Fortran source code which must be preprocessed (with the
900 traditional preprocessor).
902 @c FIXME: Descriptions of Java file types.
909 Ada source code file which contains a library unit declaration (a
910 declaration of a package, subprogram, or generic, or a generic
911 instantiation), or a library unit renaming declaration (a package,
912 generic, or subprogram renaming declaration). Such files are also
915 @itemx @var{file}.adb
916 Ada source code file containing a library unit body (a subprogram or
917 package body). Such files are also called @dfn{bodies}.
919 @c GCC also knows about some suffixes for languages not yet included:
930 Assembler code which must be preprocessed.
933 An object file to be fed straight into linking.
934 Any file name with no recognized suffix is treated this way.
938 You can specify the input language explicitly with the @option{-x} option:
941 @item -x @var{language}
942 Specify explicitly the @var{language} for the following input files
943 (rather than letting the compiler choose a default based on the file
944 name suffix). This option applies to all following input files until
945 the next @option{-x} option. Possible values for @var{language} are:
947 c c-header c-cpp-output
948 c++ c++-header c++-cpp-output
949 objective-c objective-c-header objective-c-cpp-output
950 objective-c++ objective-c++-header objective-c++-cpp-output
951 assembler assembler-with-cpp
959 Turn off any specification of a language, so that subsequent files are
960 handled according to their file name suffixes (as they are if @option{-x}
961 has not been used at all).
963 @item -pass-exit-codes
964 @opindex pass-exit-codes
965 Normally the @command{gcc} program will exit with the code of 1 if any
966 phase of the compiler returns a non-success return code. If you specify
967 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
968 numerically highest error produced by any phase that returned an error
969 indication. The C, C++, and Fortran frontends return 4, if an internal
970 compiler error is encountered.
973 If you only want some of the stages of compilation, you can use
974 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
975 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
976 @command{gcc} is to stop. Note that some combinations (for example,
977 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
982 Compile or assemble the source files, but do not link. The linking
983 stage simply is not done. The ultimate output is in the form of an
984 object file for each source file.
986 By default, the object file name for a source file is made by replacing
987 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
989 Unrecognized input files, not requiring compilation or assembly, are
994 Stop after the stage of compilation proper; do not assemble. The output
995 is in the form of an assembler code file for each non-assembler input
998 By default, the assembler file name for a source file is made by
999 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1001 Input files that don't require compilation are ignored.
1005 Stop after the preprocessing stage; do not run the compiler proper. The
1006 output is in the form of preprocessed source code, which is sent to the
1009 Input files which don't require preprocessing are ignored.
1011 @cindex output file option
1014 Place output in file @var{file}. This applies regardless to whatever
1015 sort of output is being produced, whether it be an executable file,
1016 an object file, an assembler file or preprocessed C code.
1018 If @option{-o} is not specified, the default is to put an executable
1019 file in @file{a.out}, the object file for
1020 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1021 assembler file in @file{@var{source}.s}, a precompiled header file in
1022 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1027 Print (on standard error output) the commands executed to run the stages
1028 of compilation. Also print the version number of the compiler driver
1029 program and of the preprocessor and the compiler proper.
1033 Like @option{-v} except the commands are not executed and all command
1034 arguments are quoted. This is useful for shell scripts to capture the
1035 driver-generated command lines.
1039 Use pipes rather than temporary files for communication between the
1040 various stages of compilation. This fails to work on some systems where
1041 the assembler is unable to read from a pipe; but the GNU assembler has
1046 If you are compiling multiple source files, this option tells the driver
1047 to pass all the source files to the compiler at once (for those
1048 languages for which the compiler can handle this). This will allow
1049 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1050 language for which this is supported is C@. If you pass source files for
1051 multiple languages to the driver, using this option, the driver will invoke
1052 the compiler(s) that support IMA once each, passing each compiler all the
1053 source files appropriate for it. For those languages that do not support
1054 IMA this option will be ignored, and the compiler will be invoked once for
1055 each source file in that language. If you use this option in conjunction
1056 with @option{-save-temps}, the compiler will generate multiple
1058 (one for each source file), but only one (combined) @file{.o} or
1063 Print (on the standard output) a description of the command line options
1064 understood by @command{gcc}. If the @option{-v} option is also specified
1065 then @option{--help} will also be passed on to the various processes
1066 invoked by @command{gcc}, so that they can display the command line options
1067 they accept. If the @option{-Wextra} option is also specified then command
1068 line options which have no documentation associated with them will also
1072 @opindex target-help
1073 Print (on the standard output) a description of target specific command
1074 line options for each tool.
1078 Display the version number and copyrights of the invoked GCC@.
1080 @include @value{srcdir}/../libiberty/at-file.texi
1084 @section Compiling C++ Programs
1086 @cindex suffixes for C++ source
1087 @cindex C++ source file suffixes
1088 C++ source files conventionally use one of the suffixes @samp{.C},
1089 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1090 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1091 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1092 files with these names and compiles them as C++ programs even if you
1093 call the compiler the same way as for compiling C programs (usually
1094 with the name @command{gcc}).
1098 However, the use of @command{gcc} does not add the C++ library.
1099 @command{g++} is a program that calls GCC and treats @samp{.c},
1100 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1101 files unless @option{-x} is used, and automatically specifies linking
1102 against the C++ library. This program is also useful when
1103 precompiling a C header file with a @samp{.h} extension for use in C++
1104 compilations. On many systems, @command{g++} is also installed with
1105 the name @command{c++}.
1107 @cindex invoking @command{g++}
1108 When you compile C++ programs, you may specify many of the same
1109 command-line options that you use for compiling programs in any
1110 language; or command-line options meaningful for C and related
1111 languages; or options that are meaningful only for C++ programs.
1112 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1113 explanations of options for languages related to C@.
1114 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1115 explanations of options that are meaningful only for C++ programs.
1117 @node C Dialect Options
1118 @section Options Controlling C Dialect
1119 @cindex dialect options
1120 @cindex language dialect options
1121 @cindex options, dialect
1123 The following options control the dialect of C (or languages derived
1124 from C, such as C++, Objective-C and Objective-C++) that the compiler
1128 @cindex ANSI support
1132 In C mode, support all ISO C90 programs. In C++ mode,
1133 remove GNU extensions that conflict with ISO C++.
1135 This turns off certain features of GCC that are incompatible with ISO
1136 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1137 such as the @code{asm} and @code{typeof} keywords, and
1138 predefined macros such as @code{unix} and @code{vax} that identify the
1139 type of system you are using. It also enables the undesirable and
1140 rarely used ISO trigraph feature. For the C compiler,
1141 it disables recognition of C++ style @samp{//} comments as well as
1142 the @code{inline} keyword.
1144 The alternate keywords @code{__asm__}, @code{__extension__},
1145 @code{__inline__} and @code{__typeof__} continue to work despite
1146 @option{-ansi}. You would not want to use them in an ISO C program, of
1147 course, but it is useful to put them in header files that might be included
1148 in compilations done with @option{-ansi}. Alternate predefined macros
1149 such as @code{__unix__} and @code{__vax__} are also available, with or
1150 without @option{-ansi}.
1152 The @option{-ansi} option does not cause non-ISO programs to be
1153 rejected gratuitously. For that, @option{-pedantic} is required in
1154 addition to @option{-ansi}. @xref{Warning Options}.
1156 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1157 option is used. Some header files may notice this macro and refrain
1158 from declaring certain functions or defining certain macros that the
1159 ISO standard doesn't call for; this is to avoid interfering with any
1160 programs that might use these names for other things.
1162 Functions which would normally be built in but do not have semantics
1163 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1164 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1165 built-in functions provided by GCC}, for details of the functions
1170 Determine the language standard. This option is currently only
1171 supported when compiling C or C++. A value for this option must be
1172 provided; possible values are
1177 ISO C90 (same as @option{-ansi}).
1179 @item iso9899:199409
1180 ISO C90 as modified in amendment 1.
1186 ISO C99. Note that this standard is not yet fully supported; see
1187 @w{@uref{http://gcc.gnu.org/gcc-4.2/c99status.html}} for more information. The
1188 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1191 Default, ISO C90 plus GNU extensions (including some C99 features).
1195 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1196 this will become the default. The name @samp{gnu9x} is deprecated.
1199 The 1998 ISO C++ standard plus amendments.
1202 The same as @option{-std=c++98} plus GNU extensions. This is the
1203 default for C++ code.
1206 Even when this option is not specified, you can still use some of the
1207 features of newer standards in so far as they do not conflict with
1208 previous C standards. For example, you may use @code{__restrict__} even
1209 when @option{-std=c99} is not specified.
1211 The @option{-std} options specifying some version of ISO C have the same
1212 effects as @option{-ansi}, except that features that were not in ISO C90
1213 but are in the specified version (for example, @samp{//} comments and
1214 the @code{inline} keyword in ISO C99) are not disabled.
1216 @xref{Standards,,Language Standards Supported by GCC}, for details of
1217 these standard versions.
1219 @item -fgnu89-inline
1220 @opindex fgnu89-inline
1221 The option @option{-fgnu89-inline} tells GCC to use the traditional
1222 GNU semantics for @code{inline} functions when in C99 mode.
1223 @xref{Inline,,An Inline Function is As Fast As a Macro}. Using this
1224 option is roughly equivalent to adding the @code{gnu_inline} function
1225 attribute to all inline functions (@pxref{Function Attributes}).
1227 This option is accepted by GCC versions 4.1.3 and up. In GCC versions
1228 prior to 4.3, C99 inline semantics are not supported, and thus this
1229 option is effectively assumed to be present regardless of whether or not
1230 it is specified; the only effect of specifying it explicitly is to
1231 disable warnings about using inline functions in C99 mode. Likewise,
1232 the option @option{-fno-gnu89-inline} is not supported in versions of
1233 GCC before 4.3. It will be supported only in C99 or gnu99 mode, not in
1236 The preprocesor macros @code{__GNUC_GNU_INLINE__} and
1237 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1238 in effect for @code{inline} functions. @xref{Common Predefined
1239 Macros,,,cpp,The C Preprocessor}.
1241 @item -aux-info @var{filename}
1243 Output to the given filename prototyped declarations for all functions
1244 declared and/or defined in a translation unit, including those in header
1245 files. This option is silently ignored in any language other than C@.
1247 Besides declarations, the file indicates, in comments, the origin of
1248 each declaration (source file and line), whether the declaration was
1249 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1250 @samp{O} for old, respectively, in the first character after the line
1251 number and the colon), and whether it came from a declaration or a
1252 definition (@samp{C} or @samp{F}, respectively, in the following
1253 character). In the case of function definitions, a K&R-style list of
1254 arguments followed by their declarations is also provided, inside
1255 comments, after the declaration.
1259 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1260 keyword, so that code can use these words as identifiers. You can use
1261 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1262 instead. @option{-ansi} implies @option{-fno-asm}.
1264 In C++, this switch only affects the @code{typeof} keyword, since
1265 @code{asm} and @code{inline} are standard keywords. You may want to
1266 use the @option{-fno-gnu-keywords} flag instead, which has the same
1267 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1268 switch only affects the @code{asm} and @code{typeof} keywords, since
1269 @code{inline} is a standard keyword in ISO C99.
1272 @itemx -fno-builtin-@var{function}
1273 @opindex fno-builtin
1274 @cindex built-in functions
1275 Don't recognize built-in functions that do not begin with
1276 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1277 functions provided by GCC}, for details of the functions affected,
1278 including those which are not built-in functions when @option{-ansi} or
1279 @option{-std} options for strict ISO C conformance are used because they
1280 do not have an ISO standard meaning.
1282 GCC normally generates special code to handle certain built-in functions
1283 more efficiently; for instance, calls to @code{alloca} may become single
1284 instructions that adjust the stack directly, and calls to @code{memcpy}
1285 may become inline copy loops. The resulting code is often both smaller
1286 and faster, but since the function calls no longer appear as such, you
1287 cannot set a breakpoint on those calls, nor can you change the behavior
1288 of the functions by linking with a different library. In addition,
1289 when a function is recognized as a built-in function, GCC may use
1290 information about that function to warn about problems with calls to
1291 that function, or to generate more efficient code, even if the
1292 resulting code still contains calls to that function. For example,
1293 warnings are given with @option{-Wformat} for bad calls to
1294 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1295 known not to modify global memory.
1297 With the @option{-fno-builtin-@var{function}} option
1298 only the built-in function @var{function} is
1299 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1300 function is named this is not built-in in this version of GCC, this
1301 option is ignored. There is no corresponding
1302 @option{-fbuiltin-@var{function}} option; if you wish to enable
1303 built-in functions selectively when using @option{-fno-builtin} or
1304 @option{-ffreestanding}, you may define macros such as:
1307 #define abs(n) __builtin_abs ((n))
1308 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1313 @cindex hosted environment
1315 Assert that compilation takes place in a hosted environment. This implies
1316 @option{-fbuiltin}. A hosted environment is one in which the
1317 entire standard library is available, and in which @code{main} has a return
1318 type of @code{int}. Examples are nearly everything except a kernel.
1319 This is equivalent to @option{-fno-freestanding}.
1321 @item -ffreestanding
1322 @opindex ffreestanding
1323 @cindex hosted environment
1325 Assert that compilation takes place in a freestanding environment. This
1326 implies @option{-fno-builtin}. A freestanding environment
1327 is one in which the standard library may not exist, and program startup may
1328 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1329 This is equivalent to @option{-fno-hosted}.
1331 @xref{Standards,,Language Standards Supported by GCC}, for details of
1332 freestanding and hosted environments.
1336 @cindex openmp parallel
1337 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1338 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1339 compiler generates parallel code according to the OpenMP Application
1340 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.
1342 @item -fms-extensions
1343 @opindex fms-extensions
1344 Accept some non-standard constructs used in Microsoft header files.
1346 Some cases of unnamed fields in structures and unions are only
1347 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1348 fields within structs/unions}, for details.
1352 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1353 options for strict ISO C conformance) implies @option{-trigraphs}.
1355 @item -no-integrated-cpp
1356 @opindex no-integrated-cpp
1357 Performs a compilation in two passes: preprocessing and compiling. This
1358 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1359 @option{-B} option. The user supplied compilation step can then add in
1360 an additional preprocessing step after normal preprocessing but before
1361 compiling. The default is to use the integrated cpp (internal cpp)
1363 The semantics of this option will change if "cc1", "cc1plus", and
1364 "cc1obj" are merged.
1366 @cindex traditional C language
1367 @cindex C language, traditional
1369 @itemx -traditional-cpp
1370 @opindex traditional-cpp
1371 @opindex traditional
1372 Formerly, these options caused GCC to attempt to emulate a pre-standard
1373 C compiler. They are now only supported with the @option{-E} switch.
1374 The preprocessor continues to support a pre-standard mode. See the GNU
1375 CPP manual for details.
1377 @item -fcond-mismatch
1378 @opindex fcond-mismatch
1379 Allow conditional expressions with mismatched types in the second and
1380 third arguments. The value of such an expression is void. This option
1381 is not supported for C++.
1383 @item -funsigned-char
1384 @opindex funsigned-char
1385 Let the type @code{char} be unsigned, like @code{unsigned char}.
1387 Each kind of machine has a default for what @code{char} should
1388 be. It is either like @code{unsigned char} by default or like
1389 @code{signed char} by default.
1391 Ideally, a portable program should always use @code{signed char} or
1392 @code{unsigned char} when it depends on the signedness of an object.
1393 But many programs have been written to use plain @code{char} and
1394 expect it to be signed, or expect it to be unsigned, depending on the
1395 machines they were written for. This option, and its inverse, let you
1396 make such a program work with the opposite default.
1398 The type @code{char} is always a distinct type from each of
1399 @code{signed char} or @code{unsigned char}, even though its behavior
1400 is always just like one of those two.
1403 @opindex fsigned-char
1404 Let the type @code{char} be signed, like @code{signed char}.
1406 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1407 the negative form of @option{-funsigned-char}. Likewise, the option
1408 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1410 @item -fsigned-bitfields
1411 @itemx -funsigned-bitfields
1412 @itemx -fno-signed-bitfields
1413 @itemx -fno-unsigned-bitfields
1414 @opindex fsigned-bitfields
1415 @opindex funsigned-bitfields
1416 @opindex fno-signed-bitfields
1417 @opindex fno-unsigned-bitfields
1418 These options control whether a bit-field is signed or unsigned, when the
1419 declaration does not use either @code{signed} or @code{unsigned}. By
1420 default, such a bit-field is signed, because this is consistent: the
1421 basic integer types such as @code{int} are signed types.
1424 @node C++ Dialect Options
1425 @section Options Controlling C++ Dialect
1427 @cindex compiler options, C++
1428 @cindex C++ options, command line
1429 @cindex options, C++
1430 This section describes the command-line options that are only meaningful
1431 for C++ programs; but you can also use most of the GNU compiler options
1432 regardless of what language your program is in. For example, you
1433 might compile a file @code{firstClass.C} like this:
1436 g++ -g -frepo -O -c firstClass.C
1440 In this example, only @option{-frepo} is an option meant
1441 only for C++ programs; you can use the other options with any
1442 language supported by GCC@.
1444 Here is a list of options that are @emph{only} for compiling C++ programs:
1448 @item -fabi-version=@var{n}
1449 @opindex fabi-version
1450 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1451 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1452 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1453 the version that conforms most closely to the C++ ABI specification.
1454 Therefore, the ABI obtained using version 0 will change as ABI bugs
1457 The default is version 2.
1459 @item -fno-access-control
1460 @opindex fno-access-control
1461 Turn off all access checking. This switch is mainly useful for working
1462 around bugs in the access control code.
1466 Check that the pointer returned by @code{operator new} is non-null
1467 before attempting to modify the storage allocated. This check is
1468 normally unnecessary because the C++ standard specifies that
1469 @code{operator new} will only return @code{0} if it is declared
1470 @samp{throw()}, in which case the compiler will always check the
1471 return value even without this option. In all other cases, when
1472 @code{operator new} has a non-empty exception specification, memory
1473 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1474 @samp{new (nothrow)}.
1476 @item -fconserve-space
1477 @opindex fconserve-space
1478 Put uninitialized or runtime-initialized global variables into the
1479 common segment, as C does. This saves space in the executable at the
1480 cost of not diagnosing duplicate definitions. If you compile with this
1481 flag and your program mysteriously crashes after @code{main()} has
1482 completed, you may have an object that is being destroyed twice because
1483 two definitions were merged.
1485 This option is no longer useful on most targets, now that support has
1486 been added for putting variables into BSS without making them common.
1488 @item -ffriend-injection
1489 @opindex ffriend-injection
1490 Inject friend functions into the enclosing namespace, so that they are
1491 visible outside the scope of the class in which they are declared.
1492 Friend functions were documented to work this way in the old Annotated
1493 C++ Reference Manual, and versions of G++ before 4.1 always worked
1494 that way. However, in ISO C++ a friend function which is not declared
1495 in an enclosing scope can only be found using argument dependent
1496 lookup. This option causes friends to be injected as they were in
1499 This option is for compatibility, and may be removed in a future
1502 @item -fno-elide-constructors
1503 @opindex fno-elide-constructors
1504 The C++ standard allows an implementation to omit creating a temporary
1505 which is only used to initialize another object of the same type.
1506 Specifying this option disables that optimization, and forces G++ to
1507 call the copy constructor in all cases.
1509 @item -fno-enforce-eh-specs
1510 @opindex fno-enforce-eh-specs
1511 Don't generate code to check for violation of exception specifications
1512 at runtime. This option violates the C++ standard, but may be useful
1513 for reducing code size in production builds, much like defining
1514 @samp{NDEBUG}. This does not give user code permission to throw
1515 exceptions in violation of the exception specifications; the compiler
1516 will still optimize based on the specifications, so throwing an
1517 unexpected exception will result in undefined behavior.
1520 @itemx -fno-for-scope
1522 @opindex fno-for-scope
1523 If @option{-ffor-scope} is specified, the scope of variables declared in
1524 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1525 as specified by the C++ standard.
1526 If @option{-fno-for-scope} is specified, the scope of variables declared in
1527 a @i{for-init-statement} extends to the end of the enclosing scope,
1528 as was the case in old versions of G++, and other (traditional)
1529 implementations of C++.
1531 The default if neither flag is given to follow the standard,
1532 but to allow and give a warning for old-style code that would
1533 otherwise be invalid, or have different behavior.
1535 @item -fno-gnu-keywords
1536 @opindex fno-gnu-keywords
1537 Do not recognize @code{typeof} as a keyword, so that code can use this
1538 word as an identifier. You can use the keyword @code{__typeof__} instead.
1539 @option{-ansi} implies @option{-fno-gnu-keywords}.
1541 @item -fno-implicit-templates
1542 @opindex fno-implicit-templates
1543 Never emit code for non-inline templates which are instantiated
1544 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1545 @xref{Template Instantiation}, for more information.
1547 @item -fno-implicit-inline-templates
1548 @opindex fno-implicit-inline-templates
1549 Don't emit code for implicit instantiations of inline templates, either.
1550 The default is to handle inlines differently so that compiles with and
1551 without optimization will need the same set of explicit instantiations.
1553 @item -fno-implement-inlines
1554 @opindex fno-implement-inlines
1555 To save space, do not emit out-of-line copies of inline functions
1556 controlled by @samp{#pragma implementation}. This will cause linker
1557 errors if these functions are not inlined everywhere they are called.
1559 @item -fms-extensions
1560 @opindex fms-extensions
1561 Disable pedantic warnings about constructs used in MFC, such as implicit
1562 int and getting a pointer to member function via non-standard syntax.
1564 @item -fno-nonansi-builtins
1565 @opindex fno-nonansi-builtins
1566 Disable built-in declarations of functions that are not mandated by
1567 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1568 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1570 @item -fno-operator-names
1571 @opindex fno-operator-names
1572 Do not treat the operator name keywords @code{and}, @code{bitand},
1573 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1574 synonyms as keywords.
1576 @item -fno-optional-diags
1577 @opindex fno-optional-diags
1578 Disable diagnostics that the standard says a compiler does not need to
1579 issue. Currently, the only such diagnostic issued by G++ is the one for
1580 a name having multiple meanings within a class.
1583 @opindex fpermissive
1584 Downgrade some diagnostics about nonconformant code from errors to
1585 warnings. Thus, using @option{-fpermissive} will allow some
1586 nonconforming code to compile.
1590 Enable automatic template instantiation at link time. This option also
1591 implies @option{-fno-implicit-templates}. @xref{Template
1592 Instantiation}, for more information.
1596 Disable generation of information about every class with virtual
1597 functions for use by the C++ runtime type identification features
1598 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1599 of the language, you can save some space by using this flag. Note that
1600 exception handling uses the same information, but it will generate it as
1601 needed. The @samp{dynamic_cast} operator can still be used for casts that
1602 do not require runtime type information, i.e. casts to @code{void *} or to
1603 unambiguous base classes.
1607 Emit statistics about front-end processing at the end of the compilation.
1608 This information is generally only useful to the G++ development team.
1610 @item -ftemplate-depth-@var{n}
1611 @opindex ftemplate-depth
1612 Set the maximum instantiation depth for template classes to @var{n}.
1613 A limit on the template instantiation depth is needed to detect
1614 endless recursions during template class instantiation. ANSI/ISO C++
1615 conforming programs must not rely on a maximum depth greater than 17.
1617 @item -fno-threadsafe-statics
1618 @opindex fno-threadsafe-statics
1619 Do not emit the extra code to use the routines specified in the C++
1620 ABI for thread-safe initialization of local statics. You can use this
1621 option to reduce code size slightly in code that doesn't need to be
1624 @item -fuse-cxa-atexit
1625 @opindex fuse-cxa-atexit
1626 Register destructors for objects with static storage duration with the
1627 @code{__cxa_atexit} function rather than the @code{atexit} function.
1628 This option is required for fully standards-compliant handling of static
1629 destructors, but will only work if your C library supports
1630 @code{__cxa_atexit}.
1632 @item -fno-use-cxa-get-exception-ptr
1633 @opindex fno-use-cxa-get-exception-ptr
1634 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1635 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1636 if the runtime routine is not available.
1638 @item -fvisibility-inlines-hidden
1639 @opindex fvisibility-inlines-hidden
1640 This switch declares that the user does not attempt to compare
1641 pointers to inline methods where the addresses of the two functions
1642 were taken in different shared objects.
1644 The effect of this is that GCC may, effectively, mark inline methods with
1645 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1646 appear in the export table of a DSO and do not require a PLT indirection
1647 when used within the DSO@. Enabling this option can have a dramatic effect
1648 on load and link times of a DSO as it massively reduces the size of the
1649 dynamic export table when the library makes heavy use of templates.
1651 The behaviour of this switch is not quite the same as marking the
1652 methods as hidden directly, because it does not affect static variables
1653 local to the function or cause the compiler to deduce that
1654 the function is defined in only one shared object.
1656 You may mark a method as having a visibility explicitly to negate the
1657 effect of the switch for that method. For example, if you do want to
1658 compare pointers to a particular inline method, you might mark it as
1659 having default visibility. Marking the enclosing class with explicit
1660 visibility will have no effect.
1662 Explicitly instantiated inline methods are unaffected by this option
1663 as their linkage might otherwise cross a shared library boundary.
1664 @xref{Template Instantiation}.
1668 Do not use weak symbol support, even if it is provided by the linker.
1669 By default, G++ will use weak symbols if they are available. This
1670 option exists only for testing, and should not be used by end-users;
1671 it will result in inferior code and has no benefits. This option may
1672 be removed in a future release of G++.
1676 Do not search for header files in the standard directories specific to
1677 C++, but do still search the other standard directories. (This option
1678 is used when building the C++ library.)
1681 In addition, these optimization, warning, and code generation options
1682 have meanings only for C++ programs:
1685 @item -fno-default-inline
1686 @opindex fno-default-inline
1687 Do not assume @samp{inline} for functions defined inside a class scope.
1688 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1689 functions will have linkage like inline functions; they just won't be
1692 @item -Wabi @r{(C++ only)}
1694 Warn when G++ generates code that is probably not compatible with the
1695 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1696 all such cases, there are probably some cases that are not warned about,
1697 even though G++ is generating incompatible code. There may also be
1698 cases where warnings are emitted even though the code that is generated
1701 You should rewrite your code to avoid these warnings if you are
1702 concerned about the fact that code generated by G++ may not be binary
1703 compatible with code generated by other compilers.
1705 The known incompatibilities at this point include:
1710 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1711 pack data into the same byte as a base class. For example:
1714 struct A @{ virtual void f(); int f1 : 1; @};
1715 struct B : public A @{ int f2 : 1; @};
1719 In this case, G++ will place @code{B::f2} into the same byte
1720 as@code{A::f1}; other compilers will not. You can avoid this problem
1721 by explicitly padding @code{A} so that its size is a multiple of the
1722 byte size on your platform; that will cause G++ and other compilers to
1723 layout @code{B} identically.
1726 Incorrect handling of tail-padding for virtual bases. G++ does not use
1727 tail padding when laying out virtual bases. For example:
1730 struct A @{ virtual void f(); char c1; @};
1731 struct B @{ B(); char c2; @};
1732 struct C : public A, public virtual B @{@};
1736 In this case, G++ will not place @code{B} into the tail-padding for
1737 @code{A}; other compilers will. You can avoid this problem by
1738 explicitly padding @code{A} so that its size is a multiple of its
1739 alignment (ignoring virtual base classes); that will cause G++ and other
1740 compilers to layout @code{C} identically.
1743 Incorrect handling of bit-fields with declared widths greater than that
1744 of their underlying types, when the bit-fields appear in a union. For
1748 union U @{ int i : 4096; @};
1752 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1753 union too small by the number of bits in an @code{int}.
1756 Empty classes can be placed at incorrect offsets. For example:
1766 struct C : public B, public A @{@};
1770 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1771 it should be placed at offset zero. G++ mistakenly believes that the
1772 @code{A} data member of @code{B} is already at offset zero.
1775 Names of template functions whose types involve @code{typename} or
1776 template template parameters can be mangled incorrectly.
1779 template <typename Q>
1780 void f(typename Q::X) @{@}
1782 template <template <typename> class Q>
1783 void f(typename Q<int>::X) @{@}
1787 Instantiations of these templates may be mangled incorrectly.
1791 @item -Wctor-dtor-privacy @r{(C++ only)}
1792 @opindex Wctor-dtor-privacy
1793 Warn when a class seems unusable because all the constructors or
1794 destructors in that class are private, and it has neither friends nor
1795 public static member functions.
1797 @item -Wnon-virtual-dtor @r{(C++ only)}
1798 @opindex Wnon-virtual-dtor
1799 Warn when a class appears to be polymorphic, thereby requiring a virtual
1800 destructor, yet it declares a non-virtual one. This warning is also
1801 enabled if -Weffc++ is specified.
1803 @item -Wreorder @r{(C++ only)}
1805 @cindex reordering, warning
1806 @cindex warning for reordering of member initializers
1807 Warn when the order of member initializers given in the code does not
1808 match the order in which they must be executed. For instance:
1814 A(): j (0), i (1) @{ @}
1818 The compiler will rearrange the member initializers for @samp{i}
1819 and @samp{j} to match the declaration order of the members, emitting
1820 a warning to that effect. This warning is enabled by @option{-Wall}.
1823 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1826 @item -Weffc++ @r{(C++ only)}
1828 Warn about violations of the following style guidelines from Scott Meyers'
1829 @cite{Effective C++} book:
1833 Item 11: Define a copy constructor and an assignment operator for classes
1834 with dynamically allocated memory.
1837 Item 12: Prefer initialization to assignment in constructors.
1840 Item 14: Make destructors virtual in base classes.
1843 Item 15: Have @code{operator=} return a reference to @code{*this}.
1846 Item 23: Don't try to return a reference when you must return an object.
1850 Also warn about violations of the following style guidelines from
1851 Scott Meyers' @cite{More Effective C++} book:
1855 Item 6: Distinguish between prefix and postfix forms of increment and
1856 decrement operators.
1859 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1863 When selecting this option, be aware that the standard library
1864 headers do not obey all of these guidelines; use @samp{grep -v}
1865 to filter out those warnings.
1867 @item -Wno-deprecated @r{(C++ only)}
1868 @opindex Wno-deprecated
1869 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1871 @item -Wstrict-null-sentinel @r{(C++ only)}
1872 @opindex Wstrict-null-sentinel
1873 Warn also about the use of an uncasted @code{NULL} as sentinel. When
1874 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
1875 to @code{__null}. Although it is a null pointer constant not a null pointer,
1876 it is guaranteed to of the same size as a pointer. But this use is
1877 not portable across different compilers.
1879 @item -Wno-non-template-friend @r{(C++ only)}
1880 @opindex Wno-non-template-friend
1881 Disable warnings when non-templatized friend functions are declared
1882 within a template. Since the advent of explicit template specification
1883 support in G++, if the name of the friend is an unqualified-id (i.e.,
1884 @samp{friend foo(int)}), the C++ language specification demands that the
1885 friend declare or define an ordinary, nontemplate function. (Section
1886 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1887 could be interpreted as a particular specialization of a templatized
1888 function. Because this non-conforming behavior is no longer the default
1889 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1890 check existing code for potential trouble spots and is on by default.
1891 This new compiler behavior can be turned off with
1892 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1893 but disables the helpful warning.
1895 @item -Wold-style-cast @r{(C++ only)}
1896 @opindex Wold-style-cast
1897 Warn if an old-style (C-style) cast to a non-void type is used within
1898 a C++ program. The new-style casts (@samp{dynamic_cast},
1899 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
1900 less vulnerable to unintended effects and much easier to search for.
1902 @item -Woverloaded-virtual @r{(C++ only)}
1903 @opindex Woverloaded-virtual
1904 @cindex overloaded virtual fn, warning
1905 @cindex warning for overloaded virtual fn
1906 Warn when a function declaration hides virtual functions from a
1907 base class. For example, in:
1914 struct B: public A @{
1919 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1927 will fail to compile.
1929 @item -Wno-pmf-conversions @r{(C++ only)}
1930 @opindex Wno-pmf-conversions
1931 Disable the diagnostic for converting a bound pointer to member function
1934 @item -Wsign-promo @r{(C++ only)}
1935 @opindex Wsign-promo
1936 Warn when overload resolution chooses a promotion from unsigned or
1937 enumerated type to a signed type, over a conversion to an unsigned type of
1938 the same size. Previous versions of G++ would try to preserve
1939 unsignedness, but the standard mandates the current behavior.
1944 A& operator = (int);
1954 In this example, G++ will synthesize a default @samp{A& operator =
1955 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1958 @node Objective-C and Objective-C++ Dialect Options
1959 @section Options Controlling Objective-C and Objective-C++ Dialects
1961 @cindex compiler options, Objective-C and Objective-C++
1962 @cindex Objective-C and Objective-C++ options, command line
1963 @cindex options, Objective-C and Objective-C++
1964 (NOTE: This manual does not describe the Objective-C and Objective-C++
1965 languages themselves. See @xref{Standards,,Language Standards
1966 Supported by GCC}, for references.)
1968 This section describes the command-line options that are only meaningful
1969 for Objective-C and Objective-C++ programs, but you can also use most of
1970 the language-independent GNU compiler options.
1971 For example, you might compile a file @code{some_class.m} like this:
1974 gcc -g -fgnu-runtime -O -c some_class.m
1978 In this example, @option{-fgnu-runtime} is an option meant only for
1979 Objective-C and Objective-C++ programs; you can use the other options with
1980 any language supported by GCC@.
1982 Note that since Objective-C is an extension of the C language, Objective-C
1983 compilations may also use options specific to the C front-end (e.g.,
1984 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1985 C++-specific options (e.g., @option{-Wabi}).
1987 Here is a list of options that are @emph{only} for compiling Objective-C
1988 and Objective-C++ programs:
1991 @item -fconstant-string-class=@var{class-name}
1992 @opindex fconstant-string-class
1993 Use @var{class-name} as the name of the class to instantiate for each
1994 literal string specified with the syntax @code{@@"@dots{}"}. The default
1995 class name is @code{NXConstantString} if the GNU runtime is being used, and
1996 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1997 @option{-fconstant-cfstrings} option, if also present, will override the
1998 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1999 to be laid out as constant CoreFoundation strings.
2002 @opindex fgnu-runtime
2003 Generate object code compatible with the standard GNU Objective-C
2004 runtime. This is the default for most types of systems.
2006 @item -fnext-runtime
2007 @opindex fnext-runtime
2008 Generate output compatible with the NeXT runtime. This is the default
2009 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2010 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2013 @item -fno-nil-receivers
2014 @opindex fno-nil-receivers
2015 Assume that all Objective-C message dispatches (e.g.,
2016 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2017 is not @code{nil}. This allows for more efficient entry points in the runtime
2018 to be used. Currently, this option is only available in conjunction with
2019 the NeXT runtime on Mac OS X 10.3 and later.
2021 @item -fobjc-call-cxx-cdtors
2022 @opindex fobjc-call-cxx-cdtors
2023 For each Objective-C class, check if any of its instance variables is a
2024 C++ object with a non-trivial default constructor. If so, synthesize a
2025 special @code{- (id) .cxx_construct} instance method that will run
2026 non-trivial default constructors on any such instance variables, in order,
2027 and then return @code{self}. Similarly, check if any instance variable
2028 is a C++ object with a non-trivial destructor, and if so, synthesize a
2029 special @code{- (void) .cxx_destruct} method that will run
2030 all such default destructors, in reverse order.
2032 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2033 thusly generated will only operate on instance variables declared in the
2034 current Objective-C class, and not those inherited from superclasses. It
2035 is the responsibility of the Objective-C runtime to invoke all such methods
2036 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2037 will be invoked by the runtime immediately after a new object
2038 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2039 be invoked immediately before the runtime deallocates an object instance.
2041 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2042 support for invoking the @code{- (id) .cxx_construct} and
2043 @code{- (void) .cxx_destruct} methods.
2045 @item -fobjc-direct-dispatch
2046 @opindex fobjc-direct-dispatch
2047 Allow fast jumps to the message dispatcher. On Darwin this is
2048 accomplished via the comm page.
2050 @item -fobjc-exceptions
2051 @opindex fobjc-exceptions
2052 Enable syntactic support for structured exception handling in Objective-C,
2053 similar to what is offered by C++ and Java. This option is
2054 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2063 @@catch (AnObjCClass *exc) @{
2070 @@catch (AnotherClass *exc) @{
2073 @@catch (id allOthers) @{
2083 The @code{@@throw} statement may appear anywhere in an Objective-C or
2084 Objective-C++ program; when used inside of a @code{@@catch} block, the
2085 @code{@@throw} may appear without an argument (as shown above), in which case
2086 the object caught by the @code{@@catch} will be rethrown.
2088 Note that only (pointers to) Objective-C objects may be thrown and
2089 caught using this scheme. When an object is thrown, it will be caught
2090 by the nearest @code{@@catch} clause capable of handling objects of that type,
2091 analogously to how @code{catch} blocks work in C++ and Java. A
2092 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2093 any and all Objective-C exceptions not caught by previous @code{@@catch}
2096 The @code{@@finally} clause, if present, will be executed upon exit from the
2097 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2098 regardless of whether any exceptions are thrown, caught or rethrown
2099 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2100 of the @code{finally} clause in Java.
2102 There are several caveats to using the new exception mechanism:
2106 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2107 idioms provided by the @code{NSException} class, the new
2108 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2109 systems, due to additional functionality needed in the (NeXT) Objective-C
2113 As mentioned above, the new exceptions do not support handling
2114 types other than Objective-C objects. Furthermore, when used from
2115 Objective-C++, the Objective-C exception model does not interoperate with C++
2116 exceptions at this time. This means you cannot @code{@@throw} an exception
2117 from Objective-C and @code{catch} it in C++, or vice versa
2118 (i.e., @code{throw @dots{} @@catch}).
2121 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2122 blocks for thread-safe execution:
2125 @@synchronized (ObjCClass *guard) @{
2130 Upon entering the @code{@@synchronized} block, a thread of execution shall
2131 first check whether a lock has been placed on the corresponding @code{guard}
2132 object by another thread. If it has, the current thread shall wait until
2133 the other thread relinquishes its lock. Once @code{guard} becomes available,
2134 the current thread will place its own lock on it, execute the code contained in
2135 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2136 making @code{guard} available to other threads).
2138 Unlike Java, Objective-C does not allow for entire methods to be marked
2139 @code{@@synchronized}. Note that throwing exceptions out of
2140 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2141 to be unlocked properly.
2145 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2147 @item -freplace-objc-classes
2148 @opindex freplace-objc-classes
2149 Emit a special marker instructing @command{ld(1)} not to statically link in
2150 the resulting object file, and allow @command{dyld(1)} to load it in at
2151 run time instead. This is used in conjunction with the Fix-and-Continue
2152 debugging mode, where the object file in question may be recompiled and
2153 dynamically reloaded in the course of program execution, without the need
2154 to restart the program itself. Currently, Fix-and-Continue functionality
2155 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2160 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2161 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2162 compile time) with static class references that get initialized at load time,
2163 which improves run-time performance. Specifying the @option{-fzero-link} flag
2164 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2165 to be retained. This is useful in Zero-Link debugging mode, since it allows
2166 for individual class implementations to be modified during program execution.
2170 Dump interface declarations for all classes seen in the source file to a
2171 file named @file{@var{sourcename}.decl}.
2173 @item -Wassign-intercept
2174 @opindex Wassign-intercept
2175 Warn whenever an Objective-C assignment is being intercepted by the
2179 @opindex Wno-protocol
2180 If a class is declared to implement a protocol, a warning is issued for
2181 every method in the protocol that is not implemented by the class. The
2182 default behavior is to issue a warning for every method not explicitly
2183 implemented in the class, even if a method implementation is inherited
2184 from the superclass. If you use the @option{-Wno-protocol} option, then
2185 methods inherited from the superclass are considered to be implemented,
2186 and no warning is issued for them.
2190 Warn if multiple methods of different types for the same selector are
2191 found during compilation. The check is performed on the list of methods
2192 in the final stage of compilation. Additionally, a check is performed
2193 for each selector appearing in a @code{@@selector(@dots{})}
2194 expression, and a corresponding method for that selector has been found
2195 during compilation. Because these checks scan the method table only at
2196 the end of compilation, these warnings are not produced if the final
2197 stage of compilation is not reached, for example because an error is
2198 found during compilation, or because the @option{-fsyntax-only} option is
2201 @item -Wstrict-selector-match
2202 @opindex Wstrict-selector-match
2203 Warn if multiple methods with differing argument and/or return types are
2204 found for a given selector when attempting to send a message using this
2205 selector to a receiver of type @code{id} or @code{Class}. When this flag
2206 is off (which is the default behavior), the compiler will omit such warnings
2207 if any differences found are confined to types which share the same size
2210 @item -Wundeclared-selector
2211 @opindex Wundeclared-selector
2212 Warn if a @code{@@selector(@dots{})} expression referring to an
2213 undeclared selector is found. A selector is considered undeclared if no
2214 method with that name has been declared before the
2215 @code{@@selector(@dots{})} expression, either explicitly in an
2216 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2217 an @code{@@implementation} section. This option always performs its
2218 checks as soon as a @code{@@selector(@dots{})} expression is found,
2219 while @option{-Wselector} only performs its checks in the final stage of
2220 compilation. This also enforces the coding style convention
2221 that methods and selectors must be declared before being used.
2223 @item -print-objc-runtime-info
2224 @opindex print-objc-runtime-info
2225 Generate C header describing the largest structure that is passed by
2230 @node Language Independent Options
2231 @section Options to Control Diagnostic Messages Formatting
2232 @cindex options to control diagnostics formatting
2233 @cindex diagnostic messages
2234 @cindex message formatting
2236 Traditionally, diagnostic messages have been formatted irrespective of
2237 the output device's aspect (e.g.@: its width, @dots{}). The options described
2238 below can be used to control the diagnostic messages formatting
2239 algorithm, e.g.@: how many characters per line, how often source location
2240 information should be reported. Right now, only the C++ front end can
2241 honor these options. However it is expected, in the near future, that
2242 the remaining front ends would be able to digest them correctly.
2245 @item -fmessage-length=@var{n}
2246 @opindex fmessage-length
2247 Try to format error messages so that they fit on lines of about @var{n}
2248 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2249 the front ends supported by GCC@. If @var{n} is zero, then no
2250 line-wrapping will be done; each error message will appear on a single
2253 @opindex fdiagnostics-show-location
2254 @item -fdiagnostics-show-location=once
2255 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2256 reporter to emit @emph{once} source location information; that is, in
2257 case the message is too long to fit on a single physical line and has to
2258 be wrapped, the source location won't be emitted (as prefix) again,
2259 over and over, in subsequent continuation lines. This is the default
2262 @item -fdiagnostics-show-location=every-line
2263 Only meaningful in line-wrapping mode. Instructs the diagnostic
2264 messages reporter to emit the same source location information (as
2265 prefix) for physical lines that result from the process of breaking
2266 a message which is too long to fit on a single line.
2268 @item -fdiagnostics-show-option
2269 @opindex fdiagnostics-show-option
2270 This option instructs the diagnostic machinery to add text to each
2271 diagnostic emitted, which indicates which command line option directly
2272 controls that diagnostic, when such an option is known to the
2273 diagnostic machinery.
2277 @node Warning Options
2278 @section Options to Request or Suppress Warnings
2279 @cindex options to control warnings
2280 @cindex warning messages
2281 @cindex messages, warning
2282 @cindex suppressing warnings
2284 Warnings are diagnostic messages that report constructions which
2285 are not inherently erroneous but which are risky or suggest there
2286 may have been an error.
2288 You can request many specific warnings with options beginning @samp{-W},
2289 for example @option{-Wimplicit} to request warnings on implicit
2290 declarations. Each of these specific warning options also has a
2291 negative form beginning @samp{-Wno-} to turn off warnings;
2292 for example, @option{-Wno-implicit}. This manual lists only one of the
2293 two forms, whichever is not the default.
2295 The following options control the amount and kinds of warnings produced
2296 by GCC; for further, language-specific options also refer to
2297 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2301 @cindex syntax checking
2303 @opindex fsyntax-only
2304 Check the code for syntax errors, but don't do anything beyond that.
2308 Issue all the warnings demanded by strict ISO C and ISO C++;
2309 reject all programs that use forbidden extensions, and some other
2310 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2311 version of the ISO C standard specified by any @option{-std} option used.
2313 Valid ISO C and ISO C++ programs should compile properly with or without
2314 this option (though a rare few will require @option{-ansi} or a
2315 @option{-std} option specifying the required version of ISO C)@. However,
2316 without this option, certain GNU extensions and traditional C and C++
2317 features are supported as well. With this option, they are rejected.
2319 @option{-pedantic} does not cause warning messages for use of the
2320 alternate keywords whose names begin and end with @samp{__}. Pedantic
2321 warnings are also disabled in the expression that follows
2322 @code{__extension__}. However, only system header files should use
2323 these escape routes; application programs should avoid them.
2324 @xref{Alternate Keywords}.
2326 Some users try to use @option{-pedantic} to check programs for strict ISO
2327 C conformance. They soon find that it does not do quite what they want:
2328 it finds some non-ISO practices, but not all---only those for which
2329 ISO C @emph{requires} a diagnostic, and some others for which
2330 diagnostics have been added.
2332 A feature to report any failure to conform to ISO C might be useful in
2333 some instances, but would require considerable additional work and would
2334 be quite different from @option{-pedantic}. We don't have plans to
2335 support such a feature in the near future.
2337 Where the standard specified with @option{-std} represents a GNU
2338 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2339 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2340 extended dialect is based. Warnings from @option{-pedantic} are given
2341 where they are required by the base standard. (It would not make sense
2342 for such warnings to be given only for features not in the specified GNU
2343 C dialect, since by definition the GNU dialects of C include all
2344 features the compiler supports with the given option, and there would be
2345 nothing to warn about.)
2347 @item -pedantic-errors
2348 @opindex pedantic-errors
2349 Like @option{-pedantic}, except that errors are produced rather than
2354 Inhibit all warning messages.
2358 Inhibit warning messages about the use of @samp{#import}.
2360 @item -Wchar-subscripts
2361 @opindex Wchar-subscripts
2362 Warn if an array subscript has type @code{char}. This is a common cause
2363 of error, as programmers often forget that this type is signed on some
2365 This warning is enabled by @option{-Wall}.
2369 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2370 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2371 This warning is enabled by @option{-Wall}.
2373 @item -Wfatal-errors
2374 @opindex Wfatal-errors
2375 This option causes the compiler to abort compilation on the first error
2376 occurred rather than trying to keep going and printing further error
2381 @opindex ffreestanding
2382 @opindex fno-builtin
2383 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2384 the arguments supplied have types appropriate to the format string
2385 specified, and that the conversions specified in the format string make
2386 sense. This includes standard functions, and others specified by format
2387 attributes (@pxref{Function Attributes}), in the @code{printf},
2388 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2389 not in the C standard) families (or other target-specific families).
2390 Which functions are checked without format attributes having been
2391 specified depends on the standard version selected, and such checks of
2392 functions without the attribute specified are disabled by
2393 @option{-ffreestanding} or @option{-fno-builtin}.
2395 The formats are checked against the format features supported by GNU
2396 libc version 2.2. These include all ISO C90 and C99 features, as well
2397 as features from the Single Unix Specification and some BSD and GNU
2398 extensions. Other library implementations may not support all these
2399 features; GCC does not support warning about features that go beyond a
2400 particular library's limitations. However, if @option{-pedantic} is used
2401 with @option{-Wformat}, warnings will be given about format features not
2402 in the selected standard version (but not for @code{strfmon} formats,
2403 since those are not in any version of the C standard). @xref{C Dialect
2404 Options,,Options Controlling C Dialect}.
2406 Since @option{-Wformat} also checks for null format arguments for
2407 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2409 @option{-Wformat} is included in @option{-Wall}. For more control over some
2410 aspects of format checking, the options @option{-Wformat-y2k},
2411 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2412 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2413 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2416 @opindex Wformat-y2k
2417 If @option{-Wformat} is specified, also warn about @code{strftime}
2418 formats which may yield only a two-digit year.
2420 @item -Wno-format-extra-args
2421 @opindex Wno-format-extra-args
2422 If @option{-Wformat} is specified, do not warn about excess arguments to a
2423 @code{printf} or @code{scanf} format function. The C standard specifies
2424 that such arguments are ignored.
2426 Where the unused arguments lie between used arguments that are
2427 specified with @samp{$} operand number specifications, normally
2428 warnings are still given, since the implementation could not know what
2429 type to pass to @code{va_arg} to skip the unused arguments. However,
2430 in the case of @code{scanf} formats, this option will suppress the
2431 warning if the unused arguments are all pointers, since the Single
2432 Unix Specification says that such unused arguments are allowed.
2434 @item -Wno-format-zero-length
2435 @opindex Wno-format-zero-length
2436 If @option{-Wformat} is specified, do not warn about zero-length formats.
2437 The C standard specifies that zero-length formats are allowed.
2439 @item -Wformat-nonliteral
2440 @opindex Wformat-nonliteral
2441 If @option{-Wformat} is specified, also warn if the format string is not a
2442 string literal and so cannot be checked, unless the format function
2443 takes its format arguments as a @code{va_list}.
2445 @item -Wformat-security
2446 @opindex Wformat-security
2447 If @option{-Wformat} is specified, also warn about uses of format
2448 functions that represent possible security problems. At present, this
2449 warns about calls to @code{printf} and @code{scanf} functions where the
2450 format string is not a string literal and there are no format arguments,
2451 as in @code{printf (foo);}. This may be a security hole if the format
2452 string came from untrusted input and contains @samp{%n}. (This is
2453 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2454 in future warnings may be added to @option{-Wformat-security} that are not
2455 included in @option{-Wformat-nonliteral}.)
2459 Enable @option{-Wformat} plus format checks not included in
2460 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2461 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2465 Warn about passing a null pointer for arguments marked as
2466 requiring a non-null value by the @code{nonnull} function attribute.
2468 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2469 can be disabled with the @option{-Wno-nonnull} option.
2471 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2473 Warn about uninitialized variables which are initialized with themselves.
2474 Note this option can only be used with the @option{-Wuninitialized} option,
2475 which in turn only works with @option{-O1} and above.
2477 For example, GCC will warn about @code{i} being uninitialized in the
2478 following snippet only when @option{-Winit-self} has been specified:
2489 @item -Wimplicit-int
2490 @opindex Wimplicit-int
2491 Warn when a declaration does not specify a type.
2492 This warning is enabled by @option{-Wall}.
2494 @item -Wimplicit-function-declaration
2495 @itemx -Werror-implicit-function-declaration
2496 @opindex Wimplicit-function-declaration
2497 @opindex Werror-implicit-function-declaration
2498 Give a warning (or error) whenever a function is used before being
2499 declared. The form @option{-Wno-error-implicit-function-declaration}
2501 This warning is enabled by @option{-Wall} (as a warning, not an error).
2505 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2506 This warning is enabled by @option{-Wall}.
2510 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2511 function with external linkage, returning int, taking either zero
2512 arguments, two, or three arguments of appropriate types.
2513 This warning is enabled by @option{-Wall}.
2515 @item -Wmissing-braces
2516 @opindex Wmissing-braces
2517 Warn if an aggregate or union initializer is not fully bracketed. In
2518 the following example, the initializer for @samp{a} is not fully
2519 bracketed, but that for @samp{b} is fully bracketed.
2522 int a[2][2] = @{ 0, 1, 2, 3 @};
2523 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2526 This warning is enabled by @option{-Wall}.
