1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006 Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
11 @c man begin COPYRIGHT
12 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
13 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006 Free Software Foundation, Inc.
15 Permission is granted to copy, distribute and/or modify this document
16 under the terms of the GNU Free Documentation License, Version 1.2 or
17 any later version published by the Free Software Foundation; with the
18 Invariant Sections being ``GNU General Public License'' and ``Funding
19 Free Software'', the Front-Cover texts being (a) (see below), and with
20 the Back-Cover Texts being (b) (see below). A copy of the license is
21 included in the gfdl(7) man page.
23 (a) The FSF's Front-Cover Text is:
27 (b) The FSF's Back-Cover Text is:
29 You have freedom to copy and modify this GNU Manual, like GNU
30 software. Copies published by the Free Software Foundation raise
31 funds for GNU development.
33 @c Set file name and title for the man page.
35 @settitle GNU project C and C++ compiler
37 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
38 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
39 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
40 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
41 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
42 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
43 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
45 Only the most useful options are listed here; see below for the
46 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
49 gpl(7), gfdl(7), fsf-funding(7),
50 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
51 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
52 @file{ld}, @file{binutils} and @file{gdb}.
55 For instructions on reporting bugs, see
56 @w{@uref{http://gcc.gnu.org/bugs.html}}.
59 See the Info entry for @command{gcc}, or
60 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
61 for contributors to GCC@.
66 @chapter GCC Command Options
67 @cindex GCC command options
68 @cindex command options
69 @cindex options, GCC command
71 @c man begin DESCRIPTION
72 When you invoke GCC, it normally does preprocessing, compilation,
73 assembly and linking. The ``overall options'' allow you to stop this
74 process at an intermediate stage. For example, the @option{-c} option
75 says not to run the linker. Then the output consists of object files
76 output by the assembler.
78 Other options are passed on to one stage of processing. Some options
79 control the preprocessor and others the compiler itself. Yet other
80 options control the assembler and linker; most of these are not
81 documented here, since you rarely need to use any of them.
83 @cindex C compilation options
84 Most of the command line options that you can use with GCC are useful
85 for C programs; when an option is only useful with another language
86 (usually C++), the explanation says so explicitly. If the description
87 for a particular option does not mention a source language, you can use
88 that option with all supported languages.
90 @cindex C++ compilation options
91 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
92 options for compiling C++ programs.
94 @cindex grouping options
95 @cindex options, grouping
96 The @command{gcc} program accepts options and file names as operands. Many
97 options have multi-letter names; therefore multiple single-letter options
98 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
101 @cindex order of options
102 @cindex options, order
103 You can mix options and other arguments. For the most part, the order
104 you use doesn't matter. Order does matter when you use several options
105 of the same kind; for example, if you specify @option{-L} more than once,
106 the directories are searched in the order specified.
108 Many options have long names starting with @samp{-f} or with
109 @samp{-W}---for example,
110 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
111 these have both positive and negative forms; the negative form of
112 @option{-ffoo} would be @option{-fno-foo}. This manual documents
113 only one of these two forms, whichever one is not the default.
117 @xref{Option Index}, for an index to GCC's options.
120 * Option Summary:: Brief list of all options, without explanations.
121 * Overall Options:: Controlling the kind of output:
122 an executable, object files, assembler files,
123 or preprocessed source.
124 * Invoking G++:: Compiling C++ programs.
125 * C Dialect Options:: Controlling the variant of C language compiled.
126 * C++ Dialect Options:: Variations on C++.
127 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
129 * Language Independent Options:: Controlling how diagnostics should be
131 * Warning Options:: How picky should the compiler be?
132 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
133 * Optimize Options:: How much optimization?
134 * Preprocessor Options:: Controlling header files and macro definitions.
135 Also, getting dependency information for Make.
136 * Assembler Options:: Passing options to the assembler.
137 * Link Options:: Specifying libraries and so on.
138 * Directory Options:: Where to find header files and libraries.
139 Where to find the compiler executable files.
140 * Spec Files:: How to pass switches to sub-processes.
141 * Target Options:: Running a cross-compiler, or an old version of GCC.
142 * Submodel Options:: Specifying minor hardware or convention variations,
143 such as 68010 vs 68020.
144 * Code Gen Options:: Specifying conventions for function calls, data layout
146 * Environment Variables:: Env vars that affect GCC.
147 * Precompiled Headers:: Compiling a header once, and using it many times.
148 * Running Protoize:: Automatically adding or removing function prototypes.
154 @section Option Summary
156 Here is a summary of all the options, grouped by type. Explanations are
157 in the following sections.
160 @item Overall Options
161 @xref{Overall Options,,Options Controlling the Kind of Output}.
162 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
163 -x @var{language} -v -### --help --target-help --version @@@var{file}}
165 @item C Language Options
166 @xref{C Dialect Options,,Options Controlling C Dialect}.
167 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
168 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
169 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
170 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
171 -fallow-single-precision -fcond-mismatch @gol
172 -fsigned-bitfields -fsigned-char @gol
173 -funsigned-bitfields -funsigned-char}
175 @item C++ Language Options
176 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
177 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
178 -fconserve-space -ffriend-injection @gol
179 -fno-elide-constructors @gol
180 -fno-enforce-eh-specs @gol
181 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
182 -fno-implicit-templates @gol
183 -fno-implicit-inline-templates @gol
184 -fno-implement-inlines -fms-extensions @gol
185 -fno-nonansi-builtins -fno-operator-names @gol
186 -fno-optional-diags -fpermissive @gol
187 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
188 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
189 -fno-default-inline -fvisibility-inlines-hidden @gol
190 -Wabi -Wctor-dtor-privacy @gol
191 -Wnon-virtual-dtor -Wreorder @gol
192 -Weffc++ -Wno-deprecated -Wstrict-null-sentinel @gol
193 -Wno-non-template-friend -Wold-style-cast @gol
194 -Woverloaded-virtual -Wno-pmf-conversions @gol
197 @item Objective-C and Objective-C++ Language Options
198 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
199 Objective-C and Objective-C++ Dialects}.
201 -fconstant-string-class=@var{class-name} @gol
202 -fgnu-runtime -fnext-runtime @gol
203 -fno-nil-receivers @gol
204 -fobjc-call-cxx-cdtors @gol
205 -fobjc-direct-dispatch @gol
206 -fobjc-exceptions @gol
208 -freplace-objc-classes @gol
211 -Wassign-intercept @gol
212 -Wno-protocol -Wselector @gol
213 -Wstrict-selector-match @gol
214 -Wundeclared-selector}
216 @item Language Independent Options
217 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
218 @gccoptlist{-fmessage-length=@var{n} @gol
219 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}} @gol
220 -fdiagnostics-show-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 -Waggregate-return -Walways-true -Wno-attributes @gol
226 -Wc++-compat -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
227 -Wconversion -Wno-deprecated-declarations @gol
228 -Wdisabled-optimization -Wno-div-by-zero -Wno-endif-labels @gol
229 -Werror -Werror-* -Werror-implicit-function-declaration @gol
230 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
231 -Wno-format-extra-args -Wformat-nonliteral @gol
232 -Wformat-security -Wformat-y2k @gol
233 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
234 -Wimport -Wno-import -Winit-self -Winline @gol
235 -Wno-int-to-pointer-cast @gol
236 -Wno-invalid-offsetof -Winvalid-pch @gol
237 -Wlarger-than-@var{len} -Wunsafe-loop-optimizations -Wlong-long @gol
238 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
239 -Wmissing-format-attribute -Wmissing-include-dirs @gol
240 -Wmissing-noreturn @gol
241 -Wno-multichar -Wnonnull -Wno-overflow @gol
242 -Woverlength-strings -Wpacked -Wpadded @gol
243 -Wparentheses -Wpointer-arith -Wno-pointer-to-int-cast @gol
244 -Wredundant-decls @gol
245 -Wreturn-type -Wsequence-point -Wshadow @gol
246 -Wsign-compare -Wstack-protector @gol
247 -Wstrict-aliasing -Wstrict-aliasing=2 @gol
248 -Wstring-literal-comparison @gol
249 -Wswitch -Wswitch-default -Wswitch-enum @gol
250 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
251 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
252 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
253 -Wunused-value -Wunused-variable -Wvariadic-macros @gol
254 -Wvolatile-register-var -Wwrite-strings}
256 @item C-only Warning Options
257 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
258 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
259 -Wstrict-prototypes -Wtraditional @gol
260 -Wdeclaration-after-statement -Wpointer-sign}
262 @item Debugging Options
263 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
264 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
265 -fdump-noaddr -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
266 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
267 -fdump-ipa-all -fdump-ipa-cgraph @gol
269 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
270 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
271 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
272 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
274 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
275 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
276 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
277 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
278 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
279 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
280 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
281 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
282 -fdump-tree-nrv -fdump-tree-vect @gol
283 -fdump-tree-sink @gol
284 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
285 -fdump-tree-salias @gol
286 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
288 -ftree-vectorizer-verbose=@var{n} @gol
289 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
290 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
291 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
292 -fmem-report -fprofile-arcs @gol
293 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
294 -ftest-coverage -ftime-report -fvar-tracking @gol
295 -g -g@var{level} -gcoff -gdwarf-2 @gol
296 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
297 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
298 -print-multi-directory -print-multi-lib @gol
299 -print-prog-name=@var{program} -print-search-dirs -Q @gol
302 @item Optimization Options
303 @xref{Optimize Options,,Options that Control Optimization}.
304 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
305 -falign-labels=@var{n} -falign-loops=@var{n} @gol
306 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
307 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
308 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
309 -fcaller-saves -fcprop-registers -fcse-follow-jumps @gol
310 -fcse-skip-blocks -fcx-limited-range -fdata-sections @gol
311 -fdelayed-branch -fdelete-null-pointer-checks -fearly-inlining @gol
312 -fexpensive-optimizations -ffast-math -ffloat-store @gol
313 -fforce-addr -ffunction-sections @gol
314 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
315 -fcrossjumping -fif-conversion -fif-conversion2 @gol
316 -finline-functions -finline-functions-called-once @gol
317 -finline-limit=@var{n} -fkeep-inline-functions @gol
318 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
319 -fmodulo-sched -fno-branch-count-reg @gol
320 -fno-default-inline -fno-defer-pop -fmove-loop-invariants @gol
321 -fno-function-cse -fno-guess-branch-probability @gol
322 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
323 -funsafe-math-optimizations -funsafe-loop-optimizations -ffinite-math-only @gol
324 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
325 -fomit-frame-pointer -foptimize-register-move @gol
326 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
327 -fprofile-generate -fprofile-use @gol
328 -fregmove -frename-registers @gol
329 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
330 -frerun-cse-after-loop @gol
331 -frounding-math -frtl-abstract-sequences @gol
332 -fschedule-insns -fschedule-insns2 @gol
333 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
334 -fsched-spec-load-dangerous @gol
335 -fsched-stalled-insns=@var{n} -fsched-stalled-insns-dep=@var{n} @gol
336 -fsched2-use-superblocks @gol
337 -fsched2-use-traces -fsee -freschedule-modulo-scheduled-loops @gol
338 -fsection-anchors -fsignaling-nans -fsingle-precision-constant @gol
339 -fstack-protector -fstack-protector-all @gol
340 -fstrict-aliasing -ftracer -fthread-jumps @gol
341 -funroll-all-loops -funroll-loops -fpeel-loops @gol
342 -fsplit-ivs-in-unroller -funswitch-loops @gol
343 -fvariable-expansion-in-unroller @gol
344 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
345 -ftree-loop-linear -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
346 -ftree-dominator-opts -ftree-dse -ftree-copyrename -ftree-sink @gol
347 -ftree-ch -ftree-sra -ftree-ter -ftree-lrs -ftree-fre -ftree-vectorize @gol
348 -ftree-vect-loop-version -ftree-salias -fipa-pta -fweb @gol
349 -ftree-copy-prop -ftree-store-ccp -ftree-store-copy-prop -fwhole-program @gol
350 --param @var{name}=@var{value}
351 -O -O0 -O1 -O2 -O3 -Os}
353 @item Preprocessor Options
354 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
355 @gccoptlist{-A@var{question}=@var{answer} @gol
356 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
357 -C -dD -dI -dM -dN @gol
358 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
359 -idirafter @var{dir} @gol
360 -include @var{file} -imacros @var{file} @gol
361 -iprefix @var{file} -iwithprefix @var{dir} @gol
362 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
363 -imultilib @var{dir} -isysroot @var{dir} @gol
364 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
365 -P -fworking-directory -remap @gol
366 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
367 -Xpreprocessor @var{option}}
369 @item Assembler Option
370 @xref{Assembler Options,,Passing Options to the Assembler}.
371 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
374 @xref{Link Options,,Options for Linking}.
375 @gccoptlist{@var{object-file-name} -l@var{library} @gol
376 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
377 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
378 -Wl,@var{option} -Xlinker @var{option} @gol
381 @item Directory Options
382 @xref{Directory Options,,Options for Directory Search}.
383 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
384 -specs=@var{file} -I- --sysroot=@var{dir}}
387 @c I wrote this xref this way to avoid overfull hbox. -- rms
388 @xref{Target Options}.
389 @gccoptlist{-V @var{version} -b @var{machine}}
391 @item Machine Dependent Options
392 @xref{Submodel Options,,Hardware Models and Configurations}.
393 @c This list is ordered alphanumerically by subsection name.
394 @c Try and put the significant identifier (CPU or system) first,
395 @c so users have a clue at guessing where the ones they want will be.
398 @gccoptlist{-EB -EL @gol
399 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
400 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
403 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
404 -mabi=@var{name} @gol
405 -mapcs-stack-check -mno-apcs-stack-check @gol
406 -mapcs-float -mno-apcs-float @gol
407 -mapcs-reentrant -mno-apcs-reentrant @gol
408 -msched-prolog -mno-sched-prolog @gol
409 -mlittle-endian -mbig-endian -mwords-little-endian @gol
410 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
411 -mthumb-interwork -mno-thumb-interwork @gol
412 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
413 -mstructure-size-boundary=@var{n} @gol
414 -mabort-on-noreturn @gol
415 -mlong-calls -mno-long-calls @gol
416 -msingle-pic-base -mno-single-pic-base @gol
417 -mpic-register=@var{reg} @gol
418 -mnop-fun-dllimport @gol
419 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
420 -mpoke-function-name @gol
422 -mtpcs-frame -mtpcs-leaf-frame @gol
423 -mcaller-super-interworking -mcallee-super-interworking @gol
427 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
428 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
430 @emph{Blackfin Options}
431 @gccoptlist{-momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
432 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
433 -mlow-64k -mno-low64k -mid-shared-library @gol
434 -mno-id-shared-library -mshared-library-id=@var{n} @gol
435 -mlong-calls -mno-long-calls}
438 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
439 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
440 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
441 -mstack-align -mdata-align -mconst-align @gol
442 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
443 -melf -maout -melinux -mlinux -sim -sim2 @gol
444 -mmul-bug-workaround -mno-mul-bug-workaround}
447 @gccoptlist{-mmac -mpush-args}
449 @emph{Darwin Options}
450 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
451 -arch_only -bind_at_load -bundle -bundle_loader @gol
452 -client_name -compatibility_version -current_version @gol
454 -dependency-file -dylib_file -dylinker_install_name @gol
455 -dynamic -dynamiclib -exported_symbols_list @gol
456 -filelist -flat_namespace -force_cpusubtype_ALL @gol
457 -force_flat_namespace -headerpad_max_install_names @gol
458 -image_base -init -install_name -keep_private_externs @gol
459 -multi_module -multiply_defined -multiply_defined_unused @gol
460 -noall_load -no_dead_strip_inits_and_terms @gol
461 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
462 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
463 -private_bundle -read_only_relocs -sectalign @gol
464 -sectobjectsymbols -whyload -seg1addr @gol
465 -sectcreate -sectobjectsymbols -sectorder @gol
466 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
467 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
468 -segprot -segs_read_only_addr -segs_read_write_addr @gol
469 -single_module -static -sub_library -sub_umbrella @gol
470 -twolevel_namespace -umbrella -undefined @gol
471 -unexported_symbols_list -weak_reference_mismatches @gol
472 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
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 -mssse3 -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 -mmfpgpr -mno-mfpgpr @gol
660 -mnew-mnemonics -mold-mnemonics @gol
661 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
662 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
663 -malign-power -malign-natural @gol
664 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
665 -mstring -mno-string -mupdate -mno-update @gol
666 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
667 -mstrict-align -mno-strict-align -mrelocatable @gol
668 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
669 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
670 -mdynamic-no-pic -maltivec -mswdiv @gol
671 -mprioritize-restricted-insns=@var{priority} @gol
672 -msched-costly-dep=@var{dependence_type} @gol
673 -minsert-sched-nops=@var{scheme} @gol
674 -mcall-sysv -mcall-netbsd @gol
675 -maix-struct-return -msvr4-struct-return @gol
676 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
677 -misel -mno-isel @gol
678 -misel=yes -misel=no @gol
680 -mspe=yes -mspe=no @gol
681 -mvrsave -mno-vrsave @gol
682 -mmulhw -mno-mulhw @gol
683 -mdlmzb -mno-dlmzb @gol
684 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
685 -mprototype -mno-prototype @gol
686 -msim -mmvme -mads -myellowknife -memb -msdata @gol
687 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
689 @emph{S/390 and zSeries Options}
690 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
691 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
692 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
693 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
694 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
695 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
696 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
699 @gccoptlist{-mel -mel @gol
704 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
705 -m4-nofpu -m4-single-only -m4-single -m4 @gol
706 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
707 -m5-64media -m5-64media-nofpu @gol
708 -m5-32media -m5-32media-nofpu @gol
709 -m5-compact -m5-compact-nofpu @gol
710 -mb -ml -mdalign -mrelax @gol
711 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
712 -mieee -misize -mpadstruct -mspace @gol
713 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
714 -mdivsi3_libfunc=@var{name} @gol
715 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
719 @gccoptlist{-mcpu=@var{cpu-type} @gol
720 -mtune=@var{cpu-type} @gol
721 -mcmodel=@var{code-model} @gol
722 -m32 -m64 -mapp-regs -mno-app-regs @gol
723 -mfaster-structs -mno-faster-structs @gol
724 -mfpu -mno-fpu -mhard-float -msoft-float @gol
725 -mhard-quad-float -msoft-quad-float @gol
726 -mimpure-text -mno-impure-text -mlittle-endian @gol
727 -mstack-bias -mno-stack-bias @gol
728 -munaligned-doubles -mno-unaligned-doubles @gol
729 -mv8plus -mno-v8plus -mvis -mno-vis
730 -threads -pthreads -pthread}
732 @emph{System V Options}
733 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
735 @emph{TMS320C3x/C4x Options}
736 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
737 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
738 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
739 -mparallel-insns -mparallel-mpy -mpreserve-float}
742 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
743 -mprolog-function -mno-prolog-function -mspace @gol
744 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
745 -mapp-regs -mno-app-regs @gol
746 -mdisable-callt -mno-disable-callt @gol
752 @gccoptlist{-mg -mgnu -munix}
754 @emph{x86-64 Options}
755 See i386 and x86-64 Options.
757 @emph{Xstormy16 Options}
760 @emph{Xtensa Options}
761 @gccoptlist{-mconst16 -mno-const16 @gol
762 -mfused-madd -mno-fused-madd @gol
763 -mtext-section-literals -mno-text-section-literals @gol
764 -mtarget-align -mno-target-align @gol
765 -mlongcalls -mno-longcalls}
767 @emph{zSeries Options}
768 See S/390 and zSeries Options.
770 @item Code Generation Options
771 @xref{Code Gen Options,,Options for Code Generation Conventions}.
772 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
773 -ffixed-@var{reg} -fexceptions @gol
774 -fnon-call-exceptions -funwind-tables @gol
775 -fasynchronous-unwind-tables @gol
776 -finhibit-size-directive -finstrument-functions @gol
777 -fno-common -fno-ident @gol
778 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
779 -fno-jump-tables @gol
780 -freg-struct-return -fshort-enums @gol
781 -fshort-double -fshort-wchar @gol
782 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
783 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
784 -fargument-alias -fargument-noalias @gol
785 -fargument-noalias-global -fargument-noalias-anything
786 -fleading-underscore -ftls-model=@var{model} @gol
787 -ftrapv -fwrapv -fbounds-check @gol
792 * Overall Options:: Controlling the kind of output:
793 an executable, object files, assembler files,
794 or preprocessed source.
795 * C Dialect Options:: Controlling the variant of C language compiled.
796 * C++ Dialect Options:: Variations on C++.
797 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
799 * Language Independent Options:: Controlling how diagnostics should be
801 * Warning Options:: How picky should the compiler be?
802 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
803 * Optimize Options:: How much optimization?
804 * Preprocessor Options:: Controlling header files and macro definitions.
805 Also, getting dependency information for Make.
806 * Assembler Options:: Passing options to the assembler.
807 * Link Options:: Specifying libraries and so on.
808 * Directory Options:: Where to find header files and libraries.
809 Where to find the compiler executable files.
810 * Spec Files:: How to pass switches to sub-processes.
811 * Target Options:: Running a cross-compiler, or an old version of GCC.
814 @node Overall Options
815 @section Options Controlling the Kind of Output
817 Compilation can involve up to four stages: preprocessing, compilation
818 proper, assembly and linking, always in that order. GCC is capable of
819 preprocessing and compiling several files either into several
820 assembler input files, or into one assembler input file; then each
821 assembler input file produces an object file, and linking combines all
822 the object files (those newly compiled, and those specified as input)
823 into an executable file.
825 @cindex file name suffix
826 For any given input file, the file name suffix determines what kind of
831 C source code which must be preprocessed.
834 C source code which should not be preprocessed.
837 C++ source code which should not be preprocessed.
840 Objective-C source code. Note that you must link with the @file{libobjc}
841 library to make an Objective-C program work.
844 Objective-C source code which should not be preprocessed.
848 Objective-C++ source code. Note that you must link with the @file{libobjc}
849 library to make an Objective-C++ program work. Note that @samp{.M} refers
850 to a literal capital M@.
853 Objective-C++ source code which should not be preprocessed.
856 C, C++, Objective-C or Objective-C++ header file to be turned into a
861 @itemx @var{file}.cxx
862 @itemx @var{file}.cpp
863 @itemx @var{file}.CPP
864 @itemx @var{file}.c++
866 C++ source code which must be preprocessed. Note that in @samp{.cxx},
867 the last two letters must both be literally @samp{x}. Likewise,
868 @samp{.C} refers to a literal capital C@.
872 Objective-C++ source code which must be preprocessed.
875 Objective-C++ source code which should not be preprocessed.
879 C++ header file to be turned into a precompiled header.
882 @itemx @var{file}.for
883 @itemx @var{file}.FOR
884 Fixed form Fortran source code which should not be preprocessed.
887 @itemx @var{file}.fpp
888 @itemx @var{file}.FPP
889 Fixed form Fortran source code which must be preprocessed (with the traditional
893 @itemx @var{file}.f95
894 Free form Fortran source code which should not be preprocessed.
897 @itemx @var{file}.F95
898 Free form Fortran source code which must be preprocessed (with the
899 traditional preprocessor).
901 @c FIXME: Descriptions of Java file types.
908 Ada source code file which contains a library unit declaration (a
909 declaration of a package, subprogram, or generic, or a generic
910 instantiation), or a library unit renaming declaration (a package,
911 generic, or subprogram renaming declaration). Such files are also
914 @itemx @var{file}.adb
915 Ada source code file containing a library unit body (a subprogram or
916 package body). Such files are also called @dfn{bodies}.
918 @c GCC also knows about some suffixes for languages not yet included:
929 Assembler code which must be preprocessed.
932 An object file to be fed straight into linking.
933 Any file name with no recognized suffix is treated this way.
937 You can specify the input language explicitly with the @option{-x} option:
940 @item -x @var{language}
941 Specify explicitly the @var{language} for the following input files
942 (rather than letting the compiler choose a default based on the file
943 name suffix). This option applies to all following input files until
944 the next @option{-x} option. Possible values for @var{language} are:
946 c c-header c-cpp-output
947 c++ c++-header c++-cpp-output
948 objective-c objective-c-header objective-c-cpp-output
949 objective-c++ objective-c++-header objective-c++-cpp-output
950 assembler assembler-with-cpp
958 Turn off any specification of a language, so that subsequent files are
959 handled according to their file name suffixes (as they are if @option{-x}
960 has not been used at all).
962 @item -pass-exit-codes
963 @opindex pass-exit-codes
964 Normally the @command{gcc} program will exit with the code of 1 if any
965 phase of the compiler returns a non-success return code. If you specify
966 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
967 numerically highest error produced by any phase that returned an error
968 indication. The C, C++, and Fortran frontends return 4, if an internal
969 compiler error is encountered.
972 If you only want some of the stages of compilation, you can use
973 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
974 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
975 @command{gcc} is to stop. Note that some combinations (for example,
976 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
981 Compile or assemble the source files, but do not link. The linking
982 stage simply is not done. The ultimate output is in the form of an
983 object file for each source file.
985 By default, the object file name for a source file is made by replacing
986 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
988 Unrecognized input files, not requiring compilation or assembly, are
993 Stop after the stage of compilation proper; do not assemble. The output
994 is in the form of an assembler code file for each non-assembler input
997 By default, the assembler file name for a source file is made by
998 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1000 Input files that don't require compilation are ignored.
1004 Stop after the preprocessing stage; do not run the compiler proper. The
1005 output is in the form of preprocessed source code, which is sent to the
1008 Input files which don't require preprocessing are ignored.
1010 @cindex output file option
1013 Place output in file @var{file}. This applies regardless to whatever
1014 sort of output is being produced, whether it be an executable file,
1015 an object file, an assembler file or preprocessed C code.
1017 If @option{-o} is not specified, the default is to put an executable
1018 file in @file{a.out}, the object file for
1019 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1020 assembler file in @file{@var{source}.s}, a precompiled header file in
1021 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1026 Print (on standard error output) the commands executed to run the stages
1027 of compilation. Also print the version number of the compiler driver
1028 program and of the preprocessor and the compiler proper.
1032 Like @option{-v} except the commands are not executed and all command
1033 arguments are quoted. This is useful for shell scripts to capture the
1034 driver-generated command lines.
1038 Use pipes rather than temporary files for communication between the
1039 various stages of compilation. This fails to work on some systems where
1040 the assembler is unable to read from a pipe; but the GNU assembler has
1045 If you are compiling multiple source files, this option tells the driver
1046 to pass all the source files to the compiler at once (for those
1047 languages for which the compiler can handle this). This will allow
1048 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1049 language for which this is supported is C@. If you pass source files for
1050 multiple languages to the driver, using this option, the driver will invoke
1051 the compiler(s) that support IMA once each, passing each compiler all the
1052 source files appropriate for it. For those languages that do not support
1053 IMA this option will be ignored, and the compiler will be invoked once for
1054 each source file in that language. If you use this option in conjunction
1055 with @option{-save-temps}, the compiler will generate multiple
1057 (one for each source file), but only one (combined) @file{.o} or
1062 Print (on the standard output) a description of the command line options
1063 understood by @command{gcc}. If the @option{-v} option is also specified
1064 then @option{--help} will also be passed on to the various processes
1065 invoked by @command{gcc}, so that they can display the command line options
1066 they accept. If the @option{-Wextra} option is also specified then command
1067 line options which have no documentation associated with them will also
1071 @opindex target-help
1072 Print (on the standard output) a description of target specific command
1073 line options for each tool.
1077 Display the version number and copyrights of the invoked GCC@.
1079 @include @value{srcdir}/../libiberty/at-file.texi
1083 @section Compiling C++ Programs
1085 @cindex suffixes for C++ source
1086 @cindex C++ source file suffixes
1087 C++ source files conventionally use one of the suffixes @samp{.C},
1088 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1089 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1090 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1091 files with these names and compiles them as C++ programs even if you
1092 call the compiler the same way as for compiling C programs (usually
1093 with the name @command{gcc}).
1097 However, the use of @command{gcc} does not add the C++ library.
1098 @command{g++} is a program that calls GCC and treats @samp{.c},
1099 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1100 files unless @option{-x} is used, and automatically specifies linking
1101 against the C++ library. This program is also useful when
1102 precompiling a C header file with a @samp{.h} extension for use in C++
1103 compilations. On many systems, @command{g++} is also installed with
1104 the name @command{c++}.
1106 @cindex invoking @command{g++}
1107 When you compile C++ programs, you may specify many of the same
1108 command-line options that you use for compiling programs in any
1109 language; or command-line options meaningful for C and related
1110 languages; or options that are meaningful only for C++ programs.
1111 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1112 explanations of options for languages related to C@.
1113 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1114 explanations of options that are meaningful only for C++ programs.
1116 @node C Dialect Options
1117 @section Options Controlling C Dialect
1118 @cindex dialect options
1119 @cindex language dialect options
1120 @cindex options, dialect
1122 The following options control the dialect of C (or languages derived
1123 from C, such as C++, Objective-C and Objective-C++) that the compiler
1127 @cindex ANSI support
1131 In C mode, support all ISO C90 programs. In C++ mode,
1132 remove GNU extensions that conflict with ISO C++.
1134 This turns off certain features of GCC that are incompatible with ISO
1135 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1136 such as the @code{asm} and @code{typeof} keywords, and
1137 predefined macros such as @code{unix} and @code{vax} that identify the
1138 type of system you are using. It also enables the undesirable and
1139 rarely used ISO trigraph feature. For the C compiler,
1140 it disables recognition of C++ style @samp{//} comments as well as
1141 the @code{inline} keyword.
1143 The alternate keywords @code{__asm__}, @code{__extension__},
1144 @code{__inline__} and @code{__typeof__} continue to work despite
1145 @option{-ansi}. You would not want to use them in an ISO C program, of
1146 course, but it is useful to put them in header files that might be included
1147 in compilations done with @option{-ansi}. Alternate predefined macros
1148 such as @code{__unix__} and @code{__vax__} are also available, with or
1149 without @option{-ansi}.
1151 The @option{-ansi} option does not cause non-ISO programs to be
1152 rejected gratuitously. For that, @option{-pedantic} is required in
1153 addition to @option{-ansi}. @xref{Warning Options}.
1155 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1156 option is used. Some header files may notice this macro and refrain
1157 from declaring certain functions or defining certain macros that the
1158 ISO standard doesn't call for; this is to avoid interfering with any
1159 programs that might use these names for other things.
1161 Functions which would normally be built in but do not have semantics
1162 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1163 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1164 built-in functions provided by GCC}, for details of the functions
1169 Determine the language standard. This option is currently only
1170 supported when compiling C or C++. A value for this option must be
1171 provided; possible values are
1176 ISO C90 (same as @option{-ansi}).
1178 @item iso9899:199409
1179 ISO C90 as modified in amendment 1.
1185 ISO C99. Note that this standard is not yet fully supported; see
1186 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1187 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1190 Default, ISO C90 plus GNU extensions (including some C99 features).
1194 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1195 this will become the default. The name @samp{gnu9x} is deprecated.
1198 The 1998 ISO C++ standard plus amendments.
1201 The same as @option{-std=c++98} plus GNU extensions. This is the
1202 default for C++ code.
1205 The working draft of the upcoming ISO C++0x standard. This option
1206 enables experimental features that are likely to be included in
1207 C++0x. The working draft is constantly changing, and any feature that is
1208 enabled by this flag may be removed from future versions of GCC if it is
1209 not part of the C++0x standard.
1212 The same as @option{-std=c++0x} plus GNU extensions. As with
1213 @option{-std=c++0x}, this option enables experimental features that may
1214 be removed in future versions of GCC.
1217 Even when this option is not specified, you can still use some of the
1218 features of newer standards in so far as they do not conflict with
1219 previous C standards. For example, you may use @code{__restrict__} even
1220 when @option{-std=c99} is not specified.
1222 The @option{-std} options specifying some version of ISO C have the same
1223 effects as @option{-ansi}, except that features that were not in ISO C90
1224 but are in the specified version (for example, @samp{//} comments and
1225 the @code{inline} keyword in ISO C99) are not disabled.
1227 @xref{Standards,,Language Standards Supported by GCC}, for details of
1228 these standard versions.
1230 @item -aux-info @var{filename}
1232 Output to the given filename prototyped declarations for all functions
1233 declared and/or defined in a translation unit, including those in header
1234 files. This option is silently ignored in any language other than C@.
1236 Besides declarations, the file indicates, in comments, the origin of
1237 each declaration (source file and line), whether the declaration was
1238 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1239 @samp{O} for old, respectively, in the first character after the line
1240 number and the colon), and whether it came from a declaration or a
1241 definition (@samp{C} or @samp{F}, respectively, in the following
1242 character). In the case of function definitions, a K&R-style list of
1243 arguments followed by their declarations is also provided, inside
1244 comments, after the declaration.
1248 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1249 keyword, so that code can use these words as identifiers. You can use
1250 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1251 instead. @option{-ansi} implies @option{-fno-asm}.
1253 In C++, this switch only affects the @code{typeof} keyword, since
1254 @code{asm} and @code{inline} are standard keywords. You may want to
1255 use the @option{-fno-gnu-keywords} flag instead, which has the same
1256 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1257 switch only affects the @code{asm} and @code{typeof} keywords, since
1258 @code{inline} is a standard keyword in ISO C99.
1261 @itemx -fno-builtin-@var{function}
1262 @opindex fno-builtin
1263 @cindex built-in functions
1264 Don't recognize built-in functions that do not begin with
1265 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1266 functions provided by GCC}, for details of the functions affected,
1267 including those which are not built-in functions when @option{-ansi} or
1268 @option{-std} options for strict ISO C conformance are used because they
1269 do not have an ISO standard meaning.
1271 GCC normally generates special code to handle certain built-in functions
1272 more efficiently; for instance, calls to @code{alloca} may become single
1273 instructions that adjust the stack directly, and calls to @code{memcpy}
1274 may become inline copy loops. The resulting code is often both smaller
1275 and faster, but since the function calls no longer appear as such, you
1276 cannot set a breakpoint on those calls, nor can you change the behavior
1277 of the functions by linking with a different library. In addition,
1278 when a function is recognized as a built-in function, GCC may use
1279 information about that function to warn about problems with calls to
1280 that function, or to generate more efficient code, even if the
1281 resulting code still contains calls to that function. For example,
1282 warnings are given with @option{-Wformat} for bad calls to
1283 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1284 known not to modify global memory.
1286 With the @option{-fno-builtin-@var{function}} option
1287 only the built-in function @var{function} is
1288 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1289 function is named this is not built-in in this version of GCC, this
1290 option is ignored. There is no corresponding
1291 @option{-fbuiltin-@var{function}} option; if you wish to enable
1292 built-in functions selectively when using @option{-fno-builtin} or
1293 @option{-ffreestanding}, you may define macros such as:
1296 #define abs(n) __builtin_abs ((n))
1297 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1302 @cindex hosted environment
1304 Assert that compilation takes place in a hosted environment. This implies
1305 @option{-fbuiltin}. A hosted environment is one in which the
1306 entire standard library is available, and in which @code{main} has a return
1307 type of @code{int}. Examples are nearly everything except a kernel.
1308 This is equivalent to @option{-fno-freestanding}.
1310 @item -ffreestanding
1311 @opindex ffreestanding
1312 @cindex hosted environment
1314 Assert that compilation takes place in a freestanding environment. This
1315 implies @option{-fno-builtin}. A freestanding environment
1316 is one in which the standard library may not exist, and program startup may
1317 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1318 This is equivalent to @option{-fno-hosted}.
1320 @xref{Standards,,Language Standards Supported by GCC}, for details of
1321 freestanding and hosted environments.
1325 @cindex openmp parallel
1326 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1327 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1328 compiler generates parallel code according to the OpenMP Application
1329 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.
1331 @item -fms-extensions
1332 @opindex fms-extensions
1333 Accept some non-standard constructs used in Microsoft header files.
1335 Some cases of unnamed fields in structures and unions are only
1336 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1337 fields within structs/unions}, for details.
1341 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1342 options for strict ISO C conformance) implies @option{-trigraphs}.
1344 @item -no-integrated-cpp
1345 @opindex no-integrated-cpp
1346 Performs a compilation in two passes: preprocessing and compiling. This
1347 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1348 @option{-B} option. The user supplied compilation step can then add in
1349 an additional preprocessing step after normal preprocessing but before
1350 compiling. The default is to use the integrated cpp (internal cpp)
1352 The semantics of this option will change if "cc1", "cc1plus", and
1353 "cc1obj" are merged.
1355 @cindex traditional C language
1356 @cindex C language, traditional
1358 @itemx -traditional-cpp
1359 @opindex traditional-cpp
1360 @opindex traditional
1361 Formerly, these options caused GCC to attempt to emulate a pre-standard
1362 C compiler. They are now only supported with the @option{-E} switch.
1363 The preprocessor continues to support a pre-standard mode. See the GNU
1364 CPP manual for details.
1366 @item -fcond-mismatch
1367 @opindex fcond-mismatch
1368 Allow conditional expressions with mismatched types in the second and
1369 third arguments. The value of such an expression is void. This option
1370 is not supported for C++.
1372 @item -funsigned-char
1373 @opindex funsigned-char
1374 Let the type @code{char} be unsigned, like @code{unsigned char}.
1376 Each kind of machine has a default for what @code{char} should
1377 be. It is either like @code{unsigned char} by default or like
1378 @code{signed char} by default.
1380 Ideally, a portable program should always use @code{signed char} or
1381 @code{unsigned char} when it depends on the signedness of an object.
1382 But many programs have been written to use plain @code{char} and
1383 expect it to be signed, or expect it to be unsigned, depending on the
1384 machines they were written for. This option, and its inverse, let you
1385 make such a program work with the opposite default.
1387 The type @code{char} is always a distinct type from each of
1388 @code{signed char} or @code{unsigned char}, even though its behavior
1389 is always just like one of those two.
1392 @opindex fsigned-char
1393 Let the type @code{char} be signed, like @code{signed char}.
1395 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1396 the negative form of @option{-funsigned-char}. Likewise, the option
1397 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1399 @item -fsigned-bitfields
1400 @itemx -funsigned-bitfields
1401 @itemx -fno-signed-bitfields
1402 @itemx -fno-unsigned-bitfields
1403 @opindex fsigned-bitfields
1404 @opindex funsigned-bitfields
1405 @opindex fno-signed-bitfields
1406 @opindex fno-unsigned-bitfields
1407 These options control whether a bit-field is signed or unsigned, when the
1408 declaration does not use either @code{signed} or @code{unsigned}. By
1409 default, such a bit-field is signed, because this is consistent: the
1410 basic integer types such as @code{int} are signed types.
1413 @node C++ Dialect Options
1414 @section Options Controlling C++ Dialect
1416 @cindex compiler options, C++
1417 @cindex C++ options, command line
1418 @cindex options, C++
1419 This section describes the command-line options that are only meaningful
1420 for C++ programs; but you can also use most of the GNU compiler options
1421 regardless of what language your program is in. For example, you
1422 might compile a file @code{firstClass.C} like this:
1425 g++ -g -frepo -O -c firstClass.C
1429 In this example, only @option{-frepo} is an option meant
1430 only for C++ programs; you can use the other options with any
1431 language supported by GCC@.
1433 Here is a list of options that are @emph{only} for compiling C++ programs:
1437 @item -fabi-version=@var{n}
1438 @opindex fabi-version
1439 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1440 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1441 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1442 the version that conforms most closely to the C++ ABI specification.
1443 Therefore, the ABI obtained using version 0 will change as ABI bugs
1446 The default is version 2.
1448 @item -fno-access-control
1449 @opindex fno-access-control
1450 Turn off all access checking. This switch is mainly useful for working
1451 around bugs in the access control code.
1455 Check that the pointer returned by @code{operator new} is non-null
1456 before attempting to modify the storage allocated. This check is
1457 normally unnecessary because the C++ standard specifies that
1458 @code{operator new} will only return @code{0} if it is declared
1459 @samp{throw()}, in which case the compiler will always check the
1460 return value even without this option. In all other cases, when
1461 @code{operator new} has a non-empty exception specification, memory
1462 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1463 @samp{new (nothrow)}.
1465 @item -fconserve-space
1466 @opindex fconserve-space
1467 Put uninitialized or runtime-initialized global variables into the
1468 common segment, as C does. This saves space in the executable at the
1469 cost of not diagnosing duplicate definitions. If you compile with this
1470 flag and your program mysteriously crashes after @code{main()} has
1471 completed, you may have an object that is being destroyed twice because
1472 two definitions were merged.
1474 This option is no longer useful on most targets, now that support has
1475 been added for putting variables into BSS without making them common.
1477 @item -ffriend-injection
1478 @opindex ffriend-injection
1479 Inject friend functions into the enclosing namespace, so that they are
1480 visible outside the scope of the class in which they are declared.
1481 Friend functions were documented to work this way in the old Annotated
1482 C++ Reference Manual, and versions of G++ before 4.1 always worked
1483 that way. However, in ISO C++ a friend function which is not declared
1484 in an enclosing scope can only be found using argument dependent
1485 lookup. This option causes friends to be injected as they were in
1488 This option is for compatibility, and may be removed in a future
1491 @item -fno-elide-constructors
1492 @opindex fno-elide-constructors
1493 The C++ standard allows an implementation to omit creating a temporary
1494 which is only used to initialize another object of the same type.
1495 Specifying this option disables that optimization, and forces G++ to
1496 call the copy constructor in all cases.
1498 @item -fno-enforce-eh-specs
1499 @opindex fno-enforce-eh-specs
1500 Don't generate code to check for violation of exception specifications
1501 at runtime. This option violates the C++ standard, but may be useful
1502 for reducing code size in production builds, much like defining
1503 @samp{NDEBUG}. This does not give user code permission to throw
1504 exceptions in violation of the exception specifications; the compiler
1505 will still optimize based on the specifications, so throwing an
1506 unexpected exception will result in undefined behavior.
1509 @itemx -fno-for-scope
1511 @opindex fno-for-scope
1512 If @option{-ffor-scope} is specified, the scope of variables declared in
1513 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1514 as specified by the C++ standard.
1515 If @option{-fno-for-scope} is specified, the scope of variables declared in
1516 a @i{for-init-statement} extends to the end of the enclosing scope,
1517 as was the case in old versions of G++, and other (traditional)
1518 implementations of C++.
1520 The default if neither flag is given to follow the standard,
1521 but to allow and give a warning for old-style code that would
1522 otherwise be invalid, or have different behavior.
1524 @item -fno-gnu-keywords
1525 @opindex fno-gnu-keywords
1526 Do not recognize @code{typeof} as a keyword, so that code can use this
1527 word as an identifier. You can use the keyword @code{__typeof__} instead.
1528 @option{-ansi} implies @option{-fno-gnu-keywords}.
1530 @item -fno-implicit-templates
1531 @opindex fno-implicit-templates
1532 Never emit code for non-inline templates which are instantiated
1533 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1534 @xref{Template Instantiation}, for more information.
1536 @item -fno-implicit-inline-templates
1537 @opindex fno-implicit-inline-templates
1538 Don't emit code for implicit instantiations of inline templates, either.
1539 The default is to handle inlines differently so that compiles with and
1540 without optimization will need the same set of explicit instantiations.
1542 @item -fno-implement-inlines
1543 @opindex fno-implement-inlines
1544 To save space, do not emit out-of-line copies of inline functions
1545 controlled by @samp{#pragma implementation}. This will cause linker
1546 errors if these functions are not inlined everywhere they are called.
1548 @item -fms-extensions
1549 @opindex fms-extensions
1550 Disable pedantic warnings about constructs used in MFC, such as implicit
1551 int and getting a pointer to member function via non-standard syntax.
1553 @item -fno-nonansi-builtins
1554 @opindex fno-nonansi-builtins
1555 Disable built-in declarations of functions that are not mandated by
1556 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1557 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1559 @item -fno-operator-names
1560 @opindex fno-operator-names
1561 Do not treat the operator name keywords @code{and}, @code{bitand},
1562 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1563 synonyms as keywords.
1565 @item -fno-optional-diags
1566 @opindex fno-optional-diags
1567 Disable diagnostics that the standard says a compiler does not need to
1568 issue. Currently, the only such diagnostic issued by G++ is the one for
1569 a name having multiple meanings within a class.
1572 @opindex fpermissive
1573 Downgrade some diagnostics about nonconformant code from errors to
1574 warnings. Thus, using @option{-fpermissive} will allow some
1575 nonconforming code to compile.
1579 Enable automatic template instantiation at link time. This option also
1580 implies @option{-fno-implicit-templates}. @xref{Template
1581 Instantiation}, for more information.
1585 Disable generation of information about every class with virtual
1586 functions for use by the C++ runtime type identification features
1587 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1588 of the language, you can save some space by using this flag. Note that
1589 exception handling uses the same information, but it will generate it as
1590 needed. The @samp{dynamic_cast} operator can still be used for casts that
1591 do not require runtime type information, i.e. casts to @code{void *} or to
1592 unambiguous base classes.
1596 Emit statistics about front-end processing at the end of the compilation.
1597 This information is generally only useful to the G++ development team.
1599 @item -ftemplate-depth-@var{n}
1600 @opindex ftemplate-depth
1601 Set the maximum instantiation depth for template classes to @var{n}.
1602 A limit on the template instantiation depth is needed to detect
1603 endless recursions during template class instantiation. ANSI/ISO C++
1604 conforming programs must not rely on a maximum depth greater than 17.
1606 @item -fno-threadsafe-statics
1607 @opindex fno-threadsafe-statics
1608 Do not emit the extra code to use the routines specified in the C++
1609 ABI for thread-safe initialization of local statics. You can use this
1610 option to reduce code size slightly in code that doesn't need to be
1613 @item -fuse-cxa-atexit
1614 @opindex fuse-cxa-atexit
1615 Register destructors for objects with static storage duration with the
1616 @code{__cxa_atexit} function rather than the @code{atexit} function.
1617 This option is required for fully standards-compliant handling of static
1618 destructors, but will only work if your C library supports
1619 @code{__cxa_atexit}.
1621 @item -fno-use-cxa-get-exception-ptr
1622 @opindex fno-use-cxa-get-exception-ptr
1623 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1624 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1625 if the runtime routine is not available.
1627 @item -fvisibility-inlines-hidden
1628 @opindex fvisibility-inlines-hidden
1629 This switch declares that the user does not attempt to compare
1630 pointers to inline methods where the addresses of the two functions
1631 were taken in different shared objects.
1633 The effect of this is that GCC may, effectively, mark inline methods with
1634 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1635 appear in the export table of a DSO and do not require a PLT indirection
1636 when used within the DSO@. Enabling this option can have a dramatic effect
1637 on load and link times of a DSO as it massively reduces the size of the
1638 dynamic export table when the library makes heavy use of templates.
1640 The behaviour of this switch is not quite the same as marking the
1641 methods as hidden directly, because it does not affect static variables
1642 local to the function or cause the compiler to deduce that
1643 the function is defined in only one shared object.
1645 You may mark a method as having a visibility explicitly to negate the
1646 effect of the switch for that method. For example, if you do want to
1647 compare pointers to a particular inline method, you might mark it as
1648 having default visibility. Marking the enclosing class with explicit
1649 visibility will have no effect.
1651 Explicitly instantiated inline methods are unaffected by this option
1652 as their linkage might otherwise cross a shared library boundary.
1653 @xref{Template Instantiation}.
1657 Do not use weak symbol support, even if it is provided by the linker.
1658 By default, G++ will use weak symbols if they are available. This
1659 option exists only for testing, and should not be used by end-users;
1660 it will result in inferior code and has no benefits. This option may
1661 be removed in a future release of G++.
1665 Do not search for header files in the standard directories specific to
1666 C++, but do still search the other standard directories. (This option
1667 is used when building the C++ library.)
1670 In addition, these optimization, warning, and code generation options
1671 have meanings only for C++ programs:
1674 @item -fno-default-inline
1675 @opindex fno-default-inline
1676 Do not assume @samp{inline} for functions defined inside a class scope.
1677 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1678 functions will have linkage like inline functions; they just won't be
1681 @item -Wabi @r{(C++ only)}
1683 Warn when G++ generates code that is probably not compatible with the
1684 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1685 all such cases, there are probably some cases that are not warned about,
1686 even though G++ is generating incompatible code. There may also be
1687 cases where warnings are emitted even though the code that is generated
1690 You should rewrite your code to avoid these warnings if you are
1691 concerned about the fact that code generated by G++ may not be binary
1692 compatible with code generated by other compilers.
1694 The known incompatibilities at this point include:
1699 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1700 pack data into the same byte as a base class. For example:
1703 struct A @{ virtual void f(); int f1 : 1; @};
1704 struct B : public A @{ int f2 : 1; @};
1708 In this case, G++ will place @code{B::f2} into the same byte
1709 as@code{A::f1}; other compilers will not. You can avoid this problem
1710 by explicitly padding @code{A} so that its size is a multiple of the
1711 byte size on your platform; that will cause G++ and other compilers to
1712 layout @code{B} identically.
1715 Incorrect handling of tail-padding for virtual bases. G++ does not use
1716 tail padding when laying out virtual bases. For example:
1719 struct A @{ virtual void f(); char c1; @};
1720 struct B @{ B(); char c2; @};
1721 struct C : public A, public virtual B @{@};
1725 In this case, G++ will not place @code{B} into the tail-padding for
1726 @code{A}; other compilers will. You can avoid this problem by
1727 explicitly padding @code{A} so that its size is a multiple of its
1728 alignment (ignoring virtual base classes); that will cause G++ and other
1729 compilers to layout @code{C} identically.
1732 Incorrect handling of bit-fields with declared widths greater than that
1733 of their underlying types, when the bit-fields appear in a union. For
1737 union U @{ int i : 4096; @};
1741 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1742 union too small by the number of bits in an @code{int}.
1745 Empty classes can be placed at incorrect offsets. For example:
1755 struct C : public B, public A @{@};
1759 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1760 it should be placed at offset zero. G++ mistakenly believes that the
1761 @code{A} data member of @code{B} is already at offset zero.
1764 Names of template functions whose types involve @code{typename} or
1765 template template parameters can be mangled incorrectly.
1768 template <typename Q>
1769 void f(typename Q::X) @{@}
1771 template <template <typename> class Q>
1772 void f(typename Q<int>::X) @{@}
1776 Instantiations of these templates may be mangled incorrectly.
1780 @item -Wctor-dtor-privacy @r{(C++ only)}
1781 @opindex Wctor-dtor-privacy
1782 Warn when a class seems unusable because all the constructors or
1783 destructors in that class are private, and it has neither friends nor
1784 public static member functions.
1786 @item -Wnon-virtual-dtor @r{(C++ only)}
1787 @opindex Wnon-virtual-dtor
1788 Warn when a class appears to be polymorphic, thereby requiring a virtual
1789 destructor, yet it declares a non-virtual one. This warning is also
1790 enabled if -Weffc++ is specified.
1792 @item -Wreorder @r{(C++ only)}
1794 @cindex reordering, warning
1795 @cindex warning for reordering of member initializers
1796 Warn when the order of member initializers given in the code does not
1797 match the order in which they must be executed. For instance:
1803 A(): j (0), i (1) @{ @}
1807 The compiler will rearrange the member initializers for @samp{i}
1808 and @samp{j} to match the declaration order of the members, emitting
1809 a warning to that effect. This warning is enabled by @option{-Wall}.
1812 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1815 @item -Weffc++ @r{(C++ only)}
1817 Warn about violations of the following style guidelines from Scott Meyers'
1818 @cite{Effective C++} book:
1822 Item 11: Define a copy constructor and an assignment operator for classes
1823 with dynamically allocated memory.
1826 Item 12: Prefer initialization to assignment in constructors.
1829 Item 14: Make destructors virtual in base classes.
1832 Item 15: Have @code{operator=} return a reference to @code{*this}.
1835 Item 23: Don't try to return a reference when you must return an object.
1839 Also warn about violations of the following style guidelines from
1840 Scott Meyers' @cite{More Effective C++} book:
1844 Item 6: Distinguish between prefix and postfix forms of increment and
1845 decrement operators.
1848 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1852 When selecting this option, be aware that the standard library
1853 headers do not obey all of these guidelines; use @samp{grep -v}
1854 to filter out those warnings.
1856 @item -Wno-deprecated @r{(C++ only)}
1857 @opindex Wno-deprecated
1858 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1860 @item -Wstrict-null-sentinel @r{(C++ only)}
1861 @opindex Wstrict-null-sentinel
1862 Warn also about the use of an uncasted @code{NULL} as sentinel. When
1863 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
1864 to @code{__null}. Although it is a null pointer constant not a null pointer,
1865 it is guaranteed to of the same size as a pointer. But this use is
1866 not portable across different compilers.
1868 @item -Wno-non-template-friend @r{(C++ only)}
1869 @opindex Wno-non-template-friend
1870 Disable warnings when non-templatized friend functions are declared
1871 within a template. Since the advent of explicit template specification
1872 support in G++, if the name of the friend is an unqualified-id (i.e.,
1873 @samp{friend foo(int)}), the C++ language specification demands that the
1874 friend declare or define an ordinary, nontemplate function. (Section
1875 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1876 could be interpreted as a particular specialization of a templatized
1877 function. Because this non-conforming behavior is no longer the default
1878 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1879 check existing code for potential trouble spots and is on by default.
1880 This new compiler behavior can be turned off with
1881 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1882 but disables the helpful warning.
1884 @item -Wold-style-cast @r{(C++ only)}
1885 @opindex Wold-style-cast
1886 Warn if an old-style (C-style) cast to a non-void type is used within
1887 a C++ program. The new-style casts (@samp{dynamic_cast},
1888 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
1889 less vulnerable to unintended effects and much easier to search for.
1891 @item -Woverloaded-virtual @r{(C++ only)}
1892 @opindex Woverloaded-virtual
1893 @cindex overloaded virtual fn, warning
1894 @cindex warning for overloaded virtual fn
1895 Warn when a function declaration hides virtual functions from a
1896 base class. For example, in:
1903 struct B: public A @{
1908 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1916 will fail to compile.
