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}.
200 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
201 -fgnu-runtime -fnext-runtime @gol
202 -fno-nil-receivers @gol
203 -fobjc-call-cxx-cdtors @gol
204 -fobjc-direct-dispatch @gol
205 -fobjc-exceptions @gol
207 -freplace-objc-classes @gol
210 -Wassign-intercept @gol
211 -Wno-protocol -Wselector @gol
212 -Wstrict-selector-match @gol
213 -Wundeclared-selector}
215 @item Language Independent Options
216 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
217 @gccoptlist{-fmessage-length=@var{n} @gol
218 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
219 -fdiagnostics-show-option}
221 @item Warning Options
222 @xref{Warning Options,,Options to Request or Suppress Warnings}.
223 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
224 -w -Wextra -Wall -Waggregate-return -Walways-true -Wno-attributes @gol
225 -Wc++-compat -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
226 -Wconversion -Wno-deprecated-declarations @gol
227 -Wdisabled-optimization -Wno-div-by-zero -Wno-endif-labels @gol
228 -Werror -Werror-* -Werror-implicit-function-declaration @gol
229 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
230 -Wno-format-extra-args -Wformat-nonliteral @gol
231 -Wformat-security -Wformat-y2k @gol
232 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
233 -Wimport -Wno-import -Winit-self -Winline @gol
234 -Wno-int-to-pointer-cast @gol
235 -Wno-invalid-offsetof -Winvalid-pch @gol
236 -Wlarger-than-@var{len} -Wunsafe-loop-optimizations -Wlong-long @gol
237 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
238 -Wmissing-format-attribute -Wmissing-include-dirs @gol
239 -Wmissing-noreturn @gol
240 -Wno-multichar -Wnonnull -Wno-overflow @gol
241 -Woverlength-strings -Wpacked -Wpadded @gol
242 -Wparentheses -Wpointer-arith -Wno-pointer-to-int-cast @gol
243 -Wredundant-decls @gol
244 -Wreturn-type -Wsequence-point -Wshadow @gol
245 -Wsign-compare -Wstack-protector @gol
246 -Wstrict-aliasing -Wstrict-aliasing=2 @gol
247 -Wstring-literal-comparison @gol
248 -Wswitch -Wswitch-default -Wswitch-enum @gol
249 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
250 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
251 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
252 -Wunused-value -Wunused-variable -Wvariadic-macros @gol
253 -Wvolatile-register-var -Wwrite-strings}
255 @item C-only Warning Options
256 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
257 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
258 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
259 -Wdeclaration-after-statement -Wpointer-sign}
261 @item Debugging Options
262 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
263 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
264 -fdump-noaddr -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
265 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
266 -fdump-ipa-all -fdump-ipa-cgraph @gol
268 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
269 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
270 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
271 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
273 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
274 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
275 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
276 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
277 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
278 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
279 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
280 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
281 -fdump-tree-nrv -fdump-tree-vect @gol
282 -fdump-tree-sink @gol
283 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
284 -fdump-tree-salias @gol
285 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
286 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
287 -ftree-vectorizer-verbose=@var{n} @gol
288 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
289 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
290 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
291 -fmem-report -fprofile-arcs @gol
292 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
293 -ftest-coverage -ftime-report -fvar-tracking @gol
294 -g -g@var{level} -gcoff -gdwarf-2 @gol
295 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
296 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
297 -print-multi-directory -print-multi-lib @gol
298 -print-prog-name=@var{program} -print-search-dirs -Q @gol
301 @item Optimization Options
302 @xref{Optimize Options,,Options that Control Optimization}.
303 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
304 -falign-labels=@var{n} -falign-loops=@var{n} @gol
305 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
306 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
307 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
308 -fcaller-saves -fcprop-registers -fcse-follow-jumps @gol
309 -fcse-skip-blocks -fcx-limited-range -fdata-sections @gol
310 -fdelayed-branch -fdelete-null-pointer-checks -fearly-inlining @gol
311 -fexpensive-optimizations -ffast-math -ffloat-store @gol
312 -fforce-addr -fforward-propagate -ffunction-sections @gol
313 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
314 -fcrossjumping -fif-conversion -fif-conversion2 @gol
315 -finline-functions -finline-functions-called-once @gol
316 -finline-limit=@var{n} -fkeep-inline-functions @gol
317 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
318 -fmodulo-sched -fno-branch-count-reg @gol
319 -fno-default-inline -fno-defer-pop -fmove-loop-invariants @gol
320 -fno-function-cse -fno-guess-branch-probability @gol
321 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
322 -funsafe-math-optimizations -funsafe-loop-optimizations -ffinite-math-only @gol
323 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
324 -fomit-frame-pointer -foptimize-register-move @gol
325 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
326 -fprofile-generate -fprofile-use @gol
327 -fregmove -frename-registers @gol
328 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
329 -frerun-cse-after-loop @gol
330 -frounding-math -frtl-abstract-sequences @gol
331 -fschedule-insns -fschedule-insns2 @gol
332 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
333 -fsched-spec-load-dangerous @gol
334 -fsched-stalled-insns=@var{n} -fsched-stalled-insns-dep=@var{n} @gol
335 -fsched2-use-superblocks @gol
336 -fsched2-use-traces -fsee -freschedule-modulo-scheduled-loops @gol
337 -fsection-anchors -fsignaling-nans -fsingle-precision-constant @gol
338 -fstack-protector -fstack-protector-all @gol
339 -fstrict-aliasing -ftracer -fthread-jumps @gol
340 -funroll-all-loops -funroll-loops -fpeel-loops @gol
341 -fsplit-ivs-in-unroller -funswitch-loops @gol
342 -fvariable-expansion-in-unroller @gol
343 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
344 -ftree-loop-linear -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
345 -ftree-dominator-opts -ftree-dse -ftree-copyrename -ftree-sink @gol
346 -ftree-ch -ftree-sra -ftree-ter -ftree-lrs -ftree-fre -ftree-vectorize @gol
347 -ftree-vect-loop-version -ftree-salias -fipa-pta -fweb @gol
348 -ftree-copy-prop -ftree-store-ccp -ftree-store-copy-prop -fwhole-program @gol
349 --param @var{name}=@var{value}
350 -O -O0 -O1 -O2 -O3 -Os}
352 @item Preprocessor Options
353 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
354 @gccoptlist{-A@var{question}=@var{answer} @gol
355 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
356 -C -dD -dI -dM -dN @gol
357 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
358 -idirafter @var{dir} @gol
359 -include @var{file} -imacros @var{file} @gol
360 -iprefix @var{file} -iwithprefix @var{dir} @gol
361 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
362 -imultilib @var{dir} -isysroot @var{dir} @gol
363 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
364 -P -fworking-directory -remap @gol
365 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
366 -Xpreprocessor @var{option}}
368 @item Assembler Option
369 @xref{Assembler Options,,Passing Options to the Assembler}.
370 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
373 @xref{Link Options,,Options for Linking}.
374 @gccoptlist{@var{object-file-name} -l@var{library} @gol
375 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
376 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
377 -Wl,@var{option} -Xlinker @var{option} @gol
380 @item Directory Options
381 @xref{Directory Options,,Options for Directory Search}.
382 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
383 -specs=@var{file} -I- --sysroot=@var{dir}}
386 @c I wrote this xref this way to avoid overfull hbox. -- rms
387 @xref{Target Options}.
388 @gccoptlist{-V @var{version} -b @var{machine}}
390 @item Machine Dependent Options
391 @xref{Submodel Options,,Hardware Models and Configurations}.
392 @c This list is ordered alphanumerically by subsection name.
393 @c Try and put the significant identifier (CPU or system) first,
394 @c so users have a clue at guessing where the ones they want will be.
397 @gccoptlist{-EB -EL @gol
398 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
399 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
402 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
403 -mabi=@var{name} @gol
404 -mapcs-stack-check -mno-apcs-stack-check @gol
405 -mapcs-float -mno-apcs-float @gol
406 -mapcs-reentrant -mno-apcs-reentrant @gol
407 -msched-prolog -mno-sched-prolog @gol
408 -mlittle-endian -mbig-endian -mwords-little-endian @gol
409 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
410 -mthumb-interwork -mno-thumb-interwork @gol
411 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
412 -mstructure-size-boundary=@var{n} @gol
413 -mabort-on-noreturn @gol
414 -mlong-calls -mno-long-calls @gol
415 -msingle-pic-base -mno-single-pic-base @gol
416 -mpic-register=@var{reg} @gol
417 -mnop-fun-dllimport @gol
418 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
419 -mpoke-function-name @gol
421 -mtpcs-frame -mtpcs-leaf-frame @gol
422 -mcaller-super-interworking -mcallee-super-interworking @gol
426 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
427 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
429 @emph{Blackfin Options}
430 @gccoptlist{-momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
431 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
432 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
433 -mno-id-shared-library -mshared-library-id=@var{n} @gol
434 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
435 -msep-data -mno-sep-data -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 -minline-ic_invalidate -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}
733 @gccoptlist{-mwarn-reloc -merror-reloc @gol
734 -msafe-dma -munsafe-dma @gol
736 -msmall-mem -mlarge-mem -mstdmain @gol
737 -mfixed-range=@var{register-range}}
739 @emph{System V Options}
740 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
742 @emph{TMS320C3x/C4x Options}
743 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
744 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
745 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
746 -mparallel-insns -mparallel-mpy -mpreserve-float}
749 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
750 -mprolog-function -mno-prolog-function -mspace @gol
751 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
752 -mapp-regs -mno-app-regs @gol
753 -mdisable-callt -mno-disable-callt @gol
759 @gccoptlist{-mg -mgnu -munix}
761 @emph{x86-64 Options}
762 See i386 and x86-64 Options.
764 @emph{Xstormy16 Options}
767 @emph{Xtensa Options}
768 @gccoptlist{-mconst16 -mno-const16 @gol
769 -mfused-madd -mno-fused-madd @gol
770 -mtext-section-literals -mno-text-section-literals @gol
771 -mtarget-align -mno-target-align @gol
772 -mlongcalls -mno-longcalls}
774 @emph{zSeries Options}
775 See S/390 and zSeries Options.
777 @item Code Generation Options
778 @xref{Code Gen Options,,Options for Code Generation Conventions}.
779 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
780 -ffixed-@var{reg} -fexceptions @gol
781 -fnon-call-exceptions -funwind-tables @gol
782 -fasynchronous-unwind-tables @gol
783 -finhibit-size-directive -finstrument-functions @gol
784 -fno-common -fno-ident @gol
785 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
786 -fno-jump-tables @gol
787 -frecord-gcc-switches @gol
788 -freg-struct-return -fshort-enums @gol
789 -fshort-double -fshort-wchar @gol
790 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
791 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
792 -fargument-alias -fargument-noalias @gol
793 -fargument-noalias-global -fargument-noalias-anything
794 -fleading-underscore -ftls-model=@var{model} @gol
795 -ftrapv -fwrapv -fbounds-check @gol
800 * Overall Options:: Controlling the kind of output:
801 an executable, object files, assembler files,
802 or preprocessed source.
803 * C Dialect Options:: Controlling the variant of C language compiled.
804 * C++ Dialect Options:: Variations on C++.
805 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
807 * Language Independent Options:: Controlling how diagnostics should be
809 * Warning Options:: How picky should the compiler be?
810 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
811 * Optimize Options:: How much optimization?
812 * Preprocessor Options:: Controlling header files and macro definitions.
813 Also, getting dependency information for Make.
814 * Assembler Options:: Passing options to the assembler.
815 * Link Options:: Specifying libraries and so on.
816 * Directory Options:: Where to find header files and libraries.
817 Where to find the compiler executable files.
818 * Spec Files:: How to pass switches to sub-processes.
819 * Target Options:: Running a cross-compiler, or an old version of GCC.
822 @node Overall Options
823 @section Options Controlling the Kind of Output
825 Compilation can involve up to four stages: preprocessing, compilation
826 proper, assembly and linking, always in that order. GCC is capable of
827 preprocessing and compiling several files either into several
828 assembler input files, or into one assembler input file; then each
829 assembler input file produces an object file, and linking combines all
830 the object files (those newly compiled, and those specified as input)
831 into an executable file.
833 @cindex file name suffix
834 For any given input file, the file name suffix determines what kind of
839 C source code which must be preprocessed.
842 C source code which should not be preprocessed.
845 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.
852 Objective-C source code which should not be preprocessed.
856 Objective-C++ source code. Note that you must link with the @file{libobjc}
857 library to make an Objective-C++ program work. Note that @samp{.M} refers
858 to a literal capital M@.
861 Objective-C++ source code which should not be preprocessed.
864 C, C++, Objective-C or Objective-C++ header file to be turned into a
869 @itemx @var{file}.cxx
870 @itemx @var{file}.cpp
871 @itemx @var{file}.CPP
872 @itemx @var{file}.c++
874 C++ source code which must be preprocessed. Note that in @samp{.cxx},
875 the last two letters must both be literally @samp{x}. Likewise,
876 @samp{.C} refers to a literal capital C@.
880 Objective-C++ source code which must be preprocessed.
883 Objective-C++ source code which should not be preprocessed.
887 C++ header file to be turned into a precompiled header.
890 @itemx @var{file}.for
891 @itemx @var{file}.FOR
892 Fixed form Fortran source code which should not be preprocessed.
895 @itemx @var{file}.fpp
896 @itemx @var{file}.FPP
897 Fixed form Fortran source code which must be preprocessed (with the traditional
901 @itemx @var{file}.f95
902 Free form Fortran source code which should not be preprocessed.
905 @itemx @var{file}.F95
906 Free form Fortran source code which must be preprocessed (with the
907 traditional preprocessor).
909 @c FIXME: Descriptions of Java file types.
916 Ada source code file which contains a library unit declaration (a
917 declaration of a package, subprogram, or generic, or a generic
918 instantiation), or a library unit renaming declaration (a package,
919 generic, or subprogram renaming declaration). Such files are also
922 @itemx @var{file}.adb
923 Ada source code file containing a library unit body (a subprogram or
924 package body). Such files are also called @dfn{bodies}.
926 @c GCC also knows about some suffixes for languages not yet included:
937 Assembler code which must be preprocessed.
940 An object file to be fed straight into linking.
941 Any file name with no recognized suffix is treated this way.
945 You can specify the input language explicitly with the @option{-x} option:
948 @item -x @var{language}
949 Specify explicitly the @var{language} for the following input files
950 (rather than letting the compiler choose a default based on the file
951 name suffix). This option applies to all following input files until
952 the next @option{-x} option. Possible values for @var{language} are:
954 c c-header c-cpp-output
955 c++ c++-header c++-cpp-output
956 objective-c objective-c-header objective-c-cpp-output
957 objective-c++ objective-c++-header objective-c++-cpp-output
958 assembler assembler-with-cpp
966 Turn off any specification of a language, so that subsequent files are
967 handled according to their file name suffixes (as they are if @option{-x}
968 has not been used at all).
970 @item -pass-exit-codes
971 @opindex pass-exit-codes
972 Normally the @command{gcc} program will exit with the code of 1 if any
973 phase of the compiler returns a non-success return code. If you specify
974 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
975 numerically highest error produced by any phase that returned an error
976 indication. The C, C++, and Fortran frontends return 4, if an internal
977 compiler error is encountered.
980 If you only want some of the stages of compilation, you can use
981 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
982 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
983 @command{gcc} is to stop. Note that some combinations (for example,
984 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
989 Compile or assemble the source files, but do not link. The linking
990 stage simply is not done. The ultimate output is in the form of an
991 object file for each source file.
993 By default, the object file name for a source file is made by replacing
994 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
996 Unrecognized input files, not requiring compilation or assembly, are
1001 Stop after the stage of compilation proper; do not assemble. The output
1002 is in the form of an assembler code file for each non-assembler input
1005 By default, the assembler file name for a source file is made by
1006 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1008 Input files that don't require compilation are ignored.
1012 Stop after the preprocessing stage; do not run the compiler proper. The
1013 output is in the form of preprocessed source code, which is sent to the
1016 Input files which don't require preprocessing are ignored.
1018 @cindex output file option
1021 Place output in file @var{file}. This applies regardless to whatever
1022 sort of output is being produced, whether it be an executable file,
1023 an object file, an assembler file or preprocessed C code.
1025 If @option{-o} is not specified, the default is to put an executable
1026 file in @file{a.out}, the object file for
1027 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1028 assembler file in @file{@var{source}.s}, a precompiled header file in
1029 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1034 Print (on standard error output) the commands executed to run the stages
1035 of compilation. Also print the version number of the compiler driver
1036 program and of the preprocessor and the compiler proper.
1040 Like @option{-v} except the commands are not executed and all command
1041 arguments are quoted. This is useful for shell scripts to capture the
1042 driver-generated command lines.
1046 Use pipes rather than temporary files for communication between the
1047 various stages of compilation. This fails to work on some systems where
1048 the assembler is unable to read from a pipe; but the GNU assembler has
1053 If you are compiling multiple source files, this option tells the driver
1054 to pass all the source files to the compiler at once (for those
1055 languages for which the compiler can handle this). This will allow
1056 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1057 language for which this is supported is C@. If you pass source files for
1058 multiple languages to the driver, using this option, the driver will invoke
1059 the compiler(s) that support IMA once each, passing each compiler all the
1060 source files appropriate for it. For those languages that do not support
1061 IMA this option will be ignored, and the compiler will be invoked once for
1062 each source file in that language. If you use this option in conjunction
1063 with @option{-save-temps}, the compiler will generate multiple
1065 (one for each source file), but only one (combined) @file{.o} or
1070 Print (on the standard output) a description of the command line options
1071 understood by @command{gcc}. If the @option{-v} option is also specified
1072 then @option{--help} will also be passed on to the various processes
1073 invoked by @command{gcc}, so that they can display the command line options
1074 they accept. If the @option{-Wextra} option is also specified then command
1075 line options which have no documentation associated with them will also
1079 @opindex target-help
1080 Print (on the standard output) a description of target specific command
1081 line options for each tool.
1085 Display the version number and copyrights of the invoked GCC@.
1087 @include @value{srcdir}/../libiberty/at-file.texi
1091 @section Compiling C++ Programs
1093 @cindex suffixes for C++ source
1094 @cindex C++ source file suffixes
1095 C++ source files conventionally use one of the suffixes @samp{.C},
1096 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1097 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1098 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1099 files with these names and compiles them as C++ programs even if you
1100 call the compiler the same way as for compiling C programs (usually
1101 with the name @command{gcc}).
1105 However, the use of @command{gcc} does not add the C++ library.
1106 @command{g++} is a program that calls GCC and treats @samp{.c},
1107 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1108 files unless @option{-x} is used, and automatically specifies linking
1109 against the C++ library. This program is also useful when
1110 precompiling a C header file with a @samp{.h} extension for use in C++
1111 compilations. On many systems, @command{g++} is also installed with
1112 the name @command{c++}.
1114 @cindex invoking @command{g++}
1115 When you compile C++ programs, you may specify many of the same
1116 command-line options that you use for compiling programs in any
1117 language; or command-line options meaningful for C and related
1118 languages; or options that are meaningful only for C++ programs.
1119 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1120 explanations of options for languages related to C@.
1121 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1122 explanations of options that are meaningful only for C++ programs.
1124 @node C Dialect Options
1125 @section Options Controlling C Dialect
1126 @cindex dialect options
1127 @cindex language dialect options
1128 @cindex options, dialect
1130 The following options control the dialect of C (or languages derived
1131 from C, such as C++, Objective-C and Objective-C++) that the compiler
1135 @cindex ANSI support
1139 In C mode, support all ISO C90 programs. In C++ mode,
1140 remove GNU extensions that conflict with ISO C++.
1142 This turns off certain features of GCC that are incompatible with ISO
1143 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1144 such as the @code{asm} and @code{typeof} keywords, and
1145 predefined macros such as @code{unix} and @code{vax} that identify the
1146 type of system you are using. It also enables the undesirable and
1147 rarely used ISO trigraph feature. For the C compiler,
1148 it disables recognition of C++ style @samp{//} comments as well as
1149 the @code{inline} keyword.
1151 The alternate keywords @code{__asm__}, @code{__extension__},
1152 @code{__inline__} and @code{__typeof__} continue to work despite
1153 @option{-ansi}. You would not want to use them in an ISO C program, of
1154 course, but it is useful to put them in header files that might be included
1155 in compilations done with @option{-ansi}. Alternate predefined macros
1156 such as @code{__unix__} and @code{__vax__} are also available, with or
1157 without @option{-ansi}.
1159 The @option{-ansi} option does not cause non-ISO programs to be
1160 rejected gratuitously. For that, @option{-pedantic} is required in
1161 addition to @option{-ansi}. @xref{Warning Options}.
1163 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1164 option is used. Some header files may notice this macro and refrain
1165 from declaring certain functions or defining certain macros that the
1166 ISO standard doesn't call for; this is to avoid interfering with any
1167 programs that might use these names for other things.
1169 Functions which would normally be built in but do not have semantics
1170 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1171 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1172 built-in functions provided by GCC}, for details of the functions
1177 Determine the language standard. This option is currently only
1178 supported when compiling C or C++. A value for this option must be
1179 provided; possible values are
1184 ISO C90 (same as @option{-ansi}).
1186 @item iso9899:199409
1187 ISO C90 as modified in amendment 1.
1193 ISO C99. Note that this standard is not yet fully supported; see
1194 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1195 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1198 Default, ISO C90 plus GNU extensions (including some C99 features).
1202 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1203 this will become the default. The name @samp{gnu9x} is deprecated.
1206 The 1998 ISO C++ standard plus amendments.
1209 The same as @option{-std=c++98} plus GNU extensions. This is the
1210 default for C++ code.
1213 The working draft of the upcoming ISO C++0x standard. This option
1214 enables experimental features that are likely to be included in
1215 C++0x. The working draft is constantly changing, and any feature that is
1216 enabled by this flag may be removed from future versions of GCC if it is
1217 not part of the C++0x standard.
1220 The same as @option{-std=c++0x} plus GNU extensions. As with
1221 @option{-std=c++0x}, this option enables experimental features that may
1222 be removed in future versions of GCC.
1225 Even when this option is not specified, you can still use some of the
1226 features of newer standards in so far as they do not conflict with
1227 previous C standards. For example, you may use @code{__restrict__} even
1228 when @option{-std=c99} is not specified.
1230 The @option{-std} options specifying some version of ISO C have the same
1231 effects as @option{-ansi}, except that features that were not in ISO C90
1232 but are in the specified version (for example, @samp{//} comments and
1233 the @code{inline} keyword in ISO C99) are not disabled.
1235 @xref{Standards,,Language Standards Supported by GCC}, for details of
1236 these standard versions.
1238 @item -aux-info @var{filename}
1240 Output to the given filename prototyped declarations for all functions
1241 declared and/or defined in a translation unit, including those in header
1242 files. This option is silently ignored in any language other than C@.
1244 Besides declarations, the file indicates, in comments, the origin of
1245 each declaration (source file and line), whether the declaration was
1246 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1247 @samp{O} for old, respectively, in the first character after the line
1248 number and the colon), and whether it came from a declaration or a
1249 definition (@samp{C} or @samp{F}, respectively, in the following
1250 character). In the case of function definitions, a K&R-style list of
1251 arguments followed by their declarations is also provided, inside
1252 comments, after the declaration.
1256 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1257 keyword, so that code can use these words as identifiers. You can use
1258 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1259 instead. @option{-ansi} implies @option{-fno-asm}.
1261 In C++, this switch only affects the @code{typeof} keyword, since
1262 @code{asm} and @code{inline} are standard keywords. You may want to
1263 use the @option{-fno-gnu-keywords} flag instead, which has the same
1264 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1265 switch only affects the @code{asm} and @code{typeof} keywords, since
1266 @code{inline} is a standard keyword in ISO C99.
1269 @itemx -fno-builtin-@var{function}
1270 @opindex fno-builtin
1271 @cindex built-in functions
1272 Don't recognize built-in functions that do not begin with
1273 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1274 functions provided by GCC}, for details of the functions affected,
1275 including those which are not built-in functions when @option{-ansi} or
1276 @option{-std} options for strict ISO C conformance are used because they
1277 do not have an ISO standard meaning.
1279 GCC normally generates special code to handle certain built-in functions
1280 more efficiently; for instance, calls to @code{alloca} may become single
1281 instructions that adjust the stack directly, and calls to @code{memcpy}
1282 may become inline copy loops. The resulting code is often both smaller
1283 and faster, but since the function calls no longer appear as such, you
1284 cannot set a breakpoint on those calls, nor can you change the behavior
1285 of the functions by linking with a different library. In addition,
1286 when a function is recognized as a built-in function, GCC may use
1287 information about that function to warn about problems with calls to
1288 that function, or to generate more efficient code, even if the
1289 resulting code still contains calls to that function. For example,
1290 warnings are given with @option{-Wformat} for bad calls to
1291 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1292 known not to modify global memory.
1294 With the @option{-fno-builtin-@var{function}} option
1295 only the built-in function @var{function} is
1296 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1297 function is named this is not built-in in this version of GCC, this
1298 option is ignored. There is no corresponding
1299 @option{-fbuiltin-@var{function}} option; if you wish to enable
1300 built-in functions selectively when using @option{-fno-builtin} or
1301 @option{-ffreestanding}, you may define macros such as:
1304 #define abs(n) __builtin_abs ((n))
1305 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1310 @cindex hosted environment
1312 Assert that compilation takes place in a hosted environment. This implies
1313 @option{-fbuiltin}. A hosted environment is one in which the
1314 entire standard library is available, and in which @code{main} has a return
1315 type of @code{int}. Examples are nearly everything except a kernel.
1316 This is equivalent to @option{-fno-freestanding}.
1318 @item -ffreestanding
1319 @opindex ffreestanding
1320 @cindex hosted environment
1322 Assert that compilation takes place in a freestanding environment. This
1323 implies @option{-fno-builtin}. A freestanding environment
1324 is one in which the standard library may not exist, and program startup may
1325 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1326 This is equivalent to @option{-fno-hosted}.
1328 @xref{Standards,,Language Standards Supported by GCC}, for details of
1329 freestanding and hosted environments.
1333 @cindex openmp parallel
1334 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1335 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1336 compiler generates parallel code according to the OpenMP Application
1337 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.
1339 @item -fms-extensions
1340 @opindex fms-extensions
1341 Accept some non-standard constructs used in Microsoft header files.
1343 Some cases of unnamed fields in structures and unions are only
1344 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1345 fields within structs/unions}, for details.
1349 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1350 options for strict ISO C conformance) implies @option{-trigraphs}.
1352 @item -no-integrated-cpp
1353 @opindex no-integrated-cpp
1354 Performs a compilation in two passes: preprocessing and compiling. This
1355 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1356 @option{-B} option. The user supplied compilation step can then add in
1357 an additional preprocessing step after normal preprocessing but before
1358 compiling. The default is to use the integrated cpp (internal cpp)
1360 The semantics of this option will change if "cc1", "cc1plus", and
1361 "cc1obj" are merged.
1363 @cindex traditional C language
1364 @cindex C language, traditional
1366 @itemx -traditional-cpp
1367 @opindex traditional-cpp
1368 @opindex traditional
1369 Formerly, these options caused GCC to attempt to emulate a pre-standard
1370 C compiler. They are now only supported with the @option{-E} switch.
1371 The preprocessor continues to support a pre-standard mode. See the GNU
1372 CPP manual for details.
1374 @item -fcond-mismatch
1375 @opindex fcond-mismatch
1376 Allow conditional expressions with mismatched types in the second and
1377 third arguments. The value of such an expression is void. This option
1378 is not supported for C++.
1380 @item -funsigned-char
1381 @opindex funsigned-char
1382 Let the type @code{char} be unsigned, like @code{unsigned char}.
1384 Each kind of machine has a default for what @code{char} should
1385 be. It is either like @code{unsigned char} by default or like
1386 @code{signed char} by default.
1388 Ideally, a portable program should always use @code{signed char} or
1389 @code{unsigned char} when it depends on the signedness of an object.
1390 But many programs have been written to use plain @code{char} and
1391 expect it to be signed, or expect it to be unsigned, depending on the
1392 machines they were written for. This option, and its inverse, let you
1393 make such a program work with the opposite default.
1395 The type @code{char} is always a distinct type from each of
1396 @code{signed char} or @code{unsigned char}, even though its behavior
1397 is always just like one of those two.
1400 @opindex fsigned-char
1401 Let the type @code{char} be signed, like @code{signed char}.
1403 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1404 the negative form of @option{-funsigned-char}. Likewise, the option
1405 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1407 @item -fsigned-bitfields
1408 @itemx -funsigned-bitfields
1409 @itemx -fno-signed-bitfields
1410 @itemx -fno-unsigned-bitfields
1411 @opindex fsigned-bitfields
1412 @opindex funsigned-bitfields
1413 @opindex fno-signed-bitfields
1414 @opindex fno-unsigned-bitfields
1415 These options control whether a bit-field is signed or unsigned, when the
1416 declaration does not use either @code{signed} or @code{unsigned}. By
1417 default, such a bit-field is signed, because this is consistent: the
1418 basic integer types such as @code{int} are signed types.
1421 @node C++ Dialect Options
1422 @section Options Controlling C++ Dialect
1424 @cindex compiler options, C++
1425 @cindex C++ options, command line
1426 @cindex options, C++
1427 This section describes the command-line options that are only meaningful
1428 for C++ programs; but you can also use most of the GNU compiler options
1429 regardless of what language your program is in. For example, you
1430 might compile a file @code{firstClass.C} like this:
1433 g++ -g -frepo -O -c firstClass.C
1437 In this example, only @option{-frepo} is an option meant
1438 only for C++ programs; you can use the other options with any
1439 language supported by GCC@.
1441 Here is a list of options that are @emph{only} for compiling C++ programs:
1445 @item -fabi-version=@var{n}
1446 @opindex fabi-version
1447 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1448 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1449 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1450 the version that conforms most closely to the C++ ABI specification.
1451 Therefore, the ABI obtained using version 0 will change as ABI bugs
1454 The default is version 2.
1456 @item -fno-access-control
1457 @opindex fno-access-control
1458 Turn off all access checking. This switch is mainly useful for working
1459 around bugs in the access control code.
1463 Check that the pointer returned by @code{operator new} is non-null
1464 before attempting to modify the storage allocated. This check is
1465 normally unnecessary because the C++ standard specifies that
1466 @code{operator new} will only return @code{0} if it is declared
1467 @samp{throw()}, in which case the compiler will always check the
1468 return value even without this option. In all other cases, when
1469 @code{operator new} has a non-empty exception specification, memory
1470 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1471 @samp{new (nothrow)}.