2528 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2529 @opindex Wmissing-include-dirs
2530 Warn if a user-supplied include directory does not exist.
2533 @opindex Wparentheses
2534 Warn if parentheses are omitted in certain contexts, such
2535 as when there is an assignment in a context where a truth value
2536 is expected, or when operators are nested whose precedence people
2537 often get confused about. Only the warning for an assignment used as
2538 a truth value is supported when compiling C++; the other warnings are
2539 only supported when compiling C@.
2541 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2542 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2543 interpretation from that of ordinary mathematical notation.
2545 Also warn about constructions where there may be confusion to which
2546 @code{if} statement an @code{else} branch belongs. Here is an example of
2561 In C, every @code{else} branch belongs to the innermost possible @code{if}
2562 statement, which in this example is @code{if (b)}. This is often not
2563 what the programmer expected, as illustrated in the above example by
2564 indentation the programmer chose. When there is the potential for this
2565 confusion, GCC will issue a warning when this flag is specified.
2566 To eliminate the warning, add explicit braces around the innermost
2567 @code{if} statement so there is no way the @code{else} could belong to
2568 the enclosing @code{if}. The resulting code would look like this:
2584 This warning is enabled by @option{-Wall}.
2586 @item -Wsequence-point
2587 @opindex Wsequence-point
2588 Warn about code that may have undefined semantics because of violations
2589 of sequence point rules in the C and C++ standards.
2591 The C and C++ standards defines the order in which expressions in a C/C++
2592 program are evaluated in terms of @dfn{sequence points}, which represent
2593 a partial ordering between the execution of parts of the program: those
2594 executed before the sequence point, and those executed after it. These
2595 occur after the evaluation of a full expression (one which is not part
2596 of a larger expression), after the evaluation of the first operand of a
2597 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2598 function is called (but after the evaluation of its arguments and the
2599 expression denoting the called function), and in certain other places.
2600 Other than as expressed by the sequence point rules, the order of
2601 evaluation of subexpressions of an expression is not specified. All
2602 these rules describe only a partial order rather than a total order,
2603 since, for example, if two functions are called within one expression
2604 with no sequence point between them, the order in which the functions
2605 are called is not specified. However, the standards committee have
2606 ruled that function calls do not overlap.
2608 It is not specified when between sequence points modifications to the
2609 values of objects take effect. Programs whose behavior depends on this
2610 have undefined behavior; the C and C++ standards specify that ``Between
2611 the previous and next sequence point an object shall have its stored
2612 value modified at most once by the evaluation of an expression.
2613 Furthermore, the prior value shall be read only to determine the value
2614 to be stored.''. If a program breaks these rules, the results on any
2615 particular implementation are entirely unpredictable.
2617 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2618 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2619 diagnosed by this option, and it may give an occasional false positive
2620 result, but in general it has been found fairly effective at detecting
2621 this sort of problem in programs.
2623 The standard is worded confusingly, therefore there is some debate
2624 over the precise meaning of the sequence point rules in subtle cases.
2625 Links to discussions of the problem, including proposed formal
2626 definitions, may be found on the GCC readings page, at
2627 @w{@uref{http://gcc.gnu.org/readings.html}}.
2629 This warning is enabled by @option{-Wall} for C and C++.
2632 @opindex Wreturn-type
2633 Warn whenever a function is defined with a return-type that defaults to
2634 @code{int}. Also warn about any @code{return} statement with no
2635 return-value in a function whose return-type is not @code{void}.
2637 For C, also warn if the return type of a function has a type qualifier
2638 such as @code{const}. Such a type qualifier has no effect, since the
2639 value returned by a function is not an lvalue. ISO C prohibits
2640 qualified @code{void} return types on function definitions, so such
2641 return types always receive a warning even without this option.
2643 For C++, a function without return type always produces a diagnostic
2644 message, even when @option{-Wno-return-type} is specified. The only
2645 exceptions are @samp{main} and functions defined in system headers.
2647 This warning is enabled by @option{-Wall}.
2651 Warn whenever a @code{switch} statement has an index of enumerated type
2652 and lacks a @code{case} for one or more of the named codes of that
2653 enumeration. (The presence of a @code{default} label prevents this
2654 warning.) @code{case} labels outside the enumeration range also
2655 provoke warnings when this option is used.
2656 This warning is enabled by @option{-Wall}.
2658 @item -Wswitch-default
2659 @opindex Wswitch-switch
2660 Warn whenever a @code{switch} statement does not have a @code{default}
2664 @opindex Wswitch-enum
2665 Warn whenever a @code{switch} statement has an index of enumerated type
2666 and lacks a @code{case} for one or more of the named codes of that
2667 enumeration. @code{case} labels outside the enumeration range also
2668 provoke warnings when this option is used.
2672 Warn if any trigraphs are encountered that might change the meaning of
2673 the program (trigraphs within comments are not warned about).
2674 This warning is enabled by @option{-Wall}.
2676 @item -Wunused-function
2677 @opindex Wunused-function
2678 Warn whenever a static function is declared but not defined or a
2679 non-inline static function is unused.
2680 This warning is enabled by @option{-Wall}.
2682 @item -Wunused-label
2683 @opindex Wunused-label
2684 Warn whenever a label is declared but not used.
2685 This warning is enabled by @option{-Wall}.
2687 To suppress this warning use the @samp{unused} attribute
2688 (@pxref{Variable Attributes}).
2690 @item -Wunused-parameter
2691 @opindex Wunused-parameter
2692 Warn whenever a function parameter is unused aside from its declaration.
2694 To suppress this warning use the @samp{unused} attribute
2695 (@pxref{Variable Attributes}).
2697 @item -Wunused-variable
2698 @opindex Wunused-variable
2699 Warn whenever a local variable or non-constant static variable is unused
2700 aside from its declaration.
2701 This warning is enabled by @option{-Wall}.
2703 To suppress this warning use the @samp{unused} attribute
2704 (@pxref{Variable Attributes}).
2706 @item -Wunused-value
2707 @opindex Wunused-value
2708 Warn whenever a statement computes a result that is explicitly not used.
2709 This warning is enabled by @option{-Wall}.
2711 To suppress this warning cast the expression to @samp{void}.
2715 All the above @option{-Wunused} options combined.
2717 In order to get a warning about an unused function parameter, you must
2718 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2719 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2721 @item -Wuninitialized
2722 @opindex Wuninitialized
2723 Warn if an automatic variable is used without first being initialized or
2724 if a variable may be clobbered by a @code{setjmp} call.
2726 These warnings are possible only in optimizing compilation,
2727 because they require data flow information that is computed only
2728 when optimizing. If you do not specify @option{-O}, you will not get
2729 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
2730 requiring @option{-O}.
2732 If you want to warn about code which uses the uninitialized value of the
2733 variable in its own initializer, use the @option{-Winit-self} option.
2735 These warnings occur for individual uninitialized or clobbered
2736 elements of structure, union or array variables as well as for
2737 variables which are uninitialized or clobbered as a whole. They do
2738 not occur for variables or elements declared @code{volatile}. Because
2739 these warnings depend on optimization, the exact variables or elements
2740 for which there are warnings will depend on the precise optimization
2741 options and version of GCC used.
2743 Note that there may be no warning about a variable that is used only
2744 to compute a value that itself is never used, because such
2745 computations may be deleted by data flow analysis before the warnings
2748 These warnings are made optional because GCC is not smart
2749 enough to see all the reasons why the code might be correct
2750 despite appearing to have an error. Here is one example of how
2771 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2772 always initialized, but GCC doesn't know this. Here is
2773 another common case:
2778 if (change_y) save_y = y, y = new_y;
2780 if (change_y) y = save_y;
2785 This has no bug because @code{save_y} is used only if it is set.
2787 @cindex @code{longjmp} warnings
2788 This option also warns when a non-volatile automatic variable might be
2789 changed by a call to @code{longjmp}. These warnings as well are possible
2790 only in optimizing compilation.
2792 The compiler sees only the calls to @code{setjmp}. It cannot know
2793 where @code{longjmp} will be called; in fact, a signal handler could
2794 call it at any point in the code. As a result, you may get a warning
2795 even when there is in fact no problem because @code{longjmp} cannot
2796 in fact be called at the place which would cause a problem.
2798 Some spurious warnings can be avoided if you declare all the functions
2799 you use that never return as @code{noreturn}. @xref{Function
2802 This warning is enabled by @option{-Wall}.
2804 @item -Wunknown-pragmas
2805 @opindex Wunknown-pragmas
2806 @cindex warning for unknown pragmas
2807 @cindex unknown pragmas, warning
2808 @cindex pragmas, warning of unknown
2809 Warn when a #pragma directive is encountered which is not understood by
2810 GCC@. If this command line option is used, warnings will even be issued
2811 for unknown pragmas in system header files. This is not the case if
2812 the warnings were only enabled by the @option{-Wall} command line option.
2815 @opindex Wno-pragmas
2817 Do not warn about misuses of pragmas, such as incorrect parameters,
2818 invalid syntax, or conflicts between pragmas. See also
2819 @samp{-Wunknown-pragmas}.
2821 @item -Wstrict-aliasing
2822 @opindex Wstrict-aliasing
2823 This option is only active when @option{-fstrict-aliasing} is active.
2824 It warns about code which might break the strict aliasing rules that the
2825 compiler is using for optimization. The warning does not catch all
2826 cases, but does attempt to catch the more common pitfalls. It is
2827 included in @option{-Wall}.
2829 @item -Wstrict-aliasing=2
2830 @opindex Wstrict-aliasing=2
2831 This option is only active when @option{-fstrict-aliasing} is active.
2832 It warns about code which might break the strict aliasing rules that the
2833 compiler is using for optimization. This warning catches more cases than
2834 @option{-Wstrict-aliasing}, but it will also give a warning for some ambiguous
2835 cases that are safe.
2837 @item -Wstrict-overflow
2838 @item -Wstrict-overflow=@var{n}
2839 @opindex Wstrict-overflow
2840 This option is only active when @option{-fstrict-overflow} is active.
2841 It warns about cases where the compiler optimizes based on the
2842 assumption that signed overflow does not occur. Note that it does not
2843 warn about all cases where the code might overflow: it only warns
2844 about cases where the compiler implements some optimization. Thus
2845 this warning depends on the optimization level.
2847 An optimization which assumes that signed overflow does not occur is
2848 perfectly safe if the values of the variables involved are such that
2849 overflow never does, in fact, occur. Therefore this warning can
2850 easily give a false positive: a warning about code which is not
2851 actually a problem. To help focus on important issues, several
2852 warning levels are defined. No warnings are issued for the use of
2853 undefined signed overflow when estimating how many iterations a loop
2854 will require, in particular when determining whether a loop will be
2858 @item -Wstrict-overflow=1
2859 Warn about cases which are both questionable and easy to avoid. For
2860 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
2861 compiler will simplify this to @code{1}. This level of
2862 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
2863 are not, and must be explicitly requested.
2865 @item -Wstrict-overflow=2
2866 Also warn about other cases where a comparison is simplified to a
2867 constant. For example: @code{abs (x) >= 0}. This can only be
2868 simplified when @option{-fstrict-overflow} is in effect, because
2869 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
2870 zero. @option{-Wstrict-overflow} (with no level) is the same as
2871 @option{-Wstrict-overflow=2}.
2873 @item -Wstrict-overflow=3
2874 Also warn about other cases where a comparison is simplified. For
2875 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
2877 @item -Wstrict-overflow=4
2878 Also warn about other simplifications not covered by the above cases.
2879 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
2881 @item -Wstrict-overflow=5
2882 Also warn about cases where the compiler reduces the magnitude of a
2883 constant involved in a comparison. For example: @code{x + 2 > y} will
2884 be simplified to @code{x + 1 >= y}. This is reported only at the
2885 highest warning level because this simplification applies to many
2886 comparisons, so this warning level will give a very large number of
2892 All of the above @samp{-W} options combined. This enables all the
2893 warnings about constructions that some users consider questionable, and
2894 that are easy to avoid (or modify to prevent the warning), even in
2895 conjunction with macros. This also enables some language-specific
2896 warnings described in @ref{C++ Dialect Options} and
2897 @ref{Objective-C and Objective-C++ Dialect Options}.
2900 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2901 Some of them warn about constructions that users generally do not
2902 consider questionable, but which occasionally you might wish to check
2903 for; others warn about constructions that are necessary or hard to avoid
2904 in some cases, and there is no simple way to modify the code to suppress
2911 (This option used to be called @option{-W}. The older name is still
2912 supported, but the newer name is more descriptive.) Print extra warning
2913 messages for these events:
2917 A function can return either with or without a value. (Falling
2918 off the end of the function body is considered returning without
2919 a value.) For example, this function would evoke such a
2933 An expression-statement or the left-hand side of a comma expression
2934 contains no side effects.
2935 To suppress the warning, cast the unused expression to void.
2936 For example, an expression such as @samp{x[i,j]} will cause a warning,
2937 but @samp{x[(void)i,j]} will not.
2940 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2943 Storage-class specifiers like @code{static} are not the first things in
2944 a declaration. According to the C Standard, this usage is obsolescent.
2947 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2951 A comparison between signed and unsigned values could produce an
2952 incorrect result when the signed value is converted to unsigned.
2953 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2956 An aggregate has an initializer which does not initialize all members.
2957 This warning can be independently controlled by
2958 @option{-Wmissing-field-initializers}.
2961 An initialized field without side effects is overridden when using
2962 designated initializers (@pxref{Designated Inits, , Designated
2963 Initializers}). This warning can be independently controlled by
2964 @option{-Woverride-init}.
2967 A function parameter is declared without a type specifier in K&R-style
2975 An empty body occurs in an @samp{if} or @samp{else} statement.
2978 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2979 @samp{>}, or @samp{>=}.
2982 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2984 @item @r{(C++ only)}
2985 An enumerator and a non-enumerator both appear in a conditional expression.
2987 @item @r{(C++ only)}
2988 A non-static reference or non-static @samp{const} member appears in a
2989 class without constructors.
2991 @item @r{(C++ only)}
2992 Ambiguous virtual bases.
2994 @item @r{(C++ only)}
2995 Subscripting an array which has been declared @samp{register}.
2997 @item @r{(C++ only)}
2998 Taking the address of a variable which has been declared @samp{register}.
3000 @item @r{(C++ only)}
3001 A base class is not initialized in a derived class' copy constructor.
3004 @item -Wno-div-by-zero
3005 @opindex Wno-div-by-zero
3006 @opindex Wdiv-by-zero
3007 Do not warn about compile-time integer division by zero. Floating point
3008 division by zero is not warned about, as it can be a legitimate way of
3009 obtaining infinities and NaNs.
3011 @item -Wsystem-headers
3012 @opindex Wsystem-headers
3013 @cindex warnings from system headers
3014 @cindex system headers, warnings from
3015 Print warning messages for constructs found in system header files.
3016 Warnings from system headers are normally suppressed, on the assumption
3017 that they usually do not indicate real problems and would only make the
3018 compiler output harder to read. Using this command line option tells
3019 GCC to emit warnings from system headers as if they occurred in user
3020 code. However, note that using @option{-Wall} in conjunction with this
3021 option will @emph{not} warn about unknown pragmas in system
3022 headers---for that, @option{-Wunknown-pragmas} must also be used.
3025 @opindex Wfloat-equal
3026 Warn if floating point values are used in equality comparisons.
3028 The idea behind this is that sometimes it is convenient (for the
3029 programmer) to consider floating-point values as approximations to
3030 infinitely precise real numbers. If you are doing this, then you need
3031 to compute (by analyzing the code, or in some other way) the maximum or
3032 likely maximum error that the computation introduces, and allow for it
3033 when performing comparisons (and when producing output, but that's a
3034 different problem). In particular, instead of testing for equality, you
3035 would check to see whether the two values have ranges that overlap; and
3036 this is done with the relational operators, so equality comparisons are
3039 @item -Wtraditional @r{(C only)}
3040 @opindex Wtraditional
3041 Warn about certain constructs that behave differently in traditional and
3042 ISO C@. Also warn about ISO C constructs that have no traditional C
3043 equivalent, and/or problematic constructs which should be avoided.
3047 Macro parameters that appear within string literals in the macro body.
3048 In traditional C macro replacement takes place within string literals,
3049 but does not in ISO C@.
3052 In traditional C, some preprocessor directives did not exist.
3053 Traditional preprocessors would only consider a line to be a directive
3054 if the @samp{#} appeared in column 1 on the line. Therefore
3055 @option{-Wtraditional} warns about directives that traditional C
3056 understands but would ignore because the @samp{#} does not appear as the
3057 first character on the line. It also suggests you hide directives like
3058 @samp{#pragma} not understood by traditional C by indenting them. Some
3059 traditional implementations would not recognize @samp{#elif}, so it
3060 suggests avoiding it altogether.
3063 A function-like macro that appears without arguments.
3066 The unary plus operator.
3069 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3070 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3071 constants.) Note, these suffixes appear in macros defined in the system
3072 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3073 Use of these macros in user code might normally lead to spurious
3074 warnings, however GCC's integrated preprocessor has enough context to
3075 avoid warning in these cases.
3078 A function declared external in one block and then used after the end of
3082 A @code{switch} statement has an operand of type @code{long}.
3085 A non-@code{static} function declaration follows a @code{static} one.
3086 This construct is not accepted by some traditional C compilers.
3089 The ISO type of an integer constant has a different width or
3090 signedness from its traditional type. This warning is only issued if
3091 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3092 typically represent bit patterns, are not warned about.
3095 Usage of ISO string concatenation is detected.
3098 Initialization of automatic aggregates.
3101 Identifier conflicts with labels. Traditional C lacks a separate
3102 namespace for labels.
3105 Initialization of unions. If the initializer is zero, the warning is
3106 omitted. This is done under the assumption that the zero initializer in
3107 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3108 initializer warnings and relies on default initialization to zero in the
3112 Conversions by prototypes between fixed/floating point values and vice
3113 versa. The absence of these prototypes when compiling with traditional
3114 C would cause serious problems. This is a subset of the possible
3115 conversion warnings, for the full set use @option{-Wconversion}.
3118 Use of ISO C style function definitions. This warning intentionally is
3119 @emph{not} issued for prototype declarations or variadic functions
3120 because these ISO C features will appear in your code when using
3121 libiberty's traditional C compatibility macros, @code{PARAMS} and
3122 @code{VPARAMS}. This warning is also bypassed for nested functions
3123 because that feature is already a GCC extension and thus not relevant to
3124 traditional C compatibility.
3127 @item -Wdeclaration-after-statement @r{(C only)}
3128 @opindex Wdeclaration-after-statement
3129 Warn when a declaration is found after a statement in a block. This
3130 construct, known from C++, was introduced with ISO C99 and is by default
3131 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3132 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3136 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3138 @item -Wno-endif-labels
3139 @opindex Wno-endif-labels
3140 @opindex Wendif-labels
3141 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3145 Warn whenever a local variable shadows another local variable, parameter or
3146 global variable or whenever a built-in function is shadowed.
3148 @item -Wlarger-than-@var{len}
3149 @opindex Wlarger-than
3150 Warn whenever an object of larger than @var{len} bytes is defined.
3152 @item -Wunsafe-loop-optimizations
3153 @opindex Wunsafe-loop-optimizations
3154 Warn if the loop cannot be optimized because the compiler could not
3155 assume anything on the bounds of the loop indices. With
3156 @option{-funsafe-loop-optimizations} warn if the compiler made
3159 @item -Wpointer-arith
3160 @opindex Wpointer-arith
3161 Warn about anything that depends on the ``size of'' a function type or
3162 of @code{void}. GNU C assigns these types a size of 1, for
3163 convenience in calculations with @code{void *} pointers and pointers
3166 @item -Wbad-function-cast @r{(C only)}
3167 @opindex Wbad-function-cast
3168 Warn whenever a function call is cast to a non-matching type.
3169 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3172 Warn about ISO C constructs that are outside of the common subset of
3173 ISO C and ISO C++, e.g.@: request for implicit conversion from
3174 @code{void *} to a pointer to non-@code{void} type.
3178 Warn whenever a pointer is cast so as to remove a type qualifier from
3179 the target type. For example, warn if a @code{const char *} is cast
3180 to an ordinary @code{char *}.
3183 @opindex Wcast-align
3184 Warn whenever a pointer is cast such that the required alignment of the
3185 target is increased. For example, warn if a @code{char *} is cast to
3186 an @code{int *} on machines where integers can only be accessed at
3187 two- or four-byte boundaries.
3189 @item -Wwrite-strings
3190 @opindex Wwrite-strings
3191 When compiling C, give string constants the type @code{const
3192 char[@var{length}]} so that
3193 copying the address of one into a non-@code{const} @code{char *}
3194 pointer will get a warning; when compiling C++, warn about the
3195 deprecated conversion from string literals to @code{char *}. This
3196 warning, by default, is enabled for C++ programs.
3197 These warnings will help you find at
3198 compile time code that can try to write into a string constant, but
3199 only if you have been very careful about using @code{const} in
3200 declarations and prototypes. Otherwise, it will just be a nuisance;
3201 this is why we did not make @option{-Wall} request these warnings.
3204 @opindex Wconversion
3205 Warn if a prototype causes a type conversion that is different from what
3206 would happen to the same argument in the absence of a prototype. This
3207 includes conversions of fixed point to floating and vice versa, and
3208 conversions changing the width or signedness of a fixed point argument
3209 except when the same as the default promotion.
3211 Also, warn if a negative integer constant expression is implicitly
3212 converted to an unsigned type. For example, warn about the assignment
3213 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
3214 casts like @code{(unsigned) -1}.
3216 @item -Wsign-compare
3217 @opindex Wsign-compare
3218 @cindex warning for comparison of signed and unsigned values
3219 @cindex comparison of signed and unsigned values, warning
3220 @cindex signed and unsigned values, comparison warning
3221 Warn when a comparison between signed and unsigned values could produce
3222 an incorrect result when the signed value is converted to unsigned.
3223 This warning is also enabled by @option{-Wextra}; to get the other warnings
3224 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3228 @opindex Wno-address
3229 Warn about suspicious uses of memory addresses. These include using
3230 the address of a function in a conditional expression, such as
3231 @code{void func(void); if (func)}, and comparisons against the memory
3232 address of a string literal, such as @code{if (x == "abc")}. Such
3233 uses typically indicate a programmer error: the address of a function
3234 always evaluates to true, so their use in a conditional usually
3235 indicate that the programmer forgot the parentheses in a function
3236 call; and comparisons against string literals result in unspecified
3237 behavior and are not portable in C, so they usually indicate that the
3238 programmer intended to use @code{strcmp}. This warning is enabled by
3241 @item -Waggregate-return
3242 @opindex Waggregate-return
3243 Warn if any functions that return structures or unions are defined or
3244 called. (In languages where you can return an array, this also elicits
3247 @item -Wno-attributes
3248 @opindex Wno-attributes
3249 @opindex Wattributes
3250 Do not warn if an unexpected @code{__attribute__} is used, such as
3251 unrecognized attributes, function attributes applied to variables,
3252 etc. This will not stop errors for incorrect use of supported
3255 @item -Wstrict-prototypes @r{(C only)}
3256 @opindex Wstrict-prototypes
3257 Warn if a function is declared or defined without specifying the
3258 argument types. (An old-style function definition is permitted without
3259 a warning if preceded by a declaration which specifies the argument
3262 @item -Wold-style-definition @r{(C only)}
3263 @opindex Wold-style-definition
3264 Warn if an old-style function definition is used. A warning is given
3265 even if there is a previous prototype.
3267 @item -Wmissing-prototypes @r{(C only)}
3268 @opindex Wmissing-prototypes
3269 Warn if a global function is defined without a previous prototype
3270 declaration. This warning is issued even if the definition itself
3271 provides a prototype. The aim is to detect global functions that fail
3272 to be declared in header files.
3274 @item -Wmissing-declarations @r{(C only)}
3275 @opindex Wmissing-declarations
3276 Warn if a global function is defined without a previous declaration.
3277 Do so even if the definition itself provides a prototype.
3278 Use this option to detect global functions that are not declared in
3281 @item -Wmissing-field-initializers
3282 @opindex Wmissing-field-initializers
3285 Warn if a structure's initializer has some fields missing. For
3286 example, the following code would cause such a warning, because
3287 @code{x.h} is implicitly zero:
3290 struct s @{ int f, g, h; @};
3291 struct s x = @{ 3, 4 @};
3294 This option does not warn about designated initializers, so the following
3295 modification would not trigger a warning:
3298 struct s @{ int f, g, h; @};
3299 struct s x = @{ .f = 3, .g = 4 @};
3302 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3303 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3305 @item -Wmissing-noreturn
3306 @opindex Wmissing-noreturn
3307 Warn about functions which might be candidates for attribute @code{noreturn}.
3308 Note these are only possible candidates, not absolute ones. Care should
3309 be taken to manually verify functions actually do not ever return before
3310 adding the @code{noreturn} attribute, otherwise subtle code generation
3311 bugs could be introduced. You will not get a warning for @code{main} in
3312 hosted C environments.
3314 @item -Wmissing-format-attribute
3315 @opindex Wmissing-format-attribute
3317 Warn about function pointers which might be candidates for @code{format}
3318 attributes. Note these are only possible candidates, not absolute ones.
3319 GCC will guess that function pointers with @code{format} attributes that
3320 are used in assignment, initialization, parameter passing or return
3321 statements should have a corresponding @code{format} attribute in the
3322 resulting type. I.e.@: the left-hand side of the assignment or
3323 initialization, the type of the parameter variable, or the return type
3324 of the containing function respectively should also have a @code{format}
3325 attribute to avoid the warning.
3327 GCC will also warn about function definitions which might be
3328 candidates for @code{format} attributes. Again, these are only
3329 possible candidates. GCC will guess that @code{format} attributes
3330 might be appropriate for any function that calls a function like
3331 @code{vprintf} or @code{vscanf}, but this might not always be the
3332 case, and some functions for which @code{format} attributes are
3333 appropriate may not be detected.
3335 @item -Wno-multichar
3336 @opindex Wno-multichar
3338 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3339 Usually they indicate a typo in the user's code, as they have
3340 implementation-defined values, and should not be used in portable code.
3342 @item -Wnormalized=<none|id|nfc|nfkc>
3343 @opindex Wnormalized
3346 @cindex character set, input normalization
3347 In ISO C and ISO C++, two identifiers are different if they are
3348 different sequences of characters. However, sometimes when characters
3349 outside the basic ASCII character set are used, you can have two
3350 different character sequences that look the same. To avoid confusion,
3351 the ISO 10646 standard sets out some @dfn{normalization rules} which
3352 when applied ensure that two sequences that look the same are turned into
3353 the same sequence. GCC can warn you if you are using identifiers which
3354 have not been normalized; this option controls that warning.
3356 There are four levels of warning that GCC supports. The default is
3357 @option{-Wnormalized=nfc}, which warns about any identifier which is
3358 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3359 recommended form for most uses.
3361 Unfortunately, there are some characters which ISO C and ISO C++ allow
3362 in identifiers that when turned into NFC aren't allowable as
3363 identifiers. That is, there's no way to use these symbols in portable
3364 ISO C or C++ and have all your identifiers in NFC.
3365 @option{-Wnormalized=id} suppresses the warning for these characters.
3366 It is hoped that future versions of the standards involved will correct
3367 this, which is why this option is not the default.
3369 You can switch the warning off for all characters by writing
3370 @option{-Wnormalized=none}. You would only want to do this if you
3371 were using some other normalization scheme (like ``D''), because
3372 otherwise you can easily create bugs that are literally impossible to see.
3374 Some characters in ISO 10646 have distinct meanings but look identical
3375 in some fonts or display methodologies, especially once formatting has
3376 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3377 LETTER N'', will display just like a regular @code{n} which has been
3378 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3379 normalization scheme to convert all these into a standard form as
3380 well, and GCC will warn if your code is not in NFKC if you use
3381 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3382 about every identifier that contains the letter O because it might be
3383 confused with the digit 0, and so is not the default, but may be
3384 useful as a local coding convention if the programming environment is
3385 unable to be fixed to display these characters distinctly.
3387 @item -Wno-deprecated-declarations
3388 @opindex Wno-deprecated-declarations
3389 Do not warn about uses of functions (@pxref{Function Attributes}),
3390 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3391 Attributes}) marked as deprecated by using the @code{deprecated}
3395 @opindex Wno-overflow
3396 Do not warn about compile-time overflow in constant expressions.
3398 @item -Woverride-init
3399 @opindex Woverride-init
3402 Warn if an initialized field without side effects is overridden when
3403 using designated initializers (@pxref{Designated Inits, , Designated
3406 This warning is included in @option{-Wextra}. To get other
3407 @option{-Wextra} warnings without this one, use @samp{-Wextra
3408 -Wno-override-init}.
3412 Warn if a structure is given the packed attribute, but the packed
3413 attribute has no effect on the layout or size of the structure.
3414 Such structures may be mis-aligned for little benefit. For
3415 instance, in this code, the variable @code{f.x} in @code{struct bar}
3416 will be misaligned even though @code{struct bar} does not itself
3417 have the packed attribute:
3424 @} __attribute__((packed));
3434 Warn if padding is included in a structure, either to align an element
3435 of the structure or to align the whole structure. Sometimes when this
3436 happens it is possible to rearrange the fields of the structure to
3437 reduce the padding and so make the structure smaller.
3439 @item -Wredundant-decls
3440 @opindex Wredundant-decls
3441 Warn if anything is declared more than once in the same scope, even in
3442 cases where multiple declaration is valid and changes nothing.
3444 @item -Wnested-externs @r{(C only)}
3445 @opindex Wnested-externs
3446 Warn if an @code{extern} declaration is encountered within a function.
3448 @item -Wunreachable-code
3449 @opindex Wunreachable-code
3450 Warn if the compiler detects that code will never be executed.
3452 This option is intended to warn when the compiler detects that at
3453 least a whole line of source code will never be executed, because
3454 some condition is never satisfied or because it is after a
3455 procedure that never returns.
3457 It is possible for this option to produce a warning even though there
3458 are circumstances under which part of the affected line can be executed,
3459 so care should be taken when removing apparently-unreachable code.
3461 For instance, when a function is inlined, a warning may mean that the
3462 line is unreachable in only one inlined copy of the function.
3464 This option is not made part of @option{-Wall} because in a debugging
3465 version of a program there is often substantial code which checks
3466 correct functioning of the program and is, hopefully, unreachable
3467 because the program does work. Another common use of unreachable
3468 code is to provide behavior which is selectable at compile-time.
3472 Warn if a function can not be inlined and it was declared as inline.
3473 Even with this option, the compiler will not warn about failures to
3474 inline functions declared in system headers.
3476 The compiler uses a variety of heuristics to determine whether or not
3477 to inline a function. For example, the compiler takes into account
3478 the size of the function being inlined and the amount of inlining
3479 that has already been done in the current function. Therefore,
3480 seemingly insignificant changes in the source program can cause the
3481 warnings produced by @option{-Winline} to appear or disappear.
3483 @item -Wno-invalid-offsetof @r{(C++ only)}
3484 @opindex Wno-invalid-offsetof
3485 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3486 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3487 to a non-POD type is undefined. In existing C++ implementations,
3488 however, @samp{offsetof} typically gives meaningful results even when
3489 applied to certain kinds of non-POD types. (Such as a simple
3490 @samp{struct} that fails to be a POD type only by virtue of having a
3491 constructor.) This flag is for users who are aware that they are
3492 writing nonportable code and who have deliberately chosen to ignore the
3495 The restrictions on @samp{offsetof} may be relaxed in a future version
3496 of the C++ standard.
3498 @item -Wno-int-to-pointer-cast @r{(C only)}
3499 @opindex Wno-int-to-pointer-cast
3500 Suppress warnings from casts to pointer type of an integer of a
3503 @item -Wno-pointer-to-int-cast @r{(C only)}
3504 @opindex Wno-pointer-to-int-cast
3505 Suppress warnings from casts from a pointer to an integer type of a
3509 @opindex Winvalid-pch
3510 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3511 the search path but can't be used.
3515 @opindex Wno-long-long
3516 Warn if @samp{long long} type is used. This is default. To inhibit
3517 the warning messages, use @option{-Wno-long-long}. Flags
3518 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3519 only when @option{-pedantic} flag is used.
3521 @item -Wvariadic-macros
3522 @opindex Wvariadic-macros
3523 @opindex Wno-variadic-macros
3524 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3525 alternate syntax when in pedantic ISO C99 mode. This is default.
3526 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3528 @item -Wvolatile-register-var
3529 @opindex Wvolatile-register-var
3530 @opindex Wno-volatile-register-var
3531 Warn if a register variable is declared volatile. The volatile
3532 modifier does not inhibit all optimizations that may eliminate reads
3533 and/or writes to register variables.
3535 @item -Wdisabled-optimization
3536 @opindex Wdisabled-optimization
3537 Warn if a requested optimization pass is disabled. This warning does
3538 not generally indicate that there is anything wrong with your code; it
3539 merely indicates that GCC's optimizers were unable to handle the code
3540 effectively. Often, the problem is that your code is too big or too
3541 complex; GCC will refuse to optimize programs when the optimization
3542 itself is likely to take inordinate amounts of time.
3544 @item -Wpointer-sign
3545 @opindex Wpointer-sign
3546 @opindex Wno-pointer-sign
3547 Warn for pointer argument passing or assignment with different signedness.
3548 This option is only supported for C and Objective-C@. It is implied by
3549 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3550 @option{-Wno-pointer-sign}.
3554 Make all warnings into errors.
3558 Make the specified warning into an errors. The specifier for a
3559 warning is appended, for example @option{-Werror=switch} turns the
3560 warnings controlled by @option{-Wswitch} into errors. This switch
3561 takes a negative form, to be used to negate @option{-Werror} for
3562 specific warnings, for example @option{-Wno-error=switch} makes
3563 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3564 is in effect. You can use the @option{-fdiagnostics-show-option}
3565 option to have each controllable warning amended with the option which
3566 controls it, to determine what to use with this option.
3568 Note that specifying @option{-Werror=}@var{foo} automatically implies
3569 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3572 @item -Wstack-protector
3573 @opindex Wstack-protector
3574 This option is only active when @option{-fstack-protector} is active. It
3575 warns about functions that will not be protected against stack smashing.
3577 @item -Woverlength-strings
3578 @opindex Woverlength-strings
3579 Warn about string constants which are longer than the ``minimum
3580 maximum'' length specified in the C standard. Modern compilers
3581 generally allow string constants which are much longer than the
3582 standard's minimum limit, but very portable programs should avoid
3583 using longer strings.
3585 The limit applies @emph{after} string constant concatenation, and does
3586 not count the trailing NUL@. In C89, the limit was 509 characters; in
3587 C99, it was raised to 4095. C++98 does not specify a normative
3588 minimum maximum, so we do not diagnose overlength strings in C++@.
3590 This option is implied by @option{-pedantic}, and can be disabled with
3591 @option{-Wno-overlength-strings}.
3594 @node Debugging Options
3595 @section Options for Debugging Your Program or GCC
3596 @cindex options, debugging
3597 @cindex debugging information options
3599 GCC has various special options that are used for debugging
3600 either your program or GCC:
3605 Produce debugging information in the operating system's native format
3606 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3609 On most systems that use stabs format, @option{-g} enables use of extra
3610 debugging information that only GDB can use; this extra information
3611 makes debugging work better in GDB but will probably make other debuggers
3613 refuse to read the program. If you want to control for certain whether
3614 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3615 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3617 GCC allows you to use @option{-g} with
3618 @option{-O}. The shortcuts taken by optimized code may occasionally
3619 produce surprising results: some variables you declared may not exist
3620 at all; flow of control may briefly move where you did not expect it;
3621 some statements may not be executed because they compute constant
3622 results or their values were already at hand; some statements may
3623 execute in different places because they were moved out of loops.
3625 Nevertheless it proves possible to debug optimized output. This makes
3626 it reasonable to use the optimizer for programs that might have bugs.
3628 The following options are useful when GCC is generated with the
3629 capability for more than one debugging format.
3633 Produce debugging information for use by GDB@. This means to use the
3634 most expressive format available (DWARF 2, stabs, or the native format
3635 if neither of those are supported), including GDB extensions if at all
3640 Produce debugging information in stabs format (if that is supported),
3641 without GDB extensions. This is the format used by DBX on most BSD
3642 systems. On MIPS, Alpha and System V Release 4 systems this option
3643 produces stabs debugging output which is not understood by DBX or SDB@.
3644 On System V Release 4 systems this option requires the GNU assembler.
3646 @item -feliminate-unused-debug-symbols
3647 @opindex feliminate-unused-debug-symbols
3648 Produce debugging information in stabs format (if that is supported),
3649 for only symbols that are actually used.
3651 @item -femit-class-debug-always
3652 Instead of emitting debugging information for a C++ class in only one
3653 object file, emit it in all object files using the class. This option
3654 should be used only with debuggers that are unable to handle the way GCC
3655 normally emits debugging information for classes because using this
3656 option will increase the size of debugging information by as much as a
3661 Produce debugging information in stabs format (if that is supported),
3662 using GNU extensions understood only by the GNU debugger (GDB)@. The
3663 use of these extensions is likely to make other debuggers crash or
3664 refuse to read the program.
3668 Produce debugging information in COFF format (if that is supported).
3669 This is the format used by SDB on most System V systems prior to
3674 Produce debugging information in XCOFF format (if that is supported).
3675 This is the format used by the DBX debugger on IBM RS/6000 systems.
3679 Produce debugging information in XCOFF format (if that is supported),
3680 using GNU extensions understood only by the GNU debugger (GDB)@. The
3681 use of these extensions is likely to make other debuggers crash or
3682 refuse to read the program, and may cause assemblers other than the GNU
3683 assembler (GAS) to fail with an error.
3687 Produce debugging information in DWARF version 2 format (if that is
3688 supported). This is the format used by DBX on IRIX 6. With this
3689 option, GCC uses features of DWARF version 3 when they are useful;
3690 version 3 is upward compatible with version 2, but may still cause
3691 problems for older debuggers.
3695 Produce debugging information in VMS debug format (if that is
3696 supported). This is the format used by DEBUG on VMS systems.
3699 @itemx -ggdb@var{level}
3700 @itemx -gstabs@var{level}
3701 @itemx -gcoff@var{level}
3702 @itemx -gxcoff@var{level}
3703 @itemx -gvms@var{level}
3704 Request debugging information and also use @var{level} to specify how
3705 much information. The default level is 2.
3707 Level 1 produces minimal information, enough for making backtraces in
3708 parts of the program that you don't plan to debug. This includes
3709 descriptions of functions and external variables, but no information
3710 about local variables and no line numbers.
3712 Level 3 includes extra information, such as all the macro definitions
3713 present in the program. Some debuggers support macro expansion when
3714 you use @option{-g3}.
3716 @option{-gdwarf-2} does not accept a concatenated debug level, because
3717 GCC used to support an option @option{-gdwarf} that meant to generate
3718 debug information in version 1 of the DWARF format (which is very
3719 different from version 2), and it would have been too confusing. That
3720 debug format is long obsolete, but the option cannot be changed now.
3721 Instead use an additional @option{-g@var{level}} option to change the
3722 debug level for DWARF2.
3724 @item -feliminate-dwarf2-dups
3725 @opindex feliminate-dwarf2-dups
3726 Compress DWARF2 debugging information by eliminating duplicated
3727 information about each symbol. This option only makes sense when
3728 generating DWARF2 debugging information with @option{-gdwarf-2}.
3730 @cindex @command{prof}
3733 Generate extra code to write profile information suitable for the
3734 analysis program @command{prof}. You must use this option when compiling
3735 the source files you want data about, and you must also use it when
3738 @cindex @command{gprof}
3741 Generate extra code to write profile information suitable for the
3742 analysis program @command{gprof}. You must use this option when compiling
3743 the source files you want data about, and you must also use it when
3748 Makes the compiler print out each function name as it is compiled, and
3749 print some statistics about each pass when it finishes.
3752 @opindex ftime-report
3753 Makes the compiler print some statistics about the time consumed by each
3754 pass when it finishes.
3757 @opindex fmem-report
3758 Makes the compiler print some statistics about permanent memory
3759 allocation when it finishes.
3761 @item -fprofile-arcs
3762 @opindex fprofile-arcs
3763 Add code so that program flow @dfn{arcs} are instrumented. During
3764 execution the program records how many times each branch and call is
3765 executed and how many times it is taken or returns. When the compiled
3766 program exits it saves this data to a file called
3767 @file{@var{auxname}.gcda} for each source file. The data may be used for
3768 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3769 test coverage analysis (@option{-ftest-coverage}). Each object file's
3770 @var{auxname} is generated from the name of the output file, if
3771 explicitly specified and it is not the final executable, otherwise it is
3772 the basename of the source file. In both cases any suffix is removed
3773 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3774 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3775 @xref{Cross-profiling}.
3777 @cindex @command{gcov}
3781 This option is used to compile and link code instrumented for coverage
3782 analysis. The option is a synonym for @option{-fprofile-arcs}
3783 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
3784 linking). See the documentation for those options for more details.
3789 Compile the source files with @option{-fprofile-arcs} plus optimization
3790 and code generation options. For test coverage analysis, use the
3791 additional @option{-ftest-coverage} option. You do not need to profile
3792 every source file in a program.
3795 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3796 (the latter implies the former).
3799 Run the program on a representative workload to generate the arc profile
3800 information. This may be repeated any number of times. You can run
3801 concurrent instances of your program, and provided that the file system
3802 supports locking, the data files will be correctly updated. Also
3803 @code{fork} calls are detected and correctly handled (double counting
3807 For profile-directed optimizations, compile the source files again with
3808 the same optimization and code generation options plus
3809 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3810 Control Optimization}).
3813 For test coverage analysis, use @command{gcov} to produce human readable
3814 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3815 @command{gcov} documentation for further information.
3819 With @option{-fprofile-arcs}, for each function of your program GCC
3820 creates a program flow graph, then finds a spanning tree for the graph.
3821 Only arcs that are not on the spanning tree have to be instrumented: the
3822 compiler adds code to count the number of times that these arcs are
3823 executed. When an arc is the only exit or only entrance to a block, the
3824 instrumentation code can be added to the block; otherwise, a new basic
3825 block must be created to hold the instrumentation code.
3828 @item -ftest-coverage
3829 @opindex ftest-coverage
3830 Produce a notes file that the @command{gcov} code-coverage utility
3831 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3832 show program coverage. Each source file's note file is called
3833 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3834 above for a description of @var{auxname} and instructions on how to
3835 generate test coverage data. Coverage data will match the source files
3836 more closely, if you do not optimize.
3838 @item -d@var{letters}
3839 @item -fdump-rtl-@var{pass}
3841 Says to make debugging dumps during compilation at times specified by
3842 @var{letters}. This is used for debugging the RTL-based passes of the
3843 compiler. The file names for most of the dumps are made by appending a
3844 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3845 from the name of the output file, if explicitly specified and it is not
3846 an executable, otherwise it is the basename of the source file.
3848 Most debug dumps can be enabled either passing a letter to the @option{-d}
3849 option, or with a long @option{-fdump-rtl} switch; here are the possible
3850 letters for use in @var{letters} and @var{pass}, and their meanings:
3855 Annotate the assembler output with miscellaneous debugging information.
3858 @itemx -fdump-rtl-bbro
3860 @opindex fdump-rtl-bbro
3861 Dump after block reordering, to @file{@var{file}.148r.bbro}.
3864 @itemx -fdump-rtl-combine
3866 @opindex fdump-rtl-combine
3867 Dump after instruction combination, to the file @file{@var{file}.129r.combine}.
3870 @itemx -fdump-rtl-ce1
3871 @itemx -fdump-rtl-ce2
3873 @opindex fdump-rtl-ce1
3874 @opindex fdump-rtl-ce2
3875 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3876 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
3877 and @option{-fdump-rtl-ce2} enable dumping after the second if
3878 conversion, to the file @file{@var{file}.130r.ce2}.
3881 @itemx -fdump-rtl-btl
3882 @itemx -fdump-rtl-dbr
3884 @opindex fdump-rtl-btl
3885 @opindex fdump-rtl-dbr
3886 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3887 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
3888 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3889 scheduling, to @file{@var{file}.36.dbr}.
3893 Dump all macro definitions, at the end of preprocessing, in addition to
3897 @itemx -fdump-rtl-ce3
3899 @opindex fdump-rtl-ce3
3900 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
3903 @itemx -fdump-rtl-cfg
3904 @itemx -fdump-rtl-life
3906 @opindex fdump-rtl-cfg
3907 @opindex fdump-rtl-life
3908 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3909 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
3910 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3911 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
3914 @itemx -fdump-rtl-greg
3916 @opindex fdump-rtl-greg
3917 Dump after global register allocation, to @file{@var{file}.139r.greg}.
3920 @itemx -fdump-rtl-gcse
3921 @itemx -fdump-rtl-bypass
3923 @opindex fdump-rtl-gcse
3924 @opindex fdump-rtl-bypass
3925 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3926 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3927 enable dumping after jump bypassing and control flow optimizations, to
3928 @file{@var{file}.115r.bypass}.
3931 @itemx -fdump-rtl-eh
3933 @opindex fdump-rtl-eh
3934 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3937 @itemx -fdump-rtl-sibling
3939 @opindex fdump-rtl-sibling
3940 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
3943 @itemx -fdump-rtl-jump
3945 @opindex fdump-rtl-jump
3946 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
3949 @itemx -fdump-rtl-stack
3951 @opindex fdump-rtl-stack
3952 Dump after conversion from registers to stack, to @file{@var{file}.152r.stack}.
3955 @itemx -fdump-rtl-lreg
3957 @opindex fdump-rtl-lreg
3958 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
3961 @itemx -fdump-rtl-loop2
3963 @opindex fdump-rtl-loop2
3964 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
3965 loop optimization pass, to @file{@var{file}.119r.loop2},
3966 @file{@var{file}.120r.loop2_init},
3967 @file{@var{file}.121r.loop2_invariant}, and
3968 @file{@var{file}.125r.loop2_done}.
3971 @itemx -fdump-rtl-sms
3973 @opindex fdump-rtl-sms
3974 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
3977 @itemx -fdump-rtl-mach
3979 @opindex fdump-rtl-mach
3980 Dump after performing the machine dependent reorganization pass, to
3981 @file{@var{file}.155r.mach}.
3984 @itemx -fdump-rtl-rnreg
3986 @opindex fdump-rtl-rnreg
3987 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
3990 @itemx -fdump-rtl-regmove
3992 @opindex fdump-rtl-regmove
3993 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
3996 @itemx -fdump-rtl-postreload
3998 @opindex fdump-rtl-postreload
3999 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
4002 @itemx -fdump-rtl-expand
4004 @opindex fdump-rtl-expand
4005 Dump after RTL generation, to @file{@var{file}.104r.expand}.
4008 @itemx -fdump-rtl-sched2
4010 @opindex fdump-rtl-sched2
4011 Dump after the second scheduling pass, to @file{@var{file}.150r.sched2}.
4014 @itemx -fdump-rtl-cse
4016 @opindex fdump-rtl-cse
4017 Dump after CSE (including the jump optimization that sometimes follows
4018 CSE), to @file{@var{file}.113r.cse}.
4021 @itemx -fdump-rtl-sched
4023 @opindex fdump-rtl-sched
4024 Dump after the first scheduling pass, to @file{@var{file}.21.sched}.
4027 @itemx -fdump-rtl-cse2
4029 @opindex fdump-rtl-cse2
4030 Dump after the second CSE pass (including the jump optimization that
4031 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
4034 @itemx -fdump-rtl-tracer
4036 @opindex fdump-rtl-tracer
4037 Dump after running tracer, to @file{@var{file}.118r.tracer}.
4040 @itemx -fdump-rtl-vpt
4041 @itemx -fdump-rtl-vartrack
4043 @opindex fdump-rtl-vpt
4044 @opindex fdump-rtl-vartrack
4045 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
4046 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
4047 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
4048 to @file{@var{file}.154r.vartrack}.
4051 @itemx -fdump-rtl-flow2
4053 @opindex fdump-rtl-flow2
4054 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
4057 @itemx -fdump-rtl-peephole2
4059 @opindex fdump-rtl-peephole2
4060 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
4063 @itemx -fdump-rtl-web
4065 @opindex fdump-rtl-web
4066 Dump after live range splitting, to @file{@var{file}.126r.web}.
4069 @itemx -fdump-rtl-all
4071 @opindex fdump-rtl-all
4072 Produce all the dumps listed above.
4076 Produce a core dump whenever an error occurs.
4080 Print statistics on memory usage, at the end of the run, to
4085 Annotate the assembler output with a comment indicating which
4086 pattern and alternative was used. The length of each instruction is
4091 Dump the RTL in the assembler output as a comment before each instruction.
4092 Also turns on @option{-dp} annotation.
4096 For each of the other indicated dump files (either with @option{-d} or
4097 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4098 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4102 Just generate RTL for a function instead of compiling it. Usually used
4103 with @samp{r} (@option{-fdump-rtl-expand}).
4107 Dump debugging information during parsing, to standard error.
4111 @opindex fdump-noaddr
4112 When doing debugging dumps (see @option{-d} option above), suppress
4113 address output. This makes it more feasible to use diff on debugging
4114 dumps for compiler invocations with different compiler binaries and/or
4115 different text / bss / data / heap / stack / dso start locations.
4117 @item -fdump-unnumbered
4118 @opindex fdump-unnumbered
4119 When doing debugging dumps (see @option{-d} option above), suppress instruction
4120 numbers, line number note and address output. This makes it more feasible to
4121 use diff on debugging dumps for compiler invocations with different
4122 options, in particular with and without @option{-g}.
4124 @item -fdump-translation-unit @r{(C++ only)}
4125 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4126 @opindex fdump-translation-unit
4127 Dump a representation of the tree structure for the entire translation
4128 unit to a file. The file name is made by appending @file{.tu} to the
4129 source file name. If the @samp{-@var{options}} form is used, @var{options}
4130 controls the details of the dump as described for the
4131 @option{-fdump-tree} options.
4133 @item -fdump-class-hierarchy @r{(C++ only)}
4134 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4135 @opindex fdump-class-hierarchy
4136 Dump a representation of each class's hierarchy and virtual function
4137 table layout to a file. The file name is made by appending @file{.class}
4138 to the source file name. If the @samp{-@var{options}} form is used,
4139 @var{options} controls the details of the dump as described for the
4140 @option{-fdump-tree} options.
4142 @item -fdump-ipa-@var{switch}
4144 Control the dumping at various stages of inter-procedural analysis
4145 language tree to a file. The file name is generated by appending a switch
4146 specific suffix to the source file name. The following dumps are possible:
4150 Enables all inter-procedural analysis dumps; currently the only produced
4151 dump is the @samp{cgraph} dump.
4154 Dumps information about call-graph optimization, unused function removal,
4155 and inlining decisions.
4158 @item -fdump-tree-@var{switch}
4159 @itemx -fdump-tree-@var{switch}-@var{options}
4161 Control the dumping at various stages of processing the intermediate
4162 language tree to a file. The file name is generated by appending a switch
4163 specific suffix to the source file name. If the @samp{-@var{options}}
4164 form is used, @var{options} is a list of @samp{-} separated options that
4165 control the details of the dump. Not all options are applicable to all
4166 dumps, those which are not meaningful will be ignored. The following
4167 options are available
4171 Print the address of each node. Usually this is not meaningful as it
4172 changes according to the environment and source file. Its primary use
4173 is for tying up a dump file with a debug environment.
4175 Inhibit dumping of members of a scope or body of a function merely
4176 because that scope has been reached. Only dump such items when they
4177 are directly reachable by some other path. When dumping pretty-printed
4178 trees, this option inhibits dumping the bodies of control structures.
4180 Print a raw representation of the tree. By default, trees are
4181 pretty-printed into a C-like representation.