1918 @item -Wno-pmf-conversions @r{(C++ only)}
1919 @opindex Wno-pmf-conversions
1920 Disable the diagnostic for converting a bound pointer to member function
1923 @item -Wsign-promo @r{(C++ only)}
1924 @opindex Wsign-promo
1925 Warn when overload resolution chooses a promotion from unsigned or
1926 enumerated type to a signed type, over a conversion to an unsigned type of
1927 the same size. Previous versions of G++ would try to preserve
1928 unsignedness, but the standard mandates the current behavior.
1933 A& operator = (int);
1943 In this example, G++ will synthesize a default @samp{A& operator =
1944 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1947 @node Objective-C and Objective-C++ Dialect Options
1948 @section Options Controlling Objective-C and Objective-C++ Dialects
1950 @cindex compiler options, Objective-C and Objective-C++
1951 @cindex Objective-C and Objective-C++ options, command line
1952 @cindex options, Objective-C and Objective-C++
1953 (NOTE: This manual does not describe the Objective-C and Objective-C++
1954 languages themselves. See @xref{Standards,,Language Standards
1955 Supported by GCC}, for references.)
1957 This section describes the command-line options that are only meaningful
1958 for Objective-C and Objective-C++ programs, but you can also use most of
1959 the language-independent GNU compiler options.
1960 For example, you might compile a file @code{some_class.m} like this:
1963 gcc -g -fgnu-runtime -O -c some_class.m
1967 In this example, @option{-fgnu-runtime} is an option meant only for
1968 Objective-C and Objective-C++ programs; you can use the other options with
1969 any language supported by GCC@.
1971 Note that since Objective-C is an extension of the C language, Objective-C
1972 compilations may also use options specific to the C front-end (e.g.,
1973 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1974 C++-specific options (e.g., @option{-Wabi}).
1976 Here is a list of options that are @emph{only} for compiling Objective-C
1977 and Objective-C++ programs:
1980 @item -fconstant-string-class=@var{class-name}
1981 @opindex fconstant-string-class
1982 Use @var{class-name} as the name of the class to instantiate for each
1983 literal string specified with the syntax @code{@@"@dots{}"}. The default
1984 class name is @code{NXConstantString} if the GNU runtime is being used, and
1985 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1986 @option{-fconstant-cfstrings} option, if also present, will override the
1987 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1988 to be laid out as constant CoreFoundation strings.
1991 @opindex fgnu-runtime
1992 Generate object code compatible with the standard GNU Objective-C
1993 runtime. This is the default for most types of systems.
1995 @item -fnext-runtime
1996 @opindex fnext-runtime
1997 Generate output compatible with the NeXT runtime. This is the default
1998 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1999 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2002 @item -fno-nil-receivers
2003 @opindex fno-nil-receivers
2004 Assume that all Objective-C message dispatches (e.g.,
2005 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2006 is not @code{nil}. This allows for more efficient entry points in the runtime
2007 to be used. Currently, this option is only available in conjunction with
2008 the NeXT runtime on Mac OS X 10.3 and later.
2010 @item -fobjc-call-cxx-cdtors
2011 @opindex fobjc-call-cxx-cdtors
2012 For each Objective-C class, check if any of its instance variables is a
2013 C++ object with a non-trivial default constructor. If so, synthesize a
2014 special @code{- (id) .cxx_construct} instance method that will run
2015 non-trivial default constructors on any such instance variables, in order,
2016 and then return @code{self}. Similarly, check if any instance variable
2017 is a C++ object with a non-trivial destructor, and if so, synthesize a
2018 special @code{- (void) .cxx_destruct} method that will run
2019 all such default destructors, in reverse order.
2021 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2022 thusly generated will only operate on instance variables declared in the
2023 current Objective-C class, and not those inherited from superclasses. It
2024 is the responsibility of the Objective-C runtime to invoke all such methods
2025 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2026 will be invoked by the runtime immediately after a new object
2027 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2028 be invoked immediately before the runtime deallocates an object instance.
2030 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2031 support for invoking the @code{- (id) .cxx_construct} and
2032 @code{- (void) .cxx_destruct} methods.
2034 @item -fobjc-direct-dispatch
2035 @opindex fobjc-direct-dispatch
2036 Allow fast jumps to the message dispatcher. On Darwin this is
2037 accomplished via the comm page.
2039 @item -fobjc-exceptions
2040 @opindex fobjc-exceptions
2041 Enable syntactic support for structured exception handling in Objective-C,
2042 similar to what is offered by C++ and Java. This option is
2043 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2052 @@catch (AnObjCClass *exc) @{
2059 @@catch (AnotherClass *exc) @{
2062 @@catch (id allOthers) @{
2072 The @code{@@throw} statement may appear anywhere in an Objective-C or
2073 Objective-C++ program; when used inside of a @code{@@catch} block, the
2074 @code{@@throw} may appear without an argument (as shown above), in which case
2075 the object caught by the @code{@@catch} will be rethrown.
2077 Note that only (pointers to) Objective-C objects may be thrown and
2078 caught using this scheme. When an object is thrown, it will be caught
2079 by the nearest @code{@@catch} clause capable of handling objects of that type,
2080 analogously to how @code{catch} blocks work in C++ and Java. A
2081 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2082 any and all Objective-C exceptions not caught by previous @code{@@catch}
2085 The @code{@@finally} clause, if present, will be executed upon exit from the
2086 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2087 regardless of whether any exceptions are thrown, caught or rethrown
2088 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2089 of the @code{finally} clause in Java.
2091 There are several caveats to using the new exception mechanism:
2095 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2096 idioms provided by the @code{NSException} class, the new
2097 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2098 systems, due to additional functionality needed in the (NeXT) Objective-C
2102 As mentioned above, the new exceptions do not support handling
2103 types other than Objective-C objects. Furthermore, when used from
2104 Objective-C++, the Objective-C exception model does not interoperate with C++
2105 exceptions at this time. This means you cannot @code{@@throw} an exception
2106 from Objective-C and @code{catch} it in C++, or vice versa
2107 (i.e., @code{throw @dots{} @@catch}).
2110 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2111 blocks for thread-safe execution:
2114 @@synchronized (ObjCClass *guard) @{
2119 Upon entering the @code{@@synchronized} block, a thread of execution shall
2120 first check whether a lock has been placed on the corresponding @code{guard}
2121 object by another thread. If it has, the current thread shall wait until
2122 the other thread relinquishes its lock. Once @code{guard} becomes available,
2123 the current thread will place its own lock on it, execute the code contained in
2124 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2125 making @code{guard} available to other threads).
2127 Unlike Java, Objective-C does not allow for entire methods to be marked
2128 @code{@@synchronized}. Note that throwing exceptions out of
2129 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2130 to be unlocked properly.
2134 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2136 @item -freplace-objc-classes
2137 @opindex freplace-objc-classes
2138 Emit a special marker instructing @command{ld(1)} not to statically link in
2139 the resulting object file, and allow @command{dyld(1)} to load it in at
2140 run time instead. This is used in conjunction with the Fix-and-Continue
2141 debugging mode, where the object file in question may be recompiled and
2142 dynamically reloaded in the course of program execution, without the need
2143 to restart the program itself. Currently, Fix-and-Continue functionality
2144 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2149 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2150 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2151 compile time) with static class references that get initialized at load time,
2152 which improves run-time performance. Specifying the @option{-fzero-link} flag
2153 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2154 to be retained. This is useful in Zero-Link debugging mode, since it allows
2155 for individual class implementations to be modified during program execution.
2159 Dump interface declarations for all classes seen in the source file to a
2160 file named @file{@var{sourcename}.decl}.
2162 @item -Wassign-intercept
2163 @opindex Wassign-intercept
2164 Warn whenever an Objective-C assignment is being intercepted by the
2168 @opindex Wno-protocol
2169 If a class is declared to implement a protocol, a warning is issued for
2170 every method in the protocol that is not implemented by the class. The
2171 default behavior is to issue a warning for every method not explicitly
2172 implemented in the class, even if a method implementation is inherited
2173 from the superclass. If you use the @option{-Wno-protocol} option, then
2174 methods inherited from the superclass are considered to be implemented,
2175 and no warning is issued for them.
2179 Warn if multiple methods of different types for the same selector are
2180 found during compilation. The check is performed on the list of methods
2181 in the final stage of compilation. Additionally, a check is performed
2182 for each selector appearing in a @code{@@selector(@dots{})}
2183 expression, and a corresponding method for that selector has been found
2184 during compilation. Because these checks scan the method table only at
2185 the end of compilation, these warnings are not produced if the final
2186 stage of compilation is not reached, for example because an error is
2187 found during compilation, or because the @option{-fsyntax-only} option is
2190 @item -Wstrict-selector-match
2191 @opindex Wstrict-selector-match
2192 Warn if multiple methods with differing argument and/or return types are
2193 found for a given selector when attempting to send a message using this
2194 selector to a receiver of type @code{id} or @code{Class}. When this flag
2195 is off (which is the default behavior), the compiler will omit such warnings
2196 if any differences found are confined to types which share the same size
2199 @item -Wundeclared-selector
2200 @opindex Wundeclared-selector
2201 Warn if a @code{@@selector(@dots{})} expression referring to an
2202 undeclared selector is found. A selector is considered undeclared if no
2203 method with that name has been declared before the
2204 @code{@@selector(@dots{})} expression, either explicitly in an
2205 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2206 an @code{@@implementation} section. This option always performs its
2207 checks as soon as a @code{@@selector(@dots{})} expression is found,
2208 while @option{-Wselector} only performs its checks in the final stage of
2209 compilation. This also enforces the coding style convention
2210 that methods and selectors must be declared before being used.
2212 @item -print-objc-runtime-info
2213 @opindex print-objc-runtime-info
2214 Generate C header describing the largest structure that is passed by
2219 @node Language Independent Options
2220 @section Options to Control Diagnostic Messages Formatting
2221 @cindex options to control diagnostics formatting
2222 @cindex diagnostic messages
2223 @cindex message formatting
2225 Traditionally, diagnostic messages have been formatted irrespective of
2226 the output device's aspect (e.g.@: its width, @dots{}). The options described
2227 below can be used to control the diagnostic messages formatting
2228 algorithm, e.g.@: how many characters per line, how often source location
2229 information should be reported. Right now, only the C++ front end can
2230 honor these options. However it is expected, in the near future, that
2231 the remaining front ends would be able to digest them correctly.
2234 @item -fmessage-length=@var{n}
2235 @opindex fmessage-length
2236 Try to format error messages so that they fit on lines of about @var{n}
2237 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2238 the front ends supported by GCC@. If @var{n} is zero, then no
2239 line-wrapping will be done; each error message will appear on a single
2242 @opindex fdiagnostics-show-location
2243 @item -fdiagnostics-show-location=once
2244 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2245 reporter to emit @emph{once} source location information; that is, in
2246 case the message is too long to fit on a single physical line and has to
2247 be wrapped, the source location won't be emitted (as prefix) again,
2248 over and over, in subsequent continuation lines. This is the default
2251 @item -fdiagnostics-show-location=every-line
2252 Only meaningful in line-wrapping mode. Instructs the diagnostic
2253 messages reporter to emit the same source location information (as
2254 prefix) for physical lines that result from the process of breaking
2255 a message which is too long to fit on a single line.
2257 @item -fdiagnostics-show-option
2258 @opindex fdiagnostics-show-option
2259 This option instructs the diagnostic machinery to add text to each
2260 diagnostic emitted, which indicates which command line option directly
2261 controls that diagnostic, when such an option is known to the
2262 diagnostic machinery.
2266 @node Warning Options
2267 @section Options to Request or Suppress Warnings
2268 @cindex options to control warnings
2269 @cindex warning messages
2270 @cindex messages, warning
2271 @cindex suppressing warnings
2273 Warnings are diagnostic messages that report constructions which
2274 are not inherently erroneous but which are risky or suggest there
2275 may have been an error.
2277 You can request many specific warnings with options beginning @samp{-W},
2278 for example @option{-Wimplicit} to request warnings on implicit
2279 declarations. Each of these specific warning options also has a
2280 negative form beginning @samp{-Wno-} to turn off warnings;
2281 for example, @option{-Wno-implicit}. This manual lists only one of the
2282 two forms, whichever is not the default.
2284 The following options control the amount and kinds of warnings produced
2285 by GCC; for further, language-specific options also refer to
2286 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2290 @cindex syntax checking
2292 @opindex fsyntax-only
2293 Check the code for syntax errors, but don't do anything beyond that.
2297 Issue all the warnings demanded by strict ISO C and ISO C++;
2298 reject all programs that use forbidden extensions, and some other
2299 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2300 version of the ISO C standard specified by any @option{-std} option used.
2302 Valid ISO C and ISO C++ programs should compile properly with or without
2303 this option (though a rare few will require @option{-ansi} or a
2304 @option{-std} option specifying the required version of ISO C)@. However,
2305 without this option, certain GNU extensions and traditional C and C++
2306 features are supported as well. With this option, they are rejected.
2308 @option{-pedantic} does not cause warning messages for use of the
2309 alternate keywords whose names begin and end with @samp{__}. Pedantic
2310 warnings are also disabled in the expression that follows
2311 @code{__extension__}. However, only system header files should use
2312 these escape routes; application programs should avoid them.
2313 @xref{Alternate Keywords}.
2315 Some users try to use @option{-pedantic} to check programs for strict ISO
2316 C conformance. They soon find that it does not do quite what they want:
2317 it finds some non-ISO practices, but not all---only those for which
2318 ISO C @emph{requires} a diagnostic, and some others for which
2319 diagnostics have been added.
2321 A feature to report any failure to conform to ISO C might be useful in
2322 some instances, but would require considerable additional work and would
2323 be quite different from @option{-pedantic}. We don't have plans to
2324 support such a feature in the near future.
2326 Where the standard specified with @option{-std} represents a GNU
2327 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2328 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2329 extended dialect is based. Warnings from @option{-pedantic} are given
2330 where they are required by the base standard. (It would not make sense
2331 for such warnings to be given only for features not in the specified GNU
2332 C dialect, since by definition the GNU dialects of C include all
2333 features the compiler supports with the given option, and there would be
2334 nothing to warn about.)
2336 @item -pedantic-errors
2337 @opindex pedantic-errors
2338 Like @option{-pedantic}, except that errors are produced rather than
2343 Inhibit all warning messages.
2347 Inhibit warning messages about the use of @samp{#import}.
2349 @item -Wchar-subscripts
2350 @opindex Wchar-subscripts
2351 Warn if an array subscript has type @code{char}. This is a common cause
2352 of error, as programmers often forget that this type is signed on some
2354 This warning is enabled by @option{-Wall}.
2358 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2359 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2360 This warning is enabled by @option{-Wall}.
2362 @item -Wfatal-errors
2363 @opindex Wfatal-errors
2364 This option causes the compiler to abort compilation on the first error
2365 occurred rather than trying to keep going and printing further error
2370 @opindex ffreestanding
2371 @opindex fno-builtin
2372 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2373 the arguments supplied have types appropriate to the format string
2374 specified, and that the conversions specified in the format string make
2375 sense. This includes standard functions, and others specified by format
2376 attributes (@pxref{Function Attributes}), in the @code{printf},
2377 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2378 not in the C standard) families (or other target-specific families).
2379 Which functions are checked without format attributes having been
2380 specified depends on the standard version selected, and such checks of
2381 functions without the attribute specified are disabled by
2382 @option{-ffreestanding} or @option{-fno-builtin}.
2384 The formats are checked against the format features supported by GNU
2385 libc version 2.2. These include all ISO C90 and C99 features, as well
2386 as features from the Single Unix Specification and some BSD and GNU
2387 extensions. Other library implementations may not support all these
2388 features; GCC does not support warning about features that go beyond a
2389 particular library's limitations. However, if @option{-pedantic} is used
2390 with @option{-Wformat}, warnings will be given about format features not
2391 in the selected standard version (but not for @code{strfmon} formats,
2392 since those are not in any version of the C standard). @xref{C Dialect
2393 Options,,Options Controlling C Dialect}.
2395 Since @option{-Wformat} also checks for null format arguments for
2396 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2398 @option{-Wformat} is included in @option{-Wall}. For more control over some
2399 aspects of format checking, the options @option{-Wformat-y2k},
2400 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2401 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2402 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2405 @opindex Wformat-y2k
2406 If @option{-Wformat} is specified, also warn about @code{strftime}
2407 formats which may yield only a two-digit year.
2409 @item -Wno-format-extra-args
2410 @opindex Wno-format-extra-args
2411 If @option{-Wformat} is specified, do not warn about excess arguments to a
2412 @code{printf} or @code{scanf} format function. The C standard specifies
2413 that such arguments are ignored.
2415 Where the unused arguments lie between used arguments that are
2416 specified with @samp{$} operand number specifications, normally
2417 warnings are still given, since the implementation could not know what
2418 type to pass to @code{va_arg} to skip the unused arguments. However,
2419 in the case of @code{scanf} formats, this option will suppress the
2420 warning if the unused arguments are all pointers, since the Single
2421 Unix Specification says that such unused arguments are allowed.
2423 @item -Wno-format-zero-length
2424 @opindex Wno-format-zero-length
2425 If @option{-Wformat} is specified, do not warn about zero-length formats.
2426 The C standard specifies that zero-length formats are allowed.
2428 @item -Wformat-nonliteral
2429 @opindex Wformat-nonliteral
2430 If @option{-Wformat} is specified, also warn if the format string is not a
2431 string literal and so cannot be checked, unless the format function
2432 takes its format arguments as a @code{va_list}.
2434 @item -Wformat-security
2435 @opindex Wformat-security
2436 If @option{-Wformat} is specified, also warn about uses of format
2437 functions that represent possible security problems. At present, this
2438 warns about calls to @code{printf} and @code{scanf} functions where the
2439 format string is not a string literal and there are no format arguments,
2440 as in @code{printf (foo);}. This may be a security hole if the format
2441 string came from untrusted input and contains @samp{%n}. (This is
2442 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2443 in future warnings may be added to @option{-Wformat-security} that are not
2444 included in @option{-Wformat-nonliteral}.)
2448 Enable @option{-Wformat} plus format checks not included in
2449 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2450 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2454 Warn about passing a null pointer for arguments marked as
2455 requiring a non-null value by the @code{nonnull} function attribute.
2457 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2458 can be disabled with the @option{-Wno-nonnull} option.
2460 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2462 Warn about uninitialized variables which are initialized with themselves.
2463 Note this option can only be used with the @option{-Wuninitialized} option,
2464 which in turn only works with @option{-O1} and above.
2466 For example, GCC will warn about @code{i} being uninitialized in the
2467 following snippet only when @option{-Winit-self} has been specified:
2478 @item -Wimplicit-int
2479 @opindex Wimplicit-int
2480 Warn when a declaration does not specify a type.
2481 This warning is enabled by @option{-Wall}.
2483 @item -Wimplicit-function-declaration
2484 @itemx -Werror-implicit-function-declaration
2485 @opindex Wimplicit-function-declaration
2486 @opindex Werror-implicit-function-declaration
2487 Give a warning (or error) whenever a function is used before being
2488 declared. The form @option{-Wno-error-implicit-function-declaration}
2490 This warning is enabled by @option{-Wall} (as a warning, not an error).
2494 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2495 This warning is enabled by @option{-Wall}.
2499 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2500 function with external linkage, returning int, taking either zero
2501 arguments, two, or three arguments of appropriate types.
2502 This warning is enabled by @option{-Wall}.
2504 @item -Wmissing-braces
2505 @opindex Wmissing-braces
2506 Warn if an aggregate or union initializer is not fully bracketed. In
2507 the following example, the initializer for @samp{a} is not fully
2508 bracketed, but that for @samp{b} is fully bracketed.
2511 int a[2][2] = @{ 0, 1, 2, 3 @};
2512 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2515 This warning is enabled by @option{-Wall}.
2517 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2518 @opindex Wmissing-include-dirs
2519 Warn if a user-supplied include directory does not exist.
2522 @opindex Wparentheses
2523 Warn if parentheses are omitted in certain contexts, such
2524 as when there is an assignment in a context where a truth value
2525 is expected, or when operators are nested whose precedence people
2526 often get confused about. Only the warning for an assignment used as
2527 a truth value is supported when compiling C++; the other warnings are
2528 only supported when compiling C@.
2530 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2531 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2532 interpretation from that of ordinary mathematical notation.
2534 Also warn about constructions where there may be confusion to which
2535 @code{if} statement an @code{else} branch belongs. Here is an example of
2550 In C, every @code{else} branch belongs to the innermost possible @code{if}
2551 statement, which in this example is @code{if (b)}. This is often not
2552 what the programmer expected, as illustrated in the above example by
2553 indentation the programmer chose. When there is the potential for this
2554 confusion, GCC will issue a warning when this flag is specified.
2555 To eliminate the warning, add explicit braces around the innermost
2556 @code{if} statement so there is no way the @code{else} could belong to
2557 the enclosing @code{if}. The resulting code would look like this:
2573 This warning is enabled by @option{-Wall}.
2575 @item -Wsequence-point
2576 @opindex Wsequence-point
2577 Warn about code that may have undefined semantics because of violations
2578 of sequence point rules in the C and C++ standards.
2580 The C and C++ standards defines the order in which expressions in a C/C++
2581 program are evaluated in terms of @dfn{sequence points}, which represent
2582 a partial ordering between the execution of parts of the program: those
2583 executed before the sequence point, and those executed after it. These
2584 occur after the evaluation of a full expression (one which is not part
2585 of a larger expression), after the evaluation of the first operand of a
2586 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2587 function is called (but after the evaluation of its arguments and the
2588 expression denoting the called function), and in certain other places.
2589 Other than as expressed by the sequence point rules, the order of
2590 evaluation of subexpressions of an expression is not specified. All
2591 these rules describe only a partial order rather than a total order,
2592 since, for example, if two functions are called within one expression
2593 with no sequence point between them, the order in which the functions
2594 are called is not specified. However, the standards committee have
2595 ruled that function calls do not overlap.
2597 It is not specified when between sequence points modifications to the
2598 values of objects take effect. Programs whose behavior depends on this
2599 have undefined behavior; the C and C++ standards specify that ``Between
2600 the previous and next sequence point an object shall have its stored
2601 value modified at most once by the evaluation of an expression.
2602 Furthermore, the prior value shall be read only to determine the value
2603 to be stored.''. If a program breaks these rules, the results on any
2604 particular implementation are entirely unpredictable.
2606 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2607 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2608 diagnosed by this option, and it may give an occasional false positive
2609 result, but in general it has been found fairly effective at detecting
2610 this sort of problem in programs.
2612 The standard is worded confusingly, therefore there is some debate
2613 over the precise meaning of the sequence point rules in subtle cases.
2614 Links to discussions of the problem, including proposed formal
2615 definitions, may be found on the GCC readings page, at
2616 @w{@uref{http://gcc.gnu.org/readings.html}}.
2618 This warning is enabled by @option{-Wall} for C and C++.
2621 @opindex Wreturn-type
2622 Warn whenever a function is defined with a return-type that defaults to
2623 @code{int}. Also warn about any @code{return} statement with no
2624 return-value in a function whose return-type is not @code{void}.
2626 For C, also warn if the return type of a function has a type qualifier
2627 such as @code{const}. Such a type qualifier has no effect, since the
2628 value returned by a function is not an lvalue. ISO C prohibits
2629 qualified @code{void} return types on function definitions, so such
2630 return types always receive a warning even without this option.
2632 For C++, a function without return type always produces a diagnostic
2633 message, even when @option{-Wno-return-type} is specified. The only
2634 exceptions are @samp{main} and functions defined in system headers.
2636 This warning is enabled by @option{-Wall}.
2640 Warn whenever a @code{switch} statement has an index of enumerated type
2641 and lacks a @code{case} for one or more of the named codes of that
2642 enumeration. (The presence of a @code{default} label prevents this
2643 warning.) @code{case} labels outside the enumeration range also
2644 provoke warnings when this option is used.
2645 This warning is enabled by @option{-Wall}.
2647 @item -Wswitch-default
2648 @opindex Wswitch-switch
2649 Warn whenever a @code{switch} statement does not have a @code{default}
2653 @opindex Wswitch-enum
2654 Warn whenever a @code{switch} statement has an index of enumerated type
2655 and lacks a @code{case} for one or more of the named codes of that
2656 enumeration. @code{case} labels outside the enumeration range also
2657 provoke warnings when this option is used.
2661 Warn if any trigraphs are encountered that might change the meaning of
2662 the program (trigraphs within comments are not warned about).
2663 This warning is enabled by @option{-Wall}.
2665 @item -Wunused-function
2666 @opindex Wunused-function
2667 Warn whenever a static function is declared but not defined or a
2668 non-inline static function is unused.
2669 This warning is enabled by @option{-Wall}.
2671 @item -Wunused-label
2672 @opindex Wunused-label
2673 Warn whenever a label is declared but not used.
2674 This warning is enabled by @option{-Wall}.
2676 To suppress this warning use the @samp{unused} attribute
2677 (@pxref{Variable Attributes}).
2679 @item -Wunused-parameter
2680 @opindex Wunused-parameter
2681 Warn whenever a function parameter is unused aside from its declaration.
2683 To suppress this warning use the @samp{unused} attribute
2684 (@pxref{Variable Attributes}).
2686 @item -Wunused-variable
2687 @opindex Wunused-variable
2688 Warn whenever a local variable or non-constant static variable is unused
2689 aside from its declaration.
2690 This warning is enabled by @option{-Wall}.
2692 To suppress this warning use the @samp{unused} attribute
2693 (@pxref{Variable Attributes}).
2695 @item -Wunused-value
2696 @opindex Wunused-value
2697 Warn whenever a statement computes a result that is explicitly not used.
2698 This warning is enabled by @option{-Wall}.
2700 To suppress this warning cast the expression to @samp{void}.
2704 All the above @option{-Wunused} options combined.
2706 In order to get a warning about an unused function parameter, you must
2707 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2708 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2710 @item -Wuninitialized
2711 @opindex Wuninitialized
2712 Warn if an automatic variable is used without first being initialized or
2713 if a variable may be clobbered by a @code{setjmp} call.
2715 These warnings are possible only in optimizing compilation,
2716 because they require data flow information that is computed only
2717 when optimizing. If you do not specify @option{-O}, you will not get
2718 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
2719 requiring @option{-O}.
2721 If you want to warn about code which uses the uninitialized value of the
2722 variable in its own initializer, use the @option{-Winit-self} option.
2724 These warnings occur for individual uninitialized or clobbered
2725 elements of structure, union or array variables as well as for
2726 variables which are uninitialized or clobbered as a whole. They do
2727 not occur for variables or elements declared @code{volatile}. Because
2728 these warnings depend on optimization, the exact variables or elements
2729 for which there are warnings will depend on the precise optimization
2730 options and version of GCC used.
2732 Note that there may be no warning about a variable that is used only
2733 to compute a value that itself is never used, because such
2734 computations may be deleted by data flow analysis before the warnings
2737 These warnings are made optional because GCC is not smart
2738 enough to see all the reasons why the code might be correct
2739 despite appearing to have an error. Here is one example of how
2760 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2761 always initialized, but GCC doesn't know this. Here is
2762 another common case:
2767 if (change_y) save_y = y, y = new_y;
2769 if (change_y) y = save_y;
2774 This has no bug because @code{save_y} is used only if it is set.
2776 @cindex @code{longjmp} warnings
2777 This option also warns when a non-volatile automatic variable might be
2778 changed by a call to @code{longjmp}. These warnings as well are possible
2779 only in optimizing compilation.
2781 The compiler sees only the calls to @code{setjmp}. It cannot know
2782 where @code{longjmp} will be called; in fact, a signal handler could
2783 call it at any point in the code. As a result, you may get a warning
2784 even when there is in fact no problem because @code{longjmp} cannot
2785 in fact be called at the place which would cause a problem.
2787 Some spurious warnings can be avoided if you declare all the functions
2788 you use that never return as @code{noreturn}. @xref{Function
2791 This warning is enabled by @option{-Wall}.
2793 @item -Wunknown-pragmas
2794 @opindex Wunknown-pragmas
2795 @cindex warning for unknown pragmas
2796 @cindex unknown pragmas, warning
2797 @cindex pragmas, warning of unknown
2798 Warn when a #pragma directive is encountered which is not understood by
2799 GCC@. If this command line option is used, warnings will even be issued
2800 for unknown pragmas in system header files. This is not the case if
2801 the warnings were only enabled by the @option{-Wall} command line option.
2804 @opindex Wno-pragmas
2806 Do not warn about misuses of pragmas, such as incorrect parameters,
2807 invalid syntax, or conflicts between pragmas. See also
2808 @samp{-Wunknown-pragmas}.
2810 @item -Wstrict-aliasing
2811 @opindex Wstrict-aliasing
2812 This option is only active when @option{-fstrict-aliasing} is active.
2813 It warns about code which might break the strict aliasing rules that the
2814 compiler is using for optimization. The warning does not catch all
2815 cases, but does attempt to catch the more common pitfalls. It is
2816 included in @option{-Wall}.
2818 @item -Wstrict-aliasing=2
2819 @opindex Wstrict-aliasing=2
2820 This option is only active when @option{-fstrict-aliasing} is active.
2821 It warns about code which might break the strict aliasing rules that the
2822 compiler is using for optimization. This warning catches more cases than
2823 @option{-Wstrict-aliasing}, but it will also give a warning for some ambiguous
2824 cases that are safe.
2828 All of the above @samp{-W} options combined. This enables all the
2829 warnings about constructions that some users consider questionable, and
2830 that are easy to avoid (or modify to prevent the warning), even in
2831 conjunction with macros. This also enables some language-specific
2832 warnings described in @ref{C++ Dialect Options} and
2833 @ref{Objective-C and Objective-C++ Dialect Options}.
2836 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2837 Some of them warn about constructions that users generally do not
2838 consider questionable, but which occasionally you might wish to check
2839 for; others warn about constructions that are necessary or hard to avoid
2840 in some cases, and there is no simple way to modify the code to suppress
2847 (This option used to be called @option{-W}. The older name is still
2848 supported, but the newer name is more descriptive.) Print extra warning
2849 messages for these events:
2853 A function can return either with or without a value. (Falling
2854 off the end of the function body is considered returning without
2855 a value.) For example, this function would evoke such a
2869 An expression-statement or the left-hand side of a comma expression
2870 contains no side effects.
2871 To suppress the warning, cast the unused expression to void.
2872 For example, an expression such as @samp{x[i,j]} will cause a warning,
2873 but @samp{x[(void)i,j]} will not.
2876 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2879 Storage-class specifiers like @code{static} are not the first things in
2880 a declaration. According to the C Standard, this usage is obsolescent.
2883 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2887 A comparison between signed and unsigned values could produce an
2888 incorrect result when the signed value is converted to unsigned.
2889 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2892 An aggregate has an initializer which does not initialize all members.
2893 This warning can be independently controlled by
2894 @option{-Wmissing-field-initializers}.
2897 An initialized field without side effects is overridden when using
2898 designated initializers (@pxref{Designated Inits, , Designated
2899 Initializers}). This warning can be independently controlled by
2900 @option{-Woverride-init}.
2903 A function parameter is declared without a type specifier in K&R-style
2911 An empty body occurs in an @samp{if} or @samp{else} statement.
2914 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2915 @samp{>}, or @samp{>=}.
2918 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2921 Any of several floating-point events that often indicate errors, such as
2922 overflow, underflow, loss of precision, etc.
2924 @item @r{(C++ only)}
2925 An enumerator and a non-enumerator both appear in a conditional expression.
2927 @item @r{(C++ only)}
2928 A non-static reference or non-static @samp{const} member appears in a
2929 class without constructors.
2931 @item @r{(C++ only)}
2932 Ambiguous virtual bases.
2934 @item @r{(C++ only)}
2935 Subscripting an array which has been declared @samp{register}.
2937 @item @r{(C++ only)}
2938 Taking the address of a variable which has been declared @samp{register}.
2940 @item @r{(C++ only)}
2941 A base class is not initialized in a derived class' copy constructor.
2944 @item -Wno-div-by-zero
2945 @opindex Wno-div-by-zero
2946 @opindex Wdiv-by-zero
2947 Do not warn about compile-time integer division by zero. Floating point
2948 division by zero is not warned about, as it can be a legitimate way of
2949 obtaining infinities and NaNs.
2951 @item -Wsystem-headers
2952 @opindex Wsystem-headers
2953 @cindex warnings from system headers
2954 @cindex system headers, warnings from
2955 Print warning messages for constructs found in system header files.
2956 Warnings from system headers are normally suppressed, on the assumption
2957 that they usually do not indicate real problems and would only make the
2958 compiler output harder to read. Using this command line option tells
2959 GCC to emit warnings from system headers as if they occurred in user
2960 code. However, note that using @option{-Wall} in conjunction with this
2961 option will @emph{not} warn about unknown pragmas in system
2962 headers---for that, @option{-Wunknown-pragmas} must also be used.
2965 @opindex Wfloat-equal
2966 Warn if floating point values are used in equality comparisons.
2968 The idea behind this is that sometimes it is convenient (for the
2969 programmer) to consider floating-point values as approximations to
2970 infinitely precise real numbers. If you are doing this, then you need
2971 to compute (by analyzing the code, or in some other way) the maximum or
2972 likely maximum error that the computation introduces, and allow for it
2973 when performing comparisons (and when producing output, but that's a
2974 different problem). In particular, instead of testing for equality, you
2975 would check to see whether the two values have ranges that overlap; and
2976 this is done with the relational operators, so equality comparisons are
2979 @item -Wtraditional @r{(C only)}
2980 @opindex Wtraditional
2981 Warn about certain constructs that behave differently in traditional and
2982 ISO C@. Also warn about ISO C constructs that have no traditional C
2983 equivalent, and/or problematic constructs which should be avoided.
2987 Macro parameters that appear within string literals in the macro body.
2988 In traditional C macro replacement takes place within string literals,
2989 but does not in ISO C@.
2992 In traditional C, some preprocessor directives did not exist.
2993 Traditional preprocessors would only consider a line to be a directive
2994 if the @samp{#} appeared in column 1 on the line. Therefore
2995 @option{-Wtraditional} warns about directives that traditional C
2996 understands but would ignore because the @samp{#} does not appear as the
2997 first character on the line. It also suggests you hide directives like
2998 @samp{#pragma} not understood by traditional C by indenting them. Some
2999 traditional implementations would not recognize @samp{#elif}, so it
3000 suggests avoiding it altogether.
3003 A function-like macro that appears without arguments.
3006 The unary plus operator.
3009 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3010 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3011 constants.) Note, these suffixes appear in macros defined in the system
3012 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3013 Use of these macros in user code might normally lead to spurious
3014 warnings, however GCC's integrated preprocessor has enough context to
3015 avoid warning in these cases.
3018 A function declared external in one block and then used after the end of
3022 A @code{switch} statement has an operand of type @code{long}.
3025 A non-@code{static} function declaration follows a @code{static} one.
3026 This construct is not accepted by some traditional C compilers.
3029 The ISO type of an integer constant has a different width or
3030 signedness from its traditional type. This warning is only issued if
3031 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3032 typically represent bit patterns, are not warned about.
3035 Usage of ISO string concatenation is detected.
3038 Initialization of automatic aggregates.
3041 Identifier conflicts with labels. Traditional C lacks a separate
3042 namespace for labels.
3045 Initialization of unions. If the initializer is zero, the warning is
3046 omitted. This is done under the assumption that the zero initializer in
3047 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3048 initializer warnings and relies on default initialization to zero in the
3052 Conversions by prototypes between fixed/floating point values and vice
3053 versa. The absence of these prototypes when compiling with traditional
3054 C would cause serious problems. This is a subset of the possible
3055 conversion warnings, for the full set use @option{-Wconversion}.
3058 Use of ISO C style function definitions. This warning intentionally is
3059 @emph{not} issued for prototype declarations or variadic functions
3060 because these ISO C features will appear in your code when using
3061 libiberty's traditional C compatibility macros, @code{PARAMS} and
3062 @code{VPARAMS}. This warning is also bypassed for nested functions
3063 because that feature is already a GCC extension and thus not relevant to
3064 traditional C compatibility.
3067 @item -Wdeclaration-after-statement @r{(C only)}
3068 @opindex Wdeclaration-after-statement
3069 Warn when a declaration is found after a statement in a block. This
3070 construct, known from C++, was introduced with ISO C99 and is by default
3071 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3072 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3076 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3078 @item -Wno-endif-labels
3079 @opindex Wno-endif-labels
3080 @opindex Wendif-labels
3081 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3085 Warn whenever a local variable shadows another local variable, parameter or
3086 global variable or whenever a built-in function is shadowed.
3088 @item -Wlarger-than-@var{len}
3089 @opindex Wlarger-than
3090 Warn whenever an object of larger than @var{len} bytes is defined.
3092 @item -Wunsafe-loop-optimizations
3093 @opindex Wunsafe-loop-optimizations
3094 Warn if the loop cannot be optimized because the compiler could not
3095 assume anything on the bounds of the loop indices. With
3096 @option{-funsafe-loop-optimizations} warn if the compiler made
3099 @item -Wpointer-arith
3100 @opindex Wpointer-arith
3101 Warn about anything that depends on the ``size of'' a function type or
3102 of @code{void}. GNU C assigns these types a size of 1, for
3103 convenience in calculations with @code{void *} pointers and pointers
3106 @item -Wbad-function-cast @r{(C only)}
3107 @opindex Wbad-function-cast
3108 Warn whenever a function call is cast to a non-matching type.
3109 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3112 Warn about ISO C constructs that are outside of the common subset of
3113 ISO C and ISO C++, e.g.@: request for implicit conversion from
3114 @code{void *} to a pointer to non-@code{void} type.
3118 Warn whenever a pointer is cast so as to remove a type qualifier from
3119 the target type. For example, warn if a @code{const char *} is cast
3120 to an ordinary @code{char *}.
3123 @opindex Wcast-align
3124 Warn whenever a pointer is cast such that the required alignment of the
3125 target is increased. For example, warn if a @code{char *} is cast to
3126 an @code{int *} on machines where integers can only be accessed at
3127 two- or four-byte boundaries.
3129 @item -Wwrite-strings
3130 @opindex Wwrite-strings
3131 When compiling C, give string constants the type @code{const
3132 char[@var{length}]} so that
3133 copying the address of one into a non-@code{const} @code{char *}
3134 pointer will get a warning; when compiling C++, warn about the
3135 deprecated conversion from string literals to @code{char *}. This
3136 warning, by default, is enabled for C++ programs.
3137 These warnings will help you find at
3138 compile time code that can try to write into a string constant, but
3139 only if you have been very careful about using @code{const} in
3140 declarations and prototypes. Otherwise, it will just be a nuisance;
3141 this is why we did not make @option{-Wall} request these warnings.
3144 @opindex Wconversion
3145 Warn if a prototype causes a type conversion that is different from what
3146 would happen to the same argument in the absence of a prototype. This
3147 includes conversions of fixed point to floating and vice versa, and
3148 conversions changing the width or signedness of a fixed point argument
3149 except when the same as the default promotion.
3151 Also, warn if a negative integer constant expression is implicitly
3152 converted to an unsigned type. For example, warn about the assignment
3153 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
3154 casts like @code{(unsigned) -1}.
3156 @item -Wsign-compare
3157 @opindex Wsign-compare
3158 @cindex warning for comparison of signed and unsigned values
3159 @cindex comparison of signed and unsigned values, warning
3160 @cindex signed and unsigned values, comparison warning
3161 Warn when a comparison between signed and unsigned values could produce
3162 an incorrect result when the signed value is converted to unsigned.
3163 This warning is also enabled by @option{-Wextra}; to get the other warnings
3164 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3166 @item -Waggregate-return
3167 @opindex Waggregate-return
3168 Warn if any functions that return structures or unions are defined or
3169 called. (In languages where you can return an array, this also elicits
3173 @opindex Walways-true
3174 Warn about comparisons which are always true such as testing if
3175 unsigned values are greater than or equal to zero. This warning is
3176 enabled by @option{-Wall}.
3178 @item -Wno-attributes
3179 @opindex Wno-attributes
3180 @opindex Wattributes
3181 Do not warn if an unexpected @code{__attribute__} is used, such as
3182 unrecognized attributes, function attributes applied to variables,
3183 etc. This will not stop errors for incorrect use of supported
3186 @item -Wstrict-prototypes @r{(C only)}
3187 @opindex Wstrict-prototypes
3188 Warn if a function is declared or defined without specifying the
3189 argument types. (An old-style function definition is permitted without
3190 a warning if preceded by a declaration which specifies the argument
3193 @item -Wold-style-definition @r{(C only)}
3194 @opindex Wold-style-definition
3195 Warn if an old-style function definition is used. A warning is given
3196 even if there is a previous prototype.
3198 @item -Wmissing-prototypes @r{(C only)}
3199 @opindex Wmissing-prototypes
3200 Warn if a global function is defined without a previous prototype
3201 declaration. This warning is issued even if the definition itself
3202 provides a prototype. The aim is to detect global functions that fail
3203 to be declared in header files.
3205 @item -Wmissing-declarations @r{(C only)}
3206 @opindex Wmissing-declarations
3207 Warn if a global function is defined without a previous declaration.
3208 Do so even if the definition itself provides a prototype.
3209 Use this option to detect global functions that are not declared in
3212 @item -Wmissing-field-initializers
3213 @opindex Wmissing-field-initializers
3216 Warn if a structure's initializer has some fields missing. For
3217 example, the following code would cause such a warning, because
3218 @code{x.h} is implicitly zero:
3221 struct s @{ int f, g, h; @};
3222 struct s x = @{ 3, 4 @};
3225 This option does not warn about designated initializers, so the following
3226 modification would not trigger a warning:
3229 struct s @{ int f, g, h; @};
3230 struct s x = @{ .f = 3, .g = 4 @};
3233 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3234 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3236 @item -Wmissing-noreturn
3237 @opindex Wmissing-noreturn
3238 Warn about functions which might be candidates for attribute @code{noreturn}.
3239 Note these are only possible candidates, not absolute ones. Care should
3240 be taken to manually verify functions actually do not ever return before
3241 adding the @code{noreturn} attribute, otherwise subtle code generation
3242 bugs could be introduced. You will not get a warning for @code{main} in
3243 hosted C environments.
3245 @item -Wmissing-format-attribute
3246 @opindex Wmissing-format-attribute
3248 Warn about function pointers which might be candidates for @code{format}
3249 attributes. Note these are only possible candidates, not absolute ones.
3250 GCC will guess that function pointers with @code{format} attributes that
3251 are used in assignment, initialization, parameter passing or return
3252 statements should have a corresponding @code{format} attribute in the
3253 resulting type. I.e.@: the left-hand side of the assignment or
3254 initialization, the type of the parameter variable, or the return type
3255 of the containing function respectively should also have a @code{format}
3256 attribute to avoid the warning.
3258 GCC will also warn about function definitions which might be
3259 candidates for @code{format} attributes. Again, these are only
3260 possible candidates. GCC will guess that @code{format} attributes
3261 might be appropriate for any function that calls a function like
3262 @code{vprintf} or @code{vscanf}, but this might not always be the
3263 case, and some functions for which @code{format} attributes are
3264 appropriate may not be detected.
3266 @item -Wno-multichar
3267 @opindex Wno-multichar
3269 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3270 Usually they indicate a typo in the user's code, as they have
3271 implementation-defined values, and should not be used in portable code.
3273 @item -Wnormalized=<none|id|nfc|nfkc>
3274 @opindex Wnormalized
3277 @cindex character set, input normalization
3278 In ISO C and ISO C++, two identifiers are different if they are
3279 different sequences of characters. However, sometimes when characters
3280 outside the basic ASCII character set are used, you can have two
3281 different character sequences that look the same. To avoid confusion,
3282 the ISO 10646 standard sets out some @dfn{normalization rules} which
3283 when applied ensure that two sequences that look the same are turned into
3284 the same sequence. GCC can warn you if you are using identifiers which
3285 have not been normalized; this option controls that warning.
3287 There are four levels of warning that GCC supports. The default is
3288 @option{-Wnormalized=nfc}, which warns about any identifier which is
3289 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3290 recommended form for most uses.
3292 Unfortunately, there are some characters which ISO C and ISO C++ allow
3293 in identifiers that when turned into NFC aren't allowable as
3294 identifiers. That is, there's no way to use these symbols in portable
3295 ISO C or C++ and have all your identifiers in NFC.
3296 @option{-Wnormalized=id} suppresses the warning for these characters.
3297 It is hoped that future versions of the standards involved will correct
3298 this, which is why this option is not the default.
3300 You can switch the warning off for all characters by writing
3301 @option{-Wnormalized=none}. You would only want to do this if you
3302 were using some other normalization scheme (like ``D''), because
3303 otherwise you can easily create bugs that are literally impossible to see.
3305 Some characters in ISO 10646 have distinct meanings but look identical
3306 in some fonts or display methodologies, especially once formatting has
3307 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3308 LETTER N'', will display just like a regular @code{n} which has been
3309 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3310 normalization scheme to convert all these into a standard form as
3311 well, and GCC will warn if your code is not in NFKC if you use
3312 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3313 about every identifier that contains the letter O because it might be
3314 confused with the digit 0, and so is not the default, but may be
3315 useful as a local coding convention if the programming environment is
3316 unable to be fixed to display these characters distinctly.
3318 @item -Wno-deprecated-declarations
3319 @opindex Wno-deprecated-declarations
3320 Do not warn about uses of functions (@pxref{Function Attributes}),
3321 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3322 Attributes}) marked as deprecated by using the @code{deprecated}
3326 @opindex Wno-overflow
3327 Do not warn about compile-time overflow in constant expressions.
3329 @item -Woverride-init
3330 @opindex Woverride-init
3333 Warn if an initialized field without side effects is overridden when
3334 using designated initializers (@pxref{Designated Inits, , Designated
3337 This warning is included in @option{-Wextra}. To get other
3338 @option{-Wextra} warnings without this one, use @samp{-Wextra
3339 -Wno-override-init}.
3343 Warn if a structure is given the packed attribute, but the packed
3344 attribute has no effect on the layout or size of the structure.
3345 Such structures may be mis-aligned for little benefit. For
3346 instance, in this code, the variable @code{f.x} in @code{struct bar}
3347 will be misaligned even though @code{struct bar} does not itself
3348 have the packed attribute:
3355 @} __attribute__((packed));
3365 Warn if padding is included in a structure, either to align an element
3366 of the structure or to align the whole structure. Sometimes when this
3367 happens it is possible to rearrange the fields of the structure to
3368 reduce the padding and so make the structure smaller.
3370 @item -Wredundant-decls
3371 @opindex Wredundant-decls
3372 Warn if anything is declared more than once in the same scope, even in
3373 cases where multiple declaration is valid and changes nothing.
3375 @item -Wnested-externs @r{(C only)}
3376 @opindex Wnested-externs
3377 Warn if an @code{extern} declaration is encountered within a function.
3379 @item -Wunreachable-code
3380 @opindex Wunreachable-code
3381 Warn if the compiler detects that code will never be executed.
3383 This option is intended to warn when the compiler detects that at
3384 least a whole line of source code will never be executed, because
3385 some condition is never satisfied or because it is after a
3386 procedure that never returns.
3388 It is possible for this option to produce a warning even though there
3389 are circumstances under which part of the affected line can be executed,
3390 so care should be taken when removing apparently-unreachable code.
3392 For instance, when a function is inlined, a warning may mean that the
3393 line is unreachable in only one inlined copy of the function.
3395 This option is not made part of @option{-Wall} because in a debugging
3396 version of a program there is often substantial code which checks
3397 correct functioning of the program and is, hopefully, unreachable
3398 because the program does work. Another common use of unreachable
3399 code is to provide behavior which is selectable at compile-time.
3403 Warn if a function can not be inlined and it was declared as inline.
3404 Even with this option, the compiler will not warn about failures to
3405 inline functions declared in system headers.
3407 The compiler uses a variety of heuristics to determine whether or not
3408 to inline a function. For example, the compiler takes into account
3409 the size of the function being inlined and the amount of inlining
3410 that has already been done in the current function. Therefore,
3411 seemingly insignificant changes in the source program can cause the
3412 warnings produced by @option{-Winline} to appear or disappear.
3414 @item -Wno-invalid-offsetof @r{(C++ only)}
3415 @opindex Wno-invalid-offsetof
3416 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3417 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3418 to a non-POD type is undefined. In existing C++ implementations,
3419 however, @samp{offsetof} typically gives meaningful results even when
3420 applied to certain kinds of non-POD types. (Such as a simple
3421 @samp{struct} that fails to be a POD type only by virtue of having a
3422 constructor.) This flag is for users who are aware that they are
3423 writing nonportable code and who have deliberately chosen to ignore the
3426 The restrictions on @samp{offsetof} may be relaxed in a future version
3427 of the C++ standard.
3429 @item -Wno-int-to-pointer-cast @r{(C only)}
3430 @opindex Wno-int-to-pointer-cast
3431 Suppress warnings from casts to pointer type of an integer of a
3434 @item -Wno-pointer-to-int-cast @r{(C only)}
3435 @opindex Wno-pointer-to-int-cast
3436 Suppress warnings from casts from a pointer to an integer type of a
3440 @opindex Winvalid-pch
3441 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3442 the search path but can't be used.
3446 @opindex Wno-long-long
3447 Warn if @samp{long long} type is used. This is default. To inhibit
3448 the warning messages, use @option{-Wno-long-long}. Flags
3449 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3450 only when @option{-pedantic} flag is used.
3452 @item -Wvariadic-macros
3453 @opindex Wvariadic-macros
3454 @opindex Wno-variadic-macros
3455 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3456 alternate syntax when in pedantic ISO C99 mode. This is default.
3457 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3459 @item -Wvolatile-register-var
3460 @opindex Wvolatile-register-var
3461 @opindex Wno-volatile-register-var
3462 Warn if a register variable is declared volatile. The volatile
3463 modifier does not inhibit all optimizations that may eliminate reads
3464 and/or writes to register variables.
3466 @item -Wdisabled-optimization
3467 @opindex Wdisabled-optimization
3468 Warn if a requested optimization pass is disabled. This warning does
3469 not generally indicate that there is anything wrong with your code; it
3470 merely indicates that GCC's optimizers were unable to handle the code
3471 effectively. Often, the problem is that your code is too big or too
3472 complex; GCC will refuse to optimize programs when the optimization
3473 itself is likely to take inordinate amounts of time.
3475 @item -Wpointer-sign
3476 @opindex Wpointer-sign
3477 @opindex Wno-pointer-sign
3478 Warn for pointer argument passing or assignment with different signedness.
3479 This option is only supported for C and Objective-C@. It is implied by
3480 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3481 @option{-Wno-pointer-sign}.
3485 Make all warnings into errors.
3489 Make the specified warning into an errors. The specifier for a
3490 warning is appended, for example @option{-Werror=switch} turns the
3491 warnings controlled by @option{-Wswitch} into errors. This switch
3492 takes a negative form, to be used to negate @option{-Werror} for
3493 specific warnings, for example @option{-Wno-error=switch} makes
3494 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3495 is in effect. You can use the @option{-fdiagnostics-show-option}
3496 option to have each controllable warning amended with the option which
3497 controls it, to determine what to use with this option.
3499 Note that specifying @option{-Werror=}@var{foo} automatically implies
3500 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3503 @item -Wstack-protector
3504 @opindex Wstack-protector
3505 This option is only active when @option{-fstack-protector} is active. It
3506 warns about functions that will not be protected against stack smashing.
3508 @item -Wstring-literal-comparison
3509 @opindex Wstring-literal-comparison
3510 Warn about suspicious comparisons to string literal constants. In C,
3511 direct comparisons against the memory address of a string literal, such
3512 as @code{if (x == "abc")}, typically indicate a programmer error, and
3513 even when intentional, result in unspecified behavior and are not portable.
3514 Usually these warnings alert that the programmer intended to use
3515 @code{strcmp}. This warning is enabled by @option{-Wall}.
3517 @item -Woverlength-strings
3518 @opindex Woverlength-strings
3519 Warn about string constants which are longer than the ``minimum
3520 maximum'' length specified in the C standard. Modern compilers
3521 generally allow string constants which are much longer than the
3522 standard's minimum limit, but very portable programs should avoid
3523 using longer strings.
3525 The limit applies @emph{after} string constant concatenation, and does
3526 not count the trailing NUL@. In C89, the limit was 509 characters; in
3527 C99, it was raised to 4095. C++98 does not specify a normative
3528 minimum maximum, so we do not diagnose overlength strings in C++@.
3530 This option is implied by @option{-pedantic}, and can be disabled with
3531 @option{-Wno-overlength-strings}.
3534 @node Debugging Options
3535 @section Options for Debugging Your Program or GCC
3536 @cindex options, debugging
3537 @cindex debugging information options
3539 GCC has various special options that are used for debugging
3540 either your program or GCC:
3545 Produce debugging information in the operating system's native format
3546 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3549 On most systems that use stabs format, @option{-g} enables use of extra
3550 debugging information that only GDB can use; this extra information
3551 makes debugging work better in GDB but will probably make other debuggers
3553 refuse to read the program. If you want to control for certain whether
3554 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3555 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3557 GCC allows you to use @option{-g} with
3558 @option{-O}. The shortcuts taken by optimized code may occasionally
3559 produce surprising results: some variables you declared may not exist
3560 at all; flow of control may briefly move where you did not expect it;
3561 some statements may not be executed because they compute constant
3562 results or their values were already at hand; some statements may
3563 execute in different places because they were moved out of loops.
3565 Nevertheless it proves possible to debug optimized output. This makes
3566 it reasonable to use the optimizer for programs that might have bugs.
3568 The following options are useful when GCC is generated with the
3569 capability for more than one debugging format.
3573 Produce debugging information for use by GDB@. This means to use the
3574 most expressive format available (DWARF 2, stabs, or the native format
3575 if neither of those are supported), including GDB extensions if at all
3580 Produce debugging information in stabs format (if that is supported),
3581 without GDB extensions. This is the format used by DBX on most BSD
3582 systems. On MIPS, Alpha and System V Release 4 systems this option
3583 produces stabs debugging output which is not understood by DBX or SDB@.