1473 @item -fconserve-space
1474 @opindex fconserve-space
1475 Put uninitialized or runtime-initialized global variables into the
1476 common segment, as C does. This saves space in the executable at the
1477 cost of not diagnosing duplicate definitions. If you compile with this
1478 flag and your program mysteriously crashes after @code{main()} has
1479 completed, you may have an object that is being destroyed twice because
1480 two definitions were merged.
1482 This option is no longer useful on most targets, now that support has
1483 been added for putting variables into BSS without making them common.
1485 @item -ffriend-injection
1486 @opindex ffriend-injection
1487 Inject friend functions into the enclosing namespace, so that they are
1488 visible outside the scope of the class in which they are declared.
1489 Friend functions were documented to work this way in the old Annotated
1490 C++ Reference Manual, and versions of G++ before 4.1 always worked
1491 that way. However, in ISO C++ a friend function which is not declared
1492 in an enclosing scope can only be found using argument dependent
1493 lookup. This option causes friends to be injected as they were in
1496 This option is for compatibility, and may be removed in a future
1499 @item -fno-elide-constructors
1500 @opindex fno-elide-constructors
1501 The C++ standard allows an implementation to omit creating a temporary
1502 which is only used to initialize another object of the same type.
1503 Specifying this option disables that optimization, and forces G++ to
1504 call the copy constructor in all cases.
1506 @item -fno-enforce-eh-specs
1507 @opindex fno-enforce-eh-specs
1508 Don't generate code to check for violation of exception specifications
1509 at runtime. This option violates the C++ standard, but may be useful
1510 for reducing code size in production builds, much like defining
1511 @samp{NDEBUG}. This does not give user code permission to throw
1512 exceptions in violation of the exception specifications; the compiler
1513 will still optimize based on the specifications, so throwing an
1514 unexpected exception will result in undefined behavior.
1517 @itemx -fno-for-scope
1519 @opindex fno-for-scope
1520 If @option{-ffor-scope} is specified, the scope of variables declared in
1521 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1522 as specified by the C++ standard.
1523 If @option{-fno-for-scope} is specified, the scope of variables declared in
1524 a @i{for-init-statement} extends to the end of the enclosing scope,
1525 as was the case in old versions of G++, and other (traditional)
1526 implementations of C++.
1528 The default if neither flag is given to follow the standard,
1529 but to allow and give a warning for old-style code that would
1530 otherwise be invalid, or have different behavior.
1532 @item -fno-gnu-keywords
1533 @opindex fno-gnu-keywords
1534 Do not recognize @code{typeof} as a keyword, so that code can use this
1535 word as an identifier. You can use the keyword @code{__typeof__} instead.
1536 @option{-ansi} implies @option{-fno-gnu-keywords}.
1538 @item -fno-implicit-templates
1539 @opindex fno-implicit-templates
1540 Never emit code for non-inline templates which are instantiated
1541 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1542 @xref{Template Instantiation}, for more information.
1544 @item -fno-implicit-inline-templates
1545 @opindex fno-implicit-inline-templates
1546 Don't emit code for implicit instantiations of inline templates, either.
1547 The default is to handle inlines differently so that compiles with and
1548 without optimization will need the same set of explicit instantiations.
1550 @item -fno-implement-inlines
1551 @opindex fno-implement-inlines
1552 To save space, do not emit out-of-line copies of inline functions
1553 controlled by @samp{#pragma implementation}. This will cause linker
1554 errors if these functions are not inlined everywhere they are called.
1556 @item -fms-extensions
1557 @opindex fms-extensions
1558 Disable pedantic warnings about constructs used in MFC, such as implicit
1559 int and getting a pointer to member function via non-standard syntax.
1561 @item -fno-nonansi-builtins
1562 @opindex fno-nonansi-builtins
1563 Disable built-in declarations of functions that are not mandated by
1564 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1565 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1567 @item -fno-operator-names
1568 @opindex fno-operator-names
1569 Do not treat the operator name keywords @code{and}, @code{bitand},
1570 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1571 synonyms as keywords.
1573 @item -fno-optional-diags
1574 @opindex fno-optional-diags
1575 Disable diagnostics that the standard says a compiler does not need to
1576 issue. Currently, the only such diagnostic issued by G++ is the one for
1577 a name having multiple meanings within a class.
1580 @opindex fpermissive
1581 Downgrade some diagnostics about nonconformant code from errors to
1582 warnings. Thus, using @option{-fpermissive} will allow some
1583 nonconforming code to compile.
1587 Enable automatic template instantiation at link time. This option also
1588 implies @option{-fno-implicit-templates}. @xref{Template
1589 Instantiation}, for more information.
1593 Disable generation of information about every class with virtual
1594 functions for use by the C++ runtime type identification features
1595 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1596 of the language, you can save some space by using this flag. Note that
1597 exception handling uses the same information, but it will generate it as
1598 needed. The @samp{dynamic_cast} operator can still be used for casts that
1599 do not require runtime type information, i.e. casts to @code{void *} or to
1600 unambiguous base classes.
1604 Emit statistics about front-end processing at the end of the compilation.
1605 This information is generally only useful to the G++ development team.
1607 @item -ftemplate-depth-@var{n}
1608 @opindex ftemplate-depth
1609 Set the maximum instantiation depth for template classes to @var{n}.
1610 A limit on the template instantiation depth is needed to detect
1611 endless recursions during template class instantiation. ANSI/ISO C++
1612 conforming programs must not rely on a maximum depth greater than 17.
1614 @item -fno-threadsafe-statics
1615 @opindex fno-threadsafe-statics
1616 Do not emit the extra code to use the routines specified in the C++
1617 ABI for thread-safe initialization of local statics. You can use this
1618 option to reduce code size slightly in code that doesn't need to be
1621 @item -fuse-cxa-atexit
1622 @opindex fuse-cxa-atexit
1623 Register destructors for objects with static storage duration with the
1624 @code{__cxa_atexit} function rather than the @code{atexit} function.
1625 This option is required for fully standards-compliant handling of static
1626 destructors, but will only work if your C library supports
1627 @code{__cxa_atexit}.
1629 @item -fno-use-cxa-get-exception-ptr
1630 @opindex fno-use-cxa-get-exception-ptr
1631 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1632 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1633 if the runtime routine is not available.
1635 @item -fvisibility-inlines-hidden
1636 @opindex fvisibility-inlines-hidden
1637 This switch declares that the user does not attempt to compare
1638 pointers to inline methods where the addresses of the two functions
1639 were taken in different shared objects.
1641 The effect of this is that GCC may, effectively, mark inline methods with
1642 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1643 appear in the export table of a DSO and do not require a PLT indirection
1644 when used within the DSO@. Enabling this option can have a dramatic effect
1645 on load and link times of a DSO as it massively reduces the size of the
1646 dynamic export table when the library makes heavy use of templates.
1648 The behaviour of this switch is not quite the same as marking the
1649 methods as hidden directly, because it does not affect static variables
1650 local to the function or cause the compiler to deduce that
1651 the function is defined in only one shared object.
1653 You may mark a method as having a visibility explicitly to negate the
1654 effect of the switch for that method. For example, if you do want to
1655 compare pointers to a particular inline method, you might mark it as
1656 having default visibility. Marking the enclosing class with explicit
1657 visibility will have no effect.
1659 Explicitly instantiated inline methods are unaffected by this option
1660 as their linkage might otherwise cross a shared library boundary.
1661 @xref{Template Instantiation}.
1665 Do not use weak symbol support, even if it is provided by the linker.
1666 By default, G++ will use weak symbols if they are available. This
1667 option exists only for testing, and should not be used by end-users;
1668 it will result in inferior code and has no benefits. This option may
1669 be removed in a future release of G++.
1673 Do not search for header files in the standard directories specific to
1674 C++, but do still search the other standard directories. (This option
1675 is used when building the C++ library.)
1678 In addition, these optimization, warning, and code generation options
1679 have meanings only for C++ programs:
1682 @item -fno-default-inline
1683 @opindex fno-default-inline
1684 Do not assume @samp{inline} for functions defined inside a class scope.
1685 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1686 functions will have linkage like inline functions; they just won't be
1689 @item -Wabi @r{(C++ only)}
1691 Warn when G++ generates code that is probably not compatible with the
1692 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1693 all such cases, there are probably some cases that are not warned about,
1694 even though G++ is generating incompatible code. There may also be
1695 cases where warnings are emitted even though the code that is generated
1698 You should rewrite your code to avoid these warnings if you are
1699 concerned about the fact that code generated by G++ may not be binary
1700 compatible with code generated by other compilers.
1702 The known incompatibilities at this point include:
1707 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1708 pack data into the same byte as a base class. For example:
1711 struct A @{ virtual void f(); int f1 : 1; @};
1712 struct B : public A @{ int f2 : 1; @};
1716 In this case, G++ will place @code{B::f2} into the same byte
1717 as@code{A::f1}; other compilers will not. You can avoid this problem
1718 by explicitly padding @code{A} so that its size is a multiple of the
1719 byte size on your platform; that will cause G++ and other compilers to
1720 layout @code{B} identically.
1723 Incorrect handling of tail-padding for virtual bases. G++ does not use
1724 tail padding when laying out virtual bases. For example:
1727 struct A @{ virtual void f(); char c1; @};
1728 struct B @{ B(); char c2; @};
1729 struct C : public A, public virtual B @{@};
1733 In this case, G++ will not place @code{B} into the tail-padding for
1734 @code{A}; other compilers will. You can avoid this problem by
1735 explicitly padding @code{A} so that its size is a multiple of its
1736 alignment (ignoring virtual base classes); that will cause G++ and other
1737 compilers to layout @code{C} identically.
1740 Incorrect handling of bit-fields with declared widths greater than that
1741 of their underlying types, when the bit-fields appear in a union. For
1745 union U @{ int i : 4096; @};
1749 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1750 union too small by the number of bits in an @code{int}.
1753 Empty classes can be placed at incorrect offsets. For example:
1763 struct C : public B, public A @{@};
1767 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1768 it should be placed at offset zero. G++ mistakenly believes that the
1769 @code{A} data member of @code{B} is already at offset zero.
1772 Names of template functions whose types involve @code{typename} or
1773 template template parameters can be mangled incorrectly.
1776 template <typename Q>
1777 void f(typename Q::X) @{@}
1779 template <template <typename> class Q>
1780 void f(typename Q<int>::X) @{@}
1784 Instantiations of these templates may be mangled incorrectly.
1788 @item -Wctor-dtor-privacy @r{(C++ only)}
1789 @opindex Wctor-dtor-privacy
1790 Warn when a class seems unusable because all the constructors or
1791 destructors in that class are private, and it has neither friends nor
1792 public static member functions.
1794 @item -Wnon-virtual-dtor @r{(C++ only)}
1795 @opindex Wnon-virtual-dtor
1796 Warn when a class appears to be polymorphic, thereby requiring a virtual
1797 destructor, yet it declares a non-virtual one. This warning is also
1798 enabled if -Weffc++ is specified.
1800 @item -Wreorder @r{(C++ only)}
1802 @cindex reordering, warning
1803 @cindex warning for reordering of member initializers
1804 Warn when the order of member initializers given in the code does not
1805 match the order in which they must be executed. For instance:
1811 A(): j (0), i (1) @{ @}
1815 The compiler will rearrange the member initializers for @samp{i}
1816 and @samp{j} to match the declaration order of the members, emitting
1817 a warning to that effect. This warning is enabled by @option{-Wall}.
1820 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1823 @item -Weffc++ @r{(C++ only)}
1825 Warn about violations of the following style guidelines from Scott Meyers'
1826 @cite{Effective C++} book:
1830 Item 11: Define a copy constructor and an assignment operator for classes
1831 with dynamically allocated memory.
1834 Item 12: Prefer initialization to assignment in constructors.
1837 Item 14: Make destructors virtual in base classes.
1840 Item 15: Have @code{operator=} return a reference to @code{*this}.
1843 Item 23: Don't try to return a reference when you must return an object.
1847 Also warn about violations of the following style guidelines from
1848 Scott Meyers' @cite{More Effective C++} book:
1852 Item 6: Distinguish between prefix and postfix forms of increment and
1853 decrement operators.
1856 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1860 When selecting this option, be aware that the standard library
1861 headers do not obey all of these guidelines; use @samp{grep -v}
1862 to filter out those warnings.
1864 @item -Wno-deprecated @r{(C++ only)}
1865 @opindex Wno-deprecated
1866 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1868 @item -Wstrict-null-sentinel @r{(C++ only)}
1869 @opindex Wstrict-null-sentinel
1870 Warn also about the use of an uncasted @code{NULL} as sentinel. When
1871 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
1872 to @code{__null}. Although it is a null pointer constant not a null pointer,
1873 it is guaranteed to of the same size as a pointer. But this use is
1874 not portable across different compilers.
1876 @item -Wno-non-template-friend @r{(C++ only)}
1877 @opindex Wno-non-template-friend
1878 Disable warnings when non-templatized friend functions are declared
1879 within a template. Since the advent of explicit template specification
1880 support in G++, if the name of the friend is an unqualified-id (i.e.,
1881 @samp{friend foo(int)}), the C++ language specification demands that the
1882 friend declare or define an ordinary, nontemplate function. (Section
1883 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1884 could be interpreted as a particular specialization of a templatized
1885 function. Because this non-conforming behavior is no longer the default
1886 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1887 check existing code for potential trouble spots and is on by default.
1888 This new compiler behavior can be turned off with
1889 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1890 but disables the helpful warning.
1892 @item -Wold-style-cast @r{(C++ only)}
1893 @opindex Wold-style-cast
1894 Warn if an old-style (C-style) cast to a non-void type is used within
1895 a C++ program. The new-style casts (@samp{dynamic_cast},
1896 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
1897 less vulnerable to unintended effects and much easier to search for.
1899 @item -Woverloaded-virtual @r{(C++ only)}
1900 @opindex Woverloaded-virtual
1901 @cindex overloaded virtual fn, warning
1902 @cindex warning for overloaded virtual fn
1903 Warn when a function declaration hides virtual functions from a
1904 base class. For example, in:
1911 struct B: public A @{
1916 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1924 will fail to compile.
1926 @item -Wno-pmf-conversions @r{(C++ only)}
1927 @opindex Wno-pmf-conversions
1928 Disable the diagnostic for converting a bound pointer to member function
1931 @item -Wsign-promo @r{(C++ only)}
1932 @opindex Wsign-promo
1933 Warn when overload resolution chooses a promotion from unsigned or
1934 enumerated type to a signed type, over a conversion to an unsigned type of
1935 the same size. Previous versions of G++ would try to preserve
1936 unsignedness, but the standard mandates the current behavior.
1941 A& operator = (int);
1951 In this example, G++ will synthesize a default @samp{A& operator =
1952 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1955 @node Objective-C and Objective-C++ Dialect Options
1956 @section Options Controlling Objective-C and Objective-C++ Dialects
1958 @cindex compiler options, Objective-C and Objective-C++
1959 @cindex Objective-C and Objective-C++ options, command line
1960 @cindex options, Objective-C and Objective-C++
1961 (NOTE: This manual does not describe the Objective-C and Objective-C++
1962 languages themselves. See @xref{Standards,,Language Standards
1963 Supported by GCC}, for references.)
1965 This section describes the command-line options that are only meaningful
1966 for Objective-C and Objective-C++ programs, but you can also use most of
1967 the language-independent GNU compiler options.
1968 For example, you might compile a file @code{some_class.m} like this:
1971 gcc -g -fgnu-runtime -O -c some_class.m
1975 In this example, @option{-fgnu-runtime} is an option meant only for
1976 Objective-C and Objective-C++ programs; you can use the other options with
1977 any language supported by GCC@.
1979 Note that since Objective-C is an extension of the C language, Objective-C
1980 compilations may also use options specific to the C front-end (e.g.,
1981 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1982 C++-specific options (e.g., @option{-Wabi}).
1984 Here is a list of options that are @emph{only} for compiling Objective-C
1985 and Objective-C++ programs:
1988 @item -fconstant-string-class=@var{class-name}
1989 @opindex fconstant-string-class
1990 Use @var{class-name} as the name of the class to instantiate for each
1991 literal string specified with the syntax @code{@@"@dots{}"}. The default
1992 class name is @code{NXConstantString} if the GNU runtime is being used, and
1993 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1994 @option{-fconstant-cfstrings} option, if also present, will override the
1995 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1996 to be laid out as constant CoreFoundation strings.
1999 @opindex fgnu-runtime
2000 Generate object code compatible with the standard GNU Objective-C
2001 runtime. This is the default for most types of systems.
2003 @item -fnext-runtime
2004 @opindex fnext-runtime
2005 Generate output compatible with the NeXT runtime. This is the default
2006 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2007 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2010 @item -fno-nil-receivers
2011 @opindex fno-nil-receivers
2012 Assume that all Objective-C message dispatches (e.g.,
2013 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2014 is not @code{nil}. This allows for more efficient entry points in the runtime
2015 to be used. Currently, this option is only available in conjunction with
2016 the NeXT runtime on Mac OS X 10.3 and later.
2018 @item -fobjc-call-cxx-cdtors
2019 @opindex fobjc-call-cxx-cdtors
2020 For each Objective-C class, check if any of its instance variables is a
2021 C++ object with a non-trivial default constructor. If so, synthesize a
2022 special @code{- (id) .cxx_construct} instance method that will run
2023 non-trivial default constructors on any such instance variables, in order,
2024 and then return @code{self}. Similarly, check if any instance variable
2025 is a C++ object with a non-trivial destructor, and if so, synthesize a
2026 special @code{- (void) .cxx_destruct} method that will run
2027 all such default destructors, in reverse order.
2029 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2030 thusly generated will only operate on instance variables declared in the
2031 current Objective-C class, and not those inherited from superclasses. It
2032 is the responsibility of the Objective-C runtime to invoke all such methods
2033 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2034 will be invoked by the runtime immediately after a new object
2035 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2036 be invoked immediately before the runtime deallocates an object instance.
2038 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2039 support for invoking the @code{- (id) .cxx_construct} and
2040 @code{- (void) .cxx_destruct} methods.
2042 @item -fobjc-direct-dispatch
2043 @opindex fobjc-direct-dispatch
2044 Allow fast jumps to the message dispatcher. On Darwin this is
2045 accomplished via the comm page.
2047 @item -fobjc-exceptions
2048 @opindex fobjc-exceptions
2049 Enable syntactic support for structured exception handling in Objective-C,
2050 similar to what is offered by C++ and Java. This option is
2051 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2060 @@catch (AnObjCClass *exc) @{
2067 @@catch (AnotherClass *exc) @{
2070 @@catch (id allOthers) @{
2080 The @code{@@throw} statement may appear anywhere in an Objective-C or
2081 Objective-C++ program; when used inside of a @code{@@catch} block, the
2082 @code{@@throw} may appear without an argument (as shown above), in which case
2083 the object caught by the @code{@@catch} will be rethrown.
2085 Note that only (pointers to) Objective-C objects may be thrown and
2086 caught using this scheme. When an object is thrown, it will be caught
2087 by the nearest @code{@@catch} clause capable of handling objects of that type,
2088 analogously to how @code{catch} blocks work in C++ and Java. A
2089 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2090 any and all Objective-C exceptions not caught by previous @code{@@catch}
2093 The @code{@@finally} clause, if present, will be executed upon exit from the
2094 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2095 regardless of whether any exceptions are thrown, caught or rethrown
2096 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2097 of the @code{finally} clause in Java.
2099 There are several caveats to using the new exception mechanism:
2103 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2104 idioms provided by the @code{NSException} class, the new
2105 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2106 systems, due to additional functionality needed in the (NeXT) Objective-C
2110 As mentioned above, the new exceptions do not support handling
2111 types other than Objective-C objects. Furthermore, when used from
2112 Objective-C++, the Objective-C exception model does not interoperate with C++
2113 exceptions at this time. This means you cannot @code{@@throw} an exception
2114 from Objective-C and @code{catch} it in C++, or vice versa
2115 (i.e., @code{throw @dots{} @@catch}).
2118 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2119 blocks for thread-safe execution:
2122 @@synchronized (ObjCClass *guard) @{
2127 Upon entering the @code{@@synchronized} block, a thread of execution shall
2128 first check whether a lock has been placed on the corresponding @code{guard}
2129 object by another thread. If it has, the current thread shall wait until
2130 the other thread relinquishes its lock. Once @code{guard} becomes available,
2131 the current thread will place its own lock on it, execute the code contained in
2132 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2133 making @code{guard} available to other threads).
2135 Unlike Java, Objective-C does not allow for entire methods to be marked
2136 @code{@@synchronized}. Note that throwing exceptions out of
2137 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2138 to be unlocked properly.
2142 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2144 @item -freplace-objc-classes
2145 @opindex freplace-objc-classes
2146 Emit a special marker instructing @command{ld(1)} not to statically link in
2147 the resulting object file, and allow @command{dyld(1)} to load it in at
2148 run time instead. This is used in conjunction with the Fix-and-Continue
2149 debugging mode, where the object file in question may be recompiled and
2150 dynamically reloaded in the course of program execution, without the need
2151 to restart the program itself. Currently, Fix-and-Continue functionality
2152 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2157 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2158 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2159 compile time) with static class references that get initialized at load time,
2160 which improves run-time performance. Specifying the @option{-fzero-link} flag
2161 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2162 to be retained. This is useful in Zero-Link debugging mode, since it allows
2163 for individual class implementations to be modified during program execution.
2167 Dump interface declarations for all classes seen in the source file to a
2168 file named @file{@var{sourcename}.decl}.
2170 @item -Wassign-intercept
2171 @opindex Wassign-intercept
2172 Warn whenever an Objective-C assignment is being intercepted by the
2176 @opindex Wno-protocol
2177 If a class is declared to implement a protocol, a warning is issued for
2178 every method in the protocol that is not implemented by the class. The
2179 default behavior is to issue a warning for every method not explicitly
2180 implemented in the class, even if a method implementation is inherited
2181 from the superclass. If you use the @option{-Wno-protocol} option, then
2182 methods inherited from the superclass are considered to be implemented,
2183 and no warning is issued for them.
2187 Warn if multiple methods of different types for the same selector are
2188 found during compilation. The check is performed on the list of methods
2189 in the final stage of compilation. Additionally, a check is performed
2190 for each selector appearing in a @code{@@selector(@dots{})}
2191 expression, and a corresponding method for that selector has been found
2192 during compilation. Because these checks scan the method table only at
2193 the end of compilation, these warnings are not produced if the final
2194 stage of compilation is not reached, for example because an error is
2195 found during compilation, or because the @option{-fsyntax-only} option is
2198 @item -Wstrict-selector-match
2199 @opindex Wstrict-selector-match
2200 Warn if multiple methods with differing argument and/or return types are
2201 found for a given selector when attempting to send a message using this
2202 selector to a receiver of type @code{id} or @code{Class}. When this flag
2203 is off (which is the default behavior), the compiler will omit such warnings
2204 if any differences found are confined to types which share the same size
2207 @item -Wundeclared-selector
2208 @opindex Wundeclared-selector
2209 Warn if a @code{@@selector(@dots{})} expression referring to an
2210 undeclared selector is found. A selector is considered undeclared if no
2211 method with that name has been declared before the
2212 @code{@@selector(@dots{})} expression, either explicitly in an
2213 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2214 an @code{@@implementation} section. This option always performs its
2215 checks as soon as a @code{@@selector(@dots{})} expression is found,
2216 while @option{-Wselector} only performs its checks in the final stage of
2217 compilation. This also enforces the coding style convention
2218 that methods and selectors must be declared before being used.
2220 @item -print-objc-runtime-info
2221 @opindex print-objc-runtime-info
2222 Generate C header describing the largest structure that is passed by
2227 @node Language Independent Options
2228 @section Options to Control Diagnostic Messages Formatting
2229 @cindex options to control diagnostics formatting
2230 @cindex diagnostic messages
2231 @cindex message formatting
2233 Traditionally, diagnostic messages have been formatted irrespective of
2234 the output device's aspect (e.g.@: its width, @dots{}). The options described
2235 below can be used to control the diagnostic messages formatting
2236 algorithm, e.g.@: how many characters per line, how often source location
2237 information should be reported. Right now, only the C++ front end can
2238 honor these options. However it is expected, in the near future, that
2239 the remaining front ends would be able to digest them correctly.
2242 @item -fmessage-length=@var{n}
2243 @opindex fmessage-length
2244 Try to format error messages so that they fit on lines of about @var{n}
2245 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2246 the front ends supported by GCC@. If @var{n} is zero, then no
2247 line-wrapping will be done; each error message will appear on a single
2250 @opindex fdiagnostics-show-location
2251 @item -fdiagnostics-show-location=once
2252 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2253 reporter to emit @emph{once} source location information; that is, in
2254 case the message is too long to fit on a single physical line and has to
2255 be wrapped, the source location won't be emitted (as prefix) again,
2256 over and over, in subsequent continuation lines. This is the default
2259 @item -fdiagnostics-show-location=every-line
2260 Only meaningful in line-wrapping mode. Instructs the diagnostic
2261 messages reporter to emit the same source location information (as
2262 prefix) for physical lines that result from the process of breaking
2263 a message which is too long to fit on a single line.
2265 @item -fdiagnostics-show-option
2266 @opindex fdiagnostics-show-option
2267 This option instructs the diagnostic machinery to add text to each
2268 diagnostic emitted, which indicates which command line option directly
2269 controls that diagnostic, when such an option is known to the
2270 diagnostic machinery.
2274 @node Warning Options
2275 @section Options to Request or Suppress Warnings
2276 @cindex options to control warnings
2277 @cindex warning messages
2278 @cindex messages, warning
2279 @cindex suppressing warnings
2281 Warnings are diagnostic messages that report constructions which
2282 are not inherently erroneous but which are risky or suggest there
2283 may have been an error.
2285 You can request many specific warnings with options beginning @samp{-W},
2286 for example @option{-Wimplicit} to request warnings on implicit
2287 declarations. Each of these specific warning options also has a
2288 negative form beginning @samp{-Wno-} to turn off warnings;
2289 for example, @option{-Wno-implicit}. This manual lists only one of the
2290 two forms, whichever is not the default.
2292 The following options control the amount and kinds of warnings produced
2293 by GCC; for further, language-specific options also refer to
2294 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2298 @cindex syntax checking
2300 @opindex fsyntax-only
2301 Check the code for syntax errors, but don't do anything beyond that.
2305 Issue all the warnings demanded by strict ISO C and ISO C++;
2306 reject all programs that use forbidden extensions, and some other
2307 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2308 version of the ISO C standard specified by any @option{-std} option used.
2310 Valid ISO C and ISO C++ programs should compile properly with or without
2311 this option (though a rare few will require @option{-ansi} or a
2312 @option{-std} option specifying the required version of ISO C)@. However,
2313 without this option, certain GNU extensions and traditional C and C++
2314 features are supported as well. With this option, they are rejected.
2316 @option{-pedantic} does not cause warning messages for use of the
2317 alternate keywords whose names begin and end with @samp{__}. Pedantic
2318 warnings are also disabled in the expression that follows
2319 @code{__extension__}. However, only system header files should use
2320 these escape routes; application programs should avoid them.
2321 @xref{Alternate Keywords}.
2323 Some users try to use @option{-pedantic} to check programs for strict ISO
2324 C conformance. They soon find that it does not do quite what they want:
2325 it finds some non-ISO practices, but not all---only those for which
2326 ISO C @emph{requires} a diagnostic, and some others for which
2327 diagnostics have been added.
2329 A feature to report any failure to conform to ISO C might be useful in
2330 some instances, but would require considerable additional work and would
2331 be quite different from @option{-pedantic}. We don't have plans to
2332 support such a feature in the near future.
2334 Where the standard specified with @option{-std} represents a GNU
2335 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2336 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2337 extended dialect is based. Warnings from @option{-pedantic} are given
2338 where they are required by the base standard. (It would not make sense
2339 for such warnings to be given only for features not in the specified GNU
2340 C dialect, since by definition the GNU dialects of C include all
2341 features the compiler supports with the given option, and there would be
2342 nothing to warn about.)
2344 @item -pedantic-errors
2345 @opindex pedantic-errors
2346 Like @option{-pedantic}, except that errors are produced rather than
2351 Inhibit all warning messages.
2355 Inhibit warning messages about the use of @samp{#import}.
2357 @item -Wchar-subscripts
2358 @opindex Wchar-subscripts
2359 Warn if an array subscript has type @code{char}. This is a common cause
2360 of error, as programmers often forget that this type is signed on some
2362 This warning is enabled by @option{-Wall}.
2366 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2367 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2368 This warning is enabled by @option{-Wall}.
2370 @item -Wfatal-errors
2371 @opindex Wfatal-errors
2372 This option causes the compiler to abort compilation on the first error
2373 occurred rather than trying to keep going and printing further error
2378 @opindex ffreestanding
2379 @opindex fno-builtin
2380 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2381 the arguments supplied have types appropriate to the format string
2382 specified, and that the conversions specified in the format string make
2383 sense. This includes standard functions, and others specified by format
2384 attributes (@pxref{Function Attributes}), in the @code{printf},
2385 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2386 not in the C standard) families (or other target-specific families).
2387 Which functions are checked without format attributes having been
2388 specified depends on the standard version selected, and such checks of
2389 functions without the attribute specified are disabled by
2390 @option{-ffreestanding} or @option{-fno-builtin}.
2392 The formats are checked against the format features supported by GNU
2393 libc version 2.2. These include all ISO C90 and C99 features, as well
2394 as features from the Single Unix Specification and some BSD and GNU
2395 extensions. Other library implementations may not support all these
2396 features; GCC does not support warning about features that go beyond a
2397 particular library's limitations. However, if @option{-pedantic} is used
2398 with @option{-Wformat}, warnings will be given about format features not
2399 in the selected standard version (but not for @code{strfmon} formats,
2400 since those are not in any version of the C standard). @xref{C Dialect
2401 Options,,Options Controlling C Dialect}.
2403 Since @option{-Wformat} also checks for null format arguments for
2404 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2406 @option{-Wformat} is included in @option{-Wall}. For more control over some
2407 aspects of format checking, the options @option{-Wformat-y2k},
2408 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2409 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2410 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2413 @opindex Wformat-y2k
2414 If @option{-Wformat} is specified, also warn about @code{strftime}
2415 formats which may yield only a two-digit year.
2417 @item -Wno-format-extra-args
2418 @opindex Wno-format-extra-args
2419 If @option{-Wformat} is specified, do not warn about excess arguments to a
2420 @code{printf} or @code{scanf} format function. The C standard specifies
2421 that such arguments are ignored.