4183 Enable more detailed dumps (not honored by every dump option).
4185 Enable dumping various statistics about the pass (not honored by every dump
4188 Enable showing basic block boundaries (disabled in raw dumps).
4190 Enable showing virtual operands for every statement.
4192 Enable showing line numbers for statements.
4194 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4196 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4199 The following tree dumps are possible:
4203 Dump before any tree based optimization, to @file{@var{file}.original}.
4206 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4209 Dump after function inlining, to @file{@var{file}.inlined}.
4212 @opindex fdump-tree-gimple
4213 Dump each function before and after the gimplification pass to a file. The
4214 file name is made by appending @file{.gimple} to the source file name.
4217 @opindex fdump-tree-cfg
4218 Dump the control flow graph of each function to a file. The file name is
4219 made by appending @file{.cfg} to the source file name.
4222 @opindex fdump-tree-vcg
4223 Dump the control flow graph of each function to a file in VCG format. The
4224 file name is made by appending @file{.vcg} to the source file name. Note
4225 that if the file contains more than one function, the generated file cannot
4226 be used directly by VCG@. You will need to cut and paste each function's
4227 graph into its own separate file first.
4230 @opindex fdump-tree-ch
4231 Dump each function after copying loop headers. The file name is made by
4232 appending @file{.ch} to the source file name.
4235 @opindex fdump-tree-ssa
4236 Dump SSA related information to a file. The file name is made by appending
4237 @file{.ssa} to the source file name.
4240 @opindex fdump-tree-salias
4241 Dump structure aliasing variable information to a file. This file name
4242 is made by appending @file{.salias} to the source file name.
4245 @opindex fdump-tree-alias
4246 Dump aliasing information for each function. The file name is made by
4247 appending @file{.alias} to the source file name.
4250 @opindex fdump-tree-ccp
4251 Dump each function after CCP@. The file name is made by appending
4252 @file{.ccp} to the source file name.
4255 @opindex fdump-tree-storeccp
4256 Dump each function after STORE-CCP. The file name is made by appending
4257 @file{.storeccp} to the source file name.
4260 @opindex fdump-tree-pre
4261 Dump trees after partial redundancy elimination. The file name is made
4262 by appending @file{.pre} to the source file name.
4265 @opindex fdump-tree-fre
4266 Dump trees after full redundancy elimination. The file name is made
4267 by appending @file{.fre} to the source file name.
4270 @opindex fdump-tree-copyprop
4271 Dump trees after copy propagation. The file name is made
4272 by appending @file{.copyprop} to the source file name.
4274 @item store_copyprop
4275 @opindex fdump-tree-store_copyprop
4276 Dump trees after store copy-propagation. The file name is made
4277 by appending @file{.store_copyprop} to the source file name.
4280 @opindex fdump-tree-dce
4281 Dump each function after dead code elimination. The file name is made by
4282 appending @file{.dce} to the source file name.
4285 @opindex fdump-tree-mudflap
4286 Dump each function after adding mudflap instrumentation. The file name is
4287 made by appending @file{.mudflap} to the source file name.
4290 @opindex fdump-tree-sra
4291 Dump each function after performing scalar replacement of aggregates. The
4292 file name is made by appending @file{.sra} to the source file name.
4295 @opindex fdump-tree-sink
4296 Dump each function after performing code sinking. The file name is made
4297 by appending @file{.sink} to the source file name.
4300 @opindex fdump-tree-dom
4301 Dump each function after applying dominator tree optimizations. The file
4302 name is made by appending @file{.dom} to the source file name.
4305 @opindex fdump-tree-dse
4306 Dump each function after applying dead store elimination. The file
4307 name is made by appending @file{.dse} to the source file name.
4310 @opindex fdump-tree-phiopt
4311 Dump each function after optimizing PHI nodes into straightline code. The file
4312 name is made by appending @file{.phiopt} to the source file name.
4315 @opindex fdump-tree-forwprop
4316 Dump each function after forward propagating single use variables. The file
4317 name is made by appending @file{.forwprop} to the source file name.
4320 @opindex fdump-tree-copyrename
4321 Dump each function after applying the copy rename optimization. The file
4322 name is made by appending @file{.copyrename} to the source file name.
4325 @opindex fdump-tree-nrv
4326 Dump each function after applying the named return value optimization on
4327 generic trees. The file name is made by appending @file{.nrv} to the source
4331 @opindex fdump-tree-vect
4332 Dump each function after applying vectorization of loops. The file name is
4333 made by appending @file{.vect} to the source file name.
4336 @opindex fdump-tree-vrp
4337 Dump each function after Value Range Propagation (VRP). The file name
4338 is made by appending @file{.vrp} to the source file name.
4341 @opindex fdump-tree-all
4342 Enable all the available tree dumps with the flags provided in this option.
4345 @item -ftree-vectorizer-verbose=@var{n}
4346 @opindex ftree-vectorizer-verbose
4347 This option controls the amount of debugging output the vectorizer prints.
4348 This information is written to standard error, unless
4349 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4350 in which case it is output to the usual dump listing file, @file{.vect}.
4351 For @var{n}=0 no diagnostic information is reported.
4352 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4353 and the total number of loops that got vectorized.
4354 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4355 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4356 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4357 level that @option{-fdump-tree-vect-stats} uses.
4358 Higher verbosity levels mean either more information dumped for each
4359 reported loop, or same amount of information reported for more loops:
4360 If @var{n}=3, alignment related information is added to the reports.
4361 If @var{n}=4, data-references related information (e.g. memory dependences,
4362 memory access-patterns) is added to the reports.
4363 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4364 that did not pass the first analysis phase (i.e. may not be countable, or
4365 may have complicated control-flow).
4366 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4367 For @var{n}=7, all the information the vectorizer generates during its
4368 analysis and transformation is reported. This is the same verbosity level
4369 that @option{-fdump-tree-vect-details} uses.
4371 @item -frandom-seed=@var{string}
4372 @opindex frandom-string
4373 This option provides a seed that GCC uses when it would otherwise use
4374 random numbers. It is used to generate certain symbol names
4375 that have to be different in every compiled file. It is also used to
4376 place unique stamps in coverage data files and the object files that
4377 produce them. You can use the @option{-frandom-seed} option to produce
4378 reproducibly identical object files.
4380 The @var{string} should be different for every file you compile.
4382 @item -fsched-verbose=@var{n}
4383 @opindex fsched-verbose
4384 On targets that use instruction scheduling, this option controls the
4385 amount of debugging output the scheduler prints. This information is
4386 written to standard error, unless @option{-dS} or @option{-dR} is
4387 specified, in which case it is output to the usual dump
4388 listing file, @file{.sched} or @file{.sched2} respectively. However
4389 for @var{n} greater than nine, the output is always printed to standard
4392 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4393 same information as @option{-dRS}. For @var{n} greater than one, it
4394 also output basic block probabilities, detailed ready list information
4395 and unit/insn info. For @var{n} greater than two, it includes RTL
4396 at abort point, control-flow and regions info. And for @var{n} over
4397 four, @option{-fsched-verbose} also includes dependence info.
4401 Store the usual ``temporary'' intermediate files permanently; place them
4402 in the current directory and name them based on the source file. Thus,
4403 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4404 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4405 preprocessed @file{foo.i} output file even though the compiler now
4406 normally uses an integrated preprocessor.
4408 When used in combination with the @option{-x} command line option,
4409 @option{-save-temps} is sensible enough to avoid over writing an
4410 input source file with the same extension as an intermediate file.
4411 The corresponding intermediate file may be obtained by renaming the
4412 source file before using @option{-save-temps}.
4416 Report the CPU time taken by each subprocess in the compilation
4417 sequence. For C source files, this is the compiler proper and assembler
4418 (plus the linker if linking is done). The output looks like this:
4425 The first number on each line is the ``user time'', that is time spent
4426 executing the program itself. The second number is ``system time'',
4427 time spent executing operating system routines on behalf of the program.
4428 Both numbers are in seconds.
4430 @item -fvar-tracking
4431 @opindex fvar-tracking
4432 Run variable tracking pass. It computes where variables are stored at each
4433 position in code. Better debugging information is then generated
4434 (if the debugging information format supports this information).
4436 It is enabled by default when compiling with optimization (@option{-Os},
4437 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4438 the debug info format supports it.
4440 @item -print-file-name=@var{library}
4441 @opindex print-file-name
4442 Print the full absolute name of the library file @var{library} that
4443 would be used when linking---and don't do anything else. With this
4444 option, GCC does not compile or link anything; it just prints the
4447 @item -print-multi-directory
4448 @opindex print-multi-directory
4449 Print the directory name corresponding to the multilib selected by any
4450 other switches present in the command line. This directory is supposed
4451 to exist in @env{GCC_EXEC_PREFIX}.
4453 @item -print-multi-lib
4454 @opindex print-multi-lib
4455 Print the mapping from multilib directory names to compiler switches
4456 that enable them. The directory name is separated from the switches by
4457 @samp{;}, and each switch starts with an @samp{@@} instead of the
4458 @samp{-}, without spaces between multiple switches. This is supposed to
4459 ease shell-processing.
4461 @item -print-prog-name=@var{program}
4462 @opindex print-prog-name
4463 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4465 @item -print-libgcc-file-name
4466 @opindex print-libgcc-file-name
4467 Same as @option{-print-file-name=libgcc.a}.
4469 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4470 but you do want to link with @file{libgcc.a}. You can do
4473 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4476 @item -print-search-dirs
4477 @opindex print-search-dirs
4478 Print the name of the configured installation directory and a list of
4479 program and library directories @command{gcc} will search---and don't do anything else.
4481 This is useful when @command{gcc} prints the error message
4482 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4483 To resolve this you either need to put @file{cpp0} and the other compiler
4484 components where @command{gcc} expects to find them, or you can set the environment
4485 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4486 Don't forget the trailing @samp{/}.
4487 @xref{Environment Variables}.
4490 @opindex dumpmachine
4491 Print the compiler's target machine (for example,
4492 @samp{i686-pc-linux-gnu})---and don't do anything else.
4495 @opindex dumpversion
4496 Print the compiler version (for example, @samp{3.0})---and don't do
4501 Print the compiler's built-in specs---and don't do anything else. (This
4502 is used when GCC itself is being built.) @xref{Spec Files}.
4504 @item -feliminate-unused-debug-types
4505 @opindex feliminate-unused-debug-types
4506 Normally, when producing DWARF2 output, GCC will emit debugging
4507 information for all types declared in a compilation
4508 unit, regardless of whether or not they are actually used
4509 in that compilation unit. Sometimes this is useful, such as
4510 if, in the debugger, you want to cast a value to a type that is
4511 not actually used in your program (but is declared). More often,
4512 however, this results in a significant amount of wasted space.
4513 With this option, GCC will avoid producing debug symbol output
4514 for types that are nowhere used in the source file being compiled.
4517 @node Optimize Options
4518 @section Options That Control Optimization
4519 @cindex optimize options
4520 @cindex options, optimization
4522 These options control various sorts of optimizations.
4524 Without any optimization option, the compiler's goal is to reduce the
4525 cost of compilation and to make debugging produce the expected
4526 results. Statements are independent: if you stop the program with a
4527 breakpoint between statements, you can then assign a new value to any
4528 variable or change the program counter to any other statement in the
4529 function and get exactly the results you would expect from the source
4532 Turning on optimization flags makes the compiler attempt to improve
4533 the performance and/or code size at the expense of compilation time
4534 and possibly the ability to debug the program.
4536 The compiler performs optimization based on the knowledge it has of
4537 the program. Optimization levels @option{-O} and above, in
4538 particular, enable @emph{unit-at-a-time} mode, which allows the
4539 compiler to consider information gained from later functions in
4540 the file when compiling a function. Compiling multiple files at
4541 once to a single output file in @emph{unit-at-a-time} mode allows
4542 the compiler to use information gained from all of the files when
4543 compiling each of them.
4545 Not all optimizations are controlled directly by a flag. Only
4546 optimizations that have a flag are listed.
4553 Optimize. Optimizing compilation takes somewhat more time, and a lot
4554 more memory for a large function.
4556 With @option{-O}, the compiler tries to reduce code size and execution
4557 time, without performing any optimizations that take a great deal of
4560 @option{-O} turns on the following optimization flags:
4561 @gccoptlist{-fdefer-pop @gol
4562 -fdelayed-branch @gol
4563 -fguess-branch-probability @gol
4564 -fcprop-registers @gol
4565 -fif-conversion @gol
4566 -fif-conversion2 @gol
4569 -ftree-dominator-opts @gol
4574 -ftree-copyrename @gol
4577 -funit-at-a-time @gol
4580 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4581 where doing so does not interfere with debugging.
4585 Optimize even more. GCC performs nearly all supported optimizations
4586 that do not involve a space-speed tradeoff. The compiler does not
4587 perform loop unrolling or function inlining when you specify @option{-O2}.
4588 As compared to @option{-O}, this option increases both compilation time
4589 and the performance of the generated code.
4591 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4592 also turns on the following optimization flags:
4593 @gccoptlist{-fthread-jumps @gol
4595 -foptimize-sibling-calls @gol
4596 -fcse-follow-jumps -fcse-skip-blocks @gol
4597 -fgcse -fgcse-lm @gol
4598 -fexpensive-optimizations @gol
4599 -frerun-cse-after-loop @gol
4602 -fschedule-insns -fschedule-insns2 @gol
4603 -fsched-interblock -fsched-spec @gol
4605 -fstrict-aliasing -fstrict-overflow @gol
4606 -fdelete-null-pointer-checks @gol
4607 -freorder-blocks -freorder-functions @gol
4608 -falign-functions -falign-jumps @gol
4609 -falign-loops -falign-labels @gol
4613 Please note the warning under @option{-fgcse} about
4614 invoking @option{-O2} on programs that use computed gotos.
4616 @option{-O2} doesn't turn on @option{-ftree-vrp} for the Ada compiler.
4617 This option must be explicitly specified on the command line to be
4618 enabled for the Ada compiler.
4622 Optimize yet more. @option{-O3} turns on all optimizations specified by
4623 @option{-O2} and also turns on the @option{-finline-functions},
4624 @option{-funswitch-loops} and @option{-fgcse-after-reload} options.
4628 Do not optimize. This is the default.
4632 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4633 do not typically increase code size. It also performs further
4634 optimizations designed to reduce code size.
4636 @option{-Os} disables the following optimization flags:
4637 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4638 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
4639 -fprefetch-loop-arrays -ftree-vect-loop-version}
4641 If you use multiple @option{-O} options, with or without level numbers,
4642 the last such option is the one that is effective.
4645 Options of the form @option{-f@var{flag}} specify machine-independent
4646 flags. Most flags have both positive and negative forms; the negative
4647 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4648 below, only one of the forms is listed---the one you typically will
4649 use. You can figure out the other form by either removing @samp{no-}
4652 The following options control specific optimizations. They are either
4653 activated by @option{-O} options or are related to ones that are. You
4654 can use the following flags in the rare cases when ``fine-tuning'' of
4655 optimizations to be performed is desired.
4658 @item -fno-default-inline
4659 @opindex fno-default-inline
4660 Do not make member functions inline by default merely because they are
4661 defined inside the class scope (C++ only). Otherwise, when you specify
4662 @w{@option{-O}}, member functions defined inside class scope are compiled
4663 inline by default; i.e., you don't need to add @samp{inline} in front of
4664 the member function name.
4666 @item -fno-defer-pop
4667 @opindex fno-defer-pop
4668 Always pop the arguments to each function call as soon as that function
4669 returns. For machines which must pop arguments after a function call,
4670 the compiler normally lets arguments accumulate on the stack for several
4671 function calls and pops them all at once.
4673 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4677 Force memory operands to be copied into registers before doing
4678 arithmetic on them. This produces better code by making all memory
4679 references potential common subexpressions. When they are not common
4680 subexpressions, instruction combination should eliminate the separate
4681 register-load. This option is now a nop and will be removed in 4.3.
4684 @opindex fforce-addr
4685 Force memory address constants to be copied into registers before
4686 doing arithmetic on them.
4688 @item -fomit-frame-pointer
4689 @opindex fomit-frame-pointer
4690 Don't keep the frame pointer in a register for functions that
4691 don't need one. This avoids the instructions to save, set up and
4692 restore frame pointers; it also makes an extra register available
4693 in many functions. @strong{It also makes debugging impossible on
4696 On some machines, such as the VAX, this flag has no effect, because
4697 the standard calling sequence automatically handles the frame pointer
4698 and nothing is saved by pretending it doesn't exist. The
4699 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4700 whether a target machine supports this flag. @xref{Registers,,Register
4701 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4703 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4705 @item -foptimize-sibling-calls
4706 @opindex foptimize-sibling-calls
4707 Optimize sibling and tail recursive calls.
4709 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4713 Don't pay attention to the @code{inline} keyword. Normally this option
4714 is used to keep the compiler from expanding any functions inline.
4715 Note that if you are not optimizing, no functions can be expanded inline.
4717 @item -finline-functions
4718 @opindex finline-functions
4719 Integrate all simple functions into their callers. The compiler
4720 heuristically decides which functions are simple enough to be worth
4721 integrating in this way.
4723 If all calls to a given function are integrated, and the function is
4724 declared @code{static}, then the function is normally not output as
4725 assembler code in its own right.
4727 Enabled at level @option{-O3}.
4729 @item -finline-functions-called-once
4730 @opindex finline-functions-called-once
4731 Consider all @code{static} functions called once for inlining into their
4732 caller even if they are not marked @code{inline}. If a call to a given
4733 function is integrated, then the function is not output as assembler code
4736 Enabled if @option{-funit-at-a-time} is enabled.
4738 @item -fearly-inlining
4739 @opindex fearly-inlining
4740 Inline functions marked by @code{always_inline} and functions whose body seems
4741 smaller than the function call overhead early before doing
4742 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
4743 makes profiling significantly cheaper and usually inlining faster on programs
4744 having large chains of nested wrapper functions.
4748 @item -finline-limit=@var{n}
4749 @opindex finline-limit
4750 By default, GCC limits the size of functions that can be inlined. This flag
4751 allows the control of this limit for functions that are explicitly marked as
4752 inline (i.e., marked with the inline keyword or defined within the class
4753 definition in c++). @var{n} is the size of functions that can be inlined in
4754 number of pseudo instructions (not counting parameter handling). The default
4755 value of @var{n} is 600.
4756 Increasing this value can result in more inlined code at
4757 the cost of compilation time and memory consumption. Decreasing usually makes
4758 the compilation faster and less code will be inlined (which presumably
4759 means slower programs). This option is particularly useful for programs that
4760 use inlining heavily such as those based on recursive templates with C++.
4762 Inlining is actually controlled by a number of parameters, which may be
4763 specified individually by using @option{--param @var{name}=@var{value}}.
4764 The @option{-finline-limit=@var{n}} option sets some of these parameters
4768 @item max-inline-insns-single
4769 is set to @var{n}/2.
4770 @item max-inline-insns-auto
4771 is set to @var{n}/2.
4772 @item min-inline-insns
4773 is set to 130 or @var{n}/4, whichever is smaller.
4774 @item max-inline-insns-rtl
4778 See below for a documentation of the individual
4779 parameters controlling inlining.
4781 @emph{Note:} pseudo instruction represents, in this particular context, an
4782 abstract measurement of function's size. In no way does it represent a count
4783 of assembly instructions and as such its exact meaning might change from one
4784 release to an another.
4786 @item -fkeep-inline-functions
4787 @opindex fkeep-inline-functions
4788 In C, emit @code{static} functions that are declared @code{inline}
4789 into the object file, even if the function has been inlined into all
4790 of its callers. This switch does not affect functions using the
4791 @code{extern inline} extension in GNU C@. In C++, emit any and all
4792 inline functions into the object file.
4794 @item -fkeep-static-consts
4795 @opindex fkeep-static-consts
4796 Emit variables declared @code{static const} when optimization isn't turned
4797 on, even if the variables aren't referenced.
4799 GCC enables this option by default. If you want to force the compiler to
4800 check if the variable was referenced, regardless of whether or not
4801 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4803 @item -fmerge-constants
4804 Attempt to merge identical constants (string constants and floating point
4805 constants) across compilation units.
4807 This option is the default for optimized compilation if the assembler and
4808 linker support it. Use @option{-fno-merge-constants} to inhibit this
4811 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4813 @item -fmerge-all-constants
4814 Attempt to merge identical constants and identical variables.
4816 This option implies @option{-fmerge-constants}. In addition to
4817 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4818 arrays or initialized constant variables with integral or floating point
4819 types. Languages like C or C++ require each non-automatic variable to
4820 have distinct location, so using this option will result in non-conforming
4823 @item -fmodulo-sched
4824 @opindex fmodulo-sched
4825 Perform swing modulo scheduling immediately before the first scheduling
4826 pass. This pass looks at innermost loops and reorders their
4827 instructions by overlapping different iterations.
4829 @item -fno-branch-count-reg
4830 @opindex fno-branch-count-reg
4831 Do not use ``decrement and branch'' instructions on a count register,
4832 but instead generate a sequence of instructions that decrement a
4833 register, compare it against zero, then branch based upon the result.
4834 This option is only meaningful on architectures that support such
4835 instructions, which include x86, PowerPC, IA-64 and S/390.
4837 The default is @option{-fbranch-count-reg}.
4839 @item -fno-function-cse
4840 @opindex fno-function-cse
4841 Do not put function addresses in registers; make each instruction that
4842 calls a constant function contain the function's address explicitly.
4844 This option results in less efficient code, but some strange hacks
4845 that alter the assembler output may be confused by the optimizations
4846 performed when this option is not used.
4848 The default is @option{-ffunction-cse}
4850 @item -fno-zero-initialized-in-bss
4851 @opindex fno-zero-initialized-in-bss
4852 If the target supports a BSS section, GCC by default puts variables that
4853 are initialized to zero into BSS@. This can save space in the resulting
4856 This option turns off this behavior because some programs explicitly
4857 rely on variables going to the data section. E.g., so that the
4858 resulting executable can find the beginning of that section and/or make
4859 assumptions based on that.
4861 The default is @option{-fzero-initialized-in-bss}.
4863 @item -fbounds-check
4864 @opindex fbounds-check
4865 For front-ends that support it, generate additional code to check that
4866 indices used to access arrays are within the declared range. This is
4867 currently only supported by the Java and Fortran front-ends, where
4868 this option defaults to true and false respectively.
4870 @item -fmudflap -fmudflapth -fmudflapir
4874 @cindex bounds checking
4876 For front-ends that support it (C and C++), instrument all risky
4877 pointer/array dereferencing operations, some standard library
4878 string/heap functions, and some other associated constructs with
4879 range/validity tests. Modules so instrumented should be immune to
4880 buffer overflows, invalid heap use, and some other classes of C/C++
4881 programming errors. The instrumentation relies on a separate runtime
4882 library (@file{libmudflap}), which will be linked into a program if
4883 @option{-fmudflap} is given at link time. Run-time behavior of the
4884 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4885 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4888 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4889 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4890 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4891 instrumentation should ignore pointer reads. This produces less
4892 instrumentation (and therefore faster execution) and still provides
4893 some protection against outright memory corrupting writes, but allows
4894 erroneously read data to propagate within a program.
4896 @item -fthread-jumps
4897 @opindex fthread-jumps
4898 Perform optimizations where we check to see if a jump branches to a
4899 location where another comparison subsumed by the first is found. If
4900 so, the first branch is redirected to either the destination of the
4901 second branch or a point immediately following it, depending on whether
4902 the condition is known to be true or false.
4904 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4906 @item -fcse-follow-jumps
4907 @opindex fcse-follow-jumps
4908 In common subexpression elimination, scan through jump instructions
4909 when the target of the jump is not reached by any other path. For
4910 example, when CSE encounters an @code{if} statement with an
4911 @code{else} clause, CSE will follow the jump when the condition
4914 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4916 @item -fcse-skip-blocks
4917 @opindex fcse-skip-blocks
4918 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4919 follow jumps which conditionally skip over blocks. When CSE
4920 encounters a simple @code{if} statement with no else clause,
4921 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4922 body of the @code{if}.
4924 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4926 @item -frerun-cse-after-loop
4927 @opindex frerun-cse-after-loop
4928 Re-run common subexpression elimination after loop optimizations has been
4931 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4935 Perform a global common subexpression elimination pass.
4936 This pass also performs global constant and copy propagation.
4938 @emph{Note:} When compiling a program using computed gotos, a GCC
4939 extension, you may get better runtime performance if you disable
4940 the global common subexpression elimination pass by adding
4941 @option{-fno-gcse} to the command line.
4943 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4947 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4948 attempt to move loads which are only killed by stores into themselves. This
4949 allows a loop containing a load/store sequence to be changed to a load outside
4950 the loop, and a copy/store within the loop.
4952 Enabled by default when gcse is enabled.
4956 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4957 global common subexpression elimination. This pass will attempt to move
4958 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4959 loops containing a load/store sequence can be changed to a load before
4960 the loop and a store after the loop.
4962 Not enabled at any optimization level.
4966 When @option{-fgcse-las} is enabled, the global common subexpression
4967 elimination pass eliminates redundant loads that come after stores to the
4968 same memory location (both partial and full redundancies).
4970 Not enabled at any optimization level.
4972 @item -fgcse-after-reload
4973 @opindex fgcse-after-reload
4974 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4975 pass is performed after reload. The purpose of this pass is to cleanup
4978 @item -funsafe-loop-optimizations
4979 @opindex funsafe-loop-optimizations
4980 If given, the loop optimizer will assume that loop indices do not
4981 overflow, and that the loops with nontrivial exit condition are not
4982 infinite. This enables a wider range of loop optimizations even if
4983 the loop optimizer itself cannot prove that these assumptions are valid.
4984 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
4985 if it finds this kind of loop.
4987 @item -fcrossjumping
4988 @opindex crossjumping
4989 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4990 resulting code may or may not perform better than without cross-jumping.
4992 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4994 @item -fif-conversion
4995 @opindex if-conversion
4996 Attempt to transform conditional jumps into branch-less equivalents. This
4997 include use of conditional moves, min, max, set flags and abs instructions, and
4998 some tricks doable by standard arithmetics. The use of conditional execution
4999 on chips where it is available is controlled by @code{if-conversion2}.
5001 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5003 @item -fif-conversion2
5004 @opindex if-conversion2
5005 Use conditional execution (where available) to transform conditional jumps into
5006 branch-less equivalents.
5008 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5010 @item -fdelete-null-pointer-checks
5011 @opindex fdelete-null-pointer-checks
5012 Use global dataflow analysis to identify and eliminate useless checks
5013 for null pointers. The compiler assumes that dereferencing a null
5014 pointer would have halted the program. If a pointer is checked after
5015 it has already been dereferenced, it cannot be null.
5017 In some environments, this assumption is not true, and programs can
5018 safely dereference null pointers. Use
5019 @option{-fno-delete-null-pointer-checks} to disable this optimization
5020 for programs which depend on that behavior.
5022 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5024 @item -fexpensive-optimizations
5025 @opindex fexpensive-optimizations
5026 Perform a number of minor optimizations that are relatively expensive.
5028 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5030 @item -foptimize-register-move
5032 @opindex foptimize-register-move
5034 Attempt to reassign register numbers in move instructions and as
5035 operands of other simple instructions in order to maximize the amount of
5036 register tying. This is especially helpful on machines with two-operand
5039 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5042 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5044 @item -fdelayed-branch
5045 @opindex fdelayed-branch
5046 If supported for the target machine, attempt to reorder instructions
5047 to exploit instruction slots available after delayed branch
5050 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5052 @item -fschedule-insns
5053 @opindex fschedule-insns
5054 If supported for the target machine, attempt to reorder instructions to
5055 eliminate execution stalls due to required data being unavailable. This
5056 helps machines that have slow floating point or memory load instructions
5057 by allowing other instructions to be issued until the result of the load
5058 or floating point instruction is required.
5060 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5062 @item -fschedule-insns2
5063 @opindex fschedule-insns2
5064 Similar to @option{-fschedule-insns}, but requests an additional pass of
5065 instruction scheduling after register allocation has been done. This is
5066 especially useful on machines with a relatively small number of
5067 registers and where memory load instructions take more than one cycle.
5069 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5071 @item -fno-sched-interblock
5072 @opindex fno-sched-interblock
5073 Don't schedule instructions across basic blocks. This is normally
5074 enabled by default when scheduling before register allocation, i.e.@:
5075 with @option{-fschedule-insns} or at @option{-O2} or higher.
5077 @item -fno-sched-spec
5078 @opindex fno-sched-spec
5079 Don't allow speculative motion of non-load instructions. This is normally
5080 enabled by default when scheduling before register allocation, i.e.@:
5081 with @option{-fschedule-insns} or at @option{-O2} or higher.
5083 @item -fsched-spec-load
5084 @opindex fsched-spec-load
5085 Allow speculative motion of some load instructions. This only makes
5086 sense when scheduling before register allocation, i.e.@: with
5087 @option{-fschedule-insns} or at @option{-O2} or higher.
5089 @item -fsched-spec-load-dangerous
5090 @opindex fsched-spec-load-dangerous
5091 Allow speculative motion of more load instructions. This only makes
5092 sense when scheduling before register allocation, i.e.@: with
5093 @option{-fschedule-insns} or at @option{-O2} or higher.
5095 @item -fsched-stalled-insns=@var{n}
5096 @opindex fsched-stalled-insns
5097 Define how many insns (if any) can be moved prematurely from the queue
5098 of stalled insns into the ready list, during the second scheduling pass.
5100 @item -fsched-stalled-insns-dep=@var{n}
5101 @opindex fsched-stalled-insns-dep
5102 Define how many insn groups (cycles) will be examined for a dependency
5103 on a stalled insn that is candidate for premature removal from the queue
5104 of stalled insns. Has an effect only during the second scheduling pass,
5105 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
5107 @item -fsched2-use-superblocks
5108 @opindex fsched2-use-superblocks
5109 When scheduling after register allocation, do use superblock scheduling
5110 algorithm. Superblock scheduling allows motion across basic block boundaries
5111 resulting on faster schedules. This option is experimental, as not all machine
5112 descriptions used by GCC model the CPU closely enough to avoid unreliable
5113 results from the algorithm.
5115 This only makes sense when scheduling after register allocation, i.e.@: with
5116 @option{-fschedule-insns2} or at @option{-O2} or higher.
5118 @item -fsched2-use-traces
5119 @opindex fsched2-use-traces
5120 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5121 allocation and additionally perform code duplication in order to increase the
5122 size of superblocks using tracer pass. See @option{-ftracer} for details on
5125 This mode should produce faster but significantly longer programs. Also
5126 without @option{-fbranch-probabilities} the traces constructed may not
5127 match the reality and hurt the performance. This only makes
5128 sense when scheduling after register allocation, i.e.@: with
5129 @option{-fschedule-insns2} or at @option{-O2} or higher.
5133 Eliminates redundant extension instructions and move the non redundant
5134 ones to optimal placement using LCM.
5136 @item -freschedule-modulo-scheduled-loops
5137 @opindex fscheduling-in-modulo-scheduled-loops
5138 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
5139 we may want to prevent the later scheduling passes from changing its schedule, we use this
5140 option to control that.
5142 @item -fcaller-saves
5143 @opindex fcaller-saves
5144 Enable values to be allocated in registers that will be clobbered by
5145 function calls, by emitting extra instructions to save and restore the
5146 registers around such calls. Such allocation is done only when it
5147 seems to result in better code than would otherwise be produced.
5149 This option is always enabled by default on certain machines, usually
5150 those which have no call-preserved registers to use instead.
5152 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5155 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
5156 enabled by default at @option{-O2} and @option{-O3}.
5159 Perform Full Redundancy Elimination (FRE) on trees. The difference
5160 between FRE and PRE is that FRE only considers expressions
5161 that are computed on all paths leading to the redundant computation.
5162 This analysis faster than PRE, though it exposes fewer redundancies.
5163 This flag is enabled by default at @option{-O} and higher.
5165 @item -ftree-copy-prop
5166 Perform copy propagation on trees. This pass eliminates unnecessary
5167 copy operations. This flag is enabled by default at @option{-O} and
5170 @item -ftree-store-copy-prop
5171 Perform copy propagation of memory loads and stores. This pass
5172 eliminates unnecessary copy operations in memory references
5173 (structures, global variables, arrays, etc). This flag is enabled by
5174 default at @option{-O2} and higher.
5177 Perform structural alias analysis on trees. This flag
5178 is enabled by default at @option{-O} and higher.
5181 Perform interprocedural pointer analysis.
5184 Perform forward store motion on trees. This flag is
5185 enabled by default at @option{-O} and higher.
5188 Perform sparse conditional constant propagation (CCP) on trees. This
5189 pass only operates on local scalar variables and is enabled by default
5190 at @option{-O} and higher.
5192 @item -ftree-store-ccp
5193 Perform sparse conditional constant propagation (CCP) on trees. This
5194 pass operates on both local scalar variables and memory stores and
5195 loads (global variables, structures, arrays, etc). This flag is
5196 enabled by default at @option{-O2} and higher.
5199 Perform dead code elimination (DCE) on trees. This flag is enabled by
5200 default at @option{-O} and higher.
5202 @item -ftree-dominator-opts
5203 Perform a variety of simple scalar cleanups (constant/copy
5204 propagation, redundancy elimination, range propagation and expression
5205 simplification) based on a dominator tree traversal. This also
5206 performs jump threading (to reduce jumps to jumps). This flag is
5207 enabled by default at @option{-O} and higher.
5210 Perform loop header copying on trees. This is beneficial since it increases
5211 effectiveness of code motion optimizations. It also saves one jump. This flag
5212 is enabled by default at @option{-O} and higher. It is not enabled
5213 for @option{-Os}, since it usually increases code size.
5215 @item -ftree-loop-optimize
5216 Perform loop optimizations on trees. This flag is enabled by default
5217 at @option{-O} and higher.
5219 @item -ftree-loop-linear
5220 Perform linear loop transformations on tree. This flag can improve cache
5221 performance and allow further loop optimizations to take place.
5223 @item -ftree-loop-im
5224 Perform loop invariant motion on trees. This pass moves only invariants that
5225 would be hard to handle at RTL level (function calls, operations that expand to
5226 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5227 operands of conditions that are invariant out of the loop, so that we can use
5228 just trivial invariantness analysis in loop unswitching. The pass also includes
5231 @item -ftree-loop-ivcanon
5232 Create a canonical counter for number of iterations in the loop for that
5233 determining number of iterations requires complicated analysis. Later
5234 optimizations then may determine the number easily. Useful especially
5235 in connection with unrolling.
5238 Perform induction variable optimizations (strength reduction, induction
5239 variable merging and induction variable elimination) on trees.
5242 Perform scalar replacement of aggregates. This pass replaces structure
5243 references with scalars to prevent committing structures to memory too
5244 early. This flag is enabled by default at @option{-O} and higher.
5246 @item -ftree-copyrename
5247 Perform copy renaming on trees. This pass attempts to rename compiler
5248 temporaries to other variables at copy locations, usually resulting in
5249 variable names which more closely resemble the original variables. This flag
5250 is enabled by default at @option{-O} and higher.
5253 Perform temporary expression replacement during the SSA->normal phase. Single
5254 use/single def temporaries are replaced at their use location with their
5255 defining expression. This results in non-GIMPLE code, but gives the expanders
5256 much more complex trees to work on resulting in better RTL generation. This is
5257 enabled by default at @option{-O} and higher.
5260 Perform live range splitting during the SSA->normal phase. Distinct live
5261 ranges of a variable are split into unique variables, allowing for better
5262 optimization later. This is enabled by default at @option{-O} and higher.
5264 @item -ftree-vectorize
5265 Perform loop vectorization on trees.
5267 @item -ftree-vect-loop-version
5268 @opindex ftree-vect-loop-version
5269 Perform loop versioning when doing loop vectorization on trees. When a loop
5270 appears to be vectorizable except that data alignment or data dependence cannot
5271 be determined at compile time then vectorized and non-vectorized versions of
5272 the loop are generated along with runtime checks for alignment or dependence
5273 to control which version is executed. This option is enabled by default
5274 except at level @option{-Os} where it is disabled.
5277 Perform Value Range Propagation on trees. This is similar to the
5278 constant propagation pass, but instead of values, ranges of values are
5279 propagated. This allows the optimizers to remove unnecessary range
5280 checks like array bound checks and null pointer checks. This is
5281 enabled by default at @option{-O2} and higher. Null pointer check
5282 elimination is only done if @option{-fdelete-null-pointer-checks} is
5287 Perform tail duplication to enlarge superblock size. This transformation
5288 simplifies the control flow of the function allowing other optimizations to do
5291 @item -funroll-loops
5292 @opindex funroll-loops
5293 Unroll loops whose number of iterations can be determined at compile
5294 time or upon entry to the loop. @option{-funroll-loops} implies
5295 @option{-frerun-cse-after-loop}. This option makes code larger,
5296 and may or may not make it run faster.
5298 @item -funroll-all-loops
5299 @opindex funroll-all-loops
5300 Unroll all loops, even if their number of iterations is uncertain when
5301 the loop is entered. This usually makes programs run more slowly.
5302 @option{-funroll-all-loops} implies the same options as
5303 @option{-funroll-loops},
5305 @item -fsplit-ivs-in-unroller
5306 @opindex fsplit-ivs-in-unroller
5307 Enables expressing of values of induction variables in later iterations
5308 of the unrolled loop using the value in the first iteration. This breaks
5309 long dependency chains, thus improving efficiency of the scheduling passes.
5311 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5312 same effect. However in cases the loop body is more complicated than
5313 a single basic block, this is not reliable. It also does not work at all
5314 on some of the architectures due to restrictions in the CSE pass.
5316 This optimization is enabled by default.
5318 @item -fvariable-expansion-in-unroller
5319 @opindex fvariable-expansion-in-unroller
5320 With this option, the compiler will create multiple copies of some
5321 local variables when unrolling a loop which can result in superior code.
5323 @item -fprefetch-loop-arrays
5324 @opindex fprefetch-loop-arrays
5325 If supported by the target machine, generate instructions to prefetch
5326 memory to improve the performance of loops that access large arrays.
5328 This option may generate better or worse code; results are highly
5329 dependent on the structure of loops within the source code.
5331 Disabled at level @option{-Os}.
5334 @itemx -fno-peephole2
5335 @opindex fno-peephole
5336 @opindex fno-peephole2
5337 Disable any machine-specific peephole optimizations. The difference
5338 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5339 are implemented in the compiler; some targets use one, some use the
5340 other, a few use both.
5342 @option{-fpeephole} is enabled by default.
5343 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5345 @item -fno-guess-branch-probability
5346 @opindex fno-guess-branch-probability
5347 Do not guess branch probabilities using heuristics.
5349 GCC will use heuristics to guess branch probabilities if they are
5350 not provided by profiling feedback (@option{-fprofile-arcs}). These
5351 heuristics are based on the control flow graph. If some branch probabilities
5352 are specified by @samp{__builtin_expect}, then the heuristics will be
5353 used to guess branch probabilities for the rest of the control flow graph,
5354 taking the @samp{__builtin_expect} info into account. The interactions
5355 between the heuristics and @samp{__builtin_expect} can be complex, and in
5356 some cases, it may be useful to disable the heuristics so that the effects
5357 of @samp{__builtin_expect} are easier to understand.
5359 The default is @option{-fguess-branch-probability} at levels
5360 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5362 @item -freorder-blocks
5363 @opindex freorder-blocks
5364 Reorder basic blocks in the compiled function in order to reduce number of
5365 taken branches and improve code locality.
5367 Enabled at levels @option{-O2}, @option{-O3}.
5369 @item -freorder-blocks-and-partition
5370 @opindex freorder-blocks-and-partition
5371 In addition to reordering basic blocks in the compiled function, in order
5372 to reduce number of taken branches, partitions hot and cold basic blocks
5373 into separate sections of the assembly and .o files, to improve
5374 paging and cache locality performance.
5376 This optimization is automatically turned off in the presence of
5377 exception handling, for linkonce sections, for functions with a user-defined
5378 section attribute and on any architecture that does not support named
5381 @item -freorder-functions
5382 @opindex freorder-functions
5383 Reorder functions in the object file in order to
5384 improve code locality. This is implemented by using special
5385 subsections @code{.text.hot} for most frequently executed functions and
5386 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5387 the linker so object file format must support named sections and linker must
5388 place them in a reasonable way.
5390 Also profile feedback must be available in to make this option effective. See
5391 @option{-fprofile-arcs} for details.
5393 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5395 @item -fstrict-aliasing
5396 @opindex fstrict-aliasing
5397 Allows the compiler to assume the strictest aliasing rules applicable to
5398 the language being compiled. For C (and C++), this activates
5399 optimizations based on the type of expressions. In particular, an
5400 object of one type is assumed never to reside at the same address as an
5401 object of a different type, unless the types are almost the same. For
5402 example, an @code{unsigned int} can alias an @code{int}, but not a
5403 @code{void*} or a @code{double}. A character type may alias any other
5406 Pay special attention to code like this:
5419 The practice of reading from a different union member than the one most
5420 recently written to (called ``type-punning'') is common. Even with
5421 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5422 is accessed through the union type. So, the code above will work as
5423 expected. However, this code might not:
5434 Every language that wishes to perform language-specific alias analysis
5435 should define a function that computes, given an @code{tree}
5436 node, an alias set for the node. Nodes in different alias sets are not
5437 allowed to alias. For an example, see the C front-end function
5438 @code{c_get_alias_set}.
5440 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5442 @item -fstrict-overflow
5443 @opindex fstrict-overflow
5444 Allow the compiler to assume strict signed overflow rules, depending
5445 on the language being compiled. For C (and C++) this means that
5446 overflow when doing arithmetic with signed numbers is undefined, which
5447 means that the compiler may assume that it will not happen. This
5448 permits various optimizations. For example, the compiler will assume
5449 that an expression like @code{i + 10 > i} will always be true for
5450 signed @code{i}. This assumption is only valid if signed overflow is
5451 undefined, as the expression is false if @code{i + 10} overflows when
5452 using twos complement arithmetic. When this option is in effect any
5453 attempt to determine whether an operation on signed numbers will
5454 overflow must be written carefully to not actually involve overflow.
5456 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
5457 that signed overflow is fully defined: it wraps. When
5458 @option{-fwrapv} is used, there is no difference between
5459 @option{-fstrict-overflow} and @option{-fno-strict-overflow}. With
5460 @option{-fwrapv} certain types of overflow are permitted. For
5461 example, if the compiler gets an overflow when doing arithmetic on
5462 constants, the overflowed value can still be used with
5463 @option{-fwrapv}, but not otherwise.
5465 The @option{-fstrict-overflow} option is enabled at levels
5466 @option{-O2}, @option{-O3}, @option{-Os}.
5468 @item -falign-functions
5469 @itemx -falign-functions=@var{n}
5470 @opindex falign-functions
5471 Align the start of functions to the next power-of-two greater than
5472 @var{n}, skipping up to @var{n} bytes. For instance,
5473 @option{-falign-functions=32} aligns functions to the next 32-byte
5474 boundary, but @option{-falign-functions=24} would align to the next
5475 32-byte boundary only if this can be done by skipping 23 bytes or less.
5477 @option{-fno-align-functions} and @option{-falign-functions=1} are
5478 equivalent and mean that functions will not be aligned.
5480 Some assemblers only support this flag when @var{n} is a power of two;
5481 in that case, it is rounded up.
5483 If @var{n} is not specified or is zero, use a machine-dependent default.
5485 Enabled at levels @option{-O2}, @option{-O3}.
5487 @item -falign-labels
5488 @itemx -falign-labels=@var{n}
5489 @opindex falign-labels
5490 Align all branch targets to a power-of-two boundary, skipping up to
5491 @var{n} bytes like @option{-falign-functions}. This option can easily
5492 make code slower, because it must insert dummy operations for when the
5493 branch target is reached in the usual flow of the code.
5495 @option{-fno-align-labels} and @option{-falign-labels=1} are
5496 equivalent and mean that labels will not be aligned.
5498 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
5499 are greater than this value, then their values are used instead.
5501 If @var{n} is not specified or is zero, use a machine-dependent default
5502 which is very likely to be @samp{1}, meaning no alignment.
5504 Enabled at levels @option{-O2}, @option{-O3}.
5507 @itemx -falign-loops=@var{n}
5508 @opindex falign-loops
5509 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
5510 like @option{-falign-functions}. The hope is that the loop will be
5511 executed many times, which will make up for any execution of the dummy
5514 @option{-fno-align-loops} and @option{-falign-loops=1} are
5515 equivalent and mean that loops will not be aligned.
5517 If @var{n} is not specified or is zero, use a machine-dependent default.
5519 Enabled at levels @option{-O2}, @option{-O3}.
5522 @itemx -falign-jumps=@var{n}
5523 @opindex falign-jumps
5524 Align branch targets to a power-of-two boundary, for branch targets
5525 where the targets can only be reached by jumping, skipping up to @var{n}
5526 bytes like @option{-falign-functions}. In this case, no dummy operations
5529 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
5530 equivalent and mean that loops will not be aligned.
5532 If @var{n} is not specified or is zero, use a machine-dependent default.
5534 Enabled at levels @option{-O2}, @option{-O3}.
5536 @item -funit-at-a-time
5537 @opindex funit-at-a-time
5538 Parse the whole compilation unit before starting to produce code.
5539 This allows some extra optimizations to take place but consumes
5540 more memory (in general). There are some compatibility issues
5541 with @emph{unit-at-a-time} mode:
5544 enabling @emph{unit-at-a-time} mode may change the order
5545 in which functions, variables, and top-level @code{asm} statements
5546 are emitted, and will likely break code relying on some particular
5547 ordering. The majority of such top-level @code{asm} statements,
5548 though, can be replaced by @code{section} attributes. The
5549 @option{fno-toplevel-reorder} option may be used to keep the ordering
5550 used in the input file, at the cost of some optimizations.
5553 @emph{unit-at-a-time} mode removes unreferenced static variables
5554 and functions. This may result in undefined references
5555 when an @code{asm} statement refers directly to variables or functions
5556 that are otherwise unused. In that case either the variable/function
5557 shall be listed as an operand of the @code{asm} statement operand or,
5558 in the case of top-level @code{asm} statements the attribute @code{used}
5559 shall be used on the declaration.
5562 Static functions now can use non-standard passing conventions that
5563 may break @code{asm} statements calling functions directly. Again,
5564 attribute @code{used} will prevent this behavior.
5567 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
5568 but this scheme may not be supported by future releases of GCC@.
5570 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5572 @item -fno-toplevel-reorder
5573 Do not reorder top-level functions, variables, and @code{asm}
5574 statements. Output them in the same order that they appear in the
5575 input file. When this option is used, unreferenced static variables
5576 will not be removed. This option is intended to support existing code
5577 which relies on a particular ordering. For new code, it is better to
5582 Constructs webs as commonly used for register allocation purposes and assign
5583 each web individual pseudo register. This allows the register allocation pass
5584 to operate on pseudos directly, but also strengthens several other optimization
5585 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
5586 however, make debugging impossible, since variables will no longer stay in a
5589 Enabled by default with @option{-funroll-loops}.
5591 @item -fwhole-program
5592 @opindex fwhole-program
5593 Assume that the current compilation unit represents whole program being
5594 compiled. All public functions and variables with the exception of @code{main}
5595 and those merged by attribute @code{externally_visible} become static functions
5596 and in a affect gets more aggressively optimized by interprocedural optimizers.
5597 While this option is equivalent to proper use of @code{static} keyword for
5598 programs consisting of single file, in combination with option
5599 @option{--combine} this flag can be used to compile most of smaller scale C
5600 programs since the functions and variables become local for the whole combined
5601 compilation unit, not for the single source file itself.
5604 @item -fno-cprop-registers
5605 @opindex fno-cprop-registers
5606 After register allocation and post-register allocation instruction splitting,
5607 we perform a copy-propagation pass to try to reduce scheduling dependencies
5608 and occasionally eliminate the copy.
5610 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5612 @item -fprofile-generate
5613 @opindex fprofile-generate
5615 Enable options usually used for instrumenting application to produce
5616 profile useful for later recompilation with profile feedback based
5617 optimization. You must use @option{-fprofile-generate} both when
5618 compiling and when linking your program.
5620 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
5623 @opindex fprofile-use
5624 Enable profile feedback directed optimizations, and optimizations
5625 generally profitable only with profile feedback available.
5627 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
5628 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
5632 The following options control compiler behavior regarding floating
5633 point arithmetic. These options trade off between speed and
5634 correctness. All must be specifically enabled.
5638 @opindex ffloat-store
5639 Do not store floating point variables in registers, and inhibit other
5640 options that might change whether a floating point value is taken from a
5643 @cindex floating point precision
5644 This option prevents undesirable excess precision on machines such as
5645 the 68000 where the floating registers (of the 68881) keep more
5646 precision than a @code{double} is supposed to have. Similarly for the
5647 x86 architecture. For most programs, the excess precision does only
5648 good, but a few programs rely on the precise definition of IEEE floating
5649 point. Use @option{-ffloat-store} for such programs, after modifying
5650 them to store all pertinent intermediate computations into variables.
5654 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5655 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5656 @option{-fno-rounding-math}, @option{-fno-signaling-nans}
5657 and @option{fcx-limited-range}.
5659 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5661 This option should never be turned on by any @option{-O} option since
5662 it can result in incorrect output for programs which depend on
5663 an exact implementation of IEEE or ISO rules/specifications for
5666 @item -fno-math-errno
5667 @opindex fno-math-errno
5668 Do not set ERRNO after calling math functions that are executed
5669 with a single instruction, e.g., sqrt. A program that relies on
5670 IEEE exceptions for math error handling may want to use this flag
5671 for speed while maintaining IEEE arithmetic compatibility.
5673 This option should never be turned on by any @option{-O} option since
5674 it can result in incorrect output for programs which depend on
5675 an exact implementation of IEEE or ISO rules/specifications for
5678 The default is @option{-fmath-errno}.
5680 On Darwin systems, the math library never sets @code{errno}. There is therefore
5681 no reason for the compiler to consider the possibility that it might,
5682 and @option{-fno-math-errno} is the default.
5684 @item -funsafe-math-optimizations
5685 @opindex funsafe-math-optimizations
5686 Allow optimizations for floating-point arithmetic that (a) assume
5687 that arguments and results are valid and (b) may violate IEEE or
5688 ANSI standards. When used at link-time, it may include libraries
5689 or startup files that change the default FPU control word or other
5690 similar optimizations.
5692 This option should never be turned on by any @option{-O} option since
5693 it can result in incorrect output for programs which depend on
5694 an exact implementation of IEEE or ISO rules/specifications for
5697 The default is @option{-fno-unsafe-math-optimizations}.
5699 @item -ffinite-math-only
5700 @opindex ffinite-math-only
5701 Allow optimizations for floating-point arithmetic that assume
5702 that arguments and results are not NaNs or +-Infs.
5704 This option should never be turned on by any @option{-O} option since
5705 it can result in incorrect output for programs which depend on
5706 an exact implementation of IEEE or ISO rules/specifications.
5708 The default is @option{-fno-finite-math-only}.
5710 @item -fno-trapping-math
5711 @opindex fno-trapping-math
5712 Compile code assuming that floating-point operations cannot generate
5713 user-visible traps. These traps include division by zero, overflow,
5714 underflow, inexact result and invalid operation. This option implies
5715 @option{-fno-signaling-nans}. Setting this option may allow faster
5716 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5718 This option should never be turned on by any @option{-O} option since
5719 it can result in incorrect output for programs which depend on
5720 an exact implementation of IEEE or ISO rules/specifications for
5723 The default is @option{-ftrapping-math}.
5725 @item -frounding-math
5726 @opindex frounding-math
5727 Disable transformations and optimizations that assume default floating
5728 point rounding behavior. This is round-to-zero for all floating point
5729 to integer conversions, and round-to-nearest for all other arithmetic
5730 truncations. This option should be specified for programs that change
5731 the FP rounding mode dynamically, or that may be executed with a
5732 non-default rounding mode. This option disables constant folding of
5733 floating point expressions at compile-time (which may be affected by
5734 rounding mode) and arithmetic transformations that are unsafe in the
5735 presence of sign-dependent rounding modes.