3584 On System V Release 4 systems this option requires the GNU assembler.
3586 @item -feliminate-unused-debug-symbols
3587 @opindex feliminate-unused-debug-symbols
3588 Produce debugging information in stabs format (if that is supported),
3589 for only symbols that are actually used.
3591 @item -femit-class-debug-always
3592 Instead of emitting debugging information for a C++ class in only one
3593 object file, emit it in all object files using the class. This option
3594 should be used only with debuggers that are unable to handle the way GCC
3595 normally emits debugging information for classes because using this
3596 option will increase the size of debugging information by as much as a
3601 Produce debugging information in stabs format (if that is supported),
3602 using GNU extensions understood only by the GNU debugger (GDB)@. The
3603 use of these extensions is likely to make other debuggers crash or
3604 refuse to read the program.
3608 Produce debugging information in COFF format (if that is supported).
3609 This is the format used by SDB on most System V systems prior to
3614 Produce debugging information in XCOFF format (if that is supported).
3615 This is the format used by the DBX debugger on IBM RS/6000 systems.
3619 Produce debugging information in XCOFF format (if that is supported),
3620 using GNU extensions understood only by the GNU debugger (GDB)@. The
3621 use of these extensions is likely to make other debuggers crash or
3622 refuse to read the program, and may cause assemblers other than the GNU
3623 assembler (GAS) to fail with an error.
3627 Produce debugging information in DWARF version 2 format (if that is
3628 supported). This is the format used by DBX on IRIX 6. With this
3629 option, GCC uses features of DWARF version 3 when they are useful;
3630 version 3 is upward compatible with version 2, but may still cause
3631 problems for older debuggers.
3635 Produce debugging information in VMS debug format (if that is
3636 supported). This is the format used by DEBUG on VMS systems.
3639 @itemx -ggdb@var{level}
3640 @itemx -gstabs@var{level}
3641 @itemx -gcoff@var{level}
3642 @itemx -gxcoff@var{level}
3643 @itemx -gvms@var{level}
3644 Request debugging information and also use @var{level} to specify how
3645 much information. The default level is 2.
3647 Level 1 produces minimal information, enough for making backtraces in
3648 parts of the program that you don't plan to debug. This includes
3649 descriptions of functions and external variables, but no information
3650 about local variables and no line numbers.
3652 Level 3 includes extra information, such as all the macro definitions
3653 present in the program. Some debuggers support macro expansion when
3654 you use @option{-g3}.
3656 @option{-gdwarf-2} does not accept a concatenated debug level, because
3657 GCC used to support an option @option{-gdwarf} that meant to generate
3658 debug information in version 1 of the DWARF format (which is very
3659 different from version 2), and it would have been too confusing. That
3660 debug format is long obsolete, but the option cannot be changed now.
3661 Instead use an additional @option{-g@var{level}} option to change the
3662 debug level for DWARF2.
3664 @item -feliminate-dwarf2-dups
3665 @opindex feliminate-dwarf2-dups
3666 Compress DWARF2 debugging information by eliminating duplicated
3667 information about each symbol. This option only makes sense when
3668 generating DWARF2 debugging information with @option{-gdwarf-2}.
3670 @cindex @command{prof}
3673 Generate extra code to write profile information suitable for the
3674 analysis program @command{prof}. You must use this option when compiling
3675 the source files you want data about, and you must also use it when
3678 @cindex @command{gprof}
3681 Generate extra code to write profile information suitable for the
3682 analysis program @command{gprof}. You must use this option when compiling
3683 the source files you want data about, and you must also use it when
3688 Makes the compiler print out each function name as it is compiled, and
3689 print some statistics about each pass when it finishes.
3692 @opindex ftime-report
3693 Makes the compiler print some statistics about the time consumed by each
3694 pass when it finishes.
3697 @opindex fmem-report
3698 Makes the compiler print some statistics about permanent memory
3699 allocation when it finishes.
3701 @item -fprofile-arcs
3702 @opindex fprofile-arcs
3703 Add code so that program flow @dfn{arcs} are instrumented. During
3704 execution the program records how many times each branch and call is
3705 executed and how many times it is taken or returns. When the compiled
3706 program exits it saves this data to a file called
3707 @file{@var{auxname}.gcda} for each source file. The data may be used for
3708 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3709 test coverage analysis (@option{-ftest-coverage}). Each object file's
3710 @var{auxname} is generated from the name of the output file, if
3711 explicitly specified and it is not the final executable, otherwise it is
3712 the basename of the source file. In both cases any suffix is removed
3713 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3714 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3715 @xref{Cross-profiling}.
3717 @cindex @command{gcov}
3721 This option is used to compile and link code instrumented for coverage
3722 analysis. The option is a synonym for @option{-fprofile-arcs}
3723 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
3724 linking). See the documentation for those options for more details.
3729 Compile the source files with @option{-fprofile-arcs} plus optimization
3730 and code generation options. For test coverage analysis, use the
3731 additional @option{-ftest-coverage} option. You do not need to profile
3732 every source file in a program.
3735 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3736 (the latter implies the former).
3739 Run the program on a representative workload to generate the arc profile
3740 information. This may be repeated any number of times. You can run
3741 concurrent instances of your program, and provided that the file system
3742 supports locking, the data files will be correctly updated. Also
3743 @code{fork} calls are detected and correctly handled (double counting
3747 For profile-directed optimizations, compile the source files again with
3748 the same optimization and code generation options plus
3749 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3750 Control Optimization}).
3753 For test coverage analysis, use @command{gcov} to produce human readable
3754 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3755 @command{gcov} documentation for further information.
3759 With @option{-fprofile-arcs}, for each function of your program GCC
3760 creates a program flow graph, then finds a spanning tree for the graph.
3761 Only arcs that are not on the spanning tree have to be instrumented: the
3762 compiler adds code to count the number of times that these arcs are
3763 executed. When an arc is the only exit or only entrance to a block, the
3764 instrumentation code can be added to the block; otherwise, a new basic
3765 block must be created to hold the instrumentation code.
3768 @item -ftest-coverage
3769 @opindex ftest-coverage
3770 Produce a notes file that the @command{gcov} code-coverage utility
3771 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3772 show program coverage. Each source file's note file is called
3773 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3774 above for a description of @var{auxname} and instructions on how to
3775 generate test coverage data. Coverage data will match the source files
3776 more closely, if you do not optimize.
3778 @item -d@var{letters}
3779 @item -fdump-rtl-@var{pass}
3781 Says to make debugging dumps during compilation at times specified by
3782 @var{letters}. This is used for debugging the RTL-based passes of the
3783 compiler. The file names for most of the dumps are made by appending a
3784 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3785 from the name of the output file, if explicitly specified and it is not
3786 an executable, otherwise it is the basename of the source file.
3788 Most debug dumps can be enabled either passing a letter to the @option{-d}
3789 option, or with a long @option{-fdump-rtl} switch; here are the possible
3790 letters for use in @var{letters} and @var{pass}, and their meanings:
3795 Annotate the assembler output with miscellaneous debugging information.
3798 @itemx -fdump-rtl-bbro
3800 @opindex fdump-rtl-bbro
3801 Dump after block reordering, to @file{@var{file}.148r.bbro}.
3804 @itemx -fdump-rtl-combine
3806 @opindex fdump-rtl-combine
3807 Dump after instruction combination, to the file @file{@var{file}.129r.combine}.
3810 @itemx -fdump-rtl-ce1
3811 @itemx -fdump-rtl-ce2
3813 @opindex fdump-rtl-ce1
3814 @opindex fdump-rtl-ce2
3815 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3816 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
3817 and @option{-fdump-rtl-ce2} enable dumping after the second if
3818 conversion, to the file @file{@var{file}.130r.ce2}.
3821 @itemx -fdump-rtl-btl
3822 @itemx -fdump-rtl-dbr
3824 @opindex fdump-rtl-btl
3825 @opindex fdump-rtl-dbr
3826 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3827 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
3828 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3829 scheduling, to @file{@var{file}.36.dbr}.
3833 Dump all macro definitions, at the end of preprocessing, in addition to
3837 @itemx -fdump-rtl-ce3
3839 @opindex fdump-rtl-ce3
3840 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
3843 @itemx -fdump-rtl-cfg
3844 @itemx -fdump-rtl-life
3846 @opindex fdump-rtl-cfg
3847 @opindex fdump-rtl-life
3848 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3849 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
3850 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3851 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
3854 @itemx -fdump-rtl-greg
3856 @opindex fdump-rtl-greg
3857 Dump after global register allocation, to @file{@var{file}.139r.greg}.
3860 @itemx -fdump-rtl-gcse
3861 @itemx -fdump-rtl-bypass
3863 @opindex fdump-rtl-gcse
3864 @opindex fdump-rtl-bypass
3865 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3866 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3867 enable dumping after jump bypassing and control flow optimizations, to
3868 @file{@var{file}.115r.bypass}.
3871 @itemx -fdump-rtl-eh
3873 @opindex fdump-rtl-eh
3874 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3877 @itemx -fdump-rtl-sibling
3879 @opindex fdump-rtl-sibling
3880 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
3883 @itemx -fdump-rtl-jump
3885 @opindex fdump-rtl-jump
3886 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
3889 @itemx -fdump-rtl-stack
3891 @opindex fdump-rtl-stack
3892 Dump after conversion from registers to stack, to @file{@var{file}.152r.stack}.
3895 @itemx -fdump-rtl-lreg
3897 @opindex fdump-rtl-lreg
3898 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
3901 @itemx -fdump-rtl-loop2
3903 @opindex fdump-rtl-loop2
3904 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
3905 loop optimization pass, to @file{@var{file}.119r.loop2},
3906 @file{@var{file}.120r.loop2_init},
3907 @file{@var{file}.121r.loop2_invariant}, and
3908 @file{@var{file}.125r.loop2_done}.
3911 @itemx -fdump-rtl-sms
3913 @opindex fdump-rtl-sms
3914 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
3917 @itemx -fdump-rtl-mach
3919 @opindex fdump-rtl-mach
3920 Dump after performing the machine dependent reorganization pass, to
3921 @file{@var{file}.155r.mach}.
3924 @itemx -fdump-rtl-rnreg
3926 @opindex fdump-rtl-rnreg
3927 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
3930 @itemx -fdump-rtl-regmove
3932 @opindex fdump-rtl-regmove
3933 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
3936 @itemx -fdump-rtl-postreload
3938 @opindex fdump-rtl-postreload
3939 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
3942 @itemx -fdump-rtl-expand
3944 @opindex fdump-rtl-expand
3945 Dump after RTL generation, to @file{@var{file}.104r.expand}.
3948 @itemx -fdump-rtl-sched2
3950 @opindex fdump-rtl-sched2
3951 Dump after the second scheduling pass, to @file{@var{file}.149r.sched2}.
3954 @itemx -fdump-rtl-cse
3956 @opindex fdump-rtl-cse
3957 Dump after CSE (including the jump optimization that sometimes follows
3958 CSE), to @file{@var{file}.113r.cse}.
3961 @itemx -fdump-rtl-sched1
3963 @opindex fdump-rtl-sched1
3964 Dump after the first scheduling pass, to @file{@var{file}.136r.sched1}.
3967 @itemx -fdump-rtl-cse2
3969 @opindex fdump-rtl-cse2
3970 Dump after the second CSE pass (including the jump optimization that
3971 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
3974 @itemx -fdump-rtl-tracer
3976 @opindex fdump-rtl-tracer
3977 Dump after running tracer, to @file{@var{file}.118r.tracer}.
3980 @itemx -fdump-rtl-vpt
3981 @itemx -fdump-rtl-vartrack
3983 @opindex fdump-rtl-vpt
3984 @opindex fdump-rtl-vartrack
3985 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
3986 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
3987 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
3988 to @file{@var{file}.154r.vartrack}.
3991 @itemx -fdump-rtl-flow2
3993 @opindex fdump-rtl-flow2
3994 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
3997 @itemx -fdump-rtl-peephole2
3999 @opindex fdump-rtl-peephole2
4000 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
4003 @itemx -fdump-rtl-web
4005 @opindex fdump-rtl-web
4006 Dump after live range splitting, to @file{@var{file}.126r.web}.
4009 @itemx -fdump-rtl-all
4011 @opindex fdump-rtl-all
4012 Produce all the dumps listed above.
4016 Produce a core dump whenever an error occurs.
4020 Print statistics on memory usage, at the end of the run, to
4025 Annotate the assembler output with a comment indicating which
4026 pattern and alternative was used. The length of each instruction is
4031 Dump the RTL in the assembler output as a comment before each instruction.
4032 Also turns on @option{-dp} annotation.
4036 For each of the other indicated dump files (either with @option{-d} or
4037 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4038 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4042 Just generate RTL for a function instead of compiling it. Usually used
4043 with @samp{r} (@option{-fdump-rtl-expand}).
4047 Dump debugging information during parsing, to standard error.
4051 @opindex fdump-noaddr
4052 When doing debugging dumps (see @option{-d} option above), suppress
4053 address output. This makes it more feasible to use diff on debugging
4054 dumps for compiler invocations with different compiler binaries and/or
4055 different text / bss / data / heap / stack / dso start locations.
4057 @item -fdump-unnumbered
4058 @opindex fdump-unnumbered
4059 When doing debugging dumps (see @option{-d} option above), suppress instruction
4060 numbers, line number note and address output. This makes it more feasible to
4061 use diff on debugging dumps for compiler invocations with different
4062 options, in particular with and without @option{-g}.
4064 @item -fdump-translation-unit @r{(C++ only)}
4065 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4066 @opindex fdump-translation-unit
4067 Dump a representation of the tree structure for the entire translation
4068 unit to a file. The file name is made by appending @file{.tu} to the
4069 source file name. If the @samp{-@var{options}} form is used, @var{options}
4070 controls the details of the dump as described for the
4071 @option{-fdump-tree} options.
4073 @item -fdump-class-hierarchy @r{(C++ only)}
4074 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4075 @opindex fdump-class-hierarchy
4076 Dump a representation of each class's hierarchy and virtual function
4077 table layout to a file. The file name is made by appending @file{.class}
4078 to the source file name. If the @samp{-@var{options}} form is used,
4079 @var{options} controls the details of the dump as described for the
4080 @option{-fdump-tree} options.
4082 @item -fdump-ipa-@var{switch}
4084 Control the dumping at various stages of inter-procedural analysis
4085 language tree to a file. The file name is generated by appending a switch
4086 specific suffix to the source file name. The following dumps are possible:
4090 Enables all inter-procedural analysis dumps; currently the only produced
4091 dump is the @samp{cgraph} dump.
4094 Dumps information about call-graph optimization, unused function removal,
4095 and inlining decisions.
4098 @item -fdump-tree-@var{switch}
4099 @itemx -fdump-tree-@var{switch}-@var{options}
4101 Control the dumping at various stages of processing the intermediate
4102 language tree to a file. The file name is generated by appending a switch
4103 specific suffix to the source file name. If the @samp{-@var{options}}
4104 form is used, @var{options} is a list of @samp{-} separated options that
4105 control the details of the dump. Not all options are applicable to all
4106 dumps, those which are not meaningful will be ignored. The following
4107 options are available
4111 Print the address of each node. Usually this is not meaningful as it
4112 changes according to the environment and source file. Its primary use
4113 is for tying up a dump file with a debug environment.
4115 Inhibit dumping of members of a scope or body of a function merely
4116 because that scope has been reached. Only dump such items when they
4117 are directly reachable by some other path. When dumping pretty-printed
4118 trees, this option inhibits dumping the bodies of control structures.
4120 Print a raw representation of the tree. By default, trees are
4121 pretty-printed into a C-like representation.
4123 Enable more detailed dumps (not honored by every dump option).
4125 Enable dumping various statistics about the pass (not honored by every dump
4128 Enable showing basic block boundaries (disabled in raw dumps).
4130 Enable showing virtual operands for every statement.
4132 Enable showing line numbers for statements.
4134 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4136 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4139 The following tree dumps are possible:
4143 Dump before any tree based optimization, to @file{@var{file}.original}.
4146 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4149 Dump after function inlining, to @file{@var{file}.inlined}.
4152 @opindex fdump-tree-gimple
4153 Dump each function before and after the gimplification pass to a file. The
4154 file name is made by appending @file{.gimple} to the source file name.
4157 @opindex fdump-tree-cfg
4158 Dump the control flow graph of each function to a file. The file name is
4159 made by appending @file{.cfg} to the source file name.
4162 @opindex fdump-tree-vcg
4163 Dump the control flow graph of each function to a file in VCG format. The
4164 file name is made by appending @file{.vcg} to the source file name. Note
4165 that if the file contains more than one function, the generated file cannot
4166 be used directly by VCG@. You will need to cut and paste each function's
4167 graph into its own separate file first.
4170 @opindex fdump-tree-ch
4171 Dump each function after copying loop headers. The file name is made by
4172 appending @file{.ch} to the source file name.
4175 @opindex fdump-tree-ssa
4176 Dump SSA related information to a file. The file name is made by appending
4177 @file{.ssa} to the source file name.
4180 @opindex fdump-tree-salias
4181 Dump structure aliasing variable information to a file. This file name
4182 is made by appending @file{.salias} to the source file name.
4185 @opindex fdump-tree-alias
4186 Dump aliasing information for each function. The file name is made by
4187 appending @file{.alias} to the source file name.
4190 @opindex fdump-tree-ccp
4191 Dump each function after CCP@. The file name is made by appending
4192 @file{.ccp} to the source file name.
4195 @opindex fdump-tree-storeccp
4196 Dump each function after STORE-CCP. The file name is made by appending
4197 @file{.storeccp} to the source file name.
4200 @opindex fdump-tree-pre
4201 Dump trees after partial redundancy elimination. The file name is made
4202 by appending @file{.pre} to the source file name.
4205 @opindex fdump-tree-fre
4206 Dump trees after full redundancy elimination. The file name is made
4207 by appending @file{.fre} to the source file name.
4210 @opindex fdump-tree-copyprop
4211 Dump trees after copy propagation. The file name is made
4212 by appending @file{.copyprop} to the source file name.
4214 @item store_copyprop
4215 @opindex fdump-tree-store_copyprop
4216 Dump trees after store copy-propagation. The file name is made
4217 by appending @file{.store_copyprop} to the source file name.
4220 @opindex fdump-tree-dce
4221 Dump each function after dead code elimination. The file name is made by
4222 appending @file{.dce} to the source file name.
4225 @opindex fdump-tree-mudflap
4226 Dump each function after adding mudflap instrumentation. The file name is
4227 made by appending @file{.mudflap} to the source file name.
4230 @opindex fdump-tree-sra
4231 Dump each function after performing scalar replacement of aggregates. The
4232 file name is made by appending @file{.sra} to the source file name.
4235 @opindex fdump-tree-sink
4236 Dump each function after performing code sinking. The file name is made
4237 by appending @file{.sink} to the source file name.
4240 @opindex fdump-tree-dom
4241 Dump each function after applying dominator tree optimizations. The file
4242 name is made by appending @file{.dom} to the source file name.
4245 @opindex fdump-tree-dse
4246 Dump each function after applying dead store elimination. The file
4247 name is made by appending @file{.dse} to the source file name.
4250 @opindex fdump-tree-phiopt
4251 Dump each function after optimizing PHI nodes into straightline code. The file
4252 name is made by appending @file{.phiopt} to the source file name.
4255 @opindex fdump-tree-forwprop
4256 Dump each function after forward propagating single use variables. The file
4257 name is made by appending @file{.forwprop} to the source file name.
4260 @opindex fdump-tree-copyrename
4261 Dump each function after applying the copy rename optimization. The file
4262 name is made by appending @file{.copyrename} to the source file name.
4265 @opindex fdump-tree-nrv
4266 Dump each function after applying the named return value optimization on
4267 generic trees. The file name is made by appending @file{.nrv} to the source
4271 @opindex fdump-tree-vect
4272 Dump each function after applying vectorization of loops. The file name is
4273 made by appending @file{.vect} to the source file name.
4276 @opindex fdump-tree-vrp
4277 Dump each function after Value Range Propagation (VRP). The file name
4278 is made by appending @file{.vrp} to the source file name.
4281 @opindex fdump-tree-all
4282 Enable all the available tree dumps with the flags provided in this option.
4285 @item -ftree-vectorizer-verbose=@var{n}
4286 @opindex ftree-vectorizer-verbose
4287 This option controls the amount of debugging output the vectorizer prints.
4288 This information is written to standard error, unless
4289 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4290 in which case it is output to the usual dump listing file, @file{.vect}.
4291 For @var{n}=0 no diagnostic information is reported.
4292 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4293 and the total number of loops that got vectorized.
4294 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4295 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4296 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4297 level that @option{-fdump-tree-vect-stats} uses.
4298 Higher verbosity levels mean either more information dumped for each
4299 reported loop, or same amount of information reported for more loops:
4300 If @var{n}=3, alignment related information is added to the reports.
4301 If @var{n}=4, data-references related information (e.g. memory dependences,
4302 memory access-patterns) is added to the reports.
4303 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4304 that did not pass the first analysis phase (i.e. may not be countable, or
4305 may have complicated control-flow).
4306 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4307 For @var{n}=7, all the information the vectorizer generates during its
4308 analysis and transformation is reported. This is the same verbosity level
4309 that @option{-fdump-tree-vect-details} uses.
4311 @item -frandom-seed=@var{string}
4312 @opindex frandom-string
4313 This option provides a seed that GCC uses when it would otherwise use
4314 random numbers. It is used to generate certain symbol names
4315 that have to be different in every compiled file. It is also used to
4316 place unique stamps in coverage data files and the object files that
4317 produce them. You can use the @option{-frandom-seed} option to produce
4318 reproducibly identical object files.
4320 The @var{string} should be different for every file you compile.
4322 @item -fsched-verbose=@var{n}
4323 @opindex fsched-verbose
4324 On targets that use instruction scheduling, this option controls the
4325 amount of debugging output the scheduler prints. This information is
4326 written to standard error, unless @option{-dS} or @option{-dR} is
4327 specified, in which case it is output to the usual dump
4328 listing file, @file{.sched} or @file{.sched2} respectively. However
4329 for @var{n} greater than nine, the output is always printed to standard
4332 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4333 same information as @option{-dRS}. For @var{n} greater than one, it
4334 also output basic block probabilities, detailed ready list information
4335 and unit/insn info. For @var{n} greater than two, it includes RTL
4336 at abort point, control-flow and regions info. And for @var{n} over
4337 four, @option{-fsched-verbose} also includes dependence info.
4341 Store the usual ``temporary'' intermediate files permanently; place them
4342 in the current directory and name them based on the source file. Thus,
4343 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4344 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4345 preprocessed @file{foo.i} output file even though the compiler now
4346 normally uses an integrated preprocessor.
4348 When used in combination with the @option{-x} command line option,
4349 @option{-save-temps} is sensible enough to avoid over writing an
4350 input source file with the same extension as an intermediate file.
4351 The corresponding intermediate file may be obtained by renaming the
4352 source file before using @option{-save-temps}.
4356 Report the CPU time taken by each subprocess in the compilation
4357 sequence. For C source files, this is the compiler proper and assembler
4358 (plus the linker if linking is done). The output looks like this:
4365 The first number on each line is the ``user time'', that is time spent
4366 executing the program itself. The second number is ``system time'',
4367 time spent executing operating system routines on behalf of the program.
4368 Both numbers are in seconds.
4370 @item -fvar-tracking
4371 @opindex fvar-tracking
4372 Run variable tracking pass. It computes where variables are stored at each
4373 position in code. Better debugging information is then generated
4374 (if the debugging information format supports this information).
4376 It is enabled by default when compiling with optimization (@option{-Os},
4377 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4378 the debug info format supports it.
4380 @item -print-file-name=@var{library}
4381 @opindex print-file-name
4382 Print the full absolute name of the library file @var{library} that
4383 would be used when linking---and don't do anything else. With this
4384 option, GCC does not compile or link anything; it just prints the
4387 @item -print-multi-directory
4388 @opindex print-multi-directory
4389 Print the directory name corresponding to the multilib selected by any
4390 other switches present in the command line. This directory is supposed
4391 to exist in @env{GCC_EXEC_PREFIX}.
4393 @item -print-multi-lib
4394 @opindex print-multi-lib
4395 Print the mapping from multilib directory names to compiler switches
4396 that enable them. The directory name is separated from the switches by
4397 @samp{;}, and each switch starts with an @samp{@@} instead of the
4398 @samp{-}, without spaces between multiple switches. This is supposed to
4399 ease shell-processing.
4401 @item -print-prog-name=@var{program}
4402 @opindex print-prog-name
4403 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4405 @item -print-libgcc-file-name
4406 @opindex print-libgcc-file-name
4407 Same as @option{-print-file-name=libgcc.a}.
4409 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4410 but you do want to link with @file{libgcc.a}. You can do
4413 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4416 @item -print-search-dirs
4417 @opindex print-search-dirs
4418 Print the name of the configured installation directory and a list of
4419 program and library directories @command{gcc} will search---and don't do anything else.
4421 This is useful when @command{gcc} prints the error message
4422 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4423 To resolve this you either need to put @file{cpp0} and the other compiler
4424 components where @command{gcc} expects to find them, or you can set the environment
4425 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4426 Don't forget the trailing @samp{/}.
4427 @xref{Environment Variables}.
4430 @opindex dumpmachine
4431 Print the compiler's target machine (for example,
4432 @samp{i686-pc-linux-gnu})---and don't do anything else.
4435 @opindex dumpversion
4436 Print the compiler version (for example, @samp{3.0})---and don't do
4441 Print the compiler's built-in specs---and don't do anything else. (This
4442 is used when GCC itself is being built.) @xref{Spec Files}.
4444 @item -feliminate-unused-debug-types
4445 @opindex feliminate-unused-debug-types
4446 Normally, when producing DWARF2 output, GCC will emit debugging
4447 information for all types declared in a compilation
4448 unit, regardless of whether or not they are actually used
4449 in that compilation unit. Sometimes this is useful, such as
4450 if, in the debugger, you want to cast a value to a type that is
4451 not actually used in your program (but is declared). More often,
4452 however, this results in a significant amount of wasted space.
4453 With this option, GCC will avoid producing debug symbol output
4454 for types that are nowhere used in the source file being compiled.
4457 @node Optimize Options
4458 @section Options That Control Optimization
4459 @cindex optimize options
4460 @cindex options, optimization
4462 These options control various sorts of optimizations.
4464 Without any optimization option, the compiler's goal is to reduce the
4465 cost of compilation and to make debugging produce the expected
4466 results. Statements are independent: if you stop the program with a
4467 breakpoint between statements, you can then assign a new value to any
4468 variable or change the program counter to any other statement in the
4469 function and get exactly the results you would expect from the source
4472 Turning on optimization flags makes the compiler attempt to improve
4473 the performance and/or code size at the expense of compilation time
4474 and possibly the ability to debug the program.
4476 The compiler performs optimization based on the knowledge it has of
4477 the program. Optimization levels @option{-O} and above, in
4478 particular, enable @emph{unit-at-a-time} mode, which allows the
4479 compiler to consider information gained from later functions in
4480 the file when compiling a function. Compiling multiple files at
4481 once to a single output file in @emph{unit-at-a-time} mode allows
4482 the compiler to use information gained from all of the files when
4483 compiling each of them.
4485 Not all optimizations are controlled directly by a flag. Only
4486 optimizations that have a flag are listed.
4493 Optimize. Optimizing compilation takes somewhat more time, and a lot
4494 more memory for a large function.
4496 With @option{-O}, the compiler tries to reduce code size and execution
4497 time, without performing any optimizations that take a great deal of
4500 @option{-O} turns on the following optimization flags:
4501 @gccoptlist{-fdefer-pop @gol
4502 -fdelayed-branch @gol
4503 -fguess-branch-probability @gol
4504 -fcprop-registers @gol
4505 -fif-conversion @gol
4506 -fif-conversion2 @gol
4509 -ftree-dominator-opts @gol
4514 -ftree-copyrename @gol
4517 -funit-at-a-time @gol
4520 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4521 where doing so does not interfere with debugging.
4525 Optimize even more. GCC performs nearly all supported optimizations
4526 that do not involve a space-speed tradeoff. The compiler does not
4527 perform loop unrolling or function inlining when you specify @option{-O2}.
4528 As compared to @option{-O}, this option increases both compilation time
4529 and the performance of the generated code.
4531 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4532 also turns on the following optimization flags:
4533 @gccoptlist{-fthread-jumps @gol
4535 -foptimize-sibling-calls @gol
4536 -fcse-follow-jumps -fcse-skip-blocks @gol
4537 -fgcse -fgcse-lm @gol
4538 -fexpensive-optimizations @gol
4539 -frerun-cse-after-loop @gol
4542 -fschedule-insns -fschedule-insns2 @gol
4543 -fsched-interblock -fsched-spec @gol
4545 -fstrict-aliasing @gol
4546 -fdelete-null-pointer-checks @gol
4547 -freorder-blocks -freorder-functions @gol
4548 -falign-functions -falign-jumps @gol
4549 -falign-loops -falign-labels @gol
4553 Please note the warning under @option{-fgcse} about
4554 invoking @option{-O2} on programs that use computed gotos.
4558 Optimize yet more. @option{-O3} turns on all optimizations specified by
4559 @option{-O2} and also turns on the @option{-finline-functions},
4560 @option{-funswitch-loops} and @option{-fgcse-after-reload} options.
4564 Do not optimize. This is the default.
4568 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4569 do not typically increase code size. It also performs further
4570 optimizations designed to reduce code size.
4572 @option{-Os} disables the following optimization flags:
4573 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4574 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
4575 -fprefetch-loop-arrays -ftree-vect-loop-version}
4577 If you use multiple @option{-O} options, with or without level numbers,
4578 the last such option is the one that is effective.
4581 Options of the form @option{-f@var{flag}} specify machine-independent
4582 flags. Most flags have both positive and negative forms; the negative
4583 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4584 below, only one of the forms is listed---the one you typically will
4585 use. You can figure out the other form by either removing @samp{no-}
4588 The following options control specific optimizations. They are either
4589 activated by @option{-O} options or are related to ones that are. You
4590 can use the following flags in the rare cases when ``fine-tuning'' of
4591 optimizations to be performed is desired.
4594 @item -fno-default-inline
4595 @opindex fno-default-inline
4596 Do not make member functions inline by default merely because they are
4597 defined inside the class scope (C++ only). Otherwise, when you specify
4598 @w{@option{-O}}, member functions defined inside class scope are compiled
4599 inline by default; i.e., you don't need to add @samp{inline} in front of
4600 the member function name.
4602 @item -fno-defer-pop
4603 @opindex fno-defer-pop
4604 Always pop the arguments to each function call as soon as that function
4605 returns. For machines which must pop arguments after a function call,
4606 the compiler normally lets arguments accumulate on the stack for several
4607 function calls and pops them all at once.
4609 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4613 Force memory operands to be copied into registers before doing
4614 arithmetic on them. This produces better code by making all memory
4615 references potential common subexpressions. When they are not common
4616 subexpressions, instruction combination should eliminate the separate
4617 register-load. This option is now a nop and will be removed in 4.2.
4620 @opindex fforce-addr
4621 Force memory address constants to be copied into registers before
4622 doing arithmetic on them.
4624 @item -fomit-frame-pointer
4625 @opindex fomit-frame-pointer
4626 Don't keep the frame pointer in a register for functions that
4627 don't need one. This avoids the instructions to save, set up and
4628 restore frame pointers; it also makes an extra register available
4629 in many functions. @strong{It also makes debugging impossible on
4632 On some machines, such as the VAX, this flag has no effect, because
4633 the standard calling sequence automatically handles the frame pointer
4634 and nothing is saved by pretending it doesn't exist. The
4635 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4636 whether a target machine supports this flag. @xref{Registers,,Register
4637 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4639 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4641 @item -foptimize-sibling-calls
4642 @opindex foptimize-sibling-calls
4643 Optimize sibling and tail recursive calls.
4645 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4649 Don't pay attention to the @code{inline} keyword. Normally this option
4650 is used to keep the compiler from expanding any functions inline.
4651 Note that if you are not optimizing, no functions can be expanded inline.
4653 @item -finline-functions
4654 @opindex finline-functions
4655 Integrate all simple functions into their callers. The compiler
4656 heuristically decides which functions are simple enough to be worth
4657 integrating in this way.
4659 If all calls to a given function are integrated, and the function is
4660 declared @code{static}, then the function is normally not output as
4661 assembler code in its own right.
4663 Enabled at level @option{-O3}.
4665 @item -finline-functions-called-once
4666 @opindex finline-functions-called-once
4667 Consider all @code{static} functions called once for inlining into their
4668 caller even if they are not marked @code{inline}. If a call to a given
4669 function is integrated, then the function is not output as assembler code
4672 Enabled if @option{-funit-at-a-time} is enabled.
4674 @item -fearly-inlining
4675 @opindex fearly-inlining
4676 Inline functions marked by @code{always_inline} and functions whose body seems
4677 smaller than the function call overhead early before doing
4678 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
4679 makes profiling significantly cheaper and usually inlining faster on programs
4680 having large chains of nested wrapper functions.
4684 @item -finline-limit=@var{n}
4685 @opindex finline-limit
4686 By default, GCC limits the size of functions that can be inlined. This flag
4687 allows the control of this limit for functions that are explicitly marked as
4688 inline (i.e., marked with the inline keyword or defined within the class
4689 definition in c++). @var{n} is the size of functions that can be inlined in
4690 number of pseudo instructions (not counting parameter handling). The default
4691 value of @var{n} is 600.
4692 Increasing this value can result in more inlined code at
4693 the cost of compilation time and memory consumption. Decreasing usually makes
4694 the compilation faster and less code will be inlined (which presumably
4695 means slower programs). This option is particularly useful for programs that
4696 use inlining heavily such as those based on recursive templates with C++.
4698 Inlining is actually controlled by a number of parameters, which may be
4699 specified individually by using @option{--param @var{name}=@var{value}}.
4700 The @option{-finline-limit=@var{n}} option sets some of these parameters
4704 @item max-inline-insns-single
4705 is set to @var{n}/2.
4706 @item max-inline-insns-auto
4707 is set to @var{n}/2.
4708 @item min-inline-insns
4709 is set to 130 or @var{n}/4, whichever is smaller.
4710 @item max-inline-insns-rtl
4714 See below for a documentation of the individual
4715 parameters controlling inlining.
4717 @emph{Note:} pseudo instruction represents, in this particular context, an
4718 abstract measurement of function's size. In no way does it represent a count
4719 of assembly instructions and as such its exact meaning might change from one
4720 release to an another.
4722 @item -fkeep-inline-functions
4723 @opindex fkeep-inline-functions
4724 In C, emit @code{static} functions that are declared @code{inline}
4725 into the object file, even if the function has been inlined into all
4726 of its callers. This switch does not affect functions using the
4727 @code{extern inline} extension in GNU C89@. In C++, emit any and all
4728 inline functions into the object file.
4730 @item -fkeep-static-consts
4731 @opindex fkeep-static-consts
4732 Emit variables declared @code{static const} when optimization isn't turned
4733 on, even if the variables aren't referenced.
4735 GCC enables this option by default. If you want to force the compiler to
4736 check if the variable was referenced, regardless of whether or not
4737 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4739 @item -fmerge-constants
4740 Attempt to merge identical constants (string constants and floating point
4741 constants) across compilation units.
4743 This option is the default for optimized compilation if the assembler and
4744 linker support it. Use @option{-fno-merge-constants} to inhibit this
4747 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4749 @item -fmerge-all-constants
4750 Attempt to merge identical constants and identical variables.
4752 This option implies @option{-fmerge-constants}. In addition to
4753 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4754 arrays or initialized constant variables with integral or floating point
4755 types. Languages like C or C++ require each non-automatic variable to
4756 have distinct location, so using this option will result in non-conforming
4759 @item -fmodulo-sched
4760 @opindex fmodulo-sched
4761 Perform swing modulo scheduling immediately before the first scheduling
4762 pass. This pass looks at innermost loops and reorders their
4763 instructions by overlapping different iterations.
4765 @item -fno-branch-count-reg
4766 @opindex fno-branch-count-reg
4767 Do not use ``decrement and branch'' instructions on a count register,
4768 but instead generate a sequence of instructions that decrement a
4769 register, compare it against zero, then branch based upon the result.
4770 This option is only meaningful on architectures that support such
4771 instructions, which include x86, PowerPC, IA-64 and S/390.
4773 The default is @option{-fbranch-count-reg}.
4775 @item -fno-function-cse
4776 @opindex fno-function-cse
4777 Do not put function addresses in registers; make each instruction that
4778 calls a constant function contain the function's address explicitly.
4780 This option results in less efficient code, but some strange hacks
4781 that alter the assembler output may be confused by the optimizations
4782 performed when this option is not used.
4784 The default is @option{-ffunction-cse}
4786 @item -fno-zero-initialized-in-bss
4787 @opindex fno-zero-initialized-in-bss
4788 If the target supports a BSS section, GCC by default puts variables that
4789 are initialized to zero into BSS@. This can save space in the resulting
4792 This option turns off this behavior because some programs explicitly
4793 rely on variables going to the data section. E.g., so that the
4794 resulting executable can find the beginning of that section and/or make
4795 assumptions based on that.
4797 The default is @option{-fzero-initialized-in-bss}.
4799 @item -fbounds-check
4800 @opindex fbounds-check
4801 For front-ends that support it, generate additional code to check that
4802 indices used to access arrays are within the declared range. This is
4803 currently only supported by the Java and Fortran front-ends, where
4804 this option defaults to true and false respectively.
4806 @item -fmudflap -fmudflapth -fmudflapir
4810 @cindex bounds checking
4812 For front-ends that support it (C and C++), instrument all risky
4813 pointer/array dereferencing operations, some standard library
4814 string/heap functions, and some other associated constructs with
4815 range/validity tests. Modules so instrumented should be immune to
4816 buffer overflows, invalid heap use, and some other classes of C/C++
4817 programming errors. The instrumentation relies on a separate runtime
4818 library (@file{libmudflap}), which will be linked into a program if
4819 @option{-fmudflap} is given at link time. Run-time behavior of the
4820 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4821 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4824 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4825 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4826 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4827 instrumentation should ignore pointer reads. This produces less
4828 instrumentation (and therefore faster execution) and still provides
4829 some protection against outright memory corrupting writes, but allows
4830 erroneously read data to propagate within a program.
4832 @item -fthread-jumps
4833 @opindex fthread-jumps
4834 Perform optimizations where we check to see if a jump branches to a
4835 location where another comparison subsumed by the first is found. If
4836 so, the first branch is redirected to either the destination of the
4837 second branch or a point immediately following it, depending on whether
4838 the condition is known to be true or false.
4840 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4842 @item -fcse-follow-jumps
4843 @opindex fcse-follow-jumps
4844 In common subexpression elimination, scan through jump instructions
4845 when the target of the jump is not reached by any other path. For
4846 example, when CSE encounters an @code{if} statement with an
4847 @code{else} clause, CSE will follow the jump when the condition
4850 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4852 @item -fcse-skip-blocks
4853 @opindex fcse-skip-blocks
4854 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4855 follow jumps which conditionally skip over blocks. When CSE
4856 encounters a simple @code{if} statement with no else clause,
4857 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4858 body of the @code{if}.
4860 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4862 @item -frerun-cse-after-loop
4863 @opindex frerun-cse-after-loop
4864 Re-run common subexpression elimination after loop optimizations has been
4867 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4871 Perform a global common subexpression elimination pass.
4872 This pass also performs global constant and copy propagation.
4874 @emph{Note:} When compiling a program using computed gotos, a GCC
4875 extension, you may get better runtime performance if you disable
4876 the global common subexpression elimination pass by adding
4877 @option{-fno-gcse} to the command line.
4879 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4883 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4884 attempt to move loads which are only killed by stores into themselves. This
4885 allows a loop containing a load/store sequence to be changed to a load outside
4886 the loop, and a copy/store within the loop.
4888 Enabled by default when gcse is enabled.
4892 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4893 global common subexpression elimination. This pass will attempt to move
4894 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4895 loops containing a load/store sequence can be changed to a load before
4896 the loop and a store after the loop.
4898 Not enabled at any optimization level.
4902 When @option{-fgcse-las} is enabled, the global common subexpression
4903 elimination pass eliminates redundant loads that come after stores to the
4904 same memory location (both partial and full redundancies).
4906 Not enabled at any optimization level.
4908 @item -fgcse-after-reload
4909 @opindex fgcse-after-reload
4910 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4911 pass is performed after reload. The purpose of this pass is to cleanup
4914 @item -funsafe-loop-optimizations
4915 @opindex funsafe-loop-optimizations
4916 If given, the loop optimizer will assume that loop indices do not
4917 overflow, and that the loops with nontrivial exit condition are not
4918 infinite. This enables a wider range of loop optimizations even if
4919 the loop optimizer itself cannot prove that these assumptions are valid.
4920 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
4921 if it finds this kind of loop.
4923 @item -fcrossjumping
4924 @opindex crossjumping
4925 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4926 resulting code may or may not perform better than without cross-jumping.
4928 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4930 @item -fif-conversion
4931 @opindex if-conversion
4932 Attempt to transform conditional jumps into branch-less equivalents. This
4933 include use of conditional moves, min, max, set flags and abs instructions, and
4934 some tricks doable by standard arithmetics. The use of conditional execution
4935 on chips where it is available is controlled by @code{if-conversion2}.
4937 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4939 @item -fif-conversion2
4940 @opindex if-conversion2
4941 Use conditional execution (where available) to transform conditional jumps into
4942 branch-less equivalents.
4944 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4946 @item -fdelete-null-pointer-checks
4947 @opindex fdelete-null-pointer-checks
4948 Use global dataflow analysis to identify and eliminate useless checks
4949 for null pointers. The compiler assumes that dereferencing a null
4950 pointer would have halted the program. If a pointer is checked after
4951 it has already been dereferenced, it cannot be null.
4953 In some environments, this assumption is not true, and programs can
4954 safely dereference null pointers. Use
4955 @option{-fno-delete-null-pointer-checks} to disable this optimization
4956 for programs which depend on that behavior.
4958 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4960 @item -fexpensive-optimizations
4961 @opindex fexpensive-optimizations
4962 Perform a number of minor optimizations that are relatively expensive.
4964 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4966 @item -foptimize-register-move
4968 @opindex foptimize-register-move
4970 Attempt to reassign register numbers in move instructions and as
4971 operands of other simple instructions in order to maximize the amount of
4972 register tying. This is especially helpful on machines with two-operand
4975 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4978 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4980 @item -fdelayed-branch
4981 @opindex fdelayed-branch
4982 If supported for the target machine, attempt to reorder instructions
4983 to exploit instruction slots available after delayed branch
4986 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4988 @item -fschedule-insns
4989 @opindex fschedule-insns
4990 If supported for the target machine, attempt to reorder instructions to
4991 eliminate execution stalls due to required data being unavailable. This
4992 helps machines that have slow floating point or memory load instructions
4993 by allowing other instructions to be issued until the result of the load
4994 or floating point instruction is required.
4996 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4998 @item -fschedule-insns2
4999 @opindex fschedule-insns2
5000 Similar to @option{-fschedule-insns}, but requests an additional pass of
5001 instruction scheduling after register allocation has been done. This is
5002 especially useful on machines with a relatively small number of
5003 registers and where memory load instructions take more than one cycle.
5005 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5007 @item -fno-sched-interblock
5008 @opindex fno-sched-interblock
5009 Don't schedule instructions across basic blocks. This is normally
5010 enabled by default when scheduling before register allocation, i.e.@:
5011 with @option{-fschedule-insns} or at @option{-O2} or higher.
5013 @item -fno-sched-spec
5014 @opindex fno-sched-spec
5015 Don't allow speculative motion of non-load instructions. This is normally
5016 enabled by default when scheduling before register allocation, i.e.@:
5017 with @option{-fschedule-insns} or at @option{-O2} or higher.
5019 @item -fsched-spec-load
5020 @opindex fsched-spec-load
5021 Allow speculative motion of some load instructions. This only makes
5022 sense when scheduling before register allocation, i.e.@: with
5023 @option{-fschedule-insns} or at @option{-O2} or higher.
5025 @item -fsched-spec-load-dangerous
5026 @opindex fsched-spec-load-dangerous
5027 Allow speculative motion of more load instructions. This only makes
5028 sense when scheduling before register allocation, i.e.@: with
5029 @option{-fschedule-insns} or at @option{-O2} or higher.
5031 @item -fsched-stalled-insns=@var{n}
5032 @opindex fsched-stalled-insns
5033 Define how many insns (if any) can be moved prematurely from the queue
5034 of stalled insns into the ready list, during the second scheduling pass.
5036 @item -fsched-stalled-insns-dep=@var{n}
5037 @opindex fsched-stalled-insns-dep
5038 Define how many insn groups (cycles) will be examined for a dependency
5039 on a stalled insn that is candidate for premature removal from the queue
5040 of stalled insns. Has an effect only during the second scheduling pass,
5041 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
5043 @item -fsched2-use-superblocks
5044 @opindex fsched2-use-superblocks
5045 When scheduling after register allocation, do use superblock scheduling
5046 algorithm. Superblock scheduling allows motion across basic block boundaries
5047 resulting on faster schedules. This option is experimental, as not all machine
5048 descriptions used by GCC model the CPU closely enough to avoid unreliable
5049 results from the algorithm.
5051 This only makes sense when scheduling after register allocation, i.e.@: with
5052 @option{-fschedule-insns2} or at @option{-O2} or higher.
5054 @item -fsched2-use-traces
5055 @opindex fsched2-use-traces
5056 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5057 allocation and additionally perform code duplication in order to increase the
5058 size of superblocks using tracer pass. See @option{-ftracer} for details on
5061 This mode should produce faster but significantly longer programs. Also
5062 without @option{-fbranch-probabilities} the traces constructed may not
5063 match the reality and hurt the performance. This only makes
5064 sense when scheduling after register allocation, i.e.@: with
5065 @option{-fschedule-insns2} or at @option{-O2} or higher.
5069 Eliminates redundant extension instructions and move the non redundant
5070 ones to optimal placement using LCM.
5072 @item -freschedule-modulo-scheduled-loops
5073 @opindex fscheduling-in-modulo-scheduled-loops
5074 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
5075 we may want to prevent the later scheduling passes from changing its schedule, we use this
5076 option to control that.
5078 @item -fcaller-saves
5079 @opindex fcaller-saves
5080 Enable values to be allocated in registers that will be clobbered by
5081 function calls, by emitting extra instructions to save and restore the
5082 registers around such calls. Such allocation is done only when it
5083 seems to result in better code than would otherwise be produced.
5085 This option is always enabled by default on certain machines, usually
5086 those which have no call-preserved registers to use instead.
5088 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5091 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
5092 enabled by default at @option{-O2} and @option{-O3}.
5095 Perform Full Redundancy Elimination (FRE) on trees. The difference
5096 between FRE and PRE is that FRE only considers expressions
5097 that are computed on all paths leading to the redundant computation.
5098 This analysis faster than PRE, though it exposes fewer redundancies.
5099 This flag is enabled by default at @option{-O} and higher.
5101 @item -ftree-copy-prop
5102 Perform copy propagation on trees. This pass eliminates unnecessary
5103 copy operations. This flag is enabled by default at @option{-O} and
5106 @item -ftree-store-copy-prop
5107 Perform copy propagation of memory loads and stores. This pass
5108 eliminates unnecessary copy operations in memory references
5109 (structures, global variables, arrays, etc). This flag is enabled by
5110 default at @option{-O2} and higher.
5113 Perform structural alias analysis on trees. This flag
5114 is enabled by default at @option{-O} and higher.
5117 Perform interprocedural pointer analysis.
5120 Perform forward store motion on trees. This flag is
5121 enabled by default at @option{-O} and higher.
5124 Perform sparse conditional constant propagation (CCP) on trees. This
5125 pass only operates on local scalar variables and is enabled by default
5126 at @option{-O} and higher.
5128 @item -ftree-store-ccp
5129 Perform sparse conditional constant propagation (CCP) on trees. This
5130 pass operates on both local scalar variables and memory stores and
5131 loads (global variables, structures, arrays, etc). This flag is
5132 enabled by default at @option{-O2} and higher.
5135 Perform dead code elimination (DCE) on trees. This flag is enabled by
5136 default at @option{-O} and higher.
5138 @item -ftree-dominator-opts
5139 Perform a variety of simple scalar cleanups (constant/copy
5140 propagation, redundancy elimination, range propagation and expression
5141 simplification) based on a dominator tree traversal. This also
5142 performs jump threading (to reduce jumps to jumps). This flag is
5143 enabled by default at @option{-O} and higher.
5146 Perform loop header copying on trees. This is beneficial since it increases
5147 effectiveness of code motion optimizations. It also saves one jump. This flag
5148 is enabled by default at @option{-O} and higher. It is not enabled
5149 for @option{-Os}, since it usually increases code size.
5151 @item -ftree-loop-optimize
5152 Perform loop optimizations on trees. This flag is enabled by default
5153 at @option{-O} and higher.
5155 @item -ftree-loop-linear
5156 Perform linear loop transformations on tree. This flag can improve cache
5157 performance and allow further loop optimizations to take place.
5159 @item -ftree-loop-im
5160 Perform loop invariant motion on trees. This pass moves only invariants that
5161 would be hard to handle at RTL level (function calls, operations that expand to
5162 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5163 operands of conditions that are invariant out of the loop, so that we can use
5164 just trivial invariantness analysis in loop unswitching. The pass also includes
5167 @item -ftree-loop-ivcanon
5168 Create a canonical counter for number of iterations in the loop for that
5169 determining number of iterations requires complicated analysis. Later
5170 optimizations then may determine the number easily. Useful especially
5171 in connection with unrolling.
5174 Perform induction variable optimizations (strength reduction, induction
5175 variable merging and induction variable elimination) on trees.
5178 Perform scalar replacement of aggregates. This pass replaces structure
5179 references with scalars to prevent committing structures to memory too
5180 early. This flag is enabled by default at @option{-O} and higher.
5182 @item -ftree-copyrename
5183 Perform copy renaming on trees. This pass attempts to rename compiler
5184 temporaries to other variables at copy locations, usually resulting in
5185 variable names which more closely resemble the original variables. This flag
5186 is enabled by default at @option{-O} and higher.
5189 Perform temporary expression replacement during the SSA->normal phase. Single
5190 use/single def temporaries are replaced at their use location with their
5191 defining expression. This results in non-GIMPLE code, but gives the expanders
5192 much more complex trees to work on resulting in better RTL generation. This is
5193 enabled by default at @option{-O} and higher.
5196 Perform live range splitting during the SSA->normal phase. Distinct live
5197 ranges of a variable are split into unique variables, allowing for better
5198 optimization later. This is enabled by default at @option{-O} and higher.
5200 @item -ftree-vectorize
5201 Perform loop vectorization on trees.
5203 @item -ftree-vect-loop-version
5204 @opindex ftree-vect-loop-version
5205 Perform loop versioning when doing loop vectorization on trees. When a loop
5206 appears to be vectorizable except that data alignment or data dependence cannot
5207 be determined at compile time then vectorized and non-vectorized versions of
5208 the loop are generated along with runtime checks for alignment or dependence
5209 to control which version is executed. This option is enabled by default
5210 except at level @option{-Os} where it is disabled.
5213 Perform Value Range Propagation on trees. This is similar to the
5214 constant propagation pass, but instead of values, ranges of values are
5215 propagated. This allows the optimizers to remove unnecessary range
5216 checks like array bound checks and null pointer checks. This is
5217 enabled by default at @option{-O2} and higher. Null pointer check
5218 elimination is only done if @option{-fdelete-null-pointer-checks} is
5223 Perform tail duplication to enlarge superblock size. This transformation
5224 simplifies the control flow of the function allowing other optimizations to do
5227 @item -funroll-loops
5228 @opindex funroll-loops
5229 Unroll loops whose number of iterations can be determined at compile
5230 time or upon entry to the loop. @option{-funroll-loops} implies
5231 @option{-frerun-cse-after-loop}. This option makes code larger,
5232 and may or may not make it run faster.
5234 @item -funroll-all-loops
5235 @opindex funroll-all-loops
5236 Unroll all loops, even if their number of iterations is uncertain when
5237 the loop is entered. This usually makes programs run more slowly.
5238 @option{-funroll-all-loops} implies the same options as
5239 @option{-funroll-loops},
5241 @item -fsplit-ivs-in-unroller
5242 @opindex -fsplit-ivs-in-unroller
5243 Enables expressing of values of induction variables in later iterations
5244 of the unrolled loop using the value in the first iteration. This breaks
5245 long dependency chains, thus improving efficiency of the scheduling passes.
5247 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5248 same effect. However in cases the loop body is more complicated than
5249 a single basic block, this is not reliable. It also does not work at all
5250 on some of the architectures due to restrictions in the CSE pass.
5252 This optimization is enabled by default.
5254 @item -fvariable-expansion-in-unroller
5255 @opindex -fvariable-expansion-in-unroller
5256 With this option, the compiler will create multiple copies of some
5257 local variables when unrolling a loop which can result in superior code.
5259 @item -fprefetch-loop-arrays
5260 @opindex fprefetch-loop-arrays
5261 If supported by the target machine, generate instructions to prefetch
5262 memory to improve the performance of loops that access large arrays.
5264 This option may generate better or worse code; results are highly
5265 dependent on the structure of loops within the source code.
5267 Disabled at level @option{-Os}.
5270 @itemx -fno-peephole2
5271 @opindex fno-peephole
5272 @opindex fno-peephole2
5273 Disable any machine-specific peephole optimizations. The difference
5274 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5275 are implemented in the compiler; some targets use one, some use the
5276 other, a few use both.
5278 @option{-fpeephole} is enabled by default.