2423 Where the unused arguments lie between used arguments that are
2424 specified with @samp{$} operand number specifications, normally
2425 warnings are still given, since the implementation could not know what
2426 type to pass to @code{va_arg} to skip the unused arguments. However,
2427 in the case of @code{scanf} formats, this option will suppress the
2428 warning if the unused arguments are all pointers, since the Single
2429 Unix Specification says that such unused arguments are allowed.
2431 @item -Wno-format-zero-length
2432 @opindex Wno-format-zero-length
2433 If @option{-Wformat} is specified, do not warn about zero-length formats.
2434 The C standard specifies that zero-length formats are allowed.
2436 @item -Wformat-nonliteral
2437 @opindex Wformat-nonliteral
2438 If @option{-Wformat} is specified, also warn if the format string is not a
2439 string literal and so cannot be checked, unless the format function
2440 takes its format arguments as a @code{va_list}.
2442 @item -Wformat-security
2443 @opindex Wformat-security
2444 If @option{-Wformat} is specified, also warn about uses of format
2445 functions that represent possible security problems. At present, this
2446 warns about calls to @code{printf} and @code{scanf} functions where the
2447 format string is not a string literal and there are no format arguments,
2448 as in @code{printf (foo);}. This may be a security hole if the format
2449 string came from untrusted input and contains @samp{%n}. (This is
2450 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2451 in future warnings may be added to @option{-Wformat-security} that are not
2452 included in @option{-Wformat-nonliteral}.)
2456 Enable @option{-Wformat} plus format checks not included in
2457 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2458 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2462 Warn about passing a null pointer for arguments marked as
2463 requiring a non-null value by the @code{nonnull} function attribute.
2465 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2466 can be disabled with the @option{-Wno-nonnull} option.
2468 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2470 Warn about uninitialized variables which are initialized with themselves.
2471 Note this option can only be used with the @option{-Wuninitialized} option,
2472 which in turn only works with @option{-O1} and above.
2474 For example, GCC will warn about @code{i} being uninitialized in the
2475 following snippet only when @option{-Winit-self} has been specified:
2486 @item -Wimplicit-int
2487 @opindex Wimplicit-int
2488 Warn when a declaration does not specify a type.
2489 This warning is enabled by @option{-Wall}.
2491 @item -Wimplicit-function-declaration
2492 @itemx -Werror-implicit-function-declaration
2493 @opindex Wimplicit-function-declaration
2494 @opindex Werror-implicit-function-declaration
2495 Give a warning (or error) whenever a function is used before being
2496 declared. The form @option{-Wno-error-implicit-function-declaration}
2498 This warning is enabled by @option{-Wall} (as a warning, not an error).
2502 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2503 This warning is enabled by @option{-Wall}.
2507 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2508 function with external linkage, returning int, taking either zero
2509 arguments, two, or three arguments of appropriate types.
2510 This warning is enabled by @option{-Wall}.
2512 @item -Wmissing-braces
2513 @opindex Wmissing-braces
2514 Warn if an aggregate or union initializer is not fully bracketed. In
2515 the following example, the initializer for @samp{a} is not fully
2516 bracketed, but that for @samp{b} is fully bracketed.
2519 int a[2][2] = @{ 0, 1, 2, 3 @};
2520 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2523 This warning is enabled by @option{-Wall}.
2525 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2526 @opindex Wmissing-include-dirs
2527 Warn if a user-supplied include directory does not exist.
2530 @opindex Wparentheses
2531 Warn if parentheses are omitted in certain contexts, such
2532 as when there is an assignment in a context where a truth value
2533 is expected, or when operators are nested whose precedence people
2534 often get confused about. Only the warning for an assignment used as
2535 a truth value is supported when compiling C++; the other warnings are
2536 only supported when compiling C@.
2538 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2539 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2540 interpretation from that of ordinary mathematical notation.
2542 Also warn about constructions where there may be confusion to which
2543 @code{if} statement an @code{else} branch belongs. Here is an example of
2558 In C, every @code{else} branch belongs to the innermost possible @code{if}
2559 statement, which in this example is @code{if (b)}. This is often not
2560 what the programmer expected, as illustrated in the above example by
2561 indentation the programmer chose. When there is the potential for this
2562 confusion, GCC will issue a warning when this flag is specified.
2563 To eliminate the warning, add explicit braces around the innermost
2564 @code{if} statement so there is no way the @code{else} could belong to
2565 the enclosing @code{if}. The resulting code would look like this:
2581 This warning is enabled by @option{-Wall}.
2583 @item -Wsequence-point
2584 @opindex Wsequence-point
2585 Warn about code that may have undefined semantics because of violations
2586 of sequence point rules in the C and C++ standards.
2588 The C and C++ standards defines the order in which expressions in a C/C++
2589 program are evaluated in terms of @dfn{sequence points}, which represent
2590 a partial ordering between the execution of parts of the program: those
2591 executed before the sequence point, and those executed after it. These
2592 occur after the evaluation of a full expression (one which is not part
2593 of a larger expression), after the evaluation of the first operand of a
2594 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2595 function is called (but after the evaluation of its arguments and the
2596 expression denoting the called function), and in certain other places.
2597 Other than as expressed by the sequence point rules, the order of
2598 evaluation of subexpressions of an expression is not specified. All
2599 these rules describe only a partial order rather than a total order,
2600 since, for example, if two functions are called within one expression
2601 with no sequence point between them, the order in which the functions
2602 are called is not specified. However, the standards committee have
2603 ruled that function calls do not overlap.
2605 It is not specified when between sequence points modifications to the
2606 values of objects take effect. Programs whose behavior depends on this
2607 have undefined behavior; the C and C++ standards specify that ``Between
2608 the previous and next sequence point an object shall have its stored
2609 value modified at most once by the evaluation of an expression.
2610 Furthermore, the prior value shall be read only to determine the value
2611 to be stored.''. If a program breaks these rules, the results on any
2612 particular implementation are entirely unpredictable.
2614 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2615 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2616 diagnosed by this option, and it may give an occasional false positive
2617 result, but in general it has been found fairly effective at detecting
2618 this sort of problem in programs.
2620 The standard is worded confusingly, therefore there is some debate
2621 over the precise meaning of the sequence point rules in subtle cases.
2622 Links to discussions of the problem, including proposed formal
2623 definitions, may be found on the GCC readings page, at
2624 @w{@uref{http://gcc.gnu.org/readings.html}}.
2626 This warning is enabled by @option{-Wall} for C and C++.
2629 @opindex Wreturn-type
2630 Warn whenever a function is defined with a return-type that defaults to
2631 @code{int}. Also warn about any @code{return} statement with no
2632 return-value in a function whose return-type is not @code{void}.
2634 For C, also warn if the return type of a function has a type qualifier
2635 such as @code{const}. Such a type qualifier has no effect, since the
2636 value returned by a function is not an lvalue. ISO C prohibits
2637 qualified @code{void} return types on function definitions, so such
2638 return types always receive a warning even without this option.
2640 For C++, a function without return type always produces a diagnostic
2641 message, even when @option{-Wno-return-type} is specified. The only
2642 exceptions are @samp{main} and functions defined in system headers.
2644 This warning is enabled by @option{-Wall}.
2648 Warn whenever a @code{switch} statement has an index of enumerated type
2649 and lacks a @code{case} for one or more of the named codes of that
2650 enumeration. (The presence of a @code{default} label prevents this
2651 warning.) @code{case} labels outside the enumeration range also
2652 provoke warnings when this option is used.
2653 This warning is enabled by @option{-Wall}.
2655 @item -Wswitch-default
2656 @opindex Wswitch-switch
2657 Warn whenever a @code{switch} statement does not have a @code{default}
2661 @opindex Wswitch-enum
2662 Warn whenever a @code{switch} statement has an index of enumerated type
2663 and lacks a @code{case} for one or more of the named codes of that
2664 enumeration. @code{case} labels outside the enumeration range also
2665 provoke warnings when this option is used.
2669 Warn if any trigraphs are encountered that might change the meaning of
2670 the program (trigraphs within comments are not warned about).
2671 This warning is enabled by @option{-Wall}.
2673 @item -Wunused-function
2674 @opindex Wunused-function
2675 Warn whenever a static function is declared but not defined or a
2676 non-inline static function is unused.
2677 This warning is enabled by @option{-Wall}.
2679 @item -Wunused-label
2680 @opindex Wunused-label
2681 Warn whenever a label is declared but not used.
2682 This warning is enabled by @option{-Wall}.
2684 To suppress this warning use the @samp{unused} attribute
2685 (@pxref{Variable Attributes}).
2687 @item -Wunused-parameter
2688 @opindex Wunused-parameter
2689 Warn whenever a function parameter is unused aside from its declaration.
2691 To suppress this warning use the @samp{unused} attribute
2692 (@pxref{Variable Attributes}).
2694 @item -Wunused-variable
2695 @opindex Wunused-variable
2696 Warn whenever a local variable or non-constant static variable is unused
2697 aside from its declaration.
2698 This warning is enabled by @option{-Wall}.
2700 To suppress this warning use the @samp{unused} attribute
2701 (@pxref{Variable Attributes}).
2703 @item -Wunused-value
2704 @opindex Wunused-value
2705 Warn whenever a statement computes a result that is explicitly not used.
2706 This warning is enabled by @option{-Wall}.
2708 To suppress this warning cast the expression to @samp{void}.
2712 All the above @option{-Wunused} options combined.
2714 In order to get a warning about an unused function parameter, you must
2715 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2716 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2718 @item -Wuninitialized
2719 @opindex Wuninitialized
2720 Warn if an automatic variable is used without first being initialized or
2721 if a variable may be clobbered by a @code{setjmp} call.
2723 These warnings are possible only in optimizing compilation,
2724 because they require data flow information that is computed only
2725 when optimizing. If you do not specify @option{-O}, you will not get
2726 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
2727 requiring @option{-O}.
2729 If you want to warn about code which uses the uninitialized value of the
2730 variable in its own initializer, use the @option{-Winit-self} option.
2732 These warnings occur for individual uninitialized or clobbered
2733 elements of structure, union or array variables as well as for
2734 variables which are uninitialized or clobbered as a whole. They do
2735 not occur for variables or elements declared @code{volatile}. Because
2736 these warnings depend on optimization, the exact variables or elements
2737 for which there are warnings will depend on the precise optimization
2738 options and version of GCC used.
2740 Note that there may be no warning about a variable that is used only
2741 to compute a value that itself is never used, because such
2742 computations may be deleted by data flow analysis before the warnings
2745 These warnings are made optional because GCC is not smart
2746 enough to see all the reasons why the code might be correct
2747 despite appearing to have an error. Here is one example of how
2768 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2769 always initialized, but GCC doesn't know this. Here is
2770 another common case:
2775 if (change_y) save_y = y, y = new_y;
2777 if (change_y) y = save_y;
2782 This has no bug because @code{save_y} is used only if it is set.
2784 @cindex @code{longjmp} warnings
2785 This option also warns when a non-volatile automatic variable might be
2786 changed by a call to @code{longjmp}. These warnings as well are possible
2787 only in optimizing compilation.
2789 The compiler sees only the calls to @code{setjmp}. It cannot know
2790 where @code{longjmp} will be called; in fact, a signal handler could
2791 call it at any point in the code. As a result, you may get a warning
2792 even when there is in fact no problem because @code{longjmp} cannot
2793 in fact be called at the place which would cause a problem.
2795 Some spurious warnings can be avoided if you declare all the functions
2796 you use that never return as @code{noreturn}. @xref{Function
2799 This warning is enabled by @option{-Wall}.
2801 @item -Wunknown-pragmas
2802 @opindex Wunknown-pragmas
2803 @cindex warning for unknown pragmas
2804 @cindex unknown pragmas, warning
2805 @cindex pragmas, warning of unknown
2806 Warn when a #pragma directive is encountered which is not understood by
2807 GCC@. If this command line option is used, warnings will even be issued
2808 for unknown pragmas in system header files. This is not the case if
2809 the warnings were only enabled by the @option{-Wall} command line option.
2812 @opindex Wno-pragmas
2814 Do not warn about misuses of pragmas, such as incorrect parameters,
2815 invalid syntax, or conflicts between pragmas. See also
2816 @samp{-Wunknown-pragmas}.
2818 @item -Wstrict-aliasing
2819 @opindex Wstrict-aliasing
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. The warning does not catch all
2823 cases, but does attempt to catch the more common pitfalls. It is
2824 included in @option{-Wall}.
2826 @item -Wstrict-aliasing=2
2827 @opindex Wstrict-aliasing=2
2828 This option is only active when @option{-fstrict-aliasing} is active.
2829 It warns about code which might break the strict aliasing rules that the
2830 compiler is using for optimization. This warning catches more cases than
2831 @option{-Wstrict-aliasing}, but it will also give a warning for some ambiguous
2832 cases that are safe.
2836 All of the above @samp{-W} options combined. This enables all the
2837 warnings about constructions that some users consider questionable, and
2838 that are easy to avoid (or modify to prevent the warning), even in
2839 conjunction with macros. This also enables some language-specific
2840 warnings described in @ref{C++ Dialect Options} and
2841 @ref{Objective-C and Objective-C++ Dialect Options}.
2844 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2845 Some of them warn about constructions that users generally do not
2846 consider questionable, but which occasionally you might wish to check
2847 for; others warn about constructions that are necessary or hard to avoid
2848 in some cases, and there is no simple way to modify the code to suppress
2855 (This option used to be called @option{-W}. The older name is still
2856 supported, but the newer name is more descriptive.) Print extra warning
2857 messages for these events:
2861 A function can return either with or without a value. (Falling
2862 off the end of the function body is considered returning without
2863 a value.) For example, this function would evoke such a
2877 An expression-statement or the left-hand side of a comma expression
2878 contains no side effects.
2879 To suppress the warning, cast the unused expression to void.
2880 For example, an expression such as @samp{x[i,j]} will cause a warning,
2881 but @samp{x[(void)i,j]} will not.
2884 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2887 Storage-class specifiers like @code{static} are not the first things in
2888 a declaration. According to the C Standard, this usage is obsolescent.
2891 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2895 A comparison between signed and unsigned values could produce an
2896 incorrect result when the signed value is converted to unsigned.
2897 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2900 An aggregate has an initializer which does not initialize all members.
2901 This warning can be independently controlled by
2902 @option{-Wmissing-field-initializers}.
2905 An initialized field without side effects is overridden when using
2906 designated initializers (@pxref{Designated Inits, , Designated
2907 Initializers}). This warning can be independently controlled by
2908 @option{-Woverride-init}.
2911 A function parameter is declared without a type specifier in K&R-style
2919 An empty body occurs in an @samp{if} or @samp{else} statement.
2922 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2923 @samp{>}, or @samp{>=}.
2926 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2929 Any of several floating-point events that often indicate errors, such as
2930 overflow, underflow, loss of precision, etc.
2932 @item @r{(C++ only)}
2933 An enumerator and a non-enumerator both appear in a conditional expression.
2935 @item @r{(C++ only)}
2936 A non-static reference or non-static @samp{const} member appears in a
2937 class without constructors.
2939 @item @r{(C++ only)}
2940 Ambiguous virtual bases.
2942 @item @r{(C++ only)}
2943 Subscripting an array which has been declared @samp{register}.
2945 @item @r{(C++ only)}
2946 Taking the address of a variable which has been declared @samp{register}.
2948 @item @r{(C++ only)}
2949 A base class is not initialized in a derived class' copy constructor.
2952 @item -Wno-div-by-zero
2953 @opindex Wno-div-by-zero
2954 @opindex Wdiv-by-zero
2955 Do not warn about compile-time integer division by zero. Floating point
2956 division by zero is not warned about, as it can be a legitimate way of
2957 obtaining infinities and NaNs.
2959 @item -Wsystem-headers
2960 @opindex Wsystem-headers
2961 @cindex warnings from system headers
2962 @cindex system headers, warnings from
2963 Print warning messages for constructs found in system header files.
2964 Warnings from system headers are normally suppressed, on the assumption
2965 that they usually do not indicate real problems and would only make the
2966 compiler output harder to read. Using this command line option tells
2967 GCC to emit warnings from system headers as if they occurred in user
2968 code. However, note that using @option{-Wall} in conjunction with this
2969 option will @emph{not} warn about unknown pragmas in system
2970 headers---for that, @option{-Wunknown-pragmas} must also be used.
2973 @opindex Wfloat-equal
2974 Warn if floating point values are used in equality comparisons.
2976 The idea behind this is that sometimes it is convenient (for the
2977 programmer) to consider floating-point values as approximations to
2978 infinitely precise real numbers. If you are doing this, then you need
2979 to compute (by analyzing the code, or in some other way) the maximum or
2980 likely maximum error that the computation introduces, and allow for it
2981 when performing comparisons (and when producing output, but that's a
2982 different problem). In particular, instead of testing for equality, you
2983 would check to see whether the two values have ranges that overlap; and
2984 this is done with the relational operators, so equality comparisons are
2987 @item -Wtraditional @r{(C only)}
2988 @opindex Wtraditional
2989 Warn about certain constructs that behave differently in traditional and
2990 ISO C@. Also warn about ISO C constructs that have no traditional C
2991 equivalent, and/or problematic constructs which should be avoided.
2995 Macro parameters that appear within string literals in the macro body.
2996 In traditional C macro replacement takes place within string literals,
2997 but does not in ISO C@.
3000 In traditional C, some preprocessor directives did not exist.
3001 Traditional preprocessors would only consider a line to be a directive
3002 if the @samp{#} appeared in column 1 on the line. Therefore
3003 @option{-Wtraditional} warns about directives that traditional C
3004 understands but would ignore because the @samp{#} does not appear as the
3005 first character on the line. It also suggests you hide directives like
3006 @samp{#pragma} not understood by traditional C by indenting them. Some
3007 traditional implementations would not recognize @samp{#elif}, so it
3008 suggests avoiding it altogether.
3011 A function-like macro that appears without arguments.
3014 The unary plus operator.
3017 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3018 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3019 constants.) Note, these suffixes appear in macros defined in the system
3020 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3021 Use of these macros in user code might normally lead to spurious
3022 warnings, however GCC's integrated preprocessor has enough context to
3023 avoid warning in these cases.
3026 A function declared external in one block and then used after the end of
3030 A @code{switch} statement has an operand of type @code{long}.
3033 A non-@code{static} function declaration follows a @code{static} one.
3034 This construct is not accepted by some traditional C compilers.
3037 The ISO type of an integer constant has a different width or
3038 signedness from its traditional type. This warning is only issued if
3039 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3040 typically represent bit patterns, are not warned about.
3043 Usage of ISO string concatenation is detected.
3046 Initialization of automatic aggregates.
3049 Identifier conflicts with labels. Traditional C lacks a separate
3050 namespace for labels.
3053 Initialization of unions. If the initializer is zero, the warning is
3054 omitted. This is done under the assumption that the zero initializer in
3055 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3056 initializer warnings and relies on default initialization to zero in the
3060 Conversions by prototypes between fixed/floating point values and vice
3061 versa. The absence of these prototypes when compiling with traditional
3062 C would cause serious problems. This is a subset of the possible
3063 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3066 Use of ISO C style function definitions. This warning intentionally is
3067 @emph{not} issued for prototype declarations or variadic functions
3068 because these ISO C features will appear in your code when using
3069 libiberty's traditional C compatibility macros, @code{PARAMS} and
3070 @code{VPARAMS}. This warning is also bypassed for nested functions
3071 because that feature is already a GCC extension and thus not relevant to
3072 traditional C compatibility.
3075 @item -Wtraditional-conversion @r{(C only)}
3076 @opindex Wtraditional-conversion
3077 Warn if a prototype causes a type conversion that is different from what
3078 would happen to the same argument in the absence of a prototype. This
3079 includes conversions of fixed point to floating and vice versa, and
3080 conversions changing the width or signedness of a fixed point argument
3081 except when the same as the default promotion.
3083 @item -Wdeclaration-after-statement @r{(C only)}
3084 @opindex Wdeclaration-after-statement
3085 Warn when a declaration is found after a statement in a block. This
3086 construct, known from C++, was introduced with ISO C99 and is by default
3087 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3088 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3092 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3094 @item -Wno-endif-labels
3095 @opindex Wno-endif-labels
3096 @opindex Wendif-labels
3097 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3101 Warn whenever a local variable shadows another local variable, parameter or
3102 global variable or whenever a built-in function is shadowed.
3104 @item -Wlarger-than-@var{len}
3105 @opindex Wlarger-than
3106 Warn whenever an object of larger than @var{len} bytes is defined.
3108 @item -Wunsafe-loop-optimizations
3109 @opindex Wunsafe-loop-optimizations
3110 Warn if the loop cannot be optimized because the compiler could not
3111 assume anything on the bounds of the loop indices. With
3112 @option{-funsafe-loop-optimizations} warn if the compiler made
3115 @item -Wpointer-arith
3116 @opindex Wpointer-arith
3117 Warn about anything that depends on the ``size of'' a function type or
3118 of @code{void}. GNU C assigns these types a size of 1, for
3119 convenience in calculations with @code{void *} pointers and pointers
3122 @item -Wbad-function-cast @r{(C only)}
3123 @opindex Wbad-function-cast
3124 Warn whenever a function call is cast to a non-matching type.
3125 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3128 Warn about ISO C constructs that are outside of the common subset of
3129 ISO C and ISO C++, e.g.@: request for implicit conversion from
3130 @code{void *} to a pointer to non-@code{void} type.
3134 Warn whenever a pointer is cast so as to remove a type qualifier from
3135 the target type. For example, warn if a @code{const char *} is cast
3136 to an ordinary @code{char *}.
3139 @opindex Wcast-align
3140 Warn whenever a pointer is cast such that the required alignment of the
3141 target is increased. For example, warn if a @code{char *} is cast to
3142 an @code{int *} on machines where integers can only be accessed at
3143 two- or four-byte boundaries.
3145 @item -Wwrite-strings
3146 @opindex Wwrite-strings
3147 When compiling C, give string constants the type @code{const
3148 char[@var{length}]} so that
3149 copying the address of one into a non-@code{const} @code{char *}
3150 pointer will get a warning; when compiling C++, warn about the
3151 deprecated conversion from string literals to @code{char *}. This
3152 warning, by default, is enabled for C++ programs.
3153 These warnings will help you find at
3154 compile time code that can try to write into a string constant, but
3155 only if you have been very careful about using @code{const} in
3156 declarations and prototypes. Otherwise, it will just be a nuisance;
3157 this is why we did not make @option{-Wall} request these warnings.
3160 @opindex Wconversion
3161 Warn for implicit conversions that may alter a value. This includes
3162 conversions between real and integer, like @code{abs (x)} when
3163 @code{x} is @code{double}; conversions between signed and unsigned,
3164 like @code{unsigned ui = -1}; and conversions to smaller types, like
3165 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3166 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3167 changed by the conversion like in @code{abs (2.0)}.
3169 @item -Wsign-compare
3170 @opindex Wsign-compare
3171 @cindex warning for comparison of signed and unsigned values
3172 @cindex comparison of signed and unsigned values, warning
3173 @cindex signed and unsigned values, comparison warning
3174 Warn when a comparison between signed and unsigned values could produce
3175 an incorrect result when the signed value is converted to unsigned.
3176 This warning is also enabled by @option{-Wextra}; to get the other warnings
3177 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3179 @item -Waggregate-return
3180 @opindex Waggregate-return
3181 Warn if any functions that return structures or unions are defined or
3182 called. (In languages where you can return an array, this also elicits
3186 @opindex Walways-true
3187 Warn about comparisons which are always true such as testing if
3188 unsigned values are greater than or equal to zero. This warning is
3189 enabled by @option{-Wall}.
3191 @item -Wno-attributes
3192 @opindex Wno-attributes
3193 @opindex Wattributes
3194 Do not warn if an unexpected @code{__attribute__} is used, such as
3195 unrecognized attributes, function attributes applied to variables,
3196 etc. This will not stop errors for incorrect use of supported
3199 @item -Wstrict-prototypes @r{(C only)}
3200 @opindex Wstrict-prototypes
3201 Warn if a function is declared or defined without specifying the
3202 argument types. (An old-style function definition is permitted without
3203 a warning if preceded by a declaration which specifies the argument
3206 @item -Wold-style-definition @r{(C only)}
3207 @opindex Wold-style-definition
3208 Warn if an old-style function definition is used. A warning is given
3209 even if there is a previous prototype.
3211 @item -Wmissing-prototypes @r{(C only)}
3212 @opindex Wmissing-prototypes
3213 Warn if a global function is defined without a previous prototype
3214 declaration. This warning is issued even if the definition itself
3215 provides a prototype. The aim is to detect global functions that fail
3216 to be declared in header files.
3218 @item -Wmissing-declarations @r{(C only)}
3219 @opindex Wmissing-declarations
3220 Warn if a global function is defined without a previous declaration.
3221 Do so even if the definition itself provides a prototype.
3222 Use this option to detect global functions that are not declared in
3225 @item -Wmissing-field-initializers
3226 @opindex Wmissing-field-initializers
3229 Warn if a structure's initializer has some fields missing. For
3230 example, the following code would cause such a warning, because
3231 @code{x.h} is implicitly zero:
3234 struct s @{ int f, g, h; @};
3235 struct s x = @{ 3, 4 @};
3238 This option does not warn about designated initializers, so the following
3239 modification would not trigger a warning:
3242 struct s @{ int f, g, h; @};
3243 struct s x = @{ .f = 3, .g = 4 @};
3246 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3247 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3249 @item -Wmissing-noreturn
3250 @opindex Wmissing-noreturn
3251 Warn about functions which might be candidates for attribute @code{noreturn}.
3252 Note these are only possible candidates, not absolute ones. Care should
3253 be taken to manually verify functions actually do not ever return before
3254 adding the @code{noreturn} attribute, otherwise subtle code generation
3255 bugs could be introduced. You will not get a warning for @code{main} in
3256 hosted C environments.
3258 @item -Wmissing-format-attribute
3259 @opindex Wmissing-format-attribute
3261 Warn about function pointers which might be candidates for @code{format}
3262 attributes. Note these are only possible candidates, not absolute ones.
3263 GCC will guess that function pointers with @code{format} attributes that
3264 are used in assignment, initialization, parameter passing or return
3265 statements should have a corresponding @code{format} attribute in the
3266 resulting type. I.e.@: the left-hand side of the assignment or
3267 initialization, the type of the parameter variable, or the return type
3268 of the containing function respectively should also have a @code{format}
3269 attribute to avoid the warning.
3271 GCC will also warn about function definitions which might be
3272 candidates for @code{format} attributes. Again, these are only
3273 possible candidates. GCC will guess that @code{format} attributes
3274 might be appropriate for any function that calls a function like
3275 @code{vprintf} or @code{vscanf}, but this might not always be the
3276 case, and some functions for which @code{format} attributes are
3277 appropriate may not be detected.
3279 @item -Wno-multichar
3280 @opindex Wno-multichar
3282 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3283 Usually they indicate a typo in the user's code, as they have
3284 implementation-defined values, and should not be used in portable code.
3286 @item -Wnormalized=<none|id|nfc|nfkc>
3287 @opindex Wnormalized
3290 @cindex character set, input normalization
3291 In ISO C and ISO C++, two identifiers are different if they are
3292 different sequences of characters. However, sometimes when characters
3293 outside the basic ASCII character set are used, you can have two
3294 different character sequences that look the same. To avoid confusion,
3295 the ISO 10646 standard sets out some @dfn{normalization rules} which
3296 when applied ensure that two sequences that look the same are turned into
3297 the same sequence. GCC can warn you if you are using identifiers which
3298 have not been normalized; this option controls that warning.
3300 There are four levels of warning that GCC supports. The default is
3301 @option{-Wnormalized=nfc}, which warns about any identifier which is
3302 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3303 recommended form for most uses.
3305 Unfortunately, there are some characters which ISO C and ISO C++ allow
3306 in identifiers that when turned into NFC aren't allowable as
3307 identifiers. That is, there's no way to use these symbols in portable
3308 ISO C or C++ and have all your identifiers in NFC.
3309 @option{-Wnormalized=id} suppresses the warning for these characters.
3310 It is hoped that future versions of the standards involved will correct
3311 this, which is why this option is not the default.
3313 You can switch the warning off for all characters by writing
3314 @option{-Wnormalized=none}. You would only want to do this if you
3315 were using some other normalization scheme (like ``D''), because
3316 otherwise you can easily create bugs that are literally impossible to see.
3318 Some characters in ISO 10646 have distinct meanings but look identical
3319 in some fonts or display methodologies, especially once formatting has
3320 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3321 LETTER N'', will display just like a regular @code{n} which has been
3322 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3323 normalization scheme to convert all these into a standard form as
3324 well, and GCC will warn if your code is not in NFKC if you use
3325 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3326 about every identifier that contains the letter O because it might be
3327 confused with the digit 0, and so is not the default, but may be
3328 useful as a local coding convention if the programming environment is
3329 unable to be fixed to display these characters distinctly.
3331 @item -Wno-deprecated-declarations
3332 @opindex Wno-deprecated-declarations
3333 Do not warn about uses of functions (@pxref{Function Attributes}),
3334 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3335 Attributes}) marked as deprecated by using the @code{deprecated}
3339 @opindex Wno-overflow
3340 Do not warn about compile-time overflow in constant expressions.
3342 @item -Woverride-init
3343 @opindex Woverride-init
3346 Warn if an initialized field without side effects is overridden when
3347 using designated initializers (@pxref{Designated Inits, , Designated
3350 This warning is included in @option{-Wextra}. To get other
3351 @option{-Wextra} warnings without this one, use @samp{-Wextra
3352 -Wno-override-init}.
3356 Warn if a structure is given the packed attribute, but the packed
3357 attribute has no effect on the layout or size of the structure.
3358 Such structures may be mis-aligned for little benefit. For
3359 instance, in this code, the variable @code{f.x} in @code{struct bar}
3360 will be misaligned even though @code{struct bar} does not itself
3361 have the packed attribute:
3368 @} __attribute__((packed));
3378 Warn if padding is included in a structure, either to align an element
3379 of the structure or to align the whole structure. Sometimes when this
3380 happens it is possible to rearrange the fields of the structure to
3381 reduce the padding and so make the structure smaller.
3383 @item -Wredundant-decls
3384 @opindex Wredundant-decls
3385 Warn if anything is declared more than once in the same scope, even in
3386 cases where multiple declaration is valid and changes nothing.