5737 The default is @option{-fno-rounding-math}.
5739 This option is experimental and does not currently guarantee to
5740 disable all GCC optimizations that are affected by rounding mode.
5741 Future versions of GCC may provide finer control of this setting
5742 using C99's @code{FENV_ACCESS} pragma. This command line option
5743 will be used to specify the default state for @code{FENV_ACCESS}.
5745 @item -frtl-abstract-sequences
5746 @opindex frtl-abstract-sequences
5747 It is a size optimization method. This option is to find identical
5748 sequences of code, which can be turned into pseudo-procedures and
5749 then replace all occurrences with calls to the newly created
5750 subroutine. It is kind of an opposite of @option{-finline-functions}.
5751 This optimization runs at RTL level.
5753 @item -fsignaling-nans
5754 @opindex fsignaling-nans
5755 Compile code assuming that IEEE signaling NaNs may generate user-visible
5756 traps during floating-point operations. Setting this option disables
5757 optimizations that may change the number of exceptions visible with
5758 signaling NaNs. This option implies @option{-ftrapping-math}.
5760 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5763 The default is @option{-fno-signaling-nans}.
5765 This option is experimental and does not currently guarantee to
5766 disable all GCC optimizations that affect signaling NaN behavior.
5768 @item -fsingle-precision-constant
5769 @opindex fsingle-precision-constant
5770 Treat floating point constant as single precision constant instead of
5771 implicitly converting it to double precision constant.
5773 @item -fcx-limited-range
5774 @itemx -fno-cx-limited-range
5775 @opindex fcx-limited-range
5776 @opindex fno-cx-limited-range
5777 When enabled, this option states that a range reduction step is not
5778 needed when performing complex division. The default is
5779 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
5781 This option controls the default setting of the ISO C99
5782 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
5787 The following options control optimizations that may improve
5788 performance, but are not enabled by any @option{-O} options. This
5789 section includes experimental options that may produce broken code.
5792 @item -fbranch-probabilities
5793 @opindex fbranch-probabilities
5794 After running a program compiled with @option{-fprofile-arcs}
5795 (@pxref{Debugging Options,, Options for Debugging Your Program or
5796 @command{gcc}}), you can compile it a second time using
5797 @option{-fbranch-probabilities}, to improve optimizations based on
5798 the number of times each branch was taken. When the program
5799 compiled with @option{-fprofile-arcs} exits it saves arc execution
5800 counts to a file called @file{@var{sourcename}.gcda} for each source
5801 file The information in this data file is very dependent on the
5802 structure of the generated code, so you must use the same source code
5803 and the same optimization options for both compilations.
5805 With @option{-fbranch-probabilities}, GCC puts a
5806 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5807 These can be used to improve optimization. Currently, they are only
5808 used in one place: in @file{reorg.c}, instead of guessing which path a
5809 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5810 exactly determine which path is taken more often.
5812 @item -fprofile-values
5813 @opindex fprofile-values
5814 If combined with @option{-fprofile-arcs}, it adds code so that some
5815 data about values of expressions in the program is gathered.
5817 With @option{-fbranch-probabilities}, it reads back the data gathered
5818 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5819 notes to instructions for their later usage in optimizations.
5821 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5825 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5826 a code to gather information about values of expressions.
5828 With @option{-fbranch-probabilities}, it reads back the data gathered
5829 and actually performs the optimizations based on them.
5830 Currently the optimizations include specialization of division operation
5831 using the knowledge about the value of the denominator.
5833 @item -frename-registers
5834 @opindex frename-registers
5835 Attempt to avoid false dependencies in scheduled code by making use
5836 of registers left over after register allocation. This optimization
5837 will most benefit processors with lots of registers. Depending on the
5838 debug information format adopted by the target, however, it can
5839 make debugging impossible, since variables will no longer stay in
5840 a ``home register''.
5842 Enabled by default with @option{-funroll-loops}.
5846 Perform tail duplication to enlarge superblock size. This transformation
5847 simplifies the control flow of the function allowing other optimizations to do
5850 Enabled with @option{-fprofile-use}.
5852 @item -funroll-loops
5853 @opindex funroll-loops
5854 Unroll loops whose number of iterations can be determined at compile time or
5855 upon entry to the loop. @option{-funroll-loops} implies
5856 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
5857 It also turns on complete loop peeling (i.e.@: complete removal of loops with
5858 small constant number of iterations). This option makes code larger, and may
5859 or may not make it run faster.
5861 Enabled with @option{-fprofile-use}.
5863 @item -funroll-all-loops
5864 @opindex funroll-all-loops
5865 Unroll all loops, even if their number of iterations is uncertain when
5866 the loop is entered. This usually makes programs run more slowly.
5867 @option{-funroll-all-loops} implies the same options as
5868 @option{-funroll-loops}.
5871 @opindex fpeel-loops
5872 Peels the loops for that there is enough information that they do not
5873 roll much (from profile feedback). It also turns on complete loop peeling
5874 (i.e.@: complete removal of loops with small constant number of iterations).
5876 Enabled with @option{-fprofile-use}.
5878 @item -fmove-loop-invariants
5879 @opindex fmove-loop-invariants
5880 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
5881 at level @option{-O1}
5883 @item -funswitch-loops
5884 @opindex funswitch-loops
5885 Move branches with loop invariant conditions out of the loop, with duplicates
5886 of the loop on both branches (modified according to result of the condition).
5888 @item -ffunction-sections
5889 @itemx -fdata-sections
5890 @opindex ffunction-sections
5891 @opindex fdata-sections
5892 Place each function or data item into its own section in the output
5893 file if the target supports arbitrary sections. The name of the
5894 function or the name of the data item determines the section's name
5897 Use these options on systems where the linker can perform optimizations
5898 to improve locality of reference in the instruction space. Most systems
5899 using the ELF object format and SPARC processors running Solaris 2 have
5900 linkers with such optimizations. AIX may have these optimizations in
5903 Only use these options when there are significant benefits from doing
5904 so. When you specify these options, the assembler and linker will
5905 create larger object and executable files and will also be slower.
5906 You will not be able to use @code{gprof} on all systems if you
5907 specify this option and you may have problems with debugging if
5908 you specify both this option and @option{-g}.
5910 @item -fbranch-target-load-optimize
5911 @opindex fbranch-target-load-optimize
5912 Perform branch target register load optimization before prologue / epilogue
5914 The use of target registers can typically be exposed only during reload,
5915 thus hoisting loads out of loops and doing inter-block scheduling needs
5916 a separate optimization pass.
5918 @item -fbranch-target-load-optimize2
5919 @opindex fbranch-target-load-optimize2
5920 Perform branch target register load optimization after prologue / epilogue
5923 @item -fbtr-bb-exclusive
5924 @opindex fbtr-bb-exclusive
5925 When performing branch target register load optimization, don't reuse
5926 branch target registers in within any basic block.
5928 @item -fstack-protector
5929 Emit extra code to check for buffer overflows, such as stack smashing
5930 attacks. This is done by adding a guard variable to functions with
5931 vulnerable objects. This includes functions that call alloca, and
5932 functions with buffers larger than 8 bytes. The guards are initialized
5933 when a function is entered and then checked when the function exits.
5934 If a guard check fails, an error message is printed and the program exits.
5936 @item -fstack-protector-all
5937 Like @option{-fstack-protector} except that all functions are protected.
5939 @item -fsection-anchors
5940 @opindex fsection-anchors
5941 Try to reduce the number of symbolic address calculations by using
5942 shared ``anchor'' symbols to address nearby objects. This transformation
5943 can help to reduce the number of GOT entries and GOT accesses on some
5946 For example, the implementation of the following function @code{foo}:
5950 int foo (void) @{ return a + b + c; @}
5953 would usually calculate the addresses of all three variables, but if you
5954 compile it with @option{-fsection-anchors}, it will access the variables
5955 from a common anchor point instead. The effect is similar to the
5956 following pseudocode (which isn't valid C):
5961 register int *xr = &x;
5962 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5966 Not all targets support this option.
5968 @item --param @var{name}=@var{value}
5970 In some places, GCC uses various constants to control the amount of
5971 optimization that is done. For example, GCC will not inline functions
5972 that contain more that a certain number of instructions. You can
5973 control some of these constants on the command-line using the
5974 @option{--param} option.
5976 The names of specific parameters, and the meaning of the values, are
5977 tied to the internals of the compiler, and are subject to change
5978 without notice in future releases.
5980 In each case, the @var{value} is an integer. The allowable choices for
5981 @var{name} are given in the following table:
5984 @item salias-max-implicit-fields
5985 The maximum number of fields in a variable without direct
5986 structure accesses for which structure aliasing will consider trying
5987 to track each field. The default is 5
5989 @item salias-max-array-elements
5990 The maximum number of elements an array can have and its elements
5991 still be tracked individually by structure aliasing. The default is 4
5993 @item sra-max-structure-size
5994 The maximum structure size, in bytes, at which the scalar replacement
5995 of aggregates (SRA) optimization will perform block copies. The
5996 default value, 0, implies that GCC will select the most appropriate
5999 @item sra-field-structure-ratio
6000 The threshold ratio (as a percentage) between instantiated fields and
6001 the complete structure size. We say that if the ratio of the number
6002 of bytes in instantiated fields to the number of bytes in the complete
6003 structure exceeds this parameter, then block copies are not used. The
6006 @item max-crossjump-edges
6007 The maximum number of incoming edges to consider for crossjumping.
6008 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
6009 the number of edges incoming to each block. Increasing values mean
6010 more aggressive optimization, making the compile time increase with
6011 probably small improvement in executable size.
6013 @item min-crossjump-insns
6014 The minimum number of instructions which must be matched at the end
6015 of two blocks before crossjumping will be performed on them. This
6016 value is ignored in the case where all instructions in the block being
6017 crossjumped from are matched. The default value is 5.
6019 @item max-grow-copy-bb-insns
6020 The maximum code size expansion factor when copying basic blocks
6021 instead of jumping. The expansion is relative to a jump instruction.
6022 The default value is 8.
6024 @item max-goto-duplication-insns
6025 The maximum number of instructions to duplicate to a block that jumps
6026 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
6027 passes, GCC factors computed gotos early in the compilation process,
6028 and unfactors them as late as possible. Only computed jumps at the
6029 end of a basic blocks with no more than max-goto-duplication-insns are
6030 unfactored. The default value is 8.
6032 @item max-delay-slot-insn-search
6033 The maximum number of instructions to consider when looking for an
6034 instruction to fill a delay slot. If more than this arbitrary number of
6035 instructions is searched, the time savings from filling the delay slot
6036 will be minimal so stop searching. Increasing values mean more
6037 aggressive optimization, making the compile time increase with probably
6038 small improvement in executable run time.
6040 @item max-delay-slot-live-search
6041 When trying to fill delay slots, the maximum number of instructions to
6042 consider when searching for a block with valid live register
6043 information. Increasing this arbitrarily chosen value means more
6044 aggressive optimization, increasing the compile time. This parameter
6045 should be removed when the delay slot code is rewritten to maintain the
6048 @item max-gcse-memory
6049 The approximate maximum amount of memory that will be allocated in
6050 order to perform the global common subexpression elimination
6051 optimization. If more memory than specified is required, the
6052 optimization will not be done.
6054 @item max-gcse-passes
6055 The maximum number of passes of GCSE to run. The default is 1.
6057 @item max-pending-list-length
6058 The maximum number of pending dependencies scheduling will allow
6059 before flushing the current state and starting over. Large functions
6060 with few branches or calls can create excessively large lists which
6061 needlessly consume memory and resources.
6063 @item max-inline-insns-single
6064 Several parameters control the tree inliner used in gcc.
6065 This number sets the maximum number of instructions (counted in GCC's
6066 internal representation) in a single function that the tree inliner
6067 will consider for inlining. This only affects functions declared
6068 inline and methods implemented in a class declaration (C++).
6069 The default value is 450.
6071 @item max-inline-insns-auto
6072 When you use @option{-finline-functions} (included in @option{-O3}),
6073 a lot of functions that would otherwise not be considered for inlining
6074 by the compiler will be investigated. To those functions, a different
6075 (more restrictive) limit compared to functions declared inline can
6077 The default value is 90.
6079 @item large-function-insns
6080 The limit specifying really large functions. For functions larger than this
6081 limit after inlining inlining is constrained by
6082 @option{--param large-function-growth}. This parameter is useful primarily
6083 to avoid extreme compilation time caused by non-linear algorithms used by the
6085 This parameter is ignored when @option{-funit-at-a-time} is not used.
6086 The default value is 2700.
6088 @item large-function-growth
6089 Specifies maximal growth of large function caused by inlining in percents.
6090 This parameter is ignored when @option{-funit-at-a-time} is not used.
6091 The default value is 100 which limits large function growth to 2.0 times
6094 @item large-unit-insns
6095 The limit specifying large translation unit. Growth caused by inlining of
6096 units larger than this limit is limited by @option{--param inline-unit-growth}.
6097 For small units this might be too tight (consider unit consisting of function A
6098 that is inline and B that just calls A three time. If B is small relative to
6099 A, the growth of unit is 300\% and yet such inlining is very sane. For very
6100 large units consisting of small inlininable functions however the overall unit
6101 growth limit is needed to avoid exponential explosion of code size. Thus for
6102 smaller units, the size is increased to @option{--param large-unit-insns}
6103 before applying @option{--param inline-unit-growth}. The default is 10000
6105 @item inline-unit-growth
6106 Specifies maximal overall growth of the compilation unit caused by inlining.
6107 This parameter is ignored when @option{-funit-at-a-time} is not used.
6108 The default value is 50 which limits unit growth to 1.5 times the original
6111 @item max-inline-insns-recursive
6112 @itemx max-inline-insns-recursive-auto
6113 Specifies maximum number of instructions out-of-line copy of self recursive inline
6114 function can grow into by performing recursive inlining.
6116 For functions declared inline @option{--param max-inline-insns-recursive} is
6117 taken into account. For function not declared inline, recursive inlining
6118 happens only when @option{-finline-functions} (included in @option{-O3}) is
6119 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
6120 default value is 450.
6122 @item max-inline-recursive-depth
6123 @itemx max-inline-recursive-depth-auto
6124 Specifies maximum recursion depth used by the recursive inlining.
6126 For functions declared inline @option{--param max-inline-recursive-depth} is
6127 taken into account. For function not declared inline, recursive inlining
6128 happens only when @option{-finline-functions} (included in @option{-O3}) is
6129 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6130 default value is 450.
6132 @item min-inline-recursive-probability
6133 Recursive inlining is profitable only for function having deep recursion
6134 in average and can hurt for function having little recursion depth by
6135 increasing the prologue size or complexity of function body to other
6138 When profile feedback is available (see @option{-fprofile-generate}) the actual
6139 recursion depth can be guessed from probability that function will recurse via
6140 given call expression. This parameter limits inlining only to call expression
6141 whose probability exceeds given threshold (in percents). The default value is
6144 @item inline-call-cost
6145 Specify cost of call instruction relative to simple arithmetics operations
6146 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6147 functions and at the same time increases size of leaf function that is believed to
6148 reduce function size by being inlined. In effect it increases amount of
6149 inlining for code having large abstraction penalty (many functions that just
6150 pass the arguments to other functions) and decrease inlining for code with low
6151 abstraction penalty. The default value is 16.
6153 @item max-unrolled-insns
6154 The maximum number of instructions that a loop should have if that loop
6155 is unrolled, and if the loop is unrolled, it determines how many times
6156 the loop code is unrolled.
6158 @item max-average-unrolled-insns
6159 The maximum number of instructions biased by probabilities of their execution
6160 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6161 it determines how many times the loop code is unrolled.
6163 @item max-unroll-times
6164 The maximum number of unrollings of a single loop.
6166 @item max-peeled-insns
6167 The maximum number of instructions that a loop should have if that loop
6168 is peeled, and if the loop is peeled, it determines how many times
6169 the loop code is peeled.
6171 @item max-peel-times
6172 The maximum number of peelings of a single loop.
6174 @item max-completely-peeled-insns
6175 The maximum number of insns of a completely peeled loop.
6177 @item max-completely-peel-times
6178 The maximum number of iterations of a loop to be suitable for complete peeling.
6180 @item max-unswitch-insns
6181 The maximum number of insns of an unswitched loop.
6183 @item max-unswitch-level
6184 The maximum number of branches unswitched in a single loop.
6187 The minimum cost of an expensive expression in the loop invariant motion.
6189 @item iv-consider-all-candidates-bound
6190 Bound on number of candidates for induction variables below that
6191 all candidates are considered for each use in induction variable
6192 optimizations. Only the most relevant candidates are considered
6193 if there are more candidates, to avoid quadratic time complexity.
6195 @item iv-max-considered-uses
6196 The induction variable optimizations give up on loops that contain more
6197 induction variable uses.
6199 @item iv-always-prune-cand-set-bound
6200 If number of candidates in the set is smaller than this value,
6201 we always try to remove unnecessary ivs from the set during its
6202 optimization when a new iv is added to the set.
6204 @item scev-max-expr-size
6205 Bound on size of expressions used in the scalar evolutions analyzer.
6206 Large expressions slow the analyzer.
6208 @item vect-max-version-checks
6209 The maximum number of runtime checks that can be performed when doing
6210 loop versioning in the vectorizer. See option ftree-vect-loop-version
6211 for more information.
6213 @item max-iterations-to-track
6215 The maximum number of iterations of a loop the brute force algorithm
6216 for analysis of # of iterations of the loop tries to evaluate.
6218 @item hot-bb-count-fraction
6219 Select fraction of the maximal count of repetitions of basic block in program
6220 given basic block needs to have to be considered hot.
6222 @item hot-bb-frequency-fraction
6223 Select fraction of the maximal frequency of executions of basic block in
6224 function given basic block needs to have to be considered hot
6226 @item max-predicted-iterations
6227 The maximum number of loop iterations we predict statically. This is useful
6228 in cases where function contain single loop with known bound and other loop
6229 with unknown. We predict the known number of iterations correctly, while
6230 the unknown number of iterations average to roughly 10. This means that the
6231 loop without bounds would appear artificially cold relative to the other one.
6233 @item tracer-dynamic-coverage
6234 @itemx tracer-dynamic-coverage-feedback
6236 This value is used to limit superblock formation once the given percentage of
6237 executed instructions is covered. This limits unnecessary code size
6240 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6241 feedback is available. The real profiles (as opposed to statically estimated
6242 ones) are much less balanced allowing the threshold to be larger value.
6244 @item tracer-max-code-growth
6245 Stop tail duplication once code growth has reached given percentage. This is
6246 rather hokey argument, as most of the duplicates will be eliminated later in
6247 cross jumping, so it may be set to much higher values than is the desired code
6250 @item tracer-min-branch-ratio
6252 Stop reverse growth when the reverse probability of best edge is less than this
6253 threshold (in percent).
6255 @item tracer-min-branch-ratio
6256 @itemx tracer-min-branch-ratio-feedback
6258 Stop forward growth if the best edge do have probability lower than this
6261 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6262 compilation for profile feedback and one for compilation without. The value
6263 for compilation with profile feedback needs to be more conservative (higher) in
6264 order to make tracer effective.
6266 @item max-cse-path-length
6268 Maximum number of basic blocks on path that cse considers. The default is 10.
6271 The maximum instructions CSE process before flushing. The default is 1000.
6273 @item global-var-threshold
6275 Counts the number of function calls (@var{n}) and the number of
6276 call-clobbered variables (@var{v}). If @var{n}x@var{v} is larger than this limit, a
6277 single artificial variable will be created to represent all the
6278 call-clobbered variables at function call sites. This artificial
6279 variable will then be made to alias every call-clobbered variable.
6280 (done as @code{int * size_t} on the host machine; beware overflow).
6282 @item max-aliased-vops
6284 Maximum number of virtual operands allowed to represent aliases
6285 before triggering the alias grouping heuristic. Alias grouping
6286 reduces compile times and memory consumption needed for aliasing at
6287 the expense of precision loss in alias information.
6289 @item ggc-min-expand
6291 GCC uses a garbage collector to manage its own memory allocation. This
6292 parameter specifies the minimum percentage by which the garbage
6293 collector's heap should be allowed to expand between collections.
6294 Tuning this may improve compilation speed; it has no effect on code
6297 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6298 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6299 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6300 GCC is not able to calculate RAM on a particular platform, the lower
6301 bound of 30% is used. Setting this parameter and
6302 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6303 every opportunity. This is extremely slow, but can be useful for
6306 @item ggc-min-heapsize
6308 Minimum size of the garbage collector's heap before it begins bothering
6309 to collect garbage. The first collection occurs after the heap expands
6310 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6311 tuning this may improve compilation speed, and has no effect on code
6314 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6315 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6316 with a lower bound of 4096 (four megabytes) and an upper bound of
6317 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6318 particular platform, the lower bound is used. Setting this parameter
6319 very large effectively disables garbage collection. Setting this
6320 parameter and @option{ggc-min-expand} to zero causes a full collection
6321 to occur at every opportunity.
6323 @item max-reload-search-insns
6324 The maximum number of instruction reload should look backward for equivalent
6325 register. Increasing values mean more aggressive optimization, making the
6326 compile time increase with probably slightly better performance. The default
6329 @item max-cselib-memory-locations
6330 The maximum number of memory locations cselib should take into account.
6331 Increasing values mean more aggressive optimization, making the compile time
6332 increase with probably slightly better performance. The default value is 500.
6334 @item max-flow-memory-locations
6335 Similar as @option{max-cselib-memory-locations} but for dataflow liveness.
6336 The default value is 100.
6338 @item reorder-blocks-duplicate
6339 @itemx reorder-blocks-duplicate-feedback
6341 Used by basic block reordering pass to decide whether to use unconditional
6342 branch or duplicate the code on its destination. Code is duplicated when its
6343 estimated size is smaller than this value multiplied by the estimated size of
6344 unconditional jump in the hot spots of the program.
6346 The @option{reorder-block-duplicate-feedback} is used only when profile
6347 feedback is available and may be set to higher values than
6348 @option{reorder-block-duplicate} since information about the hot spots is more
6351 @item max-sched-ready-insns
6352 The maximum number of instructions ready to be issued the scheduler should
6353 consider at any given time during the first scheduling pass. Increasing
6354 values mean more thorough searches, making the compilation time increase
6355 with probably little benefit. The default value is 100.
6357 @item max-sched-region-blocks
6358 The maximum number of blocks in a region to be considered for
6359 interblock scheduling. The default value is 10.
6361 @item max-sched-region-insns
6362 The maximum number of insns in a region to be considered for
6363 interblock scheduling. The default value is 100.
6366 The minimum probability (in percents) of reaching a source block
6367 for interblock speculative scheduling. The default value is 40.
6369 @item max-sched-extend-regions-iters
6370 The maximum number of iterations through CFG to extend regions.
6371 0 - disable region extension,
6372 N - do at most N iterations.
6373 The default value is 0.
6375 @item max-sched-insn-conflict-delay
6376 The maximum conflict delay for an insn to be considered for speculative motion.
6377 The default value is 3.
6379 @item sched-spec-prob-cutoff
6380 The minimal probability of speculation success (in percents), so that
6381 speculative insn will be scheduled.
6382 The default value is 40.
6384 @item max-last-value-rtl
6386 The maximum size measured as number of RTLs that can be recorded in an expression
6387 in combiner for a pseudo register as last known value of that register. The default
6390 @item integer-share-limit
6391 Small integer constants can use a shared data structure, reducing the
6392 compiler's memory usage and increasing its speed. This sets the maximum
6393 value of a shared integer constant's. The default value is 256.
6395 @item min-virtual-mappings
6396 Specifies the minimum number of virtual mappings in the incremental
6397 SSA updater that should be registered to trigger the virtual mappings
6398 heuristic defined by virtual-mappings-ratio. The default value is
6401 @item virtual-mappings-ratio
6402 If the number of virtual mappings is virtual-mappings-ratio bigger
6403 than the number of virtual symbols to be updated, then the incremental
6404 SSA updater switches to a full update for those symbols. The default
6407 @item ssp-buffer-size
6408 The minimum size of buffers (i.e. arrays) that will receive stack smashing
6409 protection when @option{-fstack-protection} is used.
6411 @item max-jump-thread-duplication-stmts
6412 Maximum number of statements allowed in a block that needs to be
6413 duplicated when threading jumps.
6415 @item max-fields-for-field-sensitive
6416 Maximum number of fields in a structure we will treat in
6417 a field sensitive manner during pointer analysis.
6422 @node Preprocessor Options
6423 @section Options Controlling the Preprocessor
6424 @cindex preprocessor options
6425 @cindex options, preprocessor
6427 These options control the C preprocessor, which is run on each C source
6428 file before actual compilation.
6430 If you use the @option{-E} option, nothing is done except preprocessing.
6431 Some of these options make sense only together with @option{-E} because
6432 they cause the preprocessor output to be unsuitable for actual
6437 You can use @option{-Wp,@var{option}} to bypass the compiler driver
6438 and pass @var{option} directly through to the preprocessor. If
6439 @var{option} contains commas, it is split into multiple options at the
6440 commas. However, many options are modified, translated or interpreted
6441 by the compiler driver before being passed to the preprocessor, and
6442 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
6443 interface is undocumented and subject to change, so whenever possible
6444 you should avoid using @option{-Wp} and let the driver handle the
6447 @item -Xpreprocessor @var{option}
6448 @opindex preprocessor
6449 Pass @var{option} as an option to the preprocessor. You can use this to
6450 supply system-specific preprocessor options which GCC does not know how to
6453 If you want to pass an option that takes an argument, you must use
6454 @option{-Xpreprocessor} twice, once for the option and once for the argument.
6457 @include cppopts.texi
6459 @node Assembler Options
6460 @section Passing Options to the Assembler
6462 @c prevent bad page break with this line
6463 You can pass options to the assembler.
6466 @item -Wa,@var{option}
6468 Pass @var{option} as an option to the assembler. If @var{option}
6469 contains commas, it is split into multiple options at the commas.
6471 @item -Xassembler @var{option}
6473 Pass @var{option} as an option to the assembler. You can use this to
6474 supply system-specific assembler options which GCC does not know how to
6477 If you want to pass an option that takes an argument, you must use
6478 @option{-Xassembler} twice, once for the option and once for the argument.
6483 @section Options for Linking
6484 @cindex link options
6485 @cindex options, linking
6487 These options come into play when the compiler links object files into
6488 an executable output file. They are meaningless if the compiler is
6489 not doing a link step.
6493 @item @var{object-file-name}
6494 A file name that does not end in a special recognized suffix is
6495 considered to name an object file or library. (Object files are
6496 distinguished from libraries by the linker according to the file
6497 contents.) If linking is done, these object files are used as input
6506 If any of these options is used, then the linker is not run, and
6507 object file names should not be used as arguments. @xref{Overall
6511 @item -l@var{library}
6512 @itemx -l @var{library}
6514 Search the library named @var{library} when linking. (The second
6515 alternative with the library as a separate argument is only for
6516 POSIX compliance and is not recommended.)
6518 It makes a difference where in the command you write this option; the
6519 linker searches and processes libraries and object files in the order they
6520 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
6521 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
6522 to functions in @samp{z}, those functions may not be loaded.
6524 The linker searches a standard list of directories for the library,
6525 which is actually a file named @file{lib@var{library}.a}. The linker
6526 then uses this file as if it had been specified precisely by name.
6528 The directories searched include several standard system directories
6529 plus any that you specify with @option{-L}.
6531 Normally the files found this way are library files---archive files
6532 whose members are object files. The linker handles an archive file by
6533 scanning through it for members which define symbols that have so far
6534 been referenced but not defined. But if the file that is found is an
6535 ordinary object file, it is linked in the usual fashion. The only
6536 difference between using an @option{-l} option and specifying a file name
6537 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
6538 and searches several directories.
6542 You need this special case of the @option{-l} option in order to
6543 link an Objective-C or Objective-C++ program.
6546 @opindex nostartfiles
6547 Do not use the standard system startup files when linking.
6548 The standard system libraries are used normally, unless @option{-nostdlib}
6549 or @option{-nodefaultlibs} is used.
6551 @item -nodefaultlibs
6552 @opindex nodefaultlibs
6553 Do not use the standard system libraries when linking.
6554 Only the libraries you specify will be passed to the linker.
6555 The standard startup files are used normally, unless @option{-nostartfiles}
6556 is used. The compiler may generate calls to @code{memcmp},
6557 @code{memset}, @code{memcpy} and @code{memmove}.
6558 These entries are usually resolved by entries in
6559 libc. These entry points should be supplied through some other
6560 mechanism when this option is specified.
6564 Do not use the standard system startup files or libraries when linking.
6565 No startup files and only the libraries you specify will be passed to
6566 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
6567 @code{memcpy} and @code{memmove}.
6568 These entries are usually resolved by entries in
6569 libc. These entry points should be supplied through some other
6570 mechanism when this option is specified.
6572 @cindex @option{-lgcc}, use with @option{-nostdlib}
6573 @cindex @option{-nostdlib} and unresolved references
6574 @cindex unresolved references and @option{-nostdlib}
6575 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
6576 @cindex @option{-nodefaultlibs} and unresolved references
6577 @cindex unresolved references and @option{-nodefaultlibs}
6578 One of the standard libraries bypassed by @option{-nostdlib} and
6579 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
6580 that GCC uses to overcome shortcomings of particular machines, or special
6581 needs for some languages.
6582 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
6583 Collection (GCC) Internals},
6584 for more discussion of @file{libgcc.a}.)
6585 In most cases, you need @file{libgcc.a} even when you want to avoid
6586 other standard libraries. In other words, when you specify @option{-nostdlib}
6587 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
6588 This ensures that you have no unresolved references to internal GCC
6589 library subroutines. (For example, @samp{__main}, used to ensure C++
6590 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
6591 GNU Compiler Collection (GCC) Internals}.)
6595 Produce a position independent executable on targets which support it.
6596 For predictable results, you must also specify the same set of options
6597 that were used to generate code (@option{-fpie}, @option{-fPIE},
6598 or model suboptions) when you specify this option.
6602 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
6603 that support it. This instructs the linker to add all symbols, not
6604 only used ones, to the dynamic symbol table. This option is needed
6605 for some uses of @code{dlopen} or to allow obtaining backtraces
6606 from within a program.
6610 Remove all symbol table and relocation information from the executable.
6614 On systems that support dynamic linking, this prevents linking with the shared
6615 libraries. On other systems, this option has no effect.
6619 Produce a shared object which can then be linked with other objects to
6620 form an executable. Not all systems support this option. For predictable
6621 results, you must also specify the same set of options that were used to
6622 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
6623 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
6624 needs to build supplementary stub code for constructors to work. On
6625 multi-libbed systems, @samp{gcc -shared} must select the correct support
6626 libraries to link against. Failing to supply the correct flags may lead
6627 to subtle defects. Supplying them in cases where they are not necessary
6630 @item -shared-libgcc
6631 @itemx -static-libgcc
6632 @opindex shared-libgcc
6633 @opindex static-libgcc
6634 On systems that provide @file{libgcc} as a shared library, these options
6635 force the use of either the shared or static version respectively.
6636 If no shared version of @file{libgcc} was built when the compiler was
6637 configured, these options have no effect.
6639 There are several situations in which an application should use the
6640 shared @file{libgcc} instead of the static version. The most common
6641 of these is when the application wishes to throw and catch exceptions
6642 across different shared libraries. In that case, each of the libraries
6643 as well as the application itself should use the shared @file{libgcc}.
6645 Therefore, the G++ and GCJ drivers automatically add
6646 @option{-shared-libgcc} whenever you build a shared library or a main
6647 executable, because C++ and Java programs typically use exceptions, so
6648 this is the right thing to do.
6650 If, instead, you use the GCC driver to create shared libraries, you may
6651 find that they will not always be linked with the shared @file{libgcc}.
6652 If GCC finds, at its configuration time, that you have a non-GNU linker
6653 or a GNU linker that does not support option @option{--eh-frame-hdr},
6654 it will link the shared version of @file{libgcc} into shared libraries
6655 by default. Otherwise, it will take advantage of the linker and optimize
6656 away the linking with the shared version of @file{libgcc}, linking with
6657 the static version of libgcc by default. This allows exceptions to
6658 propagate through such shared libraries, without incurring relocation
6659 costs at library load time.
6661 However, if a library or main executable is supposed to throw or catch
6662 exceptions, you must link it using the G++ or GCJ driver, as appropriate
6663 for the languages used in the program, or using the option
6664 @option{-shared-libgcc}, such that it is linked with the shared
6669 Bind references to global symbols when building a shared object. Warn
6670 about any unresolved references (unless overridden by the link editor
6671 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
6674 @item -Xlinker @var{option}
6676 Pass @var{option} as an option to the linker. You can use this to
6677 supply system-specific linker options which GCC does not know how to
6680 If you want to pass an option that takes an argument, you must use
6681 @option{-Xlinker} twice, once for the option and once for the argument.
6682 For example, to pass @option{-assert definitions}, you must write
6683 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
6684 @option{-Xlinker "-assert definitions"}, because this passes the entire
6685 string as a single argument, which is not what the linker expects.
6687 @item -Wl,@var{option}
6689 Pass @var{option} as an option to the linker. If @var{option} contains
6690 commas, it is split into multiple options at the commas.
6692 @item -u @var{symbol}
6694 Pretend the symbol @var{symbol} is undefined, to force linking of
6695 library modules to define it. You can use @option{-u} multiple times with
6696 different symbols to force loading of additional library modules.
6699 @node Directory Options
6700 @section Options for Directory Search
6701 @cindex directory options
6702 @cindex options, directory search
6705 These options specify directories to search for header files, for
6706 libraries and for parts of the compiler:
6711 Add the directory @var{dir} to the head of the list of directories to be
6712 searched for header files. This can be used to override a system header
6713 file, substituting your own version, since these directories are
6714 searched before the system header file directories. However, you should
6715 not use this option to add directories that contain vendor-supplied
6716 system header files (use @option{-isystem} for that). If you use more than
6717 one @option{-I} option, the directories are scanned in left-to-right
6718 order; the standard system directories come after.
6720 If a standard system include directory, or a directory specified with
6721 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
6722 option will be ignored. The directory will still be searched but as a
6723 system directory at its normal position in the system include chain.
6724 This is to ensure that GCC's procedure to fix buggy system headers and
6725 the ordering for the include_next directive are not inadvertently changed.
6726 If you really need to change the search order for system directories,
6727 use the @option{-nostdinc} and/or @option{-isystem} options.
6729 @item -iquote@var{dir}
6731 Add the directory @var{dir} to the head of the list of directories to
6732 be searched for header files only for the case of @samp{#include
6733 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
6734 otherwise just like @option{-I}.
6738 Add directory @var{dir} to the list of directories to be searched
6741 @item -B@var{prefix}
6743 This option specifies where to find the executables, libraries,
6744 include files, and data files of the compiler itself.
6746 The compiler driver program runs one or more of the subprograms
6747 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
6748 @var{prefix} as a prefix for each program it tries to run, both with and
6749 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
6751 For each subprogram to be run, the compiler driver first tries the
6752 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
6753 was not specified, the driver tries two standard prefixes, which are
6754 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
6755 those results in a file name that is found, the unmodified program
6756 name is searched for using the directories specified in your
6757 @env{PATH} environment variable.
6759 The compiler will check to see if the path provided by the @option{-B}
6760 refers to a directory, and if necessary it will add a directory
6761 separator character at the end of the path.
6763 @option{-B} prefixes that effectively specify directory names also apply
6764 to libraries in the linker, because the compiler translates these
6765 options into @option{-L} options for the linker. They also apply to
6766 includes files in the preprocessor, because the compiler translates these
6767 options into @option{-isystem} options for the preprocessor. In this case,
6768 the compiler appends @samp{include} to the prefix.
6770 The run-time support file @file{libgcc.a} can also be searched for using
6771 the @option{-B} prefix, if needed. If it is not found there, the two
6772 standard prefixes above are tried, and that is all. The file is left
6773 out of the link if it is not found by those means.
6775 Another way to specify a prefix much like the @option{-B} prefix is to use
6776 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
6779 As a special kludge, if the path provided by @option{-B} is
6780 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
6781 9, then it will be replaced by @file{[dir/]include}. This is to help
6782 with boot-strapping the compiler.
6784 @item -specs=@var{file}
6786 Process @var{file} after the compiler reads in the standard @file{specs}
6787 file, in order to override the defaults that the @file{gcc} driver
6788 program uses when determining what switches to pass to @file{cc1},
6789 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
6790 @option{-specs=@var{file}} can be specified on the command line, and they
6791 are processed in order, from left to right.
6793 @item --sysroot=@var{dir}
6795 Use @var{dir} as the logical root directory for headers and libraries.
6796 For example, if the compiler would normally search for headers in
6797 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
6798 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
6800 If you use both this option and the @option{-isysroot} option, then
6801 the @option{--sysroot} option will apply to libraries, but the
6802 @option{-isysroot} option will apply to header files.
6804 The GNU linker (beginning with version 2.16) has the necessary support
6805 for this option. If your linker does not support this option, the
6806 header file aspect of @option{--sysroot} will still work, but the
6807 library aspect will not.
6811 This option has been deprecated. Please use @option{-iquote} instead for
6812 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
6813 Any directories you specify with @option{-I} options before the @option{-I-}
6814 option are searched only for the case of @samp{#include "@var{file}"};
6815 they are not searched for @samp{#include <@var{file}>}.
6817 If additional directories are specified with @option{-I} options after
6818 the @option{-I-}, these directories are searched for all @samp{#include}
6819 directives. (Ordinarily @emph{all} @option{-I} directories are used
6822 In addition, the @option{-I-} option inhibits the use of the current
6823 directory (where the current input file came from) as the first search
6824 directory for @samp{#include "@var{file}"}. There is no way to
6825 override this effect of @option{-I-}. With @option{-I.} you can specify
6826 searching the directory which was current when the compiler was
6827 invoked. That is not exactly the same as what the preprocessor does
6828 by default, but it is often satisfactory.
6830 @option{-I-} does not inhibit the use of the standard system directories
6831 for header files. Thus, @option{-I-} and @option{-nostdinc} are
6838 @section Specifying subprocesses and the switches to pass to them
6841 @command{gcc} is a driver program. It performs its job by invoking a
6842 sequence of other programs to do the work of compiling, assembling and
6843 linking. GCC interprets its command-line parameters and uses these to
6844 deduce which programs it should invoke, and which command-line options
6845 it ought to place on their command lines. This behavior is controlled
6846 by @dfn{spec strings}. In most cases there is one spec string for each
6847 program that GCC can invoke, but a few programs have multiple spec
6848 strings to control their behavior. The spec strings built into GCC can
6849 be overridden by using the @option{-specs=} command-line switch to specify
6852 @dfn{Spec files} are plaintext files that are used to construct spec
6853 strings. They consist of a sequence of directives separated by blank
6854 lines. The type of directive is determined by the first non-whitespace
6855 character on the line and it can be one of the following:
6858 @item %@var{command}
6859 Issues a @var{command} to the spec file processor. The commands that can
6863 @item %include <@var{file}>
6865 Search for @var{file} and insert its text at the current point in the
6868 @item %include_noerr <@var{file}>
6869 @cindex %include_noerr
6870 Just like @samp{%include}, but do not generate an error message if the include
6871 file cannot be found.
6873 @item %rename @var{old_name} @var{new_name}
6875 Rename the spec string @var{old_name} to @var{new_name}.
6879 @item *[@var{spec_name}]:
6880 This tells the compiler to create, override or delete the named spec
6881 string. All lines after this directive up to the next directive or
6882 blank line are considered to be the text for the spec string. If this
6883 results in an empty string then the spec will be deleted. (Or, if the
6884 spec did not exist, then nothing will happened.) Otherwise, if the spec
6885 does not currently exist a new spec will be created. If the spec does
6886 exist then its contents will be overridden by the text of this
6887 directive, unless the first character of that text is the @samp{+}
6888 character, in which case the text will be appended to the spec.
6890 @item [@var{suffix}]:
6891 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6892 and up to the next directive or blank line are considered to make up the
6893 spec string for the indicated suffix. When the compiler encounters an
6894 input file with the named suffix, it will processes the spec string in
6895 order to work out how to compile that file. For example:
6902 This says that any input file whose name ends in @samp{.ZZ} should be
6903 passed to the program @samp{z-compile}, which should be invoked with the
6904 command-line switch @option{-input} and with the result of performing the
6905 @samp{%i} substitution. (See below.)
6907 As an alternative to providing a spec string, the text that follows a
6908 suffix directive can be one of the following:
6911 @item @@@var{language}
6912 This says that the suffix is an alias for a known @var{language}. This is
6913 similar to using the @option{-x} command-line switch to GCC to specify a
6914 language explicitly. For example:
6921 Says that .ZZ files are, in fact, C++ source files.
6924 This causes an error messages saying:
6927 @var{name} compiler not installed on this system.
6931 GCC already has an extensive list of suffixes built into it.
6932 This directive will add an entry to the end of the list of suffixes, but
6933 since the list is searched from the end backwards, it is effectively
6934 possible to override earlier entries using this technique.
6938 GCC has the following spec strings built into it. Spec files can
6939 override these strings or create their own. Note that individual
6940 targets can also add their own spec strings to this list.
6943 asm Options to pass to the assembler
6944 asm_final Options to pass to the assembler post-processor
6945 cpp Options to pass to the C preprocessor
6946 cc1 Options to pass to the C compiler
6947 cc1plus Options to pass to the C++ compiler
6948 endfile Object files to include at the end of the link
6949 link Options to pass to the linker
6950 lib Libraries to include on the command line to the linker
6951 libgcc Decides which GCC support library to pass to the linker
6952 linker Sets the name of the linker
6953 predefines Defines to be passed to the C preprocessor
6954 signed_char Defines to pass to CPP to say whether @code{char} is signed
6956 startfile Object files to include at the start of the link
6959 Here is a small example of a spec file:
6965 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6968 This example renames the spec called @samp{lib} to @samp{old_lib} and
6969 then overrides the previous definition of @samp{lib} with a new one.
6970 The new definition adds in some extra command-line options before
6971 including the text of the old definition.
6973 @dfn{Spec strings} are a list of command-line options to be passed to their
6974 corresponding program. In addition, the spec strings can contain
6975 @samp{%}-prefixed sequences to substitute variable text or to
6976 conditionally insert text into the command line. Using these constructs
6977 it is possible to generate quite complex command lines.
6979 Here is a table of all defined @samp{%}-sequences for spec
6980 strings. Note that spaces are not generated automatically around the
6981 results of expanding these sequences. Therefore you can concatenate them
6982 together or combine them with constant text in a single argument.
6986 Substitute one @samp{%} into the program name or argument.
6989 Substitute the name of the input file being processed.
6992 Substitute the basename of the input file being processed.
6993 This is the substring up to (and not including) the last period
6994 and not including the directory.
6997 This is the same as @samp{%b}, but include the file suffix (text after
7001 Marks the argument containing or following the @samp{%d} as a
7002 temporary file name, so that that file will be deleted if GCC exits
7003 successfully. Unlike @samp{%g}, this contributes no text to the
7006 @item %g@var{suffix}
7007 Substitute a file name that has suffix @var{suffix} and is chosen
7008 once per compilation, and mark the argument in the same way as
7009 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
7010 name is now chosen in a way that is hard to predict even when previously
7011 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
7012 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
7013 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
7014 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
7015 was simply substituted with a file name chosen once per compilation,
7016 without regard to any appended suffix (which was therefore treated
7017 just like ordinary text), making such attacks more likely to succeed.
7019 @item %u@var{suffix}
7020 Like @samp{%g}, but generates a new temporary file name even if
7021 @samp{%u@var{suffix}} was already seen.
7023 @item %U@var{suffix}
7024 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
7025 new one if there is no such last file name. In the absence of any
7026 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
7027 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
7028 would involve the generation of two distinct file names, one
7029 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
7030 simply substituted with a file name chosen for the previous @samp{%u},
7031 without regard to any appended suffix.
7033 @item %j@var{suffix}
7034 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
7035 writable, and if save-temps is off; otherwise, substitute the name
7036 of a temporary file, just like @samp{%u}. This temporary file is not
7037 meant for communication between processes, but rather as a junk
7040 @item %|@var{suffix}
7041 @itemx %m@var{suffix}
7042 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
7043 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
7044 all. These are the two most common ways to instruct a program that it
7045 should read from standard input or write to standard output. If you
7046 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
7047 construct: see for example @file{f/lang-specs.h}.
7049 @item %.@var{SUFFIX}
7050 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
7051 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
7052 terminated by the next space or %.
7055 Marks the argument containing or following the @samp{%w} as the
7056 designated output file of this compilation. This puts the argument
7057 into the sequence of arguments that @samp{%o} will substitute later.
7060 Substitutes the names of all the output files, with spaces
7061 automatically placed around them. You should write spaces
7062 around the @samp{%o} as well or the results are undefined.
7063 @samp{%o} is for use in the specs for running the linker.
7064 Input files whose names have no recognized suffix are not compiled
7065 at all, but they are included among the output files, so they will
7069 Substitutes the suffix for object files. Note that this is
7070 handled specially when it immediately follows @samp{%g, %u, or %U},
7071 because of the need for those to form complete file names. The
7072 handling is such that @samp{%O} is treated exactly as if it had already
7073 been substituted, except that @samp{%g, %u, and %U} do not currently
7074 support additional @var{suffix} characters following @samp{%O} as they would
7075 following, for example, @samp{.o}.
7078 Substitutes the standard macro predefinitions for the
7079 current target machine. Use this when running @code{cpp}.
7082 Like @samp{%p}, but puts @samp{__} before and after the name of each
7083 predefined macro, except for macros that start with @samp{__} or with
7084 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
7088 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
7089 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
7090 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
7091 and @option{-imultilib} as necessary.
7094 Current argument is the name of a library or startup file of some sort.
7095 Search for that file in a standard list of directories and substitute
7096 the full name found.
7099 Print @var{str} as an error message. @var{str} is terminated by a newline.
7100 Use this when inconsistent options are detected.
7103 Substitute the contents of spec string @var{name} at this point.
7106 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
7108 @item %x@{@var{option}@}
7109 Accumulate an option for @samp{%X}.
7112 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
7116 Output the accumulated assembler options specified by @option{-Wa}.
7119 Output the accumulated preprocessor options specified by @option{-Wp}.
7122 Process the @code{asm} spec. This is used to compute the
7123 switches to be passed to the assembler.
7126 Process the @code{asm_final} spec. This is a spec string for
7127 passing switches to an assembler post-processor, if such a program is
7131 Process the @code{link} spec. This is the spec for computing the
7132 command line passed to the linker. Typically it will make use of the
7133 @samp{%L %G %S %D and %E} sequences.
7136 Dump out a @option{-L} option for each directory that GCC believes might
7137 contain startup files. If the target supports multilibs then the
7138 current multilib directory will be prepended to each of these paths.
7141 Process the @code{lib} spec. This is a spec string for deciding which
7142 libraries should be included on the command line to the linker.
7145 Process the @code{libgcc} spec. This is a spec string for deciding
7146 which GCC support library should be included on the command line to the linker.
7149 Process the @code{startfile} spec. This is a spec for deciding which
7150 object files should be the first ones passed to the linker. Typically
7151 this might be a file named @file{crt0.o}.
7154 Process the @code{endfile} spec. This is a spec string that specifies
7155 the last object files that will be passed to the linker.
7158 Process the @code{cpp} spec. This is used to construct the arguments
7159 to be passed to the C preprocessor.
7162 Process the @code{cc1} spec. This is used to construct the options to be
7163 passed to the actual C compiler (@samp{cc1}).
7166 Process the @code{cc1plus} spec. This is used to construct the options to be
7167 passed to the actual C++ compiler (@samp{cc1plus}).
7170 Substitute the variable part of a matched option. See below.
7171 Note that each comma in the substituted string is replaced by
7175 Remove all occurrences of @code{-S} from the command line. Note---this
7176 command is position dependent. @samp{%} commands in the spec string
7177 before this one will see @code{-S}, @samp{%} commands in the spec string
7178 after this one will not.
7180 @item %:@var{function}(@var{args})
7181 Call the named function @var{function}, passing it @var{args}.
7182 @var{args} is first processed as a nested spec string, then split
7183 into an argument vector in the usual fashion. The function returns
7184 a string which is processed as if it had appeared literally as part
7185 of the current spec.
7187 The following built-in spec functions are provided:
7190 @item @code{if-exists}
7191 The @code{if-exists} spec function takes one argument, an absolute
7192 pathname to a file. If the file exists, @code{if-exists} returns the
7193 pathname. Here is a small example of its usage:
7197 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7200 @item @code{if-exists-else}
7201 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7202 spec function, except that it takes two arguments. The first argument is
7203 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7204 returns the pathname. If it does not exist, it returns the second argument.
7205 This way, @code{if-exists-else} can be used to select one file or another,
7206 based on the existence of the first. Here is a small example of its usage:
7210 crt0%O%s %:if-exists(crti%O%s) \
7211 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
7214 @item @code{replace-outfile}
7215 The @code{replace-outfile} spec function takes two arguments. It looks for the
7216 first argument in the outfiles array and replaces it with the second argument. Here
7217 is a small example of its usage:
7220 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7226 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7227 If that switch was not specified, this substitutes nothing. Note that
7228 the leading dash is omitted when specifying this option, and it is
7229 automatically inserted if the substitution is performed. Thus the spec
7230 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7231 and would output the command line option @option{-foo}.
7233 @item %W@{@code{S}@}
7234 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7237 @item %@{@code{S}*@}
7238 Substitutes all the switches specified to GCC whose names start
7239 with @code{-S}, but which also take an argument. This is used for
7240 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7241 GCC considers @option{-o foo} as being
7242 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7243 text, including the space. Thus two arguments would be generated.
7245 @item %@{@code{S}*&@code{T}*@}
7246 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7247 (the order of @code{S} and @code{T} in the spec is not significant).
7248 There can be any number of ampersand-separated variables; for each the
7249 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
7251 @item %@{@code{S}:@code{X}@}
7252 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
7254 @item %@{!@code{S}:@code{X}@}
7255 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
7257 @item %@{@code{S}*:@code{X}@}
7258 Substitutes @code{X} if one or more switches whose names start with
7259 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
7260 once, no matter how many such switches appeared. However, if @code{%*}
7261 appears somewhere in @code{X}, then @code{X} will be substituted once
7262 for each matching switch, with the @code{%*} replaced by the part of
7263 that switch that matched the @code{*}.
7265 @item %@{.@code{S}:@code{X}@}
7266 Substitutes @code{X}, if processing a file with suffix @code{S}.
7268 @item %@{!.@code{S}:@code{X}@}
7269 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
7271 @item %@{@code{S}|@code{P}:@code{X}@}
7272 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
7273 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
7274 although they have a stronger binding than the @samp{|}. If @code{%*}
7275 appears in @code{X}, all of the alternatives must be starred, and only
7276 the first matching alternative is substituted.
7278 For example, a spec string like this:
7281 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
7284 will output the following command-line options from the following input
7285 command-line options:
7290 -d fred.c -foo -baz -boggle
7291 -d jim.d -bar -baz -boggle
7294 @item %@{S:X; T:Y; :D@}
7296 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
7297 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
7298 be as many clauses as you need. This may be combined with @code{.},
7299 @code{!}, @code{|}, and @code{*} as needed.