5279 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5281 @item -fno-guess-branch-probability
5282 @opindex fno-guess-branch-probability
5283 Do not guess branch probabilities using heuristics.
5285 GCC will use heuristics to guess branch probabilities if they are
5286 not provided by profiling feedback (@option{-fprofile-arcs}). These
5287 heuristics are based on the control flow graph. If some branch probabilities
5288 are specified by @samp{__builtin_expect}, then the heuristics will be
5289 used to guess branch probabilities for the rest of the control flow graph,
5290 taking the @samp{__builtin_expect} info into account. The interactions
5291 between the heuristics and @samp{__builtin_expect} can be complex, and in
5292 some cases, it may be useful to disable the heuristics so that the effects
5293 of @samp{__builtin_expect} are easier to understand.
5295 The default is @option{-fguess-branch-probability} at levels
5296 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5298 @item -freorder-blocks
5299 @opindex freorder-blocks
5300 Reorder basic blocks in the compiled function in order to reduce number of
5301 taken branches and improve code locality.
5303 Enabled at levels @option{-O2}, @option{-O3}.
5305 @item -freorder-blocks-and-partition
5306 @opindex freorder-blocks-and-partition
5307 In addition to reordering basic blocks in the compiled function, in order
5308 to reduce number of taken branches, partitions hot and cold basic blocks
5309 into separate sections of the assembly and .o files, to improve
5310 paging and cache locality performance.
5312 This optimization is automatically turned off in the presence of
5313 exception handling, for linkonce sections, for functions with a user-defined
5314 section attribute and on any architecture that does not support named
5317 @item -freorder-functions
5318 @opindex freorder-functions
5319 Reorder functions in the object file in order to
5320 improve code locality. This is implemented by using special
5321 subsections @code{.text.hot} for most frequently executed functions and
5322 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5323 the linker so object file format must support named sections and linker must
5324 place them in a reasonable way.
5326 Also profile feedback must be available in to make this option effective. See
5327 @option{-fprofile-arcs} for details.
5329 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5331 @item -fstrict-aliasing
5332 @opindex fstrict-aliasing
5333 Allows the compiler to assume the strictest aliasing rules applicable to
5334 the language being compiled. For C (and C++), this activates
5335 optimizations based on the type of expressions. In particular, an
5336 object of one type is assumed never to reside at the same address as an
5337 object of a different type, unless the types are almost the same. For
5338 example, an @code{unsigned int} can alias an @code{int}, but not a
5339 @code{void*} or a @code{double}. A character type may alias any other
5342 Pay special attention to code like this:
5355 The practice of reading from a different union member than the one most
5356 recently written to (called ``type-punning'') is common. Even with
5357 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5358 is accessed through the union type. So, the code above will work as
5359 expected. However, this code might not:
5370 Every language that wishes to perform language-specific alias analysis
5371 should define a function that computes, given an @code{tree}
5372 node, an alias set for the node. Nodes in different alias sets are not
5373 allowed to alias. For an example, see the C front-end function
5374 @code{c_get_alias_set}.
5376 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5378 @item -falign-functions
5379 @itemx -falign-functions=@var{n}
5380 @opindex falign-functions
5381 Align the start of functions to the next power-of-two greater than
5382 @var{n}, skipping up to @var{n} bytes. For instance,
5383 @option{-falign-functions=32} aligns functions to the next 32-byte
5384 boundary, but @option{-falign-functions=24} would align to the next
5385 32-byte boundary only if this can be done by skipping 23 bytes or less.
5387 @option{-fno-align-functions} and @option{-falign-functions=1} are
5388 equivalent and mean that functions will not be aligned.
5390 Some assemblers only support this flag when @var{n} is a power of two;
5391 in that case, it is rounded up.
5393 If @var{n} is not specified or is zero, use a machine-dependent default.
5395 Enabled at levels @option{-O2}, @option{-O3}.
5397 @item -falign-labels
5398 @itemx -falign-labels=@var{n}
5399 @opindex falign-labels
5400 Align all branch targets to a power-of-two boundary, skipping up to
5401 @var{n} bytes like @option{-falign-functions}. This option can easily
5402 make code slower, because it must insert dummy operations for when the
5403 branch target is reached in the usual flow of the code.
5405 @option{-fno-align-labels} and @option{-falign-labels=1} are
5406 equivalent and mean that labels will not be aligned.
5408 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
5409 are greater than this value, then their values are used instead.
5411 If @var{n} is not specified or is zero, use a machine-dependent default
5412 which is very likely to be @samp{1}, meaning no alignment.
5414 Enabled at levels @option{-O2}, @option{-O3}.
5417 @itemx -falign-loops=@var{n}
5418 @opindex falign-loops
5419 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
5420 like @option{-falign-functions}. The hope is that the loop will be
5421 executed many times, which will make up for any execution of the dummy
5424 @option{-fno-align-loops} and @option{-falign-loops=1} are
5425 equivalent and mean that loops will not be aligned.
5427 If @var{n} is not specified or is zero, use a machine-dependent default.
5429 Enabled at levels @option{-O2}, @option{-O3}.
5432 @itemx -falign-jumps=@var{n}
5433 @opindex falign-jumps
5434 Align branch targets to a power-of-two boundary, for branch targets
5435 where the targets can only be reached by jumping, skipping up to @var{n}
5436 bytes like @option{-falign-functions}. In this case, no dummy operations
5439 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
5440 equivalent and mean that loops will not be aligned.
5442 If @var{n} is not specified or is zero, use a machine-dependent default.
5444 Enabled at levels @option{-O2}, @option{-O3}.
5446 @item -funit-at-a-time
5447 @opindex funit-at-a-time
5448 Parse the whole compilation unit before starting to produce code.
5449 This allows some extra optimizations to take place but consumes
5450 more memory (in general). There are some compatibility issues
5451 with @emph{unit-at-a-time} mode:
5454 enabling @emph{unit-at-a-time} mode may change the order
5455 in which functions, variables, and top-level @code{asm} statements
5456 are emitted, and will likely break code relying on some particular
5457 ordering. The majority of such top-level @code{asm} statements,
5458 though, can be replaced by @code{section} attributes. The
5459 @option{fno-toplevel-reorder} option may be used to keep the ordering
5460 used in the input file, at the cost of some optimizations.
5463 @emph{unit-at-a-time} mode removes unreferenced static variables
5464 and functions. This may result in undefined references
5465 when an @code{asm} statement refers directly to variables or functions
5466 that are otherwise unused. In that case either the variable/function
5467 shall be listed as an operand of the @code{asm} statement operand or,
5468 in the case of top-level @code{asm} statements the attribute @code{used}
5469 shall be used on the declaration.
5472 Static functions now can use non-standard passing conventions that
5473 may break @code{asm} statements calling functions directly. Again,
5474 attribute @code{used} will prevent this behavior.
5477 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
5478 but this scheme may not be supported by future releases of GCC@.
5480 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5482 @item -fno-toplevel-reorder
5483 Do not reorder top-level functions, variables, and @code{asm}
5484 statements. Output them in the same order that they appear in the
5485 input file. When this option is used, unreferenced static variables
5486 will not be removed. This option is intended to support existing code
5487 which relies on a particular ordering. For new code, it is better to
5492 Constructs webs as commonly used for register allocation purposes and assign
5493 each web individual pseudo register. This allows the register allocation pass
5494 to operate on pseudos directly, but also strengthens several other optimization
5495 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
5496 however, make debugging impossible, since variables will no longer stay in a
5499 Enabled by default with @option{-funroll-loops}.
5501 @item -fwhole-program
5502 @opindex fwhole-program
5503 Assume that the current compilation unit represents whole program being
5504 compiled. All public functions and variables with the exception of @code{main}
5505 and those merged by attribute @code{externally_visible} become static functions
5506 and in a affect gets more aggressively optimized by interprocedural optimizers.
5507 While this option is equivalent to proper use of @code{static} keyword for
5508 programs consisting of single file, in combination with option
5509 @option{--combine} this flag can be used to compile most of smaller scale C
5510 programs since the functions and variables become local for the whole combined
5511 compilation unit, not for the single source file itself.
5514 @item -fno-cprop-registers
5515 @opindex fno-cprop-registers
5516 After register allocation and post-register allocation instruction splitting,
5517 we perform a copy-propagation pass to try to reduce scheduling dependencies
5518 and occasionally eliminate the copy.
5520 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5522 @item -fprofile-generate
5523 @opindex fprofile-generate
5525 Enable options usually used for instrumenting application to produce
5526 profile useful for later recompilation with profile feedback based
5527 optimization. You must use @option{-fprofile-generate} both when
5528 compiling and when linking your program.
5530 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
5533 @opindex fprofile-use
5534 Enable profile feedback directed optimizations, and optimizations
5535 generally profitable only with profile feedback available.
5537 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
5538 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
5542 The following options control compiler behavior regarding floating
5543 point arithmetic. These options trade off between speed and
5544 correctness. All must be specifically enabled.
5548 @opindex ffloat-store
5549 Do not store floating point variables in registers, and inhibit other
5550 options that might change whether a floating point value is taken from a
5553 @cindex floating point precision
5554 This option prevents undesirable excess precision on machines such as
5555 the 68000 where the floating registers (of the 68881) keep more
5556 precision than a @code{double} is supposed to have. Similarly for the
5557 x86 architecture. For most programs, the excess precision does only
5558 good, but a few programs rely on the precise definition of IEEE floating
5559 point. Use @option{-ffloat-store} for such programs, after modifying
5560 them to store all pertinent intermediate computations into variables.
5564 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5565 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5566 @option{-fno-rounding-math}, @option{-fno-signaling-nans}
5567 and @option{fcx-limited-range}.
5569 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5571 This option should never be turned on by any @option{-O} option since
5572 it can result in incorrect output for programs which depend on
5573 an exact implementation of IEEE or ISO rules/specifications for
5576 @item -fno-math-errno
5577 @opindex fno-math-errno
5578 Do not set ERRNO after calling math functions that are executed
5579 with a single instruction, e.g., sqrt. A program that relies on
5580 IEEE exceptions for math error handling may want to use this flag
5581 for speed while maintaining IEEE arithmetic compatibility.
5583 This option should never be turned on by any @option{-O} option since
5584 it can result in incorrect output for programs which depend on
5585 an exact implementation of IEEE or ISO rules/specifications for
5588 The default is @option{-fmath-errno}.
5590 On Darwin systems, the math library never sets @code{errno}. There is therefore
5591 no reason for the compiler to consider the possibility that it might,
5592 and @option{-fno-math-errno} is the default.
5594 @item -funsafe-math-optimizations
5595 @opindex funsafe-math-optimizations
5596 Allow optimizations for floating-point arithmetic that (a) assume
5597 that arguments and results are valid and (b) may violate IEEE or
5598 ANSI standards. When used at link-time, it may include libraries
5599 or startup files that change the default FPU control word or other
5600 similar optimizations.
5602 This option should never be turned on by any @option{-O} option since
5603 it can result in incorrect output for programs which depend on
5604 an exact implementation of IEEE or ISO rules/specifications for
5607 The default is @option{-fno-unsafe-math-optimizations}.
5609 @item -ffinite-math-only
5610 @opindex ffinite-math-only
5611 Allow optimizations for floating-point arithmetic that assume
5612 that arguments and results are not NaNs or +-Infs.
5614 This option should never be turned on by any @option{-O} option since
5615 it can result in incorrect output for programs which depend on
5616 an exact implementation of IEEE or ISO rules/specifications.
5618 The default is @option{-fno-finite-math-only}.
5620 @item -fno-trapping-math
5621 @opindex fno-trapping-math
5622 Compile code assuming that floating-point operations cannot generate
5623 user-visible traps. These traps include division by zero, overflow,
5624 underflow, inexact result and invalid operation. This option implies
5625 @option{-fno-signaling-nans}. Setting this option may allow faster
5626 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5628 This option should never be turned on by any @option{-O} option since
5629 it can result in incorrect output for programs which depend on
5630 an exact implementation of IEEE or ISO rules/specifications for
5633 The default is @option{-ftrapping-math}.
5635 @item -frounding-math
5636 @opindex frounding-math
5637 Disable transformations and optimizations that assume default floating
5638 point rounding behavior. This is round-to-zero for all floating point
5639 to integer conversions, and round-to-nearest for all other arithmetic
5640 truncations. This option should be specified for programs that change
5641 the FP rounding mode dynamically, or that may be executed with a
5642 non-default rounding mode. This option disables constant folding of
5643 floating point expressions at compile-time (which may be affected by
5644 rounding mode) and arithmetic transformations that are unsafe in the
5645 presence of sign-dependent rounding modes.
5647 The default is @option{-fno-rounding-math}.
5649 This option is experimental and does not currently guarantee to
5650 disable all GCC optimizations that are affected by rounding mode.
5651 Future versions of GCC may provide finer control of this setting
5652 using C99's @code{FENV_ACCESS} pragma. This command line option
5653 will be used to specify the default state for @code{FENV_ACCESS}.
5655 @item -frtl-abstract-sequences
5656 @opindex frtl-abstract-sequences
5657 It is a size optimization method. This option is to find identical
5658 sequences of code, which can be turned into pseudo-procedures and
5659 then replace all occurrences with calls to the newly created
5660 subroutine. It is kind of an opposite of @option{-finline-functions}.
5661 This optimization runs at RTL level.
5663 @item -fsignaling-nans
5664 @opindex fsignaling-nans
5665 Compile code assuming that IEEE signaling NaNs may generate user-visible
5666 traps during floating-point operations. Setting this option disables
5667 optimizations that may change the number of exceptions visible with
5668 signaling NaNs. This option implies @option{-ftrapping-math}.
5670 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5673 The default is @option{-fno-signaling-nans}.
5675 This option is experimental and does not currently guarantee to
5676 disable all GCC optimizations that affect signaling NaN behavior.
5678 @item -fsingle-precision-constant
5679 @opindex fsingle-precision-constant
5680 Treat floating point constant as single precision constant instead of
5681 implicitly converting it to double precision constant.
5683 @item -fcx-limited-range
5684 @itemx -fno-cx-limited-range
5685 @opindex fcx-limited-range
5686 @opindex fno-cx-limited-range
5687 When enabled, this option states that a range reduction step is not
5688 needed when performing complex division. The default is
5689 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
5691 This option controls the default setting of the ISO C99
5692 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
5697 The following options control optimizations that may improve
5698 performance, but are not enabled by any @option{-O} options. This
5699 section includes experimental options that may produce broken code.
5702 @item -fbranch-probabilities
5703 @opindex fbranch-probabilities
5704 After running a program compiled with @option{-fprofile-arcs}
5705 (@pxref{Debugging Options,, Options for Debugging Your Program or
5706 @command{gcc}}), you can compile it a second time using
5707 @option{-fbranch-probabilities}, to improve optimizations based on
5708 the number of times each branch was taken. When the program
5709 compiled with @option{-fprofile-arcs} exits it saves arc execution
5710 counts to a file called @file{@var{sourcename}.gcda} for each source
5711 file The information in this data file is very dependent on the
5712 structure of the generated code, so you must use the same source code
5713 and the same optimization options for both compilations.
5715 With @option{-fbranch-probabilities}, GCC puts a
5716 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5717 These can be used to improve optimization. Currently, they are only
5718 used in one place: in @file{reorg.c}, instead of guessing which path a
5719 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5720 exactly determine which path is taken more often.
5722 @item -fprofile-values
5723 @opindex fprofile-values
5724 If combined with @option{-fprofile-arcs}, it adds code so that some
5725 data about values of expressions in the program is gathered.
5727 With @option{-fbranch-probabilities}, it reads back the data gathered
5728 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5729 notes to instructions for their later usage in optimizations.
5731 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5735 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5736 a code to gather information about values of expressions.
5738 With @option{-fbranch-probabilities}, it reads back the data gathered
5739 and actually performs the optimizations based on them.
5740 Currently the optimizations include specialization of division operation
5741 using the knowledge about the value of the denominator.
5743 @item -frename-registers
5744 @opindex frename-registers
5745 Attempt to avoid false dependencies in scheduled code by making use
5746 of registers left over after register allocation. This optimization
5747 will most benefit processors with lots of registers. Depending on the
5748 debug information format adopted by the target, however, it can
5749 make debugging impossible, since variables will no longer stay in
5750 a ``home register''.
5752 Enabled by default with @option{-funroll-loops}.
5756 Perform tail duplication to enlarge superblock size. This transformation
5757 simplifies the control flow of the function allowing other optimizations to do
5760 Enabled with @option{-fprofile-use}.
5762 @item -funroll-loops
5763 @opindex funroll-loops
5764 Unroll loops whose number of iterations can be determined at compile time or
5765 upon entry to the loop. @option{-funroll-loops} implies
5766 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
5767 It also turns on complete loop peeling (i.e.@: complete removal of loops with
5768 small constant number of iterations). This option makes code larger, and may
5769 or may not make it run faster.
5771 Enabled with @option{-fprofile-use}.
5773 @item -funroll-all-loops
5774 @opindex funroll-all-loops
5775 Unroll all loops, even if their number of iterations is uncertain when
5776 the loop is entered. This usually makes programs run more slowly.
5777 @option{-funroll-all-loops} implies the same options as
5778 @option{-funroll-loops}.
5781 @opindex fpeel-loops
5782 Peels the loops for that there is enough information that they do not
5783 roll much (from profile feedback). It also turns on complete loop peeling
5784 (i.e.@: complete removal of loops with small constant number of iterations).
5786 Enabled with @option{-fprofile-use}.
5788 @item -fmove-loop-invariants
5789 @opindex fmove-loop-invariants
5790 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
5791 at level @option{-O1}
5793 @item -funswitch-loops
5794 @opindex funswitch-loops
5795 Move branches with loop invariant conditions out of the loop, with duplicates
5796 of the loop on both branches (modified according to result of the condition).
5798 @item -ffunction-sections
5799 @itemx -fdata-sections
5800 @opindex ffunction-sections
5801 @opindex fdata-sections
5802 Place each function or data item into its own section in the output
5803 file if the target supports arbitrary sections. The name of the
5804 function or the name of the data item determines the section's name
5807 Use these options on systems where the linker can perform optimizations
5808 to improve locality of reference in the instruction space. Most systems
5809 using the ELF object format and SPARC processors running Solaris 2 have
5810 linkers with such optimizations. AIX may have these optimizations in
5813 Only use these options when there are significant benefits from doing
5814 so. When you specify these options, the assembler and linker will
5815 create larger object and executable files and will also be slower.
5816 You will not be able to use @code{gprof} on all systems if you
5817 specify this option and you may have problems with debugging if
5818 you specify both this option and @option{-g}.
5820 @item -fbranch-target-load-optimize
5821 @opindex fbranch-target-load-optimize
5822 Perform branch target register load optimization before prologue / epilogue
5824 The use of target registers can typically be exposed only during reload,
5825 thus hoisting loads out of loops and doing inter-block scheduling needs
5826 a separate optimization pass.
5828 @item -fbranch-target-load-optimize2
5829 @opindex fbranch-target-load-optimize2
5830 Perform branch target register load optimization after prologue / epilogue
5833 @item -fbtr-bb-exclusive
5834 @opindex fbtr-bb-exclusive
5835 When performing branch target register load optimization, don't reuse
5836 branch target registers in within any basic block.
5838 @item -fstack-protector
5839 Emit extra code to check for buffer overflows, such as stack smashing
5840 attacks. This is done by adding a guard variable to functions with
5841 vulnerable objects. This includes functions that call alloca, and
5842 functions with buffers larger than 8 bytes. The guards are initialized
5843 when a function is entered and then checked when the function exits.
5844 If a guard check fails, an error message is printed and the program exits.
5846 @item -fstack-protector-all
5847 Like @option{-fstack-protector} except that all functions are protected.
5849 @item -fsection-anchors
5850 @opindex fsection-anchors
5851 Try to reduce the number of symbolic address calculations by using
5852 shared ``anchor'' symbols to address nearby objects. This transformation
5853 can help to reduce the number of GOT entries and GOT accesses on some
5856 For example, the implementation of the following function @code{foo}:
5860 int foo (void) @{ return a + b + c; @}
5863 would usually calculate the addresses of all three variables, but if you
5864 compile it with @option{-fsection-anchors}, it will access the variables
5865 from a common anchor point instead. The effect is similar to the
5866 following pseudocode (which isn't valid C):
5871 register int *xr = &x;
5872 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5876 Not all targets support this option.
5878 @item --param @var{name}=@var{value}
5880 In some places, GCC uses various constants to control the amount of
5881 optimization that is done. For example, GCC will not inline functions
5882 that contain more that a certain number of instructions. You can
5883 control some of these constants on the command-line using the
5884 @option{--param} option.
5886 The names of specific parameters, and the meaning of the values, are
5887 tied to the internals of the compiler, and are subject to change
5888 without notice in future releases.
5890 In each case, the @var{value} is an integer. The allowable choices for
5891 @var{name} are given in the following table:
5894 @item salias-max-implicit-fields
5895 The maximum number of fields in a variable without direct
5896 structure accesses for which structure aliasing will consider trying
5897 to track each field. The default is 5
5899 @item salias-max-array-elements
5900 The maximum number of elements an array can have and its elements
5901 still be tracked individually by structure aliasing. The default is 4
5903 @item sra-max-structure-size
5904 The maximum structure size, in bytes, at which the scalar replacement
5905 of aggregates (SRA) optimization will perform block copies. The
5906 default value, 0, implies that GCC will select the most appropriate
5909 @item sra-field-structure-ratio
5910 The threshold ratio (as a percentage) between instantiated fields and
5911 the complete structure size. We say that if the ratio of the number
5912 of bytes in instantiated fields to the number of bytes in the complete
5913 structure exceeds this parameter, then block copies are not used. The
5916 @item max-crossjump-edges
5917 The maximum number of incoming edges to consider for crossjumping.
5918 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5919 the number of edges incoming to each block. Increasing values mean
5920 more aggressive optimization, making the compile time increase with
5921 probably small improvement in executable size.
5923 @item min-crossjump-insns
5924 The minimum number of instructions which must be matched at the end
5925 of two blocks before crossjumping will be performed on them. This
5926 value is ignored in the case where all instructions in the block being
5927 crossjumped from are matched. The default value is 5.
5929 @item max-grow-copy-bb-insns
5930 The maximum code size expansion factor when copying basic blocks
5931 instead of jumping. The expansion is relative to a jump instruction.
5932 The default value is 8.
5934 @item max-goto-duplication-insns
5935 The maximum number of instructions to duplicate to a block that jumps
5936 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
5937 passes, GCC factors computed gotos early in the compilation process,
5938 and unfactors them as late as possible. Only computed jumps at the
5939 end of a basic blocks with no more than max-goto-duplication-insns are
5940 unfactored. The default value is 8.
5942 @item max-delay-slot-insn-search
5943 The maximum number of instructions to consider when looking for an
5944 instruction to fill a delay slot. If more than this arbitrary number of
5945 instructions is searched, the time savings from filling the delay slot
5946 will be minimal so stop searching. Increasing values mean more
5947 aggressive optimization, making the compile time increase with probably
5948 small improvement in executable run time.
5950 @item max-delay-slot-live-search
5951 When trying to fill delay slots, the maximum number of instructions to
5952 consider when searching for a block with valid live register
5953 information. Increasing this arbitrarily chosen value means more
5954 aggressive optimization, increasing the compile time. This parameter
5955 should be removed when the delay slot code is rewritten to maintain the
5958 @item max-gcse-memory
5959 The approximate maximum amount of memory that will be allocated in
5960 order to perform the global common subexpression elimination
5961 optimization. If more memory than specified is required, the
5962 optimization will not be done.
5964 @item max-gcse-passes
5965 The maximum number of passes of GCSE to run. The default is 1.
5967 @item max-pending-list-length
5968 The maximum number of pending dependencies scheduling will allow
5969 before flushing the current state and starting over. Large functions
5970 with few branches or calls can create excessively large lists which
5971 needlessly consume memory and resources.
5973 @item max-inline-insns-single
5974 Several parameters control the tree inliner used in gcc.
5975 This number sets the maximum number of instructions (counted in GCC's
5976 internal representation) in a single function that the tree inliner
5977 will consider for inlining. This only affects functions declared
5978 inline and methods implemented in a class declaration (C++).
5979 The default value is 450.
5981 @item max-inline-insns-auto
5982 When you use @option{-finline-functions} (included in @option{-O3}),
5983 a lot of functions that would otherwise not be considered for inlining
5984 by the compiler will be investigated. To those functions, a different
5985 (more restrictive) limit compared to functions declared inline can
5987 The default value is 90.
5989 @item large-function-insns
5990 The limit specifying really large functions. For functions larger than this
5991 limit after inlining inlining is constrained by
5992 @option{--param large-function-growth}. This parameter is useful primarily
5993 to avoid extreme compilation time caused by non-linear algorithms used by the
5995 This parameter is ignored when @option{-funit-at-a-time} is not used.
5996 The default value is 2700.
5998 @item large-function-growth
5999 Specifies maximal growth of large function caused by inlining in percents.
6000 This parameter is ignored when @option{-funit-at-a-time} is not used.
6001 The default value is 100 which limits large function growth to 2.0 times
6004 @item large-unit-insns
6005 The limit specifying large translation unit. Growth caused by inlining of
6006 units larger than this limit is limited by @option{--param inline-unit-growth}.
6007 For small units this might be too tight (consider unit consisting of function A
6008 that is inline and B that just calls A three time. If B is small relative to
6009 A, the growth of unit is 300\% and yet such inlining is very sane. For very
6010 large units consisting of small inlininable functions however the overall unit
6011 growth limit is needed to avoid exponential explosion of code size. Thus for
6012 smaller units, the size is increased to @option{--param large-unit-insns}
6013 before applying @option{--param inline-unit-growth}. The default is 10000
6015 @item inline-unit-growth
6016 Specifies maximal overall growth of the compilation unit caused by inlining.
6017 This parameter is ignored when @option{-funit-at-a-time} is not used.
6018 The default value is 50 which limits unit growth to 1.5 times the original
6021 @item max-inline-insns-recursive
6022 @itemx max-inline-insns-recursive-auto
6023 Specifies maximum number of instructions out-of-line copy of self recursive inline
6024 function can grow into by performing recursive inlining.
6026 For functions declared inline @option{--param max-inline-insns-recursive} is
6027 taken into account. For function not declared inline, recursive inlining
6028 happens only when @option{-finline-functions} (included in @option{-O3}) is
6029 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
6030 default value is 450.
6032 @item max-inline-recursive-depth
6033 @itemx max-inline-recursive-depth-auto
6034 Specifies maximum recursion depth used by the recursive inlining.
6036 For functions declared inline @option{--param max-inline-recursive-depth} is
6037 taken into account. For function not declared inline, recursive inlining
6038 happens only when @option{-finline-functions} (included in @option{-O3}) is
6039 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6040 default value is 450.
6042 @item min-inline-recursive-probability
6043 Recursive inlining is profitable only for function having deep recursion
6044 in average and can hurt for function having little recursion depth by
6045 increasing the prologue size or complexity of function body to other
6048 When profile feedback is available (see @option{-fprofile-generate}) the actual
6049 recursion depth can be guessed from probability that function will recurse via
6050 given call expression. This parameter limits inlining only to call expression
6051 whose probability exceeds given threshold (in percents). The default value is
6054 @item inline-call-cost
6055 Specify cost of call instruction relative to simple arithmetics operations
6056 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6057 functions and at the same time increases size of leaf function that is believed to
6058 reduce function size by being inlined. In effect it increases amount of
6059 inlining for code having large abstraction penalty (many functions that just
6060 pass the arguments to other functions) and decrease inlining for code with low
6061 abstraction penalty. The default value is 16.
6063 @item max-unrolled-insns
6064 The maximum number of instructions that a loop should have if that loop
6065 is unrolled, and if the loop is unrolled, it determines how many times
6066 the loop code is unrolled.
6068 @item max-average-unrolled-insns
6069 The maximum number of instructions biased by probabilities of their execution
6070 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6071 it determines how many times the loop code is unrolled.
6073 @item max-unroll-times
6074 The maximum number of unrollings of a single loop.
6076 @item max-peeled-insns
6077 The maximum number of instructions that a loop should have if that loop
6078 is peeled, and if the loop is peeled, it determines how many times
6079 the loop code is peeled.
6081 @item max-peel-times
6082 The maximum number of peelings of a single loop.
6084 @item max-completely-peeled-insns
6085 The maximum number of insns of a completely peeled loop.
6087 @item max-completely-peel-times
6088 The maximum number of iterations of a loop to be suitable for complete peeling.
6090 @item max-unswitch-insns
6091 The maximum number of insns of an unswitched loop.
6093 @item max-unswitch-level
6094 The maximum number of branches unswitched in a single loop.
6097 The minimum cost of an expensive expression in the loop invariant motion.
6099 @item iv-consider-all-candidates-bound
6100 Bound on number of candidates for induction variables below that
6101 all candidates are considered for each use in induction variable
6102 optimizations. Only the most relevant candidates are considered
6103 if there are more candidates, to avoid quadratic time complexity.
6105 @item iv-max-considered-uses
6106 The induction variable optimizations give up on loops that contain more
6107 induction variable uses.
6109 @item iv-always-prune-cand-set-bound
6110 If number of candidates in the set is smaller than this value,
6111 we always try to remove unnecessary ivs from the set during its
6112 optimization when a new iv is added to the set.
6114 @item scev-max-expr-size
6115 Bound on size of expressions used in the scalar evolutions analyzer.
6116 Large expressions slow the analyzer.
6118 @item vect-max-version-checks
6119 The maximum number of runtime checks that can be performed when doing
6120 loop versioning in the vectorizer. See option ftree-vect-loop-version
6121 for more information.
6123 @item max-iterations-to-track
6125 The maximum number of iterations of a loop the brute force algorithm
6126 for analysis of # of iterations of the loop tries to evaluate.
6128 @item hot-bb-count-fraction
6129 Select fraction of the maximal count of repetitions of basic block in program
6130 given basic block needs to have to be considered hot.
6132 @item hot-bb-frequency-fraction
6133 Select fraction of the maximal frequency of executions of basic block in
6134 function given basic block needs to have to be considered hot
6136 @item max-predicted-iterations
6137 The maximum number of loop iterations we predict statically. This is useful
6138 in cases where function contain single loop with known bound and other loop
6139 with unknown. We predict the known number of iterations correctly, while
6140 the unknown number of iterations average to roughly 10. This means that the
6141 loop without bounds would appear artificially cold relative to the other one.
6143 @item tracer-dynamic-coverage
6144 @itemx tracer-dynamic-coverage-feedback
6146 This value is used to limit superblock formation once the given percentage of
6147 executed instructions is covered. This limits unnecessary code size
6150 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6151 feedback is available. The real profiles (as opposed to statically estimated
6152 ones) are much less balanced allowing the threshold to be larger value.
6154 @item tracer-max-code-growth
6155 Stop tail duplication once code growth has reached given percentage. This is
6156 rather hokey argument, as most of the duplicates will be eliminated later in
6157 cross jumping, so it may be set to much higher values than is the desired code
6160 @item tracer-min-branch-ratio
6162 Stop reverse growth when the reverse probability of best edge is less than this
6163 threshold (in percent).
6165 @item tracer-min-branch-ratio
6166 @itemx tracer-min-branch-ratio-feedback
6168 Stop forward growth if the best edge do have probability lower than this
6171 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6172 compilation for profile feedback and one for compilation without. The value
6173 for compilation with profile feedback needs to be more conservative (higher) in
6174 order to make tracer effective.
6176 @item max-cse-path-length
6178 Maximum number of basic blocks on path that cse considers. The default is 10.
6181 The maximum instructions CSE process before flushing. The default is 1000.
6183 @item global-var-threshold
6185 Counts the number of function calls (@var{n}) and the number of
6186 call-clobbered variables (@var{v}). If @var{n}x@var{v} is larger than this limit, a
6187 single artificial variable will be created to represent all the
6188 call-clobbered variables at function call sites. This artificial
6189 variable will then be made to alias every call-clobbered variable.
6190 (done as @code{int * size_t} on the host machine; beware overflow).
6192 @item max-aliased-vops
6194 Maximum number of virtual operands allowed to represent aliases
6195 before triggering the alias grouping heuristic. Alias grouping
6196 reduces compile times and memory consumption needed for aliasing at
6197 the expense of precision loss in alias information.
6199 @item ggc-min-expand
6201 GCC uses a garbage collector to manage its own memory allocation. This
6202 parameter specifies the minimum percentage by which the garbage
6203 collector's heap should be allowed to expand between collections.
6204 Tuning this may improve compilation speed; it has no effect on code
6207 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6208 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6209 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6210 GCC is not able to calculate RAM on a particular platform, the lower
6211 bound of 30% is used. Setting this parameter and
6212 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6213 every opportunity. This is extremely slow, but can be useful for
6216 @item ggc-min-heapsize
6218 Minimum size of the garbage collector's heap before it begins bothering
6219 to collect garbage. The first collection occurs after the heap expands
6220 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6221 tuning this may improve compilation speed, and has no effect on code
6224 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6225 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6226 with a lower bound of 4096 (four megabytes) and an upper bound of
6227 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6228 particular platform, the lower bound is used. Setting this parameter
6229 very large effectively disables garbage collection. Setting this
6230 parameter and @option{ggc-min-expand} to zero causes a full collection
6231 to occur at every opportunity.
6233 @item max-reload-search-insns
6234 The maximum number of instruction reload should look backward for equivalent
6235 register. Increasing values mean more aggressive optimization, making the
6236 compile time increase with probably slightly better performance. The default
6239 @item max-cselib-memory-locations
6240 The maximum number of memory locations cselib should take into account.
6241 Increasing values mean more aggressive optimization, making the compile time
6242 increase with probably slightly better performance. The default value is 500.
6244 @item max-flow-memory-locations
6245 Similar as @option{max-cselib-memory-locations} but for dataflow liveness.
6246 The default value is 100.
6248 @item reorder-blocks-duplicate
6249 @itemx reorder-blocks-duplicate-feedback
6251 Used by basic block reordering pass to decide whether to use unconditional
6252 branch or duplicate the code on its destination. Code is duplicated when its
6253 estimated size is smaller than this value multiplied by the estimated size of
6254 unconditional jump in the hot spots of the program.
6256 The @option{reorder-block-duplicate-feedback} is used only when profile
6257 feedback is available and may be set to higher values than
6258 @option{reorder-block-duplicate} since information about the hot spots is more
6261 @item max-sched-ready-insns
6262 The maximum number of instructions ready to be issued the scheduler should
6263 consider at any given time during the first scheduling pass. Increasing
6264 values mean more thorough searches, making the compilation time increase
6265 with probably little benefit. The default value is 100.
6267 @item max-sched-region-blocks
6268 The maximum number of blocks in a region to be considered for
6269 interblock scheduling. The default value is 10.
6271 @item max-sched-region-insns
6272 The maximum number of insns in a region to be considered for
6273 interblock scheduling. The default value is 100.
6276 The minimum probability (in percents) of reaching a source block
6277 for interblock speculative scheduling. The default value is 40.
6279 @item max-sched-extend-regions-iters
6280 The maximum number of iterations through CFG to extend regions.
6281 0 - disable region extension,
6282 N - do at most N iterations.
6283 The default value is 0.
6285 @item max-sched-insn-conflict-delay
6286 The maximum conflict delay for an insn to be considered for speculative motion.
6287 The default value is 3.
6289 @item sched-spec-prob-cutoff
6290 The minimal probability of speculation success (in percents), so that
6291 speculative insn will be scheduled.
6292 The default value is 40.
6294 @item max-last-value-rtl
6296 The maximum size measured as number of RTLs that can be recorded in an expression
6297 in combiner for a pseudo register as last known value of that register. The default
6300 @item integer-share-limit
6301 Small integer constants can use a shared data structure, reducing the
6302 compiler's memory usage and increasing its speed. This sets the maximum
6303 value of a shared integer constant's. The default value is 256.
6305 @item min-virtual-mappings
6306 Specifies the minimum number of virtual mappings in the incremental
6307 SSA updater that should be registered to trigger the virtual mappings
6308 heuristic defined by virtual-mappings-ratio. The default value is
6311 @item virtual-mappings-ratio
6312 If the number of virtual mappings is virtual-mappings-ratio bigger
6313 than the number of virtual symbols to be updated, then the incremental
6314 SSA updater switches to a full update for those symbols. The default
6317 @item ssp-buffer-size
6318 The minimum size of buffers (i.e. arrays) that will receive stack smashing
6319 protection when @option{-fstack-protection} is used.
6321 @item max-jump-thread-duplication-stmts
6322 Maximum number of statements allowed in a block that needs to be
6323 duplicated when threading jumps.
6325 @item max-fields-for-field-sensitive
6326 Maximum number of fields in a structure we will treat in
6327 a field sensitive manner during pointer analysis.
6332 @node Preprocessor Options
6333 @section Options Controlling the Preprocessor
6334 @cindex preprocessor options
6335 @cindex options, preprocessor
6337 These options control the C preprocessor, which is run on each C source
6338 file before actual compilation.
6340 If you use the @option{-E} option, nothing is done except preprocessing.
6341 Some of these options make sense only together with @option{-E} because
6342 they cause the preprocessor output to be unsuitable for actual
6347 You can use @option{-Wp,@var{option}} to bypass the compiler driver
6348 and pass @var{option} directly through to the preprocessor. If
6349 @var{option} contains commas, it is split into multiple options at the
6350 commas. However, many options are modified, translated or interpreted
6351 by the compiler driver before being passed to the preprocessor, and
6352 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
6353 interface is undocumented and subject to change, so whenever possible
6354 you should avoid using @option{-Wp} and let the driver handle the
6357 @item -Xpreprocessor @var{option}
6358 @opindex preprocessor
6359 Pass @var{option} as an option to the preprocessor. You can use this to
6360 supply system-specific preprocessor options which GCC does not know how to
6363 If you want to pass an option that takes an argument, you must use
6364 @option{-Xpreprocessor} twice, once for the option and once for the argument.
6367 @include cppopts.texi
6369 @node Assembler Options
6370 @section Passing Options to the Assembler
6372 @c prevent bad page break with this line
6373 You can pass options to the assembler.
6376 @item -Wa,@var{option}
6378 Pass @var{option} as an option to the assembler. If @var{option}
6379 contains commas, it is split into multiple options at the commas.
6381 @item -Xassembler @var{option}
6383 Pass @var{option} as an option to the assembler. You can use this to
6384 supply system-specific assembler options which GCC does not know how to
6387 If you want to pass an option that takes an argument, you must use
6388 @option{-Xassembler} twice, once for the option and once for the argument.
6393 @section Options for Linking
6394 @cindex link options
6395 @cindex options, linking
6397 These options come into play when the compiler links object files into
6398 an executable output file. They are meaningless if the compiler is
6399 not doing a link step.
6403 @item @var{object-file-name}
6404 A file name that does not end in a special recognized suffix is
6405 considered to name an object file or library. (Object files are
6406 distinguished from libraries by the linker according to the file
6407 contents.) If linking is done, these object files are used as input
6416 If any of these options is used, then the linker is not run, and
6417 object file names should not be used as arguments. @xref{Overall
6421 @item -l@var{library}
6422 @itemx -l @var{library}
6424 Search the library named @var{library} when linking. (The second
6425 alternative with the library as a separate argument is only for
6426 POSIX compliance and is not recommended.)
6428 It makes a difference where in the command you write this option; the
6429 linker searches and processes libraries and object files in the order they
6430 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
6431 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
6432 to functions in @samp{z}, those functions may not be loaded.
6434 The linker searches a standard list of directories for the library,
6435 which is actually a file named @file{lib@var{library}.a}. The linker
6436 then uses this file as if it had been specified precisely by name.
6438 The directories searched include several standard system directories
6439 plus any that you specify with @option{-L}.
6441 Normally the files found this way are library files---archive files
6442 whose members are object files. The linker handles an archive file by
6443 scanning through it for members which define symbols that have so far
6444 been referenced but not defined. But if the file that is found is an
6445 ordinary object file, it is linked in the usual fashion. The only
6446 difference between using an @option{-l} option and specifying a file name
6447 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
6448 and searches several directories.
6452 You need this special case of the @option{-l} option in order to
6453 link an Objective-C or Objective-C++ program.
6456 @opindex nostartfiles
6457 Do not use the standard system startup files when linking.
6458 The standard system libraries are used normally, unless @option{-nostdlib}
6459 or @option{-nodefaultlibs} is used.
6461 @item -nodefaultlibs
6462 @opindex nodefaultlibs
6463 Do not use the standard system libraries when linking.
6464 Only the libraries you specify will be passed to the linker.
6465 The standard startup files are used normally, unless @option{-nostartfiles}
6466 is used. The compiler may generate calls to @code{memcmp},
6467 @code{memset}, @code{memcpy} and @code{memmove}.
6468 These entries are usually resolved by entries in
6469 libc. These entry points should be supplied through some other
6470 mechanism when this option is specified.
6474 Do not use the standard system startup files or libraries when linking.
6475 No startup files and only the libraries you specify will be passed to
6476 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
6477 @code{memcpy} and @code{memmove}.
6478 These entries are usually resolved by entries in
6479 libc. These entry points should be supplied through some other
6480 mechanism when this option is specified.
6482 @cindex @option{-lgcc}, use with @option{-nostdlib}
6483 @cindex @option{-nostdlib} and unresolved references
6484 @cindex unresolved references and @option{-nostdlib}
6485 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
6486 @cindex @option{-nodefaultlibs} and unresolved references
6487 @cindex unresolved references and @option{-nodefaultlibs}
6488 One of the standard libraries bypassed by @option{-nostdlib} and
6489 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
6490 that GCC uses to overcome shortcomings of particular machines, or special
6491 needs for some languages.
6492 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
6493 Collection (GCC) Internals},
6494 for more discussion of @file{libgcc.a}.)
6495 In most cases, you need @file{libgcc.a} even when you want to avoid
6496 other standard libraries. In other words, when you specify @option{-nostdlib}
6497 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
6498 This ensures that you have no unresolved references to internal GCC
6499 library subroutines. (For example, @samp{__main}, used to ensure C++
6500 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
6501 GNU Compiler Collection (GCC) Internals}.)
6505 Produce a position independent executable on targets which support it.
6506 For predictable results, you must also specify the same set of options
6507 that were used to generate code (@option{-fpie}, @option{-fPIE},
6508 or model suboptions) when you specify this option.
6512 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
6513 that support it. This instructs the linker to add all symbols, not
6514 only used ones, to the dynamic symbol table. This option is needed
6515 for some uses of @code{dlopen} or to allow obtaining backtraces
6516 from within a program.
6520 Remove all symbol table and relocation information from the executable.
6524 On systems that support dynamic linking, this prevents linking with the shared
6525 libraries. On other systems, this option has no effect.
6529 Produce a shared object which can then be linked with other objects to
6530 form an executable. Not all systems support this option. For predictable
6531 results, you must also specify the same set of options that were used to
6532 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
6533 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
6534 needs to build supplementary stub code for constructors to work. On
6535 multi-libbed systems, @samp{gcc -shared} must select the correct support
6536 libraries to link against. Failing to supply the correct flags may lead
6537 to subtle defects. Supplying them in cases where they are not necessary
6540 @item -shared-libgcc
6541 @itemx -static-libgcc
6542 @opindex shared-libgcc
6543 @opindex static-libgcc
6544 On systems that provide @file{libgcc} as a shared library, these options
6545 force the use of either the shared or static version respectively.
6546 If no shared version of @file{libgcc} was built when the compiler was
6547 configured, these options have no effect.
6549 There are several situations in which an application should use the
6550 shared @file{libgcc} instead of the static version. The most common
6551 of these is when the application wishes to throw and catch exceptions
6552 across different shared libraries. In that case, each of the libraries
6553 as well as the application itself should use the shared @file{libgcc}.
6555 Therefore, the G++ and GCJ drivers automatically add
6556 @option{-shared-libgcc} whenever you build a shared library or a main
6557 executable, because C++ and Java programs typically use exceptions, so
6558 this is the right thing to do.
6560 If, instead, you use the GCC driver to create shared libraries, you may
6561 find that they will not always be linked with the shared @file{libgcc}.
6562 If GCC finds, at its configuration time, that you have a non-GNU linker
6563 or a GNU linker that does not support option @option{--eh-frame-hdr},
6564 it will link the shared version of @file{libgcc} into shared libraries
6565 by default. Otherwise, it will take advantage of the linker and optimize
6566 away the linking with the shared version of @file{libgcc}, linking with
6567 the static version of libgcc by default. This allows exceptions to
6568 propagate through such shared libraries, without incurring relocation
6569 costs at library load time.
6571 However, if a library or main executable is supposed to throw or catch
6572 exceptions, you must link it using the G++ or GCJ driver, as appropriate
6573 for the languages used in the program, or using the option
6574 @option{-shared-libgcc}, such that it is linked with the shared
6579 Bind references to global symbols when building a shared object. Warn
6580 about any unresolved references (unless overridden by the link editor
6581 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
6584 @item -Xlinker @var{option}
6586 Pass @var{option} as an option to the linker. You can use this to
6587 supply system-specific linker options which GCC does not know how to
6590 If you want to pass an option that takes an argument, you must use
6591 @option{-Xlinker} twice, once for the option and once for the argument.
6592 For example, to pass @option{-assert definitions}, you must write
6593 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
6594 @option{-Xlinker "-assert definitions"}, because this passes the entire
6595 string as a single argument, which is not what the linker expects.
6597 @item -Wl,@var{option}
6599 Pass @var{option} as an option to the linker. If @var{option} contains
6600 commas, it is split into multiple options at the commas.
6602 @item -u @var{symbol}
6604 Pretend the symbol @var{symbol} is undefined, to force linking of
6605 library modules to define it. You can use @option{-u} multiple times with
6606 different symbols to force loading of additional library modules.
6609 @node Directory Options
6610 @section Options for Directory Search
6611 @cindex directory options
6612 @cindex options, directory search
6615 These options specify directories to search for header files, for
6616 libraries and for parts of the compiler:
6621 Add the directory @var{dir} to the head of the list of directories to be
6622 searched for header files. This can be used to override a system header
6623 file, substituting your own version, since these directories are
6624 searched before the system header file directories. However, you should
6625 not use this option to add directories that contain vendor-supplied
6626 system header files (use @option{-isystem} for that). If you use more than
6627 one @option{-I} option, the directories are scanned in left-to-right
6628 order; the standard system directories come after.
6630 If a standard system include directory, or a directory specified with
6631 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
6632 option will be ignored. The directory will still be searched but as a
6633 system directory at its normal position in the system include chain.
6634 This is to ensure that GCC's procedure to fix buggy system headers and
6635 the ordering for the include_next directive are not inadvertently changed.
6636 If you really need to change the search order for system directories,
6637 use the @option{-nostdinc} and/or @option{-isystem} options.
6639 @item -iquote@var{dir}
6641 Add the directory @var{dir} to the head of the list of directories to
6642 be searched for header files only for the case of @samp{#include
6643 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
6644 otherwise just like @option{-I}.
6648 Add directory @var{dir} to the list of directories to be searched
6651 @item -B@var{prefix}
6653 This option specifies where to find the executables, libraries,
6654 include files, and data files of the compiler itself.
6656 The compiler driver program runs one or more of the subprograms
6657 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
6658 @var{prefix} as a prefix for each program it tries to run, both with and
6659 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
6661 For each subprogram to be run, the compiler driver first tries the
6662 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
6663 was not specified, the driver tries two standard prefixes, which are
6664 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
6665 those results in a file name that is found, the unmodified program
6666 name is searched for using the directories specified in your
6667 @env{PATH} environment variable.
6669 The compiler will check to see if the path provided by the @option{-B}
6670 refers to a directory, and if necessary it will add a directory
6671 separator character at the end of the path.
6673 @option{-B} prefixes that effectively specify directory names also apply
6674 to libraries in the linker, because the compiler translates these
6675 options into @option{-L} options for the linker. They also apply to
6676 includes files in the preprocessor, because the compiler translates these
6677 options into @option{-isystem} options for the preprocessor. In this case,
6678 the compiler appends @samp{include} to the prefix.
6680 The run-time support file @file{libgcc.a} can also be searched for using
6681 the @option{-B} prefix, if needed. If it is not found there, the two
6682 standard prefixes above are tried, and that is all. The file is left
6683 out of the link if it is not found by those means.
6685 Another way to specify a prefix much like the @option{-B} prefix is to use
6686 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
6689 As a special kludge, if the path provided by @option{-B} is
6690 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
6691 9, then it will be replaced by @file{[dir/]include}. This is to help
6692 with boot-strapping the compiler.
6694 @item -specs=@var{file}
6696 Process @var{file} after the compiler reads in the standard @file{specs}
6697 file, in order to override the defaults that the @file{gcc} driver
6698 program uses when determining what switches to pass to @file{cc1},
6699 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
6700 @option{-specs=@var{file}} can be specified on the command line, and they
6701 are processed in order, from left to right.
6703 @item --sysroot=@var{dir}
6705 Use @var{dir} as the logical root directory for headers and libraries.
6706 For example, if the compiler would normally search for headers in
6707 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
6708 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
6710 If you use both this option and the @option{-isysroot} option, then
6711 the @option{--sysroot} option will apply to libraries, but the
6712 @option{-isysroot} option will apply to header files.
6714 The GNU linker (beginning with version 2.16) has the necessary support
6715 for this option. If your linker does not support this option, the
6716 header file aspect of @option{--sysroot} will still work, but the
6717 library aspect will not.
6721 This option has been deprecated. Please use @option{-iquote} instead for
6722 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
6723 Any directories you specify with @option{-I} options before the @option{-I-}
6724 option are searched only for the case of @samp{#include "@var{file}"};
6725 they are not searched for @samp{#include <@var{file}>}.
6727 If additional directories are specified with @option{-I} options after
6728 the @option{-I-}, these directories are searched for all @samp{#include}
6729 directives. (Ordinarily @emph{all} @option{-I} directories are used
6732 In addition, the @option{-I-} option inhibits the use of the current
6733 directory (where the current input file came from) as the first search
6734 directory for @samp{#include "@var{file}"}. There is no way to
6735 override this effect of @option{-I-}. With @option{-I.} you can specify
6736 searching the directory which was current when the compiler was
6737 invoked. That is not exactly the same as what the preprocessor does
6738 by default, but it is often satisfactory.
6740 @option{-I-} does not inhibit the use of the standard system directories
6741 for header files. Thus, @option{-I-} and @option{-nostdinc} are
6748 @section Specifying subprocesses and the switches to pass to them
6751 @command{gcc} is a driver program. It performs its job by invoking a
6752 sequence of other programs to do the work of compiling, assembling and
6753 linking. GCC interprets its command-line parameters and uses these to
6754 deduce which programs it should invoke, and which command-line options
6755 it ought to place on their command lines. This behavior is controlled
6756 by @dfn{spec strings}. In most cases there is one spec string for each
6757 program that GCC can invoke, but a few programs have multiple spec
6758 strings to control their behavior. The spec strings built into GCC can
6759 be overridden by using the @option{-specs=} command-line switch to specify
6762 @dfn{Spec files} are plaintext files that are used to construct spec
6763 strings. They consist of a sequence of directives separated by blank
6764 lines. The type of directive is determined by the first non-whitespace
6765 character on the line and it can be one of the following:
6768 @item %@var{command}
6769 Issues a @var{command} to the spec file processor. The commands that can
6773 @item %include <@var{file}>
6775 Search for @var{file} and insert its text at the current point in the
6778 @item %include_noerr <@var{file}>
6779 @cindex %include_noerr
6780 Just like @samp{%include}, but do not generate an error message if the include
6781 file cannot be found.
6783 @item %rename @var{old_name} @var{new_name}
6785 Rename the spec string @var{old_name} to @var{new_name}.
6789 @item *[@var{spec_name}]:
6790 This tells the compiler to create, override or delete the named spec
6791 string. All lines after this directive up to the next directive or
6792 blank line are considered to be the text for the spec string. If this
6793 results in an empty string then the spec will be deleted. (Or, if the
6794 spec did not exist, then nothing will happened.) Otherwise, if the spec
6795 does not currently exist a new spec will be created. If the spec does
6796 exist then its contents will be overridden by the text of this
6797 directive, unless the first character of that text is the @samp{+}
6798 character, in which case the text will be appended to the spec.
6800 @item [@var{suffix}]:
6801 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6802 and up to the next directive or blank line are considered to make up the
6803 spec string for the indicated suffix. When the compiler encounters an
6804 input file with the named suffix, it will processes the spec string in
6805 order to work out how to compile that file. For example:
6812 This says that any input file whose name ends in @samp{.ZZ} should be
6813 passed to the program @samp{z-compile}, which should be invoked with the
6814 command-line switch @option{-input} and with the result of performing the
6815 @samp{%i} substitution. (See below.)
6817 As an alternative to providing a spec string, the text that follows a
6818 suffix directive can be one of the following:
6821 @item @@@var{language}
6822 This says that the suffix is an alias for a known @var{language}. This is
6823 similar to using the @option{-x} command-line switch to GCC to specify a
6824 language explicitly. For example:
6831 Says that .ZZ files are, in fact, C++ source files.
6834 This causes an error messages saying:
6837 @var{name} compiler not installed on this system.
6841 GCC already has an extensive list of suffixes built into it.
6842 This directive will add an entry to the end of the list of suffixes, but
6843 since the list is searched from the end backwards, it is effectively
6844 possible to override earlier entries using this technique.
6848 GCC has the following spec strings built into it. Spec files can
6849 override these strings or create their own. Note that individual
6850 targets can also add their own spec strings to this list.
6853 asm Options to pass to the assembler
6854 asm_final Options to pass to the assembler post-processor
6855 cpp Options to pass to the C preprocessor
6856 cc1 Options to pass to the C compiler
6857 cc1plus Options to pass to the C++ compiler
6858 endfile Object files to include at the end of the link
6859 link Options to pass to the linker
6860 lib Libraries to include on the command line to the linker
6861 libgcc Decides which GCC support library to pass to the linker
6862 linker Sets the name of the linker
6863 predefines Defines to be passed to the C preprocessor
6864 signed_char Defines to pass to CPP to say whether @code{char} is signed
6866 startfile Object files to include at the start of the link
6869 Here is a small example of a spec file:
6875 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6878 This example renames the spec called @samp{lib} to @samp{old_lib} and
6879 then overrides the previous definition of @samp{lib} with a new one.