3388 @item -Wnested-externs @r{(C only)}
3389 @opindex Wnested-externs
3390 Warn if an @code{extern} declaration is encountered within a function.
3392 @item -Wunreachable-code
3393 @opindex Wunreachable-code
3394 Warn if the compiler detects that code will never be executed.
3396 This option is intended to warn when the compiler detects that at
3397 least a whole line of source code will never be executed, because
3398 some condition is never satisfied or because it is after a
3399 procedure that never returns.
3401 It is possible for this option to produce a warning even though there
3402 are circumstances under which part of the affected line can be executed,
3403 so care should be taken when removing apparently-unreachable code.
3405 For instance, when a function is inlined, a warning may mean that the
3406 line is unreachable in only one inlined copy of the function.
3408 This option is not made part of @option{-Wall} because in a debugging
3409 version of a program there is often substantial code which checks
3410 correct functioning of the program and is, hopefully, unreachable
3411 because the program does work. Another common use of unreachable
3412 code is to provide behavior which is selectable at compile-time.
3416 Warn if a function can not be inlined and it was declared as inline.
3417 Even with this option, the compiler will not warn about failures to
3418 inline functions declared in system headers.
3420 The compiler uses a variety of heuristics to determine whether or not
3421 to inline a function. For example, the compiler takes into account
3422 the size of the function being inlined and the amount of inlining
3423 that has already been done in the current function. Therefore,
3424 seemingly insignificant changes in the source program can cause the
3425 warnings produced by @option{-Winline} to appear or disappear.
3427 @item -Wno-invalid-offsetof @r{(C++ only)}
3428 @opindex Wno-invalid-offsetof
3429 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3430 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3431 to a non-POD type is undefined. In existing C++ implementations,
3432 however, @samp{offsetof} typically gives meaningful results even when
3433 applied to certain kinds of non-POD types. (Such as a simple
3434 @samp{struct} that fails to be a POD type only by virtue of having a
3435 constructor.) This flag is for users who are aware that they are
3436 writing nonportable code and who have deliberately chosen to ignore the
3439 The restrictions on @samp{offsetof} may be relaxed in a future version
3440 of the C++ standard.
3442 @item -Wno-int-to-pointer-cast @r{(C only)}
3443 @opindex Wno-int-to-pointer-cast
3444 Suppress warnings from casts to pointer type of an integer of a
3447 @item -Wno-pointer-to-int-cast @r{(C only)}
3448 @opindex Wno-pointer-to-int-cast
3449 Suppress warnings from casts from a pointer to an integer type of a
3453 @opindex Winvalid-pch
3454 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3455 the search path but can't be used.
3459 @opindex Wno-long-long
3460 Warn if @samp{long long} type is used. This is default. To inhibit
3461 the warning messages, use @option{-Wno-long-long}. Flags
3462 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3463 only when @option{-pedantic} flag is used.
3465 @item -Wvariadic-macros
3466 @opindex Wvariadic-macros
3467 @opindex Wno-variadic-macros
3468 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3469 alternate syntax when in pedantic ISO C99 mode. This is default.
3470 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3472 @item -Wvolatile-register-var
3473 @opindex Wvolatile-register-var
3474 @opindex Wno-volatile-register-var
3475 Warn if a register variable is declared volatile. The volatile
3476 modifier does not inhibit all optimizations that may eliminate reads
3477 and/or writes to register variables.
3479 @item -Wdisabled-optimization
3480 @opindex Wdisabled-optimization
3481 Warn if a requested optimization pass is disabled. This warning does
3482 not generally indicate that there is anything wrong with your code; it
3483 merely indicates that GCC's optimizers were unable to handle the code
3484 effectively. Often, the problem is that your code is too big or too
3485 complex; GCC will refuse to optimize programs when the optimization
3486 itself is likely to take inordinate amounts of time.
3488 @item -Wpointer-sign
3489 @opindex Wpointer-sign
3490 @opindex Wno-pointer-sign
3491 Warn for pointer argument passing or assignment with different signedness.
3492 This option is only supported for C and Objective-C@. It is implied by
3493 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3494 @option{-Wno-pointer-sign}.
3498 Make all warnings into errors.
3502 Make the specified warning into an errors. The specifier for a
3503 warning is appended, for example @option{-Werror=switch} turns the
3504 warnings controlled by @option{-Wswitch} into errors. This switch
3505 takes a negative form, to be used to negate @option{-Werror} for
3506 specific warnings, for example @option{-Wno-error=switch} makes
3507 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3508 is in effect. You can use the @option{-fdiagnostics-show-option}
3509 option to have each controllable warning amended with the option which
3510 controls it, to determine what to use with this option.
3512 Note that specifying @option{-Werror=}@var{foo} automatically implies
3513 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3516 @item -Wstack-protector
3517 @opindex Wstack-protector
3518 This option is only active when @option{-fstack-protector} is active. It
3519 warns about functions that will not be protected against stack smashing.
3521 @item -Wstring-literal-comparison
3522 @opindex Wstring-literal-comparison
3523 Warn about suspicious comparisons to string literal constants. In C,
3524 direct comparisons against the memory address of a string literal, such
3525 as @code{if (x == "abc")}, typically indicate a programmer error, and
3526 even when intentional, result in unspecified behavior and are not portable.
3527 Usually these warnings alert that the programmer intended to use
3528 @code{strcmp}. This warning is enabled by @option{-Wall}.
3530 @item -Woverlength-strings
3531 @opindex Woverlength-strings
3532 Warn about string constants which are longer than the ``minimum
3533 maximum'' length specified in the C standard. Modern compilers
3534 generally allow string constants which are much longer than the
3535 standard's minimum limit, but very portable programs should avoid
3536 using longer strings.
3538 The limit applies @emph{after} string constant concatenation, and does
3539 not count the trailing NUL@. In C89, the limit was 509 characters; in
3540 C99, it was raised to 4095. C++98 does not specify a normative
3541 minimum maximum, so we do not diagnose overlength strings in C++@.
3543 This option is implied by @option{-pedantic}, and can be disabled with
3544 @option{-Wno-overlength-strings}.
3547 @node Debugging Options
3548 @section Options for Debugging Your Program or GCC
3549 @cindex options, debugging
3550 @cindex debugging information options
3552 GCC has various special options that are used for debugging
3553 either your program or GCC:
3558 Produce debugging information in the operating system's native format
3559 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3562 On most systems that use stabs format, @option{-g} enables use of extra
3563 debugging information that only GDB can use; this extra information
3564 makes debugging work better in GDB but will probably make other debuggers
3566 refuse to read the program. If you want to control for certain whether
3567 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3568 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3570 GCC allows you to use @option{-g} with
3571 @option{-O}. The shortcuts taken by optimized code may occasionally
3572 produce surprising results: some variables you declared may not exist
3573 at all; flow of control may briefly move where you did not expect it;
3574 some statements may not be executed because they compute constant
3575 results or their values were already at hand; some statements may
3576 execute in different places because they were moved out of loops.
3578 Nevertheless it proves possible to debug optimized output. This makes
3579 it reasonable to use the optimizer for programs that might have bugs.
3581 The following options are useful when GCC is generated with the
3582 capability for more than one debugging format.
3586 Produce debugging information for use by GDB@. This means to use the
3587 most expressive format available (DWARF 2, stabs, or the native format
3588 if neither of those are supported), including GDB extensions if at all
3593 Produce debugging information in stabs format (if that is supported),
3594 without GDB extensions. This is the format used by DBX on most BSD
3595 systems. On MIPS, Alpha and System V Release 4 systems this option
3596 produces stabs debugging output which is not understood by DBX or SDB@.
3597 On System V Release 4 systems this option requires the GNU assembler.
3599 @item -feliminate-unused-debug-symbols
3600 @opindex feliminate-unused-debug-symbols
3601 Produce debugging information in stabs format (if that is supported),
3602 for only symbols that are actually used.
3604 @item -femit-class-debug-always
3605 Instead of emitting debugging information for a C++ class in only one
3606 object file, emit it in all object files using the class. This option
3607 should be used only with debuggers that are unable to handle the way GCC
3608 normally emits debugging information for classes because using this
3609 option will increase the size of debugging information by as much as a
3614 Produce debugging information in stabs format (if that is supported),
3615 using GNU extensions understood only by the GNU debugger (GDB)@. The
3616 use of these extensions is likely to make other debuggers crash or
3617 refuse to read the program.
3621 Produce debugging information in COFF format (if that is supported).
3622 This is the format used by SDB on most System V systems prior to
3627 Produce debugging information in XCOFF format (if that is supported).
3628 This is the format used by the DBX debugger on IBM RS/6000 systems.
3632 Produce debugging information in XCOFF format (if that is supported),
3633 using GNU extensions understood only by the GNU debugger (GDB)@. The
3634 use of these extensions is likely to make other debuggers crash or
3635 refuse to read the program, and may cause assemblers other than the GNU
3636 assembler (GAS) to fail with an error.
3640 Produce debugging information in DWARF version 2 format (if that is
3641 supported). This is the format used by DBX on IRIX 6. With this
3642 option, GCC uses features of DWARF version 3 when they are useful;
3643 version 3 is upward compatible with version 2, but may still cause
3644 problems for older debuggers.
3648 Produce debugging information in VMS debug format (if that is
3649 supported). This is the format used by DEBUG on VMS systems.
3652 @itemx -ggdb@var{level}
3653 @itemx -gstabs@var{level}
3654 @itemx -gcoff@var{level}
3655 @itemx -gxcoff@var{level}
3656 @itemx -gvms@var{level}
3657 Request debugging information and also use @var{level} to specify how
3658 much information. The default level is 2.
3660 Level 1 produces minimal information, enough for making backtraces in
3661 parts of the program that you don't plan to debug. This includes
3662 descriptions of functions and external variables, but no information
3663 about local variables and no line numbers.
3665 Level 3 includes extra information, such as all the macro definitions
3666 present in the program. Some debuggers support macro expansion when
3667 you use @option{-g3}.
3669 @option{-gdwarf-2} does not accept a concatenated debug level, because
3670 GCC used to support an option @option{-gdwarf} that meant to generate
3671 debug information in version 1 of the DWARF format (which is very
3672 different from version 2), and it would have been too confusing. That
3673 debug format is long obsolete, but the option cannot be changed now.
3674 Instead use an additional @option{-g@var{level}} option to change the
3675 debug level for DWARF2.
3677 @item -feliminate-dwarf2-dups
3678 @opindex feliminate-dwarf2-dups
3679 Compress DWARF2 debugging information by eliminating duplicated
3680 information about each symbol. This option only makes sense when
3681 generating DWARF2 debugging information with @option{-gdwarf-2}.
3683 @cindex @command{prof}
3686 Generate extra code to write profile information suitable for the
3687 analysis program @command{prof}. You must use this option when compiling
3688 the source files you want data about, and you must also use it when
3691 @cindex @command{gprof}
3694 Generate extra code to write profile information suitable for the
3695 analysis program @command{gprof}. You must use this option when compiling
3696 the source files you want data about, and you must also use it when
3701 Makes the compiler print out each function name as it is compiled, and
3702 print some statistics about each pass when it finishes.
3705 @opindex ftime-report
3706 Makes the compiler print some statistics about the time consumed by each
3707 pass when it finishes.
3710 @opindex fmem-report
3711 Makes the compiler print some statistics about permanent memory
3712 allocation when it finishes.
3714 @item -fprofile-arcs
3715 @opindex fprofile-arcs
3716 Add code so that program flow @dfn{arcs} are instrumented. During
3717 execution the program records how many times each branch and call is
3718 executed and how many times it is taken or returns. When the compiled
3719 program exits it saves this data to a file called
3720 @file{@var{auxname}.gcda} for each source file. The data may be used for
3721 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3722 test coverage analysis (@option{-ftest-coverage}). Each object file's
3723 @var{auxname} is generated from the name of the output file, if
3724 explicitly specified and it is not the final executable, otherwise it is
3725 the basename of the source file. In both cases any suffix is removed
3726 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3727 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3728 @xref{Cross-profiling}.
3730 @cindex @command{gcov}
3734 This option is used to compile and link code instrumented for coverage
3735 analysis. The option is a synonym for @option{-fprofile-arcs}
3736 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
3737 linking). See the documentation for those options for more details.
3742 Compile the source files with @option{-fprofile-arcs} plus optimization
3743 and code generation options. For test coverage analysis, use the
3744 additional @option{-ftest-coverage} option. You do not need to profile
3745 every source file in a program.
3748 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3749 (the latter implies the former).
3752 Run the program on a representative workload to generate the arc profile
3753 information. This may be repeated any number of times. You can run
3754 concurrent instances of your program, and provided that the file system
3755 supports locking, the data files will be correctly updated. Also
3756 @code{fork} calls are detected and correctly handled (double counting
3760 For profile-directed optimizations, compile the source files again with
3761 the same optimization and code generation options plus
3762 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3763 Control Optimization}).
3766 For test coverage analysis, use @command{gcov} to produce human readable
3767 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3768 @command{gcov} documentation for further information.
3772 With @option{-fprofile-arcs}, for each function of your program GCC
3773 creates a program flow graph, then finds a spanning tree for the graph.
3774 Only arcs that are not on the spanning tree have to be instrumented: the
3775 compiler adds code to count the number of times that these arcs are
3776 executed. When an arc is the only exit or only entrance to a block, the
3777 instrumentation code can be added to the block; otherwise, a new basic
3778 block must be created to hold the instrumentation code.
3781 @item -ftest-coverage
3782 @opindex ftest-coverage
3783 Produce a notes file that the @command{gcov} code-coverage utility
3784 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3785 show program coverage. Each source file's note file is called
3786 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3787 above for a description of @var{auxname} and instructions on how to
3788 generate test coverage data. Coverage data will match the source files
3789 more closely, if you do not optimize.
3791 @item -d@var{letters}
3792 @item -fdump-rtl-@var{pass}
3794 Says to make debugging dumps during compilation at times specified by
3795 @var{letters}. This is used for debugging the RTL-based passes of the
3796 compiler. The file names for most of the dumps are made by appending a
3797 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3798 from the name of the output file, if explicitly specified and it is not
3799 an executable, otherwise it is the basename of the source file.
3801 Most debug dumps can be enabled either passing a letter to the @option{-d}
3802 option, or with a long @option{-fdump-rtl} switch; here are the possible
3803 letters for use in @var{letters} and @var{pass}, and their meanings:
3808 Annotate the assembler output with miscellaneous debugging information.
3811 @itemx -fdump-rtl-bbro
3813 @opindex fdump-rtl-bbro
3814 Dump after block reordering, to @file{@var{file}.148r.bbro}.
3817 @itemx -fdump-rtl-combine
3819 @opindex fdump-rtl-combine
3820 Dump after instruction combination, to the file @file{@var{file}.129r.combine}.
3823 @itemx -fdump-rtl-ce1
3824 @itemx -fdump-rtl-ce2
3826 @opindex fdump-rtl-ce1
3827 @opindex fdump-rtl-ce2
3828 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3829 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
3830 and @option{-fdump-rtl-ce2} enable dumping after the second if
3831 conversion, to the file @file{@var{file}.130r.ce2}.
3834 @itemx -fdump-rtl-btl
3835 @itemx -fdump-rtl-dbr
3837 @opindex fdump-rtl-btl
3838 @opindex fdump-rtl-dbr
3839 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3840 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
3841 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3842 scheduling, to @file{@var{file}.36.dbr}.
3846 Dump all macro definitions, at the end of preprocessing, in addition to
3850 @itemx -fdump-rtl-ce3
3852 @opindex fdump-rtl-ce3
3853 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
3856 @itemx -fdump-rtl-cfg
3857 @itemx -fdump-rtl-life
3859 @opindex fdump-rtl-cfg
3860 @opindex fdump-rtl-life
3861 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3862 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
3863 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3864 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
3867 @itemx -fdump-rtl-greg
3869 @opindex fdump-rtl-greg
3870 Dump after global register allocation, to @file{@var{file}.139r.greg}.
3873 @itemx -fdump-rtl-gcse
3874 @itemx -fdump-rtl-bypass
3876 @opindex fdump-rtl-gcse
3877 @opindex fdump-rtl-bypass
3878 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3879 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3880 enable dumping after jump bypassing and control flow optimizations, to
3881 @file{@var{file}.115r.bypass}.
3884 @itemx -fdump-rtl-eh
3886 @opindex fdump-rtl-eh
3887 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3890 @itemx -fdump-rtl-sibling
3892 @opindex fdump-rtl-sibling
3893 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
3896 @itemx -fdump-rtl-jump
3898 @opindex fdump-rtl-jump
3899 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
3902 @itemx -fdump-rtl-stack
3904 @opindex fdump-rtl-stack
3905 Dump after conversion from registers to stack, to @file{@var{file}.152r.stack}.
3908 @itemx -fdump-rtl-lreg
3910 @opindex fdump-rtl-lreg
3911 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
3914 @itemx -fdump-rtl-loop2
3916 @opindex fdump-rtl-loop2
3917 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
3918 loop optimization pass, to @file{@var{file}.119r.loop2},
3919 @file{@var{file}.120r.loop2_init},
3920 @file{@var{file}.121r.loop2_invariant}, and
3921 @file{@var{file}.125r.loop2_done}.
3924 @itemx -fdump-rtl-sms
3926 @opindex fdump-rtl-sms
3927 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
3930 @itemx -fdump-rtl-mach
3932 @opindex fdump-rtl-mach
3933 Dump after performing the machine dependent reorganization pass, to
3934 @file{@var{file}.155r.mach}.
3937 @itemx -fdump-rtl-rnreg
3939 @opindex fdump-rtl-rnreg
3940 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
3943 @itemx -fdump-rtl-regmove
3945 @opindex fdump-rtl-regmove
3946 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
3949 @itemx -fdump-rtl-postreload
3951 @opindex fdump-rtl-postreload
3952 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
3955 @itemx -fdump-rtl-expand
3957 @opindex fdump-rtl-expand
3958 Dump after RTL generation, to @file{@var{file}.104r.expand}.
3961 @itemx -fdump-rtl-sched2
3963 @opindex fdump-rtl-sched2
3964 Dump after the second scheduling pass, to @file{@var{file}.149r.sched2}.
3967 @itemx -fdump-rtl-cse
3969 @opindex fdump-rtl-cse
3970 Dump after CSE (including the jump optimization that sometimes follows
3971 CSE), to @file{@var{file}.113r.cse}.
3974 @itemx -fdump-rtl-sched1
3976 @opindex fdump-rtl-sched1
3977 Dump after the first scheduling pass, to @file{@var{file}.136r.sched1}.
3980 @itemx -fdump-rtl-cse2
3982 @opindex fdump-rtl-cse2
3983 Dump after the second CSE pass (including the jump optimization that
3984 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
3987 @itemx -fdump-rtl-tracer
3989 @opindex fdump-rtl-tracer
3990 Dump after running tracer, to @file{@var{file}.118r.tracer}.
3993 @itemx -fdump-rtl-vpt
3994 @itemx -fdump-rtl-vartrack
3996 @opindex fdump-rtl-vpt
3997 @opindex fdump-rtl-vartrack
3998 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
3999 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
4000 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
4001 to @file{@var{file}.154r.vartrack}.
4004 @itemx -fdump-rtl-flow2
4006 @opindex fdump-rtl-flow2
4007 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
4010 @itemx -fdump-rtl-peephole2
4012 @opindex fdump-rtl-peephole2
4013 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
4016 @itemx -fdump-rtl-web
4018 @opindex fdump-rtl-web
4019 Dump after live range splitting, to @file{@var{file}.126r.web}.
4022 @itemx -fdump-rtl-all
4024 @opindex fdump-rtl-all
4025 Produce all the dumps listed above.
4029 Produce a core dump whenever an error occurs.
4033 Print statistics on memory usage, at the end of the run, to
4038 Annotate the assembler output with a comment indicating which
4039 pattern and alternative was used. The length of each instruction is
4044 Dump the RTL in the assembler output as a comment before each instruction.
4045 Also turns on @option{-dp} annotation.
4049 For each of the other indicated dump files (either with @option{-d} or
4050 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4051 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4055 Just generate RTL for a function instead of compiling it. Usually used
4056 with @samp{r} (@option{-fdump-rtl-expand}).
4060 Dump debugging information during parsing, to standard error.
4064 @opindex fdump-noaddr
4065 When doing debugging dumps (see @option{-d} option above), suppress
4066 address output. This makes it more feasible to use diff on debugging
4067 dumps for compiler invocations with different compiler binaries and/or
4068 different text / bss / data / heap / stack / dso start locations.
4070 @item -fdump-unnumbered
4071 @opindex fdump-unnumbered
4072 When doing debugging dumps (see @option{-d} option above), suppress instruction
4073 numbers, line number note and address output. This makes it more feasible to
4074 use diff on debugging dumps for compiler invocations with different
4075 options, in particular with and without @option{-g}.
4077 @item -fdump-translation-unit @r{(C++ only)}
4078 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4079 @opindex fdump-translation-unit
4080 Dump a representation of the tree structure for the entire translation
4081 unit to a file. The file name is made by appending @file{.tu} to the
4082 source file name. If the @samp{-@var{options}} form is used, @var{options}
4083 controls the details of the dump as described for the
4084 @option{-fdump-tree} options.
4086 @item -fdump-class-hierarchy @r{(C++ only)}
4087 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4088 @opindex fdump-class-hierarchy
4089 Dump a representation of each class's hierarchy and virtual function
4090 table layout to a file. The file name is made by appending @file{.class}
4091 to the source file name. If the @samp{-@var{options}} form is used,
4092 @var{options} controls the details of the dump as described for the
4093 @option{-fdump-tree} options.
4095 @item -fdump-ipa-@var{switch}
4097 Control the dumping at various stages of inter-procedural analysis
4098 language tree to a file. The file name is generated by appending a switch
4099 specific suffix to the source file name. The following dumps are possible:
4103 Enables all inter-procedural analysis dumps; currently the only produced
4104 dump is the @samp{cgraph} dump.
4107 Dumps information about call-graph optimization, unused function removal,
4108 and inlining decisions.
4111 @item -fdump-tree-@var{switch}
4112 @itemx -fdump-tree-@var{switch}-@var{options}
4114 Control the dumping at various stages of processing the intermediate
4115 language tree to a file. The file name is generated by appending a switch
4116 specific suffix to the source file name. If the @samp{-@var{options}}
4117 form is used, @var{options} is a list of @samp{-} separated options that
4118 control the details of the dump. Not all options are applicable to all
4119 dumps, those which are not meaningful will be ignored. The following
4120 options are available
4124 Print the address of each node. Usually this is not meaningful as it
4125 changes according to the environment and source file. Its primary use
4126 is for tying up a dump file with a debug environment.
4128 Inhibit dumping of members of a scope or body of a function merely
4129 because that scope has been reached. Only dump such items when they
4130 are directly reachable by some other path. When dumping pretty-printed
4131 trees, this option inhibits dumping the bodies of control structures.
4133 Print a raw representation of the tree. By default, trees are
4134 pretty-printed into a C-like representation.
4136 Enable more detailed dumps (not honored by every dump option).
4138 Enable dumping various statistics about the pass (not honored by every dump
4141 Enable showing basic block boundaries (disabled in raw dumps).
4143 Enable showing virtual operands for every statement.
4145 Enable showing line numbers for statements.
4147 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4149 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4152 The following tree dumps are possible:
4156 Dump before any tree based optimization, to @file{@var{file}.original}.
4159 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4162 Dump after function inlining, to @file{@var{file}.inlined}.
4165 @opindex fdump-tree-gimple
4166 Dump each function before and after the gimplification pass to a file. The
4167 file name is made by appending @file{.gimple} to the source file name.
4170 @opindex fdump-tree-cfg
4171 Dump the control flow graph of each function to a file. The file name is
4172 made by appending @file{.cfg} to the source file name.
4175 @opindex fdump-tree-vcg
4176 Dump the control flow graph of each function to a file in VCG format. The
4177 file name is made by appending @file{.vcg} to the source file name. Note
4178 that if the file contains more than one function, the generated file cannot
4179 be used directly by VCG@. You will need to cut and paste each function's
4180 graph into its own separate file first.
4183 @opindex fdump-tree-ch
4184 Dump each function after copying loop headers. The file name is made by
4185 appending @file{.ch} to the source file name.
4188 @opindex fdump-tree-ssa
4189 Dump SSA related information to a file. The file name is made by appending
4190 @file{.ssa} to the source file name.
4193 @opindex fdump-tree-salias
4194 Dump structure aliasing variable information to a file. This file name
4195 is made by appending @file{.salias} to the source file name.
4198 @opindex fdump-tree-alias
4199 Dump aliasing information for each function. The file name is made by
4200 appending @file{.alias} to the source file name.
4203 @opindex fdump-tree-ccp
4204 Dump each function after CCP@. The file name is made by appending
4205 @file{.ccp} to the source file name.
4208 @opindex fdump-tree-storeccp
4209 Dump each function after STORE-CCP. The file name is made by appending
4210 @file{.storeccp} to the source file name.
4213 @opindex fdump-tree-pre
4214 Dump trees after partial redundancy elimination. The file name is made
4215 by appending @file{.pre} to the source file name.
4218 @opindex fdump-tree-fre
4219 Dump trees after full redundancy elimination. The file name is made
4220 by appending @file{.fre} to the source file name.
4223 @opindex fdump-tree-copyprop
4224 Dump trees after copy propagation. The file name is made
4225 by appending @file{.copyprop} to the source file name.
4227 @item store_copyprop
4228 @opindex fdump-tree-store_copyprop
4229 Dump trees after store copy-propagation. The file name is made
4230 by appending @file{.store_copyprop} to the source file name.
4233 @opindex fdump-tree-dce
4234 Dump each function after dead code elimination. The file name is made by
4235 appending @file{.dce} to the source file name.
4238 @opindex fdump-tree-mudflap
4239 Dump each function after adding mudflap instrumentation. The file name is
4240 made by appending @file{.mudflap} to the source file name.
4243 @opindex fdump-tree-sra
4244 Dump each function after performing scalar replacement of aggregates. The
4245 file name is made by appending @file{.sra} to the source file name.
4248 @opindex fdump-tree-sink
4249 Dump each function after performing code sinking. The file name is made
4250 by appending @file{.sink} to the source file name.
4253 @opindex fdump-tree-dom
4254 Dump each function after applying dominator tree optimizations. The file
4255 name is made by appending @file{.dom} to the source file name.
4258 @opindex fdump-tree-dse
4259 Dump each function after applying dead store elimination. The file
4260 name is made by appending @file{.dse} to the source file name.
4263 @opindex fdump-tree-phiopt
4264 Dump each function after optimizing PHI nodes into straightline code. The file
4265 name is made by appending @file{.phiopt} to the source file name.
4268 @opindex fdump-tree-forwprop
4269 Dump each function after forward propagating single use variables. The file
4270 name is made by appending @file{.forwprop} to the source file name.
4273 @opindex fdump-tree-copyrename
4274 Dump each function after applying the copy rename optimization. The file
4275 name is made by appending @file{.copyrename} to the source file name.
4278 @opindex fdump-tree-nrv
4279 Dump each function after applying the named return value optimization on
4280 generic trees. The file name is made by appending @file{.nrv} to the source
4284 @opindex fdump-tree-vect
4285 Dump each function after applying vectorization of loops. The file name is
4286 made by appending @file{.vect} to the source file name.
4289 @opindex fdump-tree-vrp
4290 Dump each function after Value Range Propagation (VRP). The file name
4291 is made by appending @file{.vrp} to the source file name.
4294 @opindex fdump-tree-all
4295 Enable all the available tree dumps with the flags provided in this option.
4298 @item -ftree-vectorizer-verbose=@var{n}
4299 @opindex ftree-vectorizer-verbose
4300 This option controls the amount of debugging output the vectorizer prints.
4301 This information is written to standard error, unless
4302 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4303 in which case it is output to the usual dump listing file, @file{.vect}.
4304 For @var{n}=0 no diagnostic information is reported.
4305 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4306 and the total number of loops that got vectorized.
4307 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4308 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4309 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4310 level that @option{-fdump-tree-vect-stats} uses.
4311 Higher verbosity levels mean either more information dumped for each
4312 reported loop, or same amount of information reported for more loops:
4313 If @var{n}=3, alignment related information is added to the reports.
4314 If @var{n}=4, data-references related information (e.g. memory dependences,
4315 memory access-patterns) is added to the reports.
4316 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4317 that did not pass the first analysis phase (i.e. may not be countable, or
4318 may have complicated control-flow).
4319 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4320 For @var{n}=7, all the information the vectorizer generates during its
4321 analysis and transformation is reported. This is the same verbosity level
4322 that @option{-fdump-tree-vect-details} uses.
4324 @item -frandom-seed=@var{string}
4325 @opindex frandom-string
4326 This option provides a seed that GCC uses when it would otherwise use
4327 random numbers. It is used to generate certain symbol names
4328 that have to be different in every compiled file. It is also used to
4329 place unique stamps in coverage data files and the object files that
4330 produce them. You can use the @option{-frandom-seed} option to produce
4331 reproducibly identical object files.
4333 The @var{string} should be different for every file you compile.
4335 @item -fsched-verbose=@var{n}
4336 @opindex fsched-verbose
4337 On targets that use instruction scheduling, this option controls the
4338 amount of debugging output the scheduler prints. This information is
4339 written to standard error, unless @option{-dS} or @option{-dR} is
4340 specified, in which case it is output to the usual dump
4341 listing file, @file{.sched} or @file{.sched2} respectively. However
4342 for @var{n} greater than nine, the output is always printed to standard
4345 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4346 same information as @option{-dRS}. For @var{n} greater than one, it
4347 also output basic block probabilities, detailed ready list information
4348 and unit/insn info. For @var{n} greater than two, it includes RTL
4349 at abort point, control-flow and regions info. And for @var{n} over
4350 four, @option{-fsched-verbose} also includes dependence info.
4354 Store the usual ``temporary'' intermediate files permanently; place them
4355 in the current directory and name them based on the source file. Thus,
4356 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4357 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4358 preprocessed @file{foo.i} output file even though the compiler now
4359 normally uses an integrated preprocessor.
4361 When used in combination with the @option{-x} command line option,
4362 @option{-save-temps} is sensible enough to avoid over writing an
4363 input source file with the same extension as an intermediate file.
4364 The corresponding intermediate file may be obtained by renaming the
4365 source file before using @option{-save-temps}.
4369 Report the CPU time taken by each subprocess in the compilation
4370 sequence. For C source files, this is the compiler proper and assembler
4371 (plus the linker if linking is done). The output looks like this:
4378 The first number on each line is the ``user time'', that is time spent
4379 executing the program itself. The second number is ``system time'',
4380 time spent executing operating system routines on behalf of the program.