7304 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
7305 construct may contain other nested @samp{%} constructs or spaces, or
7306 even newlines. They are processed as usual, as described above.
7307 Trailing white space in @code{X} is ignored. White space may also
7308 appear anywhere on the left side of the colon in these constructs,
7309 except between @code{.} or @code{*} and the corresponding word.
7311 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
7312 handled specifically in these constructs. If another value of
7313 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7314 @option{-W} switch is found later in the command line, the earlier
7315 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7316 just one letter, which passes all matching options.
7318 The character @samp{|} at the beginning of the predicate text is used to
7319 indicate that a command should be piped to the following command, but
7320 only if @option{-pipe} is specified.
7322 It is built into GCC which switches take arguments and which do not.
7323 (You might think it would be useful to generalize this to allow each
7324 compiler's spec to say which switches take arguments. But this cannot
7325 be done in a consistent fashion. GCC cannot even decide which input
7326 files have been specified without knowing which switches take arguments,
7327 and it must know which input files to compile in order to tell which
7330 GCC also knows implicitly that arguments starting in @option{-l} are to be
7331 treated as compiler output files, and passed to the linker in their
7332 proper position among the other output files.
7334 @c man begin OPTIONS
7336 @node Target Options
7337 @section Specifying Target Machine and Compiler Version
7338 @cindex target options
7339 @cindex cross compiling
7340 @cindex specifying machine version
7341 @cindex specifying compiler version and target machine
7342 @cindex compiler version, specifying
7343 @cindex target machine, specifying
7345 The usual way to run GCC is to run the executable called @file{gcc}, or
7346 @file{<machine>-gcc} when cross-compiling, or
7347 @file{<machine>-gcc-<version>} to run a version other than the one that
7348 was installed last. Sometimes this is inconvenient, so GCC provides
7349 options that will switch to another cross-compiler or version.
7352 @item -b @var{machine}
7354 The argument @var{machine} specifies the target machine for compilation.
7356 The value to use for @var{machine} is the same as was specified as the
7357 machine type when configuring GCC as a cross-compiler. For
7358 example, if a cross-compiler was configured with @samp{configure
7359 arm-elf}, meaning to compile for an arm processor with elf binaries,
7360 then you would specify @option{-b arm-elf} to run that cross compiler.
7361 Because there are other options beginning with @option{-b}, the
7362 configuration must contain a hyphen.
7364 @item -V @var{version}
7366 The argument @var{version} specifies which version of GCC to run.
7367 This is useful when multiple versions are installed. For example,
7368 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
7371 The @option{-V} and @option{-b} options work by running the
7372 @file{<machine>-gcc-<version>} executable, so there's no real reason to
7373 use them if you can just run that directly.
7375 @node Submodel Options
7376 @section Hardware Models and Configurations
7377 @cindex submodel options
7378 @cindex specifying hardware config
7379 @cindex hardware models and configurations, specifying
7380 @cindex machine dependent options
7382 Earlier we discussed the standard option @option{-b} which chooses among
7383 different installed compilers for completely different target
7384 machines, such as VAX vs.@: 68000 vs.@: 80386.
7386 In addition, each of these target machine types can have its own
7387 special options, starting with @samp{-m}, to choose among various
7388 hardware models or configurations---for example, 68010 vs 68020,
7389 floating coprocessor or none. A single installed version of the
7390 compiler can compile for any model or configuration, according to the
7393 Some configurations of the compiler also support additional special
7394 options, usually for compatibility with other compilers on the same
7397 @c This list is ordered alphanumerically by subsection name.
7398 @c It should be the same order and spelling as these options are listed
7399 @c in Machine Dependent Options
7405 * Blackfin Options::
7409 * DEC Alpha Options::
7410 * DEC Alpha/VMS Options::
7412 * GNU/Linux Options::
7415 * i386 and x86-64 Options::
7428 * RS/6000 and PowerPC Options::
7429 * S/390 and zSeries Options::
7433 * System V Options::
7434 * TMS320C3x/C4x Options::
7438 * Xstormy16 Options::
7444 @subsection ARC Options
7447 These options are defined for ARC implementations:
7452 Compile code for little endian mode. This is the default.
7456 Compile code for big endian mode.
7459 @opindex mmangle-cpu
7460 Prepend the name of the cpu to all public symbol names.
7461 In multiple-processor systems, there are many ARC variants with different
7462 instruction and register set characteristics. This flag prevents code
7463 compiled for one cpu to be linked with code compiled for another.
7464 No facility exists for handling variants that are ``almost identical''.
7465 This is an all or nothing option.
7467 @item -mcpu=@var{cpu}
7469 Compile code for ARC variant @var{cpu}.
7470 Which variants are supported depend on the configuration.
7471 All variants support @option{-mcpu=base}, this is the default.
7473 @item -mtext=@var{text-section}
7474 @itemx -mdata=@var{data-section}
7475 @itemx -mrodata=@var{readonly-data-section}
7479 Put functions, data, and readonly data in @var{text-section},
7480 @var{data-section}, and @var{readonly-data-section} respectively
7481 by default. This can be overridden with the @code{section} attribute.
7482 @xref{Variable Attributes}.
7487 @subsection ARM Options
7490 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
7494 @item -mabi=@var{name}
7496 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
7497 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
7500 @opindex mapcs-frame
7501 Generate a stack frame that is compliant with the ARM Procedure Call
7502 Standard for all functions, even if this is not strictly necessary for
7503 correct execution of the code. Specifying @option{-fomit-frame-pointer}
7504 with this option will cause the stack frames not to be generated for
7505 leaf functions. The default is @option{-mno-apcs-frame}.
7509 This is a synonym for @option{-mapcs-frame}.
7512 @c not currently implemented
7513 @item -mapcs-stack-check
7514 @opindex mapcs-stack-check
7515 Generate code to check the amount of stack space available upon entry to
7516 every function (that actually uses some stack space). If there is
7517 insufficient space available then either the function
7518 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
7519 called, depending upon the amount of stack space required. The run time
7520 system is required to provide these functions. The default is
7521 @option{-mno-apcs-stack-check}, since this produces smaller code.
7523 @c not currently implemented
7525 @opindex mapcs-float
7526 Pass floating point arguments using the float point registers. This is
7527 one of the variants of the APCS@. This option is recommended if the
7528 target hardware has a floating point unit or if a lot of floating point
7529 arithmetic is going to be performed by the code. The default is
7530 @option{-mno-apcs-float}, since integer only code is slightly increased in
7531 size if @option{-mapcs-float} is used.
7533 @c not currently implemented
7534 @item -mapcs-reentrant
7535 @opindex mapcs-reentrant
7536 Generate reentrant, position independent code. The default is
7537 @option{-mno-apcs-reentrant}.
7540 @item -mthumb-interwork
7541 @opindex mthumb-interwork
7542 Generate code which supports calling between the ARM and Thumb
7543 instruction sets. Without this option the two instruction sets cannot
7544 be reliably used inside one program. The default is
7545 @option{-mno-thumb-interwork}, since slightly larger code is generated
7546 when @option{-mthumb-interwork} is specified.
7548 @item -mno-sched-prolog
7549 @opindex mno-sched-prolog
7550 Prevent the reordering of instructions in the function prolog, or the
7551 merging of those instruction with the instructions in the function's
7552 body. This means that all functions will start with a recognizable set
7553 of instructions (or in fact one of a choice from a small set of
7554 different function prologues), and this information can be used to
7555 locate the start if functions inside an executable piece of code. The
7556 default is @option{-msched-prolog}.
7559 @opindex mhard-float
7560 Generate output containing floating point instructions. This is the
7564 @opindex msoft-float
7565 Generate output containing library calls for floating point.
7566 @strong{Warning:} the requisite libraries are not available for all ARM
7567 targets. Normally the facilities of the machine's usual C compiler are
7568 used, but this cannot be done directly in cross-compilation. You must make
7569 your own arrangements to provide suitable library functions for
7572 @option{-msoft-float} changes the calling convention in the output file;
7573 therefore, it is only useful if you compile @emph{all} of a program with
7574 this option. In particular, you need to compile @file{libgcc.a}, the
7575 library that comes with GCC, with @option{-msoft-float} in order for
7578 @item -mfloat-abi=@var{name}
7580 Specifies which ABI to use for floating point values. Permissible values
7581 are: @samp{soft}, @samp{softfp} and @samp{hard}.
7583 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
7584 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
7585 of floating point instructions, but still uses the soft-float calling
7588 @item -mlittle-endian
7589 @opindex mlittle-endian
7590 Generate code for a processor running in little-endian mode. This is
7591 the default for all standard configurations.
7594 @opindex mbig-endian
7595 Generate code for a processor running in big-endian mode; the default is
7596 to compile code for a little-endian processor.
7598 @item -mwords-little-endian
7599 @opindex mwords-little-endian
7600 This option only applies when generating code for big-endian processors.
7601 Generate code for a little-endian word order but a big-endian byte
7602 order. That is, a byte order of the form @samp{32107654}. Note: this
7603 option should only be used if you require compatibility with code for
7604 big-endian ARM processors generated by versions of the compiler prior to
7607 @item -mcpu=@var{name}
7609 This specifies the name of the target ARM processor. GCC uses this name
7610 to determine what kind of instructions it can emit when generating
7611 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
7612 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
7613 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
7614 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
7615 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
7616 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
7617 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
7618 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
7619 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
7620 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
7621 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
7622 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
7623 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
7624 @samp{arm1176jz-s}, @samp{arm1176jzf-s}, @samp{xscale}, @samp{iwmmxt},
7627 @itemx -mtune=@var{name}
7629 This option is very similar to the @option{-mcpu=} option, except that
7630 instead of specifying the actual target processor type, and hence
7631 restricting which instructions can be used, it specifies that GCC should
7632 tune the performance of the code as if the target were of the type
7633 specified in this option, but still choosing the instructions that it
7634 will generate based on the cpu specified by a @option{-mcpu=} option.
7635 For some ARM implementations better performance can be obtained by using
7638 @item -march=@var{name}
7640 This specifies the name of the target ARM architecture. GCC uses this
7641 name to determine what kind of instructions it can emit when generating
7642 assembly code. This option can be used in conjunction with or instead
7643 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
7644 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
7645 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
7646 @samp{iwmmxt}, @samp{ep9312}.
7648 @item -mfpu=@var{name}
7649 @itemx -mfpe=@var{number}
7650 @itemx -mfp=@var{number}
7654 This specifies what floating point hardware (or hardware emulation) is
7655 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
7656 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
7657 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
7658 with older versions of GCC@.
7660 If @option{-msoft-float} is specified this specifies the format of
7661 floating point values.
7663 @item -mstructure-size-boundary=@var{n}
7664 @opindex mstructure-size-boundary
7665 The size of all structures and unions will be rounded up to a multiple
7666 of the number of bits set by this option. Permissible values are 8, 32
7667 and 64. The default value varies for different toolchains. For the COFF
7668 targeted toolchain the default value is 8. A value of 64 is only allowed
7669 if the underlying ABI supports it.
7671 Specifying the larger number can produce faster, more efficient code, but
7672 can also increase the size of the program. Different values are potentially
7673 incompatible. Code compiled with one value cannot necessarily expect to
7674 work with code or libraries compiled with another value, if they exchange
7675 information using structures or unions.
7677 @item -mabort-on-noreturn
7678 @opindex mabort-on-noreturn
7679 Generate a call to the function @code{abort} at the end of a
7680 @code{noreturn} function. It will be executed if the function tries to
7684 @itemx -mno-long-calls
7685 @opindex mlong-calls
7686 @opindex mno-long-calls
7687 Tells the compiler to perform function calls by first loading the
7688 address of the function into a register and then performing a subroutine
7689 call on this register. This switch is needed if the target function
7690 will lie outside of the 64 megabyte addressing range of the offset based
7691 version of subroutine call instruction.
7693 Even if this switch is enabled, not all function calls will be turned
7694 into long calls. The heuristic is that static functions, functions
7695 which have the @samp{short-call} attribute, functions that are inside
7696 the scope of a @samp{#pragma no_long_calls} directive and functions whose
7697 definitions have already been compiled within the current compilation
7698 unit, will not be turned into long calls. The exception to this rule is
7699 that weak function definitions, functions with the @samp{long-call}
7700 attribute or the @samp{section} attribute, and functions that are within
7701 the scope of a @samp{#pragma long_calls} directive, will always be
7702 turned into long calls.
7704 This feature is not enabled by default. Specifying
7705 @option{-mno-long-calls} will restore the default behavior, as will
7706 placing the function calls within the scope of a @samp{#pragma
7707 long_calls_off} directive. Note these switches have no effect on how
7708 the compiler generates code to handle function calls via function
7711 @item -mnop-fun-dllimport
7712 @opindex mnop-fun-dllimport
7713 Disable support for the @code{dllimport} attribute.
7715 @item -msingle-pic-base
7716 @opindex msingle-pic-base
7717 Treat the register used for PIC addressing as read-only, rather than
7718 loading it in the prologue for each function. The run-time system is
7719 responsible for initializing this register with an appropriate value
7720 before execution begins.
7722 @item -mpic-register=@var{reg}
7723 @opindex mpic-register
7724 Specify the register to be used for PIC addressing. The default is R10
7725 unless stack-checking is enabled, when R9 is used.
7727 @item -mcirrus-fix-invalid-insns
7728 @opindex mcirrus-fix-invalid-insns
7729 @opindex mno-cirrus-fix-invalid-insns
7730 Insert NOPs into the instruction stream to in order to work around
7731 problems with invalid Maverick instruction combinations. This option
7732 is only valid if the @option{-mcpu=ep9312} option has been used to
7733 enable generation of instructions for the Cirrus Maverick floating
7734 point co-processor. This option is not enabled by default, since the
7735 problem is only present in older Maverick implementations. The default
7736 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
7739 @item -mpoke-function-name
7740 @opindex mpoke-function-name
7741 Write the name of each function into the text section, directly
7742 preceding the function prologue. The generated code is similar to this:
7746 .ascii "arm_poke_function_name", 0
7749 .word 0xff000000 + (t1 - t0)
7750 arm_poke_function_name
7752 stmfd sp!, @{fp, ip, lr, pc@}
7756 When performing a stack backtrace, code can inspect the value of
7757 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
7758 location @code{pc - 12} and the top 8 bits are set, then we know that
7759 there is a function name embedded immediately preceding this location
7760 and has length @code{((pc[-3]) & 0xff000000)}.
7764 Generate code for the 16-bit Thumb instruction set. The default is to
7765 use the 32-bit ARM instruction set.
7768 @opindex mtpcs-frame
7769 Generate a stack frame that is compliant with the Thumb Procedure Call
7770 Standard for all non-leaf functions. (A leaf function is one that does
7771 not call any other functions.) The default is @option{-mno-tpcs-frame}.
7773 @item -mtpcs-leaf-frame
7774 @opindex mtpcs-leaf-frame
7775 Generate a stack frame that is compliant with the Thumb Procedure Call
7776 Standard for all leaf functions. (A leaf function is one that does
7777 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
7779 @item -mcallee-super-interworking
7780 @opindex mcallee-super-interworking
7781 Gives all externally visible functions in the file being compiled an ARM
7782 instruction set header which switches to Thumb mode before executing the
7783 rest of the function. This allows these functions to be called from
7784 non-interworking code.
7786 @item -mcaller-super-interworking
7787 @opindex mcaller-super-interworking
7788 Allows calls via function pointers (including virtual functions) to
7789 execute correctly regardless of whether the target code has been
7790 compiled for interworking or not. There is a small overhead in the cost
7791 of executing a function pointer if this option is enabled.
7793 @item -mtp=@var{name}
7795 Specify the access model for the thread local storage pointer. The valid
7796 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
7797 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
7798 (supported in the arm6k architecture), and @option{auto}, which uses the
7799 best available method for the selected processor. The default setting is
7805 @subsection AVR Options
7808 These options are defined for AVR implementations:
7811 @item -mmcu=@var{mcu}
7813 Specify ATMEL AVR instruction set or MCU type.
7815 Instruction set avr1 is for the minimal AVR core, not supported by the C
7816 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7817 attiny11, attiny12, attiny15, attiny28).
7819 Instruction set avr2 (default) is for the classic AVR core with up to
7820 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7821 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7822 at90c8534, at90s8535).
7824 Instruction set avr3 is for the classic AVR core with up to 128K program
7825 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
7827 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7828 memory space (MCU types: atmega8, atmega83, atmega85).
7830 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7831 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
7832 atmega64, atmega128, at43usb355, at94k).
7836 Output instruction sizes to the asm file.
7838 @item -minit-stack=@var{N}
7839 @opindex minit-stack
7840 Specify the initial stack address, which may be a symbol or numeric value,
7841 @samp{__stack} is the default.
7843 @item -mno-interrupts
7844 @opindex mno-interrupts
7845 Generated code is not compatible with hardware interrupts.
7846 Code size will be smaller.
7848 @item -mcall-prologues
7849 @opindex mcall-prologues
7850 Functions prologues/epilogues expanded as call to appropriate
7851 subroutines. Code size will be smaller.
7853 @item -mno-tablejump
7854 @opindex mno-tablejump
7855 Do not generate tablejump insns which sometimes increase code size.
7858 @opindex mtiny-stack
7859 Change only the low 8 bits of the stack pointer.
7863 Assume int to be 8 bit integer. This affects the sizes of all types: A
7864 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7865 and long long will be 4 bytes. Please note that this option does not
7866 comply to the C standards, but it will provide you with smaller code
7870 @node Blackfin Options
7871 @subsection Blackfin Options
7872 @cindex Blackfin Options
7875 @item -momit-leaf-frame-pointer
7876 @opindex momit-leaf-frame-pointer
7877 Don't keep the frame pointer in a register for leaf functions. This
7878 avoids the instructions to save, set up and restore frame pointers and
7879 makes an extra register available in leaf functions. The option
7880 @option{-fomit-frame-pointer} removes the frame pointer for all functions
7881 which might make debugging harder.
7883 @item -mspecld-anomaly
7884 @opindex mspecld-anomaly
7885 When enabled, the compiler will ensure that the generated code does not
7886 contain speculative loads after jump instructions. This option is enabled
7889 @item -mno-specld-anomaly
7890 @opindex mno-specld-anomaly
7891 Don't generate extra code to prevent speculative loads from occurring.
7893 @item -mcsync-anomaly
7894 @opindex mcsync-anomaly
7895 When enabled, the compiler will ensure that the generated code does not
7896 contain CSYNC or SSYNC instructions too soon after conditional branches.
7897 This option is enabled by default.
7899 @item -mno-csync-anomaly
7900 @opindex mno-csync-anomaly
7901 Don't generate extra code to prevent CSYNC or SSYNC instructions from
7902 occurring too soon after a conditional branch.
7906 When enabled, the compiler is free to take advantage of the knowledge that
7907 the entire program fits into the low 64k of memory.
7910 @opindex mno-low-64k
7911 Assume that the program is arbitrarily large. This is the default.
7913 @item -mid-shared-library
7914 @opindex mid-shared-library
7915 Generate code that supports shared libraries via the library ID method.
7916 This allows for execute in place and shared libraries in an environment
7917 without virtual memory management. This option implies @option{-fPIC}.
7919 @item -mno-id-shared-library
7920 @opindex mno-id-shared-library
7921 Generate code that doesn't assume ID based shared libraries are being used.
7922 This is the default.
7924 @item -mshared-library-id=n
7925 @opindex mshared-library-id
7926 Specified the identification number of the ID based shared library being
7927 compiled. Specifying a value of 0 will generate more compact code, specifying
7928 other values will force the allocation of that number to the current
7929 library but is no more space or time efficient than omitting this option.
7932 @itemx -mno-long-calls
7933 @opindex mlong-calls
7934 @opindex mno-long-calls
7935 Tells the compiler to perform function calls by first loading the
7936 address of the function into a register and then performing a subroutine
7937 call on this register. This switch is needed if the target function
7938 will lie outside of the 24 bit addressing range of the offset based
7939 version of subroutine call instruction.
7941 This feature is not enabled by default. Specifying
7942 @option{-mno-long-calls} will restore the default behavior. Note these
7943 switches have no effect on how the compiler generates code to handle
7944 function calls via function pointers.
7948 @subsection CRIS Options
7949 @cindex CRIS Options
7951 These options are defined specifically for the CRIS ports.
7954 @item -march=@var{architecture-type}
7955 @itemx -mcpu=@var{architecture-type}
7958 Generate code for the specified architecture. The choices for
7959 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7960 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
7961 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7964 @item -mtune=@var{architecture-type}
7966 Tune to @var{architecture-type} everything applicable about the generated
7967 code, except for the ABI and the set of available instructions. The
7968 choices for @var{architecture-type} are the same as for
7969 @option{-march=@var{architecture-type}}.
7971 @item -mmax-stack-frame=@var{n}
7972 @opindex mmax-stack-frame
7973 Warn when the stack frame of a function exceeds @var{n} bytes.
7975 @item -melinux-stacksize=@var{n}
7976 @opindex melinux-stacksize
7977 Only available with the @samp{cris-axis-aout} target. Arranges for
7978 indications in the program to the kernel loader that the stack of the
7979 program should be set to @var{n} bytes.
7985 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7986 @option{-march=v3} and @option{-march=v8} respectively.
7988 @item -mmul-bug-workaround
7989 @itemx -mno-mul-bug-workaround
7990 @opindex mmul-bug-workaround
7991 @opindex mno-mul-bug-workaround
7992 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7993 models where it applies. This option is active by default.
7997 Enable CRIS-specific verbose debug-related information in the assembly
7998 code. This option also has the effect to turn off the @samp{#NO_APP}
7999 formatted-code indicator to the assembler at the beginning of the
8004 Do not use condition-code results from previous instruction; always emit
8005 compare and test instructions before use of condition codes.
8007 @item -mno-side-effects
8008 @opindex mno-side-effects
8009 Do not emit instructions with side-effects in addressing modes other than
8013 @itemx -mno-stack-align
8015 @itemx -mno-data-align
8016 @itemx -mconst-align
8017 @itemx -mno-const-align
8018 @opindex mstack-align
8019 @opindex mno-stack-align
8020 @opindex mdata-align
8021 @opindex mno-data-align
8022 @opindex mconst-align
8023 @opindex mno-const-align
8024 These options (no-options) arranges (eliminate arrangements) for the
8025 stack-frame, individual data and constants to be aligned for the maximum
8026 single data access size for the chosen CPU model. The default is to
8027 arrange for 32-bit alignment. ABI details such as structure layout are
8028 not affected by these options.
8036 Similar to the stack- data- and const-align options above, these options
8037 arrange for stack-frame, writable data and constants to all be 32-bit,
8038 16-bit or 8-bit aligned. The default is 32-bit alignment.
8040 @item -mno-prologue-epilogue
8041 @itemx -mprologue-epilogue
8042 @opindex mno-prologue-epilogue
8043 @opindex mprologue-epilogue
8044 With @option{-mno-prologue-epilogue}, the normal function prologue and
8045 epilogue that sets up the stack-frame are omitted and no return
8046 instructions or return sequences are generated in the code. Use this
8047 option only together with visual inspection of the compiled code: no
8048 warnings or errors are generated when call-saved registers must be saved,
8049 or storage for local variable needs to be allocated.
8055 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
8056 instruction sequences that load addresses for functions from the PLT part
8057 of the GOT rather than (traditional on other architectures) calls to the
8058 PLT@. The default is @option{-mgotplt}.
8062 Legacy no-op option only recognized with the cris-axis-aout target.
8066 Legacy no-op option only recognized with the cris-axis-elf and
8067 cris-axis-linux-gnu targets.
8071 Only recognized with the cris-axis-aout target, where it selects a
8072 GNU/linux-like multilib, include files and instruction set for
8077 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
8081 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
8082 to link with input-output functions from a simulator library. Code,
8083 initialized data and zero-initialized data are allocated consecutively.
8087 Like @option{-sim}, but pass linker options to locate initialized data at
8088 0x40000000 and zero-initialized data at 0x80000000.
8092 @subsection CRX Options
8095 These options are defined specifically for the CRX ports.
8101 Enable the use of multiply-accumulate instructions. Disabled by default.
8105 Push instructions will be used to pass outgoing arguments when functions
8106 are called. Enabled by default.
8109 @node Darwin Options
8110 @subsection Darwin Options
8111 @cindex Darwin options
8113 These options are defined for all architectures running the Darwin operating
8116 FSF GCC on Darwin does not create ``fat'' object files; it will create
8117 an object file for the single architecture that it was built to
8118 target. Apple's GCC on Darwin does create ``fat'' files if multiple
8119 @option{-arch} options are used; it does so by running the compiler or
8120 linker multiple times and joining the results together with
8123 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
8124 @samp{i686}) is determined by the flags that specify the ISA
8125 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
8126 @option{-force_cpusubtype_ALL} option can be used to override this.
8128 The Darwin tools vary in their behavior when presented with an ISA
8129 mismatch. The assembler, @file{as}, will only permit instructions to
8130 be used that are valid for the subtype of the file it is generating,
8131 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
8132 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
8133 and print an error if asked to create a shared library with a less
8134 restrictive subtype than its input files (for instance, trying to put
8135 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
8136 for executables, @file{ld}, will quietly give the executable the most
8137 restrictive subtype of any of its input files.
8142 Add the framework directory @var{dir} to the head of the list of
8143 directories to be searched for header files. These directories are
8144 interleaved with those specified by @option{-I} options and are
8145 scanned in a left-to-right order.
8147 A framework directory is a directory with frameworks in it. A
8148 framework is a directory with a @samp{"Headers"} and/or
8149 @samp{"PrivateHeaders"} directory contained directly in it that ends
8150 in @samp{".framework"}. The name of a framework is the name of this
8151 directory excluding the @samp{".framework"}. Headers associated with
8152 the framework are found in one of those two directories, with
8153 @samp{"Headers"} being searched first. A subframework is a framework
8154 directory that is in a framework's @samp{"Frameworks"} directory.
8155 Includes of subframework headers can only appear in a header of a
8156 framework that contains the subframework, or in a sibling subframework
8157 header. Two subframeworks are siblings if they occur in the same
8158 framework. A subframework should not have the same name as a
8159 framework, a warning will be issued if this is violated. Currently a
8160 subframework cannot have subframeworks, in the future, the mechanism
8161 may be extended to support this. The standard frameworks can be found
8162 in @samp{"/System/Library/Frameworks"} and
8163 @samp{"/Library/Frameworks"}. An example include looks like
8164 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
8165 the name of the framework and header.h is found in the
8166 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
8170 Emit debugging information for symbols that are used. For STABS
8171 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
8172 This is by default ON@.
8176 Emit debugging information for all symbols and types.
8178 @item -mmacosx-version-min=@var{version}
8179 The earliest version of MacOS X that this executable will run on
8180 is @var{version}. Typical values of @var{version} include @code{10.1},
8181 @code{10.2}, and @code{10.3.9}.
8183 The default for this option is to make choices that seem to be most
8188 Enable kernel development mode. The @option{-mkernel} option sets
8189 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
8190 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
8191 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
8192 applicable. This mode also sets @option{-mno-altivec},
8193 @option{-msoft-float}, @option{-fno-builtin} and
8194 @option{-mlong-branch} for PowerPC targets.
8196 @item -mone-byte-bool
8197 @opindex mone-byte-bool
8198 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
8199 By default @samp{sizeof(bool)} is @samp{4} when compiling for
8200 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
8201 option has no effect on x86.
8203 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
8204 to generate code that is not binary compatible with code generated
8205 without that switch. Using this switch may require recompiling all
8206 other modules in a program, including system libraries. Use this
8207 switch to conform to a non-default data model.
8209 @item -mfix-and-continue
8210 @itemx -ffix-and-continue
8211 @itemx -findirect-data
8212 @opindex mfix-and-continue
8213 @opindex ffix-and-continue
8214 @opindex findirect-data
8215 Generate code suitable for fast turn around development. Needed to
8216 enable gdb to dynamically load @code{.o} files into already running
8217 programs. @option{-findirect-data} and @option{-ffix-and-continue}
8218 are provided for backwards compatibility.
8222 Loads all members of static archive libraries.
8223 See man ld(1) for more information.
8225 @item -arch_errors_fatal
8226 @opindex arch_errors_fatal
8227 Cause the errors having to do with files that have the wrong architecture
8231 @opindex bind_at_load
8232 Causes the output file to be marked such that the dynamic linker will
8233 bind all undefined references when the file is loaded or launched.
8237 Produce a Mach-o bundle format file.
8238 See man ld(1) for more information.
8240 @item -bundle_loader @var{executable}
8241 @opindex bundle_loader
8242 This option specifies the @var{executable} that will be loading the build
8243 output file being linked. See man ld(1) for more information.
8247 When passed this option, GCC will produce a dynamic library instead of
8248 an executable when linking, using the Darwin @file{libtool} command.
8250 @item -force_cpusubtype_ALL
8251 @opindex force_cpusubtype_ALL
8252 This causes GCC's output file to have the @var{ALL} subtype, instead of
8253 one controlled by the @option{-mcpu} or @option{-march} option.
8255 @item -allowable_client @var{client_name}
8257 @itemx -compatibility_version
8258 @itemx -current_version
8260 @itemx -dependency-file
8262 @itemx -dylinker_install_name
8264 @itemx -exported_symbols_list
8266 @itemx -flat_namespace
8267 @itemx -force_flat_namespace
8268 @itemx -headerpad_max_install_names
8271 @itemx -install_name
8272 @itemx -keep_private_externs
8273 @itemx -multi_module
8274 @itemx -multiply_defined
8275 @itemx -multiply_defined_unused
8277 @itemx -no_dead_strip_inits_and_terms
8278 @itemx -nofixprebinding
8281 @itemx -noseglinkedit
8282 @itemx -pagezero_size
8284 @itemx -prebind_all_twolevel_modules
8285 @itemx -private_bundle
8286 @itemx -read_only_relocs
8288 @itemx -sectobjectsymbols
8292 @itemx -sectobjectsymbols
8295 @itemx -segs_read_only_addr
8296 @itemx -segs_read_write_addr
8297 @itemx -seg_addr_table
8298 @itemx -seg_addr_table_filename
8301 @itemx -segs_read_only_addr
8302 @itemx -segs_read_write_addr
8303 @itemx -single_module
8306 @itemx -sub_umbrella
8307 @itemx -twolevel_namespace
8310 @itemx -unexported_symbols_list
8311 @itemx -weak_reference_mismatches
8314 @opindex allowable_client
8315 @opindex client_name
8316 @opindex compatibility_version
8317 @opindex current_version
8319 @opindex dependency-file
8321 @opindex dylinker_install_name
8323 @opindex exported_symbols_list
8325 @opindex flat_namespace
8326 @opindex force_flat_namespace
8327 @opindex headerpad_max_install_names
8330 @opindex install_name
8331 @opindex keep_private_externs
8332 @opindex multi_module
8333 @opindex multiply_defined
8334 @opindex multiply_defined_unused
8336 @opindex no_dead_strip_inits_and_terms
8337 @opindex nofixprebinding
8338 @opindex nomultidefs
8340 @opindex noseglinkedit
8341 @opindex pagezero_size
8343 @opindex prebind_all_twolevel_modules
8344 @opindex private_bundle
8345 @opindex read_only_relocs
8347 @opindex sectobjectsymbols
8351 @opindex sectobjectsymbols
8354 @opindex segs_read_only_addr
8355 @opindex segs_read_write_addr
8356 @opindex seg_addr_table
8357 @opindex seg_addr_table_filename
8358 @opindex seglinkedit
8360 @opindex segs_read_only_addr
8361 @opindex segs_read_write_addr
8362 @opindex single_module
8364 @opindex sub_library
8365 @opindex sub_umbrella
8366 @opindex twolevel_namespace
8369 @opindex unexported_symbols_list
8370 @opindex weak_reference_mismatches
8371 @opindex whatsloaded
8373 These options are passed to the Darwin linker. The Darwin linker man page
8374 describes them in detail.
8377 @node DEC Alpha Options
8378 @subsection DEC Alpha Options
8380 These @samp{-m} options are defined for the DEC Alpha implementations:
8383 @item -mno-soft-float
8385 @opindex mno-soft-float
8386 @opindex msoft-float
8387 Use (do not use) the hardware floating-point instructions for
8388 floating-point operations. When @option{-msoft-float} is specified,
8389 functions in @file{libgcc.a} will be used to perform floating-point
8390 operations. Unless they are replaced by routines that emulate the
8391 floating-point operations, or compiled in such a way as to call such
8392 emulations routines, these routines will issue floating-point
8393 operations. If you are compiling for an Alpha without floating-point
8394 operations, you must ensure that the library is built so as not to call
8397 Note that Alpha implementations without floating-point operations are
8398 required to have floating-point registers.
8403 @opindex mno-fp-regs
8404 Generate code that uses (does not use) the floating-point register set.
8405 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8406 register set is not used, floating point operands are passed in integer
8407 registers as if they were integers and floating-point results are passed
8408 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8409 so any function with a floating-point argument or return value called by code
8410 compiled with @option{-mno-fp-regs} must also be compiled with that
8413 A typical use of this option is building a kernel that does not use,
8414 and hence need not save and restore, any floating-point registers.
8418 The Alpha architecture implements floating-point hardware optimized for
8419 maximum performance. It is mostly compliant with the IEEE floating
8420 point standard. However, for full compliance, software assistance is
8421 required. This option generates code fully IEEE compliant code
8422 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8423 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8424 defined during compilation. The resulting code is less efficient but is
8425 able to correctly support denormalized numbers and exceptional IEEE
8426 values such as not-a-number and plus/minus infinity. Other Alpha
8427 compilers call this option @option{-ieee_with_no_inexact}.
8429 @item -mieee-with-inexact
8430 @opindex mieee-with-inexact
8431 This is like @option{-mieee} except the generated code also maintains
8432 the IEEE @var{inexact-flag}. Turning on this option causes the
8433 generated code to implement fully-compliant IEEE math. In addition to
8434 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8435 macro. On some Alpha implementations the resulting code may execute
8436 significantly slower than the code generated by default. Since there is
8437 very little code that depends on the @var{inexact-flag}, you should
8438 normally not specify this option. Other Alpha compilers call this
8439 option @option{-ieee_with_inexact}.
8441 @item -mfp-trap-mode=@var{trap-mode}
8442 @opindex mfp-trap-mode
8443 This option controls what floating-point related traps are enabled.
8444 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8445 The trap mode can be set to one of four values:
8449 This is the default (normal) setting. The only traps that are enabled
8450 are the ones that cannot be disabled in software (e.g., division by zero
8454 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8458 Like @samp{u}, but the instructions are marked to be safe for software
8459 completion (see Alpha architecture manual for details).
8462 Like @samp{su}, but inexact traps are enabled as well.
8465 @item -mfp-rounding-mode=@var{rounding-mode}
8466 @opindex mfp-rounding-mode
8467 Selects the IEEE rounding mode. Other Alpha compilers call this option
8468 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8473 Normal IEEE rounding mode. Floating point numbers are rounded towards
8474 the nearest machine number or towards the even machine number in case
8478 Round towards minus infinity.
8481 Chopped rounding mode. Floating point numbers are rounded towards zero.
8484 Dynamic rounding mode. A field in the floating point control register
8485 (@var{fpcr}, see Alpha architecture reference manual) controls the
8486 rounding mode in effect. The C library initializes this register for
8487 rounding towards plus infinity. Thus, unless your program modifies the
8488 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8491 @item -mtrap-precision=@var{trap-precision}
8492 @opindex mtrap-precision
8493 In the Alpha architecture, floating point traps are imprecise. This
8494 means without software assistance it is impossible to recover from a
8495 floating trap and program execution normally needs to be terminated.
8496 GCC can generate code that can assist operating system trap handlers
8497 in determining the exact location that caused a floating point trap.
8498 Depending on the requirements of an application, different levels of
8499 precisions can be selected:
8503 Program precision. This option is the default and means a trap handler
8504 can only identify which program caused a floating point exception.
8507 Function precision. The trap handler can determine the function that
8508 caused a floating point exception.
8511 Instruction precision. The trap handler can determine the exact
8512 instruction that caused a floating point exception.
8515 Other Alpha compilers provide the equivalent options called
8516 @option{-scope_safe} and @option{-resumption_safe}.
8518 @item -mieee-conformant
8519 @opindex mieee-conformant
8520 This option marks the generated code as IEEE conformant. You must not
8521 use this option unless you also specify @option{-mtrap-precision=i} and either
8522 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8523 is to emit the line @samp{.eflag 48} in the function prologue of the
8524 generated assembly file. Under DEC Unix, this has the effect that
8525 IEEE-conformant math library routines will be linked in.
8527 @item -mbuild-constants
8528 @opindex mbuild-constants
8529 Normally GCC examines a 32- or 64-bit integer constant to
8530 see if it can construct it from smaller constants in two or three
8531 instructions. If it cannot, it will output the constant as a literal and
8532 generate code to load it from the data segment at runtime.
8534 Use this option to require GCC to construct @emph{all} integer constants
8535 using code, even if it takes more instructions (the maximum is six).
8537 You would typically use this option to build a shared library dynamic
8538 loader. Itself a shared library, it must relocate itself in memory
8539 before it can find the variables and constants in its own data segment.
8545 Select whether to generate code to be assembled by the vendor-supplied
8546 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8564 Indicate whether GCC should generate code to use the optional BWX,
8565 CIX, FIX and MAX instruction sets. The default is to use the instruction
8566 sets supported by the CPU type specified via @option{-mcpu=} option or that
8567 of the CPU on which GCC was built if none was specified.
8572 @opindex mfloat-ieee
8573 Generate code that uses (does not use) VAX F and G floating point
8574 arithmetic instead of IEEE single and double precision.
8576 @item -mexplicit-relocs
8577 @itemx -mno-explicit-relocs
8578 @opindex mexplicit-relocs
8579 @opindex mno-explicit-relocs
8580 Older Alpha assemblers provided no way to generate symbol relocations
8581 except via assembler macros. Use of these macros does not allow
8582 optimal instruction scheduling. GNU binutils as of version 2.12
8583 supports a new syntax that allows the compiler to explicitly mark
8584 which relocations should apply to which instructions. This option
8585 is mostly useful for debugging, as GCC detects the capabilities of
8586 the assembler when it is built and sets the default accordingly.
8590 @opindex msmall-data
8591 @opindex mlarge-data
8592 When @option{-mexplicit-relocs} is in effect, static data is
8593 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8594 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8595 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8596 16-bit relocations off of the @code{$gp} register. This limits the
8597 size of the small data area to 64KB, but allows the variables to be
8598 directly accessed via a single instruction.
8600 The default is @option{-mlarge-data}. With this option the data area
8601 is limited to just below 2GB@. Programs that require more than 2GB of
8602 data must use @code{malloc} or @code{mmap} to allocate the data in the
8603 heap instead of in the program's data segment.
8605 When generating code for shared libraries, @option{-fpic} implies
8606 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8610 @opindex msmall-text
8611 @opindex mlarge-text
8612 When @option{-msmall-text} is used, the compiler assumes that the
8613 code of the entire program (or shared library) fits in 4MB, and is
8614 thus reachable with a branch instruction. When @option{-msmall-data}
8615 is used, the compiler can assume that all local symbols share the
8616 same @code{$gp} value, and thus reduce the number of instructions
8617 required for a function call from 4 to 1.
8619 The default is @option{-mlarge-text}.
8621 @item -mcpu=@var{cpu_type}
8623 Set the instruction set and instruction scheduling parameters for
8624 machine type @var{cpu_type}. You can specify either the @samp{EV}
8625 style name or the corresponding chip number. GCC supports scheduling
8626 parameters for the EV4, EV5 and EV6 family of processors and will
8627 choose the default values for the instruction set from the processor
8628 you specify. If you do not specify a processor type, GCC will default
8629 to the processor on which the compiler was built.
8631 Supported values for @var{cpu_type} are
8637 Schedules as an EV4 and has no instruction set extensions.
8641 Schedules as an EV5 and has no instruction set extensions.
8645 Schedules as an EV5 and supports the BWX extension.
8650 Schedules as an EV5 and supports the BWX and MAX extensions.
8654 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8658 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8661 @item -mtune=@var{cpu_type}
8663 Set only the instruction scheduling parameters for machine type
8664 @var{cpu_type}. The instruction set is not changed.
8666 @item -mmemory-latency=@var{time}
8667 @opindex mmemory-latency
8668 Sets the latency the scheduler should assume for typical memory
8669 references as seen by the application. This number is highly
8670 dependent on the memory access patterns used by the application
8671 and the size of the external cache on the machine.
8673 Valid options for @var{time} are
8677 A decimal number representing clock cycles.
8683 The compiler contains estimates of the number of clock cycles for
8684 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8685 (also called Dcache, Scache, and Bcache), as well as to main memory.
8686 Note that L3 is only valid for EV5.
8691 @node DEC Alpha/VMS Options
8692 @subsection DEC Alpha/VMS Options
8694 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8697 @item -mvms-return-codes
8698 @opindex mvms-return-codes
8699 Return VMS condition codes from main. The default is to return POSIX
8700 style condition (e.g.@ error) codes.
8704 @subsection FRV Options
8711 Only use the first 32 general purpose registers.
8716 Use all 64 general purpose registers.
8721 Use only the first 32 floating point registers.
8726 Use all 64 floating point registers
8729 @opindex mhard-float
8731 Use hardware instructions for floating point operations.
8734 @opindex msoft-float
8736 Use library routines for floating point operations.
8741 Dynamically allocate condition code registers.
8746 Do not try to dynamically allocate condition code registers, only
8747 use @code{icc0} and @code{fcc0}.
8752 Change ABI to use double word insns.
8757 Do not use double word instructions.
8762 Use floating point double instructions.
8767 Do not use floating point double instructions.
8772 Use media instructions.
8777 Do not use media instructions.
8782 Use multiply and add/subtract instructions.
8787 Do not use multiply and add/subtract instructions.
8792 Select the FDPIC ABI, that uses function descriptors to represent
8793 pointers to functions. Without any PIC/PIE-related options, it
8794 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
8795 assumes GOT entries and small data are within a 12-bit range from the
8796 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
8797 are computed with 32 bits.
8800 @opindex minline-plt
8802 Enable inlining of PLT entries in function calls to functions that are
8803 not known to bind locally. It has no effect without @option{-mfdpic}.
8804 It's enabled by default if optimizing for speed and compiling for
8805 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
8806 optimization option such as @option{-O3} or above is present in the
8812 Assume a large TLS segment when generating thread-local code.
8817 Do not assume a large TLS segment when generating thread-local code.
8822 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
8823 that is known to be in read-only sections. It's enabled by default,
8824 except for @option{-fpic} or @option{-fpie}: even though it may help
8825 make the global offset table smaller, it trades 1 instruction for 4.
8826 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
8827 one of which may be shared by multiple symbols, and it avoids the need
8828 for a GOT entry for the referenced symbol, so it's more likely to be a
8829 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
8831 @item -multilib-library-pic
8832 @opindex multilib-library-pic
8834 Link with the (library, not FD) pic libraries. It's implied by
8835 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
8836 @option{-fpic} without @option{-mfdpic}. You should never have to use
8842 Follow the EABI requirement of always creating a frame pointer whenever
8843 a stack frame is allocated. This option is enabled by default and can
8844 be disabled with @option{-mno-linked-fp}.
8847 @opindex mlong-calls
8849 Use indirect addressing to call functions outside the current
8850 compilation unit. This allows the functions to be placed anywhere
8851 within the 32-bit address space.
8853 @item -malign-labels
8854 @opindex malign-labels
8856 Try to align labels to an 8-byte boundary by inserting nops into the
8857 previous packet. This option only has an effect when VLIW packing
8858 is enabled. It doesn't create new packets; it merely adds nops to
8862 @opindex mlibrary-pic
8864 Generate position-independent EABI code.
8869 Use only the first four media accumulator registers.
8874 Use all eight media accumulator registers.
8879 Pack VLIW instructions.
8884 Do not pack VLIW instructions.
8889 Do not mark ABI switches in e_flags.
8894 Enable the use of conditional-move instructions (default).
8896 This switch is mainly for debugging the compiler and will likely be removed
8897 in a future version.
8899 @item -mno-cond-move
8900 @opindex mno-cond-move
8902 Disable the use of conditional-move instructions.
8904 This switch is mainly for debugging the compiler and will likely be removed
8905 in a future version.
8910 Enable the use of conditional set instructions (default).
8912 This switch is mainly for debugging the compiler and will likely be removed
8913 in a future version.
8918 Disable the use of conditional set instructions.
8920 This switch is mainly for debugging the compiler and will likely be removed
8921 in a future version.
8926 Enable the use of conditional execution (default).
8928 This switch is mainly for debugging the compiler and will likely be removed
8929 in a future version.
8931 @item -mno-cond-exec
8932 @opindex mno-cond-exec
8934 Disable the use of conditional execution.
8936 This switch is mainly for debugging the compiler and will likely be removed
8937 in a future version.
8940 @opindex mvliw-branch
8942 Run a pass to pack branches into VLIW instructions (default).
8944 This switch is mainly for debugging the compiler and will likely be removed
8945 in a future version.
8947 @item -mno-vliw-branch
8948 @opindex mno-vliw-branch
8950 Do not run a pass to pack branches into VLIW instructions.
8952 This switch is mainly for debugging the compiler and will likely be removed
8953 in a future version.
8955 @item -mmulti-cond-exec
8956 @opindex mmulti-cond-exec
8958 Enable optimization of @code{&&} and @code{||} in conditional execution
8961 This switch is mainly for debugging the compiler and will likely be removed
8962 in a future version.
8964 @item -mno-multi-cond-exec
8965 @opindex mno-multi-cond-exec
8967 Disable optimization of @code{&&} and @code{||} in conditional execution.
8969 This switch is mainly for debugging the compiler and will likely be removed
8970 in a future version.
8972 @item -mnested-cond-exec
8973 @opindex mnested-cond-exec
8975 Enable nested conditional execution optimizations (default).
8977 This switch is mainly for debugging the compiler and will likely be removed
8978 in a future version.
8980 @item -mno-nested-cond-exec
8981 @opindex mno-nested-cond-exec
8983 Disable nested conditional execution optimizations.
8985 This switch is mainly for debugging the compiler and will likely be removed
8986 in a future version.
8988 @item -moptimize-membar
8989 @opindex moptimize-membar
8991 This switch removes redundant @code{membar} instructions from the
8992 compiler generated code. It is enabled by default.
8994 @item -mno-optimize-membar
8995 @opindex mno-optimize-membar
8997 This switch disables the automatic removal of redundant @code{membar}
8998 instructions from the generated code.
9000 @item -mtomcat-stats
9001 @opindex mtomcat-stats
9003 Cause gas to print out tomcat statistics.
9005 @item -mcpu=@var{cpu}
9008 Select the processor type for which to generate code. Possible values are
9009 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
9010 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
9014 @node GNU/Linux Options
9015 @subsection GNU/Linux Options
9017 These @samp{-m} options are defined for GNU/Linux targets:
9022 Use the GNU C library instead of uClibc. This is the default except
9023 on @samp{*-*-linux-*uclibc*} targets.
9027 Use uClibc instead of the GNU C library. This is the default on
9028 @samp{*-*-linux-*uclibc*} targets.
9031 @node H8/300 Options
9032 @subsection H8/300 Options
9034 These @samp{-m} options are defined for the H8/300 implementations:
9039 Shorten some address references at link time, when possible; uses the
9040 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9041 ld, Using ld}, for a fuller description.
9045 Generate code for the H8/300H@.
9049 Generate code for the H8S@.
9053 Generate code for the H8S and H8/300H in the normal mode. This switch
9054 must be used either with @option{-mh} or @option{-ms}.
9058 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9062 Make @code{int} data 32 bits by default.
9066 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9067 The default for the H8/300H and H8S is to align longs and floats on 4
9069 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9070 This option has no effect on the H8/300.
9074 @subsection HPPA Options
9075 @cindex HPPA Options
9077 These @samp{-m} options are defined for the HPPA family of computers:
9080 @item -march=@var{architecture-type}
9082 Generate code for the specified architecture. The choices for
9083 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
9084 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
9085 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
9086 architecture option for your machine. Code compiled for lower numbered
9087 architectures will run on higher numbered architectures, but not the
9091 @itemx -mpa-risc-1-1
9092 @itemx -mpa-risc-2-0
9093 @opindex mpa-risc-1-0
9094 @opindex mpa-risc-1-1
9095 @opindex mpa-risc-2-0
9096 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
9099 @opindex mbig-switch
9100 Generate code suitable for big switch tables. Use this option only if
9101 the assembler/linker complain about out of range branches within a switch
9104 @item -mjump-in-delay
9105 @opindex mjump-in-delay
9106 Fill delay slots of function calls with unconditional jump instructions
9107 by modifying the return pointer for the function call to be the target
9108 of the conditional jump.
9110 @item -mdisable-fpregs
9111 @opindex mdisable-fpregs
9112 Prevent floating point registers from being used in any manner. This is
9113 necessary for compiling kernels which perform lazy context switching of
9114 floating point registers. If you use this option and attempt to perform
9115 floating point operations, the compiler will abort.
9117 @item -mdisable-indexing
9118 @opindex mdisable-indexing
9119 Prevent the compiler from using indexing address modes. This avoids some
9120 rather obscure problems when compiling MIG generated code under MACH@.
9122 @item -mno-space-regs
9123 @opindex mno-space-regs
9124 Generate code that assumes the target has no space registers. This allows
9125 GCC to generate faster indirect calls and use unscaled index address modes.
9127 Such code is suitable for level 0 PA systems and kernels.
9129 @item -mfast-indirect-calls
9130 @opindex mfast-indirect-calls
9131 Generate code that assumes calls never cross space boundaries. This
9132 allows GCC to emit code which performs faster indirect calls.
9134 This option will not work in the presence of shared libraries or nested
9137 @item -mfixed-range=@var{register-range}
9138 @opindex mfixed-range
9139 Generate code treating the given register range as fixed registers.
9140 A fixed register is one that the register allocator can not use. This is
9141 useful when compiling kernel code. A register range is specified as
9142 two registers separated by a dash. Multiple register ranges can be
9143 specified separated by a comma.
9145 @item -mlong-load-store
9146 @opindex mlong-load-store
9147 Generate 3-instruction load and store sequences as sometimes required by
9148 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
9151 @item -mportable-runtime
9152 @opindex mportable-runtime
9153 Use the portable calling conventions proposed by HP for ELF systems.
9157 Enable the use of assembler directives only GAS understands.
9159 @item -mschedule=@var{cpu-type}
9161 Schedule code according to the constraints for the machine type
9162 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
9163 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
9164 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
9165 proper scheduling option for your machine. The default scheduling is
9169 @opindex mlinker-opt
9170 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
9171 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
9172 linkers in which they give bogus error messages when linking some programs.
9175 @opindex msoft-float
9176 Generate output containing library calls for floating point.
9177 @strong{Warning:} the requisite libraries are not available for all HPPA
9178 targets. Normally the facilities of the machine's usual C compiler are
9179 used, but this cannot be done directly in cross-compilation. You must make
9180 your own arrangements to provide suitable library functions for
9181 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
9182 does provide software floating point support.
9184 @option{-msoft-float} changes the calling convention in the output file;
9185 therefore, it is only useful if you compile @emph{all} of a program with
9186 this option. In particular, you need to compile @file{libgcc.a}, the
9187 library that comes with GCC, with @option{-msoft-float} in order for
9192 Generate the predefine, @code{_SIO}, for server IO@. The default is
9193 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
9194 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
9195 options are available under HP-UX and HI-UX@.