6880 The new definition adds in some extra command-line options before
6881 including the text of the old definition.
6883 @dfn{Spec strings} are a list of command-line options to be passed to their
6884 corresponding program. In addition, the spec strings can contain
6885 @samp{%}-prefixed sequences to substitute variable text or to
6886 conditionally insert text into the command line. Using these constructs
6887 it is possible to generate quite complex command lines.
6889 Here is a table of all defined @samp{%}-sequences for spec
6890 strings. Note that spaces are not generated automatically around the
6891 results of expanding these sequences. Therefore you can concatenate them
6892 together or combine them with constant text in a single argument.
6896 Substitute one @samp{%} into the program name or argument.
6899 Substitute the name of the input file being processed.
6902 Substitute the basename of the input file being processed.
6903 This is the substring up to (and not including) the last period
6904 and not including the directory.
6907 This is the same as @samp{%b}, but include the file suffix (text after
6911 Marks the argument containing or following the @samp{%d} as a
6912 temporary file name, so that that file will be deleted if GCC exits
6913 successfully. Unlike @samp{%g}, this contributes no text to the
6916 @item %g@var{suffix}
6917 Substitute a file name that has suffix @var{suffix} and is chosen
6918 once per compilation, and mark the argument in the same way as
6919 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
6920 name is now chosen in a way that is hard to predict even when previously
6921 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
6922 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
6923 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
6924 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
6925 was simply substituted with a file name chosen once per compilation,
6926 without regard to any appended suffix (which was therefore treated
6927 just like ordinary text), making such attacks more likely to succeed.
6929 @item %u@var{suffix}
6930 Like @samp{%g}, but generates a new temporary file name even if
6931 @samp{%u@var{suffix}} was already seen.
6933 @item %U@var{suffix}
6934 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6935 new one if there is no such last file name. In the absence of any
6936 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6937 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6938 would involve the generation of two distinct file names, one
6939 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6940 simply substituted with a file name chosen for the previous @samp{%u},
6941 without regard to any appended suffix.
6943 @item %j@var{suffix}
6944 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6945 writable, and if save-temps is off; otherwise, substitute the name
6946 of a temporary file, just like @samp{%u}. This temporary file is not
6947 meant for communication between processes, but rather as a junk
6950 @item %|@var{suffix}
6951 @itemx %m@var{suffix}
6952 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6953 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6954 all. These are the two most common ways to instruct a program that it
6955 should read from standard input or write to standard output. If you
6956 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6957 construct: see for example @file{f/lang-specs.h}.
6959 @item %.@var{SUFFIX}
6960 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6961 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6962 terminated by the next space or %.
6965 Marks the argument containing or following the @samp{%w} as the
6966 designated output file of this compilation. This puts the argument
6967 into the sequence of arguments that @samp{%o} will substitute later.
6970 Substitutes the names of all the output files, with spaces
6971 automatically placed around them. You should write spaces
6972 around the @samp{%o} as well or the results are undefined.
6973 @samp{%o} is for use in the specs for running the linker.
6974 Input files whose names have no recognized suffix are not compiled
6975 at all, but they are included among the output files, so they will
6979 Substitutes the suffix for object files. Note that this is
6980 handled specially when it immediately follows @samp{%g, %u, or %U},
6981 because of the need for those to form complete file names. The
6982 handling is such that @samp{%O} is treated exactly as if it had already
6983 been substituted, except that @samp{%g, %u, and %U} do not currently
6984 support additional @var{suffix} characters following @samp{%O} as they would
6985 following, for example, @samp{.o}.
6988 Substitutes the standard macro predefinitions for the
6989 current target machine. Use this when running @code{cpp}.
6992 Like @samp{%p}, but puts @samp{__} before and after the name of each
6993 predefined macro, except for macros that start with @samp{__} or with
6994 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
6998 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
6999 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
7000 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
7001 and @option{-imultilib} as necessary.
7004 Current argument is the name of a library or startup file of some sort.
7005 Search for that file in a standard list of directories and substitute
7006 the full name found.
7009 Print @var{str} as an error message. @var{str} is terminated by a newline.
7010 Use this when inconsistent options are detected.
7013 Substitute the contents of spec string @var{name} at this point.
7016 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
7018 @item %x@{@var{option}@}
7019 Accumulate an option for @samp{%X}.
7022 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
7026 Output the accumulated assembler options specified by @option{-Wa}.
7029 Output the accumulated preprocessor options specified by @option{-Wp}.
7032 Process the @code{asm} spec. This is used to compute the
7033 switches to be passed to the assembler.
7036 Process the @code{asm_final} spec. This is a spec string for
7037 passing switches to an assembler post-processor, if such a program is
7041 Process the @code{link} spec. This is the spec for computing the
7042 command line passed to the linker. Typically it will make use of the
7043 @samp{%L %G %S %D and %E} sequences.
7046 Dump out a @option{-L} option for each directory that GCC believes might
7047 contain startup files. If the target supports multilibs then the
7048 current multilib directory will be prepended to each of these paths.
7051 Process the @code{lib} spec. This is a spec string for deciding which
7052 libraries should be included on the command line to the linker.
7055 Process the @code{libgcc} spec. This is a spec string for deciding
7056 which GCC support library should be included on the command line to the linker.
7059 Process the @code{startfile} spec. This is a spec for deciding which
7060 object files should be the first ones passed to the linker. Typically
7061 this might be a file named @file{crt0.o}.
7064 Process the @code{endfile} spec. This is a spec string that specifies
7065 the last object files that will be passed to the linker.
7068 Process the @code{cpp} spec. This is used to construct the arguments
7069 to be passed to the C preprocessor.
7072 Process the @code{cc1} spec. This is used to construct the options to be
7073 passed to the actual C compiler (@samp{cc1}).
7076 Process the @code{cc1plus} spec. This is used to construct the options to be
7077 passed to the actual C++ compiler (@samp{cc1plus}).
7080 Substitute the variable part of a matched option. See below.
7081 Note that each comma in the substituted string is replaced by
7085 Remove all occurrences of @code{-S} from the command line. Note---this
7086 command is position dependent. @samp{%} commands in the spec string
7087 before this one will see @code{-S}, @samp{%} commands in the spec string
7088 after this one will not.
7090 @item %:@var{function}(@var{args})
7091 Call the named function @var{function}, passing it @var{args}.
7092 @var{args} is first processed as a nested spec string, then split
7093 into an argument vector in the usual fashion. The function returns
7094 a string which is processed as if it had appeared literally as part
7095 of the current spec.
7097 The following built-in spec functions are provided:
7100 @item @code{if-exists}
7101 The @code{if-exists} spec function takes one argument, an absolute
7102 pathname to a file. If the file exists, @code{if-exists} returns the
7103 pathname. Here is a small example of its usage:
7107 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7110 @item @code{if-exists-else}
7111 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7112 spec function, except that it takes two arguments. The first argument is
7113 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7114 returns the pathname. If it does not exist, it returns the second argument.
7115 This way, @code{if-exists-else} can be used to select one file or another,
7116 based on the existence of the first. Here is a small example of its usage:
7120 crt0%O%s %:if-exists(crti%O%s) \
7121 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
7124 @item @code{replace-outfile}
7125 The @code{replace-outfile} spec function takes two arguments. It looks for the
7126 first argument in the outfiles array and replaces it with the second argument. Here
7127 is a small example of its usage:
7130 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7136 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7137 If that switch was not specified, this substitutes nothing. Note that
7138 the leading dash is omitted when specifying this option, and it is
7139 automatically inserted if the substitution is performed. Thus the spec
7140 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7141 and would output the command line option @option{-foo}.
7143 @item %W@{@code{S}@}
7144 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7147 @item %@{@code{S}*@}
7148 Substitutes all the switches specified to GCC whose names start
7149 with @code{-S}, but which also take an argument. This is used for
7150 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7151 GCC considers @option{-o foo} as being
7152 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7153 text, including the space. Thus two arguments would be generated.
7155 @item %@{@code{S}*&@code{T}*@}
7156 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7157 (the order of @code{S} and @code{T} in the spec is not significant).
7158 There can be any number of ampersand-separated variables; for each the
7159 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
7161 @item %@{@code{S}:@code{X}@}
7162 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
7164 @item %@{!@code{S}:@code{X}@}
7165 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
7167 @item %@{@code{S}*:@code{X}@}
7168 Substitutes @code{X} if one or more switches whose names start with
7169 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
7170 once, no matter how many such switches appeared. However, if @code{%*}
7171 appears somewhere in @code{X}, then @code{X} will be substituted once
7172 for each matching switch, with the @code{%*} replaced by the part of
7173 that switch that matched the @code{*}.
7175 @item %@{.@code{S}:@code{X}@}
7176 Substitutes @code{X}, if processing a file with suffix @code{S}.
7178 @item %@{!.@code{S}:@code{X}@}
7179 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
7181 @item %@{@code{S}|@code{P}:@code{X}@}
7182 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
7183 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
7184 although they have a stronger binding than the @samp{|}. If @code{%*}
7185 appears in @code{X}, all of the alternatives must be starred, and only
7186 the first matching alternative is substituted.
7188 For example, a spec string like this:
7191 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
7194 will output the following command-line options from the following input
7195 command-line options:
7200 -d fred.c -foo -baz -boggle
7201 -d jim.d -bar -baz -boggle
7204 @item %@{S:X; T:Y; :D@}
7206 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
7207 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
7208 be as many clauses as you need. This may be combined with @code{.},
7209 @code{!}, @code{|}, and @code{*} as needed.
7214 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
7215 construct may contain other nested @samp{%} constructs or spaces, or
7216 even newlines. They are processed as usual, as described above.
7217 Trailing white space in @code{X} is ignored. White space may also
7218 appear anywhere on the left side of the colon in these constructs,
7219 except between @code{.} or @code{*} and the corresponding word.
7221 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
7222 handled specifically in these constructs. If another value of
7223 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7224 @option{-W} switch is found later in the command line, the earlier
7225 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7226 just one letter, which passes all matching options.
7228 The character @samp{|} at the beginning of the predicate text is used to
7229 indicate that a command should be piped to the following command, but
7230 only if @option{-pipe} is specified.
7232 It is built into GCC which switches take arguments and which do not.
7233 (You might think it would be useful to generalize this to allow each
7234 compiler's spec to say which switches take arguments. But this cannot
7235 be done in a consistent fashion. GCC cannot even decide which input
7236 files have been specified without knowing which switches take arguments,
7237 and it must know which input files to compile in order to tell which
7240 GCC also knows implicitly that arguments starting in @option{-l} are to be
7241 treated as compiler output files, and passed to the linker in their
7242 proper position among the other output files.
7244 @c man begin OPTIONS
7246 @node Target Options
7247 @section Specifying Target Machine and Compiler Version
7248 @cindex target options
7249 @cindex cross compiling
7250 @cindex specifying machine version
7251 @cindex specifying compiler version and target machine
7252 @cindex compiler version, specifying
7253 @cindex target machine, specifying
7255 The usual way to run GCC is to run the executable called @file{gcc}, or
7256 @file{<machine>-gcc} when cross-compiling, or
7257 @file{<machine>-gcc-<version>} to run a version other than the one that
7258 was installed last. Sometimes this is inconvenient, so GCC provides
7259 options that will switch to another cross-compiler or version.
7262 @item -b @var{machine}
7264 The argument @var{machine} specifies the target machine for compilation.
7266 The value to use for @var{machine} is the same as was specified as the
7267 machine type when configuring GCC as a cross-compiler. For
7268 example, if a cross-compiler was configured with @samp{configure
7269 arm-elf}, meaning to compile for an arm processor with elf binaries,
7270 then you would specify @option{-b arm-elf} to run that cross compiler.
7271 Because there are other options beginning with @option{-b}, the
7272 configuration must contain a hyphen.
7274 @item -V @var{version}
7276 The argument @var{version} specifies which version of GCC to run.
7277 This is useful when multiple versions are installed. For example,
7278 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
7281 The @option{-V} and @option{-b} options work by running the
7282 @file{<machine>-gcc-<version>} executable, so there's no real reason to
7283 use them if you can just run that directly.
7285 @node Submodel Options
7286 @section Hardware Models and Configurations
7287 @cindex submodel options
7288 @cindex specifying hardware config
7289 @cindex hardware models and configurations, specifying
7290 @cindex machine dependent options
7292 Earlier we discussed the standard option @option{-b} which chooses among
7293 different installed compilers for completely different target
7294 machines, such as VAX vs.@: 68000 vs.@: 80386.
7296 In addition, each of these target machine types can have its own
7297 special options, starting with @samp{-m}, to choose among various
7298 hardware models or configurations---for example, 68010 vs 68020,
7299 floating coprocessor or none. A single installed version of the
7300 compiler can compile for any model or configuration, according to the
7303 Some configurations of the compiler also support additional special
7304 options, usually for compatibility with other compilers on the same
7307 @c This list is ordered alphanumerically by subsection name.
7308 @c It should be the same order and spelling as these options are listed
7309 @c in Machine Dependent Options
7315 * Blackfin Options::
7319 * DEC Alpha Options::
7320 * DEC Alpha/VMS Options::
7322 * GNU/Linux Options::
7325 * i386 and x86-64 Options::
7338 * RS/6000 and PowerPC Options::
7339 * S/390 and zSeries Options::
7343 * System V Options::
7344 * TMS320C3x/C4x Options::
7348 * Xstormy16 Options::
7354 @subsection ARC Options
7357 These options are defined for ARC implementations:
7362 Compile code for little endian mode. This is the default.
7366 Compile code for big endian mode.
7369 @opindex mmangle-cpu
7370 Prepend the name of the cpu to all public symbol names.
7371 In multiple-processor systems, there are many ARC variants with different
7372 instruction and register set characteristics. This flag prevents code
7373 compiled for one cpu to be linked with code compiled for another.
7374 No facility exists for handling variants that are ``almost identical''.
7375 This is an all or nothing option.
7377 @item -mcpu=@var{cpu}
7379 Compile code for ARC variant @var{cpu}.
7380 Which variants are supported depend on the configuration.
7381 All variants support @option{-mcpu=base}, this is the default.
7383 @item -mtext=@var{text-section}
7384 @itemx -mdata=@var{data-section}
7385 @itemx -mrodata=@var{readonly-data-section}
7389 Put functions, data, and readonly data in @var{text-section},
7390 @var{data-section}, and @var{readonly-data-section} respectively
7391 by default. This can be overridden with the @code{section} attribute.
7392 @xref{Variable Attributes}.
7397 @subsection ARM Options
7400 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
7404 @item -mabi=@var{name}
7406 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
7407 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
7410 @opindex mapcs-frame
7411 Generate a stack frame that is compliant with the ARM Procedure Call
7412 Standard for all functions, even if this is not strictly necessary for
7413 correct execution of the code. Specifying @option{-fomit-frame-pointer}
7414 with this option will cause the stack frames not to be generated for
7415 leaf functions. The default is @option{-mno-apcs-frame}.
7419 This is a synonym for @option{-mapcs-frame}.
7422 @c not currently implemented
7423 @item -mapcs-stack-check
7424 @opindex mapcs-stack-check
7425 Generate code to check the amount of stack space available upon entry to
7426 every function (that actually uses some stack space). If there is
7427 insufficient space available then either the function
7428 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
7429 called, depending upon the amount of stack space required. The run time
7430 system is required to provide these functions. The default is
7431 @option{-mno-apcs-stack-check}, since this produces smaller code.
7433 @c not currently implemented
7435 @opindex mapcs-float
7436 Pass floating point arguments using the float point registers. This is
7437 one of the variants of the APCS@. This option is recommended if the
7438 target hardware has a floating point unit or if a lot of floating point
7439 arithmetic is going to be performed by the code. The default is
7440 @option{-mno-apcs-float}, since integer only code is slightly increased in
7441 size if @option{-mapcs-float} is used.
7443 @c not currently implemented
7444 @item -mapcs-reentrant
7445 @opindex mapcs-reentrant
7446 Generate reentrant, position independent code. The default is
7447 @option{-mno-apcs-reentrant}.
7450 @item -mthumb-interwork
7451 @opindex mthumb-interwork
7452 Generate code which supports calling between the ARM and Thumb
7453 instruction sets. Without this option the two instruction sets cannot
7454 be reliably used inside one program. The default is
7455 @option{-mno-thumb-interwork}, since slightly larger code is generated
7456 when @option{-mthumb-interwork} is specified.
7458 @item -mno-sched-prolog
7459 @opindex mno-sched-prolog
7460 Prevent the reordering of instructions in the function prolog, or the
7461 merging of those instruction with the instructions in the function's
7462 body. This means that all functions will start with a recognizable set
7463 of instructions (or in fact one of a choice from a small set of
7464 different function prologues), and this information can be used to
7465 locate the start if functions inside an executable piece of code. The
7466 default is @option{-msched-prolog}.
7469 @opindex mhard-float
7470 Generate output containing floating point instructions. This is the
7474 @opindex msoft-float
7475 Generate output containing library calls for floating point.
7476 @strong{Warning:} the requisite libraries are not available for all ARM
7477 targets. Normally the facilities of the machine's usual C compiler are
7478 used, but this cannot be done directly in cross-compilation. You must make
7479 your own arrangements to provide suitable library functions for
7482 @option{-msoft-float} changes the calling convention in the output file;
7483 therefore, it is only useful if you compile @emph{all} of a program with
7484 this option. In particular, you need to compile @file{libgcc.a}, the
7485 library that comes with GCC, with @option{-msoft-float} in order for
7488 @item -mfloat-abi=@var{name}
7490 Specifies which ABI to use for floating point values. Permissible values
7491 are: @samp{soft}, @samp{softfp} and @samp{hard}.
7493 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
7494 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
7495 of floating point instructions, but still uses the soft-float calling
7498 @item -mlittle-endian
7499 @opindex mlittle-endian
7500 Generate code for a processor running in little-endian mode. This is
7501 the default for all standard configurations.
7504 @opindex mbig-endian
7505 Generate code for a processor running in big-endian mode; the default is
7506 to compile code for a little-endian processor.
7508 @item -mwords-little-endian
7509 @opindex mwords-little-endian
7510 This option only applies when generating code for big-endian processors.
7511 Generate code for a little-endian word order but a big-endian byte
7512 order. That is, a byte order of the form @samp{32107654}. Note: this
7513 option should only be used if you require compatibility with code for
7514 big-endian ARM processors generated by versions of the compiler prior to
7517 @item -mcpu=@var{name}
7519 This specifies the name of the target ARM processor. GCC uses this name
7520 to determine what kind of instructions it can emit when generating
7521 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
7522 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
7523 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
7524 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
7525 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
7526 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
7527 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
7528 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
7529 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
7530 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
7531 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
7532 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
7533 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
7534 @samp{arm1176jz-s}, @samp{arm1176jzf-s}, @samp{xscale}, @samp{iwmmxt},
7537 @itemx -mtune=@var{name}
7539 This option is very similar to the @option{-mcpu=} option, except that
7540 instead of specifying the actual target processor type, and hence
7541 restricting which instructions can be used, it specifies that GCC should
7542 tune the performance of the code as if the target were of the type
7543 specified in this option, but still choosing the instructions that it
7544 will generate based on the cpu specified by a @option{-mcpu=} option.
7545 For some ARM implementations better performance can be obtained by using
7548 @item -march=@var{name}
7550 This specifies the name of the target ARM architecture. GCC uses this
7551 name to determine what kind of instructions it can emit when generating
7552 assembly code. This option can be used in conjunction with or instead
7553 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
7554 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
7555 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
7556 @samp{iwmmxt}, @samp{ep9312}.
7558 @item -mfpu=@var{name}
7559 @itemx -mfpe=@var{number}
7560 @itemx -mfp=@var{number}
7564 This specifies what floating point hardware (or hardware emulation) is
7565 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
7566 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
7567 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
7568 with older versions of GCC@.
7570 If @option{-msoft-float} is specified this specifies the format of
7571 floating point values.
7573 @item -mstructure-size-boundary=@var{n}
7574 @opindex mstructure-size-boundary
7575 The size of all structures and unions will be rounded up to a multiple
7576 of the number of bits set by this option. Permissible values are 8, 32
7577 and 64. The default value varies for different toolchains. For the COFF
7578 targeted toolchain the default value is 8. A value of 64 is only allowed
7579 if the underlying ABI supports it.
7581 Specifying the larger number can produce faster, more efficient code, but
7582 can also increase the size of the program. Different values are potentially
7583 incompatible. Code compiled with one value cannot necessarily expect to
7584 work with code or libraries compiled with another value, if they exchange
7585 information using structures or unions.
7587 @item -mabort-on-noreturn
7588 @opindex mabort-on-noreturn
7589 Generate a call to the function @code{abort} at the end of a
7590 @code{noreturn} function. It will be executed if the function tries to
7594 @itemx -mno-long-calls
7595 @opindex mlong-calls
7596 @opindex mno-long-calls
7597 Tells the compiler to perform function calls by first loading the
7598 address of the function into a register and then performing a subroutine
7599 call on this register. This switch is needed if the target function
7600 will lie outside of the 64 megabyte addressing range of the offset based
7601 version of subroutine call instruction.
7603 Even if this switch is enabled, not all function calls will be turned
7604 into long calls. The heuristic is that static functions, functions
7605 which have the @samp{short-call} attribute, functions that are inside
7606 the scope of a @samp{#pragma no_long_calls} directive and functions whose
7607 definitions have already been compiled within the current compilation
7608 unit, will not be turned into long calls. The exception to this rule is
7609 that weak function definitions, functions with the @samp{long-call}
7610 attribute or the @samp{section} attribute, and functions that are within
7611 the scope of a @samp{#pragma long_calls} directive, will always be
7612 turned into long calls.
7614 This feature is not enabled by default. Specifying
7615 @option{-mno-long-calls} will restore the default behavior, as will
7616 placing the function calls within the scope of a @samp{#pragma
7617 long_calls_off} directive. Note these switches have no effect on how
7618 the compiler generates code to handle function calls via function
7621 @item -mnop-fun-dllimport
7622 @opindex mnop-fun-dllimport
7623 Disable support for the @code{dllimport} attribute.
7625 @item -msingle-pic-base
7626 @opindex msingle-pic-base
7627 Treat the register used for PIC addressing as read-only, rather than
7628 loading it in the prologue for each function. The run-time system is
7629 responsible for initializing this register with an appropriate value
7630 before execution begins.
7632 @item -mpic-register=@var{reg}
7633 @opindex mpic-register
7634 Specify the register to be used for PIC addressing. The default is R10
7635 unless stack-checking is enabled, when R9 is used.
7637 @item -mcirrus-fix-invalid-insns
7638 @opindex mcirrus-fix-invalid-insns
7639 @opindex mno-cirrus-fix-invalid-insns
7640 Insert NOPs into the instruction stream to in order to work around
7641 problems with invalid Maverick instruction combinations. This option
7642 is only valid if the @option{-mcpu=ep9312} option has been used to
7643 enable generation of instructions for the Cirrus Maverick floating
7644 point co-processor. This option is not enabled by default, since the
7645 problem is only present in older Maverick implementations. The default
7646 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
7649 @item -mpoke-function-name
7650 @opindex mpoke-function-name
7651 Write the name of each function into the text section, directly
7652 preceding the function prologue. The generated code is similar to this:
7656 .ascii "arm_poke_function_name", 0
7659 .word 0xff000000 + (t1 - t0)
7660 arm_poke_function_name
7662 stmfd sp!, @{fp, ip, lr, pc@}
7666 When performing a stack backtrace, code can inspect the value of
7667 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
7668 location @code{pc - 12} and the top 8 bits are set, then we know that
7669 there is a function name embedded immediately preceding this location
7670 and has length @code{((pc[-3]) & 0xff000000)}.
7674 Generate code for the 16-bit Thumb instruction set. The default is to
7675 use the 32-bit ARM instruction set.
7678 @opindex mtpcs-frame
7679 Generate a stack frame that is compliant with the Thumb Procedure Call
7680 Standard for all non-leaf functions. (A leaf function is one that does
7681 not call any other functions.) The default is @option{-mno-tpcs-frame}.
7683 @item -mtpcs-leaf-frame
7684 @opindex mtpcs-leaf-frame
7685 Generate a stack frame that is compliant with the Thumb Procedure Call
7686 Standard for all leaf functions. (A leaf function is one that does
7687 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
7689 @item -mcallee-super-interworking
7690 @opindex mcallee-super-interworking
7691 Gives all externally visible functions in the file being compiled an ARM
7692 instruction set header which switches to Thumb mode before executing the
7693 rest of the function. This allows these functions to be called from
7694 non-interworking code.
7696 @item -mcaller-super-interworking
7697 @opindex mcaller-super-interworking
7698 Allows calls via function pointers (including virtual functions) to
7699 execute correctly regardless of whether the target code has been
7700 compiled for interworking or not. There is a small overhead in the cost
7701 of executing a function pointer if this option is enabled.
7703 @item -mtp=@var{name}
7705 Specify the access model for the thread local storage pointer. The valid
7706 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
7707 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
7708 (supported in the arm6k architecture), and @option{auto}, which uses the
7709 best available method for the selected processor. The default setting is
7715 @subsection AVR Options
7718 These options are defined for AVR implementations:
7721 @item -mmcu=@var{mcu}
7723 Specify ATMEL AVR instruction set or MCU type.
7725 Instruction set avr1 is for the minimal AVR core, not supported by the C
7726 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7727 attiny11, attiny12, attiny15, attiny28).
7729 Instruction set avr2 (default) is for the classic AVR core with up to
7730 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7731 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7732 at90c8534, at90s8535).
7734 Instruction set avr3 is for the classic AVR core with up to 128K program
7735 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
7737 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7738 memory space (MCU types: atmega8, atmega83, atmega85).
7740 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7741 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
7742 atmega64, atmega128, at43usb355, at94k).
7746 Output instruction sizes to the asm file.
7748 @item -minit-stack=@var{N}
7749 @opindex minit-stack
7750 Specify the initial stack address, which may be a symbol or numeric value,
7751 @samp{__stack} is the default.
7753 @item -mno-interrupts
7754 @opindex mno-interrupts
7755 Generated code is not compatible with hardware interrupts.
7756 Code size will be smaller.
7758 @item -mcall-prologues
7759 @opindex mcall-prologues
7760 Functions prologues/epilogues expanded as call to appropriate
7761 subroutines. Code size will be smaller.
7763 @item -mno-tablejump
7764 @opindex mno-tablejump
7765 Do not generate tablejump insns which sometimes increase code size.
7768 @opindex mtiny-stack
7769 Change only the low 8 bits of the stack pointer.
7773 Assume int to be 8 bit integer. This affects the sizes of all types: A
7774 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7775 and long long will be 4 bytes. Please note that this option does not
7776 comply to the C standards, but it will provide you with smaller code
7780 @node Blackfin Options
7781 @subsection Blackfin Options
7782 @cindex Blackfin Options
7785 @item -momit-leaf-frame-pointer
7786 @opindex momit-leaf-frame-pointer
7787 Don't keep the frame pointer in a register for leaf functions. This
7788 avoids the instructions to save, set up and restore frame pointers and
7789 makes an extra register available in leaf functions. The option
7790 @option{-fomit-frame-pointer} removes the frame pointer for all functions
7791 which might make debugging harder.
7793 @item -mspecld-anomaly
7794 @opindex mspecld-anomaly
7795 When enabled, the compiler will ensure that the generated code does not
7796 contain speculative loads after jump instructions. This option is enabled
7799 @item -mno-specld-anomaly
7800 @opindex mno-specld-anomaly
7801 Don't generate extra code to prevent speculative loads from occurring.
7803 @item -mcsync-anomaly
7804 @opindex mcsync-anomaly
7805 When enabled, the compiler will ensure that the generated code does not
7806 contain CSYNC or SSYNC instructions too soon after conditional branches.
7807 This option is enabled by default.
7809 @item -mno-csync-anomaly
7810 @opindex mno-csync-anomaly
7811 Don't generate extra code to prevent CSYNC or SSYNC instructions from
7812 occurring too soon after a conditional branch.
7816 When enabled, the compiler is free to take advantage of the knowledge that
7817 the entire program fits into the low 64k of memory.
7820 @opindex mno-low-64k
7821 Assume that the program is arbitrarily large. This is the default.
7823 @item -mid-shared-library
7824 @opindex mid-shared-library
7825 Generate code that supports shared libraries via the library ID method.
7826 This allows for execute in place and shared libraries in an environment
7827 without virtual memory management. This option implies @option{-fPIC}.
7829 @item -mno-id-shared-library
7830 @opindex mno-id-shared-library
7831 Generate code that doesn't assume ID based shared libraries are being used.
7832 This is the default.
7834 @item -mshared-library-id=n
7835 @opindex mshared-library-id
7836 Specified the identification number of the ID based shared library being
7837 compiled. Specifying a value of 0 will generate more compact code, specifying
7838 other values will force the allocation of that number to the current
7839 library but is no more space or time efficient than omitting this option.
7842 @itemx -mno-long-calls
7843 @opindex mlong-calls
7844 @opindex mno-long-calls
7845 Tells the compiler to perform function calls by first loading the
7846 address of the function into a register and then performing a subroutine
7847 call on this register. This switch is needed if the target function
7848 will lie outside of the 24 bit addressing range of the offset based
7849 version of subroutine call instruction.
7851 This feature is not enabled by default. Specifying
7852 @option{-mno-long-calls} will restore the default behavior. Note these
7853 switches have no effect on how the compiler generates code to handle
7854 function calls via function pointers.
7858 @subsection CRIS Options
7859 @cindex CRIS Options
7861 These options are defined specifically for the CRIS ports.
7864 @item -march=@var{architecture-type}
7865 @itemx -mcpu=@var{architecture-type}
7868 Generate code for the specified architecture. The choices for
7869 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7870 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
7871 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7874 @item -mtune=@var{architecture-type}
7876 Tune to @var{architecture-type} everything applicable about the generated
7877 code, except for the ABI and the set of available instructions. The
7878 choices for @var{architecture-type} are the same as for
7879 @option{-march=@var{architecture-type}}.
7881 @item -mmax-stack-frame=@var{n}
7882 @opindex mmax-stack-frame
7883 Warn when the stack frame of a function exceeds @var{n} bytes.
7885 @item -melinux-stacksize=@var{n}
7886 @opindex melinux-stacksize
7887 Only available with the @samp{cris-axis-aout} target. Arranges for
7888 indications in the program to the kernel loader that the stack of the
7889 program should be set to @var{n} bytes.
7895 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7896 @option{-march=v3} and @option{-march=v8} respectively.
7898 @item -mmul-bug-workaround
7899 @itemx -mno-mul-bug-workaround
7900 @opindex mmul-bug-workaround
7901 @opindex mno-mul-bug-workaround
7902 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7903 models where it applies. This option is active by default.
7907 Enable CRIS-specific verbose debug-related information in the assembly
7908 code. This option also has the effect to turn off the @samp{#NO_APP}
7909 formatted-code indicator to the assembler at the beginning of the
7914 Do not use condition-code results from previous instruction; always emit
7915 compare and test instructions before use of condition codes.
7917 @item -mno-side-effects
7918 @opindex mno-side-effects
7919 Do not emit instructions with side-effects in addressing modes other than
7923 @itemx -mno-stack-align
7925 @itemx -mno-data-align
7926 @itemx -mconst-align
7927 @itemx -mno-const-align
7928 @opindex mstack-align
7929 @opindex mno-stack-align
7930 @opindex mdata-align
7931 @opindex mno-data-align
7932 @opindex mconst-align
7933 @opindex mno-const-align
7934 These options (no-options) arranges (eliminate arrangements) for the
7935 stack-frame, individual data and constants to be aligned for the maximum
7936 single data access size for the chosen CPU model. The default is to
7937 arrange for 32-bit alignment. ABI details such as structure layout are
7938 not affected by these options.
7946 Similar to the stack- data- and const-align options above, these options
7947 arrange for stack-frame, writable data and constants to all be 32-bit,
7948 16-bit or 8-bit aligned. The default is 32-bit alignment.
7950 @item -mno-prologue-epilogue
7951 @itemx -mprologue-epilogue
7952 @opindex mno-prologue-epilogue
7953 @opindex mprologue-epilogue
7954 With @option{-mno-prologue-epilogue}, the normal function prologue and
7955 epilogue that sets up the stack-frame are omitted and no return
7956 instructions or return sequences are generated in the code. Use this
7957 option only together with visual inspection of the compiled code: no
7958 warnings or errors are generated when call-saved registers must be saved,
7959 or storage for local variable needs to be allocated.
7965 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
7966 instruction sequences that load addresses for functions from the PLT part
7967 of the GOT rather than (traditional on other architectures) calls to the
7968 PLT@. The default is @option{-mgotplt}.
7972 Legacy no-op option only recognized with the cris-axis-aout target.
7976 Legacy no-op option only recognized with the cris-axis-elf and
7977 cris-axis-linux-gnu targets.
7981 Only recognized with the cris-axis-aout target, where it selects a
7982 GNU/linux-like multilib, include files and instruction set for
7987 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
7991 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
7992 to link with input-output functions from a simulator library. Code,
7993 initialized data and zero-initialized data are allocated consecutively.
7997 Like @option{-sim}, but pass linker options to locate initialized data at
7998 0x40000000 and zero-initialized data at 0x80000000.
8002 @subsection CRX Options
8005 These options are defined specifically for the CRX ports.
8011 Enable the use of multiply-accumulate instructions. Disabled by default.
8015 Push instructions will be used to pass outgoing arguments when functions
8016 are called. Enabled by default.
8019 @node Darwin Options
8020 @subsection Darwin Options
8021 @cindex Darwin options
8023 These options are defined for all architectures running the Darwin operating
8026 FSF GCC on Darwin does not create ``fat'' object files; it will create
8027 an object file for the single architecture that it was built to
8028 target. Apple's GCC on Darwin does create ``fat'' files if multiple
8029 @option{-arch} options are used; it does so by running the compiler or
8030 linker multiple times and joining the results together with
8033 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
8034 @samp{i686}) is determined by the flags that specify the ISA
8035 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
8036 @option{-force_cpusubtype_ALL} option can be used to override this.
8038 The Darwin tools vary in their behavior when presented with an ISA
8039 mismatch. The assembler, @file{as}, will only permit instructions to
8040 be used that are valid for the subtype of the file it is generating,
8041 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
8042 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
8043 and print an error if asked to create a shared library with a less
8044 restrictive subtype than its input files (for instance, trying to put
8045 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
8046 for executables, @file{ld}, will quietly give the executable the most
8047 restrictive subtype of any of its input files.
8052 Add the framework directory @var{dir} to the head of the list of
8053 directories to be searched for header files. These directories are
8054 interleaved with those specified by @option{-I} options and are
8055 scanned in a left-to-right order.
8057 A framework directory is a directory with frameworks in it. A
8058 framework is a directory with a @samp{"Headers"} and/or
8059 @samp{"PrivateHeaders"} directory contained directly in it that ends
8060 in @samp{".framework"}. The name of a framework is the name of this
8061 directory excluding the @samp{".framework"}. Headers associated with
8062 the framework are found in one of those two directories, with
8063 @samp{"Headers"} being searched first. A subframework is a framework
8064 directory that is in a framework's @samp{"Frameworks"} directory.
8065 Includes of subframework headers can only appear in a header of a
8066 framework that contains the subframework, or in a sibling subframework
8067 header. Two subframeworks are siblings if they occur in the same
8068 framework. A subframework should not have the same name as a
8069 framework, a warning will be issued if this is violated. Currently a
8070 subframework cannot have subframeworks, in the future, the mechanism
8071 may be extended to support this. The standard frameworks can be found
8072 in @samp{"/System/Library/Frameworks"} and
8073 @samp{"/Library/Frameworks"}. An example include looks like
8074 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
8075 the name of the framework and header.h is found in the
8076 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
8080 Emit debugging information for symbols that are used. For STABS
8081 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
8082 This is by default ON@.
8086 Emit debugging information for all symbols and types.
8088 @item -mmacosx-version-min=@var{version}
8089 The earliest version of MacOS X that this executable will run on
8090 is @var{version}. Typical values of @var{version} include @code{10.1},
8091 @code{10.2}, and @code{10.3.9}.
8093 The default for this option is to make choices that seem to be most
8098 Enable kernel development mode. The @option{-mkernel} option sets
8099 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
8100 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
8101 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
8102 applicable. This mode also sets @option{-mno-altivec},
8103 @option{-msoft-float}, @option{-fno-builtin} and
8104 @option{-mlong-branch} for PowerPC targets.
8106 @item -mone-byte-bool
8107 @opindex -mone-byte-bool
8108 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
8109 By default @samp{sizeof(bool)} is @samp{4} when compiling for
8110 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
8111 option has no effect on x86.
8113 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
8114 to generate code that is not binary compatible with code generated
8115 without that switch. Using this switch may require recompiling all
8116 other modules in a program, including system libraries. Use this
8117 switch to conform to a non-default data model.
8119 @item -mfix-and-continue
8120 @itemx -ffix-and-continue
8121 @itemx -findirect-data
8122 @opindex mfix-and-continue
8123 @opindex ffix-and-continue
8124 @opindex findirect-data
8125 Generate code suitable for fast turn around development. Needed to
8126 enable gdb to dynamically load @code{.o} files into already running
8127 programs. @option{-findirect-data} and @option{-ffix-and-continue}
8128 are provided for backwards compatibility.
8132 Loads all members of static archive libraries.
8133 See man ld(1) for more information.
8135 @item -arch_errors_fatal
8136 @opindex arch_errors_fatal
8137 Cause the errors having to do with files that have the wrong architecture
8141 @opindex bind_at_load
8142 Causes the output file to be marked such that the dynamic linker will
8143 bind all undefined references when the file is loaded or launched.
8147 Produce a Mach-o bundle format file.
8148 See man ld(1) for more information.
8150 @item -bundle_loader @var{executable}
8151 @opindex bundle_loader
8152 This option specifies the @var{executable} that will be loading the build
8153 output file being linked. See man ld(1) for more information.
8156 @opindex -dynamiclib
8157 When passed this option, GCC will produce a dynamic library instead of
8158 an executable when linking, using the Darwin @file{libtool} command.
8160 @item -force_cpusubtype_ALL
8161 @opindex -force_cpusubtype_ALL
8162 This causes GCC's output file to have the @var{ALL} subtype, instead of
8163 one controlled by the @option{-mcpu} or @option{-march} option.
8165 @item -allowable_client @var{client_name}
8167 @itemx -compatibility_version
8168 @itemx -current_version
8170 @itemx -dependency-file
8172 @itemx -dylinker_install_name
8174 @itemx -exported_symbols_list
8176 @itemx -flat_namespace
8177 @itemx -force_flat_namespace
8178 @itemx -headerpad_max_install_names
8181 @itemx -install_name
8182 @itemx -keep_private_externs
8183 @itemx -multi_module
8184 @itemx -multiply_defined
8185 @itemx -multiply_defined_unused
8187 @itemx -no_dead_strip_inits_and_terms
8188 @itemx -nofixprebinding
8191 @itemx -noseglinkedit
8192 @itemx -pagezero_size
8194 @itemx -prebind_all_twolevel_modules
8195 @itemx -private_bundle
8196 @itemx -read_only_relocs
8198 @itemx -sectobjectsymbols
8202 @itemx -sectobjectsymbols
8205 @itemx -segs_read_only_addr
8206 @itemx -segs_read_write_addr
8207 @itemx -seg_addr_table
8208 @itemx -seg_addr_table_filename
8211 @itemx -segs_read_only_addr
8212 @itemx -segs_read_write_addr
8213 @itemx -single_module
8216 @itemx -sub_umbrella
8217 @itemx -twolevel_namespace
8220 @itemx -unexported_symbols_list
8221 @itemx -weak_reference_mismatches
8224 @opindex allowable_client
8225 @opindex client_name
8226 @opindex compatibility_version
8227 @opindex current_version
8229 @opindex dependency-file
8231 @opindex dylinker_install_name
8233 @opindex exported_symbols_list
8235 @opindex flat_namespace
8236 @opindex force_flat_namespace
8237 @opindex headerpad_max_install_names
8240 @opindex install_name
8241 @opindex keep_private_externs
8242 @opindex multi_module
8243 @opindex multiply_defined
8244 @opindex multiply_defined_unused
8246 @opindex no_dead_strip_inits_and_terms
8247 @opindex nofixprebinding
8248 @opindex nomultidefs
8250 @opindex noseglinkedit
8251 @opindex pagezero_size
8253 @opindex prebind_all_twolevel_modules
8254 @opindex private_bundle
8255 @opindex read_only_relocs
8257 @opindex sectobjectsymbols
8261 @opindex sectobjectsymbols
8264 @opindex segs_read_only_addr
8265 @opindex segs_read_write_addr
8266 @opindex seg_addr_table
8267 @opindex seg_addr_table_filename
8268 @opindex seglinkedit
8270 @opindex segs_read_only_addr
8271 @opindex segs_read_write_addr
8272 @opindex single_module
8274 @opindex sub_library
8275 @opindex sub_umbrella
8276 @opindex twolevel_namespace
8279 @opindex unexported_symbols_list
8280 @opindex weak_reference_mismatches
8281 @opindex whatsloaded
8283 These options are passed to the Darwin linker. The Darwin linker man page
8284 describes them in detail.
8287 @node DEC Alpha Options
8288 @subsection DEC Alpha Options
8290 These @samp{-m} options are defined for the DEC Alpha implementations:
8293 @item -mno-soft-float
8295 @opindex mno-soft-float
8296 @opindex msoft-float
8297 Use (do not use) the hardware floating-point instructions for
8298 floating-point operations. When @option{-msoft-float} is specified,
8299 functions in @file{libgcc.a} will be used to perform floating-point
8300 operations. Unless they are replaced by routines that emulate the
8301 floating-point operations, or compiled in such a way as to call such
8302 emulations routines, these routines will issue floating-point
8303 operations. If you are compiling for an Alpha without floating-point
8304 operations, you must ensure that the library is built so as not to call
8307 Note that Alpha implementations without floating-point operations are
8308 required to have floating-point registers.
8313 @opindex mno-fp-regs
8314 Generate code that uses (does not use) the floating-point register set.
8315 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8316 register set is not used, floating point operands are passed in integer
8317 registers as if they were integers and floating-point results are passed
8318 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8319 so any function with a floating-point argument or return value called by code
8320 compiled with @option{-mno-fp-regs} must also be compiled with that
8323 A typical use of this option is building a kernel that does not use,
8324 and hence need not save and restore, any floating-point registers.
8328 The Alpha architecture implements floating-point hardware optimized for
8329 maximum performance. It is mostly compliant with the IEEE floating
8330 point standard. However, for full compliance, software assistance is
8331 required. This option generates code fully IEEE compliant code
8332 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8333 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8334 defined during compilation. The resulting code is less efficient but is
8335 able to correctly support denormalized numbers and exceptional IEEE
8336 values such as not-a-number and plus/minus infinity. Other Alpha
8337 compilers call this option @option{-ieee_with_no_inexact}.
8339 @item -mieee-with-inexact
8340 @opindex mieee-with-inexact
8341 This is like @option{-mieee} except the generated code also maintains
8342 the IEEE @var{inexact-flag}. Turning on this option causes the
8343 generated code to implement fully-compliant IEEE math. In addition to
8344 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8345 macro. On some Alpha implementations the resulting code may execute
8346 significantly slower than the code generated by default. Since there is
8347 very little code that depends on the @var{inexact-flag}, you should
8348 normally not specify this option. Other Alpha compilers call this
8349 option @option{-ieee_with_inexact}.
8351 @item -mfp-trap-mode=@var{trap-mode}
8352 @opindex mfp-trap-mode
8353 This option controls what floating-point related traps are enabled.
8354 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8355 The trap mode can be set to one of four values:
8359 This is the default (normal) setting. The only traps that are enabled
8360 are the ones that cannot be disabled in software (e.g., division by zero
8364 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8368 Like @samp{su}, but the instructions are marked to be safe for software
8369 completion (see Alpha architecture manual for details).
8372 Like @samp{su}, but inexact traps are enabled as well.
8375 @item -mfp-rounding-mode=@var{rounding-mode}
8376 @opindex mfp-rounding-mode
8377 Selects the IEEE rounding mode. Other Alpha compilers call this option
8378 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8383 Normal IEEE rounding mode. Floating point numbers are rounded towards
8384 the nearest machine number or towards the even machine number in case
8388 Round towards minus infinity.
8391 Chopped rounding mode. Floating point numbers are rounded towards zero.
8394 Dynamic rounding mode. A field in the floating point control register
8395 (@var{fpcr}, see Alpha architecture reference manual) controls the
8396 rounding mode in effect. The C library initializes this register for
8397 rounding towards plus infinity. Thus, unless your program modifies the
8398 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8401 @item -mtrap-precision=@var{trap-precision}
8402 @opindex mtrap-precision
8403 In the Alpha architecture, floating point traps are imprecise. This
8404 means without software assistance it is impossible to recover from a
8405 floating trap and program execution normally needs to be terminated.
8406 GCC can generate code that can assist operating system trap handlers
8407 in determining the exact location that caused a floating point trap.
8408 Depending on the requirements of an application, different levels of
8409 precisions can be selected:
8413 Program precision. This option is the default and means a trap handler
8414 can only identify which program caused a floating point exception.
8417 Function precision. The trap handler can determine the function that
8418 caused a floating point exception.
8421 Instruction precision. The trap handler can determine the exact
8422 instruction that caused a floating point exception.
8425 Other Alpha compilers provide the equivalent options called
8426 @option{-scope_safe} and @option{-resumption_safe}.
8428 @item -mieee-conformant
8429 @opindex mieee-conformant
8430 This option marks the generated code as IEEE conformant. You must not
8431 use this option unless you also specify @option{-mtrap-precision=i} and either
8432 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8433 is to emit the line @samp{.eflag 48} in the function prologue of the
8434 generated assembly file. Under DEC Unix, this has the effect that
8435 IEEE-conformant math library routines will be linked in.
8437 @item -mbuild-constants
8438 @opindex mbuild-constants
8439 Normally GCC examines a 32- or 64-bit integer constant to
8440 see if it can construct it from smaller constants in two or three
8441 instructions. If it cannot, it will output the constant as a literal and
8442 generate code to load it from the data segment at runtime.
8444 Use this option to require GCC to construct @emph{all} integer constants
8445 using code, even if it takes more instructions (the maximum is six).
8447 You would typically use this option to build a shared library dynamic
8448 loader. Itself a shared library, it must relocate itself in memory
8449 before it can find the variables and constants in its own data segment.
8455 Select whether to generate code to be assembled by the vendor-supplied
8456 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8474 Indicate whether GCC should generate code to use the optional BWX,
8475 CIX, FIX and MAX instruction sets. The default is to use the instruction
8476 sets supported by the CPU type specified via @option{-mcpu=} option or that
8477 of the CPU on which GCC was built if none was specified.
8482 @opindex mfloat-ieee
8483 Generate code that uses (does not use) VAX F and G floating point
8484 arithmetic instead of IEEE single and double precision.
8486 @item -mexplicit-relocs
8487 @itemx -mno-explicit-relocs
8488 @opindex mexplicit-relocs
8489 @opindex mno-explicit-relocs
8490 Older Alpha assemblers provided no way to generate symbol relocations
8491 except via assembler macros. Use of these macros does not allow
8492 optimal instruction scheduling. GNU binutils as of version 2.12
8493 supports a new syntax that allows the compiler to explicitly mark
8494 which relocations should apply to which instructions. This option
8495 is mostly useful for debugging, as GCC detects the capabilities of
8496 the assembler when it is built and sets the default accordingly.
8500 @opindex msmall-data
8501 @opindex mlarge-data
8502 When @option{-mexplicit-relocs} is in effect, static data is
8503 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8504 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8505 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8506 16-bit relocations off of the @code{$gp} register. This limits the
8507 size of the small data area to 64KB, but allows the variables to be
8508 directly accessed via a single instruction.
8510 The default is @option{-mlarge-data}. With this option the data area
8511 is limited to just below 2GB@. Programs that require more than 2GB of
8512 data must use @code{malloc} or @code{mmap} to allocate the data in the
8513 heap instead of in the program's data segment.
8515 When generating code for shared libraries, @option{-fpic} implies
8516 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8520 @opindex msmall-text
8521 @opindex mlarge-text
8522 When @option{-msmall-text} is used, the compiler assumes that the
8523 code of the entire program (or shared library) fits in 4MB, and is
8524 thus reachable with a branch instruction. When @option{-msmall-data}
8525 is used, the compiler can assume that all local symbols share the
8526 same @code{$gp} value, and thus reduce the number of instructions
8527 required for a function call from 4 to 1.
8529 The default is @option{-mlarge-text}.
8531 @item -mcpu=@var{cpu_type}
8533 Set the instruction set and instruction scheduling parameters for
8534 machine type @var{cpu_type}. You can specify either the @samp{EV}
8535 style name or the corresponding chip number. GCC supports scheduling
8536 parameters for the EV4, EV5 and EV6 family of processors and will
8537 choose the default values for the instruction set from the processor
8538 you specify. If you do not specify a processor type, GCC will default
8539 to the processor on which the compiler was built.
8541 Supported values for @var{cpu_type} are
8547 Schedules as an EV4 and has no instruction set extensions.
8551 Schedules as an EV5 and has no instruction set extensions.
8555 Schedules as an EV5 and supports the BWX extension.
8560 Schedules as an EV5 and supports the BWX and MAX extensions.
8564 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8568 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8571 @item -mtune=@var{cpu_type}
8573 Set only the instruction scheduling parameters for machine type
8574 @var{cpu_type}. The instruction set is not changed.
8576 @item -mmemory-latency=@var{time}
8577 @opindex mmemory-latency
8578 Sets the latency the scheduler should assume for typical memory
8579 references as seen by the application. This number is highly
8580 dependent on the memory access patterns used by the application
8581 and the size of the external cache on the machine.
8583 Valid options for @var{time} are
8587 A decimal number representing clock cycles.
8593 The compiler contains estimates of the number of clock cycles for
8594 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8595 (also called Dcache, Scache, and Bcache), as well as to main memory.
8596 Note that L3 is only valid for EV5.
8601 @node DEC Alpha/VMS Options
8602 @subsection DEC Alpha/VMS Options
8604 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8607 @item -mvms-return-codes
8608 @opindex mvms-return-codes
8609 Return VMS condition codes from main. The default is to return POSIX
8610 style condition (e.g.@ error) codes.
8614 @subsection FRV Options
8621 Only use the first 32 general purpose registers.
8626 Use all 64 general purpose registers.
8631 Use only the first 32 floating point registers.
8636 Use all 64 floating point registers
8639 @opindex mhard-float
8641 Use hardware instructions for floating point operations.
8644 @opindex msoft-float
8646 Use library routines for floating point operations.
8651 Dynamically allocate condition code registers.
8656 Do not try to dynamically allocate condition code registers, only
8657 use @code{icc0} and @code{fcc0}.
8662 Change ABI to use double word insns.
8667 Do not use double word instructions.
8672 Use floating point double instructions.
8677 Do not use floating point double instructions.
8682 Use media instructions.
8687 Do not use media instructions.
8692 Use multiply and add/subtract instructions.
8697 Do not use multiply and add/subtract instructions.
8702 Select the FDPIC ABI, that uses function descriptors to represent
8703 pointers to functions. Without any PIC/PIE-related options, it
8704 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
8705 assumes GOT entries and small data are within a 12-bit range from the
8706 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
8707 are computed with 32 bits.
8710 @opindex minline-plt
8712 Enable inlining of PLT entries in function calls to functions that are
8713 not known to bind locally. It has no effect without @option{-mfdpic}.
8714 It's enabled by default if optimizing for speed and compiling for
8715 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
8716 optimization option such as @option{-O3} or above is present in the
8722 Assume a large TLS segment when generating thread-local code.
8727 Do not assume a large TLS segment when generating thread-local code.
8732 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
8733 that is known to be in read-only sections. It's enabled by default,
8734 except for @option{-fpic} or @option{-fpie}: even though it may help
8735 make the global offset table smaller, it trades 1 instruction for 4.
8736 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
8737 one of which may be shared by multiple symbols, and it avoids the need
8738 for a GOT entry for the referenced symbol, so it's more likely to be a
8739 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
8741 @item -multilib-library-pic
8742 @opindex multilib-library-pic
8744 Link with the (library, not FD) pic libraries. It's implied by
8745 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
8746 @option{-fpic} without @option{-mfdpic}. You should never have to use
8752 Follow the EABI requirement of always creating a frame pointer whenever
8753 a stack frame is allocated. This option is enabled by default and can
8754 be disabled with @option{-mno-linked-fp}.
8757 @opindex mlong-calls
8759 Use indirect addressing to call functions outside the current
8760 compilation unit. This allows the functions to be placed anywhere
8761 within the 32-bit address space.
8763 @item -malign-labels
8764 @opindex malign-labels
8766 Try to align labels to an 8-byte boundary by inserting nops into the
8767 previous packet. This option only has an effect when VLIW packing
8768 is enabled. It doesn't create new packets; it merely adds nops to
8772 @opindex mlibrary-pic
8774 Generate position-independent EABI code.
8779 Use only the first four media accumulator registers.
8784 Use all eight media accumulator registers.
8789 Pack VLIW instructions.
8794 Do not pack VLIW instructions.
8799 Do not mark ABI switches in e_flags.
8804 Enable the use of conditional-move instructions (default).
8806 This switch is mainly for debugging the compiler and will likely be removed
8807 in a future version.
8809 @item -mno-cond-move
8810 @opindex mno-cond-move
8812 Disable the use of conditional-move instructions.
8814 This switch is mainly for debugging the compiler and will likely be removed
8815 in a future version.
8820 Enable the use of conditional set instructions (default).
8822 This switch is mainly for debugging the compiler and will likely be removed
8823 in a future version.
8828 Disable the use of conditional set instructions.
8830 This switch is mainly for debugging the compiler and will likely be removed
8831 in a future version.
8836 Enable the use of conditional execution (default).
8838 This switch is mainly for debugging the compiler and will likely be removed
8839 in a future version.