4381 Both numbers are in seconds.
4383 @item -fvar-tracking
4384 @opindex fvar-tracking
4385 Run variable tracking pass. It computes where variables are stored at each
4386 position in code. Better debugging information is then generated
4387 (if the debugging information format supports this information).
4389 It is enabled by default when compiling with optimization (@option{-Os},
4390 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4391 the debug info format supports it.
4393 @item -print-file-name=@var{library}
4394 @opindex print-file-name
4395 Print the full absolute name of the library file @var{library} that
4396 would be used when linking---and don't do anything else. With this
4397 option, GCC does not compile or link anything; it just prints the
4400 @item -print-multi-directory
4401 @opindex print-multi-directory
4402 Print the directory name corresponding to the multilib selected by any
4403 other switches present in the command line. This directory is supposed
4404 to exist in @env{GCC_EXEC_PREFIX}.
4406 @item -print-multi-lib
4407 @opindex print-multi-lib
4408 Print the mapping from multilib directory names to compiler switches
4409 that enable them. The directory name is separated from the switches by
4410 @samp{;}, and each switch starts with an @samp{@@} instead of the
4411 @samp{-}, without spaces between multiple switches. This is supposed to
4412 ease shell-processing.
4414 @item -print-prog-name=@var{program}
4415 @opindex print-prog-name
4416 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4418 @item -print-libgcc-file-name
4419 @opindex print-libgcc-file-name
4420 Same as @option{-print-file-name=libgcc.a}.
4422 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4423 but you do want to link with @file{libgcc.a}. You can do
4426 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4429 @item -print-search-dirs
4430 @opindex print-search-dirs
4431 Print the name of the configured installation directory and a list of
4432 program and library directories @command{gcc} will search---and don't do anything else.
4434 This is useful when @command{gcc} prints the error message
4435 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4436 To resolve this you either need to put @file{cpp0} and the other compiler
4437 components where @command{gcc} expects to find them, or you can set the environment
4438 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4439 Don't forget the trailing @samp{/}.
4440 @xref{Environment Variables}.
4443 @opindex dumpmachine
4444 Print the compiler's target machine (for example,
4445 @samp{i686-pc-linux-gnu})---and don't do anything else.
4448 @opindex dumpversion
4449 Print the compiler version (for example, @samp{3.0})---and don't do
4454 Print the compiler's built-in specs---and don't do anything else. (This
4455 is used when GCC itself is being built.) @xref{Spec Files}.
4457 @item -feliminate-unused-debug-types
4458 @opindex feliminate-unused-debug-types
4459 Normally, when producing DWARF2 output, GCC will emit debugging
4460 information for all types declared in a compilation
4461 unit, regardless of whether or not they are actually used
4462 in that compilation unit. Sometimes this is useful, such as
4463 if, in the debugger, you want to cast a value to a type that is
4464 not actually used in your program (but is declared). More often,
4465 however, this results in a significant amount of wasted space.
4466 With this option, GCC will avoid producing debug symbol output
4467 for types that are nowhere used in the source file being compiled.
4470 @node Optimize Options
4471 @section Options That Control Optimization
4472 @cindex optimize options
4473 @cindex options, optimization
4475 These options control various sorts of optimizations.
4477 Without any optimization option, the compiler's goal is to reduce the
4478 cost of compilation and to make debugging produce the expected
4479 results. Statements are independent: if you stop the program with a
4480 breakpoint between statements, you can then assign a new value to any
4481 variable or change the program counter to any other statement in the
4482 function and get exactly the results you would expect from the source
4485 Turning on optimization flags makes the compiler attempt to improve
4486 the performance and/or code size at the expense of compilation time
4487 and possibly the ability to debug the program.
4489 The compiler performs optimization based on the knowledge it has of
4490 the program. Optimization levels @option{-O} and above, in
4491 particular, enable @emph{unit-at-a-time} mode, which allows the
4492 compiler to consider information gained from later functions in
4493 the file when compiling a function. Compiling multiple files at
4494 once to a single output file in @emph{unit-at-a-time} mode allows
4495 the compiler to use information gained from all of the files when
4496 compiling each of them.
4498 Not all optimizations are controlled directly by a flag. Only
4499 optimizations that have a flag are listed.
4506 Optimize. Optimizing compilation takes somewhat more time, and a lot
4507 more memory for a large function.
4509 With @option{-O}, the compiler tries to reduce code size and execution
4510 time, without performing any optimizations that take a great deal of
4513 @option{-O} turns on the following optimization flags:
4514 @gccoptlist{-fdefer-pop @gol
4515 -fdelayed-branch @gol
4516 -fguess-branch-probability @gol
4517 -fcprop-registers @gol
4518 -fif-conversion @gol
4519 -fif-conversion2 @gol
4522 -ftree-dominator-opts @gol
4527 -ftree-copyrename @gol
4530 -funit-at-a-time @gol
4533 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4534 where doing so does not interfere with debugging.
4538 Optimize even more. GCC performs nearly all supported optimizations
4539 that do not involve a space-speed tradeoff. The compiler does not
4540 perform loop unrolling or function inlining when you specify @option{-O2}.
4541 As compared to @option{-O}, this option increases both compilation time
4542 and the performance of the generated code.
4544 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4545 also turns on the following optimization flags:
4546 @gccoptlist{-fthread-jumps @gol
4548 -foptimize-sibling-calls @gol
4549 -fcse-follow-jumps -fcse-skip-blocks @gol
4550 -fgcse -fgcse-lm @gol
4551 -fexpensive-optimizations @gol
4552 -frerun-cse-after-loop @gol
4555 -fschedule-insns -fschedule-insns2 @gol
4556 -fsched-interblock -fsched-spec @gol
4558 -fstrict-aliasing @gol
4559 -fdelete-null-pointer-checks @gol
4560 -freorder-blocks -freorder-functions @gol
4561 -falign-functions -falign-jumps @gol
4562 -falign-loops -falign-labels @gol
4566 Please note the warning under @option{-fgcse} about
4567 invoking @option{-O2} on programs that use computed gotos.
4571 Optimize yet more. @option{-O3} turns on all optimizations specified by
4572 @option{-O2} and also turns on the @option{-finline-functions},
4573 @option{-funswitch-loops} and @option{-fgcse-after-reload} options.
4577 Do not optimize. This is the default.
4581 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4582 do not typically increase code size. It also performs further
4583 optimizations designed to reduce code size.
4585 @option{-Os} disables the following optimization flags:
4586 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4587 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
4588 -fprefetch-loop-arrays -ftree-vect-loop-version}
4590 If you use multiple @option{-O} options, with or without level numbers,
4591 the last such option is the one that is effective.
4594 Options of the form @option{-f@var{flag}} specify machine-independent
4595 flags. Most flags have both positive and negative forms; the negative
4596 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4597 below, only one of the forms is listed---the one you typically will
4598 use. You can figure out the other form by either removing @samp{no-}
4601 The following options control specific optimizations. They are either
4602 activated by @option{-O} options or are related to ones that are. You
4603 can use the following flags in the rare cases when ``fine-tuning'' of
4604 optimizations to be performed is desired.
4607 @item -fno-default-inline
4608 @opindex fno-default-inline
4609 Do not make member functions inline by default merely because they are
4610 defined inside the class scope (C++ only). Otherwise, when you specify
4611 @w{@option{-O}}, member functions defined inside class scope are compiled
4612 inline by default; i.e., you don't need to add @samp{inline} in front of
4613 the member function name.
4615 @item -fno-defer-pop
4616 @opindex fno-defer-pop
4617 Always pop the arguments to each function call as soon as that function
4618 returns. For machines which must pop arguments after a function call,
4619 the compiler normally lets arguments accumulate on the stack for several
4620 function calls and pops them all at once.
4622 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4626 Force memory operands to be copied into registers before doing
4627 arithmetic on them. This produces better code by making all memory
4628 references potential common subexpressions. When they are not common
4629 subexpressions, instruction combination should eliminate the separate
4630 register-load. This option is now a nop and will be removed in 4.2.
4633 @opindex fforce-addr
4634 Force memory address constants to be copied into registers before
4635 doing arithmetic on them.
4637 @item -fforward-propagate
4638 @opindex fforward-propagate
4639 Perform a forward propagation pass on RTL. The pass tries to combine two
4640 instructions and checks if the result can be simplified. If loop unrolling
4641 is active, two passes are performed and the second is scheduled after
4644 This option is enabled by default at optimization levels @option{-O2},
4645 @option{-O3}, @option{-Os}.
4647 @item -fomit-frame-pointer
4648 @opindex fomit-frame-pointer
4649 Don't keep the frame pointer in a register for functions that
4650 don't need one. This avoids the instructions to save, set up and
4651 restore frame pointers; it also makes an extra register available
4652 in many functions. @strong{It also makes debugging impossible on
4655 On some machines, such as the VAX, this flag has no effect, because
4656 the standard calling sequence automatically handles the frame pointer
4657 and nothing is saved by pretending it doesn't exist. The
4658 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4659 whether a target machine supports this flag. @xref{Registers,,Register
4660 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4662 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4664 @item -foptimize-sibling-calls
4665 @opindex foptimize-sibling-calls
4666 Optimize sibling and tail recursive calls.
4668 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4672 Don't pay attention to the @code{inline} keyword. Normally this option
4673 is used to keep the compiler from expanding any functions inline.
4674 Note that if you are not optimizing, no functions can be expanded inline.
4676 @item -finline-functions
4677 @opindex finline-functions
4678 Integrate all simple functions into their callers. The compiler
4679 heuristically decides which functions are simple enough to be worth
4680 integrating in this way.
4682 If all calls to a given function are integrated, and the function is
4683 declared @code{static}, then the function is normally not output as
4684 assembler code in its own right.
4686 Enabled at level @option{-O3}.
4688 @item -finline-functions-called-once
4689 @opindex finline-functions-called-once
4690 Consider all @code{static} functions called once for inlining into their
4691 caller even if they are not marked @code{inline}. If a call to a given
4692 function is integrated, then the function is not output as assembler code
4695 Enabled if @option{-funit-at-a-time} is enabled.
4697 @item -fearly-inlining
4698 @opindex fearly-inlining
4699 Inline functions marked by @code{always_inline} and functions whose body seems
4700 smaller than the function call overhead early before doing
4701 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
4702 makes profiling significantly cheaper and usually inlining faster on programs
4703 having large chains of nested wrapper functions.
4707 @item -finline-limit=@var{n}
4708 @opindex finline-limit
4709 By default, GCC limits the size of functions that can be inlined. This flag
4710 allows the control of this limit for functions that are explicitly marked as
4711 inline (i.e., marked with the inline keyword or defined within the class
4712 definition in c++). @var{n} is the size of functions that can be inlined in
4713 number of pseudo instructions (not counting parameter handling). The default
4714 value of @var{n} is 600.
4715 Increasing this value can result in more inlined code at
4716 the cost of compilation time and memory consumption. Decreasing usually makes
4717 the compilation faster and less code will be inlined (which presumably
4718 means slower programs). This option is particularly useful for programs that
4719 use inlining heavily such as those based on recursive templates with C++.
4721 Inlining is actually controlled by a number of parameters, which may be
4722 specified individually by using @option{--param @var{name}=@var{value}}.
4723 The @option{-finline-limit=@var{n}} option sets some of these parameters
4727 @item max-inline-insns-single
4728 is set to @var{n}/2.
4729 @item max-inline-insns-auto
4730 is set to @var{n}/2.
4731 @item min-inline-insns
4732 is set to 130 or @var{n}/4, whichever is smaller.
4733 @item max-inline-insns-rtl
4737 See below for a documentation of the individual
4738 parameters controlling inlining.
4740 @emph{Note:} pseudo instruction represents, in this particular context, an
4741 abstract measurement of function's size. In no way does it represent a count
4742 of assembly instructions and as such its exact meaning might change from one
4743 release to an another.
4745 @item -fkeep-inline-functions
4746 @opindex fkeep-inline-functions
4747 In C, emit @code{static} functions that are declared @code{inline}
4748 into the object file, even if the function has been inlined into all
4749 of its callers. This switch does not affect functions using the
4750 @code{extern inline} extension in GNU C89@. In C++, emit any and all
4751 inline functions into the object file.
4753 @item -fkeep-static-consts
4754 @opindex fkeep-static-consts
4755 Emit variables declared @code{static const} when optimization isn't turned
4756 on, even if the variables aren't referenced.
4758 GCC enables this option by default. If you want to force the compiler to
4759 check if the variable was referenced, regardless of whether or not
4760 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4762 @item -fmerge-constants
4763 Attempt to merge identical constants (string constants and floating point
4764 constants) across compilation units.
4766 This option is the default for optimized compilation if the assembler and
4767 linker support it. Use @option{-fno-merge-constants} to inhibit this
4770 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4772 @item -fmerge-all-constants
4773 Attempt to merge identical constants and identical variables.
4775 This option implies @option{-fmerge-constants}. In addition to
4776 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4777 arrays or initialized constant variables with integral or floating point
4778 types. Languages like C or C++ require each non-automatic variable to
4779 have distinct location, so using this option will result in non-conforming
4782 @item -fmodulo-sched
4783 @opindex fmodulo-sched
4784 Perform swing modulo scheduling immediately before the first scheduling
4785 pass. This pass looks at innermost loops and reorders their
4786 instructions by overlapping different iterations.
4788 @item -fno-branch-count-reg
4789 @opindex fno-branch-count-reg
4790 Do not use ``decrement and branch'' instructions on a count register,
4791 but instead generate a sequence of instructions that decrement a
4792 register, compare it against zero, then branch based upon the result.
4793 This option is only meaningful on architectures that support such
4794 instructions, which include x86, PowerPC, IA-64 and S/390.
4796 The default is @option{-fbranch-count-reg}.
4798 @item -fno-function-cse
4799 @opindex fno-function-cse
4800 Do not put function addresses in registers; make each instruction that
4801 calls a constant function contain the function's address explicitly.
4803 This option results in less efficient code, but some strange hacks
4804 that alter the assembler output may be confused by the optimizations
4805 performed when this option is not used.
4807 The default is @option{-ffunction-cse}
4809 @item -fno-zero-initialized-in-bss
4810 @opindex fno-zero-initialized-in-bss
4811 If the target supports a BSS section, GCC by default puts variables that
4812 are initialized to zero into BSS@. This can save space in the resulting
4815 This option turns off this behavior because some programs explicitly
4816 rely on variables going to the data section. E.g., so that the
4817 resulting executable can find the beginning of that section and/or make
4818 assumptions based on that.
4820 The default is @option{-fzero-initialized-in-bss}.
4822 @item -fbounds-check
4823 @opindex fbounds-check
4824 For front-ends that support it, generate additional code to check that
4825 indices used to access arrays are within the declared range. This is
4826 currently only supported by the Java and Fortran front-ends, where
4827 this option defaults to true and false respectively.
4829 @item -fmudflap -fmudflapth -fmudflapir
4833 @cindex bounds checking
4835 For front-ends that support it (C and C++), instrument all risky
4836 pointer/array dereferencing operations, some standard library
4837 string/heap functions, and some other associated constructs with
4838 range/validity tests. Modules so instrumented should be immune to
4839 buffer overflows, invalid heap use, and some other classes of C/C++
4840 programming errors. The instrumentation relies on a separate runtime
4841 library (@file{libmudflap}), which will be linked into a program if
4842 @option{-fmudflap} is given at link time. Run-time behavior of the
4843 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4844 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4847 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4848 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4849 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4850 instrumentation should ignore pointer reads. This produces less
4851 instrumentation (and therefore faster execution) and still provides
4852 some protection against outright memory corrupting writes, but allows
4853 erroneously read data to propagate within a program.
4855 @item -fthread-jumps
4856 @opindex fthread-jumps
4857 Perform optimizations where we check to see if a jump branches to a
4858 location where another comparison subsumed by the first is found. If
4859 so, the first branch is redirected to either the destination of the
4860 second branch or a point immediately following it, depending on whether
4861 the condition is known to be true or false.
4863 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4865 @item -fcse-follow-jumps
4866 @opindex fcse-follow-jumps
4867 In common subexpression elimination, scan through jump instructions
4868 when the target of the jump is not reached by any other path. For
4869 example, when CSE encounters an @code{if} statement with an
4870 @code{else} clause, CSE will follow the jump when the condition
4873 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4875 @item -fcse-skip-blocks
4876 @opindex fcse-skip-blocks
4877 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4878 follow jumps which conditionally skip over blocks. When CSE
4879 encounters a simple @code{if} statement with no else clause,
4880 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4881 body of the @code{if}.
4883 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4885 @item -frerun-cse-after-loop
4886 @opindex frerun-cse-after-loop
4887 Re-run common subexpression elimination after loop optimizations has been
4890 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4894 Perform a global common subexpression elimination pass.
4895 This pass also performs global constant and copy propagation.
4897 @emph{Note:} When compiling a program using computed gotos, a GCC
4898 extension, you may get better runtime performance if you disable
4899 the global common subexpression elimination pass by adding
4900 @option{-fno-gcse} to the command line.
4902 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4906 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4907 attempt to move loads which are only killed by stores into themselves. This
4908 allows a loop containing a load/store sequence to be changed to a load outside
4909 the loop, and a copy/store within the loop.
4911 Enabled by default when gcse is enabled.
4915 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4916 global common subexpression elimination. This pass will attempt to move
4917 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4918 loops containing a load/store sequence can be changed to a load before
4919 the loop and a store after the loop.
4921 Not enabled at any optimization level.
4925 When @option{-fgcse-las} is enabled, the global common subexpression
4926 elimination pass eliminates redundant loads that come after stores to the
4927 same memory location (both partial and full redundancies).
4929 Not enabled at any optimization level.
4931 @item -fgcse-after-reload
4932 @opindex fgcse-after-reload
4933 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4934 pass is performed after reload. The purpose of this pass is to cleanup
4937 @item -funsafe-loop-optimizations
4938 @opindex funsafe-loop-optimizations
4939 If given, the loop optimizer will assume that loop indices do not
4940 overflow, and that the loops with nontrivial exit condition are not
4941 infinite. This enables a wider range of loop optimizations even if
4942 the loop optimizer itself cannot prove that these assumptions are valid.
4943 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
4944 if it finds this kind of loop.
4946 @item -fcrossjumping
4947 @opindex crossjumping
4948 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4949 resulting code may or may not perform better than without cross-jumping.
4951 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4953 @item -fif-conversion
4954 @opindex if-conversion
4955 Attempt to transform conditional jumps into branch-less equivalents. This
4956 include use of conditional moves, min, max, set flags and abs instructions, and
4957 some tricks doable by standard arithmetics. The use of conditional execution
4958 on chips where it is available is controlled by @code{if-conversion2}.
4960 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4962 @item -fif-conversion2
4963 @opindex if-conversion2
4964 Use conditional execution (where available) to transform conditional jumps into
4965 branch-less equivalents.
4967 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4969 @item -fdelete-null-pointer-checks
4970 @opindex fdelete-null-pointer-checks
4971 Use global dataflow analysis to identify and eliminate useless checks
4972 for null pointers. The compiler assumes that dereferencing a null
4973 pointer would have halted the program. If a pointer is checked after
4974 it has already been dereferenced, it cannot be null.
4976 In some environments, this assumption is not true, and programs can
4977 safely dereference null pointers. Use
4978 @option{-fno-delete-null-pointer-checks} to disable this optimization
4979 for programs which depend on that behavior.
4981 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4983 @item -fexpensive-optimizations
4984 @opindex fexpensive-optimizations
4985 Perform a number of minor optimizations that are relatively expensive.
4987 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4989 @item -foptimize-register-move
4991 @opindex foptimize-register-move
4993 Attempt to reassign register numbers in move instructions and as
4994 operands of other simple instructions in order to maximize the amount of
4995 register tying. This is especially helpful on machines with two-operand
4998 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5001 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5003 @item -fdelayed-branch
5004 @opindex fdelayed-branch
5005 If supported for the target machine, attempt to reorder instructions
5006 to exploit instruction slots available after delayed branch
5009 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5011 @item -fschedule-insns
5012 @opindex fschedule-insns
5013 If supported for the target machine, attempt to reorder instructions to
5014 eliminate execution stalls due to required data being unavailable. This
5015 helps machines that have slow floating point or memory load instructions
5016 by allowing other instructions to be issued until the result of the load
5017 or floating point instruction is required.
5019 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5021 @item -fschedule-insns2
5022 @opindex fschedule-insns2
5023 Similar to @option{-fschedule-insns}, but requests an additional pass of
5024 instruction scheduling after register allocation has been done. This is
5025 especially useful on machines with a relatively small number of
5026 registers and where memory load instructions take more than one cycle.
5028 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5030 @item -fno-sched-interblock
5031 @opindex fno-sched-interblock
5032 Don't schedule instructions across basic blocks. This is normally
5033 enabled by default when scheduling before register allocation, i.e.@:
5034 with @option{-fschedule-insns} or at @option{-O2} or higher.
5036 @item -fno-sched-spec
5037 @opindex fno-sched-spec
5038 Don't allow speculative motion of non-load instructions. This is normally
5039 enabled by default when scheduling before register allocation, i.e.@:
5040 with @option{-fschedule-insns} or at @option{-O2} or higher.
5042 @item -fsched-spec-load
5043 @opindex fsched-spec-load
5044 Allow speculative motion of some load instructions. This only makes
5045 sense when scheduling before register allocation, i.e.@: with
5046 @option{-fschedule-insns} or at @option{-O2} or higher.
5048 @item -fsched-spec-load-dangerous
5049 @opindex fsched-spec-load-dangerous
5050 Allow speculative motion of more load instructions. This only makes
5051 sense when scheduling before register allocation, i.e.@: with
5052 @option{-fschedule-insns} or at @option{-O2} or higher.
5054 @item -fsched-stalled-insns=@var{n}
5055 @opindex fsched-stalled-insns
5056 Define how many insns (if any) can be moved prematurely from the queue
5057 of stalled insns into the ready list, during the second scheduling pass.
5059 @item -fsched-stalled-insns-dep=@var{n}
5060 @opindex fsched-stalled-insns-dep
5061 Define how many insn groups (cycles) will be examined for a dependency
5062 on a stalled insn that is candidate for premature removal from the queue
5063 of stalled insns. Has an effect only during the second scheduling pass,
5064 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
5066 @item -fsched2-use-superblocks
5067 @opindex fsched2-use-superblocks
5068 When scheduling after register allocation, do use superblock scheduling
5069 algorithm. Superblock scheduling allows motion across basic block boundaries
5070 resulting on faster schedules. This option is experimental, as not all machine
5071 descriptions used by GCC model the CPU closely enough to avoid unreliable
5072 results from the algorithm.
5074 This only makes sense when scheduling after register allocation, i.e.@: with
5075 @option{-fschedule-insns2} or at @option{-O2} or higher.
5077 @item -fsched2-use-traces
5078 @opindex fsched2-use-traces
5079 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5080 allocation and additionally perform code duplication in order to increase the
5081 size of superblocks using tracer pass. See @option{-ftracer} for details on
5084 This mode should produce faster but significantly longer programs. Also
5085 without @option{-fbranch-probabilities} the traces constructed may not
5086 match the reality and hurt the performance. This only makes
5087 sense when scheduling after register allocation, i.e.@: with
5088 @option{-fschedule-insns2} or at @option{-O2} or higher.
5092 Eliminates redundant extension instructions and move the non redundant
5093 ones to optimal placement using LCM.
5095 @item -freschedule-modulo-scheduled-loops
5096 @opindex fscheduling-in-modulo-scheduled-loops
5097 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
5098 we may want to prevent the later scheduling passes from changing its schedule, we use this
5099 option to control that.
5101 @item -fcaller-saves
5102 @opindex fcaller-saves
5103 Enable values to be allocated in registers that will be clobbered by
5104 function calls, by emitting extra instructions to save and restore the
5105 registers around such calls. Such allocation is done only when it
5106 seems to result in better code than would otherwise be produced.
5108 This option is always enabled by default on certain machines, usually
5109 those which have no call-preserved registers to use instead.
5111 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5114 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
5115 enabled by default at @option{-O2} and @option{-O3}.
5118 Perform Full Redundancy Elimination (FRE) on trees. The difference
5119 between FRE and PRE is that FRE only considers expressions
5120 that are computed on all paths leading to the redundant computation.
5121 This analysis faster than PRE, though it exposes fewer redundancies.
5122 This flag is enabled by default at @option{-O} and higher.
5124 @item -ftree-copy-prop
5125 Perform copy propagation on trees. This pass eliminates unnecessary
5126 copy operations. This flag is enabled by default at @option{-O} and
5129 @item -ftree-store-copy-prop
5130 Perform copy propagation of memory loads and stores. This pass
5131 eliminates unnecessary copy operations in memory references
5132 (structures, global variables, arrays, etc). This flag is enabled by
5133 default at @option{-O2} and higher.
5136 Perform structural alias analysis on trees. This flag
5137 is enabled by default at @option{-O} and higher.
5140 Perform interprocedural pointer analysis.
5143 Perform forward store motion on trees. This flag is
5144 enabled by default at @option{-O} and higher.
5147 Perform sparse conditional constant propagation (CCP) on trees. This
5148 pass only operates on local scalar variables and is enabled by default
5149 at @option{-O} and higher.
5151 @item -ftree-store-ccp
5152 Perform sparse conditional constant propagation (CCP) on trees. This
5153 pass operates on both local scalar variables and memory stores and
5154 loads (global variables, structures, arrays, etc). This flag is
5155 enabled by default at @option{-O2} and higher.
5158 Perform dead code elimination (DCE) on trees. This flag is enabled by
5159 default at @option{-O} and higher.
5161 @item -ftree-dominator-opts
5162 Perform a variety of simple scalar cleanups (constant/copy
5163 propagation, redundancy elimination, range propagation and expression
5164 simplification) based on a dominator tree traversal. This also
5165 performs jump threading (to reduce jumps to jumps). This flag is
5166 enabled by default at @option{-O} and higher.
5169 Perform loop header copying on trees. This is beneficial since it increases
5170 effectiveness of code motion optimizations. It also saves one jump. This flag
5171 is enabled by default at @option{-O} and higher. It is not enabled
5172 for @option{-Os}, since it usually increases code size.
5174 @item -ftree-loop-optimize
5175 Perform loop optimizations on trees. This flag is enabled by default
5176 at @option{-O} and higher.
5178 @item -ftree-loop-linear
5179 Perform linear loop transformations on tree. This flag can improve cache
5180 performance and allow further loop optimizations to take place.
5182 @item -ftree-loop-im
5183 Perform loop invariant motion on trees. This pass moves only invariants that
5184 would be hard to handle at RTL level (function calls, operations that expand to
5185 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5186 operands of conditions that are invariant out of the loop, so that we can use
5187 just trivial invariantness analysis in loop unswitching. The pass also includes
5190 @item -ftree-loop-ivcanon
5191 Create a canonical counter for number of iterations in the loop for that
5192 determining number of iterations requires complicated analysis. Later
5193 optimizations then may determine the number easily. Useful especially
5194 in connection with unrolling.
5197 Perform induction variable optimizations (strength reduction, induction
5198 variable merging and induction variable elimination) on trees.
5201 Perform scalar replacement of aggregates. This pass replaces structure
5202 references with scalars to prevent committing structures to memory too
5203 early. This flag is enabled by default at @option{-O} and higher.
5205 @item -ftree-copyrename
5206 Perform copy renaming on trees. This pass attempts to rename compiler
5207 temporaries to other variables at copy locations, usually resulting in
5208 variable names which more closely resemble the original variables. This flag
5209 is enabled by default at @option{-O} and higher.
5212 Perform temporary expression replacement during the SSA->normal phase. Single
5213 use/single def temporaries are replaced at their use location with their
5214 defining expression. This results in non-GIMPLE code, but gives the expanders
5215 much more complex trees to work on resulting in better RTL generation. This is
5216 enabled by default at @option{-O} and higher.
5219 Perform live range splitting during the SSA->normal phase. Distinct live
5220 ranges of a variable are split into unique variables, allowing for better
5221 optimization later. This is enabled by default at @option{-O} and higher.
5223 @item -ftree-vectorize
5224 Perform loop vectorization on trees.
5226 @item -ftree-vect-loop-version
5227 @opindex ftree-vect-loop-version
5228 Perform loop versioning when doing loop vectorization on trees. When a loop
5229 appears to be vectorizable except that data alignment or data dependence cannot
5230 be determined at compile time then vectorized and non-vectorized versions of
5231 the loop are generated along with runtime checks for alignment or dependence
5232 to control which version is executed. This option is enabled by default
5233 except at level @option{-Os} where it is disabled.
5236 Perform Value Range Propagation on trees. This is similar to the
5237 constant propagation pass, but instead of values, ranges of values are
5238 propagated. This allows the optimizers to remove unnecessary range
5239 checks like array bound checks and null pointer checks. This is
5240 enabled by default at @option{-O2} and higher. Null pointer check
5241 elimination is only done if @option{-fdelete-null-pointer-checks} is
5246 Perform tail duplication to enlarge superblock size. This transformation
5247 simplifies the control flow of the function allowing other optimizations to do
5250 @item -funroll-loops
5251 @opindex funroll-loops
5252 Unroll loops whose number of iterations can be determined at compile
5253 time or upon entry to the loop. @option{-funroll-loops} implies
5254 @option{-frerun-cse-after-loop}. This option makes code larger,
5255 and may or may not make it run faster.
5257 @item -funroll-all-loops
5258 @opindex funroll-all-loops
5259 Unroll all loops, even if their number of iterations is uncertain when
5260 the loop is entered. This usually makes programs run more slowly.
5261 @option{-funroll-all-loops} implies the same options as
5262 @option{-funroll-loops},
5264 @item -fsplit-ivs-in-unroller
5265 @opindex -fsplit-ivs-in-unroller
5266 Enables expressing of values of induction variables in later iterations
5267 of the unrolled loop using the value in the first iteration. This breaks
5268 long dependency chains, thus improving efficiency of the scheduling passes.
5270 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5271 same effect. However in cases the loop body is more complicated than
5272 a single basic block, this is not reliable. It also does not work at all
5273 on some of the architectures due to restrictions in the CSE pass.
5275 This optimization is enabled by default.
5277 @item -fvariable-expansion-in-unroller
5278 @opindex -fvariable-expansion-in-unroller
5279 With this option, the compiler will create multiple copies of some
5280 local variables when unrolling a loop which can result in superior code.
5282 @item -fprefetch-loop-arrays
5283 @opindex fprefetch-loop-arrays
5284 If supported by the target machine, generate instructions to prefetch
5285 memory to improve the performance of loops that access large arrays.
5287 This option may generate better or worse code; results are highly
5288 dependent on the structure of loops within the source code.