9199 Use GNU ld specific options. This passes @option{-shared} to ld when
9200 building a shared library. It is the default when GCC is configured,
9201 explicitly or implicitly, with the GNU linker. This option does not
9202 have any affect on which ld is called, it only changes what parameters
9203 are passed to that ld. The ld that is called is determined by the
9204 @option{--with-ld} configure option, GCC's program search path, and
9205 finally by the user's @env{PATH}. The linker used by GCC can be printed
9206 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
9207 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9211 Use HP ld specific options. This passes @option{-b} to ld when building
9212 a shared library and passes @option{+Accept TypeMismatch} to ld on all
9213 links. It is the default when GCC is configured, explicitly or
9214 implicitly, with the HP linker. This option does not have any affect on
9215 which ld is called, it only changes what parameters are passed to that
9216 ld. The ld that is called is determined by the @option{--with-ld}
9217 configure option, GCC's program search path, and finally by the user's
9218 @env{PATH}. The linker used by GCC can be printed using @samp{which
9219 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
9220 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9223 @opindex mno-long-calls
9224 Generate code that uses long call sequences. This ensures that a call
9225 is always able to reach linker generated stubs. The default is to generate
9226 long calls only when the distance from the call site to the beginning
9227 of the function or translation unit, as the case may be, exceeds a
9228 predefined limit set by the branch type being used. The limits for
9229 normal calls are 7,600,000 and 240,000 bytes, respectively for the
9230 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
9233 Distances are measured from the beginning of functions when using the
9234 @option{-ffunction-sections} option, or when using the @option{-mgas}
9235 and @option{-mno-portable-runtime} options together under HP-UX with
9238 It is normally not desirable to use this option as it will degrade
9239 performance. However, it may be useful in large applications,
9240 particularly when partial linking is used to build the application.
9242 The types of long calls used depends on the capabilities of the
9243 assembler and linker, and the type of code being generated. The
9244 impact on systems that support long absolute calls, and long pic
9245 symbol-difference or pc-relative calls should be relatively small.
9246 However, an indirect call is used on 32-bit ELF systems in pic code
9247 and it is quite long.
9249 @item -munix=@var{unix-std}
9251 Generate compiler predefines and select a startfile for the specified
9252 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
9253 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
9254 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
9255 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
9256 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
9259 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
9260 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
9261 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
9262 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
9263 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
9264 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
9266 It is @emph{important} to note that this option changes the interfaces
9267 for various library routines. It also affects the operational behavior
9268 of the C library. Thus, @emph{extreme} care is needed in using this
9271 Library code that is intended to operate with more than one UNIX
9272 standard must test, set and restore the variable @var{__xpg4_extended_mask}
9273 as appropriate. Most GNU software doesn't provide this capability.
9277 Suppress the generation of link options to search libdld.sl when the
9278 @option{-static} option is specified on HP-UX 10 and later.
9282 The HP-UX implementation of setlocale in libc has a dependency on
9283 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9284 when the @option{-static} option is specified, special link options
9285 are needed to resolve this dependency.
9287 On HP-UX 10 and later, the GCC driver adds the necessary options to
9288 link with libdld.sl when the @option{-static} option is specified.
9289 This causes the resulting binary to be dynamic. On the 64-bit port,
9290 the linkers generate dynamic binaries by default in any case. The
9291 @option{-nolibdld} option can be used to prevent the GCC driver from
9292 adding these link options.
9296 Add support for multithreading with the @dfn{dce thread} library
9297 under HP-UX@. This option sets flags for both the preprocessor and
9301 @node i386 and x86-64 Options
9302 @subsection Intel 386 and AMD x86-64 Options
9303 @cindex i386 Options
9304 @cindex x86-64 Options
9305 @cindex Intel 386 Options
9306 @cindex AMD x86-64 Options
9308 These @samp{-m} options are defined for the i386 and x86-64 family of
9312 @item -mtune=@var{cpu-type}
9314 Tune to @var{cpu-type} everything applicable about the generated code, except
9315 for the ABI and the set of available instructions. The choices for
9319 Produce code optimized for the most common IA32/AMD64/EM64T processors.
9320 If you know the CPU on which your code will run, then you should use
9321 the corresponding @option{-mtune} option instead of
9322 @option{-mtune=generic}. But, if you do not know exactly what CPU users
9323 of your application will have, then you should use this option.
9325 As new processors are deployed in the marketplace, the behavior of this
9326 option will change. Therefore, if you upgrade to a newer version of
9327 GCC, the code generated option will change to reflect the processors
9328 that were most common when that version of GCC was released.
9330 There is no @option{-march=generic} option because @option{-march}
9331 indicates the instruction set the compiler can use, and there is no
9332 generic instruction set applicable to all processors. In contrast,
9333 @option{-mtune} indicates the processor (or, in this case, collection of
9334 processors) for which the code is optimized.
9336 This selects the CPU to tune for at compilation time by determining
9337 the processor type of the compiling machine. Using @option{-mtune=native}
9338 will produce code optimized for the local machine under the constraints
9339 of the selected instruction set. Using @option{-march=native} will
9340 enable all instruction subsets supported by the local machine (hence
9341 the result might not run on different machines).
9343 Original Intel's i386 CPU@.
9345 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
9347 Intel Pentium CPU with no MMX support.
9349 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
9351 Intel PentiumPro CPU@.
9353 Same as @code{generic}, but when used as @code{march} option, PentiumPro
9354 instruction set will be used, so the code will run on all i686 family chips.
9356 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
9357 @item pentium3, pentium3m
9358 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
9361 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
9362 support. Used by Centrino notebooks.
9363 @item pentium4, pentium4m
9364 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
9366 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
9369 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
9370 SSE2 and SSE3 instruction set support.
9372 AMD K6 CPU with MMX instruction set support.
9374 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
9375 @item athlon, athlon-tbird
9376 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
9378 @item athlon-4, athlon-xp, athlon-mp
9379 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
9380 instruction set support.
9381 @item k8, opteron, athlon64, athlon-fx
9382 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
9383 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
9385 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
9388 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
9389 instruction set support.
9391 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
9392 implemented for this chip.)
9394 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
9395 implemented for this chip.)
9398 While picking a specific @var{cpu-type} will schedule things appropriately
9399 for that particular chip, the compiler will not generate any code that
9400 does not run on the i386 without the @option{-march=@var{cpu-type}} option
9403 @item -march=@var{cpu-type}
9405 Generate instructions for the machine type @var{cpu-type}. The choices
9406 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
9407 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
9409 @item -mcpu=@var{cpu-type}
9411 A deprecated synonym for @option{-mtune}.
9420 @opindex mpentiumpro
9421 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
9422 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
9423 These synonyms are deprecated.
9425 @item -mfpmath=@var{unit}
9427 Generate floating point arithmetics for selected unit @var{unit}. The choices
9432 Use the standard 387 floating point coprocessor present majority of chips and
9433 emulated otherwise. Code compiled with this option will run almost everywhere.
9434 The temporary results are computed in 80bit precision instead of precision
9435 specified by the type resulting in slightly different results compared to most
9436 of other chips. See @option{-ffloat-store} for more detailed description.
9438 This is the default choice for i386 compiler.
9441 Use scalar floating point instructions present in the SSE instruction set.
9442 This instruction set is supported by Pentium3 and newer chips, in the AMD line
9443 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
9444 instruction set supports only single precision arithmetics, thus the double and
9445 extended precision arithmetics is still done using 387. Later version, present
9446 only in Pentium4 and the future AMD x86-64 chips supports double precision
9449 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
9450 or @option{-msse2} switches to enable SSE extensions and make this option
9451 effective. For the x86-64 compiler, these extensions are enabled by default.
9453 The resulting code should be considerably faster in the majority of cases and avoid
9454 the numerical instability problems of 387 code, but may break some existing
9455 code that expects temporaries to be 80bit.
9457 This is the default choice for the x86-64 compiler.
9460 Attempt to utilize both instruction sets at once. This effectively double the
9461 amount of available registers and on chips with separate execution units for
9462 387 and SSE the execution resources too. Use this option with care, as it is
9463 still experimental, because the GCC register allocator does not model separate
9464 functional units well resulting in instable performance.
9467 @item -masm=@var{dialect}
9468 @opindex masm=@var{dialect}
9469 Output asm instructions using selected @var{dialect}. Supported
9470 choices are @samp{intel} or @samp{att} (the default one). Darwin does
9471 not support @samp{intel}.
9476 @opindex mno-ieee-fp
9477 Control whether or not the compiler uses IEEE floating point
9478 comparisons. These handle correctly the case where the result of a
9479 comparison is unordered.
9482 @opindex msoft-float
9483 Generate output containing library calls for floating point.
9484 @strong{Warning:} the requisite libraries are not part of GCC@.
9485 Normally the facilities of the machine's usual C compiler are used, but
9486 this can't be done directly in cross-compilation. You must make your
9487 own arrangements to provide suitable library functions for
9490 On machines where a function returns floating point results in the 80387
9491 register stack, some floating point opcodes may be emitted even if
9492 @option{-msoft-float} is used.
9494 @item -mno-fp-ret-in-387
9495 @opindex mno-fp-ret-in-387
9496 Do not use the FPU registers for return values of functions.
9498 The usual calling convention has functions return values of types
9499 @code{float} and @code{double} in an FPU register, even if there
9500 is no FPU@. The idea is that the operating system should emulate
9503 The option @option{-mno-fp-ret-in-387} causes such values to be returned
9504 in ordinary CPU registers instead.
9506 @item -mno-fancy-math-387
9507 @opindex mno-fancy-math-387
9508 Some 387 emulators do not support the @code{sin}, @code{cos} and
9509 @code{sqrt} instructions for the 387. Specify this option to avoid
9510 generating those instructions. This option is the default on FreeBSD,
9511 OpenBSD and NetBSD@. This option is overridden when @option{-march}
9512 indicates that the target cpu will always have an FPU and so the
9513 instruction will not need emulation. As of revision 2.6.1, these
9514 instructions are not generated unless you also use the
9515 @option{-funsafe-math-optimizations} switch.
9517 @item -malign-double
9518 @itemx -mno-align-double
9519 @opindex malign-double
9520 @opindex mno-align-double
9521 Control whether GCC aligns @code{double}, @code{long double}, and
9522 @code{long long} variables on a two word boundary or a one word
9523 boundary. Aligning @code{double} variables on a two word boundary will
9524 produce code that runs somewhat faster on a @samp{Pentium} at the
9525 expense of more memory.
9527 On x86-64, @option{-malign-double} is enabled by default.
9529 @strong{Warning:} if you use the @option{-malign-double} switch,
9530 structures containing the above types will be aligned differently than
9531 the published application binary interface specifications for the 386
9532 and will not be binary compatible with structures in code compiled
9533 without that switch.
9535 @item -m96bit-long-double
9536 @itemx -m128bit-long-double
9537 @opindex m96bit-long-double
9538 @opindex m128bit-long-double
9539 These switches control the size of @code{long double} type. The i386
9540 application binary interface specifies the size to be 96 bits,
9541 so @option{-m96bit-long-double} is the default in 32 bit mode.
9543 Modern architectures (Pentium and newer) would prefer @code{long double}
9544 to be aligned to an 8 or 16 byte boundary. In arrays or structures
9545 conforming to the ABI, this would not be possible. So specifying a
9546 @option{-m128bit-long-double} will align @code{long double}
9547 to a 16 byte boundary by padding the @code{long double} with an additional
9550 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
9551 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
9553 Notice that neither of these options enable any extra precision over the x87
9554 standard of 80 bits for a @code{long double}.
9556 @strong{Warning:} if you override the default value for your target ABI, the
9557 structures and arrays containing @code{long double} variables will change
9558 their size as well as function calling convention for function taking
9559 @code{long double} will be modified. Hence they will not be binary
9560 compatible with arrays or structures in code compiled without that switch.
9562 @item -mmlarge-data-threshold=@var{number}
9563 @opindex mlarge-data-threshold=@var{number}
9564 When @option{-mcmodel=medium} is specified, the data greater than
9565 @var{threshold} are placed in large data section. This value must be the
9566 same across all object linked into the binary and defaults to 65535.
9569 @itemx -mno-svr3-shlib
9570 @opindex msvr3-shlib
9571 @opindex mno-svr3-shlib
9572 Control whether GCC places uninitialized local variables into the
9573 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
9574 into @code{bss}. These options are meaningful only on System V Release 3.
9578 Use a different function-calling convention, in which functions that
9579 take a fixed number of arguments return with the @code{ret} @var{num}
9580 instruction, which pops their arguments while returning. This saves one
9581 instruction in the caller since there is no need to pop the arguments
9584 You can specify that an individual function is called with this calling
9585 sequence with the function attribute @samp{stdcall}. You can also
9586 override the @option{-mrtd} option by using the function attribute
9587 @samp{cdecl}. @xref{Function Attributes}.
9589 @strong{Warning:} this calling convention is incompatible with the one
9590 normally used on Unix, so you cannot use it if you need to call
9591 libraries compiled with the Unix compiler.
9593 Also, you must provide function prototypes for all functions that
9594 take variable numbers of arguments (including @code{printf});
9595 otherwise incorrect code will be generated for calls to those
9598 In addition, seriously incorrect code will result if you call a
9599 function with too many arguments. (Normally, extra arguments are
9600 harmlessly ignored.)
9602 @item -mregparm=@var{num}
9604 Control how many registers are used to pass integer arguments. By
9605 default, no registers are used to pass arguments, and at most 3
9606 registers can be used. You can control this behavior for a specific
9607 function by using the function attribute @samp{regparm}.
9608 @xref{Function Attributes}.
9610 @strong{Warning:} if you use this switch, and
9611 @var{num} is nonzero, then you must build all modules with the same
9612 value, including any libraries. This includes the system libraries and
9616 @opindex msseregparm
9617 Use SSE register passing conventions for float and double arguments
9618 and return values. You can control this behavior for a specific
9619 function by using the function attribute @samp{sseregparm}.
9620 @xref{Function Attributes}.
9622 @strong{Warning:} if you use this switch then you must build all
9623 modules with the same value, including any libraries. This includes
9624 the system libraries and startup modules.
9626 @item -mstackrealign
9627 @opindex mstackrealign
9628 Realign the stack at entry. On the Intel x86, the
9629 @option{-mstackrealign} option will generate an alternate prologue and
9630 epilogue that realigns the runtime stack. This supports mixing legacy
9631 codes that keep a 4-byte aligned stack with modern codes that keep a
9632 16-byte stack for SSE compatibility. The alternate prologue and
9633 epilogue are slower and bigger than the regular ones, and the
9634 alternate prologue requires an extra scratch register; this lowers the
9635 number of registers available if used in conjunction with the
9636 @code{regparm} attribute. The @option{-mstackrealign} option is
9637 incompatible with the nested function prologue; this is considered a
9638 hard error. See also the attribute @code{force_align_arg_pointer},
9639 applicable to individual functions.
9641 @item -mpreferred-stack-boundary=@var{num}
9642 @opindex mpreferred-stack-boundary
9643 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
9644 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
9645 the default is 4 (16 bytes or 128 bits).
9647 On Pentium and PentiumPro, @code{double} and @code{long double} values
9648 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
9649 suffer significant run time performance penalties. On Pentium III, the
9650 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
9651 properly if it is not 16 byte aligned.
9653 To ensure proper alignment of this values on the stack, the stack boundary
9654 must be as aligned as that required by any value stored on the stack.
9655 Further, every function must be generated such that it keeps the stack
9656 aligned. Thus calling a function compiled with a higher preferred
9657 stack boundary from a function compiled with a lower preferred stack
9658 boundary will most likely misalign the stack. It is recommended that
9659 libraries that use callbacks always use the default setting.
9661 This extra alignment does consume extra stack space, and generally
9662 increases code size. Code that is sensitive to stack space usage, such
9663 as embedded systems and operating system kernels, may want to reduce the
9664 preferred alignment to @option{-mpreferred-stack-boundary=2}.
9682 These switches enable or disable the use of instructions in the MMX,
9683 SSE, SSE2 or 3DNow! extended instruction sets. These extensions are
9684 also available as built-in functions: see @ref{X86 Built-in Functions},
9685 for details of the functions enabled and disabled by these switches.
9687 To have SSE/SSE2 instructions generated automatically from floating-point
9688 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
9690 These options will enable GCC to use these extended instructions in
9691 generated code, even without @option{-mfpmath=sse}. Applications which
9692 perform runtime CPU detection must compile separate files for each
9693 supported architecture, using the appropriate flags. In particular,
9694 the file containing the CPU detection code should be compiled without
9698 @itemx -mno-push-args
9700 @opindex mno-push-args
9701 Use PUSH operations to store outgoing parameters. This method is shorter
9702 and usually equally fast as method using SUB/MOV operations and is enabled
9703 by default. In some cases disabling it may improve performance because of
9704 improved scheduling and reduced dependencies.
9706 @item -maccumulate-outgoing-args
9707 @opindex maccumulate-outgoing-args
9708 If enabled, the maximum amount of space required for outgoing arguments will be
9709 computed in the function prologue. This is faster on most modern CPUs
9710 because of reduced dependencies, improved scheduling and reduced stack usage
9711 when preferred stack boundary is not equal to 2. The drawback is a notable
9712 increase in code size. This switch implies @option{-mno-push-args}.
9716 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
9717 on thread-safe exception handling must compile and link all code with the
9718 @option{-mthreads} option. When compiling, @option{-mthreads} defines
9719 @option{-D_MT}; when linking, it links in a special thread helper library
9720 @option{-lmingwthrd} which cleans up per thread exception handling data.
9722 @item -mno-align-stringops
9723 @opindex mno-align-stringops
9724 Do not align destination of inlined string operations. This switch reduces
9725 code size and improves performance in case the destination is already aligned,
9726 but GCC doesn't know about it.
9728 @item -minline-all-stringops
9729 @opindex minline-all-stringops
9730 By default GCC inlines string operations only when destination is known to be
9731 aligned at least to 4 byte boundary. This enables more inlining, increase code
9732 size, but may improve performance of code that depends on fast memcpy, strlen
9733 and memset for short lengths.
9735 @item -momit-leaf-frame-pointer
9736 @opindex momit-leaf-frame-pointer
9737 Don't keep the frame pointer in a register for leaf functions. This
9738 avoids the instructions to save, set up and restore frame pointers and
9739 makes an extra register available in leaf functions. The option
9740 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9741 which might make debugging harder.
9743 @item -mtls-direct-seg-refs
9744 @itemx -mno-tls-direct-seg-refs
9745 @opindex mtls-direct-seg-refs
9746 Controls whether TLS variables may be accessed with offsets from the
9747 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
9748 or whether the thread base pointer must be added. Whether or not this
9749 is legal depends on the operating system, and whether it maps the
9750 segment to cover the entire TLS area.
9752 For systems that use GNU libc, the default is on.
9755 These @samp{-m} switches are supported in addition to the above
9756 on AMD x86-64 processors in 64-bit environments.
9763 Generate code for a 32-bit or 64-bit environment.
9764 The 32-bit environment sets int, long and pointer to 32 bits and
9765 generates code that runs on any i386 system.
9766 The 64-bit environment sets int to 32 bits and long and pointer
9767 to 64 bits and generates code for AMD's x86-64 architecture. For
9768 darwin only the -m64 option turns off the @option{-fno-pic} and
9769 @option{-mdynamic-no-pic} options.
9772 @opindex no-red-zone
9773 Do not use a so called red zone for x86-64 code. The red zone is mandated
9774 by the x86-64 ABI, it is a 128-byte area beyond the location of the
9775 stack pointer that will not be modified by signal or interrupt handlers
9776 and therefore can be used for temporary data without adjusting the stack
9777 pointer. The flag @option{-mno-red-zone} disables this red zone.
9779 @item -mcmodel=small
9780 @opindex mcmodel=small
9781 Generate code for the small code model: the program and its symbols must
9782 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
9783 Programs can be statically or dynamically linked. This is the default
9786 @item -mcmodel=kernel
9787 @opindex mcmodel=kernel
9788 Generate code for the kernel code model. The kernel runs in the
9789 negative 2 GB of the address space.
9790 This model has to be used for Linux kernel code.
9792 @item -mcmodel=medium
9793 @opindex mcmodel=medium
9794 Generate code for the medium model: The program is linked in the lower 2
9795 GB of the address space but symbols can be located anywhere in the
9796 address space. Programs can be statically or dynamically linked, but
9797 building of shared libraries are not supported with the medium model.
9799 @item -mcmodel=large
9800 @opindex mcmodel=large
9801 Generate code for the large model: This model makes no assumptions
9802 about addresses and sizes of sections. Currently GCC does not implement
9807 @subsection IA-64 Options
9808 @cindex IA-64 Options
9810 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9814 @opindex mbig-endian
9815 Generate code for a big endian target. This is the default for HP-UX@.
9817 @item -mlittle-endian
9818 @opindex mlittle-endian
9819 Generate code for a little endian target. This is the default for AIX5
9826 Generate (or don't) code for the GNU assembler. This is the default.
9827 @c Also, this is the default if the configure option @option{--with-gnu-as}
9834 Generate (or don't) code for the GNU linker. This is the default.
9835 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9840 Generate code that does not use a global pointer register. The result
9841 is not position independent code, and violates the IA-64 ABI@.
9843 @item -mvolatile-asm-stop
9844 @itemx -mno-volatile-asm-stop
9845 @opindex mvolatile-asm-stop
9846 @opindex mno-volatile-asm-stop
9847 Generate (or don't) a stop bit immediately before and after volatile asm
9850 @item -mregister-names
9851 @itemx -mno-register-names
9852 @opindex mregister-names
9853 @opindex mno-register-names
9854 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9855 the stacked registers. This may make assembler output more readable.
9861 Disable (or enable) optimizations that use the small data section. This may
9862 be useful for working around optimizer bugs.
9865 @opindex mconstant-gp
9866 Generate code that uses a single constant global pointer value. This is
9867 useful when compiling kernel code.
9871 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9872 This is useful when compiling firmware code.
9874 @item -minline-float-divide-min-latency
9875 @opindex minline-float-divide-min-latency
9876 Generate code for inline divides of floating point values
9877 using the minimum latency algorithm.
9879 @item -minline-float-divide-max-throughput
9880 @opindex minline-float-divide-max-throughput
9881 Generate code for inline divides of floating point values
9882 using the maximum throughput algorithm.
9884 @item -minline-int-divide-min-latency
9885 @opindex minline-int-divide-min-latency
9886 Generate code for inline divides of integer values
9887 using the minimum latency algorithm.
9889 @item -minline-int-divide-max-throughput
9890 @opindex minline-int-divide-max-throughput
9891 Generate code for inline divides of integer values
9892 using the maximum throughput algorithm.
9894 @item -minline-sqrt-min-latency
9895 @opindex minline-sqrt-min-latency
9896 Generate code for inline square roots
9897 using the minimum latency algorithm.
9899 @item -minline-sqrt-max-throughput
9900 @opindex minline-sqrt-max-throughput
9901 Generate code for inline square roots
9902 using the maximum throughput algorithm.
9904 @item -mno-dwarf2-asm
9906 @opindex mno-dwarf2-asm
9907 @opindex mdwarf2-asm
9908 Don't (or do) generate assembler code for the DWARF2 line number debugging
9909 info. This may be useful when not using the GNU assembler.
9911 @item -mearly-stop-bits
9912 @itemx -mno-early-stop-bits
9913 @opindex mearly-stop-bits
9914 @opindex mno-early-stop-bits
9915 Allow stop bits to be placed earlier than immediately preceding the
9916 instruction that triggered the stop bit. This can improve instruction
9917 scheduling, but does not always do so.
9919 @item -mfixed-range=@var{register-range}
9920 @opindex mfixed-range
9921 Generate code treating the given register range as fixed registers.
9922 A fixed register is one that the register allocator can not use. This is
9923 useful when compiling kernel code. A register range is specified as
9924 two registers separated by a dash. Multiple register ranges can be
9925 specified separated by a comma.
9927 @item -mtls-size=@var{tls-size}
9929 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
9932 @item -mtune=@var{cpu-type}
9934 Tune the instruction scheduling for a particular CPU, Valid values are
9935 itanium, itanium1, merced, itanium2, and mckinley.
9941 Add support for multithreading using the POSIX threads library. This
9942 option sets flags for both the preprocessor and linker. It does
9943 not affect the thread safety of object code produced by the compiler or
9944 that of libraries supplied with it. These are HP-UX specific flags.
9950 Generate code for a 32-bit or 64-bit environment.
9951 The 32-bit environment sets int, long and pointer to 32 bits.
9952 The 64-bit environment sets int to 32 bits and long and pointer
9953 to 64 bits. These are HP-UX specific flags.
9955 @item -mno-sched-br-data-spec
9956 @itemx -msched-br-data-spec
9957 @opindex mno-sched-br-data-spec
9958 @opindex msched-br-data-spec
9959 (Dis/En)able data speculative scheduling before reload.
9960 This will result in generation of the ld.a instructions and
9961 the corresponding check instructions (ld.c / chk.a).
9962 The default is 'disable'.
9964 @item -msched-ar-data-spec
9965 @itemx -mno-sched-ar-data-spec
9966 @opindex msched-ar-data-spec
9967 @opindex mno-sched-ar-data-spec
9968 (En/Dis)able data speculative scheduling after reload.
9969 This will result in generation of the ld.a instructions and
9970 the corresponding check instructions (ld.c / chk.a).
9971 The default is 'enable'.
9973 @item -mno-sched-control-spec
9974 @itemx -msched-control-spec
9975 @opindex mno-sched-control-spec
9976 @opindex msched-control-spec
9977 (Dis/En)able control speculative scheduling. This feature is
9978 available only during region scheduling (i.e. before reload).
9979 This will result in generation of the ld.s instructions and
9980 the corresponding check instructions chk.s .
9981 The default is 'disable'.
9983 @item -msched-br-in-data-spec
9984 @itemx -mno-sched-br-in-data-spec
9985 @opindex msched-br-in-data-spec
9986 @opindex mno-sched-br-in-data-spec
9987 (En/Dis)able speculative scheduling of the instructions that
9988 are dependent on the data speculative loads before reload.
9989 This is effective only with @option{-msched-br-data-spec} enabled.
9990 The default is 'enable'.
9992 @item -msched-ar-in-data-spec
9993 @itemx -mno-sched-ar-in-data-spec
9994 @opindex msched-ar-in-data-spec
9995 @opindex mno-sched-ar-in-data-spec
9996 (En/Dis)able speculative scheduling of the instructions that
9997 are dependent on the data speculative loads after reload.
9998 This is effective only with @option{-msched-ar-data-spec} enabled.
9999 The default is 'enable'.
10001 @item -msched-in-control-spec
10002 @itemx -mno-sched-in-control-spec
10003 @opindex msched-in-control-spec
10004 @opindex mno-sched-in-control-spec
10005 (En/Dis)able speculative scheduling of the instructions that
10006 are dependent on the control speculative loads.
10007 This is effective only with @option{-msched-control-spec} enabled.
10008 The default is 'enable'.
10011 @itemx -mno-sched-ldc
10012 @opindex msched-ldc
10013 @opindex mno-sched-ldc
10014 (En/Dis)able use of simple data speculation checks ld.c .
10015 If disabled, only chk.a instructions will be emitted to check
10016 data speculative loads.
10017 The default is 'enable'.
10019 @item -mno-sched-control-ldc
10020 @itemx -msched-control-ldc
10021 @opindex mno-sched-control-ldc
10022 @opindex msched-control-ldc
10023 (Dis/En)able use of ld.c instructions to check control speculative loads.
10024 If enabled, in case of control speculative load with no speculatively
10025 scheduled dependent instructions this load will be emitted as ld.sa and
10026 ld.c will be used to check it.
10027 The default is 'disable'.
10029 @item -mno-sched-spec-verbose
10030 @itemx -msched-spec-verbose
10031 @opindex mno-sched-spec-verbose
10032 @opindex msched-spec-verbose
10033 (Dis/En)able printing of the information about speculative motions.
10035 @item -mno-sched-prefer-non-data-spec-insns
10036 @itemx -msched-prefer-non-data-spec-insns
10037 @opindex mno-sched-prefer-non-data-spec-insns
10038 @opindex msched-prefer-non-data-spec-insns
10039 If enabled, data speculative instructions will be chosen for schedule
10040 only if there are no other choices at the moment. This will make
10041 the use of the data speculation much more conservative.
10042 The default is 'disable'.
10044 @item -mno-sched-prefer-non-control-spec-insns
10045 @itemx -msched-prefer-non-control-spec-insns
10046 @opindex mno-sched-prefer-non-control-spec-insns
10047 @opindex msched-prefer-non-control-spec-insns
10048 If enabled, control speculative instructions will be chosen for schedule
10049 only if there are no other choices at the moment. This will make
10050 the use of the control speculation much more conservative.
10051 The default is 'disable'.
10053 @item -mno-sched-count-spec-in-critical-path
10054 @itemx -msched-count-spec-in-critical-path
10055 @opindex mno-sched-count-spec-in-critical-path
10056 @opindex msched-count-spec-in-critical-path
10057 If enabled, speculative dependencies will be considered during
10058 computation of the instructions priorities. This will make the use of the
10059 speculation a bit more conservative.
10060 The default is 'disable'.
10065 @subsection M32C Options
10066 @cindex M32C options
10069 @item -mcpu=@var{name}
10071 Select the CPU for which code is generated. @var{name} may be one of
10072 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
10073 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
10074 the M32C/80 series.
10078 Specifies that the program will be run on the simulator. This causes
10079 an alternate runtime library to be linked in which supports, for
10080 example, file I/O. You must not use this option when generating
10081 programs that will run on real hardware; you must provide your own
10082 runtime library for whatever I/O functions are needed.
10084 @item -memregs=@var{number}
10086 Specifies the number of memory-based pseudo-registers GCC will use
10087 during code generation. These pseudo-registers will be used like real
10088 registers, so there is a tradeoff between GCC's ability to fit the
10089 code into available registers, and the performance penalty of using
10090 memory instead of registers. Note that all modules in a program must
10091 be compiled with the same value for this option. Because of that, you
10092 must not use this option with the default runtime libraries gcc
10097 @node M32R/D Options
10098 @subsection M32R/D Options
10099 @cindex M32R/D options
10101 These @option{-m} options are defined for Renesas M32R/D architectures:
10106 Generate code for the M32R/2@.
10110 Generate code for the M32R/X@.
10114 Generate code for the M32R@. This is the default.
10116 @item -mmodel=small
10117 @opindex mmodel=small
10118 Assume all objects live in the lower 16MB of memory (so that their addresses
10119 can be loaded with the @code{ld24} instruction), and assume all subroutines
10120 are reachable with the @code{bl} instruction.
10121 This is the default.
10123 The addressability of a particular object can be set with the
10124 @code{model} attribute.
10126 @item -mmodel=medium
10127 @opindex mmodel=medium
10128 Assume objects may be anywhere in the 32-bit address space (the compiler
10129 will generate @code{seth/add3} instructions to load their addresses), and
10130 assume all subroutines are reachable with the @code{bl} instruction.
10132 @item -mmodel=large
10133 @opindex mmodel=large
10134 Assume objects may be anywhere in the 32-bit address space (the compiler
10135 will generate @code{seth/add3} instructions to load their addresses), and
10136 assume subroutines may not be reachable with the @code{bl} instruction
10137 (the compiler will generate the much slower @code{seth/add3/jl}
10138 instruction sequence).
10141 @opindex msdata=none
10142 Disable use of the small data area. Variables will be put into
10143 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
10144 @code{section} attribute has been specified).
10145 This is the default.
10147 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
10148 Objects may be explicitly put in the small data area with the
10149 @code{section} attribute using one of these sections.
10151 @item -msdata=sdata
10152 @opindex msdata=sdata
10153 Put small global and static data in the small data area, but do not
10154 generate special code to reference them.
10157 @opindex msdata=use
10158 Put small global and static data in the small data area, and generate
10159 special instructions to reference them.
10163 @cindex smaller data references
10164 Put global and static objects less than or equal to @var{num} bytes
10165 into the small data or bss sections instead of the normal data or bss
10166 sections. The default value of @var{num} is 8.
10167 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
10168 for this option to have any effect.
10170 All modules should be compiled with the same @option{-G @var{num}} value.
10171 Compiling with different values of @var{num} may or may not work; if it
10172 doesn't the linker will give an error message---incorrect code will not be
10177 Makes the M32R specific code in the compiler display some statistics
10178 that might help in debugging programs.
10180 @item -malign-loops
10181 @opindex malign-loops
10182 Align all loops to a 32-byte boundary.
10184 @item -mno-align-loops
10185 @opindex mno-align-loops
10186 Do not enforce a 32-byte alignment for loops. This is the default.
10188 @item -missue-rate=@var{number}
10189 @opindex missue-rate=@var{number}
10190 Issue @var{number} instructions per cycle. @var{number} can only be 1
10193 @item -mbranch-cost=@var{number}
10194 @opindex mbranch-cost=@var{number}
10195 @var{number} can only be 1 or 2. If it is 1 then branches will be
10196 preferred over conditional code, if it is 2, then the opposite will
10199 @item -mflush-trap=@var{number}
10200 @opindex mflush-trap=@var{number}
10201 Specifies the trap number to use to flush the cache. The default is
10202 12. Valid numbers are between 0 and 15 inclusive.
10204 @item -mno-flush-trap
10205 @opindex mno-flush-trap
10206 Specifies that the cache cannot be flushed by using a trap.
10208 @item -mflush-func=@var{name}
10209 @opindex mflush-func=@var{name}
10210 Specifies the name of the operating system function to call to flush
10211 the cache. The default is @emph{_flush_cache}, but a function call
10212 will only be used if a trap is not available.
10214 @item -mno-flush-func
10215 @opindex mno-flush-func
10216 Indicates that there is no OS function for flushing the cache.
10220 @node M680x0 Options
10221 @subsection M680x0 Options
10222 @cindex M680x0 options
10224 These are the @samp{-m} options defined for the 68000 series. The default
10225 values for these options depends on which style of 68000 was selected when
10226 the compiler was configured; the defaults for the most common choices are
10234 Generate output for a 68000. This is the default
10235 when the compiler is configured for 68000-based systems.
10237 Use this option for microcontrollers with a 68000 or EC000 core,
10238 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
10244 Generate output for a 68020. This is the default
10245 when the compiler is configured for 68020-based systems.
10249 Generate output containing 68881 instructions for floating point.
10250 This is the default for most 68020 systems unless @option{--nfp} was
10251 specified when the compiler was configured.
10255 Generate output for a 68030. This is the default when the compiler is
10256 configured for 68030-based systems.
10260 Generate output for a 68040. This is the default when the compiler is
10261 configured for 68040-based systems.
10263 This option inhibits the use of 68881/68882 instructions that have to be
10264 emulated by software on the 68040. Use this option if your 68040 does not
10265 have code to emulate those instructions.
10269 Generate output for a 68060. This is the default when the compiler is
10270 configured for 68060-based systems.
10272 This option inhibits the use of 68020 and 68881/68882 instructions that
10273 have to be emulated by software on the 68060. Use this option if your 68060
10274 does not have code to emulate those instructions.
10278 Generate output for a CPU32. This is the default
10279 when the compiler is configured for CPU32-based systems.
10281 Use this option for microcontrollers with a
10282 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
10283 68336, 68340, 68341, 68349 and 68360.
10287 Generate output for a 520X ``coldfire'' family cpu. This is the default
10288 when the compiler is configured for 520X-based systems.
10290 Use this option for microcontroller with a 5200 core, including
10291 the MCF5202, MCF5203, MCF5204 and MCF5202.
10295 Generate output for a ColdFire V4e family cpu (e.g.@: 547x/548x).
10296 This includes use of hardware floating point instructions.
10300 Generate output for a 68040, without using any of the new instructions.
10301 This results in code which can run relatively efficiently on either a
10302 68020/68881 or a 68030 or a 68040. The generated code does use the
10303 68881 instructions that are emulated on the 68040.
10307 Generate output for a 68060, without using any of the new instructions.
10308 This results in code which can run relatively efficiently on either a
10309 68020/68881 or a 68030 or a 68040. The generated code does use the
10310 68881 instructions that are emulated on the 68060.
10313 @opindex msoft-float
10314 Generate output containing library calls for floating point.
10315 @strong{Warning:} the requisite libraries are not available for all m68k
10316 targets. Normally the facilities of the machine's usual C compiler are
10317 used, but this can't be done directly in cross-compilation. You must
10318 make your own arrangements to provide suitable library functions for
10319 cross-compilation. The embedded targets @samp{m68k-*-aout} and
10320 @samp{m68k-*-coff} do provide software floating point support.
10324 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10325 Additionally, parameters passed on the stack are also aligned to a
10326 16-bit boundary even on targets whose API mandates promotion to 32-bit.
10329 @opindex mnobitfield
10330 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
10331 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
10335 Do use the bit-field instructions. The @option{-m68020} option implies
10336 @option{-mbitfield}. This is the default if you use a configuration
10337 designed for a 68020.
10341 Use a different function-calling convention, in which functions
10342 that take a fixed number of arguments return with the @code{rtd}
10343 instruction, which pops their arguments while returning. This
10344 saves one instruction in the caller since there is no need to pop
10345 the arguments there.
10347 This calling convention is incompatible with the one normally
10348 used on Unix, so you cannot use it if you need to call libraries
10349 compiled with the Unix compiler.
10351 Also, you must provide function prototypes for all functions that
10352 take variable numbers of arguments (including @code{printf});
10353 otherwise incorrect code will be generated for calls to those
10356 In addition, seriously incorrect code will result if you call a
10357 function with too many arguments. (Normally, extra arguments are
10358 harmlessly ignored.)
10360 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
10361 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
10364 @itemx -mno-align-int
10365 @opindex malign-int
10366 @opindex mno-align-int
10367 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
10368 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
10369 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
10370 Aligning variables on 32-bit boundaries produces code that runs somewhat
10371 faster on processors with 32-bit busses at the expense of more memory.
10373 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
10374 align structures containing the above types differently than
10375 most published application binary interface specifications for the m68k.
10379 Use the pc-relative addressing mode of the 68000 directly, instead of
10380 using a global offset table. At present, this option implies @option{-fpic},
10381 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
10382 not presently supported with @option{-mpcrel}, though this could be supported for
10383 68020 and higher processors.
10385 @item -mno-strict-align
10386 @itemx -mstrict-align
10387 @opindex mno-strict-align
10388 @opindex mstrict-align
10389 Do not (do) assume that unaligned memory references will be handled by
10393 Generate code that allows the data segment to be located in a different
10394 area of memory from the text segment. This allows for execute in place in
10395 an environment without virtual memory management. This option implies
10398 @item -mno-sep-data
10399 Generate code that assumes that the data segment follows the text segment.
10400 This is the default.
10402 @item -mid-shared-library
10403 Generate code that supports shared libraries via the library ID method.
10404 This allows for execute in place and shared libraries in an environment
10405 without virtual memory management. This option implies @option{-fPIC}.
10407 @item -mno-id-shared-library
10408 Generate code that doesn't assume ID based shared libraries are being used.
10409 This is the default.
10411 @item -mshared-library-id=n
10412 Specified the identification number of the ID based shared library being
10413 compiled. Specifying a value of 0 will generate more compact code, specifying
10414 other values will force the allocation of that number to the current
10415 library but is no more space or time efficient than omitting this option.
10419 @node M68hc1x Options
10420 @subsection M68hc1x Options
10421 @cindex M68hc1x options
10423 These are the @samp{-m} options defined for the 68hc11 and 68hc12
10424 microcontrollers. The default values for these options depends on
10425 which style of microcontroller was selected when the compiler was configured;
10426 the defaults for the most common choices are given below.
10433 Generate output for a 68HC11. This is the default
10434 when the compiler is configured for 68HC11-based systems.
10440 Generate output for a 68HC12. This is the default
10441 when the compiler is configured for 68HC12-based systems.
10447 Generate output for a 68HCS12.
10449 @item -mauto-incdec
10450 @opindex mauto-incdec
10451 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
10458 Enable the use of 68HC12 min and max instructions.
10461 @itemx -mno-long-calls
10462 @opindex mlong-calls
10463 @opindex mno-long-calls
10464 Treat all calls as being far away (near). If calls are assumed to be
10465 far away, the compiler will use the @code{call} instruction to
10466 call a function and the @code{rtc} instruction for returning.
10470 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10472 @item -msoft-reg-count=@var{count}
10473 @opindex msoft-reg-count
10474 Specify the number of pseudo-soft registers which are used for the
10475 code generation. The maximum number is 32. Using more pseudo-soft
10476 register may or may not result in better code depending on the program.
10477 The default is 4 for 68HC11 and 2 for 68HC12.
10481 @node MCore Options
10482 @subsection MCore Options
10483 @cindex MCore options
10485 These are the @samp{-m} options defined for the Motorola M*Core
10491 @itemx -mno-hardlit
10493 @opindex mno-hardlit
10494 Inline constants into the code stream if it can be done in two
10495 instructions or less.
10501 Use the divide instruction. (Enabled by default).
10503 @item -mrelax-immediate
10504 @itemx -mno-relax-immediate
10505 @opindex mrelax-immediate
10506 @opindex mno-relax-immediate
10507 Allow arbitrary sized immediates in bit operations.
10509 @item -mwide-bitfields
10510 @itemx -mno-wide-bitfields
10511 @opindex mwide-bitfields
10512 @opindex mno-wide-bitfields
10513 Always treat bit-fields as int-sized.
10515 @item -m4byte-functions
10516 @itemx -mno-4byte-functions
10517 @opindex m4byte-functions
10518 @opindex mno-4byte-functions
10519 Force all functions to be aligned to a four byte boundary.
10521 @item -mcallgraph-data
10522 @itemx -mno-callgraph-data
10523 @opindex mcallgraph-data
10524 @opindex mno-callgraph-data
10525 Emit callgraph information.
10528 @itemx -mno-slow-bytes
10529 @opindex mslow-bytes
10530 @opindex mno-slow-bytes
10531 Prefer word access when reading byte quantities.
10533 @item -mlittle-endian
10534 @itemx -mbig-endian
10535 @opindex mlittle-endian
10536 @opindex mbig-endian
10537 Generate code for a little endian target.
10543 Generate code for the 210 processor.
10547 @subsection MIPS Options
10548 @cindex MIPS options
10554 Generate big-endian code.
10558 Generate little-endian code. This is the default for @samp{mips*el-*-*}
10561 @item -march=@var{arch}
10563 Generate code that will run on @var{arch}, which can be the name of a
10564 generic MIPS ISA, or the name of a particular processor.
10566 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
10567 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
10568 The processor names are:
10569 @samp{4kc}, @samp{4km}, @samp{4kp},
10570 @samp{5kc}, @samp{5kf},
10572 @samp{24k}, @samp{24kc}, @samp{24kf}, @samp{24kx},
10575 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
10576 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
10577 @samp{rm7000}, @samp{rm9000},
10580 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
10581 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
10582 The special value @samp{from-abi} selects the
10583 most compatible architecture for the selected ABI (that is,
10584 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
10586 In processor names, a final @samp{000} can be abbreviated as @samp{k}
10587 (for example, @samp{-march=r2k}). Prefixes are optional, and
10588 @samp{vr} may be written @samp{r}.
10590 GCC defines two macros based on the value of this option. The first
10591 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
10592 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
10593 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
10594 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
10595 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
10597 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
10598 above. In other words, it will have the full prefix and will not
10599 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
10600 the macro names the resolved architecture (either @samp{"mips1"} or
10601 @samp{"mips3"}). It names the default architecture when no
10602 @option{-march} option is given.
10604 @item -mtune=@var{arch}
10606 Optimize for @var{arch}. Among other things, this option controls
10607 the way instructions are scheduled, and the perceived cost of arithmetic
10608 operations. The list of @var{arch} values is the same as for
10611 When this option is not used, GCC will optimize for the processor
10612 specified by @option{-march}. By using @option{-march} and
10613 @option{-mtune} together, it is possible to generate code that will
10614 run on a family of processors, but optimize the code for one
10615 particular member of that family.
10617 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
10618 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
10619 @samp{-march} ones described above.
10623 Equivalent to @samp{-march=mips1}.
10627 Equivalent to @samp{-march=mips2}.
10631 Equivalent to @samp{-march=mips3}.
10635 Equivalent to @samp{-march=mips4}.
10639 Equivalent to @samp{-march=mips32}.
10643 Equivalent to @samp{-march=mips32r2}.
10647 Equivalent to @samp{-march=mips64}.
10652 @opindex mno-mips16
10653 Generate (do not generate) MIPS16 code. If GCC is targetting a
10654 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
10666 Generate code for the given ABI@.
10668 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
10669 generates 64-bit code when you select a 64-bit architecture, but you
10670 can use @option{-mgp32} to get 32-bit code instead.
10672 For information about the O64 ABI, see
10673 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
10676 @itemx -mno-abicalls
10678 @opindex mno-abicalls
10679 Generate (do not generate) code that is suitable for SVR4-style
10680 dynamic objects. @option{-mabicalls} is the default for SVR4-based
10685 Generate (do not generate) code that is fully position-independent,
10686 and that can therefore be linked into shared libraries. This option
10687 only affects @option{-mabicalls}.
10689 All @option{-mabicalls} code has traditionally been position-independent,
10690 regardless of options like @option{-fPIC} and @option{-fpic}. However,
10691 as an extension, the GNU toolchain allows executables to use absolute
10692 accesses for locally-binding symbols. It can also use shorter GP
10693 initialization sequences and generate direct calls to locally-defined
10694 functions. This mode is selected by @option{-mno-shared}.
10696 @option{-mno-shared} depends on binutils 2.16 or higher and generates
10697 objects that can only be linked by the GNU linker. However, the option
10698 does not affect the ABI of the final executable; it only affects the ABI
10699 of relocatable objects. Using @option{-mno-shared} will generally make
10700 executables both smaller and quicker.
10702 @option{-mshared} is the default.
10708 Lift (do not lift) the usual restrictions on the size of the global
10711 GCC normally uses a single instruction to load values from the GOT@.
10712 While this is relatively efficient, it will only work if the GOT
10713 is smaller than about 64k. Anything larger will cause the linker
10714 to report an error such as:
10716 @cindex relocation truncated to fit (MIPS)
10718 relocation truncated to fit: R_MIPS_GOT16 foobar
10721 If this happens, you should recompile your code with @option{-mxgot}.
10722 It should then work with very large GOTs, although it will also be
10723 less efficient, since it will take three instructions to fetch the
10724 value of a global symbol.
10726 Note that some linkers can create multiple GOTs. If you have such a
10727 linker, you should only need to use @option{-mxgot} when a single object
10728 file accesses more than 64k's worth of GOT entries. Very few do.
10730 These options have no effect unless GCC is generating position
10735 Assume that general-purpose registers are 32 bits wide.
10739 Assume that general-purpose registers are 64 bits wide.
10743 Assume that floating-point registers are 32 bits wide.
10747 Assume that floating-point registers are 64 bits wide.
10750 @opindex mhard-float
10751 Use floating-point coprocessor instructions.
10754 @opindex msoft-float
10755 Do not use floating-point coprocessor instructions. Implement
10756 floating-point calculations using library calls instead.
10758 @item -msingle-float
10759 @opindex msingle-float
10760 Assume that the floating-point coprocessor only supports single-precision
10763 @itemx -mdouble-float
10764 @opindex mdouble-float
10765 Assume that the floating-point coprocessor supports double-precision
10766 operations. This is the default.
10772 Use (do not use) the MIPS DSP ASE. @xref{MIPS DSP Built-in Functions}.
10774 @itemx -mpaired-single
10775 @itemx -mno-paired-single
10776 @opindex mpaired-single
10777 @opindex mno-paired-single
10778 Use (do not use) paired-single floating-point instructions.
10779 @xref{MIPS Paired-Single Support}. This option can only be used
10780 when generating 64-bit code and requires hardware floating-point
10781 support to be enabled.
10786 @opindex mno-mips3d
10787 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
10788 The option @option{-mips3d} implies @option{-mpaired-single}.
10792 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
10793 an explanation of the default and the way that the pointer size is
10798 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
10800 The default size of @code{int}s, @code{long}s and pointers depends on
10801 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
10802 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
10803 32-bit @code{long}s. Pointers are the same size as @code{long}s,
10804 or the same size as integer registers, whichever is smaller.
10810 Assume (do not assume) that all symbols have 32-bit values, regardless
10811 of the selected ABI@. This option is useful in combination with
10812 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
10813 to generate shorter and faster references to symbolic addresses.
10817 @cindex smaller data references (MIPS)
10818 @cindex gp-relative references (MIPS)
10819 Put global and static items less than or equal to @var{num} bytes into
10820 the small data or bss section instead of the normal data or bss section.
10821 This allows the data to be accessed using a single instruction.
10823 All modules should be compiled with the same @option{-G @var{num}}
10826 @item -membedded-data
10827 @itemx -mno-embedded-data
10828 @opindex membedded-data
10829 @opindex mno-embedded-data
10830 Allocate variables to the read-only data section first if possible, then
10831 next in the small data section if possible, otherwise in data. This gives
10832 slightly slower code than the default, but reduces the amount of RAM required
10833 when executing, and thus may be preferred for some embedded systems.
10835 @item -muninit-const-in-rodata
10836 @itemx -mno-uninit-const-in-rodata
10837 @opindex muninit-const-in-rodata
10838 @opindex mno-uninit-const-in-rodata
10839 Put uninitialized @code{const} variables in the read-only data section.
10840 This option is only meaningful in conjunction with @option{-membedded-data}.
10842 @item -msplit-addresses
10843 @itemx -mno-split-addresses
10844 @opindex msplit-addresses
10845 @opindex mno-split-addresses
10846 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
10847 relocation operators. This option has been superseded by
10848 @option{-mexplicit-relocs} but is retained for backwards compatibility.
10850 @item -mexplicit-relocs
10851 @itemx -mno-explicit-relocs
10852 @opindex mexplicit-relocs
10853 @opindex mno-explicit-relocs
10854 Use (do not use) assembler relocation operators when dealing with symbolic
10855 addresses. The alternative, selected by @option{-mno-explicit-relocs},
10856 is to use assembler macros instead.
10858 @option{-mexplicit-relocs} is the default if GCC was configured
10859 to use an assembler that supports relocation operators.
10861 @item -mcheck-zero-division
10862 @itemx -mno-check-zero-division
10863 @opindex mcheck-zero-division
10864 @opindex mno-check-zero-division
10865 Trap (do not trap) on integer division by zero. The default is
10866 @option{-mcheck-zero-division}.
10868 @item -mdivide-traps
10869 @itemx -mdivide-breaks
10870 @opindex mdivide-traps
10871 @opindex mdivide-breaks
10872 MIPS systems check for division by zero by generating either a
10873 conditional trap or a break instruction. Using traps results in
10874 smaller code, but is only supported on MIPS II and later. Also, some
10875 versions of the Linux kernel have a bug that prevents trap from
10876 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
10877 allow conditional traps on architectures that support them and
10878 @option{-mdivide-breaks} to force the use of breaks.
10880 The default is usually @option{-mdivide-traps}, but this can be
10881 overridden at configure time using @option{--with-divide=breaks}.
10882 Divide-by-zero checks can be completely disabled using
10883 @option{-mno-check-zero-division}.
10888 @opindex mno-memcpy
10889 Force (do not force) the use of @code{memcpy()} for non-trivial block
10890 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
10891 most constant-sized copies.
10894 @itemx -mno-long-calls
10895 @opindex mlong-calls
10896 @opindex mno-long-calls
10897 Disable (do not disable) use of the @code{jal} instruction. Calling
10898 functions using @code{jal} is more efficient but requires the caller
10899 and callee to be in the same 256 megabyte segment.
10901 This option has no effect on abicalls code. The default is
10902 @option{-mno-long-calls}.
10908 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
10909 instructions, as provided by the R4650 ISA@.