8841 @item -mno-cond-exec
8842 @opindex mno-cond-exec
8844 Disable the use of conditional execution.
8846 This switch is mainly for debugging the compiler and will likely be removed
8847 in a future version.
8850 @opindex mvliw-branch
8852 Run a pass to pack branches into VLIW instructions (default).
8854 This switch is mainly for debugging the compiler and will likely be removed
8855 in a future version.
8857 @item -mno-vliw-branch
8858 @opindex mno-vliw-branch
8860 Do not run a pass to pack branches into VLIW instructions.
8862 This switch is mainly for debugging the compiler and will likely be removed
8863 in a future version.
8865 @item -mmulti-cond-exec
8866 @opindex mmulti-cond-exec
8868 Enable optimization of @code{&&} and @code{||} in conditional execution
8871 This switch is mainly for debugging the compiler and will likely be removed
8872 in a future version.
8874 @item -mno-multi-cond-exec
8875 @opindex mno-multi-cond-exec
8877 Disable optimization of @code{&&} and @code{||} in conditional execution.
8879 This switch is mainly for debugging the compiler and will likely be removed
8880 in a future version.
8882 @item -mnested-cond-exec
8883 @opindex mnested-cond-exec
8885 Enable nested conditional execution optimizations (default).
8887 This switch is mainly for debugging the compiler and will likely be removed
8888 in a future version.
8890 @item -mno-nested-cond-exec
8891 @opindex mno-nested-cond-exec
8893 Disable nested conditional execution optimizations.
8895 This switch is mainly for debugging the compiler and will likely be removed
8896 in a future version.
8898 @item -moptimize-membar
8899 @opindex moptimize-membar
8901 This switch removes redundant @code{membar} instructions from the
8902 compiler generated code. It is enabled by default.
8904 @item -mno-optimize-membar
8905 @opindex mno-optimize-membar
8907 This switch disables the automatic removal of redundant @code{membar}
8908 instructions from the generated code.
8910 @item -mtomcat-stats
8911 @opindex mtomcat-stats
8913 Cause gas to print out tomcat statistics.
8915 @item -mcpu=@var{cpu}
8918 Select the processor type for which to generate code. Possible values are
8919 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
8920 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
8924 @node GNU/Linux Options
8925 @subsection GNU/Linux Options
8927 These @samp{-m} options are defined for GNU/Linux targets:
8932 Use the GNU C library instead of uClibc. This is the default except
8933 on @samp{*-*-linux-*uclibc*} targets.
8937 Use uClibc instead of the GNU C library. This is the default on
8938 @samp{*-*-linux-*uclibc*} targets.
8941 @node H8/300 Options
8942 @subsection H8/300 Options
8944 These @samp{-m} options are defined for the H8/300 implementations:
8949 Shorten some address references at link time, when possible; uses the
8950 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8951 ld, Using ld}, for a fuller description.
8955 Generate code for the H8/300H@.
8959 Generate code for the H8S@.
8963 Generate code for the H8S and H8/300H in the normal mode. This switch
8964 must be used either with @option{-mh} or @option{-ms}.
8968 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
8972 Make @code{int} data 32 bits by default.
8976 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
8977 The default for the H8/300H and H8S is to align longs and floats on 4
8979 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8980 This option has no effect on the H8/300.
8984 @subsection HPPA Options
8985 @cindex HPPA Options
8987 These @samp{-m} options are defined for the HPPA family of computers:
8990 @item -march=@var{architecture-type}
8992 Generate code for the specified architecture. The choices for
8993 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8994 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8995 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8996 architecture option for your machine. Code compiled for lower numbered
8997 architectures will run on higher numbered architectures, but not the
9001 @itemx -mpa-risc-1-1
9002 @itemx -mpa-risc-2-0
9003 @opindex mpa-risc-1-0
9004 @opindex mpa-risc-1-1
9005 @opindex mpa-risc-2-0
9006 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
9009 @opindex mbig-switch
9010 Generate code suitable for big switch tables. Use this option only if
9011 the assembler/linker complain about out of range branches within a switch
9014 @item -mjump-in-delay
9015 @opindex mjump-in-delay
9016 Fill delay slots of function calls with unconditional jump instructions
9017 by modifying the return pointer for the function call to be the target
9018 of the conditional jump.
9020 @item -mdisable-fpregs
9021 @opindex mdisable-fpregs
9022 Prevent floating point registers from being used in any manner. This is
9023 necessary for compiling kernels which perform lazy context switching of
9024 floating point registers. If you use this option and attempt to perform
9025 floating point operations, the compiler will abort.
9027 @item -mdisable-indexing
9028 @opindex mdisable-indexing
9029 Prevent the compiler from using indexing address modes. This avoids some
9030 rather obscure problems when compiling MIG generated code under MACH@.
9032 @item -mno-space-regs
9033 @opindex mno-space-regs
9034 Generate code that assumes the target has no space registers. This allows
9035 GCC to generate faster indirect calls and use unscaled index address modes.
9037 Such code is suitable for level 0 PA systems and kernels.
9039 @item -mfast-indirect-calls
9040 @opindex mfast-indirect-calls
9041 Generate code that assumes calls never cross space boundaries. This
9042 allows GCC to emit code which performs faster indirect calls.
9044 This option will not work in the presence of shared libraries or nested
9047 @item -mfixed-range=@var{register-range}
9048 @opindex mfixed-range
9049 Generate code treating the given register range as fixed registers.
9050 A fixed register is one that the register allocator can not use. This is
9051 useful when compiling kernel code. A register range is specified as
9052 two registers separated by a dash. Multiple register ranges can be
9053 specified separated by a comma.
9055 @item -mlong-load-store
9056 @opindex mlong-load-store
9057 Generate 3-instruction load and store sequences as sometimes required by
9058 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
9061 @item -mportable-runtime
9062 @opindex mportable-runtime
9063 Use the portable calling conventions proposed by HP for ELF systems.
9067 Enable the use of assembler directives only GAS understands.
9069 @item -mschedule=@var{cpu-type}
9071 Schedule code according to the constraints for the machine type
9072 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
9073 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
9074 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
9075 proper scheduling option for your machine. The default scheduling is
9079 @opindex mlinker-opt
9080 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
9081 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
9082 linkers in which they give bogus error messages when linking some programs.
9085 @opindex msoft-float
9086 Generate output containing library calls for floating point.
9087 @strong{Warning:} the requisite libraries are not available for all HPPA
9088 targets. Normally the facilities of the machine's usual C compiler are
9089 used, but this cannot be done directly in cross-compilation. You must make
9090 your own arrangements to provide suitable library functions for
9091 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
9092 does provide software floating point support.
9094 @option{-msoft-float} changes the calling convention in the output file;
9095 therefore, it is only useful if you compile @emph{all} of a program with
9096 this option. In particular, you need to compile @file{libgcc.a}, the
9097 library that comes with GCC, with @option{-msoft-float} in order for
9102 Generate the predefine, @code{_SIO}, for server IO@. The default is
9103 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
9104 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
9105 options are available under HP-UX and HI-UX@.
9109 Use GNU ld specific options. This passes @option{-shared} to ld when
9110 building a shared library. It is the default when GCC is configured,
9111 explicitly or implicitly, with the GNU linker. This option does not
9112 have any affect on which ld is called, it only changes what parameters
9113 are passed to that ld. The ld that is called is determined by the
9114 @option{--with-ld} configure option, GCC's program search path, and
9115 finally by the user's @env{PATH}. The linker used by GCC can be printed
9116 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
9117 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9121 Use HP ld specific options. This passes @option{-b} to ld when building
9122 a shared library and passes @option{+Accept TypeMismatch} to ld on all
9123 links. It is the default when GCC is configured, explicitly or
9124 implicitly, with the HP linker. This option does not have any affect on
9125 which ld is called, it only changes what parameters are passed to that
9126 ld. The ld that is called is determined by the @option{--with-ld}
9127 configure option, GCC's program search path, and finally by the user's
9128 @env{PATH}. The linker used by GCC can be printed using @samp{which
9129 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
9130 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9133 @opindex mno-long-calls
9134 Generate code that uses long call sequences. This ensures that a call
9135 is always able to reach linker generated stubs. The default is to generate
9136 long calls only when the distance from the call site to the beginning
9137 of the function or translation unit, as the case may be, exceeds a
9138 predefined limit set by the branch type being used. The limits for
9139 normal calls are 7,600,000 and 240,000 bytes, respectively for the
9140 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
9143 Distances are measured from the beginning of functions when using the
9144 @option{-ffunction-sections} option, or when using the @option{-mgas}
9145 and @option{-mno-portable-runtime} options together under HP-UX with
9148 It is normally not desirable to use this option as it will degrade
9149 performance. However, it may be useful in large applications,
9150 particularly when partial linking is used to build the application.
9152 The types of long calls used depends on the capabilities of the
9153 assembler and linker, and the type of code being generated. The
9154 impact on systems that support long absolute calls, and long pic
9155 symbol-difference or pc-relative calls should be relatively small.
9156 However, an indirect call is used on 32-bit ELF systems in pic code
9157 and it is quite long.
9159 @item -munix=@var{unix-std}
9161 Generate compiler predefines and select a startfile for the specified
9162 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
9163 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
9164 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
9165 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
9166 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
9169 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
9170 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
9171 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
9172 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
9173 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
9174 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
9176 It is @emph{important} to note that this option changes the interfaces
9177 for various library routines. It also affects the operational behavior
9178 of the C library. Thus, @emph{extreme} care is needed in using this
9181 Library code that is intended to operate with more than one UNIX
9182 standard must test, set and restore the variable @var{__xpg4_extended_mask}
9183 as appropriate. Most GNU software doesn't provide this capability.
9187 Suppress the generation of link options to search libdld.sl when the
9188 @option{-static} option is specified on HP-UX 10 and later.
9192 The HP-UX implementation of setlocale in libc has a dependency on
9193 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9194 when the @option{-static} option is specified, special link options
9195 are needed to resolve this dependency.
9197 On HP-UX 10 and later, the GCC driver adds the necessary options to
9198 link with libdld.sl when the @option{-static} option is specified.
9199 This causes the resulting binary to be dynamic. On the 64-bit port,
9200 the linkers generate dynamic binaries by default in any case. The
9201 @option{-nolibdld} option can be used to prevent the GCC driver from
9202 adding these link options.
9206 Add support for multithreading with the @dfn{dce thread} library
9207 under HP-UX@. This option sets flags for both the preprocessor and
9211 @node i386 and x86-64 Options
9212 @subsection Intel 386 and AMD x86-64 Options
9213 @cindex i386 Options
9214 @cindex x86-64 Options
9215 @cindex Intel 386 Options
9216 @cindex AMD x86-64 Options
9218 These @samp{-m} options are defined for the i386 and x86-64 family of
9222 @item -mtune=@var{cpu-type}
9224 Tune to @var{cpu-type} everything applicable about the generated code, except
9225 for the ABI and the set of available instructions. The choices for
9229 Produce code optimized for the most common IA32/AMD64/EM64T processors.
9230 If you know the CPU on which your code will run, then you should use
9231 the corresponding @option{-mtune} option instead of
9232 @option{-mtune=generic}. But, if you do not know exactly what CPU users
9233 of your application will have, then you should use this option.
9235 As new processors are deployed in the marketplace, the behavior of this
9236 option will change. Therefore, if you upgrade to a newer version of
9237 GCC, the code generated option will change to reflect the processors
9238 that were most common when that version of GCC was released.
9240 There is no @option{-march=generic} option because @option{-march}
9241 indicates the instruction set the compiler can use, and there is no
9242 generic instruction set applicable to all processors. In contrast,
9243 @option{-mtune} indicates the processor (or, in this case, collection of
9244 processors) for which the code is optimized.
9246 This selects the CPU to tune for at compilation time by determining
9247 the processor type of the compiling machine. Using @option{-mtune=native}
9248 will produce code optimized for the local machine under the constraints
9249 of the selected instruction set. Using @option{-march=native} will
9250 enable all instruction subsets supported by the local machine (hence
9251 the result might not run on different machines).
9253 Original Intel's i386 CPU@.
9255 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
9257 Intel Pentium CPU with no MMX support.
9259 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
9261 Intel PentiumPro CPU@.
9263 Same as @code{generic}, but when used as @code{march} option, PentiumPro
9264 instruction set will be used, so the code will run on all i686 family chips.
9266 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
9267 @item pentium3, pentium3m
9268 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
9271 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
9272 support. Used by Centrino notebooks.
9273 @item pentium4, pentium4m
9274 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
9276 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
9279 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
9280 SSE2 and SSE3 instruction set support.
9282 AMD K6 CPU with MMX instruction set support.
9284 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
9285 @item athlon, athlon-tbird
9286 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
9288 @item athlon-4, athlon-xp, athlon-mp
9289 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
9290 instruction set support.
9291 @item k8, opteron, athlon64, athlon-fx
9292 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
9293 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
9295 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
9298 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
9299 instruction set support.
9301 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
9302 implemented for this chip.)
9304 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
9305 implemented for this chip.)
9307 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
9310 While picking a specific @var{cpu-type} will schedule things appropriately
9311 for that particular chip, the compiler will not generate any code that
9312 does not run on the i386 without the @option{-march=@var{cpu-type}} option
9315 @item -march=@var{cpu-type}
9317 Generate instructions for the machine type @var{cpu-type}. The choices
9318 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
9319 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
9321 @item -mcpu=@var{cpu-type}
9323 A deprecated synonym for @option{-mtune}.
9332 @opindex mpentiumpro
9333 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
9334 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
9335 These synonyms are deprecated.
9337 @item -mfpmath=@var{unit}
9339 Generate floating point arithmetics for selected unit @var{unit}. The choices
9344 Use the standard 387 floating point coprocessor present majority of chips and
9345 emulated otherwise. Code compiled with this option will run almost everywhere.
9346 The temporary results are computed in 80bit precision instead of precision
9347 specified by the type resulting in slightly different results compared to most
9348 of other chips. See @option{-ffloat-store} for more detailed description.
9350 This is the default choice for i386 compiler.
9353 Use scalar floating point instructions present in the SSE instruction set.
9354 This instruction set is supported by Pentium3 and newer chips, in the AMD line
9355 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
9356 instruction set supports only single precision arithmetics, thus the double and
9357 extended precision arithmetics is still done using 387. Later version, present
9358 only in Pentium4 and the future AMD x86-64 chips supports double precision
9361 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
9362 or @option{-msse2} switches to enable SSE extensions and make this option
9363 effective. For the x86-64 compiler, these extensions are enabled by default.
9365 The resulting code should be considerably faster in the majority of cases and avoid
9366 the numerical instability problems of 387 code, but may break some existing
9367 code that expects temporaries to be 80bit.
9369 This is the default choice for the x86-64 compiler.
9372 Attempt to utilize both instruction sets at once. This effectively double the
9373 amount of available registers and on chips with separate execution units for
9374 387 and SSE the execution resources too. Use this option with care, as it is
9375 still experimental, because the GCC register allocator does not model separate
9376 functional units well resulting in instable performance.
9379 @item -masm=@var{dialect}
9380 @opindex masm=@var{dialect}
9381 Output asm instructions using selected @var{dialect}. Supported
9382 choices are @samp{intel} or @samp{att} (the default one). Darwin does
9383 not support @samp{intel}.
9388 @opindex mno-ieee-fp
9389 Control whether or not the compiler uses IEEE floating point
9390 comparisons. These handle correctly the case where the result of a
9391 comparison is unordered.
9394 @opindex msoft-float
9395 Generate output containing library calls for floating point.
9396 @strong{Warning:} the requisite libraries are not part of GCC@.
9397 Normally the facilities of the machine's usual C compiler are used, but
9398 this can't be done directly in cross-compilation. You must make your
9399 own arrangements to provide suitable library functions for
9402 On machines where a function returns floating point results in the 80387
9403 register stack, some floating point opcodes may be emitted even if
9404 @option{-msoft-float} is used.
9406 @item -mno-fp-ret-in-387
9407 @opindex mno-fp-ret-in-387
9408 Do not use the FPU registers for return values of functions.
9410 The usual calling convention has functions return values of types
9411 @code{float} and @code{double} in an FPU register, even if there
9412 is no FPU@. The idea is that the operating system should emulate
9415 The option @option{-mno-fp-ret-in-387} causes such values to be returned
9416 in ordinary CPU registers instead.
9418 @item -mno-fancy-math-387
9419 @opindex mno-fancy-math-387
9420 Some 387 emulators do not support the @code{sin}, @code{cos} and
9421 @code{sqrt} instructions for the 387. Specify this option to avoid
9422 generating those instructions. This option is the default on FreeBSD,
9423 OpenBSD and NetBSD@. This option is overridden when @option{-march}
9424 indicates that the target cpu will always have an FPU and so the
9425 instruction will not need emulation. As of revision 2.6.1, these
9426 instructions are not generated unless you also use the
9427 @option{-funsafe-math-optimizations} switch.
9429 @item -malign-double
9430 @itemx -mno-align-double
9431 @opindex malign-double
9432 @opindex mno-align-double
9433 Control whether GCC aligns @code{double}, @code{long double}, and
9434 @code{long long} variables on a two word boundary or a one word
9435 boundary. Aligning @code{double} variables on a two word boundary will
9436 produce code that runs somewhat faster on a @samp{Pentium} at the
9437 expense of more memory.
9439 On x86-64, @option{-malign-double} is enabled by default.
9441 @strong{Warning:} if you use the @option{-malign-double} switch,
9442 structures containing the above types will be aligned differently than
9443 the published application binary interface specifications for the 386
9444 and will not be binary compatible with structures in code compiled
9445 without that switch.
9447 @item -m96bit-long-double
9448 @itemx -m128bit-long-double
9449 @opindex m96bit-long-double
9450 @opindex m128bit-long-double
9451 These switches control the size of @code{long double} type. The i386
9452 application binary interface specifies the size to be 96 bits,
9453 so @option{-m96bit-long-double} is the default in 32 bit mode.
9455 Modern architectures (Pentium and newer) would prefer @code{long double}
9456 to be aligned to an 8 or 16 byte boundary. In arrays or structures
9457 conforming to the ABI, this would not be possible. So specifying a
9458 @option{-m128bit-long-double} will align @code{long double}
9459 to a 16 byte boundary by padding the @code{long double} with an additional
9462 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
9463 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
9465 Notice that neither of these options enable any extra precision over the x87
9466 standard of 80 bits for a @code{long double}.
9468 @strong{Warning:} if you override the default value for your target ABI, the
9469 structures and arrays containing @code{long double} variables will change
9470 their size as well as function calling convention for function taking
9471 @code{long double} will be modified. Hence they will not be binary
9472 compatible with arrays or structures in code compiled without that switch.
9474 @item -mmlarge-data-threshold=@var{number}
9475 @opindex mlarge-data-threshold=@var{number}
9476 When @option{-mcmodel=medium} is specified, the data greater than
9477 @var{threshold} are placed in large data section. This value must be the
9478 same across all object linked into the binary and defaults to 65535.
9481 @itemx -mno-svr3-shlib
9482 @opindex msvr3-shlib
9483 @opindex mno-svr3-shlib
9484 Control whether GCC places uninitialized local variables into the
9485 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
9486 into @code{bss}. These options are meaningful only on System V Release 3.
9490 Use a different function-calling convention, in which functions that
9491 take a fixed number of arguments return with the @code{ret} @var{num}
9492 instruction, which pops their arguments while returning. This saves one
9493 instruction in the caller since there is no need to pop the arguments
9496 You can specify that an individual function is called with this calling
9497 sequence with the function attribute @samp{stdcall}. You can also
9498 override the @option{-mrtd} option by using the function attribute
9499 @samp{cdecl}. @xref{Function Attributes}.
9501 @strong{Warning:} this calling convention is incompatible with the one
9502 normally used on Unix, so you cannot use it if you need to call
9503 libraries compiled with the Unix compiler.
9505 Also, you must provide function prototypes for all functions that
9506 take variable numbers of arguments (including @code{printf});
9507 otherwise incorrect code will be generated for calls to those
9510 In addition, seriously incorrect code will result if you call a
9511 function with too many arguments. (Normally, extra arguments are
9512 harmlessly ignored.)
9514 @item -mregparm=@var{num}
9516 Control how many registers are used to pass integer arguments. By
9517 default, no registers are used to pass arguments, and at most 3
9518 registers can be used. You can control this behavior for a specific
9519 function by using the function attribute @samp{regparm}.
9520 @xref{Function Attributes}.
9522 @strong{Warning:} if you use this switch, and
9523 @var{num} is nonzero, then you must build all modules with the same
9524 value, including any libraries. This includes the system libraries and
9528 @opindex msseregparm
9529 Use SSE register passing conventions for float and double arguments
9530 and return values. You can control this behavior for a specific
9531 function by using the function attribute @samp{sseregparm}.
9532 @xref{Function Attributes}.
9534 @strong{Warning:} if you use this switch then you must build all
9535 modules with the same value, including any libraries. This includes
9536 the system libraries and startup modules.
9538 @item -mstackrealign
9539 @opindex mstackrealign
9540 Realign the stack at entry. On the Intel x86, the
9541 @option{-mstackrealign} option will generate an alternate prologue and
9542 epilogue that realigns the runtime stack. This supports mixing legacy
9543 codes that keep a 4-byte aligned stack with modern codes that keep a
9544 16-byte stack for SSE compatibility. The alternate prologue and
9545 epilogue are slower and bigger than the regular ones, and the
9546 alternate prologue requires an extra scratch register; this lowers the
9547 number of registers available if used in conjunction with the
9548 @code{regparm} attribute. The @option{-mstackrealign} option is
9549 incompatible with the nested function prologue; this is considered a
9550 hard error. See also the attribute @code{force_align_arg_pointer},
9551 applicable to individual functions.
9553 @item -mpreferred-stack-boundary=@var{num}
9554 @opindex mpreferred-stack-boundary
9555 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
9556 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
9557 the default is 4 (16 bytes or 128 bits).
9559 On Pentium and PentiumPro, @code{double} and @code{long double} values
9560 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
9561 suffer significant run time performance penalties. On Pentium III, the
9562 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
9563 properly if it is not 16 byte aligned.
9565 To ensure proper alignment of this values on the stack, the stack boundary
9566 must be as aligned as that required by any value stored on the stack.
9567 Further, every function must be generated such that it keeps the stack
9568 aligned. Thus calling a function compiled with a higher preferred
9569 stack boundary from a function compiled with a lower preferred stack
9570 boundary will most likely misalign the stack. It is recommended that
9571 libraries that use callbacks always use the default setting.
9573 This extra alignment does consume extra stack space, and generally
9574 increases code size. Code that is sensitive to stack space usage, such
9575 as embedded systems and operating system kernels, may want to reduce the
9576 preferred alignment to @option{-mpreferred-stack-boundary=2}.
9596 These switches enable or disable the use of instructions in the MMX,
9597 SSE, SSE2, SSE3, SSSE3 or 3DNow! extended instruction sets.
9598 These extensions are also available as built-in functions: see
9599 @ref{X86 Built-in Functions}, for details of the functions enabled and
9600 disabled by these switches.
9602 To have SSE/SSE2 instructions generated automatically from floating-point
9603 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
9605 These options will enable GCC to use these extended instructions in
9606 generated code, even without @option{-mfpmath=sse}. Applications which
9607 perform runtime CPU detection must compile separate files for each
9608 supported architecture, using the appropriate flags. In particular,
9609 the file containing the CPU detection code should be compiled without
9613 @itemx -mno-push-args
9615 @opindex mno-push-args
9616 Use PUSH operations to store outgoing parameters. This method is shorter
9617 and usually equally fast as method using SUB/MOV operations and is enabled
9618 by default. In some cases disabling it may improve performance because of
9619 improved scheduling and reduced dependencies.
9621 @item -maccumulate-outgoing-args
9622 @opindex maccumulate-outgoing-args
9623 If enabled, the maximum amount of space required for outgoing arguments will be
9624 computed in the function prologue. This is faster on most modern CPUs
9625 because of reduced dependencies, improved scheduling and reduced stack usage
9626 when preferred stack boundary is not equal to 2. The drawback is a notable
9627 increase in code size. This switch implies @option{-mno-push-args}.
9631 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
9632 on thread-safe exception handling must compile and link all code with the
9633 @option{-mthreads} option. When compiling, @option{-mthreads} defines
9634 @option{-D_MT}; when linking, it links in a special thread helper library
9635 @option{-lmingwthrd} which cleans up per thread exception handling data.
9637 @item -mno-align-stringops
9638 @opindex mno-align-stringops
9639 Do not align destination of inlined string operations. This switch reduces
9640 code size and improves performance in case the destination is already aligned,
9641 but GCC doesn't know about it.
9643 @item -minline-all-stringops
9644 @opindex minline-all-stringops
9645 By default GCC inlines string operations only when destination is known to be
9646 aligned at least to 4 byte boundary. This enables more inlining, increase code
9647 size, but may improve performance of code that depends on fast memcpy, strlen
9648 and memset for short lengths.
9650 @item -momit-leaf-frame-pointer
9651 @opindex momit-leaf-frame-pointer
9652 Don't keep the frame pointer in a register for leaf functions. This
9653 avoids the instructions to save, set up and restore frame pointers and
9654 makes an extra register available in leaf functions. The option
9655 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9656 which might make debugging harder.
9658 @item -mtls-direct-seg-refs
9659 @itemx -mno-tls-direct-seg-refs
9660 @opindex mtls-direct-seg-refs
9661 Controls whether TLS variables may be accessed with offsets from the
9662 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
9663 or whether the thread base pointer must be added. Whether or not this
9664 is legal depends on the operating system, and whether it maps the
9665 segment to cover the entire TLS area.
9667 For systems that use GNU libc, the default is on.
9670 These @samp{-m} switches are supported in addition to the above
9671 on AMD x86-64 processors in 64-bit environments.
9678 Generate code for a 32-bit or 64-bit environment.
9679 The 32-bit environment sets int, long and pointer to 32 bits and
9680 generates code that runs on any i386 system.
9681 The 64-bit environment sets int to 32 bits and long and pointer
9682 to 64 bits and generates code for AMD's x86-64 architecture.
9685 @opindex no-red-zone
9686 Do not use a so called red zone for x86-64 code. The red zone is mandated
9687 by the x86-64 ABI, it is a 128-byte area beyond the location of the
9688 stack pointer that will not be modified by signal or interrupt handlers
9689 and therefore can be used for temporary data without adjusting the stack
9690 pointer. The flag @option{-mno-red-zone} disables this red zone.
9692 @item -mcmodel=small
9693 @opindex mcmodel=small
9694 Generate code for the small code model: the program and its symbols must
9695 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
9696 Programs can be statically or dynamically linked. This is the default
9699 @item -mcmodel=kernel
9700 @opindex mcmodel=kernel
9701 Generate code for the kernel code model. The kernel runs in the
9702 negative 2 GB of the address space.
9703 This model has to be used for Linux kernel code.
9705 @item -mcmodel=medium
9706 @opindex mcmodel=medium
9707 Generate code for the medium model: The program is linked in the lower 2
9708 GB of the address space but symbols can be located anywhere in the
9709 address space. Programs can be statically or dynamically linked, but
9710 building of shared libraries are not supported with the medium model.
9712 @item -mcmodel=large
9713 @opindex mcmodel=large
9714 Generate code for the large model: This model makes no assumptions
9715 about addresses and sizes of sections. Currently GCC does not implement
9720 @subsection IA-64 Options
9721 @cindex IA-64 Options
9723 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9727 @opindex mbig-endian
9728 Generate code for a big endian target. This is the default for HP-UX@.
9730 @item -mlittle-endian
9731 @opindex mlittle-endian
9732 Generate code for a little endian target. This is the default for AIX5
9739 Generate (or don't) code for the GNU assembler. This is the default.
9740 @c Also, this is the default if the configure option @option{--with-gnu-as}
9747 Generate (or don't) code for the GNU linker. This is the default.
9748 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9753 Generate code that does not use a global pointer register. The result
9754 is not position independent code, and violates the IA-64 ABI@.
9756 @item -mvolatile-asm-stop
9757 @itemx -mno-volatile-asm-stop
9758 @opindex mvolatile-asm-stop
9759 @opindex mno-volatile-asm-stop
9760 Generate (or don't) a stop bit immediately before and after volatile asm
9763 @item -mregister-names
9764 @itemx -mno-register-names
9765 @opindex mregister-names
9766 @opindex mno-register-names
9767 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9768 the stacked registers. This may make assembler output more readable.
9774 Disable (or enable) optimizations that use the small data section. This may
9775 be useful for working around optimizer bugs.
9778 @opindex mconstant-gp
9779 Generate code that uses a single constant global pointer value. This is
9780 useful when compiling kernel code.
9784 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9785 This is useful when compiling firmware code.
9787 @item -minline-float-divide-min-latency
9788 @opindex minline-float-divide-min-latency
9789 Generate code for inline divides of floating point values
9790 using the minimum latency algorithm.
9792 @item -minline-float-divide-max-throughput
9793 @opindex minline-float-divide-max-throughput
9794 Generate code for inline divides of floating point values
9795 using the maximum throughput algorithm.
9797 @item -minline-int-divide-min-latency
9798 @opindex minline-int-divide-min-latency
9799 Generate code for inline divides of integer values
9800 using the minimum latency algorithm.
9802 @item -minline-int-divide-max-throughput
9803 @opindex minline-int-divide-max-throughput
9804 Generate code for inline divides of integer values
9805 using the maximum throughput algorithm.
9807 @item -minline-sqrt-min-latency
9808 @opindex minline-sqrt-min-latency
9809 Generate code for inline square roots
9810 using the minimum latency algorithm.
9812 @item -minline-sqrt-max-throughput
9813 @opindex minline-sqrt-max-throughput
9814 Generate code for inline square roots
9815 using the maximum throughput algorithm.
9817 @item -mno-dwarf2-asm
9819 @opindex mno-dwarf2-asm
9820 @opindex mdwarf2-asm
9821 Don't (or do) generate assembler code for the DWARF2 line number debugging
9822 info. This may be useful when not using the GNU assembler.
9824 @item -mearly-stop-bits
9825 @itemx -mno-early-stop-bits
9826 @opindex mearly-stop-bits
9827 @opindex mno-early-stop-bits
9828 Allow stop bits to be placed earlier than immediately preceding the
9829 instruction that triggered the stop bit. This can improve instruction
9830 scheduling, but does not always do so.
9832 @item -mfixed-range=@var{register-range}
9833 @opindex mfixed-range
9834 Generate code treating the given register range as fixed registers.
9835 A fixed register is one that the register allocator can not use. This is
9836 useful when compiling kernel code. A register range is specified as
9837 two registers separated by a dash. Multiple register ranges can be
9838 specified separated by a comma.
9840 @item -mtls-size=@var{tls-size}
9842 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
9845 @item -mtune=@var{cpu-type}
9847 Tune the instruction scheduling for a particular CPU, Valid values are
9848 itanium, itanium1, merced, itanium2, and mckinley.
9854 Add support for multithreading using the POSIX threads library. This
9855 option sets flags for both the preprocessor and linker. It does
9856 not affect the thread safety of object code produced by the compiler or
9857 that of libraries supplied with it. These are HP-UX specific flags.
9863 Generate code for a 32-bit or 64-bit environment.
9864 The 32-bit environment sets int, long and pointer to 32 bits.
9865 The 64-bit environment sets int to 32 bits and long and pointer
9866 to 64 bits. These are HP-UX specific flags.
9868 @item -mno-sched-br-data-spec
9869 @itemx -msched-br-data-spec
9870 @opindex -mno-sched-br-data-spec
9871 @opindex -msched-br-data-spec
9872 (Dis/En)able data speculative scheduling before reload.
9873 This will result in generation of the ld.a instructions and
9874 the corresponding check instructions (ld.c / chk.a).
9875 The default is 'disable'.
9877 @item -msched-ar-data-spec
9878 @itemx -mno-sched-ar-data-spec
9879 @opindex -msched-ar-data-spec
9880 @opindex -mno-sched-ar-data-spec
9881 (En/Dis)able data speculative scheduling after reload.
9882 This will result in generation of the ld.a instructions and
9883 the corresponding check instructions (ld.c / chk.a).
9884 The default is 'enable'.
9886 @item -mno-sched-control-spec
9887 @itemx -msched-control-spec
9888 @opindex -mno-sched-control-spec
9889 @opindex -msched-control-spec
9890 (Dis/En)able control speculative scheduling. This feature is
9891 available only during region scheduling (i.e. before reload).
9892 This will result in generation of the ld.s instructions and
9893 the corresponding check instructions chk.s .
9894 The default is 'disable'.
9896 @item -msched-br-in-data-spec
9897 @itemx -mno-sched-br-in-data-spec
9898 @opindex -msched-br-in-data-spec
9899 @opindex -mno-sched-br-in-data-spec
9900 (En/Dis)able speculative scheduling of the instructions that
9901 are dependent on the data speculative loads before reload.
9902 This is effective only with @option{-msched-br-data-spec} enabled.
9903 The default is 'enable'.
9905 @item -msched-ar-in-data-spec
9906 @itemx -mno-sched-ar-in-data-spec
9907 @opindex -msched-ar-in-data-spec
9908 @opindex -mno-sched-ar-in-data-spec
9909 (En/Dis)able speculative scheduling of the instructions that
9910 are dependent on the data speculative loads after reload.
9911 This is effective only with @option{-msched-ar-data-spec} enabled.
9912 The default is 'enable'.
9914 @item -msched-in-control-spec
9915 @itemx -mno-sched-in-control-spec
9916 @opindex -msched-in-control-spec
9917 @opindex -mno-sched-in-control-spec
9918 (En/Dis)able speculative scheduling of the instructions that
9919 are dependent on the control speculative loads.
9920 This is effective only with @option{-msched-control-spec} enabled.
9921 The default is 'enable'.
9924 @itemx -mno-sched-ldc
9925 @opindex -msched-ldc
9926 @opindex -mno-sched-ldc
9927 (En/Dis)able use of simple data speculation checks ld.c .
9928 If disabled, only chk.a instructions will be emitted to check
9929 data speculative loads.
9930 The default is 'enable'.
9932 @item -mno-sched-control-ldc
9933 @itemx -msched-control-ldc
9934 @opindex -mno-sched-control-ldc
9935 @opindex -msched-control-ldc
9936 (Dis/En)able use of ld.c instructions to check control speculative loads.
9937 If enabled, in case of control speculative load with no speculatively
9938 scheduled dependent instructions this load will be emitted as ld.sa and
9939 ld.c will be used to check it.
9940 The default is 'disable'.
9942 @item -mno-sched-spec-verbose
9943 @itemx -msched-spec-verbose
9944 @opindex -mno-sched-spec-verbose
9945 @opindex -msched-spec-verbose
9946 (Dis/En)able printing of the information about speculative motions.
9948 @item -mno-sched-prefer-non-data-spec-insns
9949 @itemx -msched-prefer-non-data-spec-insns
9950 @opindex -mno-sched-prefer-non-data-spec-insns
9951 @opindex -msched-prefer-non-data-spec-insns
9952 If enabled, data speculative instructions will be chosen for schedule
9953 only if there are no other choices at the moment. This will make
9954 the use of the data speculation much more conservative.
9955 The default is 'disable'.
9957 @item -mno-sched-prefer-non-control-spec-insns
9958 @itemx -msched-prefer-non-control-spec-insns
9959 @opindex -mno-sched-prefer-non-control-spec-insns
9960 @opindex -msched-prefer-non-control-spec-insns
9961 If enabled, control speculative instructions will be chosen for schedule
9962 only if there are no other choices at the moment. This will make
9963 the use of the control speculation much more conservative.
9964 The default is 'disable'.
9966 @item -mno-sched-count-spec-in-critical-path
9967 @itemx -msched-count-spec-in-critical-path
9968 @opindex -mno-sched-count-spec-in-critical-path
9969 @opindex -msched-count-spec-in-critical-path
9970 If enabled, speculative dependencies will be considered during
9971 computation of the instructions priorities. This will make the use of the
9972 speculation a bit more conservative.
9973 The default is 'disable'.
9978 @subsection M32C Options
9979 @cindex M32C options
9982 @item -mcpu=@var{name}
9984 Select the CPU for which code is generated. @var{name} may be one of
9985 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
9986 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
9991 Specifies that the program will be run on the simulator. This causes
9992 an alternate runtime library to be linked in which supports, for
9993 example, file I/O. You must not use this option when generating
9994 programs that will run on real hardware; you must provide your own
9995 runtime library for whatever I/O functions are needed.
9997 @item -memregs=@var{number}
9999 Specifies the number of memory-based pseudo-registers GCC will use
10000 during code generation. These pseudo-registers will be used like real
10001 registers, so there is a tradeoff between GCC's ability to fit the
10002 code into available registers, and the performance penalty of using
10003 memory instead of registers. Note that all modules in a program must
10004 be compiled with the same value for this option. Because of that, you
10005 must not use this option with the default runtime libraries gcc
10010 @node M32R/D Options
10011 @subsection M32R/D Options
10012 @cindex M32R/D options
10014 These @option{-m} options are defined for Renesas M32R/D architectures:
10019 Generate code for the M32R/2@.
10023 Generate code for the M32R/X@.
10027 Generate code for the M32R@. This is the default.
10029 @item -mmodel=small
10030 @opindex mmodel=small
10031 Assume all objects live in the lower 16MB of memory (so that their addresses
10032 can be loaded with the @code{ld24} instruction), and assume all subroutines
10033 are reachable with the @code{bl} instruction.
10034 This is the default.
10036 The addressability of a particular object can be set with the
10037 @code{model} attribute.
10039 @item -mmodel=medium
10040 @opindex mmodel=medium
10041 Assume objects may be anywhere in the 32-bit address space (the compiler
10042 will generate @code{seth/add3} instructions to load their addresses), and
10043 assume all subroutines are reachable with the @code{bl} instruction.
10045 @item -mmodel=large
10046 @opindex mmodel=large
10047 Assume objects may be anywhere in the 32-bit address space (the compiler
10048 will generate @code{seth/add3} instructions to load their addresses), and
10049 assume subroutines may not be reachable with the @code{bl} instruction
10050 (the compiler will generate the much slower @code{seth/add3/jl}
10051 instruction sequence).
10054 @opindex msdata=none
10055 Disable use of the small data area. Variables will be put into
10056 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
10057 @code{section} attribute has been specified).
10058 This is the default.
10060 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
10061 Objects may be explicitly put in the small data area with the
10062 @code{section} attribute using one of these sections.
10064 @item -msdata=sdata
10065 @opindex msdata=sdata
10066 Put small global and static data in the small data area, but do not
10067 generate special code to reference them.
10070 @opindex msdata=use
10071 Put small global and static data in the small data area, and generate
10072 special instructions to reference them.
10076 @cindex smaller data references
10077 Put global and static objects less than or equal to @var{num} bytes
10078 into the small data or bss sections instead of the normal data or bss
10079 sections. The default value of @var{num} is 8.
10080 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
10081 for this option to have any effect.
10083 All modules should be compiled with the same @option{-G @var{num}} value.
10084 Compiling with different values of @var{num} may or may not work; if it
10085 doesn't the linker will give an error message---incorrect code will not be
10090 Makes the M32R specific code in the compiler display some statistics
10091 that might help in debugging programs.
10093 @item -malign-loops
10094 @opindex malign-loops
10095 Align all loops to a 32-byte boundary.
10097 @item -mno-align-loops
10098 @opindex mno-align-loops
10099 Do not enforce a 32-byte alignment for loops. This is the default.
10101 @item -missue-rate=@var{number}
10102 @opindex missue-rate=@var{number}
10103 Issue @var{number} instructions per cycle. @var{number} can only be 1
10106 @item -mbranch-cost=@var{number}
10107 @opindex mbranch-cost=@var{number}
10108 @var{number} can only be 1 or 2. If it is 1 then branches will be
10109 preferred over conditional code, if it is 2, then the opposite will
10112 @item -mflush-trap=@var{number}
10113 @opindex mflush-trap=@var{number}
10114 Specifies the trap number to use to flush the cache. The default is
10115 12. Valid numbers are between 0 and 15 inclusive.
10117 @item -mno-flush-trap
10118 @opindex mno-flush-trap
10119 Specifies that the cache cannot be flushed by using a trap.
10121 @item -mflush-func=@var{name}
10122 @opindex mflush-func=@var{name}
10123 Specifies the name of the operating system function to call to flush
10124 the cache. The default is @emph{_flush_cache}, but a function call
10125 will only be used if a trap is not available.
10127 @item -mno-flush-func
10128 @opindex mno-flush-func
10129 Indicates that there is no OS function for flushing the cache.
10133 @node M680x0 Options
10134 @subsection M680x0 Options
10135 @cindex M680x0 options
10137 These are the @samp{-m} options defined for the 68000 series. The default
10138 values for these options depends on which style of 68000 was selected when
10139 the compiler was configured; the defaults for the most common choices are
10147 Generate output for a 68000. This is the default
10148 when the compiler is configured for 68000-based systems.
10150 Use this option for microcontrollers with a 68000 or EC000 core,
10151 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
10157 Generate output for a 68020. This is the default
10158 when the compiler is configured for 68020-based systems.
10162 Generate output containing 68881 instructions for floating point.
10163 This is the default for most 68020 systems unless @option{--nfp} was
10164 specified when the compiler was configured.
10168 Generate output for a 68030. This is the default when the compiler is
10169 configured for 68030-based systems.
10173 Generate output for a 68040. This is the default when the compiler is
10174 configured for 68040-based systems.
10176 This option inhibits the use of 68881/68882 instructions that have to be
10177 emulated by software on the 68040. Use this option if your 68040 does not
10178 have code to emulate those instructions.
10182 Generate output for a 68060. This is the default when the compiler is
10183 configured for 68060-based systems.
10185 This option inhibits the use of 68020 and 68881/68882 instructions that
10186 have to be emulated by software on the 68060. Use this option if your 68060
10187 does not have code to emulate those instructions.
10191 Generate output for a CPU32. This is the default
10192 when the compiler is configured for CPU32-based systems.
10194 Use this option for microcontrollers with a
10195 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
10196 68336, 68340, 68341, 68349 and 68360.
10200 Generate output for a 520X ``coldfire'' family cpu. This is the default
10201 when the compiler is configured for 520X-based systems.
10203 Use this option for microcontroller with a 5200 core, including
10204 the MCF5202, MCF5203, MCF5204 and MCF5202.
10208 Generate output for a ColdFire V4e family cpu (e.g.@: 547x/548x).
10209 This includes use of hardware floating point instructions.
10213 Generate output for a 68040, without using any of the new instructions.
10214 This results in code which can run relatively efficiently on either a
10215 68020/68881 or a 68030 or a 68040. The generated code does use the
10216 68881 instructions that are emulated on the 68040.
10220 Generate output for a 68060, without using any of the new instructions.
10221 This results in code which can run relatively efficiently on either a
10222 68020/68881 or a 68030 or a 68040. The generated code does use the
10223 68881 instructions that are emulated on the 68060.
10226 @opindex msoft-float
10227 Generate output containing library calls for floating point.
10228 @strong{Warning:} the requisite libraries are not available for all m68k
10229 targets. Normally the facilities of the machine's usual C compiler are
10230 used, but this can't be done directly in cross-compilation. You must
10231 make your own arrangements to provide suitable library functions for
10232 cross-compilation. The embedded targets @samp{m68k-*-aout} and
10233 @samp{m68k-*-coff} do provide software floating point support.
10237 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10238 Additionally, parameters passed on the stack are also aligned to a
10239 16-bit boundary even on targets whose API mandates promotion to 32-bit.
10242 @opindex mnobitfield
10243 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
10244 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
10248 Do use the bit-field instructions. The @option{-m68020} option implies
10249 @option{-mbitfield}. This is the default if you use a configuration
10250 designed for a 68020.
10254 Use a different function-calling convention, in which functions
10255 that take a fixed number of arguments return with the @code{rtd}
10256 instruction, which pops their arguments while returning. This
10257 saves one instruction in the caller since there is no need to pop
10258 the arguments there.
10260 This calling convention is incompatible with the one normally
10261 used on Unix, so you cannot use it if you need to call libraries
10262 compiled with the Unix compiler.
10264 Also, you must provide function prototypes for all functions that
10265 take variable numbers of arguments (including @code{printf});
10266 otherwise incorrect code will be generated for calls to those
10269 In addition, seriously incorrect code will result if you call a
10270 function with too many arguments. (Normally, extra arguments are
10271 harmlessly ignored.)
10273 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
10274 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
10277 @itemx -mno-align-int
10278 @opindex malign-int
10279 @opindex mno-align-int
10280 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
10281 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
10282 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
10283 Aligning variables on 32-bit boundaries produces code that runs somewhat
10284 faster on processors with 32-bit busses at the expense of more memory.
10286 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
10287 align structures containing the above types differently than
10288 most published application binary interface specifications for the m68k.
10292 Use the pc-relative addressing mode of the 68000 directly, instead of
10293 using a global offset table. At present, this option implies @option{-fpic},
10294 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
10295 not presently supported with @option{-mpcrel}, though this could be supported for
10296 68020 and higher processors.
10298 @item -mno-strict-align
10299 @itemx -mstrict-align
10300 @opindex mno-strict-align
10301 @opindex mstrict-align
10302 Do not (do) assume that unaligned memory references will be handled by
10306 Generate code that allows the data segment to be located in a different
10307 area of memory from the text segment. This allows for execute in place in
10308 an environment without virtual memory management. This option implies
10311 @item -mno-sep-data
10312 Generate code that assumes that the data segment follows the text segment.
10313 This is the default.
10315 @item -mid-shared-library
10316 Generate code that supports shared libraries via the library ID method.
10317 This allows for execute in place and shared libraries in an environment
10318 without virtual memory management. This option implies @option{-fPIC}.
10320 @item -mno-id-shared-library
10321 Generate code that doesn't assume ID based shared libraries are being used.
10322 This is the default.
10324 @item -mshared-library-id=n
10325 Specified the identification number of the ID based shared library being
10326 compiled. Specifying a value of 0 will generate more compact code, specifying
10327 other values will force the allocation of that number to the current
10328 library but is no more space or time efficient than omitting this option.
10332 @node M68hc1x Options
10333 @subsection M68hc1x Options
10334 @cindex M68hc1x options
10336 These are the @samp{-m} options defined for the 68hc11 and 68hc12
10337 microcontrollers. The default values for these options depends on
10338 which style of microcontroller was selected when the compiler was configured;
10339 the defaults for the most common choices are given below.
10346 Generate output for a 68HC11. This is the default
10347 when the compiler is configured for 68HC11-based systems.
10353 Generate output for a 68HC12. This is the default
10354 when the compiler is configured for 68HC12-based systems.
10360 Generate output for a 68HCS12.
10362 @item -mauto-incdec
10363 @opindex mauto-incdec
10364 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
10371 Enable the use of 68HC12 min and max instructions.
10374 @itemx -mno-long-calls
10375 @opindex mlong-calls
10376 @opindex mno-long-calls
10377 Treat all calls as being far away (near). If calls are assumed to be
10378 far away, the compiler will use the @code{call} instruction to
10379 call a function and the @code{rtc} instruction for returning.
10383 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10385 @item -msoft-reg-count=@var{count}
10386 @opindex msoft-reg-count
10387 Specify the number of pseudo-soft registers which are used for the
10388 code generation. The maximum number is 32. Using more pseudo-soft
10389 register may or may not result in better code depending on the program.
10390 The default is 4 for 68HC11 and 2 for 68HC12.
10394 @node MCore Options
10395 @subsection MCore Options
10396 @cindex MCore options
10398 These are the @samp{-m} options defined for the Motorola M*Core
10404 @itemx -mno-hardlit
10406 @opindex mno-hardlit
10407 Inline constants into the code stream if it can be done in two
10408 instructions or less.
10414 Use the divide instruction. (Enabled by default).
10416 @item -mrelax-immediate
10417 @itemx -mno-relax-immediate
10418 @opindex mrelax-immediate
10419 @opindex mno-relax-immediate
10420 Allow arbitrary sized immediates in bit operations.
10422 @item -mwide-bitfields
10423 @itemx -mno-wide-bitfields
10424 @opindex mwide-bitfields
10425 @opindex mno-wide-bitfields
10426 Always treat bit-fields as int-sized.
10428 @item -m4byte-functions
10429 @itemx -mno-4byte-functions
10430 @opindex m4byte-functions
10431 @opindex mno-4byte-functions
10432 Force all functions to be aligned to a four byte boundary.
10434 @item -mcallgraph-data
10435 @itemx -mno-callgraph-data
10436 @opindex mcallgraph-data
10437 @opindex mno-callgraph-data
10438 Emit callgraph information.
10441 @itemx -mno-slow-bytes
10442 @opindex mslow-bytes
10443 @opindex mno-slow-bytes
10444 Prefer word access when reading byte quantities.
10446 @item -mlittle-endian
10447 @itemx -mbig-endian
10448 @opindex mlittle-endian
10449 @opindex mbig-endian
10450 Generate code for a little endian target.
10456 Generate code for the 210 processor.
10460 @subsection MIPS Options
10461 @cindex MIPS options
10467 Generate big-endian code.
10471 Generate little-endian code. This is the default for @samp{mips*el-*-*}
10474 @item -march=@var{arch}
10476 Generate code that will run on @var{arch}, which can be the name of a
10477 generic MIPS ISA, or the name of a particular processor.
10479 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
10480 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
10481 The processor names are:
10482 @samp{4kc}, @samp{4km}, @samp{4kp},
10483 @samp{5kc}, @samp{5kf},
10485 @samp{24k}, @samp{24kc}, @samp{24kf}, @samp{24kx},
10488 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
10489 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
10490 @samp{rm7000}, @samp{rm9000},
10493 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
10494 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
10495 The special value @samp{from-abi} selects the
10496 most compatible architecture for the selected ABI (that is,
10497 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
10499 In processor names, a final @samp{000} can be abbreviated as @samp{k}
10500 (for example, @samp{-march=r2k}). Prefixes are optional, and
10501 @samp{vr} may be written @samp{r}.
10503 GCC defines two macros based on the value of this option. The first
10504 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
10505 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
10506 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
10507 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
10508 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
10510 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
10511 above. In other words, it will have the full prefix and will not
10512 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
10513 the macro names the resolved architecture (either @samp{"mips1"} or
10514 @samp{"mips3"}). It names the default architecture when no
10515 @option{-march} option is given.
10517 @item -mtune=@var{arch}
10519 Optimize for @var{arch}. Among other things, this option controls
10520 the way instructions are scheduled, and the perceived cost of arithmetic
10521 operations. The list of @var{arch} values is the same as for
10524 When this option is not used, GCC will optimize for the processor
10525 specified by @option{-march}. By using @option{-march} and
10526 @option{-mtune} together, it is possible to generate code that will
10527 run on a family of processors, but optimize the code for one
10528 particular member of that family.
10530 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
10531 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
10532 @samp{-march} ones described above.
10536 Equivalent to @samp{-march=mips1}.
10540 Equivalent to @samp{-march=mips2}.
10544 Equivalent to @samp{-march=mips3}.
10548 Equivalent to @samp{-march=mips4}.
10552 Equivalent to @samp{-march=mips32}.
10556 Equivalent to @samp{-march=mips32r2}.
10560 Equivalent to @samp{-march=mips64}.
10565 @opindex mno-mips16
10566 Generate (do not generate) MIPS16 code. If GCC is targetting a
10567 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
10579 Generate code for the given ABI@.
10581 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
10582 generates 64-bit code when you select a 64-bit architecture, but you
10583 can use @option{-mgp32} to get 32-bit code instead.
10585 For information about the O64 ABI, see
10586 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
10589 @itemx -mno-abicalls
10591 @opindex mno-abicalls
10592 Generate (do not generate) code that is suitable for SVR4-style
10593 dynamic objects. @option{-mabicalls} is the default for SVR4-based
10598 Generate (do not generate) code that is fully position-independent,
10599 and that can therefore be linked into shared libraries. This option
10600 only affects @option{-mabicalls}.
10602 All @option{-mabicalls} code has traditionally been position-independent,
10603 regardless of options like @option{-fPIC} and @option{-fpic}. However,
10604 as an extension, the GNU toolchain allows executables to use absolute
10605 accesses for locally-binding symbols. It can also use shorter GP
10606 initialization sequences and generate direct calls to locally-defined
10607 functions. This mode is selected by @option{-mno-shared}.
10609 @option{-mno-shared} depends on binutils 2.16 or higher and generates
10610 objects that can only be linked by the GNU linker. However, the option
10611 does not affect the ABI of the final executable; it only affects the ABI
10612 of relocatable objects. Using @option{-mno-shared} will generally make
10613 executables both smaller and quicker.
10615 @option{-mshared} is the default.
10621 Lift (do not lift) the usual restrictions on the size of the global
10624 GCC normally uses a single instruction to load values from the GOT@.
10625 While this is relatively efficient, it will only work if the GOT
10626 is smaller than about 64k. Anything larger will cause the linker
10627 to report an error such as:
10629 @cindex relocation truncated to fit (MIPS)
10631 relocation truncated to fit: R_MIPS_GOT16 foobar
10634 If this happens, you should recompile your code with @option{-mxgot}.
10635 It should then work with very large GOTs, although it will also be
10636 less efficient, since it will take three instructions to fetch the
10637 value of a global symbol.
10639 Note that some linkers can create multiple GOTs. If you have such a
10640 linker, you should only need to use @option{-mxgot} when a single object
10641 file accesses more than 64k's worth of GOT entries. Very few do.
10643 These options have no effect unless GCC is generating position
10648 Assume that general-purpose registers are 32 bits wide.
10652 Assume that general-purpose registers are 64 bits wide.
10656 Assume that floating-point registers are 32 bits wide.
10660 Assume that floating-point registers are 64 bits wide.
10663 @opindex mhard-float
10664 Use floating-point coprocessor instructions.
10667 @opindex msoft-float
10668 Do not use floating-point coprocessor instructions. Implement
10669 floating-point calculations using library calls instead.