5290 Disabled at level @option{-Os}.
5293 @itemx -fno-peephole2
5294 @opindex fno-peephole
5295 @opindex fno-peephole2
5296 Disable any machine-specific peephole optimizations. The difference
5297 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5298 are implemented in the compiler; some targets use one, some use the
5299 other, a few use both.
5301 @option{-fpeephole} is enabled by default.
5302 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5304 @item -fno-guess-branch-probability
5305 @opindex fno-guess-branch-probability
5306 Do not guess branch probabilities using heuristics.
5308 GCC will use heuristics to guess branch probabilities if they are
5309 not provided by profiling feedback (@option{-fprofile-arcs}). These
5310 heuristics are based on the control flow graph. If some branch probabilities
5311 are specified by @samp{__builtin_expect}, then the heuristics will be
5312 used to guess branch probabilities for the rest of the control flow graph,
5313 taking the @samp{__builtin_expect} info into account. The interactions
5314 between the heuristics and @samp{__builtin_expect} can be complex, and in
5315 some cases, it may be useful to disable the heuristics so that the effects
5316 of @samp{__builtin_expect} are easier to understand.
5318 The default is @option{-fguess-branch-probability} at levels
5319 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5321 @item -freorder-blocks
5322 @opindex freorder-blocks
5323 Reorder basic blocks in the compiled function in order to reduce number of
5324 taken branches and improve code locality.
5326 Enabled at levels @option{-O2}, @option{-O3}.
5328 @item -freorder-blocks-and-partition
5329 @opindex freorder-blocks-and-partition
5330 In addition to reordering basic blocks in the compiled function, in order
5331 to reduce number of taken branches, partitions hot and cold basic blocks
5332 into separate sections of the assembly and .o files, to improve
5333 paging and cache locality performance.
5335 This optimization is automatically turned off in the presence of
5336 exception handling, for linkonce sections, for functions with a user-defined
5337 section attribute and on any architecture that does not support named
5340 @item -freorder-functions
5341 @opindex freorder-functions
5342 Reorder functions in the object file in order to
5343 improve code locality. This is implemented by using special
5344 subsections @code{.text.hot} for most frequently executed functions and
5345 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5346 the linker so object file format must support named sections and linker must
5347 place them in a reasonable way.
5349 Also profile feedback must be available in to make this option effective. See
5350 @option{-fprofile-arcs} for details.
5352 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5354 @item -fstrict-aliasing
5355 @opindex fstrict-aliasing
5356 Allows the compiler to assume the strictest aliasing rules applicable to
5357 the language being compiled. For C (and C++), this activates
5358 optimizations based on the type of expressions. In particular, an
5359 object of one type is assumed never to reside at the same address as an
5360 object of a different type, unless the types are almost the same. For
5361 example, an @code{unsigned int} can alias an @code{int}, but not a
5362 @code{void*} or a @code{double}. A character type may alias any other
5365 Pay special attention to code like this:
5378 The practice of reading from a different union member than the one most
5379 recently written to (called ``type-punning'') is common. Even with
5380 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5381 is accessed through the union type. So, the code above will work as
5382 expected. However, this code might not:
5393 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5395 @item -falign-functions
5396 @itemx -falign-functions=@var{n}
5397 @opindex falign-functions
5398 Align the start of functions to the next power-of-two greater than
5399 @var{n}, skipping up to @var{n} bytes. For instance,
5400 @option{-falign-functions=32} aligns functions to the next 32-byte
5401 boundary, but @option{-falign-functions=24} would align to the next
5402 32-byte boundary only if this can be done by skipping 23 bytes or less.
5404 @option{-fno-align-functions} and @option{-falign-functions=1} are
5405 equivalent and mean that functions will not be aligned.
5407 Some assemblers only support this flag when @var{n} is a power of two;
5408 in that case, it is rounded up.
5410 If @var{n} is not specified or is zero, use a machine-dependent default.
5412 Enabled at levels @option{-O2}, @option{-O3}.
5414 @item -falign-labels
5415 @itemx -falign-labels=@var{n}
5416 @opindex falign-labels
5417 Align all branch targets to a power-of-two boundary, skipping up to
5418 @var{n} bytes like @option{-falign-functions}. This option can easily
5419 make code slower, because it must insert dummy operations for when the
5420 branch target is reached in the usual flow of the code.
5422 @option{-fno-align-labels} and @option{-falign-labels=1} are
5423 equivalent and mean that labels will not be aligned.
5425 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
5426 are greater than this value, then their values are used instead.
5428 If @var{n} is not specified or is zero, use a machine-dependent default
5429 which is very likely to be @samp{1}, meaning no alignment.
5431 Enabled at levels @option{-O2}, @option{-O3}.
5434 @itemx -falign-loops=@var{n}
5435 @opindex falign-loops
5436 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
5437 like @option{-falign-functions}. The hope is that the loop will be
5438 executed many times, which will make up for any execution of the dummy
5441 @option{-fno-align-loops} and @option{-falign-loops=1} are
5442 equivalent and mean that loops will not be aligned.
5444 If @var{n} is not specified or is zero, use a machine-dependent default.
5446 Enabled at levels @option{-O2}, @option{-O3}.
5449 @itemx -falign-jumps=@var{n}
5450 @opindex falign-jumps
5451 Align branch targets to a power-of-two boundary, for branch targets
5452 where the targets can only be reached by jumping, skipping up to @var{n}
5453 bytes like @option{-falign-functions}. In this case, no dummy operations
5456 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
5457 equivalent and mean that loops will not be aligned.
5459 If @var{n} is not specified or is zero, use a machine-dependent default.
5461 Enabled at levels @option{-O2}, @option{-O3}.
5463 @item -funit-at-a-time
5464 @opindex funit-at-a-time
5465 Parse the whole compilation unit before starting to produce code.
5466 This allows some extra optimizations to take place but consumes
5467 more memory (in general). There are some compatibility issues
5468 with @emph{unit-at-a-time} mode:
5471 enabling @emph{unit-at-a-time} mode may change the order
5472 in which functions, variables, and top-level @code{asm} statements
5473 are emitted, and will likely break code relying on some particular
5474 ordering. The majority of such top-level @code{asm} statements,
5475 though, can be replaced by @code{section} attributes. The
5476 @option{fno-toplevel-reorder} option may be used to keep the ordering
5477 used in the input file, at the cost of some optimizations.
5480 @emph{unit-at-a-time} mode removes unreferenced static variables
5481 and functions. This may result in undefined references
5482 when an @code{asm} statement refers directly to variables or functions
5483 that are otherwise unused. In that case either the variable/function
5484 shall be listed as an operand of the @code{asm} statement operand or,
5485 in the case of top-level @code{asm} statements the attribute @code{used}
5486 shall be used on the declaration.
5489 Static functions now can use non-standard passing conventions that
5490 may break @code{asm} statements calling functions directly. Again,
5491 attribute @code{used} will prevent this behavior.
5494 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
5495 but this scheme may not be supported by future releases of GCC@.
5497 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5499 @item -fno-toplevel-reorder
5500 Do not reorder top-level functions, variables, and @code{asm}
5501 statements. Output them in the same order that they appear in the
5502 input file. When this option is used, unreferenced static variables
5503 will not be removed. This option is intended to support existing code
5504 which relies on a particular ordering. For new code, it is better to
5509 Constructs webs as commonly used for register allocation purposes and assign
5510 each web individual pseudo register. This allows the register allocation pass
5511 to operate on pseudos directly, but also strengthens several other optimization
5512 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
5513 however, make debugging impossible, since variables will no longer stay in a
5516 Enabled by default with @option{-funroll-loops}.
5518 @item -fwhole-program
5519 @opindex fwhole-program
5520 Assume that the current compilation unit represents whole program being
5521 compiled. All public functions and variables with the exception of @code{main}
5522 and those merged by attribute @code{externally_visible} become static functions
5523 and in a affect gets more aggressively optimized by interprocedural optimizers.
5524 While this option is equivalent to proper use of @code{static} keyword for
5525 programs consisting of single file, in combination with option
5526 @option{--combine} this flag can be used to compile most of smaller scale C
5527 programs since the functions and variables become local for the whole combined
5528 compilation unit, not for the single source file itself.
5531 @item -fno-cprop-registers
5532 @opindex fno-cprop-registers
5533 After register allocation and post-register allocation instruction splitting,
5534 we perform a copy-propagation pass to try to reduce scheduling dependencies
5535 and occasionally eliminate the copy.
5537 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5539 @item -fprofile-generate
5540 @opindex fprofile-generate
5542 Enable options usually used for instrumenting application to produce
5543 profile useful for later recompilation with profile feedback based
5544 optimization. You must use @option{-fprofile-generate} both when
5545 compiling and when linking your program.
5547 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
5550 @opindex fprofile-use
5551 Enable profile feedback directed optimizations, and optimizations
5552 generally profitable only with profile feedback available.
5554 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
5555 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
5559 The following options control compiler behavior regarding floating
5560 point arithmetic. These options trade off between speed and
5561 correctness. All must be specifically enabled.
5565 @opindex ffloat-store
5566 Do not store floating point variables in registers, and inhibit other
5567 options that might change whether a floating point value is taken from a
5570 @cindex floating point precision
5571 This option prevents undesirable excess precision on machines such as
5572 the 68000 where the floating registers (of the 68881) keep more
5573 precision than a @code{double} is supposed to have. Similarly for the
5574 x86 architecture. For most programs, the excess precision does only
5575 good, but a few programs rely on the precise definition of IEEE floating
5576 point. Use @option{-ffloat-store} for such programs, after modifying
5577 them to store all pertinent intermediate computations into variables.
5581 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5582 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5583 @option{-fno-rounding-math}, @option{-fno-signaling-nans}
5584 and @option{fcx-limited-range}.
5586 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5588 This option should never be turned on by any @option{-O} option since
5589 it can result in incorrect output for programs which depend on
5590 an exact implementation of IEEE or ISO rules/specifications for
5593 @item -fno-math-errno
5594 @opindex fno-math-errno
5595 Do not set ERRNO after calling math functions that are executed
5596 with a single instruction, e.g., sqrt. A program that relies on
5597 IEEE exceptions for math error handling may want to use this flag
5598 for speed while maintaining IEEE arithmetic compatibility.
5600 This option should never be turned on by any @option{-O} option since
5601 it can result in incorrect output for programs which depend on
5602 an exact implementation of IEEE or ISO rules/specifications for
5605 The default is @option{-fmath-errno}.
5607 On Darwin systems, the math library never sets @code{errno}. There is therefore
5608 no reason for the compiler to consider the possibility that it might,
5609 and @option{-fno-math-errno} is the default.
5611 @item -funsafe-math-optimizations
5612 @opindex funsafe-math-optimizations
5613 Allow optimizations for floating-point arithmetic that (a) assume
5614 that arguments and results are valid and (b) may violate IEEE or
5615 ANSI standards. When used at link-time, it may include libraries
5616 or startup files that change the default FPU control word or other
5617 similar optimizations.
5619 This option should never be turned on by any @option{-O} option since
5620 it can result in incorrect output for programs which depend on
5621 an exact implementation of IEEE or ISO rules/specifications for
5624 The default is @option{-fno-unsafe-math-optimizations}.
5626 @item -ffinite-math-only
5627 @opindex ffinite-math-only
5628 Allow optimizations for floating-point arithmetic that assume
5629 that arguments and results are not NaNs or +-Infs.
5631 This option should never be turned on by any @option{-O} option since
5632 it can result in incorrect output for programs which depend on
5633 an exact implementation of IEEE or ISO rules/specifications.
5635 The default is @option{-fno-finite-math-only}.
5637 @item -fno-trapping-math
5638 @opindex fno-trapping-math
5639 Compile code assuming that floating-point operations cannot generate
5640 user-visible traps. These traps include division by zero, overflow,
5641 underflow, inexact result and invalid operation. This option implies
5642 @option{-fno-signaling-nans}. Setting this option may allow faster
5643 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5645 This option should never be turned on by any @option{-O} option since
5646 it can result in incorrect output for programs which depend on
5647 an exact implementation of IEEE or ISO rules/specifications for
5650 The default is @option{-ftrapping-math}.
5652 @item -frounding-math
5653 @opindex frounding-math
5654 Disable transformations and optimizations that assume default floating
5655 point rounding behavior. This is round-to-zero for all floating point
5656 to integer conversions, and round-to-nearest for all other arithmetic
5657 truncations. This option should be specified for programs that change
5658 the FP rounding mode dynamically, or that may be executed with a
5659 non-default rounding mode. This option disables constant folding of
5660 floating point expressions at compile-time (which may be affected by
5661 rounding mode) and arithmetic transformations that are unsafe in the
5662 presence of sign-dependent rounding modes.
5664 The default is @option{-fno-rounding-math}.
5666 This option is experimental and does not currently guarantee to
5667 disable all GCC optimizations that are affected by rounding mode.
5668 Future versions of GCC may provide finer control of this setting
5669 using C99's @code{FENV_ACCESS} pragma. This command line option
5670 will be used to specify the default state for @code{FENV_ACCESS}.
5672 @item -frtl-abstract-sequences
5673 @opindex frtl-abstract-sequences
5674 It is a size optimization method. This option is to find identical
5675 sequences of code, which can be turned into pseudo-procedures and
5676 then replace all occurrences with calls to the newly created
5677 subroutine. It is kind of an opposite of @option{-finline-functions}.
5678 This optimization runs at RTL level.
5680 @item -fsignaling-nans
5681 @opindex fsignaling-nans
5682 Compile code assuming that IEEE signaling NaNs may generate user-visible
5683 traps during floating-point operations. Setting this option disables
5684 optimizations that may change the number of exceptions visible with
5685 signaling NaNs. This option implies @option{-ftrapping-math}.
5687 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5690 The default is @option{-fno-signaling-nans}.
5692 This option is experimental and does not currently guarantee to
5693 disable all GCC optimizations that affect signaling NaN behavior.
5695 @item -fsingle-precision-constant
5696 @opindex fsingle-precision-constant
5697 Treat floating point constant as single precision constant instead of
5698 implicitly converting it to double precision constant.
5700 @item -fcx-limited-range
5701 @itemx -fno-cx-limited-range
5702 @opindex fcx-limited-range
5703 @opindex fno-cx-limited-range
5704 When enabled, this option states that a range reduction step is not
5705 needed when performing complex division. The default is
5706 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
5708 This option controls the default setting of the ISO C99
5709 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
5714 The following options control optimizations that may improve
5715 performance, but are not enabled by any @option{-O} options. This
5716 section includes experimental options that may produce broken code.
5719 @item -fbranch-probabilities
5720 @opindex fbranch-probabilities
5721 After running a program compiled with @option{-fprofile-arcs}
5722 (@pxref{Debugging Options,, Options for Debugging Your Program or
5723 @command{gcc}}), you can compile it a second time using
5724 @option{-fbranch-probabilities}, to improve optimizations based on
5725 the number of times each branch was taken. When the program
5726 compiled with @option{-fprofile-arcs} exits it saves arc execution
5727 counts to a file called @file{@var{sourcename}.gcda} for each source
5728 file The information in this data file is very dependent on the
5729 structure of the generated code, so you must use the same source code
5730 and the same optimization options for both compilations.
5732 With @option{-fbranch-probabilities}, GCC puts a
5733 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5734 These can be used to improve optimization. Currently, they are only
5735 used in one place: in @file{reorg.c}, instead of guessing which path a
5736 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5737 exactly determine which path is taken more often.
5739 @item -fprofile-values
5740 @opindex fprofile-values
5741 If combined with @option{-fprofile-arcs}, it adds code so that some
5742 data about values of expressions in the program is gathered.
5744 With @option{-fbranch-probabilities}, it reads back the data gathered
5745 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5746 notes to instructions for their later usage in optimizations.
5748 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5752 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5753 a code to gather information about values of expressions.
5755 With @option{-fbranch-probabilities}, it reads back the data gathered
5756 and actually performs the optimizations based on them.
5757 Currently the optimizations include specialization of division operation
5758 using the knowledge about the value of the denominator.
5760 @item -frename-registers
5761 @opindex frename-registers
5762 Attempt to avoid false dependencies in scheduled code by making use
5763 of registers left over after register allocation. This optimization
5764 will most benefit processors with lots of registers. Depending on the
5765 debug information format adopted by the target, however, it can
5766 make debugging impossible, since variables will no longer stay in
5767 a ``home register''.
5769 Enabled by default with @option{-funroll-loops}.
5773 Perform tail duplication to enlarge superblock size. This transformation
5774 simplifies the control flow of the function allowing other optimizations to do
5777 Enabled with @option{-fprofile-use}.
5779 @item -funroll-loops
5780 @opindex funroll-loops
5781 Unroll loops whose number of iterations can be determined at compile time or
5782 upon entry to the loop. @option{-funroll-loops} implies
5783 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
5784 It also turns on complete loop peeling (i.e.@: complete removal of loops with
5785 small constant number of iterations). This option makes code larger, and may
5786 or may not make it run faster.
5788 Enabled with @option{-fprofile-use}.
5790 @item -funroll-all-loops
5791 @opindex funroll-all-loops
5792 Unroll all loops, even if their number of iterations is uncertain when
5793 the loop is entered. This usually makes programs run more slowly.
5794 @option{-funroll-all-loops} implies the same options as
5795 @option{-funroll-loops}.
5798 @opindex fpeel-loops
5799 Peels the loops for that there is enough information that they do not
5800 roll much (from profile feedback). It also turns on complete loop peeling
5801 (i.e.@: complete removal of loops with small constant number of iterations).
5803 Enabled with @option{-fprofile-use}.
5805 @item -fmove-loop-invariants
5806 @opindex fmove-loop-invariants
5807 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
5808 at level @option{-O1}
5810 @item -funswitch-loops
5811 @opindex funswitch-loops
5812 Move branches with loop invariant conditions out of the loop, with duplicates
5813 of the loop on both branches (modified according to result of the condition).
5815 @item -ffunction-sections
5816 @itemx -fdata-sections
5817 @opindex ffunction-sections
5818 @opindex fdata-sections
5819 Place each function or data item into its own section in the output
5820 file if the target supports arbitrary sections. The name of the
5821 function or the name of the data item determines the section's name
5824 Use these options on systems where the linker can perform optimizations
5825 to improve locality of reference in the instruction space. Most systems
5826 using the ELF object format and SPARC processors running Solaris 2 have
5827 linkers with such optimizations. AIX may have these optimizations in
5830 Only use these options when there are significant benefits from doing
5831 so. When you specify these options, the assembler and linker will
5832 create larger object and executable files and will also be slower.
5833 You will not be able to use @code{gprof} on all systems if you
5834 specify this option and you may have problems with debugging if
5835 you specify both this option and @option{-g}.
5837 @item -fbranch-target-load-optimize
5838 @opindex fbranch-target-load-optimize
5839 Perform branch target register load optimization before prologue / epilogue
5841 The use of target registers can typically be exposed only during reload,
5842 thus hoisting loads out of loops and doing inter-block scheduling needs
5843 a separate optimization pass.
5845 @item -fbranch-target-load-optimize2
5846 @opindex fbranch-target-load-optimize2
5847 Perform branch target register load optimization after prologue / epilogue
5850 @item -fbtr-bb-exclusive
5851 @opindex fbtr-bb-exclusive
5852 When performing branch target register load optimization, don't reuse
5853 branch target registers in within any basic block.
5855 @item -fstack-protector
5856 Emit extra code to check for buffer overflows, such as stack smashing
5857 attacks. This is done by adding a guard variable to functions with
5858 vulnerable objects. This includes functions that call alloca, and
5859 functions with buffers larger than 8 bytes. The guards are initialized
5860 when a function is entered and then checked when the function exits.
5861 If a guard check fails, an error message is printed and the program exits.
5863 @item -fstack-protector-all
5864 Like @option{-fstack-protector} except that all functions are protected.
5866 @item -fsection-anchors
5867 @opindex fsection-anchors
5868 Try to reduce the number of symbolic address calculations by using
5869 shared ``anchor'' symbols to address nearby objects. This transformation
5870 can help to reduce the number of GOT entries and GOT accesses on some
5873 For example, the implementation of the following function @code{foo}:
5877 int foo (void) @{ return a + b + c; @}
5880 would usually calculate the addresses of all three variables, but if you
5881 compile it with @option{-fsection-anchors}, it will access the variables
5882 from a common anchor point instead. The effect is similar to the
5883 following pseudocode (which isn't valid C):
5888 register int *xr = &x;
5889 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5893 Not all targets support this option.
5895 @item --param @var{name}=@var{value}
5897 In some places, GCC uses various constants to control the amount of
5898 optimization that is done. For example, GCC will not inline functions
5899 that contain more that a certain number of instructions. You can
5900 control some of these constants on the command-line using the
5901 @option{--param} option.
5903 The names of specific parameters, and the meaning of the values, are
5904 tied to the internals of the compiler, and are subject to change
5905 without notice in future releases.
5907 In each case, the @var{value} is an integer. The allowable choices for
5908 @var{name} are given in the following table:
5911 @item salias-max-implicit-fields
5912 The maximum number of fields in a variable without direct
5913 structure accesses for which structure aliasing will consider trying
5914 to track each field. The default is 5
5916 @item salias-max-array-elements
5917 The maximum number of elements an array can have and its elements
5918 still be tracked individually by structure aliasing. The default is 4
5920 @item sra-max-structure-size
5921 The maximum structure size, in bytes, at which the scalar replacement
5922 of aggregates (SRA) optimization will perform block copies. The
5923 default value, 0, implies that GCC will select the most appropriate
5926 @item sra-field-structure-ratio
5927 The threshold ratio (as a percentage) between instantiated fields and
5928 the complete structure size. We say that if the ratio of the number
5929 of bytes in instantiated fields to the number of bytes in the complete
5930 structure exceeds this parameter, then block copies are not used. The
5933 @item max-crossjump-edges
5934 The maximum number of incoming edges to consider for crossjumping.
5935 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5936 the number of edges incoming to each block. Increasing values mean
5937 more aggressive optimization, making the compile time increase with
5938 probably small improvement in executable size.
5940 @item min-crossjump-insns
5941 The minimum number of instructions which must be matched at the end
5942 of two blocks before crossjumping will be performed on them. This
5943 value is ignored in the case where all instructions in the block being
5944 crossjumped from are matched. The default value is 5.
5946 @item max-grow-copy-bb-insns
5947 The maximum code size expansion factor when copying basic blocks
5948 instead of jumping. The expansion is relative to a jump instruction.
5949 The default value is 8.
5951 @item max-goto-duplication-insns
5952 The maximum number of instructions to duplicate to a block that jumps
5953 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
5954 passes, GCC factors computed gotos early in the compilation process,
5955 and unfactors them as late as possible. Only computed jumps at the
5956 end of a basic blocks with no more than max-goto-duplication-insns are
5957 unfactored. The default value is 8.
5959 @item max-delay-slot-insn-search
5960 The maximum number of instructions to consider when looking for an
5961 instruction to fill a delay slot. If more than this arbitrary number of
5962 instructions is searched, the time savings from filling the delay slot
5963 will be minimal so stop searching. Increasing values mean more
5964 aggressive optimization, making the compile time increase with probably
5965 small improvement in executable run time.
5967 @item max-delay-slot-live-search
5968 When trying to fill delay slots, the maximum number of instructions to
5969 consider when searching for a block with valid live register
5970 information. Increasing this arbitrarily chosen value means more
5971 aggressive optimization, increasing the compile time. This parameter
5972 should be removed when the delay slot code is rewritten to maintain the
5975 @item max-gcse-memory
5976 The approximate maximum amount of memory that will be allocated in
5977 order to perform the global common subexpression elimination
5978 optimization. If more memory than specified is required, the
5979 optimization will not be done.
5981 @item max-gcse-passes
5982 The maximum number of passes of GCSE to run. The default is 1.
5984 @item max-pending-list-length
5985 The maximum number of pending dependencies scheduling will allow
5986 before flushing the current state and starting over. Large functions
5987 with few branches or calls can create excessively large lists which
5988 needlessly consume memory and resources.
5990 @item max-inline-insns-single
5991 Several parameters control the tree inliner used in gcc.
5992 This number sets the maximum number of instructions (counted in GCC's
5993 internal representation) in a single function that the tree inliner
5994 will consider for inlining. This only affects functions declared
5995 inline and methods implemented in a class declaration (C++).
5996 The default value is 450.
5998 @item max-inline-insns-auto
5999 When you use @option{-finline-functions} (included in @option{-O3}),
6000 a lot of functions that would otherwise not be considered for inlining
6001 by the compiler will be investigated. To those functions, a different
6002 (more restrictive) limit compared to functions declared inline can
6004 The default value is 90.
6006 @item large-function-insns
6007 The limit specifying really large functions. For functions larger than this
6008 limit after inlining inlining is constrained by
6009 @option{--param large-function-growth}. This parameter is useful primarily
6010 to avoid extreme compilation time caused by non-linear algorithms used by the
6012 This parameter is ignored when @option{-funit-at-a-time} is not used.
6013 The default value is 2700.
6015 @item large-function-growth
6016 Specifies maximal growth of large function caused by inlining in percents.
6017 This parameter is ignored when @option{-funit-at-a-time} is not used.
6018 The default value is 100 which limits large function growth to 2.0 times
6021 @item large-unit-insns
6022 The limit specifying large translation unit. Growth caused by inlining of
6023 units larger than this limit is limited by @option{--param inline-unit-growth}.
6024 For small units this might be too tight (consider unit consisting of function A
6025 that is inline and B that just calls A three time. If B is small relative to
6026 A, the growth of unit is 300\% and yet such inlining is very sane. For very
6027 large units consisting of small inlininable functions however the overall unit
6028 growth limit is needed to avoid exponential explosion of code size. Thus for
6029 smaller units, the size is increased to @option{--param large-unit-insns}
6030 before applying @option{--param inline-unit-growth}. The default is 10000
6032 @item inline-unit-growth
6033 Specifies maximal overall growth of the compilation unit caused by inlining.
6034 This parameter is ignored when @option{-funit-at-a-time} is not used.
6035 The default value is 50 which limits unit growth to 1.5 times the original
6038 @item large-stack-frame
6039 The limit specifying large stack frames. While inlining the algorithm is trying
6040 to not grow past this limit too much. Default value is 256 bytes.
6042 @item large-stack-frame-growth
6043 Specifies maximal growth of large stack frames caused by inlining in percents.
6044 The default value is 1000 which limits large stack frame growth to 11 times
6047 @item max-inline-insns-recursive
6048 @itemx max-inline-insns-recursive-auto
6049 Specifies maximum number of instructions out-of-line copy of self recursive inline
6050 function can grow into by performing recursive inlining.
6052 For functions declared inline @option{--param max-inline-insns-recursive} is
6053 taken into account. For function not declared inline, recursive inlining
6054 happens only when @option{-finline-functions} (included in @option{-O3}) is
6055 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
6056 default value is 450.
6058 @item max-inline-recursive-depth
6059 @itemx max-inline-recursive-depth-auto
6060 Specifies maximum recursion depth used by the recursive inlining.
6062 For functions declared inline @option{--param max-inline-recursive-depth} is
6063 taken into account. For function not declared inline, recursive inlining
6064 happens only when @option{-finline-functions} (included in @option{-O3}) is
6065 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6066 default value is 450.
6068 @item min-inline-recursive-probability
6069 Recursive inlining is profitable only for function having deep recursion
6070 in average and can hurt for function having little recursion depth by
6071 increasing the prologue size or complexity of function body to other
6074 When profile feedback is available (see @option{-fprofile-generate}) the actual
6075 recursion depth can be guessed from probability that function will recurse via
6076 given call expression. This parameter limits inlining only to call expression
6077 whose probability exceeds given threshold (in percents). The default value is
6080 @item inline-call-cost
6081 Specify cost of call instruction relative to simple arithmetics operations
6082 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6083 functions and at the same time increases size of leaf function that is believed to
6084 reduce function size by being inlined. In effect it increases amount of
6085 inlining for code having large abstraction penalty (many functions that just
6086 pass the arguments to other functions) and decrease inlining for code with low
6087 abstraction penalty. The default value is 16.
6089 @item max-unrolled-insns
6090 The maximum number of instructions that a loop should have if that loop
6091 is unrolled, and if the loop is unrolled, it determines how many times
6092 the loop code is unrolled.
6094 @item max-average-unrolled-insns
6095 The maximum number of instructions biased by probabilities of their execution
6096 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6097 it determines how many times the loop code is unrolled.
6099 @item max-unroll-times
6100 The maximum number of unrollings of a single loop.
6102 @item max-peeled-insns
6103 The maximum number of instructions that a loop should have if that loop
6104 is peeled, and if the loop is peeled, it determines how many times
6105 the loop code is peeled.
6107 @item max-peel-times
6108 The maximum number of peelings of a single loop.
6110 @item max-completely-peeled-insns
6111 The maximum number of insns of a completely peeled loop.
6113 @item max-completely-peel-times
6114 The maximum number of iterations of a loop to be suitable for complete peeling.
6116 @item max-unswitch-insns
6117 The maximum number of insns of an unswitched loop.
6119 @item max-unswitch-level
6120 The maximum number of branches unswitched in a single loop.
6123 The minimum cost of an expensive expression in the loop invariant motion.
6125 @item iv-consider-all-candidates-bound
6126 Bound on number of candidates for induction variables below that
6127 all candidates are considered for each use in induction variable
6128 optimizations. Only the most relevant candidates are considered
6129 if there are more candidates, to avoid quadratic time complexity.
6131 @item iv-max-considered-uses
6132 The induction variable optimizations give up on loops that contain more
6133 induction variable uses.
6135 @item iv-always-prune-cand-set-bound
6136 If number of candidates in the set is smaller than this value,
6137 we always try to remove unnecessary ivs from the set during its
6138 optimization when a new iv is added to the set.
6140 @item scev-max-expr-size
6141 Bound on size of expressions used in the scalar evolutions analyzer.
6142 Large expressions slow the analyzer.
6144 @item vect-max-version-checks
6145 The maximum number of runtime checks that can be performed when doing
6146 loop versioning in the vectorizer. See option ftree-vect-loop-version
6147 for more information.
6149 @item max-iterations-to-track
6151 The maximum number of iterations of a loop the brute force algorithm
6152 for analysis of # of iterations of the loop tries to evaluate.
6154 @item hot-bb-count-fraction
6155 Select fraction of the maximal count of repetitions of basic block in program
6156 given basic block needs to have to be considered hot.