10912 @itemx -mno-fused-madd
10913 @opindex mfused-madd
10914 @opindex mno-fused-madd
10915 Enable (disable) use of the floating point multiply-accumulate
10916 instructions, when they are available. The default is
10917 @option{-mfused-madd}.
10919 When multiply-accumulate instructions are used, the intermediate
10920 product is calculated to infinite precision and is not subject to
10921 the FCSR Flush to Zero bit. This may be undesirable in some
10926 Tell the MIPS assembler to not run its preprocessor over user
10927 assembler files (with a @samp{.s} suffix) when assembling them.
10930 @itemx -mno-fix-r4000
10931 @opindex mfix-r4000
10932 @opindex mno-fix-r4000
10933 Work around certain R4000 CPU errata:
10936 A double-word or a variable shift may give an incorrect result if executed
10937 immediately after starting an integer division.
10939 A double-word or a variable shift may give an incorrect result if executed
10940 while an integer multiplication is in progress.
10942 An integer division may give an incorrect result if started in a delay slot
10943 of a taken branch or a jump.
10947 @itemx -mno-fix-r4400
10948 @opindex mfix-r4400
10949 @opindex mno-fix-r4400
10950 Work around certain R4400 CPU errata:
10953 A double-word or a variable shift may give an incorrect result if executed
10954 immediately after starting an integer division.
10958 @itemx -mno-fix-vr4120
10959 @opindex mfix-vr4120
10960 Work around certain VR4120 errata:
10963 @code{dmultu} does not always produce the correct result.
10965 @code{div} and @code{ddiv} do not always produce the correct result if one
10966 of the operands is negative.
10968 The workarounds for the division errata rely on special functions in
10969 @file{libgcc.a}. At present, these functions are only provided by
10970 the @code{mips64vr*-elf} configurations.
10972 Other VR4120 errata require a nop to be inserted between certain pairs of
10973 instructions. These errata are handled by the assembler, not by GCC itself.
10976 @opindex mfix-vr4130
10977 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
10978 workarounds are implemented by the assembler rather than by GCC,
10979 although GCC will avoid using @code{mflo} and @code{mfhi} if the
10980 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
10981 instructions are available instead.
10984 @itemx -mno-fix-sb1
10986 Work around certain SB-1 CPU core errata.
10987 (This flag currently works around the SB-1 revision 2
10988 ``F1'' and ``F2'' floating point errata.)
10990 @item -mflush-func=@var{func}
10991 @itemx -mno-flush-func
10992 @opindex mflush-func
10993 Specifies the function to call to flush the I and D caches, or to not
10994 call any such function. If called, the function must take the same
10995 arguments as the common @code{_flush_func()}, that is, the address of the
10996 memory range for which the cache is being flushed, the size of the
10997 memory range, and the number 3 (to flush both caches). The default
10998 depends on the target GCC was configured for, but commonly is either
10999 @samp{_flush_func} or @samp{__cpu_flush}.
11001 @item -mbranch-likely
11002 @itemx -mno-branch-likely
11003 @opindex mbranch-likely
11004 @opindex mno-branch-likely
11005 Enable or disable use of Branch Likely instructions, regardless of the
11006 default for the selected architecture. By default, Branch Likely
11007 instructions may be generated if they are supported by the selected
11008 architecture. An exception is for the MIPS32 and MIPS64 architectures
11009 and processors which implement those architectures; for those, Branch
11010 Likely instructions will not be generated by default because the MIPS32
11011 and MIPS64 architectures specifically deprecate their use.
11013 @item -mfp-exceptions
11014 @itemx -mno-fp-exceptions
11015 @opindex mfp-exceptions
11016 Specifies whether FP exceptions are enabled. This affects how we schedule
11017 FP instructions for some processors. The default is that FP exceptions are
11020 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
11021 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
11024 @item -mvr4130-align
11025 @itemx -mno-vr4130-align
11026 @opindex mvr4130-align
11027 The VR4130 pipeline is two-way superscalar, but can only issue two
11028 instructions together if the first one is 8-byte aligned. When this
11029 option is enabled, GCC will align pairs of instructions that it
11030 thinks should execute in parallel.
11032 This option only has an effect when optimizing for the VR4130.
11033 It normally makes code faster, but at the expense of making it bigger.
11034 It is enabled by default at optimization level @option{-O3}.
11038 @subsection MMIX Options
11039 @cindex MMIX Options
11041 These options are defined for the MMIX:
11045 @itemx -mno-libfuncs
11047 @opindex mno-libfuncs
11048 Specify that intrinsic library functions are being compiled, passing all
11049 values in registers, no matter the size.
11052 @itemx -mno-epsilon
11054 @opindex mno-epsilon
11055 Generate floating-point comparison instructions that compare with respect
11056 to the @code{rE} epsilon register.
11058 @item -mabi=mmixware
11060 @opindex mabi-mmixware
11062 Generate code that passes function parameters and return values that (in
11063 the called function) are seen as registers @code{$0} and up, as opposed to
11064 the GNU ABI which uses global registers @code{$231} and up.
11066 @item -mzero-extend
11067 @itemx -mno-zero-extend
11068 @opindex mzero-extend
11069 @opindex mno-zero-extend
11070 When reading data from memory in sizes shorter than 64 bits, use (do not
11071 use) zero-extending load instructions by default, rather than
11072 sign-extending ones.
11075 @itemx -mno-knuthdiv
11077 @opindex mno-knuthdiv
11078 Make the result of a division yielding a remainder have the same sign as
11079 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
11080 remainder follows the sign of the dividend. Both methods are
11081 arithmetically valid, the latter being almost exclusively used.
11083 @item -mtoplevel-symbols
11084 @itemx -mno-toplevel-symbols
11085 @opindex mtoplevel-symbols
11086 @opindex mno-toplevel-symbols
11087 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
11088 code can be used with the @code{PREFIX} assembly directive.
11092 Generate an executable in the ELF format, rather than the default
11093 @samp{mmo} format used by the @command{mmix} simulator.
11095 @item -mbranch-predict
11096 @itemx -mno-branch-predict
11097 @opindex mbranch-predict
11098 @opindex mno-branch-predict
11099 Use (do not use) the probable-branch instructions, when static branch
11100 prediction indicates a probable branch.
11102 @item -mbase-addresses
11103 @itemx -mno-base-addresses
11104 @opindex mbase-addresses
11105 @opindex mno-base-addresses
11106 Generate (do not generate) code that uses @emph{base addresses}. Using a
11107 base address automatically generates a request (handled by the assembler
11108 and the linker) for a constant to be set up in a global register. The
11109 register is used for one or more base address requests within the range 0
11110 to 255 from the value held in the register. The generally leads to short
11111 and fast code, but the number of different data items that can be
11112 addressed is limited. This means that a program that uses lots of static
11113 data may require @option{-mno-base-addresses}.
11115 @item -msingle-exit
11116 @itemx -mno-single-exit
11117 @opindex msingle-exit
11118 @opindex mno-single-exit
11119 Force (do not force) generated code to have a single exit point in each
11123 @node MN10300 Options
11124 @subsection MN10300 Options
11125 @cindex MN10300 options
11127 These @option{-m} options are defined for Matsushita MN10300 architectures:
11132 Generate code to avoid bugs in the multiply instructions for the MN10300
11133 processors. This is the default.
11135 @item -mno-mult-bug
11136 @opindex mno-mult-bug
11137 Do not generate code to avoid bugs in the multiply instructions for the
11138 MN10300 processors.
11142 Generate code which uses features specific to the AM33 processor.
11146 Do not generate code which uses features specific to the AM33 processor. This
11149 @item -mreturn-pointer-on-d0
11150 @opindex mreturn-pointer-on-d0
11151 When generating a function which returns a pointer, return the pointer
11152 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
11153 only in a0, and attempts to call such functions without a prototype
11154 would result in errors. Note that this option is on by default; use
11155 @option{-mno-return-pointer-on-d0} to disable it.
11159 Do not link in the C run-time initialization object file.
11163 Indicate to the linker that it should perform a relaxation optimization pass
11164 to shorten branches, calls and absolute memory addresses. This option only
11165 has an effect when used on the command line for the final link step.
11167 This option makes symbolic debugging impossible.
11171 @subsection MT Options
11174 These @option{-m} options are defined for Morpho MT architectures:
11178 @item -march=@var{cpu-type}
11180 Generate code that will run on @var{cpu-type}, which is the name of a system
11181 representing a certain processor type. Possible values for
11182 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
11183 @samp{ms1-16-003} and @samp{ms2}.
11185 When this option is not used, the default is @option{-march=ms1-16-002}.
11189 Use byte loads and stores when generating code.
11193 Do not use byte loads and stores when generating code.
11197 Use simulator runtime
11201 Do not link in the C run-time initialization object file
11202 @file{crti.o}. Other run-time initialization and termination files
11203 such as @file{startup.o} and @file{exit.o} are still included on the
11204 linker command line.
11208 @node PDP-11 Options
11209 @subsection PDP-11 Options
11210 @cindex PDP-11 Options
11212 These options are defined for the PDP-11:
11217 Use hardware FPP floating point. This is the default. (FIS floating
11218 point on the PDP-11/40 is not supported.)
11221 @opindex msoft-float
11222 Do not use hardware floating point.
11226 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
11230 Return floating-point results in memory. This is the default.
11234 Generate code for a PDP-11/40.
11238 Generate code for a PDP-11/45. This is the default.
11242 Generate code for a PDP-11/10.
11244 @item -mbcopy-builtin
11245 @opindex bcopy-builtin
11246 Use inline @code{movmemhi} patterns for copying memory. This is the
11251 Do not use inline @code{movmemhi} patterns for copying memory.
11257 Use 16-bit @code{int}. This is the default.
11263 Use 32-bit @code{int}.
11266 @itemx -mno-float32
11268 @opindex mno-float32
11269 Use 64-bit @code{float}. This is the default.
11272 @itemx -mno-float64
11274 @opindex mno-float64
11275 Use 32-bit @code{float}.
11279 Use @code{abshi2} pattern. This is the default.
11283 Do not use @code{abshi2} pattern.
11285 @item -mbranch-expensive
11286 @opindex mbranch-expensive
11287 Pretend that branches are expensive. This is for experimenting with
11288 code generation only.
11290 @item -mbranch-cheap
11291 @opindex mbranch-cheap
11292 Do not pretend that branches are expensive. This is the default.
11296 Generate code for a system with split I&D@.
11300 Generate code for a system without split I&D@. This is the default.
11304 Use Unix assembler syntax. This is the default when configured for
11305 @samp{pdp11-*-bsd}.
11309 Use DEC assembler syntax. This is the default when configured for any
11310 PDP-11 target other than @samp{pdp11-*-bsd}.
11313 @node PowerPC Options
11314 @subsection PowerPC Options
11315 @cindex PowerPC options
11317 These are listed under @xref{RS/6000 and PowerPC Options}.
11319 @node RS/6000 and PowerPC Options
11320 @subsection IBM RS/6000 and PowerPC Options
11321 @cindex RS/6000 and PowerPC Options
11322 @cindex IBM RS/6000 and PowerPC Options
11324 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
11331 @itemx -mno-powerpc
11332 @itemx -mpowerpc-gpopt
11333 @itemx -mno-powerpc-gpopt
11334 @itemx -mpowerpc-gfxopt
11335 @itemx -mno-powerpc-gfxopt
11337 @itemx -mno-powerpc64
11341 @itemx -mno-popcntb
11347 @opindex mno-power2
11349 @opindex mno-powerpc
11350 @opindex mpowerpc-gpopt
11351 @opindex mno-powerpc-gpopt
11352 @opindex mpowerpc-gfxopt
11353 @opindex mno-powerpc-gfxopt
11354 @opindex mpowerpc64
11355 @opindex mno-powerpc64
11359 @opindex mno-popcntb
11362 GCC supports two related instruction set architectures for the
11363 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
11364 instructions supported by the @samp{rios} chip set used in the original
11365 RS/6000 systems and the @dfn{PowerPC} instruction set is the
11366 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
11367 the IBM 4xx, 6xx, and follow-on microprocessors.
11369 Neither architecture is a subset of the other. However there is a
11370 large common subset of instructions supported by both. An MQ
11371 register is included in processors supporting the POWER architecture.
11373 You use these options to specify which instructions are available on the
11374 processor you are using. The default value of these options is
11375 determined when configuring GCC@. Specifying the
11376 @option{-mcpu=@var{cpu_type}} overrides the specification of these
11377 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
11378 rather than the options listed above.
11380 The @option{-mpower} option allows GCC to generate instructions that
11381 are found only in the POWER architecture and to use the MQ register.
11382 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
11383 to generate instructions that are present in the POWER2 architecture but
11384 not the original POWER architecture.
11386 The @option{-mpowerpc} option allows GCC to generate instructions that
11387 are found only in the 32-bit subset of the PowerPC architecture.
11388 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
11389 GCC to use the optional PowerPC architecture instructions in the
11390 General Purpose group, including floating-point square root. Specifying
11391 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
11392 use the optional PowerPC architecture instructions in the Graphics
11393 group, including floating-point select.
11395 The @option{-mmfcrf} option allows GCC to generate the move from
11396 condition register field instruction implemented on the POWER4
11397 processor and other processors that support the PowerPC V2.01
11399 The @option{-mpopcntb} option allows GCC to generate the popcount and
11400 double precision FP reciprocal estimate instruction implemented on the
11401 POWER5 processor and other processors that support the PowerPC V2.02
11403 The @option{-mfprnd} option allows GCC to generate the FP round to
11404 integer instructions implemented on the POWER5+ processor and other
11405 processors that support the PowerPC V2.03 architecture.
11407 The @option{-mpowerpc64} option allows GCC to generate the additional
11408 64-bit instructions that are found in the full PowerPC64 architecture
11409 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
11410 @option{-mno-powerpc64}.
11412 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
11413 will use only the instructions in the common subset of both
11414 architectures plus some special AIX common-mode calls, and will not use
11415 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
11416 permits GCC to use any instruction from either architecture and to
11417 allow use of the MQ register; specify this for the Motorola MPC601.
11419 @item -mnew-mnemonics
11420 @itemx -mold-mnemonics
11421 @opindex mnew-mnemonics
11422 @opindex mold-mnemonics
11423 Select which mnemonics to use in the generated assembler code. With
11424 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
11425 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
11426 assembler mnemonics defined for the POWER architecture. Instructions
11427 defined in only one architecture have only one mnemonic; GCC uses that
11428 mnemonic irrespective of which of these options is specified.
11430 GCC defaults to the mnemonics appropriate for the architecture in
11431 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
11432 value of these option. Unless you are building a cross-compiler, you
11433 should normally not specify either @option{-mnew-mnemonics} or
11434 @option{-mold-mnemonics}, but should instead accept the default.
11436 @item -mcpu=@var{cpu_type}
11438 Set architecture type, register usage, choice of mnemonics, and
11439 instruction scheduling parameters for machine type @var{cpu_type}.
11440 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
11441 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
11442 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
11443 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
11444 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
11445 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
11446 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
11447 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6},
11448 @samp{common}, @samp{powerpc}, @samp{powerpc64},
11449 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
11451 @option{-mcpu=common} selects a completely generic processor. Code
11452 generated under this option will run on any POWER or PowerPC processor.
11453 GCC will use only the instructions in the common subset of both
11454 architectures, and will not use the MQ register. GCC assumes a generic
11455 processor model for scheduling purposes.
11457 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
11458 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
11459 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
11460 types, with an appropriate, generic processor model assumed for
11461 scheduling purposes.
11463 The other options specify a specific processor. Code generated under
11464 those options will run best on that processor, and may not run at all on
11467 The @option{-mcpu} options automatically enable or disable the
11468 following options: @option{-maltivec}, @option{-mfprnd},
11469 @option{-mhard-float}, @option{-mmfcrf}, @option{-mmultiple},
11470 @option{-mnew-mnemonics}, @option{-mpopcntb}, @option{-mpower},
11471 @option{-mpower2}, @option{-mpowerpc64}, @option{-mpowerpc-gpopt},
11472 @option{-mpowerpc-gfxopt}, @option{-mstring}, @option{-mmulhw}, @option{-mdlmzb}.
11473 The particular options
11474 set for any particular CPU will vary between compiler versions,
11475 depending on what setting seems to produce optimal code for that CPU;
11476 it doesn't necessarily reflect the actual hardware's capabilities. If
11477 you wish to set an individual option to a particular value, you may
11478 specify it after the @option{-mcpu} option, like @samp{-mcpu=970
11481 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
11482 not enabled or disabled by the @option{-mcpu} option at present because
11483 AIX does not have full support for these options. You may still
11484 enable or disable them individually if you're sure it'll work in your
11487 @item -mtune=@var{cpu_type}
11489 Set the instruction scheduling parameters for machine type
11490 @var{cpu_type}, but do not set the architecture type, register usage, or
11491 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
11492 values for @var{cpu_type} are used for @option{-mtune} as for
11493 @option{-mcpu}. If both are specified, the code generated will use the
11494 architecture, registers, and mnemonics set by @option{-mcpu}, but the
11495 scheduling parameters set by @option{-mtune}.
11501 Generate code to compute division as reciprocal estimate and iterative
11502 refinement, creating opportunities for increased throughput. This
11503 feature requires: optional PowerPC Graphics instruction set for single
11504 precision and FRE instruction for double precision, assuming divides
11505 cannot generate user-visible traps, and the domain values not include
11506 Infinities, denormals or zero denominator.
11509 @itemx -mno-altivec
11511 @opindex mno-altivec
11512 Generate code that uses (does not use) AltiVec instructions, and also
11513 enable the use of built-in functions that allow more direct access to
11514 the AltiVec instruction set. You may also need to set
11515 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
11521 @opindex mno-vrsave
11522 Generate VRSAVE instructions when generating AltiVec code.
11525 @opindex msecure-plt
11526 Generate code that allows ld and ld.so to build executables and shared
11527 libraries with non-exec .plt and .got sections. This is a PowerPC
11528 32-bit SYSV ABI option.
11532 Generate code that uses a BSS .plt section that ld.so fills in, and
11533 requires .plt and .got sections that are both writable and executable.
11534 This is a PowerPC 32-bit SYSV ABI option.
11540 This switch enables or disables the generation of ISEL instructions.
11542 @item -misel=@var{yes/no}
11543 This switch has been deprecated. Use @option{-misel} and
11544 @option{-mno-isel} instead.
11550 This switch enables or disables the generation of SPE simd
11553 @item -mspe=@var{yes/no}
11554 This option has been deprecated. Use @option{-mspe} and
11555 @option{-mno-spe} instead.
11557 @item -mfloat-gprs=@var{yes/single/double/no}
11558 @itemx -mfloat-gprs
11559 @opindex mfloat-gprs
11560 This switch enables or disables the generation of floating point
11561 operations on the general purpose registers for architectures that
11564 The argument @var{yes} or @var{single} enables the use of
11565 single-precision floating point operations.
11567 The argument @var{double} enables the use of single and
11568 double-precision floating point operations.
11570 The argument @var{no} disables floating point operations on the
11571 general purpose registers.
11573 This option is currently only available on the MPC854x.
11579 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
11580 targets (including GNU/Linux). The 32-bit environment sets int, long
11581 and pointer to 32 bits and generates code that runs on any PowerPC
11582 variant. The 64-bit environment sets int to 32 bits and long and
11583 pointer to 64 bits, and generates code for PowerPC64, as for
11584 @option{-mpowerpc64}.
11587 @itemx -mno-fp-in-toc
11588 @itemx -mno-sum-in-toc
11589 @itemx -mminimal-toc
11591 @opindex mno-fp-in-toc
11592 @opindex mno-sum-in-toc
11593 @opindex mminimal-toc
11594 Modify generation of the TOC (Table Of Contents), which is created for
11595 every executable file. The @option{-mfull-toc} option is selected by
11596 default. In that case, GCC will allocate at least one TOC entry for
11597 each unique non-automatic variable reference in your program. GCC
11598 will also place floating-point constants in the TOC@. However, only
11599 16,384 entries are available in the TOC@.
11601 If you receive a linker error message that saying you have overflowed
11602 the available TOC space, you can reduce the amount of TOC space used
11603 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
11604 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
11605 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
11606 generate code to calculate the sum of an address and a constant at
11607 run-time instead of putting that sum into the TOC@. You may specify one
11608 or both of these options. Each causes GCC to produce very slightly
11609 slower and larger code at the expense of conserving TOC space.
11611 If you still run out of space in the TOC even when you specify both of
11612 these options, specify @option{-mminimal-toc} instead. This option causes
11613 GCC to make only one TOC entry for every file. When you specify this
11614 option, GCC will produce code that is slower and larger but which
11615 uses extremely little TOC space. You may wish to use this option
11616 only on files that contain less frequently executed code.
11622 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
11623 @code{long} type, and the infrastructure needed to support them.
11624 Specifying @option{-maix64} implies @option{-mpowerpc64} and
11625 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
11626 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
11629 @itemx -mno-xl-compat
11630 @opindex mxl-compat
11631 @opindex mno-xl-compat
11632 Produce code that conforms more closely to IBM XL compiler semantics
11633 when using AIX-compatible ABI. Pass floating-point arguments to
11634 prototyped functions beyond the register save area (RSA) on the stack
11635 in addition to argument FPRs. Do not assume that most significant
11636 double in 128-bit long double value is properly rounded when comparing
11637 values and converting to double. Use XL symbol names for long double
11640 The AIX calling convention was extended but not initially documented to
11641 handle an obscure K&R C case of calling a function that takes the
11642 address of its arguments with fewer arguments than declared. IBM XL
11643 compilers access floating point arguments which do not fit in the
11644 RSA from the stack when a subroutine is compiled without
11645 optimization. Because always storing floating-point arguments on the
11646 stack is inefficient and rarely needed, this option is not enabled by
11647 default and only is necessary when calling subroutines compiled by IBM
11648 XL compilers without optimization.
11652 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
11653 application written to use message passing with special startup code to
11654 enable the application to run. The system must have PE installed in the
11655 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
11656 must be overridden with the @option{-specs=} option to specify the
11657 appropriate directory location. The Parallel Environment does not
11658 support threads, so the @option{-mpe} option and the @option{-pthread}
11659 option are incompatible.
11661 @item -malign-natural
11662 @itemx -malign-power
11663 @opindex malign-natural
11664 @opindex malign-power
11665 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
11666 @option{-malign-natural} overrides the ABI-defined alignment of larger
11667 types, such as floating-point doubles, on their natural size-based boundary.
11668 The option @option{-malign-power} instructs GCC to follow the ABI-specified
11669 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
11671 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
11675 @itemx -mhard-float
11676 @opindex msoft-float
11677 @opindex mhard-float
11678 Generate code that does not use (uses) the floating-point register set.
11679 Software floating point emulation is provided if you use the
11680 @option{-msoft-float} option, and pass the option to GCC when linking.
11683 @itemx -mno-multiple
11685 @opindex mno-multiple
11686 Generate code that uses (does not use) the load multiple word
11687 instructions and the store multiple word instructions. These
11688 instructions are generated by default on POWER systems, and not
11689 generated on PowerPC systems. Do not use @option{-mmultiple} on little
11690 endian PowerPC systems, since those instructions do not work when the
11691 processor is in little endian mode. The exceptions are PPC740 and
11692 PPC750 which permit the instructions usage in little endian mode.
11697 @opindex mno-string
11698 Generate code that uses (does not use) the load string instructions
11699 and the store string word instructions to save multiple registers and
11700 do small block moves. These instructions are generated by default on
11701 POWER systems, and not generated on PowerPC systems. Do not use
11702 @option{-mstring} on little endian PowerPC systems, since those
11703 instructions do not work when the processor is in little endian mode.
11704 The exceptions are PPC740 and PPC750 which permit the instructions
11705 usage in little endian mode.
11710 @opindex mno-update
11711 Generate code that uses (does not use) the load or store instructions
11712 that update the base register to the address of the calculated memory
11713 location. These instructions are generated by default. If you use
11714 @option{-mno-update}, there is a small window between the time that the
11715 stack pointer is updated and the address of the previous frame is
11716 stored, which means code that walks the stack frame across interrupts or
11717 signals may get corrupted data.
11720 @itemx -mno-fused-madd
11721 @opindex mfused-madd
11722 @opindex mno-fused-madd
11723 Generate code that uses (does not use) the floating point multiply and
11724 accumulate instructions. These instructions are generated by default if
11725 hardware floating is used.
11731 Generate code that uses (does not use) the half-word multiply and
11732 multiply-accumulate instructions on the IBM 405 and 440 processors.
11733 These instructions are generated by default when targetting those
11740 Generate code that uses (does not use) the string-search @samp{dlmzb}
11741 instruction on the IBM 405 and 440 processors. This instruction is
11742 generated by default when targetting those processors.
11744 @item -mno-bit-align
11746 @opindex mno-bit-align
11747 @opindex mbit-align
11748 On System V.4 and embedded PowerPC systems do not (do) force structures
11749 and unions that contain bit-fields to be aligned to the base type of the
11752 For example, by default a structure containing nothing but 8
11753 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
11754 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
11755 the structure would be aligned to a 1 byte boundary and be one byte in
11758 @item -mno-strict-align
11759 @itemx -mstrict-align
11760 @opindex mno-strict-align
11761 @opindex mstrict-align
11762 On System V.4 and embedded PowerPC systems do not (do) assume that
11763 unaligned memory references will be handled by the system.
11765 @item -mrelocatable
11766 @itemx -mno-relocatable
11767 @opindex mrelocatable
11768 @opindex mno-relocatable
11769 On embedded PowerPC systems generate code that allows (does not allow)
11770 the program to be relocated to a different address at runtime. If you
11771 use @option{-mrelocatable} on any module, all objects linked together must
11772 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
11774 @item -mrelocatable-lib
11775 @itemx -mno-relocatable-lib
11776 @opindex mrelocatable-lib
11777 @opindex mno-relocatable-lib
11778 On embedded PowerPC systems generate code that allows (does not allow)
11779 the program to be relocated to a different address at runtime. Modules
11780 compiled with @option{-mrelocatable-lib} can be linked with either modules
11781 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
11782 with modules compiled with the @option{-mrelocatable} options.
11788 On System V.4 and embedded PowerPC systems do not (do) assume that
11789 register 2 contains a pointer to a global area pointing to the addresses
11790 used in the program.
11793 @itemx -mlittle-endian
11795 @opindex mlittle-endian
11796 On System V.4 and embedded PowerPC systems compile code for the
11797 processor in little endian mode. The @option{-mlittle-endian} option is
11798 the same as @option{-mlittle}.
11801 @itemx -mbig-endian
11803 @opindex mbig-endian
11804 On System V.4 and embedded PowerPC systems compile code for the
11805 processor in big endian mode. The @option{-mbig-endian} option is
11806 the same as @option{-mbig}.
11808 @item -mdynamic-no-pic
11809 @opindex mdynamic-no-pic
11810 On Darwin and Mac OS X systems, compile code so that it is not
11811 relocatable, but that its external references are relocatable. The
11812 resulting code is suitable for applications, but not shared
11815 @item -mprioritize-restricted-insns=@var{priority}
11816 @opindex mprioritize-restricted-insns
11817 This option controls the priority that is assigned to
11818 dispatch-slot restricted instructions during the second scheduling
11819 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
11820 @var{no/highest/second-highest} priority to dispatch slot restricted
11823 @item -msched-costly-dep=@var{dependence_type}
11824 @opindex msched-costly-dep
11825 This option controls which dependences are considered costly
11826 by the target during instruction scheduling. The argument
11827 @var{dependence_type} takes one of the following values:
11828 @var{no}: no dependence is costly,
11829 @var{all}: all dependences are costly,
11830 @var{true_store_to_load}: a true dependence from store to load is costly,
11831 @var{store_to_load}: any dependence from store to load is costly,
11832 @var{number}: any dependence which latency >= @var{number} is costly.
11834 @item -minsert-sched-nops=@var{scheme}
11835 @opindex minsert-sched-nops
11836 This option controls which nop insertion scheme will be used during
11837 the second scheduling pass. The argument @var{scheme} takes one of the
11839 @var{no}: Don't insert nops.
11840 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
11841 according to the scheduler's grouping.
11842 @var{regroup_exact}: Insert nops to force costly dependent insns into
11843 separate groups. Insert exactly as many nops as needed to force an insn
11844 to a new group, according to the estimated processor grouping.
11845 @var{number}: Insert nops to force costly dependent insns into
11846 separate groups. Insert @var{number} nops to force an insn to a new group.
11849 @opindex mcall-sysv
11850 On System V.4 and embedded PowerPC systems compile code using calling
11851 conventions that adheres to the March 1995 draft of the System V
11852 Application Binary Interface, PowerPC processor supplement. This is the
11853 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
11855 @item -mcall-sysv-eabi
11856 @opindex mcall-sysv-eabi
11857 Specify both @option{-mcall-sysv} and @option{-meabi} options.
11859 @item -mcall-sysv-noeabi
11860 @opindex mcall-sysv-noeabi
11861 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
11863 @item -mcall-solaris
11864 @opindex mcall-solaris
11865 On System V.4 and embedded PowerPC systems compile code for the Solaris
11869 @opindex mcall-linux
11870 On System V.4 and embedded PowerPC systems compile code for the
11871 Linux-based GNU system.
11875 On System V.4 and embedded PowerPC systems compile code for the
11876 Hurd-based GNU system.
11878 @item -mcall-netbsd
11879 @opindex mcall-netbsd
11880 On System V.4 and embedded PowerPC systems compile code for the
11881 NetBSD operating system.
11883 @item -maix-struct-return
11884 @opindex maix-struct-return
11885 Return all structures in memory (as specified by the AIX ABI)@.
11887 @item -msvr4-struct-return
11888 @opindex msvr4-struct-return
11889 Return structures smaller than 8 bytes in registers (as specified by the
11892 @item -mabi=@var{abi-type}
11894 Extend the current ABI with a particular extension, or remove such extension.
11895 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
11896 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
11900 Extend the current ABI with SPE ABI extensions. This does not change
11901 the default ABI, instead it adds the SPE ABI extensions to the current
11905 @opindex mabi=no-spe
11906 Disable Booke SPE ABI extensions for the current ABI@.
11908 @item -mabi=ibmlongdouble
11909 @opindex mabi=ibmlongdouble
11910 Change the current ABI to use IBM extended precision long double.
11911 This is a PowerPC 32-bit SYSV ABI option.
11913 @item -mabi=ieeelongdouble
11914 @opindex mabi=ieeelongdouble
11915 Change the current ABI to use IEEE extended precision long double.
11916 This is a PowerPC 32-bit Linux ABI option.
11919 @itemx -mno-prototype
11920 @opindex mprototype
11921 @opindex mno-prototype
11922 On System V.4 and embedded PowerPC systems assume that all calls to
11923 variable argument functions are properly prototyped. Otherwise, the
11924 compiler must insert an instruction before every non prototyped call to
11925 set or clear bit 6 of the condition code register (@var{CR}) to
11926 indicate whether floating point values were passed in the floating point
11927 registers in case the function takes a variable arguments. With
11928 @option{-mprototype}, only calls to prototyped variable argument functions
11929 will set or clear the bit.
11933 On embedded PowerPC systems, assume that the startup module is called
11934 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
11935 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
11940 On embedded PowerPC systems, assume that the startup module is called
11941 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
11946 On embedded PowerPC systems, assume that the startup module is called
11947 @file{crt0.o} and the standard C libraries are @file{libads.a} and
11950 @item -myellowknife
11951 @opindex myellowknife
11952 On embedded PowerPC systems, assume that the startup module is called
11953 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
11958 On System V.4 and embedded PowerPC systems, specify that you are
11959 compiling for a VxWorks system.
11963 Specify that you are compiling for the WindISS simulation environment.
11967 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
11968 header to indicate that @samp{eabi} extended relocations are used.
11974 On System V.4 and embedded PowerPC systems do (do not) adhere to the
11975 Embedded Applications Binary Interface (eabi) which is a set of
11976 modifications to the System V.4 specifications. Selecting @option{-meabi}
11977 means that the stack is aligned to an 8 byte boundary, a function
11978 @code{__eabi} is called to from @code{main} to set up the eabi
11979 environment, and the @option{-msdata} option can use both @code{r2} and
11980 @code{r13} to point to two separate small data areas. Selecting
11981 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
11982 do not call an initialization function from @code{main}, and the
11983 @option{-msdata} option will only use @code{r13} to point to a single
11984 small data area. The @option{-meabi} option is on by default if you
11985 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
11988 @opindex msdata=eabi
11989 On System V.4 and embedded PowerPC systems, put small initialized
11990 @code{const} global and static data in the @samp{.sdata2} section, which
11991 is pointed to by register @code{r2}. Put small initialized
11992 non-@code{const} global and static data in the @samp{.sdata} section,
11993 which is pointed to by register @code{r13}. Put small uninitialized
11994 global and static data in the @samp{.sbss} section, which is adjacent to
11995 the @samp{.sdata} section. The @option{-msdata=eabi} option is
11996 incompatible with the @option{-mrelocatable} option. The
11997 @option{-msdata=eabi} option also sets the @option{-memb} option.
12000 @opindex msdata=sysv
12001 On System V.4 and embedded PowerPC systems, put small global and static
12002 data in the @samp{.sdata} section, which is pointed to by register
12003 @code{r13}. Put small uninitialized global and static data in the
12004 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
12005 The @option{-msdata=sysv} option is incompatible with the
12006 @option{-mrelocatable} option.
12008 @item -msdata=default
12010 @opindex msdata=default
12012 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
12013 compile code the same as @option{-msdata=eabi}, otherwise compile code the
12014 same as @option{-msdata=sysv}.
12017 @opindex msdata-data
12018 On System V.4 and embedded PowerPC systems, put small global
12019 data in the @samp{.sdata} section. Put small uninitialized global
12020 data in the @samp{.sbss} section. Do not use register @code{r13}
12021 to address small data however. This is the default behavior unless
12022 other @option{-msdata} options are used.
12026 @opindex msdata=none
12028 On embedded PowerPC systems, put all initialized global and static data
12029 in the @samp{.data} section, and all uninitialized data in the
12030 @samp{.bss} section.
12034 @cindex smaller data references (PowerPC)
12035 @cindex .sdata/.sdata2 references (PowerPC)
12036 On embedded PowerPC systems, put global and static items less than or
12037 equal to @var{num} bytes into the small data or bss sections instead of
12038 the normal data or bss section. By default, @var{num} is 8. The
12039 @option{-G @var{num}} switch is also passed to the linker.
12040 All modules should be compiled with the same @option{-G @var{num}} value.
12043 @itemx -mno-regnames
12045 @opindex mno-regnames
12046 On System V.4 and embedded PowerPC systems do (do not) emit register
12047 names in the assembly language output using symbolic forms.
12050 @itemx -mno-longcall
12052 @opindex mno-longcall
12053 By default assume that all calls are far away so that a longer more
12054 expensive calling sequence is required. This is required for calls
12055 further than 32 megabytes (33,554,432 bytes) from the current location.
12056 A short call will be generated if the compiler knows
12057 the call cannot be that far away. This setting can be overridden by
12058 the @code{shortcall} function attribute, or by @code{#pragma
12061 Some linkers are capable of detecting out-of-range calls and generating
12062 glue code on the fly. On these systems, long calls are unnecessary and
12063 generate slower code. As of this writing, the AIX linker can do this,
12064 as can the GNU linker for PowerPC/64. It is planned to add this feature
12065 to the GNU linker for 32-bit PowerPC systems as well.
12067 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
12068 callee, L42'', plus a ``branch island'' (glue code). The two target
12069 addresses represent the callee and the ``branch island''. The
12070 Darwin/PPC linker will prefer the first address and generate a ``bl
12071 callee'' if the PPC ``bl'' instruction will reach the callee directly;
12072 otherwise, the linker will generate ``bl L42'' to call the ``branch
12073 island''. The ``branch island'' is appended to the body of the
12074 calling function; it computes the full 32-bit address of the callee
12077 On Mach-O (Darwin) systems, this option directs the compiler emit to
12078 the glue for every direct call, and the Darwin linker decides whether
12079 to use or discard it.
12081 In the future, we may cause GCC to ignore all longcall specifications
12082 when the linker is known to generate glue.
12086 Adds support for multithreading with the @dfn{pthreads} library.
12087 This option sets flags for both the preprocessor and linker.
12091 @node S/390 and zSeries Options
12092 @subsection S/390 and zSeries Options
12093 @cindex S/390 and zSeries Options
12095 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
12099 @itemx -msoft-float
12100 @opindex mhard-float
12101 @opindex msoft-float
12102 Use (do not use) the hardware floating-point instructions and registers
12103 for floating-point operations. When @option{-msoft-float} is specified,
12104 functions in @file{libgcc.a} will be used to perform floating-point
12105 operations. When @option{-mhard-float} is specified, the compiler
12106 generates IEEE floating-point instructions. This is the default.
12108 @item -mlong-double-64
12109 @itemx -mlong-double-128
12110 @opindex mlong-double-64
12111 @opindex mlong-double-128
12112 These switches control the size of @code{long double} type. A size
12113 of 64bit makes the @code{long double} type equivalent to the @code{double}
12114 type. This is the default.
12117 @itemx -mno-backchain
12118 @opindex mbackchain
12119 @opindex mno-backchain
12120 Store (do not store) the address of the caller's frame as backchain pointer
12121 into the callee's stack frame.
12122 A backchain may be needed to allow debugging using tools that do not understand
12123 DWARF-2 call frame information.
12124 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
12125 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
12126 the backchain is placed into the topmost word of the 96/160 byte register
12129 In general, code compiled with @option{-mbackchain} is call-compatible with
12130 code compiled with @option{-mmo-backchain}; however, use of the backchain
12131 for debugging purposes usually requires that the whole binary is built with
12132 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
12133 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
12134 to build a linux kernel use @option{-msoft-float}.
12136 The default is to not maintain the backchain.
12138 @item -mpacked-stack
12139 @item -mno-packed-stack
12140 @opindex mpacked-stack
12141 @opindex mno-packed-stack
12142 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
12143 specified, the compiler uses the all fields of the 96/160 byte register save
12144 area only for their default purpose; unused fields still take up stack space.
12145 When @option{-mpacked-stack} is specified, register save slots are densely
12146 packed at the top of the register save area; unused space is reused for other
12147 purposes, allowing for more efficient use of the available stack space.
12148 However, when @option{-mbackchain} is also in effect, the topmost word of
12149 the save area is always used to store the backchain, and the return address
12150 register is always saved two words below the backchain.
12152 As long as the stack frame backchain is not used, code generated with
12153 @option{-mpacked-stack} is call-compatible with code generated with
12154 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
12155 S/390 or zSeries generated code that uses the stack frame backchain at run
12156 time, not just for debugging purposes. Such code is not call-compatible
12157 with code compiled with @option{-mpacked-stack}. Also, note that the
12158 combination of @option{-mbackchain},
12159 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
12160 to build a linux kernel use @option{-msoft-float}.
12162 The default is to not use the packed stack layout.
12165 @itemx -mno-small-exec
12166 @opindex msmall-exec
12167 @opindex mno-small-exec
12168 Generate (or do not generate) code using the @code{bras} instruction
12169 to do subroutine calls.
12170 This only works reliably if the total executable size does not
12171 exceed 64k. The default is to use the @code{basr} instruction instead,
12172 which does not have this limitation.
12178 When @option{-m31} is specified, generate code compliant to the
12179 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
12180 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
12181 particular to generate 64-bit instructions. For the @samp{s390}
12182 targets, the default is @option{-m31}, while the @samp{s390x}
12183 targets default to @option{-m64}.
12189 When @option{-mzarch} is specified, generate code using the
12190 instructions available on z/Architecture.
12191 When @option{-mesa} is specified, generate code using the
12192 instructions available on ESA/390. Note that @option{-mesa} is
12193 not possible with @option{-m64}.
12194 When generating code compliant to the GNU/Linux for S/390 ABI,
12195 the default is @option{-mesa}. When generating code compliant
12196 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
12202 Generate (or do not generate) code using the @code{mvcle} instruction
12203 to perform block moves. When @option{-mno-mvcle} is specified,
12204 use a @code{mvc} loop instead. This is the default unless optimizing for
12211 Print (or do not print) additional debug information when compiling.
12212 The default is to not print debug information.
12214 @item -march=@var{cpu-type}
12216 Generate code that will run on @var{cpu-type}, which is the name of a system
12217 representing a certain processor type. Possible values for
12218 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
12219 When generating code using the instructions available on z/Architecture,
12220 the default is @option{-march=z900}. Otherwise, the default is
12221 @option{-march=g5}.
12223 @item -mtune=@var{cpu-type}
12225 Tune to @var{cpu-type} everything applicable about the generated code,
12226 except for the ABI and the set of available instructions.
12227 The list of @var{cpu-type} values is the same as for @option{-march}.
12228 The default is the value used for @option{-march}.
12231 @itemx -mno-tpf-trace
12232 @opindex mtpf-trace
12233 @opindex mno-tpf-trace
12234 Generate code that adds (does not add) in TPF OS specific branches to trace
12235 routines in the operating system. This option is off by default, even
12236 when compiling for the TPF OS@.
12239 @itemx -mno-fused-madd
12240 @opindex mfused-madd
12241 @opindex mno-fused-madd
12242 Generate code that uses (does not use) the floating point multiply and
12243 accumulate instructions. These instructions are generated by default if
12244 hardware floating point is used.
12246 @item -mwarn-framesize=@var{framesize}
12247 @opindex mwarn-framesize
12248 Emit a warning if the current function exceeds the given frame size. Because
12249 this is a compile time check it doesn't need to be a real problem when the program
12250 runs. It is intended to identify functions which most probably cause
12251 a stack overflow. It is useful to be used in an environment with limited stack
12252 size e.g.@: the linux kernel.
12254 @item -mwarn-dynamicstack
12255 @opindex mwarn-dynamicstack
12256 Emit a warning if the function calls alloca or uses dynamically
12257 sized arrays. This is generally a bad idea with a limited stack size.
12259 @item -mstack-guard=@var{stack-guard}
12260 @item -mstack-size=@var{stack-size}
12261 @opindex mstack-guard
12262 @opindex mstack-size
12263 These arguments always have to be used in conjunction. If they are present the s390
12264 back end emits additional instructions in the function prologue which trigger a trap
12265 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
12266 (remember that the stack on s390 grows downward). These options are intended to
12267 be used to help debugging stack overflow problems. The additionally emitted code
12268 causes only little overhead and hence can also be used in production like systems
12269 without greater performance degradation. The given values have to be exact
12270 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard} without
12272 In order to be efficient the extra code makes the assumption that the stack starts
12273 at an address aligned to the value given by @var{stack-size}.
12276 @node Score Options
12277 @subsection Score Options
12278 @cindex Score Options
12280 These options are defined for Score implementations:
12285 Compile code for big endian mode. This is the default.
12289 Compile code for little endian mode.
12293 Disable generate bcnz instruction.
12297 Enable generate unaligned load and store instruction.
12301 Enable the use of multiply-accumulate instructions. Disabled by default.
12305 Specify the SCORE5 as the target architecture.
12309 Specify the SCORE5U of the target architecture.
12313 Specify the SCORE7 as the target architecture. This is the default.
12317 Specify the SCORE7D as the target architecture.
12321 @subsection SH Options
12323 These @samp{-m} options are defined for the SH implementations:
12328 Generate code for the SH1.
12332 Generate code for the SH2.
12335 Generate code for the SH2e.
12339 Generate code for the SH3.
12343 Generate code for the SH3e.
12347 Generate code for the SH4 without a floating-point unit.
12349 @item -m4-single-only
12350 @opindex m4-single-only
12351 Generate code for the SH4 with a floating-point unit that only
12352 supports single-precision arithmetic.
12356 Generate code for the SH4 assuming the floating-point unit is in
12357 single-precision mode by default.
12361 Generate code for the SH4.
12365 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
12366 floating-point unit is not used.
12368 @item -m4a-single-only
12369 @opindex m4a-single-only
12370 Generate code for the SH4a, in such a way that no double-precision
12371 floating point operations are used.
12374 @opindex m4a-single
12375 Generate code for the SH4a assuming the floating-point unit is in
12376 single-precision mode by default.
12380 Generate code for the SH4a.
12384 Same as @option{-m4a-nofpu}, except that it implicitly passes
12385 @option{-dsp} to the assembler. GCC doesn't generate any DSP
12386 instructions at the moment.
12390 Compile code for the processor in big endian mode.
12394 Compile code for the processor in little endian mode.
12398 Align doubles at 64-bit boundaries. Note that this changes the calling
12399 conventions, and thus some functions from the standard C library will
12400 not work unless you recompile it first with @option{-mdalign}.
12404 Shorten some address references at link time, when possible; uses the
12405 linker option @option{-relax}.
12409 Use 32-bit offsets in @code{switch} tables. The default is to use
12414 Enable the use of the instruction @code{fmovd}.
12418 Comply with the calling conventions defined by Renesas.
12422 Comply with the calling conventions defined by Renesas.
12426 Comply with the calling conventions defined for GCC before the Renesas
12427 conventions were available. This option is the default for all
12428 targets of the SH toolchain except for @samp{sh-symbianelf}.
12431 @opindex mnomacsave
12432 Mark the @code{MAC} register as call-clobbered, even if
12433 @option{-mhitachi} is given.
12437 Increase IEEE-compliance of floating-point code.
12438 At the moment, this is equivalent to @option{-fno-finite-math-only}.
12439 When generating 16 bit SH opcodes, getting IEEE-conforming results for
12440 comparisons of NANs / infinities incurs extra overhead in every
12441 floating point comparison, therefore the default is set to
12442 @option{-ffinite-math-only}.
12446 Dump instruction size and location in the assembly code.
12449 @opindex mpadstruct
12450 This option is deprecated. It pads structures to multiple of 4 bytes,
12451 which is incompatible with the SH ABI@.
12455 Optimize for space instead of speed. Implied by @option{-Os}.
12458 @opindex mprefergot
12459 When generating position-independent code, emit function calls using
12460 the Global Offset Table instead of the Procedure Linkage Table.
12464 Generate a library function call to invalidate instruction cache
12465 entries, after fixing up a trampoline. This library function call
12466 doesn't assume it can write to the whole memory address space. This
12467 is the default when the target is @code{sh-*-linux*}.
12469 @item -multcost=@var{number}
12470 @opindex multcost=@var{number}
12471 Set the cost to assume for a multiply insn.
12473 @item -mdiv=@var{strategy}
12474 @opindex mdiv=@var{strategy}
12475 Set the division strategy to use for SHmedia code. @var{strategy} must be
12476 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
12477 inv:call2, inv:fp .
12478 "fp" performs the operation in floating point. This has a very high latency,
12479 but needs only a few instructions, so it might be a good choice if
12480 your code has enough easily exploitable ILP to allow the compiler to
12481 schedule the floating point instructions together with other instructions.
12482 Division by zero causes a floating point exception.
12483 "inv" uses integer operations to calculate the inverse of the divisor,
12484 and then multiplies the dividend with the inverse. This strategy allows
12485 cse and hoisting of the inverse calculation. Division by zero calculates
12486 an unspecified result, but does not trap.
12487 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
12488 have been found, or if the entire operation has been hoisted to the same
12489 place, the last stages of the inverse calculation are intertwined with the
12490 final multiply to reduce the overall latency, at the expense of using a few
12491 more instructions, and thus offering fewer scheduling opportunities with
12493 "call" calls a library function that usually implements the inv:minlat
12495 This gives high code density for m5-*media-nofpu compilations.
12496 "call2" uses a different entry point of the same library function, where it
12497 assumes that a pointer to a lookup table has already been set up, which
12498 exposes the pointer load to cse / code hoisting optimizations.
12499 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
12500 code generation, but if the code stays unoptimized, revert to the "call",
12501 "call2", or "fp" strategies, respectively. Note that the
12502 potentially-trapping side effect of division by zero is carried by a
12503 separate instruction, so it is possible that all the integer instructions
12504 are hoisted out, but the marker for the side effect stays where it is.
12505 A recombination to fp operations or a call is not possible in that case.
12506 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
12507 that the inverse calculation was nor separated from the multiply, they speed
12508 up division where the dividend fits into 20 bits (plus sign where applicable),
12509 by inserting a test to skip a number of operations in this case; this test
12510 slows down the case of larger dividends. inv20u assumes the case of a such
12511 a small dividend to be unlikely, and inv20l assumes it to be likely.
12513 @item -mdivsi3_libfunc=@var{name}
12514 @opindex mdivsi3_libfunc=@var{name}
12515 Set the name of the library function used for 32 bit signed division to
12516 @var{name}. This only affect the name used in the call and inv:call
12517 division strategies, and the compiler will still expect the same
12518 sets of input/output/clobbered registers as if this option was not present.
12520 @item -madjust-unroll
12521 @opindex madjust-unroll
12522 Throttle unrolling to avoid thrashing target registers.
12523 This option only has an effect if the gcc code base supports the
12524 TARGET_ADJUST_UNROLL_MAX target hook.
12526 @item -mindexed-addressing
12527 @opindex mindexed-addressing
12528 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
12529 This is only safe if the hardware and/or OS implement 32 bit wrap-around
12530 semantics for the indexed addressing mode. The architecture allows the
12531 implementation of processors with 64 bit MMU, which the OS could use to
12532 get 32 bit addressing, but since no current hardware implementation supports
12533 this or any other way to make the indexed addressing mode safe to use in
12534 the 32 bit ABI, the default is -mno-indexed-addressing.
12536 @item -mgettrcost=@var{number}
12537 @opindex mgettrcost=@var{number}
12538 Set the cost assumed for the gettr instruction to @var{number}.
12539 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
12543 Assume pt* instructions won't trap. This will generally generate better
12544 scheduled code, but is unsafe on current hardware. The current architecture
12545 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
12546 This has the unintentional effect of making it unsafe to schedule ptabs /
12547 ptrel before a branch, or hoist it out of a loop. For example,
12548 __do_global_ctors, a part of libgcc that runs constructors at program
12549 startup, calls functions in a list which is delimited by -1. With the
12550 -mpt-fixed option, the ptabs will be done before testing against -1.
12551 That means that all the constructors will be run a bit quicker, but when
12552 the loop comes to the end of the list, the program crashes because ptabs
12553 loads -1 into a target register. Since this option is unsafe for any
12554 hardware implementing the current architecture specification, the default
12555 is -mno-pt-fixed. Unless the user specifies a specific cost with
12556 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
12557 this deters register allocation using target registers for storing
12560 @item -minvalid-symbols
12561 @opindex minvalid-symbols
12562 Assume symbols might be invalid. Ordinary function symbols generated by
12563 the compiler will always be valid to load with movi/shori/ptabs or
12564 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
12565 to generate symbols that will cause ptabs / ptrel to trap.
12566 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
12567 It will then prevent cross-basic-block cse, hoisting and most scheduling
12568 of symbol loads. The default is @option{-mno-invalid-symbols}.
12571 @node SPARC Options
12572 @subsection SPARC Options
12573 @cindex SPARC options
12575 These @samp{-m} options are supported on the SPARC:
12578 @item -mno-app-regs
12580 @opindex mno-app-regs
12582 Specify @option{-mapp-regs} to generate output using the global registers
12583 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
12586 To be fully SVR4 ABI compliant at the cost of some performance loss,
12587 specify @option{-mno-app-regs}. You should compile libraries and system
12588 software with this option.
12591 @itemx -mhard-float
12593 @opindex mhard-float
12594 Generate output containing floating point instructions. This is the
12598 @itemx -msoft-float
12600 @opindex msoft-float
12601 Generate output containing library calls for floating point.
12602 @strong{Warning:} the requisite libraries are not available for all SPARC
12603 targets. Normally the facilities of the machine's usual C compiler are
12604 used, but this cannot be done directly in cross-compilation. You must make
12605 your own arrangements to provide suitable library functions for
12606 cross-compilation. The embedded targets @samp{sparc-*-aout} and
12607 @samp{sparclite-*-*} do provide software floating point support.