10671 @item -msingle-float
10672 @opindex msingle-float
10673 Assume that the floating-point coprocessor only supports single-precision
10676 @itemx -mdouble-float
10677 @opindex mdouble-float
10678 Assume that the floating-point coprocessor supports double-precision
10679 operations. This is the default.
10685 Use (do not use) the MIPS DSP ASE. @xref{MIPS DSP Built-in Functions}.
10687 @itemx -mpaired-single
10688 @itemx -mno-paired-single
10689 @opindex mpaired-single
10690 @opindex mno-paired-single
10691 Use (do not use) paired-single floating-point instructions.
10692 @xref{MIPS Paired-Single Support}. This option can only be used
10693 when generating 64-bit code and requires hardware floating-point
10694 support to be enabled.
10699 @opindex mno-mips3d
10700 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
10701 The option @option{-mips3d} implies @option{-mpaired-single}.
10705 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
10706 an explanation of the default and the way that the pointer size is
10711 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
10713 The default size of @code{int}s, @code{long}s and pointers depends on
10714 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
10715 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
10716 32-bit @code{long}s. Pointers are the same size as @code{long}s,
10717 or the same size as integer registers, whichever is smaller.
10723 Assume (do not assume) that all symbols have 32-bit values, regardless
10724 of the selected ABI@. This option is useful in combination with
10725 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
10726 to generate shorter and faster references to symbolic addresses.
10730 @cindex smaller data references (MIPS)
10731 @cindex gp-relative references (MIPS)
10732 Put global and static items less than or equal to @var{num} bytes into
10733 the small data or bss section instead of the normal data or bss section.
10734 This allows the data to be accessed using a single instruction.
10736 All modules should be compiled with the same @option{-G @var{num}}
10739 @item -membedded-data
10740 @itemx -mno-embedded-data
10741 @opindex membedded-data
10742 @opindex mno-embedded-data
10743 Allocate variables to the read-only data section first if possible, then
10744 next in the small data section if possible, otherwise in data. This gives
10745 slightly slower code than the default, but reduces the amount of RAM required
10746 when executing, and thus may be preferred for some embedded systems.
10748 @item -muninit-const-in-rodata
10749 @itemx -mno-uninit-const-in-rodata
10750 @opindex muninit-const-in-rodata
10751 @opindex mno-uninit-const-in-rodata
10752 Put uninitialized @code{const} variables in the read-only data section.
10753 This option is only meaningful in conjunction with @option{-membedded-data}.
10755 @item -msplit-addresses
10756 @itemx -mno-split-addresses
10757 @opindex msplit-addresses
10758 @opindex mno-split-addresses
10759 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
10760 relocation operators. This option has been superseded by
10761 @option{-mexplicit-relocs} but is retained for backwards compatibility.
10763 @item -mexplicit-relocs
10764 @itemx -mno-explicit-relocs
10765 @opindex mexplicit-relocs
10766 @opindex mno-explicit-relocs
10767 Use (do not use) assembler relocation operators when dealing with symbolic
10768 addresses. The alternative, selected by @option{-mno-explicit-relocs},
10769 is to use assembler macros instead.
10771 @option{-mexplicit-relocs} is the default if GCC was configured
10772 to use an assembler that supports relocation operators.
10774 @item -mcheck-zero-division
10775 @itemx -mno-check-zero-division
10776 @opindex mcheck-zero-division
10777 @opindex mno-check-zero-division
10778 Trap (do not trap) on integer division by zero. The default is
10779 @option{-mcheck-zero-division}.
10781 @item -mdivide-traps
10782 @itemx -mdivide-breaks
10783 @opindex mdivide-traps
10784 @opindex mdivide-breaks
10785 MIPS systems check for division by zero by generating either a
10786 conditional trap or a break instruction. Using traps results in
10787 smaller code, but is only supported on MIPS II and later. Also, some
10788 versions of the Linux kernel have a bug that prevents trap from
10789 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
10790 allow conditional traps on architectures that support them and
10791 @option{-mdivide-breaks} to force the use of breaks.
10793 The default is usually @option{-mdivide-traps}, but this can be
10794 overridden at configure time using @option{--with-divide=breaks}.
10795 Divide-by-zero checks can be completely disabled using
10796 @option{-mno-check-zero-division}.
10801 @opindex mno-memcpy
10802 Force (do not force) the use of @code{memcpy()} for non-trivial block
10803 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
10804 most constant-sized copies.
10807 @itemx -mno-long-calls
10808 @opindex mlong-calls
10809 @opindex mno-long-calls
10810 Disable (do not disable) use of the @code{jal} instruction. Calling
10811 functions using @code{jal} is more efficient but requires the caller
10812 and callee to be in the same 256 megabyte segment.
10814 This option has no effect on abicalls code. The default is
10815 @option{-mno-long-calls}.
10821 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
10822 instructions, as provided by the R4650 ISA@.
10825 @itemx -mno-fused-madd
10826 @opindex mfused-madd
10827 @opindex mno-fused-madd
10828 Enable (disable) use of the floating point multiply-accumulate
10829 instructions, when they are available. The default is
10830 @option{-mfused-madd}.
10832 When multiply-accumulate instructions are used, the intermediate
10833 product is calculated to infinite precision and is not subject to
10834 the FCSR Flush to Zero bit. This may be undesirable in some
10839 Tell the MIPS assembler to not run its preprocessor over user
10840 assembler files (with a @samp{.s} suffix) when assembling them.
10843 @itemx -mno-fix-r4000
10844 @opindex mfix-r4000
10845 @opindex mno-fix-r4000
10846 Work around certain R4000 CPU errata:
10849 A double-word or a variable shift may give an incorrect result if executed
10850 immediately after starting an integer division.
10852 A double-word or a variable shift may give an incorrect result if executed
10853 while an integer multiplication is in progress.
10855 An integer division may give an incorrect result if started in a delay slot
10856 of a taken branch or a jump.
10860 @itemx -mno-fix-r4400
10861 @opindex mfix-r4400
10862 @opindex mno-fix-r4400
10863 Work around certain R4400 CPU errata:
10866 A double-word or a variable shift may give an incorrect result if executed
10867 immediately after starting an integer division.
10871 @itemx -mno-fix-vr4120
10872 @opindex mfix-vr4120
10873 Work around certain VR4120 errata:
10876 @code{dmultu} does not always produce the correct result.
10878 @code{div} and @code{ddiv} do not always produce the correct result if one
10879 of the operands is negative.
10881 The workarounds for the division errata rely on special functions in
10882 @file{libgcc.a}. At present, these functions are only provided by
10883 the @code{mips64vr*-elf} configurations.
10885 Other VR4120 errata require a nop to be inserted between certain pairs of
10886 instructions. These errata are handled by the assembler, not by GCC itself.
10889 @opindex mfix-vr4130
10890 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
10891 workarounds are implemented by the assembler rather than by GCC,
10892 although GCC will avoid using @code{mflo} and @code{mfhi} if the
10893 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
10894 instructions are available instead.
10897 @itemx -mno-fix-sb1
10899 Work around certain SB-1 CPU core errata.
10900 (This flag currently works around the SB-1 revision 2
10901 ``F1'' and ``F2'' floating point errata.)
10903 @item -mflush-func=@var{func}
10904 @itemx -mno-flush-func
10905 @opindex mflush-func
10906 Specifies the function to call to flush the I and D caches, or to not
10907 call any such function. If called, the function must take the same
10908 arguments as the common @code{_flush_func()}, that is, the address of the
10909 memory range for which the cache is being flushed, the size of the
10910 memory range, and the number 3 (to flush both caches). The default
10911 depends on the target GCC was configured for, but commonly is either
10912 @samp{_flush_func} or @samp{__cpu_flush}.
10914 @item -mbranch-likely
10915 @itemx -mno-branch-likely
10916 @opindex mbranch-likely
10917 @opindex mno-branch-likely
10918 Enable or disable use of Branch Likely instructions, regardless of the
10919 default for the selected architecture. By default, Branch Likely
10920 instructions may be generated if they are supported by the selected
10921 architecture. An exception is for the MIPS32 and MIPS64 architectures
10922 and processors which implement those architectures; for those, Branch
10923 Likely instructions will not be generated by default because the MIPS32
10924 and MIPS64 architectures specifically deprecate their use.
10926 @item -mfp-exceptions
10927 @itemx -mno-fp-exceptions
10928 @opindex mfp-exceptions
10929 Specifies whether FP exceptions are enabled. This affects how we schedule
10930 FP instructions for some processors. The default is that FP exceptions are
10933 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
10934 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
10937 @item -mvr4130-align
10938 @itemx -mno-vr4130-align
10939 @opindex mvr4130-align
10940 The VR4130 pipeline is two-way superscalar, but can only issue two
10941 instructions together if the first one is 8-byte aligned. When this
10942 option is enabled, GCC will align pairs of instructions that it
10943 thinks should execute in parallel.
10945 This option only has an effect when optimizing for the VR4130.
10946 It normally makes code faster, but at the expense of making it bigger.
10947 It is enabled by default at optimization level @option{-O3}.
10951 @subsection MMIX Options
10952 @cindex MMIX Options
10954 These options are defined for the MMIX:
10958 @itemx -mno-libfuncs
10960 @opindex mno-libfuncs
10961 Specify that intrinsic library functions are being compiled, passing all
10962 values in registers, no matter the size.
10965 @itemx -mno-epsilon
10967 @opindex mno-epsilon
10968 Generate floating-point comparison instructions that compare with respect
10969 to the @code{rE} epsilon register.
10971 @item -mabi=mmixware
10973 @opindex mabi-mmixware
10975 Generate code that passes function parameters and return values that (in
10976 the called function) are seen as registers @code{$0} and up, as opposed to
10977 the GNU ABI which uses global registers @code{$231} and up.
10979 @item -mzero-extend
10980 @itemx -mno-zero-extend
10981 @opindex mzero-extend
10982 @opindex mno-zero-extend
10983 When reading data from memory in sizes shorter than 64 bits, use (do not
10984 use) zero-extending load instructions by default, rather than
10985 sign-extending ones.
10988 @itemx -mno-knuthdiv
10990 @opindex mno-knuthdiv
10991 Make the result of a division yielding a remainder have the same sign as
10992 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10993 remainder follows the sign of the dividend. Both methods are
10994 arithmetically valid, the latter being almost exclusively used.
10996 @item -mtoplevel-symbols
10997 @itemx -mno-toplevel-symbols
10998 @opindex mtoplevel-symbols
10999 @opindex mno-toplevel-symbols
11000 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
11001 code can be used with the @code{PREFIX} assembly directive.
11005 Generate an executable in the ELF format, rather than the default
11006 @samp{mmo} format used by the @command{mmix} simulator.
11008 @item -mbranch-predict
11009 @itemx -mno-branch-predict
11010 @opindex mbranch-predict
11011 @opindex mno-branch-predict
11012 Use (do not use) the probable-branch instructions, when static branch
11013 prediction indicates a probable branch.
11015 @item -mbase-addresses
11016 @itemx -mno-base-addresses
11017 @opindex mbase-addresses
11018 @opindex mno-base-addresses
11019 Generate (do not generate) code that uses @emph{base addresses}. Using a
11020 base address automatically generates a request (handled by the assembler
11021 and the linker) for a constant to be set up in a global register. The
11022 register is used for one or more base address requests within the range 0
11023 to 255 from the value held in the register. The generally leads to short
11024 and fast code, but the number of different data items that can be
11025 addressed is limited. This means that a program that uses lots of static
11026 data may require @option{-mno-base-addresses}.
11028 @item -msingle-exit
11029 @itemx -mno-single-exit
11030 @opindex msingle-exit
11031 @opindex mno-single-exit
11032 Force (do not force) generated code to have a single exit point in each
11036 @node MN10300 Options
11037 @subsection MN10300 Options
11038 @cindex MN10300 options
11040 These @option{-m} options are defined for Matsushita MN10300 architectures:
11045 Generate code to avoid bugs in the multiply instructions for the MN10300
11046 processors. This is the default.
11048 @item -mno-mult-bug
11049 @opindex mno-mult-bug
11050 Do not generate code to avoid bugs in the multiply instructions for the
11051 MN10300 processors.
11055 Generate code which uses features specific to the AM33 processor.
11059 Do not generate code which uses features specific to the AM33 processor. This
11062 @item -mreturn-pointer-on-d0
11063 @opindex mreturn-pointer-on-d0
11064 When generating a function which returns a pointer, return the pointer
11065 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
11066 only in a0, and attempts to call such functions without a prototype
11067 would result in errors. Note that this option is on by default; use
11068 @option{-mno-return-pointer-on-d0} to disable it.
11072 Do not link in the C run-time initialization object file.
11076 Indicate to the linker that it should perform a relaxation optimization pass
11077 to shorten branches, calls and absolute memory addresses. This option only
11078 has an effect when used on the command line for the final link step.
11080 This option makes symbolic debugging impossible.
11084 @subsection MT Options
11087 These @option{-m} options are defined for Morpho MT architectures:
11091 @item -march=@var{cpu-type}
11093 Generate code that will run on @var{cpu-type}, which is the name of a system
11094 representing a certain processor type. Possible values for
11095 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
11096 @samp{ms1-16-003} and @samp{ms2}.
11098 When this option is not used, the default is @option{-march=ms1-16-002}.
11102 Use byte loads and stores when generating code.
11106 Do not use byte loads and stores when generating code.
11110 Use simulator runtime
11114 Do not link in the C run-time initialization object file
11115 @file{crti.o}. Other run-time initialization and termination files
11116 such as @file{startup.o} and @file{exit.o} are still included on the
11117 linker command line.
11121 @node PDP-11 Options
11122 @subsection PDP-11 Options
11123 @cindex PDP-11 Options
11125 These options are defined for the PDP-11:
11130 Use hardware FPP floating point. This is the default. (FIS floating
11131 point on the PDP-11/40 is not supported.)
11134 @opindex msoft-float
11135 Do not use hardware floating point.
11139 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
11143 Return floating-point results in memory. This is the default.
11147 Generate code for a PDP-11/40.
11151 Generate code for a PDP-11/45. This is the default.
11155 Generate code for a PDP-11/10.
11157 @item -mbcopy-builtin
11158 @opindex bcopy-builtin
11159 Use inline @code{movmemhi} patterns for copying memory. This is the
11164 Do not use inline @code{movmemhi} patterns for copying memory.
11170 Use 16-bit @code{int}. This is the default.
11176 Use 32-bit @code{int}.
11179 @itemx -mno-float32
11181 @opindex mno-float32
11182 Use 64-bit @code{float}. This is the default.
11185 @itemx -mno-float64
11187 @opindex mno-float64
11188 Use 32-bit @code{float}.
11192 Use @code{abshi2} pattern. This is the default.
11196 Do not use @code{abshi2} pattern.
11198 @item -mbranch-expensive
11199 @opindex mbranch-expensive
11200 Pretend that branches are expensive. This is for experimenting with
11201 code generation only.
11203 @item -mbranch-cheap
11204 @opindex mbranch-cheap
11205 Do not pretend that branches are expensive. This is the default.
11209 Generate code for a system with split I&D@.
11213 Generate code for a system without split I&D@. This is the default.
11217 Use Unix assembler syntax. This is the default when configured for
11218 @samp{pdp11-*-bsd}.
11222 Use DEC assembler syntax. This is the default when configured for any
11223 PDP-11 target other than @samp{pdp11-*-bsd}.
11226 @node PowerPC Options
11227 @subsection PowerPC Options
11228 @cindex PowerPC options
11230 These are listed under @xref{RS/6000 and PowerPC Options}.
11232 @node RS/6000 and PowerPC Options
11233 @subsection IBM RS/6000 and PowerPC Options
11234 @cindex RS/6000 and PowerPC Options
11235 @cindex IBM RS/6000 and PowerPC Options
11237 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
11244 @itemx -mno-powerpc
11245 @itemx -mpowerpc-gpopt
11246 @itemx -mno-powerpc-gpopt
11247 @itemx -mpowerpc-gfxopt
11248 @itemx -mno-powerpc-gfxopt
11250 @itemx -mno-powerpc64
11254 @itemx -mno-popcntb
11262 @opindex mno-power2
11264 @opindex mno-powerpc
11265 @opindex mpowerpc-gpopt
11266 @opindex mno-powerpc-gpopt
11267 @opindex mpowerpc-gfxopt
11268 @opindex mno-powerpc-gfxopt
11269 @opindex mpowerpc64
11270 @opindex mno-powerpc64
11274 @opindex mno-popcntb
11278 @opindex mno-mfpgpr
11279 GCC supports two related instruction set architectures for the
11280 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
11281 instructions supported by the @samp{rios} chip set used in the original
11282 RS/6000 systems and the @dfn{PowerPC} instruction set is the
11283 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
11284 the IBM 4xx, 6xx, and follow-on microprocessors.
11286 Neither architecture is a subset of the other. However there is a
11287 large common subset of instructions supported by both. An MQ
11288 register is included in processors supporting the POWER architecture.
11290 You use these options to specify which instructions are available on the
11291 processor you are using. The default value of these options is
11292 determined when configuring GCC@. Specifying the
11293 @option{-mcpu=@var{cpu_type}} overrides the specification of these
11294 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
11295 rather than the options listed above.
11297 The @option{-mpower} option allows GCC to generate instructions that
11298 are found only in the POWER architecture and to use the MQ register.
11299 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
11300 to generate instructions that are present in the POWER2 architecture but
11301 not the original POWER architecture.
11303 The @option{-mpowerpc} option allows GCC to generate instructions that
11304 are found only in the 32-bit subset of the PowerPC architecture.
11305 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
11306 GCC to use the optional PowerPC architecture instructions in the
11307 General Purpose group, including floating-point square root. Specifying
11308 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
11309 use the optional PowerPC architecture instructions in the Graphics
11310 group, including floating-point select.
11312 The @option{-mmfcrf} option allows GCC to generate the move from
11313 condition register field instruction implemented on the POWER4
11314 processor and other processors that support the PowerPC V2.01
11316 The @option{-mpopcntb} option allows GCC to generate the popcount and
11317 double precision FP reciprocal estimate instruction implemented on the
11318 POWER5 processor and other processors that support the PowerPC V2.02
11320 The @option{-mfprnd} option allows GCC to generate the FP round to
11321 integer instructions implemented on the POWER5+ processor and other
11322 processors that support the PowerPC V2.03 architecture.
11323 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
11324 general purpose register instructions implemented on the POWER6X
11325 processor and other processors that support the extended PowerPC V2.05
11328 The @option{-mpowerpc64} option allows GCC to generate the additional
11329 64-bit instructions that are found in the full PowerPC64 architecture
11330 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
11331 @option{-mno-powerpc64}.
11333 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
11334 will use only the instructions in the common subset of both
11335 architectures plus some special AIX common-mode calls, and will not use
11336 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
11337 permits GCC to use any instruction from either architecture and to
11338 allow use of the MQ register; specify this for the Motorola MPC601.
11340 @item -mnew-mnemonics
11341 @itemx -mold-mnemonics
11342 @opindex mnew-mnemonics
11343 @opindex mold-mnemonics
11344 Select which mnemonics to use in the generated assembler code. With
11345 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
11346 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
11347 assembler mnemonics defined for the POWER architecture. Instructions
11348 defined in only one architecture have only one mnemonic; GCC uses that
11349 mnemonic irrespective of which of these options is specified.
11351 GCC defaults to the mnemonics appropriate for the architecture in
11352 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
11353 value of these option. Unless you are building a cross-compiler, you
11354 should normally not specify either @option{-mnew-mnemonics} or
11355 @option{-mold-mnemonics}, but should instead accept the default.
11357 @item -mcpu=@var{cpu_type}
11359 Set architecture type, register usage, choice of mnemonics, and
11360 instruction scheduling parameters for machine type @var{cpu_type}.
11361 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
11362 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
11363 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
11364 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
11365 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
11366 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
11367 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
11368 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6},
11369 @samp{power6x}, @samp{common}, @samp{powerpc}, @samp{powerpc64},
11370 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
11372 @option{-mcpu=common} selects a completely generic processor. Code
11373 generated under this option will run on any POWER or PowerPC processor.
11374 GCC will use only the instructions in the common subset of both
11375 architectures, and will not use the MQ register. GCC assumes a generic
11376 processor model for scheduling purposes.
11378 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
11379 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
11380 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
11381 types, with an appropriate, generic processor model assumed for
11382 scheduling purposes.
11384 The other options specify a specific processor. Code generated under
11385 those options will run best on that processor, and may not run at all on
11388 The @option{-mcpu} options automatically enable or disable the
11389 following options: @option{-maltivec}, @option{-mfprnd},
11390 @option{-mhard-float}, @option{-mmfcrf}, @option{-mmultiple},
11391 @option{-mnew-mnemonics}, @option{-mpopcntb}, @option{-mpower},
11392 @option{-mpower2}, @option{-mpowerpc64}, @option{-mpowerpc-gpopt},
11393 @option{-mpowerpc-gfxopt}, @option{-mstring}, @option{-mmulhw},
11394 @option{-mdlmzb}, @option{-mmfpgpr}.
11395 The particular options set for any particular CPU will vary between
11396 compiler versions, depending on what setting seems to produce optimal
11397 code for that CPU; it doesn't necessarily reflect the actual hardware's
11398 capabilities. If you wish to set an individual option to a particular
11399 value, you may specify it after the @option{-mcpu} option, like
11400 @samp{-mcpu=970 -mno-altivec}.
11402 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
11403 not enabled or disabled by the @option{-mcpu} option at present because
11404 AIX does not have full support for these options. You may still
11405 enable or disable them individually if you're sure it'll work in your
11408 @item -mtune=@var{cpu_type}
11410 Set the instruction scheduling parameters for machine type
11411 @var{cpu_type}, but do not set the architecture type, register usage, or
11412 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
11413 values for @var{cpu_type} are used for @option{-mtune} as for
11414 @option{-mcpu}. If both are specified, the code generated will use the
11415 architecture, registers, and mnemonics set by @option{-mcpu}, but the
11416 scheduling parameters set by @option{-mtune}.
11422 Generate code to compute division as reciprocal estimate and iterative
11423 refinement, creating opportunities for increased throughput. This
11424 feature requires: optional PowerPC Graphics instruction set for single
11425 precision and FRE instruction for double precision, assuming divides
11426 cannot generate user-visible traps, and the domain values not include
11427 Infinities, denormals or zero denominator.
11430 @itemx -mno-altivec
11432 @opindex mno-altivec
11433 Generate code that uses (does not use) AltiVec instructions, and also
11434 enable the use of built-in functions that allow more direct access to
11435 the AltiVec instruction set. You may also need to set
11436 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
11442 @opindex mno-vrsave
11443 Generate VRSAVE instructions when generating AltiVec code.
11446 @opindex msecure-plt
11447 Generate code that allows ld and ld.so to build executables and shared
11448 libraries with non-exec .plt and .got sections. This is a PowerPC
11449 32-bit SYSV ABI option.
11453 Generate code that uses a BSS .plt section that ld.so fills in, and
11454 requires .plt and .got sections that are both writable and executable.
11455 This is a PowerPC 32-bit SYSV ABI option.
11461 This switch enables or disables the generation of ISEL instructions.
11463 @item -misel=@var{yes/no}
11464 This switch has been deprecated. Use @option{-misel} and
11465 @option{-mno-isel} instead.
11471 This switch enables or disables the generation of SPE simd
11474 @item -mspe=@var{yes/no}
11475 This option has been deprecated. Use @option{-mspe} and
11476 @option{-mno-spe} instead.
11478 @item -mfloat-gprs=@var{yes/single/double/no}
11479 @itemx -mfloat-gprs
11480 @opindex mfloat-gprs
11481 This switch enables or disables the generation of floating point
11482 operations on the general purpose registers for architectures that
11485 The argument @var{yes} or @var{single} enables the use of
11486 single-precision floating point operations.
11488 The argument @var{double} enables the use of single and
11489 double-precision floating point operations.
11491 The argument @var{no} disables floating point operations on the
11492 general purpose registers.
11494 This option is currently only available on the MPC854x.
11500 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
11501 targets (including GNU/Linux). The 32-bit environment sets int, long
11502 and pointer to 32 bits and generates code that runs on any PowerPC
11503 variant. The 64-bit environment sets int to 32 bits and long and
11504 pointer to 64 bits, and generates code for PowerPC64, as for
11505 @option{-mpowerpc64}.
11508 @itemx -mno-fp-in-toc
11509 @itemx -mno-sum-in-toc
11510 @itemx -mminimal-toc
11512 @opindex mno-fp-in-toc
11513 @opindex mno-sum-in-toc
11514 @opindex mminimal-toc
11515 Modify generation of the TOC (Table Of Contents), which is created for
11516 every executable file. The @option{-mfull-toc} option is selected by
11517 default. In that case, GCC will allocate at least one TOC entry for
11518 each unique non-automatic variable reference in your program. GCC
11519 will also place floating-point constants in the TOC@. However, only
11520 16,384 entries are available in the TOC@.
11522 If you receive a linker error message that saying you have overflowed
11523 the available TOC space, you can reduce the amount of TOC space used
11524 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
11525 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
11526 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
11527 generate code to calculate the sum of an address and a constant at
11528 run-time instead of putting that sum into the TOC@. You may specify one
11529 or both of these options. Each causes GCC to produce very slightly
11530 slower and larger code at the expense of conserving TOC space.
11532 If you still run out of space in the TOC even when you specify both of
11533 these options, specify @option{-mminimal-toc} instead. This option causes
11534 GCC to make only one TOC entry for every file. When you specify this
11535 option, GCC will produce code that is slower and larger but which
11536 uses extremely little TOC space. You may wish to use this option
11537 only on files that contain less frequently executed code.
11543 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
11544 @code{long} type, and the infrastructure needed to support them.
11545 Specifying @option{-maix64} implies @option{-mpowerpc64} and
11546 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
11547 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
11550 @itemx -mno-xl-compat
11551 @opindex mxl-compat
11552 @opindex mno-xl-compat
11553 Produce code that conforms more closely to IBM XL compiler semantics
11554 when using AIX-compatible ABI. Pass floating-point arguments to
11555 prototyped functions beyond the register save area (RSA) on the stack
11556 in addition to argument FPRs. Do not assume that most significant
11557 double in 128-bit long double value is properly rounded when comparing
11558 values and converting to double. Use XL symbol names for long double
11561 The AIX calling convention was extended but not initially documented to
11562 handle an obscure K&R C case of calling a function that takes the
11563 address of its arguments with fewer arguments than declared. IBM XL
11564 compilers access floating point arguments which do not fit in the
11565 RSA from the stack when a subroutine is compiled without
11566 optimization. Because always storing floating-point arguments on the
11567 stack is inefficient and rarely needed, this option is not enabled by
11568 default and only is necessary when calling subroutines compiled by IBM
11569 XL compilers without optimization.
11573 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
11574 application written to use message passing with special startup code to
11575 enable the application to run. The system must have PE installed in the
11576 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
11577 must be overridden with the @option{-specs=} option to specify the
11578 appropriate directory location. The Parallel Environment does not
11579 support threads, so the @option{-mpe} option and the @option{-pthread}
11580 option are incompatible.
11582 @item -malign-natural
11583 @itemx -malign-power
11584 @opindex malign-natural
11585 @opindex malign-power
11586 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
11587 @option{-malign-natural} overrides the ABI-defined alignment of larger
11588 types, such as floating-point doubles, on their natural size-based boundary.
11589 The option @option{-malign-power} instructs GCC to follow the ABI-specified
11590 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
11592 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
11596 @itemx -mhard-float
11597 @opindex msoft-float
11598 @opindex mhard-float
11599 Generate code that does not use (uses) the floating-point register set.
11600 Software floating point emulation is provided if you use the
11601 @option{-msoft-float} option, and pass the option to GCC when linking.
11604 @itemx -mno-multiple
11606 @opindex mno-multiple
11607 Generate code that uses (does not use) the load multiple word
11608 instructions and the store multiple word instructions. These
11609 instructions are generated by default on POWER systems, and not
11610 generated on PowerPC systems. Do not use @option{-mmultiple} on little
11611 endian PowerPC systems, since those instructions do not work when the
11612 processor is in little endian mode. The exceptions are PPC740 and
11613 PPC750 which permit the instructions usage in little endian mode.
11618 @opindex mno-string
11619 Generate code that uses (does not use) the load string instructions
11620 and the store string word instructions to save multiple registers and
11621 do small block moves. These instructions are generated by default on
11622 POWER systems, and not generated on PowerPC systems. Do not use
11623 @option{-mstring} on little endian PowerPC systems, since those
11624 instructions do not work when the processor is in little endian mode.
11625 The exceptions are PPC740 and PPC750 which permit the instructions
11626 usage in little endian mode.
11631 @opindex mno-update
11632 Generate code that uses (does not use) the load or store instructions
11633 that update the base register to the address of the calculated memory
11634 location. These instructions are generated by default. If you use
11635 @option{-mno-update}, there is a small window between the time that the
11636 stack pointer is updated and the address of the previous frame is
11637 stored, which means code that walks the stack frame across interrupts or
11638 signals may get corrupted data.
11641 @itemx -mno-fused-madd
11642 @opindex mfused-madd
11643 @opindex mno-fused-madd
11644 Generate code that uses (does not use) the floating point multiply and
11645 accumulate instructions. These instructions are generated by default if
11646 hardware floating is used.
11652 Generate code that uses (does not use) the half-word multiply and
11653 multiply-accumulate instructions on the IBM 405 and 440 processors.
11654 These instructions are generated by default when targetting those
11661 Generate code that uses (does not use) the string-search @samp{dlmzb}
11662 instruction on the IBM 405 and 440 processors. This instruction is
11663 generated by default when targetting those processors.
11665 @item -mno-bit-align
11667 @opindex mno-bit-align
11668 @opindex mbit-align
11669 On System V.4 and embedded PowerPC systems do not (do) force structures
11670 and unions that contain bit-fields to be aligned to the base type of the
11673 For example, by default a structure containing nothing but 8
11674 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
11675 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
11676 the structure would be aligned to a 1 byte boundary and be one byte in
11679 @item -mno-strict-align
11680 @itemx -mstrict-align
11681 @opindex mno-strict-align
11682 @opindex mstrict-align
11683 On System V.4 and embedded PowerPC systems do not (do) assume that
11684 unaligned memory references will be handled by the system.
11686 @item -mrelocatable
11687 @itemx -mno-relocatable
11688 @opindex mrelocatable
11689 @opindex mno-relocatable
11690 On embedded PowerPC systems generate code that allows (does not allow)
11691 the program to be relocated to a different address at runtime. If you
11692 use @option{-mrelocatable} on any module, all objects linked together must
11693 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
11695 @item -mrelocatable-lib
11696 @itemx -mno-relocatable-lib
11697 @opindex mrelocatable-lib
11698 @opindex mno-relocatable-lib
11699 On embedded PowerPC systems generate code that allows (does not allow)
11700 the program to be relocated to a different address at runtime. Modules
11701 compiled with @option{-mrelocatable-lib} can be linked with either modules
11702 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
11703 with modules compiled with the @option{-mrelocatable} options.
11709 On System V.4 and embedded PowerPC systems do not (do) assume that
11710 register 2 contains a pointer to a global area pointing to the addresses
11711 used in the program.
11714 @itemx -mlittle-endian
11716 @opindex mlittle-endian
11717 On System V.4 and embedded PowerPC systems compile code for the
11718 processor in little endian mode. The @option{-mlittle-endian} option is
11719 the same as @option{-mlittle}.
11722 @itemx -mbig-endian
11724 @opindex mbig-endian
11725 On System V.4 and embedded PowerPC systems compile code for the
11726 processor in big endian mode. The @option{-mbig-endian} option is
11727 the same as @option{-mbig}.
11729 @item -mdynamic-no-pic
11730 @opindex mdynamic-no-pic
11731 On Darwin and Mac OS X systems, compile code so that it is not
11732 relocatable, but that its external references are relocatable. The
11733 resulting code is suitable for applications, but not shared
11736 @item -mprioritize-restricted-insns=@var{priority}
11737 @opindex mprioritize-restricted-insns
11738 This option controls the priority that is assigned to
11739 dispatch-slot restricted instructions during the second scheduling
11740 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
11741 @var{no/highest/second-highest} priority to dispatch slot restricted
11744 @item -msched-costly-dep=@var{dependence_type}
11745 @opindex msched-costly-dep
11746 This option controls which dependences are considered costly
11747 by the target during instruction scheduling. The argument
11748 @var{dependence_type} takes one of the following values:
11749 @var{no}: no dependence is costly,
11750 @var{all}: all dependences are costly,
11751 @var{true_store_to_load}: a true dependence from store to load is costly,
11752 @var{store_to_load}: any dependence from store to load is costly,
11753 @var{number}: any dependence which latency >= @var{number} is costly.
11755 @item -minsert-sched-nops=@var{scheme}
11756 @opindex minsert-sched-nops
11757 This option controls which nop insertion scheme will be used during
11758 the second scheduling pass. The argument @var{scheme} takes one of the
11760 @var{no}: Don't insert nops.
11761 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
11762 according to the scheduler's grouping.
11763 @var{regroup_exact}: Insert nops to force costly dependent insns into
11764 separate groups. Insert exactly as many nops as needed to force an insn
11765 to a new group, according to the estimated processor grouping.
11766 @var{number}: Insert nops to force costly dependent insns into
11767 separate groups. Insert @var{number} nops to force an insn to a new group.
11770 @opindex mcall-sysv
11771 On System V.4 and embedded PowerPC systems compile code using calling
11772 conventions that adheres to the March 1995 draft of the System V
11773 Application Binary Interface, PowerPC processor supplement. This is the
11774 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
11776 @item -mcall-sysv-eabi
11777 @opindex mcall-sysv-eabi
11778 Specify both @option{-mcall-sysv} and @option{-meabi} options.
11780 @item -mcall-sysv-noeabi
11781 @opindex mcall-sysv-noeabi
11782 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
11784 @item -mcall-solaris
11785 @opindex mcall-solaris
11786 On System V.4 and embedded PowerPC systems compile code for the Solaris
11790 @opindex mcall-linux
11791 On System V.4 and embedded PowerPC systems compile code for the
11792 Linux-based GNU system.
11796 On System V.4 and embedded PowerPC systems compile code for the
11797 Hurd-based GNU system.
11799 @item -mcall-netbsd
11800 @opindex mcall-netbsd
11801 On System V.4 and embedded PowerPC systems compile code for the
11802 NetBSD operating system.
11804 @item -maix-struct-return
11805 @opindex maix-struct-return
11806 Return all structures in memory (as specified by the AIX ABI)@.
11808 @item -msvr4-struct-return
11809 @opindex msvr4-struct-return
11810 Return structures smaller than 8 bytes in registers (as specified by the
11813 @item -mabi=@var{abi-type}
11815 Extend the current ABI with a particular extension, or remove such extension.
11816 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
11817 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
11821 Extend the current ABI with SPE ABI extensions. This does not change
11822 the default ABI, instead it adds the SPE ABI extensions to the current
11826 @opindex mabi=no-spe
11827 Disable Booke SPE ABI extensions for the current ABI@.
11829 @item -mabi=ibmlongdouble
11830 @opindex mabi=ibmlongdouble
11831 Change the current ABI to use IBM extended precision long double.
11832 This is a PowerPC 32-bit SYSV ABI option.
11834 @item -mabi=ieeelongdouble
11835 @opindex mabi=ieeelongdouble
11836 Change the current ABI to use IEEE extended precision long double.
11837 This is a PowerPC 32-bit Linux ABI option.
11840 @itemx -mno-prototype
11841 @opindex mprototype
11842 @opindex mno-prototype
11843 On System V.4 and embedded PowerPC systems assume that all calls to
11844 variable argument functions are properly prototyped. Otherwise, the
11845 compiler must insert an instruction before every non prototyped call to
11846 set or clear bit 6 of the condition code register (@var{CR}) to
11847 indicate whether floating point values were passed in the floating point
11848 registers in case the function takes a variable arguments. With
11849 @option{-mprototype}, only calls to prototyped variable argument functions
11850 will set or clear the bit.
11854 On embedded PowerPC systems, assume that the startup module is called
11855 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
11856 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
11861 On embedded PowerPC systems, assume that the startup module is called
11862 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
11867 On embedded PowerPC systems, assume that the startup module is called
11868 @file{crt0.o} and the standard C libraries are @file{libads.a} and
11871 @item -myellowknife
11872 @opindex myellowknife
11873 On embedded PowerPC systems, assume that the startup module is called
11874 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
11879 On System V.4 and embedded PowerPC systems, specify that you are
11880 compiling for a VxWorks system.
11884 Specify that you are compiling for the WindISS simulation environment.
11888 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
11889 header to indicate that @samp{eabi} extended relocations are used.
11895 On System V.4 and embedded PowerPC systems do (do not) adhere to the
11896 Embedded Applications Binary Interface (eabi) which is a set of
11897 modifications to the System V.4 specifications. Selecting @option{-meabi}
11898 means that the stack is aligned to an 8 byte boundary, a function
11899 @code{__eabi} is called to from @code{main} to set up the eabi
11900 environment, and the @option{-msdata} option can use both @code{r2} and
11901 @code{r13} to point to two separate small data areas. Selecting
11902 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
11903 do not call an initialization function from @code{main}, and the
11904 @option{-msdata} option will only use @code{r13} to point to a single
11905 small data area. The @option{-meabi} option is on by default if you
11906 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
11909 @opindex msdata=eabi
11910 On System V.4 and embedded PowerPC systems, put small initialized
11911 @code{const} global and static data in the @samp{.sdata2} section, which
11912 is pointed to by register @code{r2}. Put small initialized
11913 non-@code{const} global and static data in the @samp{.sdata} section,
11914 which is pointed to by register @code{r13}. Put small uninitialized
11915 global and static data in the @samp{.sbss} section, which is adjacent to
11916 the @samp{.sdata} section. The @option{-msdata=eabi} option is
11917 incompatible with the @option{-mrelocatable} option. The
11918 @option{-msdata=eabi} option also sets the @option{-memb} option.
11921 @opindex msdata=sysv
11922 On System V.4 and embedded PowerPC systems, put small global and static
11923 data in the @samp{.sdata} section, which is pointed to by register
11924 @code{r13}. Put small uninitialized global and static data in the
11925 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
11926 The @option{-msdata=sysv} option is incompatible with the
11927 @option{-mrelocatable} option.
11929 @item -msdata=default
11931 @opindex msdata=default
11933 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
11934 compile code the same as @option{-msdata=eabi}, otherwise compile code the
11935 same as @option{-msdata=sysv}.
11938 @opindex msdata-data
11939 On System V.4 and embedded PowerPC systems, put small global
11940 data in the @samp{.sdata} section. Put small uninitialized global
11941 data in the @samp{.sbss} section. Do not use register @code{r13}
11942 to address small data however. This is the default behavior unless
11943 other @option{-msdata} options are used.
11947 @opindex msdata=none
11949 On embedded PowerPC systems, put all initialized global and static data
11950 in the @samp{.data} section, and all uninitialized data in the
11951 @samp{.bss} section.
11955 @cindex smaller data references (PowerPC)
11956 @cindex .sdata/.sdata2 references (PowerPC)
11957 On embedded PowerPC systems, put global and static items less than or
11958 equal to @var{num} bytes into the small data or bss sections instead of
11959 the normal data or bss section. By default, @var{num} is 8. The
11960 @option{-G @var{num}} switch is also passed to the linker.
11961 All modules should be compiled with the same @option{-G @var{num}} value.
11964 @itemx -mno-regnames
11966 @opindex mno-regnames
11967 On System V.4 and embedded PowerPC systems do (do not) emit register
11968 names in the assembly language output using symbolic forms.
11971 @itemx -mno-longcall
11973 @opindex mno-longcall
11974 By default assume that all calls are far away so that a longer more
11975 expensive calling sequence is required. This is required for calls
11976 further than 32 megabytes (33,554,432 bytes) from the current location.
11977 A short call will be generated if the compiler knows
11978 the call cannot be that far away. This setting can be overridden by
11979 the @code{shortcall} function attribute, or by @code{#pragma
11982 Some linkers are capable of detecting out-of-range calls and generating
11983 glue code on the fly. On these systems, long calls are unnecessary and
11984 generate slower code. As of this writing, the AIX linker can do this,
11985 as can the GNU linker for PowerPC/64. It is planned to add this feature
11986 to the GNU linker for 32-bit PowerPC systems as well.
11988 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
11989 callee, L42'', plus a ``branch island'' (glue code). The two target
11990 addresses represent the callee and the ``branch island''. The
11991 Darwin/PPC linker will prefer the first address and generate a ``bl
11992 callee'' if the PPC ``bl'' instruction will reach the callee directly;
11993 otherwise, the linker will generate ``bl L42'' to call the ``branch
11994 island''. The ``branch island'' is appended to the body of the
11995 calling function; it computes the full 32-bit address of the callee
11998 On Mach-O (Darwin) systems, this option directs the compiler emit to
11999 the glue for every direct call, and the Darwin linker decides whether
12000 to use or discard it.
12002 In the future, we may cause GCC to ignore all longcall specifications
12003 when the linker is known to generate glue.
12007 Adds support for multithreading with the @dfn{pthreads} library.
12008 This option sets flags for both the preprocessor and linker.
12012 @node S/390 and zSeries Options
12013 @subsection S/390 and zSeries Options
12014 @cindex S/390 and zSeries Options
12016 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
12020 @itemx -msoft-float
12021 @opindex mhard-float
12022 @opindex msoft-float
12023 Use (do not use) the hardware floating-point instructions and registers
12024 for floating-point operations. When @option{-msoft-float} is specified,
12025 functions in @file{libgcc.a} will be used to perform floating-point
12026 operations. When @option{-mhard-float} is specified, the compiler
12027 generates IEEE floating-point instructions. This is the default.
12029 @item -mlong-double-64
12030 @itemx -mlong-double-128
12031 @opindex mlong-double-64
12032 @opindex mlong-double-128
12033 These switches control the size of @code{long double} type. A size
12034 of 64bit makes the @code{long double} type equivalent to the @code{double}
12035 type. This is the default.
12038 @itemx -mno-backchain
12039 @opindex mbackchain
12040 @opindex mno-backchain
12041 Store (do not store) the address of the caller's frame as backchain pointer
12042 into the callee's stack frame.
12043 A backchain may be needed to allow debugging using tools that do not understand
12044 DWARF-2 call frame information.
12045 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
12046 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
12047 the backchain is placed into the topmost word of the 96/160 byte register
12050 In general, code compiled with @option{-mbackchain} is call-compatible with
12051 code compiled with @option{-mmo-backchain}; however, use of the backchain
12052 for debugging purposes usually requires that the whole binary is built with
12053 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
12054 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
12055 to build a linux kernel use @option{-msoft-float}.
12057 The default is to not maintain the backchain.
12059 @item -mpacked-stack
12060 @item -mno-packed-stack
12061 @opindex mpacked-stack
12062 @opindex mno-packed-stack
12063 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
12064 specified, the compiler uses the all fields of the 96/160 byte register save
12065 area only for their default purpose; unused fields still take up stack space.
12066 When @option{-mpacked-stack} is specified, register save slots are densely
12067 packed at the top of the register save area; unused space is reused for other
12068 purposes, allowing for more efficient use of the available stack space.
12069 However, when @option{-mbackchain} is also in effect, the topmost word of
12070 the save area is always used to store the backchain, and the return address
12071 register is always saved two words below the backchain.
12073 As long as the stack frame backchain is not used, code generated with
12074 @option{-mpacked-stack} is call-compatible with code generated with
12075 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
12076 S/390 or zSeries generated code that uses the stack frame backchain at run
12077 time, not just for debugging purposes. Such code is not call-compatible
12078 with code compiled with @option{-mpacked-stack}. Also, note that the
12079 combination of @option{-mbackchain},
12080 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
12081 to build a linux kernel use @option{-msoft-float}.
12083 The default is to not use the packed stack layout.
12086 @itemx -mno-small-exec
12087 @opindex msmall-exec
12088 @opindex mno-small-exec
12089 Generate (or do not generate) code using the @code{bras} instruction
12090 to do subroutine calls.
12091 This only works reliably if the total executable size does not
12092 exceed 64k. The default is to use the @code{basr} instruction instead,
12093 which does not have this limitation.
12099 When @option{-m31} is specified, generate code compliant to the
12100 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
12101 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
12102 particular to generate 64-bit instructions. For the @samp{s390}
12103 targets, the default is @option{-m31}, while the @samp{s390x}
12104 targets default to @option{-m64}.
12110 When @option{-mzarch} is specified, generate code using the
12111 instructions available on z/Architecture.
12112 When @option{-mesa} is specified, generate code using the
12113 instructions available on ESA/390. Note that @option{-mesa} is
12114 not possible with @option{-m64}.
12115 When generating code compliant to the GNU/Linux for S/390 ABI,
12116 the default is @option{-mesa}. When generating code compliant
12117 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
12123 Generate (or do not generate) code using the @code{mvcle} instruction
12124 to perform block moves. When @option{-mno-mvcle} is specified,
12125 use a @code{mvc} loop instead. This is the default unless optimizing for
12132 Print (or do not print) additional debug information when compiling.
12133 The default is to not print debug information.
12135 @item -march=@var{cpu-type}
12137 Generate code that will run on @var{cpu-type}, which is the name of a system
12138 representing a certain processor type. Possible values for
12139 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
12140 When generating code using the instructions available on z/Architecture,
12141 the default is @option{-march=z900}. Otherwise, the default is
12142 @option{-march=g5}.
12144 @item -mtune=@var{cpu-type}
12146 Tune to @var{cpu-type} everything applicable about the generated code,
12147 except for the ABI and the set of available instructions.
12148 The list of @var{cpu-type} values is the same as for @option{-march}.
12149 The default is the value used for @option{-march}.
12152 @itemx -mno-tpf-trace
12153 @opindex mtpf-trace
12154 @opindex mno-tpf-trace
12155 Generate code that adds (does not add) in TPF OS specific branches to trace
12156 routines in the operating system. This option is off by default, even
12157 when compiling for the TPF OS@.
12160 @itemx -mno-fused-madd
12161 @opindex mfused-madd
12162 @opindex mno-fused-madd
12163 Generate code that uses (does not use) the floating point multiply and
12164 accumulate instructions. These instructions are generated by default if
12165 hardware floating point is used.
12167 @item -mwarn-framesize=@var{framesize}
12168 @opindex mwarn-framesize
12169 Emit a warning if the current function exceeds the given frame size. Because
12170 this is a compile time check it doesn't need to be a real problem when the program
12171 runs. It is intended to identify functions which most probably cause
12172 a stack overflow. It is useful to be used in an environment with limited stack
12173 size e.g.@: the linux kernel.
12175 @item -mwarn-dynamicstack
12176 @opindex mwarn-dynamicstack
12177 Emit a warning if the function calls alloca or uses dynamically
12178 sized arrays. This is generally a bad idea with a limited stack size.
12180 @item -mstack-guard=@var{stack-guard}
12181 @item -mstack-size=@var{stack-size}
12182 @opindex mstack-guard
12183 @opindex mstack-size
12184 These arguments always have to be used in conjunction. If they are present the s390
12185 back end emits additional instructions in the function prologue which trigger a trap
12186 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
12187 (remember that the stack on s390 grows downward). These options are intended to
12188 be used to help debugging stack overflow problems. The additionally emitted code
12189 causes only little overhead and hence can also be used in production like systems
12190 without greater performance degradation. The given values have to be exact
12191 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard} without
12193 In order to be efficient the extra code makes the assumption that the stack starts
12194 at an address aligned to the value given by @var{stack-size}.
12197 @node Score Options
12198 @subsection Score Options
12199 @cindex Score Options
12201 These options are defined for Score implementations:
12206 Compile code for little endian mode.
12210 Compile code for big endian mode. This is the default.
12214 Enable the use of multiply-accumulate instructions. Disabled by default.
12218 Specify the SCORE5U of the target architecture.
12222 Specify the SCORE7 of the target architecture. This is the default.
12226 @subsection SH Options
12228 These @samp{-m} options are defined for the SH implementations:
12233 Generate code for the SH1.
12237 Generate code for the SH2.
12240 Generate code for the SH2e.
12244 Generate code for the SH3.
12248 Generate code for the SH3e.
12252 Generate code for the SH4 without a floating-point unit.
12254 @item -m4-single-only
12255 @opindex m4-single-only
12256 Generate code for the SH4 with a floating-point unit that only
12257 supports single-precision arithmetic.
12261 Generate code for the SH4 assuming the floating-point unit is in
12262 single-precision mode by default.
12266 Generate code for the SH4.
12270 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
12271 floating-point unit is not used.
12273 @item -m4a-single-only
12274 @opindex m4a-single-only
12275 Generate code for the SH4a, in such a way that no double-precision
12276 floating point operations are used.
12279 @opindex m4a-single
12280 Generate code for the SH4a assuming the floating-point unit is in
12281 single-precision mode by default.
12285 Generate code for the SH4a.
12289 Same as @option{-m4a-nofpu}, except that it implicitly passes
12290 @option{-dsp} to the assembler. GCC doesn't generate any DSP
12291 instructions at the moment.
12295 Compile code for the processor in big endian mode.
12299 Compile code for the processor in little endian mode.
12303 Align doubles at 64-bit boundaries. Note that this changes the calling
12304 conventions, and thus some functions from the standard C library will
12305 not work unless you recompile it first with @option{-mdalign}.
12309 Shorten some address references at link time, when possible; uses the
12310 linker option @option{-relax}.
12314 Use 32-bit offsets in @code{switch} tables. The default is to use
12319 Enable the use of the instruction @code{fmovd}.
12323 Comply with the calling conventions defined by Renesas.
12327 Comply with the calling conventions defined by Renesas.
12331 Comply with the calling conventions defined for GCC before the Renesas
12332 conventions were available. This option is the default for all
12333 targets of the SH toolchain except for @samp{sh-symbianelf}.
12336 @opindex mnomacsave
12337 Mark the @code{MAC} register as call-clobbered, even if
12338 @option{-mhitachi} is given.
12342 Increase IEEE-compliance of floating-point code.
12343 At the moment, this is equivalent to @option{-fno-finite-math-only}.
12344 When generating 16 bit SH opcodes, getting IEEE-conforming results for
12345 comparisons of NANs / infinities incurs extra overhead in every
12346 floating point comparison, therefore the default is set to
12347 @option{-ffinite-math-only}.
12351 Dump instruction size and location in the assembly code.
12354 @opindex mpadstruct
12355 This option is deprecated. It pads structures to multiple of 4 bytes,
12356 which is incompatible with the SH ABI@.
12360 Optimize for space instead of speed. Implied by @option{-Os}.
12363 @opindex mprefergot
12364 When generating position-independent code, emit function calls using
12365 the Global Offset Table instead of the Procedure Linkage Table.
12369 Generate a library function call to invalidate instruction cache
12370 entries, after fixing up a trampoline. This library function call
12371 doesn't assume it can write to the whole memory address space. This
12372 is the default when the target is @code{sh-*-linux*}.
12374 @item -multcost=@var{number}
12375 @opindex multcost=@var{number}
12376 Set the cost to assume for a multiply insn.
12378 @item -mdiv=@var{strategy}
12379 @opindex mdiv=@var{strategy}
12380 Set the division strategy to use for SHmedia code. @var{strategy} must be
12381 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
12382 inv:call2, inv:fp .
12383 "fp" performs the operation in floating point. This has a very high latency,
12384 but needs only a few instructions, so it might be a good choice if
12385 your code has enough easily exploitable ILP to allow the compiler to
12386 schedule the floating point instructions together with other instructions.
12387 Division by zero causes a floating point exception.
12388 "inv" uses integer operations to calculate the inverse of the divisor,
12389 and then multiplies the dividend with the inverse. This strategy allows
12390 cse and hoisting of the inverse calculation. Division by zero calculates
12391 an unspecified result, but does not trap.
12392 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
12393 have been found, or if the entire operation has been hoisted to the same
12394 place, the last stages of the inverse calculation are intertwined with the
12395 final multiply to reduce the overall latency, at the expense of using a few
12396 more instructions, and thus offering fewer scheduling opportunities with
12398 "call" calls a library function that usually implements the inv:minlat
12400 This gives high code density for m5-*media-nofpu compilations.
12401 "call2" uses a different entry point of the same library function, where it
12402 assumes that a pointer to a lookup table has already been set up, which
12403 exposes the pointer load to cse / code hoisting optimizations.
12404 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
12405 code generation, but if the code stays unoptimized, revert to the "call",
12406 "call2", or "fp" strategies, respectively. Note that the
12407 potentially-trapping side effect of division by zero is carried by a
12408 separate instruction, so it is possible that all the integer instructions
12409 are hoisted out, but the marker for the side effect stays where it is.
12410 A recombination to fp operations or a call is not possible in that case.
12411 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
12412 that the inverse calculation was nor separated from the multiply, they speed
12413 up division where the dividend fits into 20 bits (plus sign where applicable),
12414 by inserting a test to skip a number of operations in this case; this test
12415 slows down the case of larger dividends. inv20u assumes the case of a such
12416 a small dividend to be unlikely, and inv20l assumes it to be likely.
12418 @item -mdivsi3_libfunc=@var{name}
12419 @opindex mdivsi3_libfunc=@var{name}
12420 Set the name of the library function used for 32 bit signed division to
12421 @var{name}. This only affect the name used in the call and inv:call
12422 division strategies, and the compiler will still expect the same
12423 sets of input/output/clobbered registers as if this option was not present.
12425 @item -madjust-unroll
12426 @opindex madjust-unroll
12427 Throttle unrolling to avoid thrashing target registers.
12428 This option only has an effect if the gcc code base supports the
12429 TARGET_ADJUST_UNROLL_MAX target hook.
12431 @item -mindexed-addressing
12432 @opindex mindexed-addressing
12433 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
12434 This is only safe if the hardware and/or OS implement 32 bit wrap-around
12435 semantics for the indexed addressing mode. The architecture allows the
12436 implementation of processors with 64 bit MMU, which the OS could use to
12437 get 32 bit addressing, but since no current hardware implementation supports
12438 this or any other way to make the indexed addressing mode safe to use in
12439 the 32 bit ABI, the default is -mno-indexed-addressing.