6158 @item hot-bb-frequency-fraction
6159 Select fraction of the maximal frequency of executions of basic block in
6160 function given basic block needs to have to be considered hot
6162 @item max-predicted-iterations
6163 The maximum number of loop iterations we predict statically. This is useful
6164 in cases where function contain single loop with known bound and other loop
6165 with unknown. We predict the known number of iterations correctly, while
6166 the unknown number of iterations average to roughly 10. This means that the
6167 loop without bounds would appear artificially cold relative to the other one.
6169 @item tracer-dynamic-coverage
6170 @itemx tracer-dynamic-coverage-feedback
6172 This value is used to limit superblock formation once the given percentage of
6173 executed instructions is covered. This limits unnecessary code size
6176 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6177 feedback is available. The real profiles (as opposed to statically estimated
6178 ones) are much less balanced allowing the threshold to be larger value.
6180 @item tracer-max-code-growth
6181 Stop tail duplication once code growth has reached given percentage. This is
6182 rather hokey argument, as most of the duplicates will be eliminated later in
6183 cross jumping, so it may be set to much higher values than is the desired code
6186 @item tracer-min-branch-ratio
6188 Stop reverse growth when the reverse probability of best edge is less than this
6189 threshold (in percent).
6191 @item tracer-min-branch-ratio
6192 @itemx tracer-min-branch-ratio-feedback
6194 Stop forward growth if the best edge do have probability lower than this
6197 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6198 compilation for profile feedback and one for compilation without. The value
6199 for compilation with profile feedback needs to be more conservative (higher) in
6200 order to make tracer effective.
6202 @item max-cse-path-length
6204 Maximum number of basic blocks on path that cse considers. The default is 10.
6207 The maximum instructions CSE process before flushing. The default is 1000.
6209 @item max-aliased-vops
6211 Maximum number of virtual operands per statement allowed to represent
6212 aliases before triggering the alias grouping heuristic. Alias
6213 grouping reduces compile times and memory consumption needed for
6214 aliasing at the expense of precision loss in alias information.
6216 @item ggc-min-expand
6218 GCC uses a garbage collector to manage its own memory allocation. This
6219 parameter specifies the minimum percentage by which the garbage
6220 collector's heap should be allowed to expand between collections.
6221 Tuning this may improve compilation speed; it has no effect on code
6224 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6225 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6226 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6227 GCC is not able to calculate RAM on a particular platform, the lower
6228 bound of 30% is used. Setting this parameter and
6229 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6230 every opportunity. This is extremely slow, but can be useful for
6233 @item ggc-min-heapsize
6235 Minimum size of the garbage collector's heap before it begins bothering
6236 to collect garbage. The first collection occurs after the heap expands
6237 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6238 tuning this may improve compilation speed, and has no effect on code
6241 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6242 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6243 with a lower bound of 4096 (four megabytes) and an upper bound of
6244 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6245 particular platform, the lower bound is used. Setting this parameter
6246 very large effectively disables garbage collection. Setting this
6247 parameter and @option{ggc-min-expand} to zero causes a full collection
6248 to occur at every opportunity.
6250 @item max-reload-search-insns
6251 The maximum number of instruction reload should look backward for equivalent
6252 register. Increasing values mean more aggressive optimization, making the
6253 compile time increase with probably slightly better performance. The default
6256 @item max-cselib-memory-locations
6257 The maximum number of memory locations cselib should take into account.
6258 Increasing values mean more aggressive optimization, making the compile time
6259 increase with probably slightly better performance. The default value is 500.
6261 @item max-flow-memory-locations
6262 Similar as @option{max-cselib-memory-locations} but for dataflow liveness.
6263 The default value is 100.
6265 @item reorder-blocks-duplicate
6266 @itemx reorder-blocks-duplicate-feedback
6268 Used by basic block reordering pass to decide whether to use unconditional
6269 branch or duplicate the code on its destination. Code is duplicated when its
6270 estimated size is smaller than this value multiplied by the estimated size of
6271 unconditional jump in the hot spots of the program.
6273 The @option{reorder-block-duplicate-feedback} is used only when profile
6274 feedback is available and may be set to higher values than
6275 @option{reorder-block-duplicate} since information about the hot spots is more
6278 @item max-sched-ready-insns
6279 The maximum number of instructions ready to be issued the scheduler should
6280 consider at any given time during the first scheduling pass. Increasing
6281 values mean more thorough searches, making the compilation time increase
6282 with probably little benefit. The default value is 100.
6284 @item max-sched-region-blocks
6285 The maximum number of blocks in a region to be considered for
6286 interblock scheduling. The default value is 10.
6288 @item max-sched-region-insns
6289 The maximum number of insns in a region to be considered for
6290 interblock scheduling. The default value is 100.
6293 The minimum probability (in percents) of reaching a source block
6294 for interblock speculative scheduling. The default value is 40.
6296 @item max-sched-extend-regions-iters
6297 The maximum number of iterations through CFG to extend regions.
6298 0 - disable region extension,
6299 N - do at most N iterations.
6300 The default value is 0.
6302 @item max-sched-insn-conflict-delay
6303 The maximum conflict delay for an insn to be considered for speculative motion.
6304 The default value is 3.
6306 @item sched-spec-prob-cutoff
6307 The minimal probability of speculation success (in percents), so that
6308 speculative insn will be scheduled.
6309 The default value is 40.
6311 @item max-last-value-rtl
6313 The maximum size measured as number of RTLs that can be recorded in an expression
6314 in combiner for a pseudo register as last known value of that register. The default
6317 @item integer-share-limit
6318 Small integer constants can use a shared data structure, reducing the
6319 compiler's memory usage and increasing its speed. This sets the maximum
6320 value of a shared integer constant's. The default value is 256.
6322 @item min-virtual-mappings
6323 Specifies the minimum number of virtual mappings in the incremental
6324 SSA updater that should be registered to trigger the virtual mappings
6325 heuristic defined by virtual-mappings-ratio. The default value is
6328 @item virtual-mappings-ratio
6329 If the number of virtual mappings is virtual-mappings-ratio bigger
6330 than the number of virtual symbols to be updated, then the incremental
6331 SSA updater switches to a full update for those symbols. The default
6334 @item ssp-buffer-size
6335 The minimum size of buffers (i.e. arrays) that will receive stack smashing
6336 protection when @option{-fstack-protection} is used.
6338 @item max-jump-thread-duplication-stmts
6339 Maximum number of statements allowed in a block that needs to be
6340 duplicated when threading jumps.
6342 @item max-fields-for-field-sensitive
6343 Maximum number of fields in a structure we will treat in
6344 a field sensitive manner during pointer analysis.
6346 @item prefetch-latency
6347 Estimate on average number of instructions that are executed before
6348 prefetch finishes. The distance we prefetch ahead is proportional
6349 to this constant. Increasing this number may also lead to less
6350 streams being prefetched (see @option{simultaneous-prefetches}).
6352 @item simultaneous-prefetches
6353 Maximum number of prefetches that can run at the same time.
6355 @item l1-cache-line-size
6356 The size of cache line in L1 cache, in bytes.
6359 The number of cache lines in L1 cache.
6364 @node Preprocessor Options
6365 @section Options Controlling the Preprocessor
6366 @cindex preprocessor options
6367 @cindex options, preprocessor
6369 These options control the C preprocessor, which is run on each C source
6370 file before actual compilation.
6372 If you use the @option{-E} option, nothing is done except preprocessing.
6373 Some of these options make sense only together with @option{-E} because
6374 they cause the preprocessor output to be unsuitable for actual
6379 You can use @option{-Wp,@var{option}} to bypass the compiler driver
6380 and pass @var{option} directly through to the preprocessor. If
6381 @var{option} contains commas, it is split into multiple options at the
6382 commas. However, many options are modified, translated or interpreted
6383 by the compiler driver before being passed to the preprocessor, and
6384 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
6385 interface is undocumented and subject to change, so whenever possible
6386 you should avoid using @option{-Wp} and let the driver handle the
6389 @item -Xpreprocessor @var{option}
6390 @opindex preprocessor
6391 Pass @var{option} as an option to the preprocessor. You can use this to
6392 supply system-specific preprocessor options which GCC does not know how to
6395 If you want to pass an option that takes an argument, you must use
6396 @option{-Xpreprocessor} twice, once for the option and once for the argument.
6399 @include cppopts.texi
6401 @node Assembler Options
6402 @section Passing Options to the Assembler
6404 @c prevent bad page break with this line
6405 You can pass options to the assembler.
6408 @item -Wa,@var{option}
6410 Pass @var{option} as an option to the assembler. If @var{option}
6411 contains commas, it is split into multiple options at the commas.
6413 @item -Xassembler @var{option}
6415 Pass @var{option} as an option to the assembler. You can use this to
6416 supply system-specific assembler options which GCC does not know how to
6419 If you want to pass an option that takes an argument, you must use
6420 @option{-Xassembler} twice, once for the option and once for the argument.
6425 @section Options for Linking
6426 @cindex link options
6427 @cindex options, linking
6429 These options come into play when the compiler links object files into
6430 an executable output file. They are meaningless if the compiler is
6431 not doing a link step.
6435 @item @var{object-file-name}
6436 A file name that does not end in a special recognized suffix is
6437 considered to name an object file or library. (Object files are
6438 distinguished from libraries by the linker according to the file
6439 contents.) If linking is done, these object files are used as input
6448 If any of these options is used, then the linker is not run, and
6449 object file names should not be used as arguments. @xref{Overall
6453 @item -l@var{library}
6454 @itemx -l @var{library}
6456 Search the library named @var{library} when linking. (The second
6457 alternative with the library as a separate argument is only for
6458 POSIX compliance and is not recommended.)
6460 It makes a difference where in the command you write this option; the
6461 linker searches and processes libraries and object files in the order they
6462 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
6463 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
6464 to functions in @samp{z}, those functions may not be loaded.
6466 The linker searches a standard list of directories for the library,
6467 which is actually a file named @file{lib@var{library}.a}. The linker
6468 then uses this file as if it had been specified precisely by name.
6470 The directories searched include several standard system directories
6471 plus any that you specify with @option{-L}.
6473 Normally the files found this way are library files---archive files
6474 whose members are object files. The linker handles an archive file by
6475 scanning through it for members which define symbols that have so far
6476 been referenced but not defined. But if the file that is found is an
6477 ordinary object file, it is linked in the usual fashion. The only
6478 difference between using an @option{-l} option and specifying a file name
6479 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
6480 and searches several directories.
6484 You need this special case of the @option{-l} option in order to
6485 link an Objective-C or Objective-C++ program.
6488 @opindex nostartfiles
6489 Do not use the standard system startup files when linking.
6490 The standard system libraries are used normally, unless @option{-nostdlib}
6491 or @option{-nodefaultlibs} is used.
6493 @item -nodefaultlibs
6494 @opindex nodefaultlibs
6495 Do not use the standard system libraries when linking.
6496 Only the libraries you specify will be passed to the linker.
6497 The standard startup files are used normally, unless @option{-nostartfiles}
6498 is used. The compiler may generate calls to @code{memcmp},
6499 @code{memset}, @code{memcpy} and @code{memmove}.
6500 These entries are usually resolved by entries in
6501 libc. These entry points should be supplied through some other
6502 mechanism when this option is specified.
6506 Do not use the standard system startup files or libraries when linking.
6507 No startup files and only the libraries you specify will be passed to
6508 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
6509 @code{memcpy} and @code{memmove}.
6510 These entries are usually resolved by entries in
6511 libc. These entry points should be supplied through some other
6512 mechanism when this option is specified.
6514 @cindex @option{-lgcc}, use with @option{-nostdlib}
6515 @cindex @option{-nostdlib} and unresolved references
6516 @cindex unresolved references and @option{-nostdlib}
6517 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
6518 @cindex @option{-nodefaultlibs} and unresolved references
6519 @cindex unresolved references and @option{-nodefaultlibs}
6520 One of the standard libraries bypassed by @option{-nostdlib} and
6521 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
6522 that GCC uses to overcome shortcomings of particular machines, or special
6523 needs for some languages.
6524 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
6525 Collection (GCC) Internals},
6526 for more discussion of @file{libgcc.a}.)
6527 In most cases, you need @file{libgcc.a} even when you want to avoid
6528 other standard libraries. In other words, when you specify @option{-nostdlib}
6529 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
6530 This ensures that you have no unresolved references to internal GCC
6531 library subroutines. (For example, @samp{__main}, used to ensure C++
6532 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
6533 GNU Compiler Collection (GCC) Internals}.)
6537 Produce a position independent executable on targets which support it.
6538 For predictable results, you must also specify the same set of options
6539 that were used to generate code (@option{-fpie}, @option{-fPIE},
6540 or model suboptions) when you specify this option.
6544 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
6545 that support it. This instructs the linker to add all symbols, not
6546 only used ones, to the dynamic symbol table. This option is needed
6547 for some uses of @code{dlopen} or to allow obtaining backtraces
6548 from within a program.
6552 Remove all symbol table and relocation information from the executable.
6556 On systems that support dynamic linking, this prevents linking with the shared
6557 libraries. On other systems, this option has no effect.
6561 Produce a shared object which can then be linked with other objects to
6562 form an executable. Not all systems support this option. For predictable
6563 results, you must also specify the same set of options that were used to
6564 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
6565 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
6566 needs to build supplementary stub code for constructors to work. On
6567 multi-libbed systems, @samp{gcc -shared} must select the correct support
6568 libraries to link against. Failing to supply the correct flags may lead
6569 to subtle defects. Supplying them in cases where they are not necessary
6572 @item -shared-libgcc
6573 @itemx -static-libgcc
6574 @opindex shared-libgcc
6575 @opindex static-libgcc
6576 On systems that provide @file{libgcc} as a shared library, these options
6577 force the use of either the shared or static version respectively.
6578 If no shared version of @file{libgcc} was built when the compiler was
6579 configured, these options have no effect.
6581 There are several situations in which an application should use the
6582 shared @file{libgcc} instead of the static version. The most common
6583 of these is when the application wishes to throw and catch exceptions
6584 across different shared libraries. In that case, each of the libraries
6585 as well as the application itself should use the shared @file{libgcc}.
6587 Therefore, the G++ and GCJ drivers automatically add
6588 @option{-shared-libgcc} whenever you build a shared library or a main
6589 executable, because C++ and Java programs typically use exceptions, so
6590 this is the right thing to do.
6592 If, instead, you use the GCC driver to create shared libraries, you may
6593 find that they will not always be linked with the shared @file{libgcc}.
6594 If GCC finds, at its configuration time, that you have a non-GNU linker
6595 or a GNU linker that does not support option @option{--eh-frame-hdr},
6596 it will link the shared version of @file{libgcc} into shared libraries
6597 by default. Otherwise, it will take advantage of the linker and optimize
6598 away the linking with the shared version of @file{libgcc}, linking with
6599 the static version of libgcc by default. This allows exceptions to
6600 propagate through such shared libraries, without incurring relocation
6601 costs at library load time.
6603 However, if a library or main executable is supposed to throw or catch
6604 exceptions, you must link it using the G++ or GCJ driver, as appropriate
6605 for the languages used in the program, or using the option
6606 @option{-shared-libgcc}, such that it is linked with the shared
6611 Bind references to global symbols when building a shared object. Warn
6612 about any unresolved references (unless overridden by the link editor
6613 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
6616 @item -Xlinker @var{option}
6618 Pass @var{option} as an option to the linker. You can use this to
6619 supply system-specific linker options which GCC does not know how to
6622 If you want to pass an option that takes an argument, you must use
6623 @option{-Xlinker} twice, once for the option and once for the argument.
6624 For example, to pass @option{-assert definitions}, you must write
6625 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
6626 @option{-Xlinker "-assert definitions"}, because this passes the entire
6627 string as a single argument, which is not what the linker expects.
6629 @item -Wl,@var{option}
6631 Pass @var{option} as an option to the linker. If @var{option} contains
6632 commas, it is split into multiple options at the commas.
6634 @item -u @var{symbol}
6636 Pretend the symbol @var{symbol} is undefined, to force linking of
6637 library modules to define it. You can use @option{-u} multiple times with
6638 different symbols to force loading of additional library modules.
6641 @node Directory Options
6642 @section Options for Directory Search
6643 @cindex directory options
6644 @cindex options, directory search
6647 These options specify directories to search for header files, for
6648 libraries and for parts of the compiler:
6653 Add the directory @var{dir} to the head of the list of directories to be
6654 searched for header files. This can be used to override a system header
6655 file, substituting your own version, since these directories are
6656 searched before the system header file directories. However, you should
6657 not use this option to add directories that contain vendor-supplied
6658 system header files (use @option{-isystem} for that). If you use more than
6659 one @option{-I} option, the directories are scanned in left-to-right
6660 order; the standard system directories come after.
6662 If a standard system include directory, or a directory specified with
6663 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
6664 option will be ignored. The directory will still be searched but as a
6665 system directory at its normal position in the system include chain.
6666 This is to ensure that GCC's procedure to fix buggy system headers and
6667 the ordering for the include_next directive are not inadvertently changed.
6668 If you really need to change the search order for system directories,
6669 use the @option{-nostdinc} and/or @option{-isystem} options.
6671 @item -iquote@var{dir}
6673 Add the directory @var{dir} to the head of the list of directories to
6674 be searched for header files only for the case of @samp{#include
6675 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
6676 otherwise just like @option{-I}.
6680 Add directory @var{dir} to the list of directories to be searched
6683 @item -B@var{prefix}
6685 This option specifies where to find the executables, libraries,
6686 include files, and data files of the compiler itself.
6688 The compiler driver program runs one or more of the subprograms
6689 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
6690 @var{prefix} as a prefix for each program it tries to run, both with and
6691 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
6693 For each subprogram to be run, the compiler driver first tries the
6694 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
6695 was not specified, the driver tries two standard prefixes, which are
6696 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
6697 those results in a file name that is found, the unmodified program
6698 name is searched for using the directories specified in your
6699 @env{PATH} environment variable.
6701 The compiler will check to see if the path provided by the @option{-B}
6702 refers to a directory, and if necessary it will add a directory
6703 separator character at the end of the path.
6705 @option{-B} prefixes that effectively specify directory names also apply
6706 to libraries in the linker, because the compiler translates these
6707 options into @option{-L} options for the linker. They also apply to
6708 includes files in the preprocessor, because the compiler translates these
6709 options into @option{-isystem} options for the preprocessor. In this case,
6710 the compiler appends @samp{include} to the prefix.
6712 The run-time support file @file{libgcc.a} can also be searched for using
6713 the @option{-B} prefix, if needed. If it is not found there, the two
6714 standard prefixes above are tried, and that is all. The file is left
6715 out of the link if it is not found by those means.
6717 Another way to specify a prefix much like the @option{-B} prefix is to use
6718 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
6721 As a special kludge, if the path provided by @option{-B} is
6722 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
6723 9, then it will be replaced by @file{[dir/]include}. This is to help
6724 with boot-strapping the compiler.
6726 @item -specs=@var{file}
6728 Process @var{file} after the compiler reads in the standard @file{specs}
6729 file, in order to override the defaults that the @file{gcc} driver
6730 program uses when determining what switches to pass to @file{cc1},
6731 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
6732 @option{-specs=@var{file}} can be specified on the command line, and they
6733 are processed in order, from left to right.
6735 @item --sysroot=@var{dir}
6737 Use @var{dir} as the logical root directory for headers and libraries.
6738 For example, if the compiler would normally search for headers in
6739 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
6740 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
6742 If you use both this option and the @option{-isysroot} option, then
6743 the @option{--sysroot} option will apply to libraries, but the
6744 @option{-isysroot} option will apply to header files.
6746 The GNU linker (beginning with version 2.16) has the necessary support
6747 for this option. If your linker does not support this option, the
6748 header file aspect of @option{--sysroot} will still work, but the
6749 library aspect will not.
6753 This option has been deprecated. Please use @option{-iquote} instead for
6754 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
6755 Any directories you specify with @option{-I} options before the @option{-I-}
6756 option are searched only for the case of @samp{#include "@var{file}"};
6757 they are not searched for @samp{#include <@var{file}>}.
6759 If additional directories are specified with @option{-I} options after
6760 the @option{-I-}, these directories are searched for all @samp{#include}
6761 directives. (Ordinarily @emph{all} @option{-I} directories are used
6764 In addition, the @option{-I-} option inhibits the use of the current
6765 directory (where the current input file came from) as the first search
6766 directory for @samp{#include "@var{file}"}. There is no way to
6767 override this effect of @option{-I-}. With @option{-I.} you can specify
6768 searching the directory which was current when the compiler was
6769 invoked. That is not exactly the same as what the preprocessor does
6770 by default, but it is often satisfactory.
6772 @option{-I-} does not inhibit the use of the standard system directories
6773 for header files. Thus, @option{-I-} and @option{-nostdinc} are
6780 @section Specifying subprocesses and the switches to pass to them
6783 @command{gcc} is a driver program. It performs its job by invoking a
6784 sequence of other programs to do the work of compiling, assembling and
6785 linking. GCC interprets its command-line parameters and uses these to
6786 deduce which programs it should invoke, and which command-line options
6787 it ought to place on their command lines. This behavior is controlled
6788 by @dfn{spec strings}. In most cases there is one spec string for each
6789 program that GCC can invoke, but a few programs have multiple spec
6790 strings to control their behavior. The spec strings built into GCC can
6791 be overridden by using the @option{-specs=} command-line switch to specify
6794 @dfn{Spec files} are plaintext files that are used to construct spec
6795 strings. They consist of a sequence of directives separated by blank
6796 lines. The type of directive is determined by the first non-whitespace
6797 character on the line and it can be one of the following:
6800 @item %@var{command}
6801 Issues a @var{command} to the spec file processor. The commands that can
6805 @item %include <@var{file}>
6807 Search for @var{file} and insert its text at the current point in the
6810 @item %include_noerr <@var{file}>
6811 @cindex %include_noerr
6812 Just like @samp{%include}, but do not generate an error message if the include
6813 file cannot be found.
6815 @item %rename @var{old_name} @var{new_name}
6817 Rename the spec string @var{old_name} to @var{new_name}.
6821 @item *[@var{spec_name}]:
6822 This tells the compiler to create, override or delete the named spec
6823 string. All lines after this directive up to the next directive or
6824 blank line are considered to be the text for the spec string. If this
6825 results in an empty string then the spec will be deleted. (Or, if the
6826 spec did not exist, then nothing will happened.) Otherwise, if the spec
6827 does not currently exist a new spec will be created. If the spec does
6828 exist then its contents will be overridden by the text of this
6829 directive, unless the first character of that text is the @samp{+}
6830 character, in which case the text will be appended to the spec.
6832 @item [@var{suffix}]:
6833 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6834 and up to the next directive or blank line are considered to make up the
6835 spec string for the indicated suffix. When the compiler encounters an
6836 input file with the named suffix, it will processes the spec string in
6837 order to work out how to compile that file. For example:
6844 This says that any input file whose name ends in @samp{.ZZ} should be
6845 passed to the program @samp{z-compile}, which should be invoked with the
6846 command-line switch @option{-input} and with the result of performing the
6847 @samp{%i} substitution. (See below.)
6849 As an alternative to providing a spec string, the text that follows a
6850 suffix directive can be one of the following:
6853 @item @@@var{language}
6854 This says that the suffix is an alias for a known @var{language}. This is
6855 similar to using the @option{-x} command-line switch to GCC to specify a
6856 language explicitly. For example:
6863 Says that .ZZ files are, in fact, C++ source files.
6866 This causes an error messages saying:
6869 @var{name} compiler not installed on this system.
6873 GCC already has an extensive list of suffixes built into it.
6874 This directive will add an entry to the end of the list of suffixes, but
6875 since the list is searched from the end backwards, it is effectively
6876 possible to override earlier entries using this technique.
6880 GCC has the following spec strings built into it. Spec files can
6881 override these strings or create their own. Note that individual
6882 targets can also add their own spec strings to this list.
6885 asm Options to pass to the assembler
6886 asm_final Options to pass to the assembler post-processor
6887 cpp Options to pass to the C preprocessor
6888 cc1 Options to pass to the C compiler
6889 cc1plus Options to pass to the C++ compiler
6890 endfile Object files to include at the end of the link
6891 link Options to pass to the linker
6892 lib Libraries to include on the command line to the linker
6893 libgcc Decides which GCC support library to pass to the linker
6894 linker Sets the name of the linker
6895 predefines Defines to be passed to the C preprocessor
6896 signed_char Defines to pass to CPP to say whether @code{char} is signed
6898 startfile Object files to include at the start of the link
6901 Here is a small example of a spec file:
6907 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6910 This example renames the spec called @samp{lib} to @samp{old_lib} and
6911 then overrides the previous definition of @samp{lib} with a new one.
6912 The new definition adds in some extra command-line options before
6913 including the text of the old definition.
6915 @dfn{Spec strings} are a list of command-line options to be passed to their
6916 corresponding program. In addition, the spec strings can contain
6917 @samp{%}-prefixed sequences to substitute variable text or to
6918 conditionally insert text into the command line. Using these constructs
6919 it is possible to generate quite complex command lines.
6921 Here is a table of all defined @samp{%}-sequences for spec
6922 strings. Note that spaces are not generated automatically around the
6923 results of expanding these sequences. Therefore you can concatenate them
6924 together or combine them with constant text in a single argument.
6928 Substitute one @samp{%} into the program name or argument.
6931 Substitute the name of the input file being processed.
6934 Substitute the basename of the input file being processed.
6935 This is the substring up to (and not including) the last period
6936 and not including the directory.
6939 This is the same as @samp{%b}, but include the file suffix (text after
6943 Marks the argument containing or following the @samp{%d} as a
6944 temporary file name, so that that file will be deleted if GCC exits
6945 successfully. Unlike @samp{%g}, this contributes no text to the
6948 @item %g@var{suffix}
6949 Substitute a file name that has suffix @var{suffix} and is chosen
6950 once per compilation, and mark the argument in the same way as
6951 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
6952 name is now chosen in a way that is hard to predict even when previously
6953 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
6954 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
6955 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
6956 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
6957 was simply substituted with a file name chosen once per compilation,
6958 without regard to any appended suffix (which was therefore treated
6959 just like ordinary text), making such attacks more likely to succeed.
6961 @item %u@var{suffix}
6962 Like @samp{%g}, but generates a new temporary file name even if
6963 @samp{%u@var{suffix}} was already seen.
6965 @item %U@var{suffix}
6966 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6967 new one if there is no such last file name. In the absence of any
6968 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6969 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6970 would involve the generation of two distinct file names, one
6971 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6972 simply substituted with a file name chosen for the previous @samp{%u},
6973 without regard to any appended suffix.
6975 @item %j@var{suffix}
6976 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6977 writable, and if save-temps is off; otherwise, substitute the name
6978 of a temporary file, just like @samp{%u}. This temporary file is not
6979 meant for communication between processes, but rather as a junk
6982 @item %|@var{suffix}
6983 @itemx %m@var{suffix}
6984 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6985 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6986 all. These are the two most common ways to instruct a program that it
6987 should read from standard input or write to standard output. If you
6988 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6989 construct: see for example @file{f/lang-specs.h}.
6991 @item %.@var{SUFFIX}
6992 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6993 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6994 terminated by the next space or %.
6997 Marks the argument containing or following the @samp{%w} as the
6998 designated output file of this compilation. This puts the argument
6999 into the sequence of arguments that @samp{%o} will substitute later.
7002 Substitutes the names of all the output files, with spaces
7003 automatically placed around them. You should write spaces
7004 around the @samp{%o} as well or the results are undefined.
7005 @samp{%o} is for use in the specs for running the linker.
7006 Input files whose names have no recognized suffix are not compiled
7007 at all, but they are included among the output files, so they will
7011 Substitutes the suffix for object files. Note that this is
7012 handled specially when it immediately follows @samp{%g, %u, or %U},
7013 because of the need for those to form complete file names. The
7014 handling is such that @samp{%O} is treated exactly as if it had already
7015 been substituted, except that @samp{%g, %u, and %U} do not currently
7016 support additional @var{suffix} characters following @samp{%O} as they would
7017 following, for example, @samp{.o}.
7020 Substitutes the standard macro predefinitions for the
7021 current target machine. Use this when running @code{cpp}.
7024 Like @samp{%p}, but puts @samp{__} before and after the name of each
7025 predefined macro, except for macros that start with @samp{__} or with
7026 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
7030 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
7031 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
7032 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
7033 and @option{-imultilib} as necessary.
7036 Current argument is the name of a library or startup file of some sort.
7037 Search for that file in a standard list of directories and substitute
7038 the full name found.
7041 Print @var{str} as an error message. @var{str} is terminated by a newline.
7042 Use this when inconsistent options are detected.
7045 Substitute the contents of spec string @var{name} at this point.
7048 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
7050 @item %x@{@var{option}@}
7051 Accumulate an option for @samp{%X}.
7054 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
7058 Output the accumulated assembler options specified by @option{-Wa}.
7061 Output the accumulated preprocessor options specified by @option{-Wp}.
7064 Process the @code{asm} spec. This is used to compute the
7065 switches to be passed to the assembler.
7068 Process the @code{asm_final} spec. This is a spec string for
7069 passing switches to an assembler post-processor, if such a program is
7073 Process the @code{link} spec. This is the spec for computing the
7074 command line passed to the linker. Typically it will make use of the
7075 @samp{%L %G %S %D and %E} sequences.
7078 Dump out a @option{-L} option for each directory that GCC believes might
7079 contain startup files. If the target supports multilibs then the
7080 current multilib directory will be prepended to each of these paths.
7083 Process the @code{lib} spec. This is a spec string for deciding which
7084 libraries should be included on the command line to the linker.
7087 Process the @code{libgcc} spec. This is a spec string for deciding
7088 which GCC support library should be included on the command line to the linker.
7091 Process the @code{startfile} spec. This is a spec for deciding which
7092 object files should be the first ones passed to the linker. Typically
7093 this might be a file named @file{crt0.o}.
7096 Process the @code{endfile} spec. This is a spec string that specifies
7097 the last object files that will be passed to the linker.
7100 Process the @code{cpp} spec. This is used to construct the arguments
7101 to be passed to the C preprocessor.
7104 Process the @code{cc1} spec. This is used to construct the options to be
7105 passed to the actual C compiler (@samp{cc1}).
7108 Process the @code{cc1plus} spec. This is used to construct the options to be
7109 passed to the actual C++ compiler (@samp{cc1plus}).
7112 Substitute the variable part of a matched option. See below.
7113 Note that each comma in the substituted string is replaced by
7117 Remove all occurrences of @code{-S} from the command line. Note---this
7118 command is position dependent. @samp{%} commands in the spec string
7119 before this one will see @code{-S}, @samp{%} commands in the spec string
7120 after this one will not.