12609 @option{-msoft-float} changes the calling convention in the output file;
12610 therefore, it is only useful if you compile @emph{all} of a program with
12611 this option. In particular, you need to compile @file{libgcc.a}, the
12612 library that comes with GCC, with @option{-msoft-float} in order for
12615 @item -mhard-quad-float
12616 @opindex mhard-quad-float
12617 Generate output containing quad-word (long double) floating point
12620 @item -msoft-quad-float
12621 @opindex msoft-quad-float
12622 Generate output containing library calls for quad-word (long double)
12623 floating point instructions. The functions called are those specified
12624 in the SPARC ABI@. This is the default.
12626 As of this writing, there are no SPARC implementations that have hardware
12627 support for the quad-word floating point instructions. They all invoke
12628 a trap handler for one of these instructions, and then the trap handler
12629 emulates the effect of the instruction. Because of the trap handler overhead,
12630 this is much slower than calling the ABI library routines. Thus the
12631 @option{-msoft-quad-float} option is the default.
12633 @item -mno-unaligned-doubles
12634 @itemx -munaligned-doubles
12635 @opindex mno-unaligned-doubles
12636 @opindex munaligned-doubles
12637 Assume that doubles have 8 byte alignment. This is the default.
12639 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
12640 alignment only if they are contained in another type, or if they have an
12641 absolute address. Otherwise, it assumes they have 4 byte alignment.
12642 Specifying this option avoids some rare compatibility problems with code
12643 generated by other compilers. It is not the default because it results
12644 in a performance loss, especially for floating point code.
12646 @item -mno-faster-structs
12647 @itemx -mfaster-structs
12648 @opindex mno-faster-structs
12649 @opindex mfaster-structs
12650 With @option{-mfaster-structs}, the compiler assumes that structures
12651 should have 8 byte alignment. This enables the use of pairs of
12652 @code{ldd} and @code{std} instructions for copies in structure
12653 assignment, in place of twice as many @code{ld} and @code{st} pairs.
12654 However, the use of this changed alignment directly violates the SPARC
12655 ABI@. Thus, it's intended only for use on targets where the developer
12656 acknowledges that their resulting code will not be directly in line with
12657 the rules of the ABI@.
12659 @item -mimpure-text
12660 @opindex mimpure-text
12661 @option{-mimpure-text}, used in addition to @option{-shared}, tells
12662 the compiler to not pass @option{-z text} to the linker when linking a
12663 shared object. Using this option, you can link position-dependent
12664 code into a shared object.
12666 @option{-mimpure-text} suppresses the ``relocations remain against
12667 allocatable but non-writable sections'' linker error message.
12668 However, the necessary relocations will trigger copy-on-write, and the
12669 shared object is not actually shared across processes. Instead of
12670 using @option{-mimpure-text}, you should compile all source code with
12671 @option{-fpic} or @option{-fPIC}.
12673 This option is only available on SunOS and Solaris.
12675 @item -mcpu=@var{cpu_type}
12677 Set the instruction set, register set, and instruction scheduling parameters
12678 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
12679 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
12680 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
12681 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
12682 @samp{ultrasparc3}, and @samp{niagara}.
12684 Default instruction scheduling parameters are used for values that select
12685 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
12686 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
12688 Here is a list of each supported architecture and their supported
12693 v8: supersparc, hypersparc
12694 sparclite: f930, f934, sparclite86x
12696 v9: ultrasparc, ultrasparc3, niagara
12699 By default (unless configured otherwise), GCC generates code for the V7
12700 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
12701 additionally optimizes it for the Cypress CY7C602 chip, as used in the
12702 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
12703 SPARCStation 1, 2, IPX etc.
12705 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
12706 architecture. The only difference from V7 code is that the compiler emits
12707 the integer multiply and integer divide instructions which exist in SPARC-V8
12708 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
12709 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
12712 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
12713 the SPARC architecture. This adds the integer multiply, integer divide step
12714 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
12715 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
12716 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
12717 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
12718 MB86934 chip, which is the more recent SPARClite with FPU@.
12720 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
12721 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
12722 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
12723 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
12724 optimizes it for the TEMIC SPARClet chip.
12726 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
12727 architecture. This adds 64-bit integer and floating-point move instructions,
12728 3 additional floating-point condition code registers and conditional move
12729 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
12730 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
12731 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
12732 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
12733 @option{-mcpu=niagara}, the compiler additionally optimizes it for
12734 Sun UltraSPARC T1 chips.
12736 @item -mtune=@var{cpu_type}
12738 Set the instruction scheduling parameters for machine type
12739 @var{cpu_type}, but do not set the instruction set or register set that the
12740 option @option{-mcpu=@var{cpu_type}} would.
12742 The same values for @option{-mcpu=@var{cpu_type}} can be used for
12743 @option{-mtune=@var{cpu_type}}, but the only useful values are those
12744 that select a particular cpu implementation. Those are @samp{cypress},
12745 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
12746 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
12747 @samp{ultrasparc3}, and @samp{niagara}.
12752 @opindex mno-v8plus
12753 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
12754 difference from the V8 ABI is that the global and out registers are
12755 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
12756 mode for all SPARC-V9 processors.
12762 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
12763 Visual Instruction Set extensions. The default is @option{-mno-vis}.
12766 These @samp{-m} options are supported in addition to the above
12767 on SPARC-V9 processors in 64-bit environments:
12770 @item -mlittle-endian
12771 @opindex mlittle-endian
12772 Generate code for a processor running in little-endian mode. It is only
12773 available for a few configurations and most notably not on Solaris and Linux.
12779 Generate code for a 32-bit or 64-bit environment.
12780 The 32-bit environment sets int, long and pointer to 32 bits.
12781 The 64-bit environment sets int to 32 bits and long and pointer
12784 @item -mcmodel=medlow
12785 @opindex mcmodel=medlow
12786 Generate code for the Medium/Low code model: 64-bit addresses, programs
12787 must be linked in the low 32 bits of memory. Programs can be statically
12788 or dynamically linked.
12790 @item -mcmodel=medmid
12791 @opindex mcmodel=medmid
12792 Generate code for the Medium/Middle code model: 64-bit addresses, programs
12793 must be linked in the low 44 bits of memory, the text and data segments must
12794 be less than 2GB in size and the data segment must be located within 2GB of
12797 @item -mcmodel=medany
12798 @opindex mcmodel=medany
12799 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
12800 may be linked anywhere in memory, the text and data segments must be less
12801 than 2GB in size and the data segment must be located within 2GB of the
12804 @item -mcmodel=embmedany
12805 @opindex mcmodel=embmedany
12806 Generate code for the Medium/Anywhere code model for embedded systems:
12807 64-bit addresses, the text and data segments must be less than 2GB in
12808 size, both starting anywhere in memory (determined at link time). The
12809 global register %g4 points to the base of the data segment. Programs
12810 are statically linked and PIC is not supported.
12813 @itemx -mno-stack-bias
12814 @opindex mstack-bias
12815 @opindex mno-stack-bias
12816 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
12817 frame pointer if present, are offset by @minus{}2047 which must be added back
12818 when making stack frame references. This is the default in 64-bit mode.
12819 Otherwise, assume no such offset is present.
12822 These switches are supported in addition to the above on Solaris:
12827 Add support for multithreading using the Solaris threads library. This
12828 option sets flags for both the preprocessor and linker. This option does
12829 not affect the thread safety of object code produced by the compiler or
12830 that of libraries supplied with it.
12834 Add support for multithreading using the POSIX threads library. This
12835 option sets flags for both the preprocessor and linker. This option does
12836 not affect the thread safety of object code produced by the compiler or
12837 that of libraries supplied with it.
12841 This is a synonym for @option{-pthreads}.
12844 @node System V Options
12845 @subsection Options for System V
12847 These additional options are available on System V Release 4 for
12848 compatibility with other compilers on those systems:
12853 Create a shared object.
12854 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
12858 Identify the versions of each tool used by the compiler, in a
12859 @code{.ident} assembler directive in the output.
12863 Refrain from adding @code{.ident} directives to the output file (this is
12866 @item -YP,@var{dirs}
12868 Search the directories @var{dirs}, and no others, for libraries
12869 specified with @option{-l}.
12871 @item -Ym,@var{dir}
12873 Look in the directory @var{dir} to find the M4 preprocessor.
12874 The assembler uses this option.
12875 @c This is supposed to go with a -Yd for predefined M4 macro files, but
12876 @c the generic assembler that comes with Solaris takes just -Ym.
12879 @node TMS320C3x/C4x Options
12880 @subsection TMS320C3x/C4x Options
12881 @cindex TMS320C3x/C4x Options
12883 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
12887 @item -mcpu=@var{cpu_type}
12889 Set the instruction set, register set, and instruction scheduling
12890 parameters for machine type @var{cpu_type}. Supported values for
12891 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
12892 @samp{c44}. The default is @samp{c40} to generate code for the
12897 @itemx -msmall-memory
12899 @opindex mbig-memory
12901 @opindex msmall-memory
12903 Generates code for the big or small memory model. The small memory
12904 model assumed that all data fits into one 64K word page. At run-time
12905 the data page (DP) register must be set to point to the 64K page
12906 containing the .bss and .data program sections. The big memory model is
12907 the default and requires reloading of the DP register for every direct
12914 Allow (disallow) allocation of general integer operands into the block
12915 count register BK@.
12921 Enable (disable) generation of code using decrement and branch,
12922 DBcond(D), instructions. This is enabled by default for the C4x. To be
12923 on the safe side, this is disabled for the C3x, since the maximum
12924 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
12925 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
12926 that it can utilize the decrement and branch instruction, but will give
12927 up if there is more than one memory reference in the loop. Thus a loop
12928 where the loop counter is decremented can generate slightly more
12929 efficient code, in cases where the RPTB instruction cannot be utilized.
12931 @item -mdp-isr-reload
12933 @opindex mdp-isr-reload
12935 Force the DP register to be saved on entry to an interrupt service
12936 routine (ISR), reloaded to point to the data section, and restored on
12937 exit from the ISR@. This should not be required unless someone has
12938 violated the small memory model by modifying the DP register, say within
12945 For the C3x use the 24-bit MPYI instruction for integer multiplies
12946 instead of a library call to guarantee 32-bit results. Note that if one
12947 of the operands is a constant, then the multiplication will be performed
12948 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
12949 then squaring operations are performed inline instead of a library call.
12952 @itemx -mno-fast-fix
12954 @opindex mno-fast-fix
12955 The C3x/C4x FIX instruction to convert a floating point value to an
12956 integer value chooses the nearest integer less than or equal to the
12957 floating point value rather than to the nearest integer. Thus if the
12958 floating point number is negative, the result will be incorrectly
12959 truncated an additional code is necessary to detect and correct this
12960 case. This option can be used to disable generation of the additional
12961 code required to correct the result.
12967 Enable (disable) generation of repeat block sequences using the RPTB
12968 instruction for zero overhead looping. The RPTB construct is only used
12969 for innermost loops that do not call functions or jump across the loop
12970 boundaries. There is no advantage having nested RPTB loops due to the
12971 overhead required to save and restore the RC, RS, and RE registers.
12972 This is enabled by default with @option{-O2}.
12974 @item -mrpts=@var{count}
12978 Enable (disable) the use of the single instruction repeat instruction
12979 RPTS@. If a repeat block contains a single instruction, and the loop
12980 count can be guaranteed to be less than the value @var{count}, GCC will
12981 emit a RPTS instruction instead of a RPTB@. If no value is specified,
12982 then a RPTS will be emitted even if the loop count cannot be determined
12983 at compile time. Note that the repeated instruction following RPTS does
12984 not have to be reloaded from memory each iteration, thus freeing up the
12985 CPU buses for operands. However, since interrupts are blocked by this
12986 instruction, it is disabled by default.
12988 @item -mloop-unsigned
12989 @itemx -mno-loop-unsigned
12990 @opindex mloop-unsigned
12991 @opindex mno-loop-unsigned
12992 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
12993 is @math{2^{31} + 1} since these instructions test if the iteration count is
12994 negative to terminate the loop. If the iteration count is unsigned
12995 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
12996 exceeded. This switch allows an unsigned iteration count.
13000 Try to emit an assembler syntax that the TI assembler (asm30) is happy
13001 with. This also enforces compatibility with the API employed by the TI
13002 C3x C compiler. For example, long doubles are passed as structures
13003 rather than in floating point registers.
13009 Generate code that uses registers (stack) for passing arguments to functions.
13010 By default, arguments are passed in registers where possible rather
13011 than by pushing arguments on to the stack.
13013 @item -mparallel-insns
13014 @itemx -mno-parallel-insns
13015 @opindex mparallel-insns
13016 @opindex mno-parallel-insns
13017 Allow the generation of parallel instructions. This is enabled by
13018 default with @option{-O2}.
13020 @item -mparallel-mpy
13021 @itemx -mno-parallel-mpy
13022 @opindex mparallel-mpy
13023 @opindex mno-parallel-mpy
13024 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
13025 provided @option{-mparallel-insns} is also specified. These instructions have
13026 tight register constraints which can pessimize the code generation
13027 of large functions.
13032 @subsection V850 Options
13033 @cindex V850 Options
13035 These @samp{-m} options are defined for V850 implementations:
13039 @itemx -mno-long-calls
13040 @opindex mlong-calls
13041 @opindex mno-long-calls
13042 Treat all calls as being far away (near). If calls are assumed to be
13043 far away, the compiler will always load the functions address up into a
13044 register, and call indirect through the pointer.
13050 Do not optimize (do optimize) basic blocks that use the same index
13051 pointer 4 or more times to copy pointer into the @code{ep} register, and
13052 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
13053 option is on by default if you optimize.
13055 @item -mno-prolog-function
13056 @itemx -mprolog-function
13057 @opindex mno-prolog-function
13058 @opindex mprolog-function
13059 Do not use (do use) external functions to save and restore registers
13060 at the prologue and epilogue of a function. The external functions
13061 are slower, but use less code space if more than one function saves
13062 the same number of registers. The @option{-mprolog-function} option
13063 is on by default if you optimize.
13067 Try to make the code as small as possible. At present, this just turns
13068 on the @option{-mep} and @option{-mprolog-function} options.
13070 @item -mtda=@var{n}
13072 Put static or global variables whose size is @var{n} bytes or less into
13073 the tiny data area that register @code{ep} points to. The tiny data
13074 area can hold up to 256 bytes in total (128 bytes for byte references).
13076 @item -msda=@var{n}
13078 Put static or global variables whose size is @var{n} bytes or less into
13079 the small data area that register @code{gp} points to. The small data
13080 area can hold up to 64 kilobytes.
13082 @item -mzda=@var{n}
13084 Put static or global variables whose size is @var{n} bytes or less into
13085 the first 32 kilobytes of memory.
13089 Specify that the target processor is the V850.
13092 @opindex mbig-switch
13093 Generate code suitable for big switch tables. Use this option only if
13094 the assembler/linker complain about out of range branches within a switch
13099 This option will cause r2 and r5 to be used in the code generated by
13100 the compiler. This setting is the default.
13102 @item -mno-app-regs
13103 @opindex mno-app-regs
13104 This option will cause r2 and r5 to be treated as fixed registers.
13108 Specify that the target processor is the V850E1. The preprocessor
13109 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
13110 this option is used.
13114 Specify that the target processor is the V850E@. The preprocessor
13115 constant @samp{__v850e__} will be defined if this option is used.
13117 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
13118 are defined then a default target processor will be chosen and the
13119 relevant @samp{__v850*__} preprocessor constant will be defined.
13121 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
13122 defined, regardless of which processor variant is the target.
13124 @item -mdisable-callt
13125 @opindex mdisable-callt
13126 This option will suppress generation of the CALLT instruction for the
13127 v850e and v850e1 flavors of the v850 architecture. The default is
13128 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
13133 @subsection VAX Options
13134 @cindex VAX options
13136 These @samp{-m} options are defined for the VAX:
13141 Do not output certain jump instructions (@code{aobleq} and so on)
13142 that the Unix assembler for the VAX cannot handle across long
13147 Do output those jump instructions, on the assumption that you
13148 will assemble with the GNU assembler.
13152 Output code for g-format floating point numbers instead of d-format.
13155 @node x86-64 Options
13156 @subsection x86-64 Options
13157 @cindex x86-64 options
13159 These are listed under @xref{i386 and x86-64 Options}.
13161 @node Xstormy16 Options
13162 @subsection Xstormy16 Options
13163 @cindex Xstormy16 Options
13165 These options are defined for Xstormy16:
13170 Choose startup files and linker script suitable for the simulator.
13173 @node Xtensa Options
13174 @subsection Xtensa Options
13175 @cindex Xtensa Options
13177 These options are supported for Xtensa targets:
13181 @itemx -mno-const16
13183 @opindex mno-const16
13184 Enable or disable use of @code{CONST16} instructions for loading
13185 constant values. The @code{CONST16} instruction is currently not a
13186 standard option from Tensilica. When enabled, @code{CONST16}
13187 instructions are always used in place of the standard @code{L32R}
13188 instructions. The use of @code{CONST16} is enabled by default only if
13189 the @code{L32R} instruction is not available.
13192 @itemx -mno-fused-madd
13193 @opindex mfused-madd
13194 @opindex mno-fused-madd
13195 Enable or disable use of fused multiply/add and multiply/subtract
13196 instructions in the floating-point option. This has no effect if the
13197 floating-point option is not also enabled. Disabling fused multiply/add
13198 and multiply/subtract instructions forces the compiler to use separate
13199 instructions for the multiply and add/subtract operations. This may be
13200 desirable in some cases where strict IEEE 754-compliant results are
13201 required: the fused multiply add/subtract instructions do not round the
13202 intermediate result, thereby producing results with @emph{more} bits of
13203 precision than specified by the IEEE standard. Disabling fused multiply
13204 add/subtract instructions also ensures that the program output is not
13205 sensitive to the compiler's ability to combine multiply and add/subtract
13208 @item -mtext-section-literals
13209 @itemx -mno-text-section-literals
13210 @opindex mtext-section-literals
13211 @opindex mno-text-section-literals
13212 Control the treatment of literal pools. The default is
13213 @option{-mno-text-section-literals}, which places literals in a separate
13214 section in the output file. This allows the literal pool to be placed
13215 in a data RAM/ROM, and it also allows the linker to combine literal
13216 pools from separate object files to remove redundant literals and
13217 improve code size. With @option{-mtext-section-literals}, the literals
13218 are interspersed in the text section in order to keep them as close as
13219 possible to their references. This may be necessary for large assembly
13222 @item -mtarget-align
13223 @itemx -mno-target-align
13224 @opindex mtarget-align
13225 @opindex mno-target-align
13226 When this option is enabled, GCC instructs the assembler to
13227 automatically align instructions to reduce branch penalties at the
13228 expense of some code density. The assembler attempts to widen density
13229 instructions to align branch targets and the instructions following call
13230 instructions. If there are not enough preceding safe density
13231 instructions to align a target, no widening will be performed. The
13232 default is @option{-mtarget-align}. These options do not affect the
13233 treatment of auto-aligned instructions like @code{LOOP}, which the
13234 assembler will always align, either by widening density instructions or
13235 by inserting no-op instructions.
13238 @itemx -mno-longcalls
13239 @opindex mlongcalls
13240 @opindex mno-longcalls
13241 When this option is enabled, GCC instructs the assembler to translate
13242 direct calls to indirect calls unless it can determine that the target
13243 of a direct call is in the range allowed by the call instruction. This
13244 translation typically occurs for calls to functions in other source
13245 files. Specifically, the assembler translates a direct @code{CALL}
13246 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
13247 The default is @option{-mno-longcalls}. This option should be used in
13248 programs where the call target can potentially be out of range. This
13249 option is implemented in the assembler, not the compiler, so the
13250 assembly code generated by GCC will still show direct call
13251 instructions---look at the disassembled object code to see the actual
13252 instructions. Note that the assembler will use an indirect call for
13253 every cross-file call, not just those that really will be out of range.
13256 @node zSeries Options
13257 @subsection zSeries Options
13258 @cindex zSeries options
13260 These are listed under @xref{S/390 and zSeries Options}.
13262 @node Code Gen Options
13263 @section Options for Code Generation Conventions
13264 @cindex code generation conventions
13265 @cindex options, code generation
13266 @cindex run-time options
13268 These machine-independent options control the interface conventions
13269 used in code generation.
13271 Most of them have both positive and negative forms; the negative form
13272 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
13273 one of the forms is listed---the one which is not the default. You
13274 can figure out the other form by either removing @samp{no-} or adding
13278 @item -fbounds-check
13279 @opindex fbounds-check
13280 For front-ends that support it, generate additional code to check that
13281 indices used to access arrays are within the declared range. This is
13282 currently only supported by the Java and Fortran front-ends, where
13283 this option defaults to true and false respectively.
13287 This option generates traps for signed overflow on addition, subtraction,
13288 multiplication operations.
13292 This option instructs the compiler to assume that signed arithmetic
13293 overflow of addition, subtraction and multiplication wraps around
13294 using twos-complement representation. This flag enables some optimizations
13295 and disables others. This option is enabled by default for the Java
13296 front-end, as required by the Java language specification.
13299 @opindex fexceptions
13300 Enable exception handling. Generates extra code needed to propagate
13301 exceptions. For some targets, this implies GCC will generate frame
13302 unwind information for all functions, which can produce significant data
13303 size overhead, although it does not affect execution. If you do not
13304 specify this option, GCC will enable it by default for languages like
13305 C++ which normally require exception handling, and disable it for
13306 languages like C that do not normally require it. However, you may need
13307 to enable this option when compiling C code that needs to interoperate
13308 properly with exception handlers written in C++. You may also wish to
13309 disable this option if you are compiling older C++ programs that don't
13310 use exception handling.
13312 @item -fnon-call-exceptions
13313 @opindex fnon-call-exceptions
13314 Generate code that allows trapping instructions to throw exceptions.
13315 Note that this requires platform-specific runtime support that does
13316 not exist everywhere. Moreover, it only allows @emph{trapping}
13317 instructions to throw exceptions, i.e.@: memory references or floating
13318 point instructions. It does not allow exceptions to be thrown from
13319 arbitrary signal handlers such as @code{SIGALRM}.
13321 @item -funwind-tables
13322 @opindex funwind-tables
13323 Similar to @option{-fexceptions}, except that it will just generate any needed
13324 static data, but will not affect the generated code in any other way.
13325 You will normally not enable this option; instead, a language processor
13326 that needs this handling would enable it on your behalf.
13328 @item -fasynchronous-unwind-tables
13329 @opindex fasynchronous-unwind-tables
13330 Generate unwind table in dwarf2 format, if supported by target machine. The
13331 table is exact at each instruction boundary, so it can be used for stack
13332 unwinding from asynchronous events (such as debugger or garbage collector).
13334 @item -fpcc-struct-return
13335 @opindex fpcc-struct-return
13336 Return ``short'' @code{struct} and @code{union} values in memory like
13337 longer ones, rather than in registers. This convention is less
13338 efficient, but it has the advantage of allowing intercallability between
13339 GCC-compiled files and files compiled with other compilers, particularly
13340 the Portable C Compiler (pcc).
13342 The precise convention for returning structures in memory depends
13343 on the target configuration macros.
13345 Short structures and unions are those whose size and alignment match
13346 that of some integer type.
13348 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
13349 switch is not binary compatible with code compiled with the
13350 @option{-freg-struct-return} switch.
13351 Use it to conform to a non-default application binary interface.
13353 @item -freg-struct-return
13354 @opindex freg-struct-return
13355 Return @code{struct} and @code{union} values in registers when possible.
13356 This is more efficient for small structures than
13357 @option{-fpcc-struct-return}.
13359 If you specify neither @option{-fpcc-struct-return} nor
13360 @option{-freg-struct-return}, GCC defaults to whichever convention is
13361 standard for the target. If there is no standard convention, GCC
13362 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
13363 the principal compiler. In those cases, we can choose the standard, and
13364 we chose the more efficient register return alternative.
13366 @strong{Warning:} code compiled with the @option{-freg-struct-return}
13367 switch is not binary compatible with code compiled with the
13368 @option{-fpcc-struct-return} switch.
13369 Use it to conform to a non-default application binary interface.
13371 @item -fshort-enums
13372 @opindex fshort-enums
13373 Allocate to an @code{enum} type only as many bytes as it needs for the
13374 declared range of possible values. Specifically, the @code{enum} type
13375 will be equivalent to the smallest integer type which has enough room.
13377 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
13378 code that is not binary compatible with code generated without that switch.
13379 Use it to conform to a non-default application binary interface.
13381 @item -fshort-double
13382 @opindex fshort-double
13383 Use the same size for @code{double} as for @code{float}.
13385 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
13386 code that is not binary compatible with code generated without that switch.
13387 Use it to conform to a non-default application binary interface.
13389 @item -fshort-wchar
13390 @opindex fshort-wchar
13391 Override the underlying type for @samp{wchar_t} to be @samp{short
13392 unsigned int} instead of the default for the target. This option is
13393 useful for building programs to run under WINE@.
13395 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
13396 code that is not binary compatible with code generated without that switch.
13397 Use it to conform to a non-default application binary interface.
13400 @opindex fno-common
13401 In C, allocate even uninitialized global variables in the data section of the
13402 object file, rather than generating them as common blocks. This has the
13403 effect that if the same variable is declared (without @code{extern}) in
13404 two different compilations, you will get an error when you link them.
13405 The only reason this might be useful is if you wish to verify that the
13406 program will work on other systems which always work this way.
13410 Ignore the @samp{#ident} directive.
13412 @item -finhibit-size-directive
13413 @opindex finhibit-size-directive
13414 Don't output a @code{.size} assembler directive, or anything else that
13415 would cause trouble if the function is split in the middle, and the
13416 two halves are placed at locations far apart in memory. This option is
13417 used when compiling @file{crtstuff.c}; you should not need to use it
13420 @item -fverbose-asm
13421 @opindex fverbose-asm
13422 Put extra commentary information in the generated assembly code to
13423 make it more readable. This option is generally only of use to those
13424 who actually need to read the generated assembly code (perhaps while
13425 debugging the compiler itself).
13427 @option{-fno-verbose-asm}, the default, causes the
13428 extra information to be omitted and is useful when comparing two assembler
13433 @cindex global offset table
13435 Generate position-independent code (PIC) suitable for use in a shared
13436 library, if supported for the target machine. Such code accesses all
13437 constant addresses through a global offset table (GOT)@. The dynamic
13438 loader resolves the GOT entries when the program starts (the dynamic
13439 loader is not part of GCC; it is part of the operating system). If
13440 the GOT size for the linked executable exceeds a machine-specific
13441 maximum size, you get an error message from the linker indicating that
13442 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
13443 instead. (These maximums are 8k on the SPARC and 32k
13444 on the m68k and RS/6000. The 386 has no such limit.)
13446 Position-independent code requires special support, and therefore works
13447 only on certain machines. For the 386, GCC supports PIC for System V
13448 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
13449 position-independent.
13451 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13456 If supported for the target machine, emit position-independent code,
13457 suitable for dynamic linking and avoiding any limit on the size of the
13458 global offset table. This option makes a difference on the m68k,
13459 PowerPC and SPARC@.
13461 Position-independent code requires special support, and therefore works
13462 only on certain machines.
13464 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13471 These options are similar to @option{-fpic} and @option{-fPIC}, but
13472 generated position independent code can be only linked into executables.
13473 Usually these options are used when @option{-pie} GCC option will be
13474 used during linking.
13476 @item -fno-jump-tables
13477 @opindex fno-jump-tables
13478 Do not use jump tables for switch statements even where it would be
13479 more efficient than other code generation strategies. This option is
13480 of use in conjunction with @option{-fpic} or @option{-fPIC} for
13481 building code which forms part of a dynamic linker and cannot
13482 reference the address of a jump table. On some targets, jump tables
13483 do not require a GOT and this option is not needed.
13485 @item -ffixed-@var{reg}
13487 Treat the register named @var{reg} as a fixed register; generated code
13488 should never refer to it (except perhaps as a stack pointer, frame
13489 pointer or in some other fixed role).
13491 @var{reg} must be the name of a register. The register names accepted
13492 are machine-specific and are defined in the @code{REGISTER_NAMES}
13493 macro in the machine description macro file.
13495 This flag does not have a negative form, because it specifies a
13498 @item -fcall-used-@var{reg}
13499 @opindex fcall-used
13500 Treat the register named @var{reg} as an allocable register that is
13501 clobbered by function calls. It may be allocated for temporaries or
13502 variables that do not live across a call. Functions compiled this way
13503 will not save and restore the register @var{reg}.
13505 It is an error to used this flag with the frame pointer or stack pointer.
13506 Use of this flag for other registers that have fixed pervasive roles in
13507 the machine's execution model will produce disastrous results.
13509 This flag does not have a negative form, because it specifies a
13512 @item -fcall-saved-@var{reg}
13513 @opindex fcall-saved
13514 Treat the register named @var{reg} as an allocable register saved by
13515 functions. It may be allocated even for temporaries or variables that
13516 live across a call. Functions compiled this way will save and restore
13517 the register @var{reg} if they use it.
13519 It is an error to used this flag with the frame pointer or stack pointer.
13520 Use of this flag for other registers that have fixed pervasive roles in
13521 the machine's execution model will produce disastrous results.
13523 A different sort of disaster will result from the use of this flag for
13524 a register in which function values may be returned.
13526 This flag does not have a negative form, because it specifies a
13529 @item -fpack-struct[=@var{n}]
13530 @opindex fpack-struct
13531 Without a value specified, pack all structure members together without
13532 holes. When a value is specified (which must be a small power of two), pack
13533 structure members according to this value, representing the maximum
13534 alignment (that is, objects with default alignment requirements larger than
13535 this will be output potentially unaligned at the next fitting location.
13537 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
13538 code that is not binary compatible with code generated without that switch.
13539 Additionally, it makes the code suboptimal.
13540 Use it to conform to a non-default application binary interface.
13542 @item -finstrument-functions
13543 @opindex finstrument-functions
13544 Generate instrumentation calls for entry and exit to functions. Just
13545 after function entry and just before function exit, the following
13546 profiling functions will be called with the address of the current
13547 function and its call site. (On some platforms,
13548 @code{__builtin_return_address} does not work beyond the current
13549 function, so the call site information may not be available to the
13550 profiling functions otherwise.)
13553 void __cyg_profile_func_enter (void *this_fn,
13555 void __cyg_profile_func_exit (void *this_fn,
13559 The first argument is the address of the start of the current function,
13560 which may be looked up exactly in the symbol table.
13562 This instrumentation is also done for functions expanded inline in other
13563 functions. The profiling calls will indicate where, conceptually, the
13564 inline function is entered and exited. This means that addressable
13565 versions of such functions must be available. If all your uses of a
13566 function are expanded inline, this may mean an additional expansion of
13567 code size. If you use @samp{extern inline} in your C code, an
13568 addressable version of such functions must be provided. (This is
13569 normally the case anyways, but if you get lucky and the optimizer always
13570 expands the functions inline, you might have gotten away without
13571 providing static copies.)
13573 A function may be given the attribute @code{no_instrument_function}, in
13574 which case this instrumentation will not be done. This can be used, for
13575 example, for the profiling functions listed above, high-priority
13576 interrupt routines, and any functions from which the profiling functions
13577 cannot safely be called (perhaps signal handlers, if the profiling
13578 routines generate output or allocate memory).
13580 @item -fstack-check
13581 @opindex fstack-check
13582 Generate code to verify that you do not go beyond the boundary of the
13583 stack. You should specify this flag if you are running in an
13584 environment with multiple threads, but only rarely need to specify it in
13585 a single-threaded environment since stack overflow is automatically
13586 detected on nearly all systems if there is only one stack.
13588 Note that this switch does not actually cause checking to be done; the
13589 operating system must do that. The switch causes generation of code
13590 to ensure that the operating system sees the stack being extended.
13592 @item -fstack-limit-register=@var{reg}
13593 @itemx -fstack-limit-symbol=@var{sym}
13594 @itemx -fno-stack-limit
13595 @opindex fstack-limit-register
13596 @opindex fstack-limit-symbol
13597 @opindex fno-stack-limit
13598 Generate code to ensure that the stack does not grow beyond a certain value,
13599 either the value of a register or the address of a symbol. If the stack
13600 would grow beyond the value, a signal is raised. For most targets,
13601 the signal is raised before the stack overruns the boundary, so
13602 it is possible to catch the signal without taking special precautions.
13604 For instance, if the stack starts at absolute address @samp{0x80000000}
13605 and grows downwards, you can use the flags
13606 @option{-fstack-limit-symbol=__stack_limit} and
13607 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
13608 of 128KB@. Note that this may only work with the GNU linker.
13610 @cindex aliasing of parameters
13611 @cindex parameters, aliased
13612 @item -fargument-alias
13613 @itemx -fargument-noalias
13614 @itemx -fargument-noalias-global
13615 @itemx -fargument-noalias-anything
13616 @opindex fargument-alias
13617 @opindex fargument-noalias
13618 @opindex fargument-noalias-global
13619 @opindex fargument-noalias-anything
13620 Specify the possible relationships among parameters and between
13621 parameters and global data.
13623 @option{-fargument-alias} specifies that arguments (parameters) may
13624 alias each other and may alias global storage.@*
13625 @option{-fargument-noalias} specifies that arguments do not alias
13626 each other, but may alias global storage.@*
13627 @option{-fargument-noalias-global} specifies that arguments do not
13628 alias each other and do not alias global storage.
13629 @option{-fargument-noalias-anything} specifies that arguments do not
13630 alias any other storage.
13632 Each language will automatically use whatever option is required by
13633 the language standard. You should not need to use these options yourself.
13635 @item -fleading-underscore
13636 @opindex fleading-underscore
13637 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
13638 change the way C symbols are represented in the object file. One use
13639 is to help link with legacy assembly code.
13641 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
13642 generate code that is not binary compatible with code generated without that
13643 switch. Use it to conform to a non-default application binary interface.
13644 Not all targets provide complete support for this switch.
13646 @item -ftls-model=@var{model}
13647 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
13648 The @var{model} argument should be one of @code{global-dynamic},
13649 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
13651 The default without @option{-fpic} is @code{initial-exec}; with
13652 @option{-fpic} the default is @code{global-dynamic}.
13654 @item -fvisibility=@var{default|internal|hidden|protected}
13655 @opindex fvisibility
13656 Set the default ELF image symbol visibility to the specified option---all
13657 symbols will be marked with this unless overridden within the code.
13658 Using this feature can very substantially improve linking and
13659 load times of shared object libraries, produce more optimized
13660 code, provide near-perfect API export and prevent symbol clashes.
13661 It is @strong{strongly} recommended that you use this in any shared objects
13664 Despite the nomenclature, @code{default} always means public ie;
13665 available to be linked against from outside the shared object.
13666 @code{protected} and @code{internal} are pretty useless in real-world
13667 usage so the only other commonly used option will be @code{hidden}.
13668 The default if @option{-fvisibility} isn't specified is
13669 @code{default}, i.e., make every
13670 symbol public---this causes the same behavior as previous versions of
13673 A good explanation of the benefits offered by ensuring ELF
13674 symbols have the correct visibility is given by ``How To Write
13675 Shared Libraries'' by Ulrich Drepper (which can be found at
13676 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
13677 solution made possible by this option to marking things hidden when
13678 the default is public is to make the default hidden and mark things
13679 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
13680 and @code{__attribute__ ((visibility("default")))} instead of
13681 @code{__declspec(dllexport)} you get almost identical semantics with
13682 identical syntax. This is a great boon to those working with
13683 cross-platform projects.
13685 For those adding visibility support to existing code, you may find
13686 @samp{#pragma GCC visibility} of use. This works by you enclosing
13687 the declarations you wish to set visibility for with (for example)
13688 @samp{#pragma GCC visibility push(hidden)} and
13689 @samp{#pragma GCC visibility pop}.
13690 Bear in mind that symbol visibility should be viewed @strong{as
13691 part of the API interface contract} and thus all new code should
13692 always specify visibility when it is not the default ie; declarations
13693 only for use within the local DSO should @strong{always} be marked explicitly
13694 as hidden as so to avoid PLT indirection overheads---making this
13695 abundantly clear also aids readability and self-documentation of the code.
13696 Note that due to ISO C++ specification requirements, operator new and
13697 operator delete must always be of default visibility.
13699 Be aware that headers from outside your project, in particular system
13700 headers and headers from any other library you use, may not be
13701 expecting to be compiled with visibility other than the default. You
13702 may need to explicitly say @samp{#pragma GCC visibility push(default)}
13703 before including any such headers.
13705 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
13706 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
13707 no modifications. However, this means that calls to @samp{extern}
13708 functions with no explicit visibility will use the PLT, so it is more
13709 effective to use @samp{__attribute ((visibility))} and/or
13710 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
13711 declarations should be treated as hidden.
13713 Note that @samp{-fvisibility} does affect C++ vague linkage
13714 entities. This means that, for instance, an exception class that will
13715 be thrown between DSOs must be explicitly marked with default
13716 visibility so that the @samp{type_info} nodes will be unified between
13719 An overview of these techniques, their benefits and how to use them
13720 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
13726 @node Environment Variables
13727 @section Environment Variables Affecting GCC
13728 @cindex environment variables
13730 @c man begin ENVIRONMENT
13731 This section describes several environment variables that affect how GCC
13732 operates. Some of them work by specifying directories or prefixes to use
13733 when searching for various kinds of files. Some are used to specify other
13734 aspects of the compilation environment.
13736 Note that you can also specify places to search using options such as
13737 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
13738 take precedence over places specified using environment variables, which
13739 in turn take precedence over those specified by the configuration of GCC@.
13740 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
13741 GNU Compiler Collection (GCC) Internals}.
13746 @c @itemx LC_COLLATE
13748 @c @itemx LC_MONETARY
13749 @c @itemx LC_NUMERIC
13754 @c @findex LC_COLLATE
13755 @findex LC_MESSAGES
13756 @c @findex LC_MONETARY
13757 @c @findex LC_NUMERIC
13761 These environment variables control the way that GCC uses
13762 localization information that allow GCC to work with different
13763 national conventions. GCC inspects the locale categories
13764 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
13765 so. These locale categories can be set to any value supported by your
13766 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
13767 Kingdom encoded in UTF-8.
13769 The @env{LC_CTYPE} environment variable specifies character
13770 classification. GCC uses it to determine the character boundaries in
13771 a string; this is needed for some multibyte encodings that contain quote
13772 and escape characters that would otherwise be interpreted as a string
13775 The @env{LC_MESSAGES} environment variable specifies the language to
13776 use in diagnostic messages.
13778 If the @env{LC_ALL} environment variable is set, it overrides the value
13779 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
13780 and @env{LC_MESSAGES} default to the value of the @env{LANG}
13781 environment variable. If none of these variables are set, GCC
13782 defaults to traditional C English behavior.
13786 If @env{TMPDIR} is set, it specifies the directory to use for temporary
13787 files. GCC uses temporary files to hold the output of one stage of
13788 compilation which is to be used as input to the next stage: for example,
13789 the output of the preprocessor, which is the input to the compiler
13792 @item GCC_EXEC_PREFIX
13793 @findex GCC_EXEC_PREFIX
13794 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
13795 names of the subprograms executed by the compiler. No slash is added
13796 when this prefix is combined with the name of a subprogram, but you can
13797 specify a prefix that ends with a slash if you wish.
13799 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
13800 an appropriate prefix to use based on the pathname it was invoked with.
13802 If GCC cannot find the subprogram using the specified prefix, it
13803 tries looking in the usual places for the subprogram.
13805 The default value of @env{GCC_EXEC_PREFIX} is
13806 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
13807 of @code{prefix} when you ran the @file{configure} script.
13809 Other prefixes specified with @option{-B} take precedence over this prefix.
13811 This prefix is also used for finding files such as @file{crt0.o} that are
13814 In addition, the prefix is used in an unusual way in finding the
13815 directories to search for header files. For each of the standard
13816 directories whose name normally begins with @samp{/usr/local/lib/gcc}
13817 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
13818 replacing that beginning with the specified prefix to produce an
13819 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
13820 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
13821 These alternate directories are searched first; the standard directories
13824 @item COMPILER_PATH
13825 @findex COMPILER_PATH
13826 The value of @env{COMPILER_PATH} is a colon-separated list of
13827 directories, much like @env{PATH}. GCC tries the directories thus
13828 specified when searching for subprograms, if it can't find the
13829 subprograms using @env{GCC_EXEC_PREFIX}.
13832 @findex LIBRARY_PATH
13833 The value of @env{LIBRARY_PATH} is a colon-separated list of
13834 directories, much like @env{PATH}. When configured as a native compiler,
13835 GCC tries the directories thus specified when searching for special
13836 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
13837 using GCC also uses these directories when searching for ordinary
13838 libraries for the @option{-l} option (but directories specified with
13839 @option{-L} come first).
13843 @cindex locale definition
13844 This variable is used to pass locale information to the compiler. One way in
13845 which this information is used is to determine the character set to be used
13846 when character literals, string literals and comments are parsed in C and C++.
13847 When the compiler is configured to allow multibyte characters,
13848 the following values for @env{LANG} are recognized:
13852 Recognize JIS characters.
13854 Recognize SJIS characters.
13856 Recognize EUCJP characters.
13859 If @env{LANG} is not defined, or if it has some other value, then the
13860 compiler will use mblen and mbtowc as defined by the default locale to
13861 recognize and translate multibyte characters.
13865 Some additional environments variables affect the behavior of the
13868 @include cppenv.texi
13872 @node Precompiled Headers
13873 @section Using Precompiled Headers
13874 @cindex precompiled headers
13875 @cindex speed of compilation
13877 Often large projects have many header files that are included in every
13878 source file. The time the compiler takes to process these header files
13879 over and over again can account for nearly all of the time required to
13880 build the project. To make builds faster, GCC allows users to
13881 `precompile' a header file; then, if builds can use the precompiled
13882 header file they will be much faster.
13884 To create a precompiled header file, simply compile it as you would any
13885 other file, if necessary using the @option{-x} option to make the driver
13886 treat it as a C or C++ header file. You will probably want to use a
13887 tool like @command{make} to keep the precompiled header up-to-date when
13888 the headers it contains change.
13890 A precompiled header file will be searched for when @code{#include} is
13891 seen in the compilation. As it searches for the included file
13892 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
13893 compiler looks for a precompiled header in each directory just before it
13894 looks for the include file in that directory. The name searched for is
13895 the name specified in the @code{#include} with @samp{.gch} appended. If
13896 the precompiled header file can't be used, it is ignored.
13898 For instance, if you have @code{#include "all.h"}, and you have
13899 @file{all.h.gch} in the same directory as @file{all.h}, then the
13900 precompiled header file will be used if possible, and the original
13901 header will be used otherwise.
13903 Alternatively, you might decide to put the precompiled header file in a
13904 directory and use @option{-I} to ensure that directory is searched
13905 before (or instead of) the directory containing the original header.
13906 Then, if you want to check that the precompiled header file is always
13907 used, you can put a file of the same name as the original header in this
13908 directory containing an @code{#error} command.
13910 This also works with @option{-include}. So yet another way to use
13911 precompiled headers, good for projects not designed with precompiled
13912 header files in mind, is to simply take most of the header files used by
13913 a project, include them from another header file, precompile that header
13914 file, and @option{-include} the precompiled header. If the header files
13915 have guards against multiple inclusion, they will be skipped because
13916 they've already been included (in the precompiled header).
13918 If you need to precompile the same header file for different
13919 languages, targets, or compiler options, you can instead make a
13920 @emph{directory} named like @file{all.h.gch}, and put each precompiled
13921 header in the directory, perhaps using @option{-o}. It doesn't matter
13922 what you call the files in the directory, every precompiled header in
13923 the directory will be considered. The first precompiled header
13924 encountered in the directory that is valid for this compilation will
13925 be used; they're searched in no particular order.
13927 There are many other possibilities, limited only by your imagination,
13928 good sense, and the constraints of your build system.
13930 A precompiled header file can be used only when these conditions apply:
13934 Only one precompiled header can be used in a particular compilation.
13937 A precompiled header can't be used once the first C token is seen. You
13938 can have preprocessor directives before a precompiled header; you can
13939 even include a precompiled header from inside another header, so long as
13940 there are no C tokens before the @code{#include}.
13943 The precompiled header file must be produced for the same language as
13944 the current compilation. You can't use a C precompiled header for a C++
13948 The precompiled header file must have been produced by the same compiler
13949 binary as the current compilation is using.
13952 Any macros defined before the precompiled header is included must
13953 either be defined in the same way as when the precompiled header was
13954 generated, or must not affect the precompiled header, which usually
13955 means that they don't appear in the precompiled header at all.
13957 The @option{-D} option is one way to define a macro before a
13958 precompiled header is included; using a @code{#define} can also do it.
13959 There are also some options that define macros implicitly, like
13960 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
13963 @item If debugging information is output when using the precompiled
13964 header, using @option{-g} or similar, the same kind of debugging information
13965 must have been output when building the precompiled header. However,
13966 a precompiled header built using @option{-g} can be used in a compilation
13967 when no debugging information is being output.
13969 @item The same @option{-m} options must generally be used when building
13970 and using the precompiled header. @xref{Submodel Options},
13971 for any cases where this rule is relaxed.
13973 @item Each of the following options must be the same when building and using
13974 the precompiled header:
13976 @gccoptlist{-fexceptions -funit-at-a-time}
13979 Some other command-line options starting with @option{-f},
13980 @option{-p}, or @option{-O} must be defined in the same way as when
13981 the precompiled header was generated. At present, it's not clear
13982 which options are safe to change and which are not; the safest choice
13983 is to use exactly the same options when generating and using the
13984 precompiled header. The following are known to be safe:
13986 @gccoptlist{-fmessage-length= -fpreprocessed
13987 -fsched-interblock -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous
13988 -fsched-verbose=<number> -fschedule-insns -fvisibility=
13993 For all of these except the last, the compiler will automatically
13994 ignore the precompiled header if the conditions aren't met. If you
13995 find an option combination that doesn't work and doesn't cause the
13996 precompiled header to be ignored, please consider filing a bug report,
13999 If you do use differing options when generating and using the
14000 precompiled header, the actual behavior will be a mixture of the
14001 behavior for the options. For instance, if you use @option{-g} to
14002 generate the precompiled header but not when using it, you may or may
14003 not get debugging information for routines in the precompiled header.
14005 @node Running Protoize
14006 @section Running Protoize
14008 The program @code{protoize} is an optional part of GCC@. You can use
14009 it to add prototypes to a program, thus converting the program to ISO
14010 C in one respect. The companion program @code{unprotoize} does the
14011 reverse: it removes argument types from any prototypes that are found.
14013 When you run these programs, you must specify a set of source files as
14014 command line arguments. The conversion programs start out by compiling
14015 these files to see what functions they define. The information gathered
14016 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
14018 After scanning comes actual conversion. The specified files are all
14019 eligible to be converted; any files they include (whether sources or
14020 just headers) are eligible as well.
14022 But not all the eligible files are converted. By default,
14023 @code{protoize} and @code{unprotoize} convert only source and header
14024 files in the current directory. You can specify additional directories
14025 whose files should be converted with the @option{-d @var{directory}}
14026 option. You can also specify particular files to exclude with the
14027 @option{-x @var{file}} option. A file is converted if it is eligible, its
14028 directory name matches one of the specified directory names, and its
14029 name within the directory has not been excluded.
14031 Basic conversion with @code{protoize} consists of rewriting most
14032 function definitions and function declarations to specify the types of
14033 the arguments. The only ones not rewritten are those for varargs
14036 @code{protoize} optionally inserts prototype declarations at the
14037 beginning of the source file, to make them available for any calls that
14038 precede the function's definition. Or it can insert prototype
14039 declarations with block scope in the blocks where undeclared functions
14042 Basic conversion with @code{unprotoize} consists of rewriting most
14043 function declarations to remove any argument types, and rewriting
14044 function definitions to the old-style pre-ISO form.
14046 Both conversion programs print a warning for any function declaration or
14047 definition that they can't convert. You can suppress these warnings
14050 The output from @code{protoize} or @code{unprotoize} replaces the
14051 original source file. The original file is renamed to a name ending
14052 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
14053 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
14054 for DOS) file already exists, then the source file is simply discarded.
14056 @code{protoize} and @code{unprotoize} both depend on GCC itself to
14057 scan the program and collect information about the functions it uses.
14058 So neither of these programs will work until GCC is installed.
14060 Here is a table of the options you can use with @code{protoize} and
14061 @code{unprotoize}. Each option works with both programs unless
14065 @item -B @var{directory}
14066 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
14067 usual directory (normally @file{/usr/local/lib}). This file contains
14068 prototype information about standard system functions. This option
14069 applies only to @code{protoize}.
14071 @item -c @var{compilation-options}
14072 Use @var{compilation-options} as the options when running @command{gcc} to
14073 produce the @samp{.X} files. The special option @option{-aux-info} is
14074 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
14076 Note that the compilation options must be given as a single argument to
14077 @code{protoize} or @code{unprotoize}. If you want to specify several
14078 @command{gcc} options, you must quote the entire set of compilation options
14079 to make them a single word in the shell.
14081 There are certain @command{gcc} arguments that you cannot use, because they
14082 would produce the wrong kind of output. These include @option{-g},
14083 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
14084 the @var{compilation-options}, they are ignored.
14087 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
14088 systems) instead of @samp{.c}. This is convenient if you are converting
14089 a C program to C++. This option applies only to @code{protoize}.
14092 Add explicit global declarations. This means inserting explicit
14093 declarations at the beginning of each source file for each function
14094 that is called in the file and was not declared. These declarations
14095 precede the first function definition that contains a call to an
14096 undeclared function. This option applies only to @code{protoize}.
14098 @item -i @var{string}
14099 Indent old-style parameter declarations with the string @var{string}.
14100 This option applies only to @code{protoize}.
14102 @code{unprotoize} converts prototyped function definitions to old-style
14103 function definitions, where the arguments are declared between the
14104 argument list and the initial @samp{@{}. By default, @code{unprotoize}
14105 uses five spaces as the indentation. If you want to indent with just
14106 one space instead, use @option{-i " "}.
14109 Keep the @samp{.X} files. Normally, they are deleted after conversion
14113 Add explicit local declarations. @code{protoize} with @option{-l} inserts
14114 a prototype declaration for each function in each block which calls the
14115 function without any declaration. This option applies only to
14119 Make no real changes. This mode just prints information about the conversions
14120 that would have been done without @option{-n}.
14123 Make no @samp{.save} files. The original files are simply deleted.
14124 Use this option with caution.
14126 @item -p @var{program}
14127 Use the program @var{program} as the compiler. Normally, the name
14128 @file{gcc} is used.
14131 Work quietly. Most warnings are suppressed.
14134 Print the version number, just like @option{-v} for @command{gcc}.
14137 If you need special compiler options to compile one of your program's
14138 source files, then you should generate that file's @samp{.X} file
14139 specially, by running @command{gcc} on that source file with the
14140 appropriate options and the option @option{-aux-info}. Then run
14141 @code{protoize} on the entire set of files. @code{protoize} will use
14142 the existing @samp{.X} file because it is newer than the source file.
14146 gcc -Dfoo=bar file1.c -aux-info file1.X
14151 You need to include the special files along with the rest in the
14152 @code{protoize} command, even though their @samp{.X} files already
14153 exist, because otherwise they won't get converted.
14155 @xref{Protoize Caveats}, for more information on how to use
14156 @code{protoize} successfully.