12441 @item -mgettrcost=@var{number}
12442 @opindex mgettrcost=@var{number}
12443 Set the cost assumed for the gettr instruction to @var{number}.
12444 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
12448 Assume pt* instructions won't trap. This will generally generate better
12449 scheduled code, but is unsafe on current hardware. The current architecture
12450 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
12451 This has the unintentional effect of making it unsafe to schedule ptabs /
12452 ptrel before a branch, or hoist it out of a loop. For example,
12453 __do_global_ctors, a part of libgcc that runs constructors at program
12454 startup, calls functions in a list which is delimited by -1. With the
12455 -mpt-fixed option, the ptabs will be done before testing against -1.
12456 That means that all the constructors will be run a bit quicker, but when
12457 the loop comes to the end of the list, the program crashes because ptabs
12458 loads -1 into a target register. Since this option is unsafe for any
12459 hardware implementing the current architecture specification, the default
12460 is -mno-pt-fixed. Unless the user specifies a specific cost with
12461 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
12462 this deters register allocation using target registers for storing
12465 @item -minvalid-symbols
12466 @opindex minvalid-symbols
12467 Assume symbols might be invalid. Ordinary function symbols generated by
12468 the compiler will always be valid to load with movi/shori/ptabs or
12469 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
12470 to generate symbols that will cause ptabs / ptrel to trap.
12471 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
12472 It will then prevent cross-basic-block cse, hoisting and most scheduling
12473 of symbol loads. The default is @option{-mno-invalid-symbols}.
12476 @node SPARC Options
12477 @subsection SPARC Options
12478 @cindex SPARC options
12480 These @samp{-m} options are supported on the SPARC:
12483 @item -mno-app-regs
12485 @opindex mno-app-regs
12487 Specify @option{-mapp-regs} to generate output using the global registers
12488 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
12491 To be fully SVR4 ABI compliant at the cost of some performance loss,
12492 specify @option{-mno-app-regs}. You should compile libraries and system
12493 software with this option.
12496 @itemx -mhard-float
12498 @opindex mhard-float
12499 Generate output containing floating point instructions. This is the
12503 @itemx -msoft-float
12505 @opindex msoft-float
12506 Generate output containing library calls for floating point.
12507 @strong{Warning:} the requisite libraries are not available for all SPARC
12508 targets. Normally the facilities of the machine's usual C compiler are
12509 used, but this cannot be done directly in cross-compilation. You must make
12510 your own arrangements to provide suitable library functions for
12511 cross-compilation. The embedded targets @samp{sparc-*-aout} and
12512 @samp{sparclite-*-*} do provide software floating point support.
12514 @option{-msoft-float} changes the calling convention in the output file;
12515 therefore, it is only useful if you compile @emph{all} of a program with
12516 this option. In particular, you need to compile @file{libgcc.a}, the
12517 library that comes with GCC, with @option{-msoft-float} in order for
12520 @item -mhard-quad-float
12521 @opindex mhard-quad-float
12522 Generate output containing quad-word (long double) floating point
12525 @item -msoft-quad-float
12526 @opindex msoft-quad-float
12527 Generate output containing library calls for quad-word (long double)
12528 floating point instructions. The functions called are those specified
12529 in the SPARC ABI@. This is the default.
12531 As of this writing, there are no SPARC implementations that have hardware
12532 support for the quad-word floating point instructions. They all invoke
12533 a trap handler for one of these instructions, and then the trap handler
12534 emulates the effect of the instruction. Because of the trap handler overhead,
12535 this is much slower than calling the ABI library routines. Thus the
12536 @option{-msoft-quad-float} option is the default.
12538 @item -mno-unaligned-doubles
12539 @itemx -munaligned-doubles
12540 @opindex mno-unaligned-doubles
12541 @opindex munaligned-doubles
12542 Assume that doubles have 8 byte alignment. This is the default.
12544 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
12545 alignment only if they are contained in another type, or if they have an
12546 absolute address. Otherwise, it assumes they have 4 byte alignment.
12547 Specifying this option avoids some rare compatibility problems with code
12548 generated by other compilers. It is not the default because it results
12549 in a performance loss, especially for floating point code.
12551 @item -mno-faster-structs
12552 @itemx -mfaster-structs
12553 @opindex mno-faster-structs
12554 @opindex mfaster-structs
12555 With @option{-mfaster-structs}, the compiler assumes that structures
12556 should have 8 byte alignment. This enables the use of pairs of
12557 @code{ldd} and @code{std} instructions for copies in structure
12558 assignment, in place of twice as many @code{ld} and @code{st} pairs.
12559 However, the use of this changed alignment directly violates the SPARC
12560 ABI@. Thus, it's intended only for use on targets where the developer
12561 acknowledges that their resulting code will not be directly in line with
12562 the rules of the ABI@.
12564 @item -mimpure-text
12565 @opindex mimpure-text
12566 @option{-mimpure-text}, used in addition to @option{-shared}, tells
12567 the compiler to not pass @option{-z text} to the linker when linking a
12568 shared object. Using this option, you can link position-dependent
12569 code into a shared object.
12571 @option{-mimpure-text} suppresses the ``relocations remain against
12572 allocatable but non-writable sections'' linker error message.
12573 However, the necessary relocations will trigger copy-on-write, and the
12574 shared object is not actually shared across processes. Instead of
12575 using @option{-mimpure-text}, you should compile all source code with
12576 @option{-fpic} or @option{-fPIC}.
12578 This option is only available on SunOS and Solaris.
12580 @item -mcpu=@var{cpu_type}
12582 Set the instruction set, register set, and instruction scheduling parameters
12583 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
12584 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
12585 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
12586 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
12587 @samp{ultrasparc3}, and @samp{niagara}.
12589 Default instruction scheduling parameters are used for values that select
12590 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
12591 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
12593 Here is a list of each supported architecture and their supported
12598 v8: supersparc, hypersparc
12599 sparclite: f930, f934, sparclite86x
12601 v9: ultrasparc, ultrasparc3, niagara
12604 By default (unless configured otherwise), GCC generates code for the V7
12605 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
12606 additionally optimizes it for the Cypress CY7C602 chip, as used in the
12607 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
12608 SPARCStation 1, 2, IPX etc.
12610 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
12611 architecture. The only difference from V7 code is that the compiler emits
12612 the integer multiply and integer divide instructions which exist in SPARC-V8
12613 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
12614 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
12617 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
12618 the SPARC architecture. This adds the integer multiply, integer divide step
12619 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
12620 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
12621 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
12622 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
12623 MB86934 chip, which is the more recent SPARClite with FPU@.
12625 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
12626 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
12627 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
12628 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
12629 optimizes it for the TEMIC SPARClet chip.
12631 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
12632 architecture. This adds 64-bit integer and floating-point move instructions,
12633 3 additional floating-point condition code registers and conditional move
12634 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
12635 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
12636 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
12637 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
12638 @option{-mcpu=niagara}, the compiler additionally optimizes it for
12639 Sun UltraSPARC T1 chips.
12641 @item -mtune=@var{cpu_type}
12643 Set the instruction scheduling parameters for machine type
12644 @var{cpu_type}, but do not set the instruction set or register set that the
12645 option @option{-mcpu=@var{cpu_type}} would.
12647 The same values for @option{-mcpu=@var{cpu_type}} can be used for
12648 @option{-mtune=@var{cpu_type}}, but the only useful values are those
12649 that select a particular cpu implementation. Those are @samp{cypress},
12650 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
12651 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
12652 @samp{ultrasparc3}, and @samp{niagara}.
12657 @opindex mno-v8plus
12658 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
12659 difference from the V8 ABI is that the global and out registers are
12660 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
12661 mode for all SPARC-V9 processors.
12667 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
12668 Visual Instruction Set extensions. The default is @option{-mno-vis}.
12671 These @samp{-m} options are supported in addition to the above
12672 on SPARC-V9 processors in 64-bit environments:
12675 @item -mlittle-endian
12676 @opindex mlittle-endian
12677 Generate code for a processor running in little-endian mode. It is only
12678 available for a few configurations and most notably not on Solaris and Linux.
12684 Generate code for a 32-bit or 64-bit environment.
12685 The 32-bit environment sets int, long and pointer to 32 bits.
12686 The 64-bit environment sets int to 32 bits and long and pointer
12689 @item -mcmodel=medlow
12690 @opindex mcmodel=medlow
12691 Generate code for the Medium/Low code model: 64-bit addresses, programs
12692 must be linked in the low 32 bits of memory. Programs can be statically
12693 or dynamically linked.
12695 @item -mcmodel=medmid
12696 @opindex mcmodel=medmid
12697 Generate code for the Medium/Middle code model: 64-bit addresses, programs
12698 must be linked in the low 44 bits of memory, the text and data segments must
12699 be less than 2GB in size and the data segment must be located within 2GB of
12702 @item -mcmodel=medany
12703 @opindex mcmodel=medany
12704 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
12705 may be linked anywhere in memory, the text and data segments must be less
12706 than 2GB in size and the data segment must be located within 2GB of the
12709 @item -mcmodel=embmedany
12710 @opindex mcmodel=embmedany
12711 Generate code for the Medium/Anywhere code model for embedded systems:
12712 64-bit addresses, the text and data segments must be less than 2GB in
12713 size, both starting anywhere in memory (determined at link time). The
12714 global register %g4 points to the base of the data segment. Programs
12715 are statically linked and PIC is not supported.
12718 @itemx -mno-stack-bias
12719 @opindex mstack-bias
12720 @opindex mno-stack-bias
12721 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
12722 frame pointer if present, are offset by @minus{}2047 which must be added back
12723 when making stack frame references. This is the default in 64-bit mode.
12724 Otherwise, assume no such offset is present.
12727 These switches are supported in addition to the above on Solaris:
12732 Add support for multithreading using the Solaris threads library. This
12733 option sets flags for both the preprocessor and linker. This option does
12734 not affect the thread safety of object code produced by the compiler or
12735 that of libraries supplied with it.
12739 Add support for multithreading using the POSIX threads library. This
12740 option sets flags for both the preprocessor and linker. This option does
12741 not affect the thread safety of object code produced by the compiler or
12742 that of libraries supplied with it.
12746 This is a synonym for @option{-pthreads}.
12749 @node System V Options
12750 @subsection Options for System V
12752 These additional options are available on System V Release 4 for
12753 compatibility with other compilers on those systems:
12758 Create a shared object.
12759 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
12763 Identify the versions of each tool used by the compiler, in a
12764 @code{.ident} assembler directive in the output.
12768 Refrain from adding @code{.ident} directives to the output file (this is
12771 @item -YP,@var{dirs}
12773 Search the directories @var{dirs}, and no others, for libraries
12774 specified with @option{-l}.
12776 @item -Ym,@var{dir}
12778 Look in the directory @var{dir} to find the M4 preprocessor.
12779 The assembler uses this option.
12780 @c This is supposed to go with a -Yd for predefined M4 macro files, but
12781 @c the generic assembler that comes with Solaris takes just -Ym.
12784 @node TMS320C3x/C4x Options
12785 @subsection TMS320C3x/C4x Options
12786 @cindex TMS320C3x/C4x Options
12788 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
12792 @item -mcpu=@var{cpu_type}
12794 Set the instruction set, register set, and instruction scheduling
12795 parameters for machine type @var{cpu_type}. Supported values for
12796 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
12797 @samp{c44}. The default is @samp{c40} to generate code for the
12802 @itemx -msmall-memory
12804 @opindex mbig-memory
12806 @opindex msmall-memory
12808 Generates code for the big or small memory model. The small memory
12809 model assumed that all data fits into one 64K word page. At run-time
12810 the data page (DP) register must be set to point to the 64K page
12811 containing the .bss and .data program sections. The big memory model is
12812 the default and requires reloading of the DP register for every direct
12819 Allow (disallow) allocation of general integer operands into the block
12820 count register BK@.
12826 Enable (disable) generation of code using decrement and branch,
12827 DBcond(D), instructions. This is enabled by default for the C4x. To be
12828 on the safe side, this is disabled for the C3x, since the maximum
12829 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
12830 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
12831 that it can utilize the decrement and branch instruction, but will give
12832 up if there is more than one memory reference in the loop. Thus a loop
12833 where the loop counter is decremented can generate slightly more
12834 efficient code, in cases where the RPTB instruction cannot be utilized.
12836 @item -mdp-isr-reload
12838 @opindex mdp-isr-reload
12840 Force the DP register to be saved on entry to an interrupt service
12841 routine (ISR), reloaded to point to the data section, and restored on
12842 exit from the ISR@. This should not be required unless someone has
12843 violated the small memory model by modifying the DP register, say within
12850 For the C3x use the 24-bit MPYI instruction for integer multiplies
12851 instead of a library call to guarantee 32-bit results. Note that if one
12852 of the operands is a constant, then the multiplication will be performed
12853 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
12854 then squaring operations are performed inline instead of a library call.
12857 @itemx -mno-fast-fix
12859 @opindex mno-fast-fix
12860 The C3x/C4x FIX instruction to convert a floating point value to an
12861 integer value chooses the nearest integer less than or equal to the
12862 floating point value rather than to the nearest integer. Thus if the
12863 floating point number is negative, the result will be incorrectly
12864 truncated an additional code is necessary to detect and correct this
12865 case. This option can be used to disable generation of the additional
12866 code required to correct the result.
12872 Enable (disable) generation of repeat block sequences using the RPTB
12873 instruction for zero overhead looping. The RPTB construct is only used
12874 for innermost loops that do not call functions or jump across the loop
12875 boundaries. There is no advantage having nested RPTB loops due to the
12876 overhead required to save and restore the RC, RS, and RE registers.
12877 This is enabled by default with @option{-O2}.
12879 @item -mrpts=@var{count}
12883 Enable (disable) the use of the single instruction repeat instruction
12884 RPTS@. If a repeat block contains a single instruction, and the loop
12885 count can be guaranteed to be less than the value @var{count}, GCC will
12886 emit a RPTS instruction instead of a RPTB@. If no value is specified,
12887 then a RPTS will be emitted even if the loop count cannot be determined
12888 at compile time. Note that the repeated instruction following RPTS does
12889 not have to be reloaded from memory each iteration, thus freeing up the
12890 CPU buses for operands. However, since interrupts are blocked by this
12891 instruction, it is disabled by default.
12893 @item -mloop-unsigned
12894 @itemx -mno-loop-unsigned
12895 @opindex mloop-unsigned
12896 @opindex mno-loop-unsigned
12897 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
12898 is @math{2^{31} + 1} since these instructions test if the iteration count is
12899 negative to terminate the loop. If the iteration count is unsigned
12900 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
12901 exceeded. This switch allows an unsigned iteration count.
12905 Try to emit an assembler syntax that the TI assembler (asm30) is happy
12906 with. This also enforces compatibility with the API employed by the TI
12907 C3x C compiler. For example, long doubles are passed as structures
12908 rather than in floating point registers.
12914 Generate code that uses registers (stack) for passing arguments to functions.
12915 By default, arguments are passed in registers where possible rather
12916 than by pushing arguments on to the stack.
12918 @item -mparallel-insns
12919 @itemx -mno-parallel-insns
12920 @opindex mparallel-insns
12921 @opindex mno-parallel-insns
12922 Allow the generation of parallel instructions. This is enabled by
12923 default with @option{-O2}.
12925 @item -mparallel-mpy
12926 @itemx -mno-parallel-mpy
12927 @opindex mparallel-mpy
12928 @opindex mno-parallel-mpy
12929 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
12930 provided @option{-mparallel-insns} is also specified. These instructions have
12931 tight register constraints which can pessimize the code generation
12932 of large functions.
12937 @subsection V850 Options
12938 @cindex V850 Options
12940 These @samp{-m} options are defined for V850 implementations:
12944 @itemx -mno-long-calls
12945 @opindex mlong-calls
12946 @opindex mno-long-calls
12947 Treat all calls as being far away (near). If calls are assumed to be
12948 far away, the compiler will always load the functions address up into a
12949 register, and call indirect through the pointer.
12955 Do not optimize (do optimize) basic blocks that use the same index
12956 pointer 4 or more times to copy pointer into the @code{ep} register, and
12957 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
12958 option is on by default if you optimize.
12960 @item -mno-prolog-function
12961 @itemx -mprolog-function
12962 @opindex mno-prolog-function
12963 @opindex mprolog-function
12964 Do not use (do use) external functions to save and restore registers
12965 at the prologue and epilogue of a function. The external functions
12966 are slower, but use less code space if more than one function saves
12967 the same number of registers. The @option{-mprolog-function} option
12968 is on by default if you optimize.
12972 Try to make the code as small as possible. At present, this just turns
12973 on the @option{-mep} and @option{-mprolog-function} options.
12975 @item -mtda=@var{n}
12977 Put static or global variables whose size is @var{n} bytes or less into
12978 the tiny data area that register @code{ep} points to. The tiny data
12979 area can hold up to 256 bytes in total (128 bytes for byte references).
12981 @item -msda=@var{n}
12983 Put static or global variables whose size is @var{n} bytes or less into
12984 the small data area that register @code{gp} points to. The small data
12985 area can hold up to 64 kilobytes.
12987 @item -mzda=@var{n}
12989 Put static or global variables whose size is @var{n} bytes or less into
12990 the first 32 kilobytes of memory.
12994 Specify that the target processor is the V850.
12997 @opindex mbig-switch
12998 Generate code suitable for big switch tables. Use this option only if
12999 the assembler/linker complain about out of range branches within a switch
13004 This option will cause r2 and r5 to be used in the code generated by
13005 the compiler. This setting is the default.
13007 @item -mno-app-regs
13008 @opindex mno-app-regs
13009 This option will cause r2 and r5 to be treated as fixed registers.
13013 Specify that the target processor is the V850E1. The preprocessor
13014 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
13015 this option is used.
13019 Specify that the target processor is the V850E@. The preprocessor
13020 constant @samp{__v850e__} will be defined if this option is used.
13022 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
13023 are defined then a default target processor will be chosen and the
13024 relevant @samp{__v850*__} preprocessor constant will be defined.
13026 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
13027 defined, regardless of which processor variant is the target.
13029 @item -mdisable-callt
13030 @opindex mdisable-callt
13031 This option will suppress generation of the CALLT instruction for the
13032 v850e and v850e1 flavors of the v850 architecture. The default is
13033 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
13038 @subsection VAX Options
13039 @cindex VAX options
13041 These @samp{-m} options are defined for the VAX:
13046 Do not output certain jump instructions (@code{aobleq} and so on)
13047 that the Unix assembler for the VAX cannot handle across long
13052 Do output those jump instructions, on the assumption that you
13053 will assemble with the GNU assembler.
13057 Output code for g-format floating point numbers instead of d-format.
13060 @node x86-64 Options
13061 @subsection x86-64 Options
13062 @cindex x86-64 options
13064 These are listed under @xref{i386 and x86-64 Options}.
13066 @node Xstormy16 Options
13067 @subsection Xstormy16 Options
13068 @cindex Xstormy16 Options
13070 These options are defined for Xstormy16:
13075 Choose startup files and linker script suitable for the simulator.
13078 @node Xtensa Options
13079 @subsection Xtensa Options
13080 @cindex Xtensa Options
13082 These options are supported for Xtensa targets:
13086 @itemx -mno-const16
13088 @opindex mno-const16
13089 Enable or disable use of @code{CONST16} instructions for loading
13090 constant values. The @code{CONST16} instruction is currently not a
13091 standard option from Tensilica. When enabled, @code{CONST16}
13092 instructions are always used in place of the standard @code{L32R}
13093 instructions. The use of @code{CONST16} is enabled by default only if
13094 the @code{L32R} instruction is not available.
13097 @itemx -mno-fused-madd
13098 @opindex mfused-madd
13099 @opindex mno-fused-madd
13100 Enable or disable use of fused multiply/add and multiply/subtract
13101 instructions in the floating-point option. This has no effect if the
13102 floating-point option is not also enabled. Disabling fused multiply/add
13103 and multiply/subtract instructions forces the compiler to use separate
13104 instructions for the multiply and add/subtract operations. This may be
13105 desirable in some cases where strict IEEE 754-compliant results are
13106 required: the fused multiply add/subtract instructions do not round the
13107 intermediate result, thereby producing results with @emph{more} bits of
13108 precision than specified by the IEEE standard. Disabling fused multiply
13109 add/subtract instructions also ensures that the program output is not
13110 sensitive to the compiler's ability to combine multiply and add/subtract
13113 @item -mtext-section-literals
13114 @itemx -mno-text-section-literals
13115 @opindex mtext-section-literals
13116 @opindex mno-text-section-literals
13117 Control the treatment of literal pools. The default is
13118 @option{-mno-text-section-literals}, which places literals in a separate
13119 section in the output file. This allows the literal pool to be placed
13120 in a data RAM/ROM, and it also allows the linker to combine literal
13121 pools from separate object files to remove redundant literals and
13122 improve code size. With @option{-mtext-section-literals}, the literals
13123 are interspersed in the text section in order to keep them as close as
13124 possible to their references. This may be necessary for large assembly
13127 @item -mtarget-align
13128 @itemx -mno-target-align
13129 @opindex mtarget-align
13130 @opindex mno-target-align
13131 When this option is enabled, GCC instructs the assembler to
13132 automatically align instructions to reduce branch penalties at the
13133 expense of some code density. The assembler attempts to widen density
13134 instructions to align branch targets and the instructions following call
13135 instructions. If there are not enough preceding safe density
13136 instructions to align a target, no widening will be performed. The
13137 default is @option{-mtarget-align}. These options do not affect the
13138 treatment of auto-aligned instructions like @code{LOOP}, which the
13139 assembler will always align, either by widening density instructions or
13140 by inserting no-op instructions.
13143 @itemx -mno-longcalls
13144 @opindex mlongcalls
13145 @opindex mno-longcalls
13146 When this option is enabled, GCC instructs the assembler to translate
13147 direct calls to indirect calls unless it can determine that the target
13148 of a direct call is in the range allowed by the call instruction. This
13149 translation typically occurs for calls to functions in other source
13150 files. Specifically, the assembler translates a direct @code{CALL}
13151 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
13152 The default is @option{-mno-longcalls}. This option should be used in
13153 programs where the call target can potentially be out of range. This
13154 option is implemented in the assembler, not the compiler, so the
13155 assembly code generated by GCC will still show direct call
13156 instructions---look at the disassembled object code to see the actual
13157 instructions. Note that the assembler will use an indirect call for
13158 every cross-file call, not just those that really will be out of range.
13161 @node zSeries Options
13162 @subsection zSeries Options
13163 @cindex zSeries options
13165 These are listed under @xref{S/390 and zSeries Options}.
13167 @node Code Gen Options
13168 @section Options for Code Generation Conventions
13169 @cindex code generation conventions
13170 @cindex options, code generation
13171 @cindex run-time options
13173 These machine-independent options control the interface conventions
13174 used in code generation.
13176 Most of them have both positive and negative forms; the negative form
13177 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
13178 one of the forms is listed---the one which is not the default. You
13179 can figure out the other form by either removing @samp{no-} or adding
13183 @item -fbounds-check
13184 @opindex fbounds-check
13185 For front-ends that support it, generate additional code to check that
13186 indices used to access arrays are within the declared range. This is
13187 currently only supported by the Java and Fortran front-ends, where
13188 this option defaults to true and false respectively.
13192 This option generates traps for signed overflow on addition, subtraction,
13193 multiplication operations.
13197 This option instructs the compiler to assume that signed arithmetic
13198 overflow of addition, subtraction and multiplication wraps around
13199 using twos-complement representation. This flag enables some optimizations
13200 and disables others. This option is enabled by default for the Java
13201 front-end, as required by the Java language specification.
13204 @opindex fexceptions
13205 Enable exception handling. Generates extra code needed to propagate
13206 exceptions. For some targets, this implies GCC will generate frame
13207 unwind information for all functions, which can produce significant data
13208 size overhead, although it does not affect execution. If you do not
13209 specify this option, GCC will enable it by default for languages like
13210 C++ which normally require exception handling, and disable it for
13211 languages like C that do not normally require it. However, you may need
13212 to enable this option when compiling C code that needs to interoperate
13213 properly with exception handlers written in C++. You may also wish to
13214 disable this option if you are compiling older C++ programs that don't
13215 use exception handling.
13217 @item -fnon-call-exceptions
13218 @opindex fnon-call-exceptions
13219 Generate code that allows trapping instructions to throw exceptions.
13220 Note that this requires platform-specific runtime support that does
13221 not exist everywhere. Moreover, it only allows @emph{trapping}
13222 instructions to throw exceptions, i.e.@: memory references or floating
13223 point instructions. It does not allow exceptions to be thrown from
13224 arbitrary signal handlers such as @code{SIGALRM}.
13226 @item -funwind-tables
13227 @opindex funwind-tables
13228 Similar to @option{-fexceptions}, except that it will just generate any needed
13229 static data, but will not affect the generated code in any other way.
13230 You will normally not enable this option; instead, a language processor
13231 that needs this handling would enable it on your behalf.
13233 @item -fasynchronous-unwind-tables
13234 @opindex fasynchronous-unwind-tables
13235 Generate unwind table in dwarf2 format, if supported by target machine. The
13236 table is exact at each instruction boundary, so it can be used for stack
13237 unwinding from asynchronous events (such as debugger or garbage collector).
13239 @item -fpcc-struct-return
13240 @opindex fpcc-struct-return
13241 Return ``short'' @code{struct} and @code{union} values in memory like
13242 longer ones, rather than in registers. This convention is less
13243 efficient, but it has the advantage of allowing intercallability between
13244 GCC-compiled files and files compiled with other compilers, particularly
13245 the Portable C Compiler (pcc).
13247 The precise convention for returning structures in memory depends
13248 on the target configuration macros.
13250 Short structures and unions are those whose size and alignment match
13251 that of some integer type.
13253 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
13254 switch is not binary compatible with code compiled with the
13255 @option{-freg-struct-return} switch.
13256 Use it to conform to a non-default application binary interface.
13258 @item -freg-struct-return
13259 @opindex freg-struct-return
13260 Return @code{struct} and @code{union} values in registers when possible.
13261 This is more efficient for small structures than
13262 @option{-fpcc-struct-return}.
13264 If you specify neither @option{-fpcc-struct-return} nor
13265 @option{-freg-struct-return}, GCC defaults to whichever convention is
13266 standard for the target. If there is no standard convention, GCC
13267 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
13268 the principal compiler. In those cases, we can choose the standard, and
13269 we chose the more efficient register return alternative.
13271 @strong{Warning:} code compiled with the @option{-freg-struct-return}
13272 switch is not binary compatible with code compiled with the
13273 @option{-fpcc-struct-return} switch.
13274 Use it to conform to a non-default application binary interface.
13276 @item -fshort-enums
13277 @opindex fshort-enums
13278 Allocate to an @code{enum} type only as many bytes as it needs for the
13279 declared range of possible values. Specifically, the @code{enum} type
13280 will be equivalent to the smallest integer type which has enough room.
13282 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
13283 code that is not binary compatible with code generated without that switch.
13284 Use it to conform to a non-default application binary interface.
13286 @item -fshort-double
13287 @opindex fshort-double
13288 Use the same size for @code{double} as for @code{float}.
13290 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
13291 code that is not binary compatible with code generated without that switch.
13292 Use it to conform to a non-default application binary interface.
13294 @item -fshort-wchar
13295 @opindex fshort-wchar
13296 Override the underlying type for @samp{wchar_t} to be @samp{short
13297 unsigned int} instead of the default for the target. This option is
13298 useful for building programs to run under WINE@.
13300 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
13301 code that is not binary compatible with code generated without that switch.
13302 Use it to conform to a non-default application binary interface.
13305 @opindex fno-common
13306 In C, allocate even uninitialized global variables in the data section of the
13307 object file, rather than generating them as common blocks. This has the
13308 effect that if the same variable is declared (without @code{extern}) in
13309 two different compilations, you will get an error when you link them.
13310 The only reason this might be useful is if you wish to verify that the
13311 program will work on other systems which always work this way.
13315 Ignore the @samp{#ident} directive.
13317 @item -finhibit-size-directive
13318 @opindex finhibit-size-directive
13319 Don't output a @code{.size} assembler directive, or anything else that
13320 would cause trouble if the function is split in the middle, and the
13321 two halves are placed at locations far apart in memory. This option is
13322 used when compiling @file{crtstuff.c}; you should not need to use it
13325 @item -fverbose-asm
13326 @opindex fverbose-asm
13327 Put extra commentary information in the generated assembly code to
13328 make it more readable. This option is generally only of use to those
13329 who actually need to read the generated assembly code (perhaps while
13330 debugging the compiler itself).
13332 @option{-fno-verbose-asm}, the default, causes the
13333 extra information to be omitted and is useful when comparing two assembler
13338 @cindex global offset table
13340 Generate position-independent code (PIC) suitable for use in a shared
13341 library, if supported for the target machine. Such code accesses all
13342 constant addresses through a global offset table (GOT)@. The dynamic
13343 loader resolves the GOT entries when the program starts (the dynamic
13344 loader is not part of GCC; it is part of the operating system). If
13345 the GOT size for the linked executable exceeds a machine-specific
13346 maximum size, you get an error message from the linker indicating that
13347 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
13348 instead. (These maximums are 8k on the SPARC and 32k
13349 on the m68k and RS/6000. The 386 has no such limit.)
13351 Position-independent code requires special support, and therefore works
13352 only on certain machines. For the 386, GCC supports PIC for System V
13353 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
13354 position-independent.
13356 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13361 If supported for the target machine, emit position-independent code,
13362 suitable for dynamic linking and avoiding any limit on the size of the
13363 global offset table. This option makes a difference on the m68k,
13364 PowerPC and SPARC@.
13366 Position-independent code requires special support, and therefore works
13367 only on certain machines.
13369 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13376 These options are similar to @option{-fpic} and @option{-fPIC}, but
13377 generated position independent code can be only linked into executables.
13378 Usually these options are used when @option{-pie} GCC option will be
13379 used during linking.
13381 @item -fno-jump-tables
13382 @opindex fno-jump-tables
13383 Do not use jump tables for switch statements even where it would be
13384 more efficient than other code generation strategies. This option is
13385 of use in conjunction with @option{-fpic} or @option{-fPIC} for
13386 building code which forms part of a dynamic linker and cannot
13387 reference the address of a jump table. On some targets, jump tables
13388 do not require a GOT and this option is not needed.
13390 @item -ffixed-@var{reg}
13392 Treat the register named @var{reg} as a fixed register; generated code
13393 should never refer to it (except perhaps as a stack pointer, frame
13394 pointer or in some other fixed role).
13396 @var{reg} must be the name of a register. The register names accepted
13397 are machine-specific and are defined in the @code{REGISTER_NAMES}
13398 macro in the machine description macro file.
13400 This flag does not have a negative form, because it specifies a
13403 @item -fcall-used-@var{reg}
13404 @opindex fcall-used
13405 Treat the register named @var{reg} as an allocable register that is
13406 clobbered by function calls. It may be allocated for temporaries or
13407 variables that do not live across a call. Functions compiled this way
13408 will not save and restore the register @var{reg}.
13410 It is an error to used this flag with the frame pointer or stack pointer.
13411 Use of this flag for other registers that have fixed pervasive roles in
13412 the machine's execution model will produce disastrous results.
13414 This flag does not have a negative form, because it specifies a
13417 @item -fcall-saved-@var{reg}
13418 @opindex fcall-saved
13419 Treat the register named @var{reg} as an allocable register saved by
13420 functions. It may be allocated even for temporaries or variables that
13421 live across a call. Functions compiled this way will save and restore
13422 the register @var{reg} if they use it.
13424 It is an error to used this flag with the frame pointer or stack pointer.
13425 Use of this flag for other registers that have fixed pervasive roles in
13426 the machine's execution model will produce disastrous results.
13428 A different sort of disaster will result from the use of this flag for
13429 a register in which function values may be returned.
13431 This flag does not have a negative form, because it specifies a
13434 @item -fpack-struct[=@var{n}]
13435 @opindex fpack-struct
13436 Without a value specified, pack all structure members together without
13437 holes. When a value is specified (which must be a small power of two), pack
13438 structure members according to this value, representing the maximum
13439 alignment (that is, objects with default alignment requirements larger than
13440 this will be output potentially unaligned at the next fitting location.
13442 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
13443 code that is not binary compatible with code generated without that switch.
13444 Additionally, it makes the code suboptimal.
13445 Use it to conform to a non-default application binary interface.
13447 @item -finstrument-functions
13448 @opindex finstrument-functions
13449 Generate instrumentation calls for entry and exit to functions. Just
13450 after function entry and just before function exit, the following
13451 profiling functions will be called with the address of the current
13452 function and its call site. (On some platforms,
13453 @code{__builtin_return_address} does not work beyond the current
13454 function, so the call site information may not be available to the
13455 profiling functions otherwise.)
13458 void __cyg_profile_func_enter (void *this_fn,
13460 void __cyg_profile_func_exit (void *this_fn,
13464 The first argument is the address of the start of the current function,
13465 which may be looked up exactly in the symbol table.
13467 This instrumentation is also done for functions expanded inline in other
13468 functions. The profiling calls will indicate where, conceptually, the
13469 inline function is entered and exited. This means that addressable
13470 versions of such functions must be available. If all your uses of a
13471 function are expanded inline, this may mean an additional expansion of
13472 code size. If you use @samp{extern inline} in your C code, an
13473 addressable version of such functions must be provided. (This is
13474 normally the case anyways, but if you get lucky and the optimizer always
13475 expands the functions inline, you might have gotten away without
13476 providing static copies.)
13478 A function may be given the attribute @code{no_instrument_function}, in
13479 which case this instrumentation will not be done. This can be used, for
13480 example, for the profiling functions listed above, high-priority
13481 interrupt routines, and any functions from which the profiling functions
13482 cannot safely be called (perhaps signal handlers, if the profiling
13483 routines generate output or allocate memory).
13485 @item -fstack-check
13486 @opindex fstack-check
13487 Generate code to verify that you do not go beyond the boundary of the
13488 stack. You should specify this flag if you are running in an
13489 environment with multiple threads, but only rarely need to specify it in
13490 a single-threaded environment since stack overflow is automatically
13491 detected on nearly all systems if there is only one stack.
13493 Note that this switch does not actually cause checking to be done; the
13494 operating system must do that. The switch causes generation of code
13495 to ensure that the operating system sees the stack being extended.
13497 @item -fstack-limit-register=@var{reg}
13498 @itemx -fstack-limit-symbol=@var{sym}
13499 @itemx -fno-stack-limit
13500 @opindex fstack-limit-register
13501 @opindex fstack-limit-symbol
13502 @opindex fno-stack-limit
13503 Generate code to ensure that the stack does not grow beyond a certain value,
13504 either the value of a register or the address of a symbol. If the stack
13505 would grow beyond the value, a signal is raised. For most targets,
13506 the signal is raised before the stack overruns the boundary, so
13507 it is possible to catch the signal without taking special precautions.
13509 For instance, if the stack starts at absolute address @samp{0x80000000}
13510 and grows downwards, you can use the flags
13511 @option{-fstack-limit-symbol=__stack_limit} and
13512 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
13513 of 128KB@. Note that this may only work with the GNU linker.
13515 @cindex aliasing of parameters
13516 @cindex parameters, aliased
13517 @item -fargument-alias
13518 @itemx -fargument-noalias
13519 @itemx -fargument-noalias-global
13520 @itemx -fargument-noalias-anything
13521 @opindex fargument-alias
13522 @opindex fargument-noalias
13523 @opindex fargument-noalias-global
13524 @opindex fargument-noalias-anything
13525 Specify the possible relationships among parameters and between
13526 parameters and global data.
13528 @option{-fargument-alias} specifies that arguments (parameters) may
13529 alias each other and may alias global storage.@*
13530 @option{-fargument-noalias} specifies that arguments do not alias
13531 each other, but may alias global storage.@*
13532 @option{-fargument-noalias-global} specifies that arguments do not
13533 alias each other and do not alias global storage.
13534 @option{-fargument-noalias-anything} specifies that arguments do not
13535 alias any other storage.
13537 Each language will automatically use whatever option is required by
13538 the language standard. You should not need to use these options yourself.
13540 @item -fleading-underscore
13541 @opindex fleading-underscore
13542 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
13543 change the way C symbols are represented in the object file. One use
13544 is to help link with legacy assembly code.
13546 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
13547 generate code that is not binary compatible with code generated without that
13548 switch. Use it to conform to a non-default application binary interface.
13549 Not all targets provide complete support for this switch.
13551 @item -ftls-model=@var{model}
13552 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
13553 The @var{model} argument should be one of @code{global-dynamic},
13554 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
13556 The default without @option{-fpic} is @code{initial-exec}; with
13557 @option{-fpic} the default is @code{global-dynamic}.
13559 @item -fvisibility=@var{default|internal|hidden|protected}
13560 @opindex fvisibility
13561 Set the default ELF image symbol visibility to the specified option---all
13562 symbols will be marked with this unless overridden within the code.
13563 Using this feature can very substantially improve linking and
13564 load times of shared object libraries, produce more optimized
13565 code, provide near-perfect API export and prevent symbol clashes.
13566 It is @strong{strongly} recommended that you use this in any shared objects
13569 Despite the nomenclature, @code{default} always means public ie;
13570 available to be linked against from outside the shared object.
13571 @code{protected} and @code{internal} are pretty useless in real-world
13572 usage so the only other commonly used option will be @code{hidden}.
13573 The default if @option{-fvisibility} isn't specified is
13574 @code{default}, i.e., make every
13575 symbol public---this causes the same behavior as previous versions of
13578 A good explanation of the benefits offered by ensuring ELF
13579 symbols have the correct visibility is given by ``How To Write
13580 Shared Libraries'' by Ulrich Drepper (which can be found at
13581 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
13582 solution made possible by this option to marking things hidden when
13583 the default is public is to make the default hidden and mark things
13584 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
13585 and @code{__attribute__ ((visibility("default")))} instead of
13586 @code{__declspec(dllexport)} you get almost identical semantics with
13587 identical syntax. This is a great boon to those working with
13588 cross-platform projects.
13590 For those adding visibility support to existing code, you may find
13591 @samp{#pragma GCC visibility} of use. This works by you enclosing
13592 the declarations you wish to set visibility for with (for example)
13593 @samp{#pragma GCC visibility push(hidden)} and
13594 @samp{#pragma GCC visibility pop}.
13595 Bear in mind that symbol visibility should be viewed @strong{as
13596 part of the API interface contract} and thus all new code should
13597 always specify visibility when it is not the default ie; declarations
13598 only for use within the local DSO should @strong{always} be marked explicitly
13599 as hidden as so to avoid PLT indirection overheads---making this
13600 abundantly clear also aids readability and self-documentation of the code.
13601 Note that due to ISO C++ specification requirements, operator new and
13602 operator delete must always be of default visibility.
13604 Be aware that headers from outside your project, in particular system
13605 headers and headers from any other library you use, may not be
13606 expecting to be compiled with visibility other than the default. You
13607 may need to explicitly say @samp{#pragma GCC visibility push(default)}
13608 before including any such headers.
13610 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
13611 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
13612 no modifications. However, this means that calls to @samp{extern}
13613 functions with no explicit visibility will use the PLT, so it is more
13614 effective to use @samp{__attribute ((visibility))} and/or
13615 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
13616 declarations should be treated as hidden.
13618 Note that @samp{-fvisibility} does affect C++ vague linkage
13619 entities. This means that, for instance, an exception class that will
13620 be thrown between DSOs must be explicitly marked with default
13621 visibility so that the @samp{type_info} nodes will be unified between
13624 An overview of these techniques, their benefits and how to use them
13625 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
13631 @node Environment Variables
13632 @section Environment Variables Affecting GCC
13633 @cindex environment variables
13635 @c man begin ENVIRONMENT
13636 This section describes several environment variables that affect how GCC
13637 operates. Some of them work by specifying directories or prefixes to use
13638 when searching for various kinds of files. Some are used to specify other
13639 aspects of the compilation environment.
13641 Note that you can also specify places to search using options such as
13642 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
13643 take precedence over places specified using environment variables, which
13644 in turn take precedence over those specified by the configuration of GCC@.
13645 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
13646 GNU Compiler Collection (GCC) Internals}.
13651 @c @itemx LC_COLLATE
13653 @c @itemx LC_MONETARY
13654 @c @itemx LC_NUMERIC
13659 @c @findex LC_COLLATE
13660 @findex LC_MESSAGES
13661 @c @findex LC_MONETARY
13662 @c @findex LC_NUMERIC
13666 These environment variables control the way that GCC uses
13667 localization information that allow GCC to work with different
13668 national conventions. GCC inspects the locale categories
13669 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
13670 so. These locale categories can be set to any value supported by your
13671 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
13672 Kingdom encoded in UTF-8.
13674 The @env{LC_CTYPE} environment variable specifies character
13675 classification. GCC uses it to determine the character boundaries in
13676 a string; this is needed for some multibyte encodings that contain quote
13677 and escape characters that would otherwise be interpreted as a string
13680 The @env{LC_MESSAGES} environment variable specifies the language to
13681 use in diagnostic messages.
13683 If the @env{LC_ALL} environment variable is set, it overrides the value
13684 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
13685 and @env{LC_MESSAGES} default to the value of the @env{LANG}
13686 environment variable. If none of these variables are set, GCC
13687 defaults to traditional C English behavior.
13691 If @env{TMPDIR} is set, it specifies the directory to use for temporary
13692 files. GCC uses temporary files to hold the output of one stage of
13693 compilation which is to be used as input to the next stage: for example,
13694 the output of the preprocessor, which is the input to the compiler
13697 @item GCC_EXEC_PREFIX
13698 @findex GCC_EXEC_PREFIX
13699 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
13700 names of the subprograms executed by the compiler. No slash is added
13701 when this prefix is combined with the name of a subprogram, but you can
13702 specify a prefix that ends with a slash if you wish.
13704 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
13705 an appropriate prefix to use based on the pathname it was invoked with.
13707 If GCC cannot find the subprogram using the specified prefix, it
13708 tries looking in the usual places for the subprogram.
13710 The default value of @env{GCC_EXEC_PREFIX} is
13711 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
13712 of @code{prefix} when you ran the @file{configure} script.
13714 Other prefixes specified with @option{-B} take precedence over this prefix.
13716 This prefix is also used for finding files such as @file{crt0.o} that are
13719 In addition, the prefix is used in an unusual way in finding the
13720 directories to search for header files. For each of the standard
13721 directories whose name normally begins with @samp{/usr/local/lib/gcc}
13722 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
13723 replacing that beginning with the specified prefix to produce an
13724 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
13725 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
13726 These alternate directories are searched first; the standard directories
13729 @item COMPILER_PATH
13730 @findex COMPILER_PATH
13731 The value of @env{COMPILER_PATH} is a colon-separated list of
13732 directories, much like @env{PATH}. GCC tries the directories thus
13733 specified when searching for subprograms, if it can't find the
13734 subprograms using @env{GCC_EXEC_PREFIX}.
13737 @findex LIBRARY_PATH
13738 The value of @env{LIBRARY_PATH} is a colon-separated list of
13739 directories, much like @env{PATH}. When configured as a native compiler,
13740 GCC tries the directories thus specified when searching for special
13741 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
13742 using GCC also uses these directories when searching for ordinary
13743 libraries for the @option{-l} option (but directories specified with
13744 @option{-L} come first).
13748 @cindex locale definition
13749 This variable is used to pass locale information to the compiler. One way in
13750 which this information is used is to determine the character set to be used
13751 when character literals, string literals and comments are parsed in C and C++.
13752 When the compiler is configured to allow multibyte characters,
13753 the following values for @env{LANG} are recognized:
13757 Recognize JIS characters.
13759 Recognize SJIS characters.
13761 Recognize EUCJP characters.
13764 If @env{LANG} is not defined, or if it has some other value, then the
13765 compiler will use mblen and mbtowc as defined by the default locale to
13766 recognize and translate multibyte characters.
13770 Some additional environments variables affect the behavior of the
13773 @include cppenv.texi
13777 @node Precompiled Headers
13778 @section Using Precompiled Headers
13779 @cindex precompiled headers
13780 @cindex speed of compilation
13782 Often large projects have many header files that are included in every
13783 source file. The time the compiler takes to process these header files
13784 over and over again can account for nearly all of the time required to
13785 build the project. To make builds faster, GCC allows users to
13786 `precompile' a header file; then, if builds can use the precompiled
13787 header file they will be much faster.
13789 To create a precompiled header file, simply compile it as you would any
13790 other file, if necessary using the @option{-x} option to make the driver
13791 treat it as a C or C++ header file. You will probably want to use a
13792 tool like @command{make} to keep the precompiled header up-to-date when
13793 the headers it contains change.
13795 A precompiled header file will be searched for when @code{#include} is
13796 seen in the compilation. As it searches for the included file
13797 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
13798 compiler looks for a precompiled header in each directory just before it
13799 looks for the include file in that directory. The name searched for is
13800 the name specified in the @code{#include} with @samp{.gch} appended. If
13801 the precompiled header file can't be used, it is ignored.
13803 For instance, if you have @code{#include "all.h"}, and you have
13804 @file{all.h.gch} in the same directory as @file{all.h}, then the
13805 precompiled header file will be used if possible, and the original
13806 header will be used otherwise.
13808 Alternatively, you might decide to put the precompiled header file in a
13809 directory and use @option{-I} to ensure that directory is searched
13810 before (or instead of) the directory containing the original header.
13811 Then, if you want to check that the precompiled header file is always
13812 used, you can put a file of the same name as the original header in this
13813 directory containing an @code{#error} command.
13815 This also works with @option{-include}. So yet another way to use
13816 precompiled headers, good for projects not designed with precompiled
13817 header files in mind, is to simply take most of the header files used by
13818 a project, include them from another header file, precompile that header
13819 file, and @option{-include} the precompiled header. If the header files
13820 have guards against multiple inclusion, they will be skipped because
13821 they've already been included (in the precompiled header).
13823 If you need to precompile the same header file for different
13824 languages, targets, or compiler options, you can instead make a
13825 @emph{directory} named like @file{all.h.gch}, and put each precompiled
13826 header in the directory, perhaps using @option{-o}. It doesn't matter
13827 what you call the files in the directory, every precompiled header in
13828 the directory will be considered. The first precompiled header
13829 encountered in the directory that is valid for this compilation will
13830 be used; they're searched in no particular order.
13832 There are many other possibilities, limited only by your imagination,
13833 good sense, and the constraints of your build system.
13835 A precompiled header file can be used only when these conditions apply:
13839 Only one precompiled header can be used in a particular compilation.
13842 A precompiled header can't be used once the first C token is seen. You
13843 can have preprocessor directives before a precompiled header; you can
13844 even include a precompiled header from inside another header, so long as
13845 there are no C tokens before the @code{#include}.
13848 The precompiled header file must be produced for the same language as
13849 the current compilation. You can't use a C precompiled header for a C++
13853 The precompiled header file must have been produced by the same compiler
13854 binary as the current compilation is using.
13857 Any macros defined before the precompiled header is included must
13858 either be defined in the same way as when the precompiled header was
13859 generated, or must not affect the precompiled header, which usually
13860 means that they don't appear in the precompiled header at all.
13862 The @option{-D} option is one way to define a macro before a
13863 precompiled header is included; using a @code{#define} can also do it.
13864 There are also some options that define macros implicitly, like
13865 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
13868 @item If debugging information is output when using the precompiled
13869 header, using @option{-g} or similar, the same kind of debugging information
13870 must have been output when building the precompiled header. However,
13871 a precompiled header built using @option{-g} can be used in a compilation
13872 when no debugging information is being output.
13874 @item The same @option{-m} options must generally be used when building
13875 and using the precompiled header. @xref{Submodel Options},
13876 for any cases where this rule is relaxed.
13878 @item Each of the following options must be the same when building and using
13879 the precompiled header:
13881 @gccoptlist{-fexceptions -funit-at-a-time}
13884 Some other command-line options starting with @option{-f},
13885 @option{-p}, or @option{-O} must be defined in the same way as when
13886 the precompiled header was generated. At present, it's not clear
13887 which options are safe to change and which are not; the safest choice
13888 is to use exactly the same options when generating and using the
13889 precompiled header. The following are known to be safe:
13891 @gccoptlist{-fmessage-length= -fpreprocessed
13892 -fsched-interblock -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous
13893 -fsched-verbose=<number> -fschedule-insns -fvisibility=
13898 For all of these except the last, the compiler will automatically
13899 ignore the precompiled header if the conditions aren't met. If you
13900 find an option combination that doesn't work and doesn't cause the
13901 precompiled header to be ignored, please consider filing a bug report,
13904 If you do use differing options when generating and using the
13905 precompiled header, the actual behavior will be a mixture of the
13906 behavior for the options. For instance, if you use @option{-g} to
13907 generate the precompiled header but not when using it, you may or may
13908 not get debugging information for routines in the precompiled header.
13910 @node Running Protoize
13911 @section Running Protoize
13913 The program @code{protoize} is an optional part of GCC@. You can use
13914 it to add prototypes to a program, thus converting the program to ISO
13915 C in one respect. The companion program @code{unprotoize} does the
13916 reverse: it removes argument types from any prototypes that are found.
13918 When you run these programs, you must specify a set of source files as
13919 command line arguments. The conversion programs start out by compiling
13920 these files to see what functions they define. The information gathered
13921 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
13923 After scanning comes actual conversion. The specified files are all
13924 eligible to be converted; any files they include (whether sources or
13925 just headers) are eligible as well.
13927 But not all the eligible files are converted. By default,
13928 @code{protoize} and @code{unprotoize} convert only source and header
13929 files in the current directory. You can specify additional directories
13930 whose files should be converted with the @option{-d @var{directory}}
13931 option. You can also specify particular files to exclude with the
13932 @option{-x @var{file}} option. A file is converted if it is eligible, its
13933 directory name matches one of the specified directory names, and its
13934 name within the directory has not been excluded.
13936 Basic conversion with @code{protoize} consists of rewriting most
13937 function definitions and function declarations to specify the types of
13938 the arguments. The only ones not rewritten are those for varargs
13941 @code{protoize} optionally inserts prototype declarations at the
13942 beginning of the source file, to make them available for any calls that
13943 precede the function's definition. Or it can insert prototype
13944 declarations with block scope in the blocks where undeclared functions
13947 Basic conversion with @code{unprotoize} consists of rewriting most
13948 function declarations to remove any argument types, and rewriting
13949 function definitions to the old-style pre-ISO form.
13951 Both conversion programs print a warning for any function declaration or
13952 definition that they can't convert. You can suppress these warnings
13955 The output from @code{protoize} or @code{unprotoize} replaces the
13956 original source file. The original file is renamed to a name ending
13957 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
13958 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
13959 for DOS) file already exists, then the source file is simply discarded.
13961 @code{protoize} and @code{unprotoize} both depend on GCC itself to
13962 scan the program and collect information about the functions it uses.
13963 So neither of these programs will work until GCC is installed.
13965 Here is a table of the options you can use with @code{protoize} and
13966 @code{unprotoize}. Each option works with both programs unless
13970 @item -B @var{directory}
13971 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
13972 usual directory (normally @file{/usr/local/lib}). This file contains
13973 prototype information about standard system functions. This option
13974 applies only to @code{protoize}.
13976 @item -c @var{compilation-options}
13977 Use @var{compilation-options} as the options when running @command{gcc} to
13978 produce the @samp{.X} files. The special option @option{-aux-info} is
13979 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
13981 Note that the compilation options must be given as a single argument to
13982 @code{protoize} or @code{unprotoize}. If you want to specify several
13983 @command{gcc} options, you must quote the entire set of compilation options
13984 to make them a single word in the shell.
13986 There are certain @command{gcc} arguments that you cannot use, because they
13987 would produce the wrong kind of output. These include @option{-g},
13988 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
13989 the @var{compilation-options}, they are ignored.
13992 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
13993 systems) instead of @samp{.c}. This is convenient if you are converting
13994 a C program to C++. This option applies only to @code{protoize}.
13997 Add explicit global declarations. This means inserting explicit
13998 declarations at the beginning of each source file for each function
13999 that is called in the file and was not declared. These declarations
14000 precede the first function definition that contains a call to an
14001 undeclared function. This option applies only to @code{protoize}.
14003 @item -i @var{string}
14004 Indent old-style parameter declarations with the string @var{string}.
14005 This option applies only to @code{protoize}.
14007 @code{unprotoize} converts prototyped function definitions to old-style
14008 function definitions, where the arguments are declared between the
14009 argument list and the initial @samp{@{}. By default, @code{unprotoize}
14010 uses five spaces as the indentation. If you want to indent with just
14011 one space instead, use @option{-i " "}.
14014 Keep the @samp{.X} files. Normally, they are deleted after conversion
14018 Add explicit local declarations. @code{protoize} with @option{-l} inserts
14019 a prototype declaration for each function in each block which calls the
14020 function without any declaration. This option applies only to
14024 Make no real changes. This mode just prints information about the conversions
14025 that would have been done without @option{-n}.
14028 Make no @samp{.save} files. The original files are simply deleted.
14029 Use this option with caution.
14031 @item -p @var{program}
14032 Use the program @var{program} as the compiler. Normally, the name
14033 @file{gcc} is used.
14036 Work quietly. Most warnings are suppressed.
14039 Print the version number, just like @option{-v} for @command{gcc}.
14042 If you need special compiler options to compile one of your program's
14043 source files, then you should generate that file's @samp{.X} file
14044 specially, by running @command{gcc} on that source file with the
14045 appropriate options and the option @option{-aux-info}. Then run
14046 @code{protoize} on the entire set of files. @code{protoize} will use
14047 the existing @samp{.X} file because it is newer than the source file.
14051 gcc -Dfoo=bar file1.c -aux-info file1.X
14056 You need to include the special files along with the rest in the
14057 @code{protoize} command, even though their @samp{.X} files already
14058 exist, because otherwise they won't get converted.
14060 @xref{Protoize Caveats}, for more information on how to use
14061 @code{protoize} successfully.