7122 @item %:@var{function}(@var{args})
7123 Call the named function @var{function}, passing it @var{args}.
7124 @var{args} is first processed as a nested spec string, then split
7125 into an argument vector in the usual fashion. The function returns
7126 a string which is processed as if it had appeared literally as part
7127 of the current spec.
7129 The following built-in spec functions are provided:
7132 @item @code{if-exists}
7133 The @code{if-exists} spec function takes one argument, an absolute
7134 pathname to a file. If the file exists, @code{if-exists} returns the
7135 pathname. Here is a small example of its usage:
7139 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7142 @item @code{if-exists-else}
7143 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7144 spec function, except that it takes two arguments. The first argument is
7145 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7146 returns the pathname. If it does not exist, it returns the second argument.
7147 This way, @code{if-exists-else} can be used to select one file or another,
7148 based on the existence of the first. Here is a small example of its usage:
7152 crt0%O%s %:if-exists(crti%O%s) \
7153 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
7156 @item @code{replace-outfile}
7157 The @code{replace-outfile} spec function takes two arguments. It looks for the
7158 first argument in the outfiles array and replaces it with the second argument. Here
7159 is a small example of its usage:
7162 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7168 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7169 If that switch was not specified, this substitutes nothing. Note that
7170 the leading dash is omitted when specifying this option, and it is
7171 automatically inserted if the substitution is performed. Thus the spec
7172 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7173 and would output the command line option @option{-foo}.
7175 @item %W@{@code{S}@}
7176 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7179 @item %@{@code{S}*@}
7180 Substitutes all the switches specified to GCC whose names start
7181 with @code{-S}, but which also take an argument. This is used for
7182 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7183 GCC considers @option{-o foo} as being
7184 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7185 text, including the space. Thus two arguments would be generated.
7187 @item %@{@code{S}*&@code{T}*@}
7188 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7189 (the order of @code{S} and @code{T} in the spec is not significant).
7190 There can be any number of ampersand-separated variables; for each the
7191 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
7193 @item %@{@code{S}:@code{X}@}
7194 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
7196 @item %@{!@code{S}:@code{X}@}
7197 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
7199 @item %@{@code{S}*:@code{X}@}
7200 Substitutes @code{X} if one or more switches whose names start with
7201 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
7202 once, no matter how many such switches appeared. However, if @code{%*}
7203 appears somewhere in @code{X}, then @code{X} will be substituted once
7204 for each matching switch, with the @code{%*} replaced by the part of
7205 that switch that matched the @code{*}.
7207 @item %@{.@code{S}:@code{X}@}
7208 Substitutes @code{X}, if processing a file with suffix @code{S}.
7210 @item %@{!.@code{S}:@code{X}@}
7211 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
7213 @item %@{@code{S}|@code{P}:@code{X}@}
7214 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
7215 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
7216 although they have a stronger binding than the @samp{|}. If @code{%*}
7217 appears in @code{X}, all of the alternatives must be starred, and only
7218 the first matching alternative is substituted.
7220 For example, a spec string like this:
7223 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
7226 will output the following command-line options from the following input
7227 command-line options:
7232 -d fred.c -foo -baz -boggle
7233 -d jim.d -bar -baz -boggle
7236 @item %@{S:X; T:Y; :D@}
7238 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
7239 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
7240 be as many clauses as you need. This may be combined with @code{.},
7241 @code{!}, @code{|}, and @code{*} as needed.
7246 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
7247 construct may contain other nested @samp{%} constructs or spaces, or
7248 even newlines. They are processed as usual, as described above.
7249 Trailing white space in @code{X} is ignored. White space may also
7250 appear anywhere on the left side of the colon in these constructs,
7251 except between @code{.} or @code{*} and the corresponding word.
7253 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
7254 handled specifically in these constructs. If another value of
7255 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7256 @option{-W} switch is found later in the command line, the earlier
7257 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7258 just one letter, which passes all matching options.
7260 The character @samp{|} at the beginning of the predicate text is used to
7261 indicate that a command should be piped to the following command, but
7262 only if @option{-pipe} is specified.
7264 It is built into GCC which switches take arguments and which do not.
7265 (You might think it would be useful to generalize this to allow each
7266 compiler's spec to say which switches take arguments. But this cannot
7267 be done in a consistent fashion. GCC cannot even decide which input
7268 files have been specified without knowing which switches take arguments,
7269 and it must know which input files to compile in order to tell which
7272 GCC also knows implicitly that arguments starting in @option{-l} are to be
7273 treated as compiler output files, and passed to the linker in their
7274 proper position among the other output files.
7276 @c man begin OPTIONS
7278 @node Target Options
7279 @section Specifying Target Machine and Compiler Version
7280 @cindex target options
7281 @cindex cross compiling
7282 @cindex specifying machine version
7283 @cindex specifying compiler version and target machine
7284 @cindex compiler version, specifying
7285 @cindex target machine, specifying
7287 The usual way to run GCC is to run the executable called @file{gcc}, or
7288 @file{<machine>-gcc} when cross-compiling, or
7289 @file{<machine>-gcc-<version>} to run a version other than the one that
7290 was installed last. Sometimes this is inconvenient, so GCC provides
7291 options that will switch to another cross-compiler or version.
7294 @item -b @var{machine}
7296 The argument @var{machine} specifies the target machine for compilation.
7298 The value to use for @var{machine} is the same as was specified as the
7299 machine type when configuring GCC as a cross-compiler. For
7300 example, if a cross-compiler was configured with @samp{configure
7301 arm-elf}, meaning to compile for an arm processor with elf binaries,
7302 then you would specify @option{-b arm-elf} to run that cross compiler.
7303 Because there are other options beginning with @option{-b}, the
7304 configuration must contain a hyphen.
7306 @item -V @var{version}
7308 The argument @var{version} specifies which version of GCC to run.
7309 This is useful when multiple versions are installed. For example,
7310 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
7313 The @option{-V} and @option{-b} options work by running the
7314 @file{<machine>-gcc-<version>} executable, so there's no real reason to
7315 use them if you can just run that directly.
7317 @node Submodel Options
7318 @section Hardware Models and Configurations
7319 @cindex submodel options
7320 @cindex specifying hardware config
7321 @cindex hardware models and configurations, specifying
7322 @cindex machine dependent options
7324 Earlier we discussed the standard option @option{-b} which chooses among
7325 different installed compilers for completely different target
7326 machines, such as VAX vs.@: 68000 vs.@: 80386.
7328 In addition, each of these target machine types can have its own
7329 special options, starting with @samp{-m}, to choose among various
7330 hardware models or configurations---for example, 68010 vs 68020,
7331 floating coprocessor or none. A single installed version of the
7332 compiler can compile for any model or configuration, according to the
7335 Some configurations of the compiler also support additional special
7336 options, usually for compatibility with other compilers on the same
7339 @c This list is ordered alphanumerically by subsection name.
7340 @c It should be the same order and spelling as these options are listed
7341 @c in Machine Dependent Options
7347 * Blackfin Options::
7351 * DEC Alpha Options::
7352 * DEC Alpha/VMS Options::
7354 * GNU/Linux Options::
7357 * i386 and x86-64 Options::
7370 * RS/6000 and PowerPC Options::
7371 * S/390 and zSeries Options::
7376 * System V Options::
7377 * TMS320C3x/C4x Options::
7381 * Xstormy16 Options::
7387 @subsection ARC Options
7390 These options are defined for ARC implementations:
7395 Compile code for little endian mode. This is the default.
7399 Compile code for big endian mode.
7402 @opindex mmangle-cpu
7403 Prepend the name of the cpu to all public symbol names.
7404 In multiple-processor systems, there are many ARC variants with different
7405 instruction and register set characteristics. This flag prevents code
7406 compiled for one cpu to be linked with code compiled for another.
7407 No facility exists for handling variants that are ``almost identical''.
7408 This is an all or nothing option.
7410 @item -mcpu=@var{cpu}
7412 Compile code for ARC variant @var{cpu}.
7413 Which variants are supported depend on the configuration.
7414 All variants support @option{-mcpu=base}, this is the default.
7416 @item -mtext=@var{text-section}
7417 @itemx -mdata=@var{data-section}
7418 @itemx -mrodata=@var{readonly-data-section}
7422 Put functions, data, and readonly data in @var{text-section},
7423 @var{data-section}, and @var{readonly-data-section} respectively
7424 by default. This can be overridden with the @code{section} attribute.
7425 @xref{Variable Attributes}.
7430 @subsection ARM Options
7433 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
7437 @item -mabi=@var{name}
7439 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
7440 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
7443 @opindex mapcs-frame
7444 Generate a stack frame that is compliant with the ARM Procedure Call
7445 Standard for all functions, even if this is not strictly necessary for
7446 correct execution of the code. Specifying @option{-fomit-frame-pointer}
7447 with this option will cause the stack frames not to be generated for
7448 leaf functions. The default is @option{-mno-apcs-frame}.
7452 This is a synonym for @option{-mapcs-frame}.
7455 @c not currently implemented
7456 @item -mapcs-stack-check
7457 @opindex mapcs-stack-check
7458 Generate code to check the amount of stack space available upon entry to
7459 every function (that actually uses some stack space). If there is
7460 insufficient space available then either the function
7461 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
7462 called, depending upon the amount of stack space required. The run time
7463 system is required to provide these functions. The default is
7464 @option{-mno-apcs-stack-check}, since this produces smaller code.
7466 @c not currently implemented
7468 @opindex mapcs-float
7469 Pass floating point arguments using the float point registers. This is
7470 one of the variants of the APCS@. This option is recommended if the
7471 target hardware has a floating point unit or if a lot of floating point
7472 arithmetic is going to be performed by the code. The default is
7473 @option{-mno-apcs-float}, since integer only code is slightly increased in
7474 size if @option{-mapcs-float} is used.
7476 @c not currently implemented
7477 @item -mapcs-reentrant
7478 @opindex mapcs-reentrant
7479 Generate reentrant, position independent code. The default is
7480 @option{-mno-apcs-reentrant}.
7483 @item -mthumb-interwork
7484 @opindex mthumb-interwork
7485 Generate code which supports calling between the ARM and Thumb
7486 instruction sets. Without this option the two instruction sets cannot
7487 be reliably used inside one program. The default is
7488 @option{-mno-thumb-interwork}, since slightly larger code is generated
7489 when @option{-mthumb-interwork} is specified.
7491 @item -mno-sched-prolog
7492 @opindex mno-sched-prolog
7493 Prevent the reordering of instructions in the function prolog, or the
7494 merging of those instruction with the instructions in the function's
7495 body. This means that all functions will start with a recognizable set
7496 of instructions (or in fact one of a choice from a small set of
7497 different function prologues), and this information can be used to
7498 locate the start if functions inside an executable piece of code. The
7499 default is @option{-msched-prolog}.
7502 @opindex mhard-float
7503 Generate output containing floating point instructions. This is the
7507 @opindex msoft-float
7508 Generate output containing library calls for floating point.
7509 @strong{Warning:} the requisite libraries are not available for all ARM
7510 targets. Normally the facilities of the machine's usual C compiler are
7511 used, but this cannot be done directly in cross-compilation. You must make
7512 your own arrangements to provide suitable library functions for
7515 @option{-msoft-float} changes the calling convention in the output file;
7516 therefore, it is only useful if you compile @emph{all} of a program with
7517 this option. In particular, you need to compile @file{libgcc.a}, the
7518 library that comes with GCC, with @option{-msoft-float} in order for
7521 @item -mfloat-abi=@var{name}
7523 Specifies which ABI to use for floating point values. Permissible values
7524 are: @samp{soft}, @samp{softfp} and @samp{hard}.
7526 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
7527 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
7528 of floating point instructions, but still uses the soft-float calling
7531 @item -mlittle-endian
7532 @opindex mlittle-endian
7533 Generate code for a processor running in little-endian mode. This is
7534 the default for all standard configurations.
7537 @opindex mbig-endian
7538 Generate code for a processor running in big-endian mode; the default is
7539 to compile code for a little-endian processor.
7541 @item -mwords-little-endian
7542 @opindex mwords-little-endian
7543 This option only applies when generating code for big-endian processors.
7544 Generate code for a little-endian word order but a big-endian byte
7545 order. That is, a byte order of the form @samp{32107654}. Note: this
7546 option should only be used if you require compatibility with code for
7547 big-endian ARM processors generated by versions of the compiler prior to
7550 @item -mcpu=@var{name}
7552 This specifies the name of the target ARM processor. GCC uses this name
7553 to determine what kind of instructions it can emit when generating
7554 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
7555 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
7556 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
7557 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
7558 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
7559 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
7560 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
7561 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
7562 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
7563 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
7564 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
7565 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
7566 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
7567 @samp{arm1176jz-s}, @samp{arm1176jzf-s}, @samp{xscale}, @samp{iwmmxt},
7570 @itemx -mtune=@var{name}
7572 This option is very similar to the @option{-mcpu=} option, except that
7573 instead of specifying the actual target processor type, and hence
7574 restricting which instructions can be used, it specifies that GCC should
7575 tune the performance of the code as if the target were of the type
7576 specified in this option, but still choosing the instructions that it
7577 will generate based on the cpu specified by a @option{-mcpu=} option.
7578 For some ARM implementations better performance can be obtained by using
7581 @item -march=@var{name}
7583 This specifies the name of the target ARM architecture. GCC uses this
7584 name to determine what kind of instructions it can emit when generating
7585 assembly code. This option can be used in conjunction with or instead
7586 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
7587 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
7588 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
7589 @samp{iwmmxt}, @samp{ep9312}.
7591 @item -mfpu=@var{name}
7592 @itemx -mfpe=@var{number}
7593 @itemx -mfp=@var{number}
7597 This specifies what floating point hardware (or hardware emulation) is
7598 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
7599 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
7600 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
7601 with older versions of GCC@.
7603 If @option{-msoft-float} is specified this specifies the format of
7604 floating point values.
7606 @item -mstructure-size-boundary=@var{n}
7607 @opindex mstructure-size-boundary
7608 The size of all structures and unions will be rounded up to a multiple
7609 of the number of bits set by this option. Permissible values are 8, 32
7610 and 64. The default value varies for different toolchains. For the COFF
7611 targeted toolchain the default value is 8. A value of 64 is only allowed
7612 if the underlying ABI supports it.
7614 Specifying the larger number can produce faster, more efficient code, but
7615 can also increase the size of the program. Different values are potentially
7616 incompatible. Code compiled with one value cannot necessarily expect to
7617 work with code or libraries compiled with another value, if they exchange
7618 information using structures or unions.
7620 @item -mabort-on-noreturn
7621 @opindex mabort-on-noreturn
7622 Generate a call to the function @code{abort} at the end of a
7623 @code{noreturn} function. It will be executed if the function tries to
7627 @itemx -mno-long-calls
7628 @opindex mlong-calls
7629 @opindex mno-long-calls
7630 Tells the compiler to perform function calls by first loading the
7631 address of the function into a register and then performing a subroutine
7632 call on this register. This switch is needed if the target function
7633 will lie outside of the 64 megabyte addressing range of the offset based
7634 version of subroutine call instruction.
7636 Even if this switch is enabled, not all function calls will be turned
7637 into long calls. The heuristic is that static functions, functions
7638 which have the @samp{short-call} attribute, functions that are inside
7639 the scope of a @samp{#pragma no_long_calls} directive and functions whose
7640 definitions have already been compiled within the current compilation
7641 unit, will not be turned into long calls. The exception to this rule is
7642 that weak function definitions, functions with the @samp{long-call}
7643 attribute or the @samp{section} attribute, and functions that are within
7644 the scope of a @samp{#pragma long_calls} directive, will always be
7645 turned into long calls.
7647 This feature is not enabled by default. Specifying
7648 @option{-mno-long-calls} will restore the default behavior, as will
7649 placing the function calls within the scope of a @samp{#pragma
7650 long_calls_off} directive. Note these switches have no effect on how
7651 the compiler generates code to handle function calls via function
7654 @item -mnop-fun-dllimport
7655 @opindex mnop-fun-dllimport
7656 Disable support for the @code{dllimport} attribute.
7658 @item -msingle-pic-base
7659 @opindex msingle-pic-base
7660 Treat the register used for PIC addressing as read-only, rather than
7661 loading it in the prologue for each function. The run-time system is
7662 responsible for initializing this register with an appropriate value
7663 before execution begins.
7665 @item -mpic-register=@var{reg}
7666 @opindex mpic-register
7667 Specify the register to be used for PIC addressing. The default is R10
7668 unless stack-checking is enabled, when R9 is used.
7670 @item -mcirrus-fix-invalid-insns
7671 @opindex mcirrus-fix-invalid-insns
7672 @opindex mno-cirrus-fix-invalid-insns
7673 Insert NOPs into the instruction stream to in order to work around
7674 problems with invalid Maverick instruction combinations. This option
7675 is only valid if the @option{-mcpu=ep9312} option has been used to
7676 enable generation of instructions for the Cirrus Maverick floating
7677 point co-processor. This option is not enabled by default, since the
7678 problem is only present in older Maverick implementations. The default
7679 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
7682 @item -mpoke-function-name
7683 @opindex mpoke-function-name
7684 Write the name of each function into the text section, directly
7685 preceding the function prologue. The generated code is similar to this:
7689 .ascii "arm_poke_function_name", 0
7692 .word 0xff000000 + (t1 - t0)
7693 arm_poke_function_name
7695 stmfd sp!, @{fp, ip, lr, pc@}
7699 When performing a stack backtrace, code can inspect the value of
7700 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
7701 location @code{pc - 12} and the top 8 bits are set, then we know that
7702 there is a function name embedded immediately preceding this location
7703 and has length @code{((pc[-3]) & 0xff000000)}.
7707 Generate code for the 16-bit Thumb instruction set. The default is to
7708 use the 32-bit ARM instruction set.
7711 @opindex mtpcs-frame
7712 Generate a stack frame that is compliant with the Thumb Procedure Call
7713 Standard for all non-leaf functions. (A leaf function is one that does
7714 not call any other functions.) The default is @option{-mno-tpcs-frame}.
7716 @item -mtpcs-leaf-frame
7717 @opindex mtpcs-leaf-frame
7718 Generate a stack frame that is compliant with the Thumb Procedure Call
7719 Standard for all leaf functions. (A leaf function is one that does
7720 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
7722 @item -mcallee-super-interworking
7723 @opindex mcallee-super-interworking
7724 Gives all externally visible functions in the file being compiled an ARM
7725 instruction set header which switches to Thumb mode before executing the
7726 rest of the function. This allows these functions to be called from
7727 non-interworking code.
7729 @item -mcaller-super-interworking
7730 @opindex mcaller-super-interworking
7731 Allows calls via function pointers (including virtual functions) to
7732 execute correctly regardless of whether the target code has been
7733 compiled for interworking or not. There is a small overhead in the cost
7734 of executing a function pointer if this option is enabled.
7736 @item -mtp=@var{name}
7738 Specify the access model for the thread local storage pointer. The valid
7739 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
7740 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
7741 (supported in the arm6k architecture), and @option{auto}, which uses the
7742 best available method for the selected processor. The default setting is
7748 @subsection AVR Options
7751 These options are defined for AVR implementations:
7754 @item -mmcu=@var{mcu}
7756 Specify ATMEL AVR instruction set or MCU type.
7758 Instruction set avr1 is for the minimal AVR core, not supported by the C
7759 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7760 attiny11, attiny12, attiny15, attiny28).
7762 Instruction set avr2 (default) is for the classic AVR core with up to
7763 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7764 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7765 at90c8534, at90s8535).
7767 Instruction set avr3 is for the classic AVR core with up to 128K program
7768 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
7770 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7771 memory space (MCU types: atmega8, atmega83, atmega85).
7773 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7774 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
7775 atmega64, atmega128, at43usb355, at94k).
7779 Output instruction sizes to the asm file.
7781 @item -minit-stack=@var{N}
7782 @opindex minit-stack
7783 Specify the initial stack address, which may be a symbol or numeric value,
7784 @samp{__stack} is the default.
7786 @item -mno-interrupts
7787 @opindex mno-interrupts
7788 Generated code is not compatible with hardware interrupts.
7789 Code size will be smaller.
7791 @item -mcall-prologues
7792 @opindex mcall-prologues
7793 Functions prologues/epilogues expanded as call to appropriate
7794 subroutines. Code size will be smaller.
7796 @item -mno-tablejump
7797 @opindex mno-tablejump
7798 Do not generate tablejump insns which sometimes increase code size.
7801 @opindex mtiny-stack
7802 Change only the low 8 bits of the stack pointer.
7806 Assume int to be 8 bit integer. This affects the sizes of all types: A
7807 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7808 and long long will be 4 bytes. Please note that this option does not
7809 comply to the C standards, but it will provide you with smaller code
7813 @node Blackfin Options
7814 @subsection Blackfin Options
7815 @cindex Blackfin Options
7818 @item -momit-leaf-frame-pointer
7819 @opindex momit-leaf-frame-pointer
7820 Don't keep the frame pointer in a register for leaf functions. This
7821 avoids the instructions to save, set up and restore frame pointers and
7822 makes an extra register available in leaf functions. The option
7823 @option{-fomit-frame-pointer} removes the frame pointer for all functions
7824 which might make debugging harder.
7826 @item -mspecld-anomaly
7827 @opindex mspecld-anomaly
7828 When enabled, the compiler will ensure that the generated code does not
7829 contain speculative loads after jump instructions. This option is enabled
7832 @item -mno-specld-anomaly
7833 @opindex mno-specld-anomaly
7834 Don't generate extra code to prevent speculative loads from occurring.
7836 @item -mcsync-anomaly
7837 @opindex mcsync-anomaly
7838 When enabled, the compiler will ensure that the generated code does not
7839 contain CSYNC or SSYNC instructions too soon after conditional branches.
7840 This option is enabled by default.
7842 @item -mno-csync-anomaly
7843 @opindex mno-csync-anomaly
7844 Don't generate extra code to prevent CSYNC or SSYNC instructions from
7845 occurring too soon after a conditional branch.
7849 When enabled, the compiler is free to take advantage of the knowledge that
7850 the entire program fits into the low 64k of memory.
7853 @opindex mno-low-64k
7854 Assume that the program is arbitrarily large. This is the default.
7856 @item -mstack-check-l1
7857 @opindex mstack-check-l1
7858 Do stack checking using information placed into L1 scratchpad memory by the
7861 @item -mid-shared-library
7862 @opindex mid-shared-library
7863 Generate code that supports shared libraries via the library ID method.
7864 This allows for execute in place and shared libraries in an environment
7865 without virtual memory management. This option implies @option{-fPIC}.
7867 @item -mno-id-shared-library
7868 @opindex mno-id-shared-library
7869 Generate code that doesn't assume ID based shared libraries are being used.
7870 This is the default.
7872 @item -mleaf-id-shared-library
7873 @opindex mleaf-id-shared-library
7874 Generate code that supports shared libraries via the library ID method,
7875 but assumes that this library or executable won't link against any other
7876 ID shared libraries. That allows the compiler to use faster code for jumps
7879 @item -mno-leaf-id-shared-library
7880 @opindex mno-leaf-id-shared-library
7881 Do not assume that the code being compiled won't link against any ID shared
7882 libraries. Slower code will be generated for jump and call insns.
7884 @item -mshared-library-id=n
7885 @opindex mshared-library-id
7886 Specified the identification number of the ID based shared library being
7887 compiled. Specifying a value of 0 will generate more compact code, specifying
7888 other values will force the allocation of that number to the current
7889 library but is no more space or time efficient than omitting this option.
7893 Generate code that allows the data segment to be located in a different
7894 area of memory from the text segment. This allows for execute in place in
7895 an environment without virtual memory management by eliminating relocations
7896 against the text section.
7899 @opindex mno-sep-data
7900 Generate code that assumes that the data segment follows the text segment.
7901 This is the default.
7904 @itemx -mno-long-calls
7905 @opindex mlong-calls
7906 @opindex mno-long-calls
7907 Tells the compiler to perform function calls by first loading the
7908 address of the function into a register and then performing a subroutine
7909 call on this register. This switch is needed if the target function
7910 will lie outside of the 24 bit addressing range of the offset based
7911 version of subroutine call instruction.
7913 This feature is not enabled by default. Specifying
7914 @option{-mno-long-calls} will restore the default behavior. Note these
7915 switches have no effect on how the compiler generates code to handle
7916 function calls via function pointers.
7920 @subsection CRIS Options
7921 @cindex CRIS Options
7923 These options are defined specifically for the CRIS ports.
7926 @item -march=@var{architecture-type}
7927 @itemx -mcpu=@var{architecture-type}
7930 Generate code for the specified architecture. The choices for
7931 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7932 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
7933 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7936 @item -mtune=@var{architecture-type}
7938 Tune to @var{architecture-type} everything applicable about the generated
7939 code, except for the ABI and the set of available instructions. The
7940 choices for @var{architecture-type} are the same as for
7941 @option{-march=@var{architecture-type}}.
7943 @item -mmax-stack-frame=@var{n}
7944 @opindex mmax-stack-frame
7945 Warn when the stack frame of a function exceeds @var{n} bytes.
7947 @item -melinux-stacksize=@var{n}
7948 @opindex melinux-stacksize
7949 Only available with the @samp{cris-axis-aout} target. Arranges for
7950 indications in the program to the kernel loader that the stack of the
7951 program should be set to @var{n} bytes.
7957 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7958 @option{-march=v3} and @option{-march=v8} respectively.
7960 @item -mmul-bug-workaround
7961 @itemx -mno-mul-bug-workaround
7962 @opindex mmul-bug-workaround
7963 @opindex mno-mul-bug-workaround
7964 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7965 models where it applies. This option is active by default.
7969 Enable CRIS-specific verbose debug-related information in the assembly
7970 code. This option also has the effect to turn off the @samp{#NO_APP}
7971 formatted-code indicator to the assembler at the beginning of the
7976 Do not use condition-code results from previous instruction; always emit
7977 compare and test instructions before use of condition codes.
7979 @item -mno-side-effects
7980 @opindex mno-side-effects
7981 Do not emit instructions with side-effects in addressing modes other than
7985 @itemx -mno-stack-align
7987 @itemx -mno-data-align
7988 @itemx -mconst-align
7989 @itemx -mno-const-align
7990 @opindex mstack-align
7991 @opindex mno-stack-align
7992 @opindex mdata-align
7993 @opindex mno-data-align
7994 @opindex mconst-align
7995 @opindex mno-const-align
7996 These options (no-options) arranges (eliminate arrangements) for the
7997 stack-frame, individual data and constants to be aligned for the maximum
7998 single data access size for the chosen CPU model. The default is to
7999 arrange for 32-bit alignment. ABI details such as structure layout are
8000 not affected by these options.
8008 Similar to the stack- data- and const-align options above, these options
8009 arrange for stack-frame, writable data and constants to all be 32-bit,
8010 16-bit or 8-bit aligned. The default is 32-bit alignment.
8012 @item -mno-prologue-epilogue
8013 @itemx -mprologue-epilogue
8014 @opindex mno-prologue-epilogue
8015 @opindex mprologue-epilogue
8016 With @option{-mno-prologue-epilogue}, the normal function prologue and
8017 epilogue that sets up the stack-frame are omitted and no return
8018 instructions or return sequences are generated in the code. Use this
8019 option only together with visual inspection of the compiled code: no
8020 warnings or errors are generated when call-saved registers must be saved,
8021 or storage for local variable needs to be allocated.
8027 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
8028 instruction sequences that load addresses for functions from the PLT part
8029 of the GOT rather than (traditional on other architectures) calls to the
8030 PLT@. The default is @option{-mgotplt}.
8034 Legacy no-op option only recognized with the cris-axis-aout target.
8038 Legacy no-op option only recognized with the cris-axis-elf and
8039 cris-axis-linux-gnu targets.
8043 Only recognized with the cris-axis-aout target, where it selects a
8044 GNU/linux-like multilib, include files and instruction set for
8049 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
8053 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
8054 to link with input-output functions from a simulator library. Code,
8055 initialized data and zero-initialized data are allocated consecutively.
8059 Like @option{-sim}, but pass linker options to locate initialized data at
8060 0x40000000 and zero-initialized data at 0x80000000.
8064 @subsection CRX Options
8067 These options are defined specifically for the CRX ports.
8073 Enable the use of multiply-accumulate instructions. Disabled by default.
8077 Push instructions will be used to pass outgoing arguments when functions
8078 are called. Enabled by default.
8081 @node Darwin Options
8082 @subsection Darwin Options
8083 @cindex Darwin options
8085 These options are defined for all architectures running the Darwin operating
8088 FSF GCC on Darwin does not create ``fat'' object files; it will create
8089 an object file for the single architecture that it was built to
8090 target. Apple's GCC on Darwin does create ``fat'' files if multiple
8091 @option{-arch} options are used; it does so by running the compiler or
8092 linker multiple times and joining the results together with
8095 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
8096 @samp{i686}) is determined by the flags that specify the ISA
8097 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
8098 @option{-force_cpusubtype_ALL} option can be used to override this.
8100 The Darwin tools vary in their behavior when presented with an ISA
8101 mismatch. The assembler, @file{as}, will only permit instructions to
8102 be used that are valid for the subtype of the file it is generating,
8103 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
8104 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
8105 and print an error if asked to create a shared library with a less
8106 restrictive subtype than its input files (for instance, trying to put
8107 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
8108 for executables, @file{ld}, will quietly give the executable the most
8109 restrictive subtype of any of its input files.
8114 Add the framework directory @var{dir} to the head of the list of
8115 directories to be searched for header files. These directories are
8116 interleaved with those specified by @option{-I} options and are
8117 scanned in a left-to-right order.
8119 A framework directory is a directory with frameworks in it. A
8120 framework is a directory with a @samp{"Headers"} and/or
8121 @samp{"PrivateHeaders"} directory contained directly in it that ends
8122 in @samp{".framework"}. The name of a framework is the name of this
8123 directory excluding the @samp{".framework"}. Headers associated with
8124 the framework are found in one of those two directories, with
8125 @samp{"Headers"} being searched first. A subframework is a framework
8126 directory that is in a framework's @samp{"Frameworks"} directory.
8127 Includes of subframework headers can only appear in a header of a
8128 framework that contains the subframework, or in a sibling subframework
8129 header. Two subframeworks are siblings if they occur in the same
8130 framework. A subframework should not have the same name as a
8131 framework, a warning will be issued if this is violated. Currently a
8132 subframework cannot have subframeworks, in the future, the mechanism
8133 may be extended to support this. The standard frameworks can be found
8134 in @samp{"/System/Library/Frameworks"} and
8135 @samp{"/Library/Frameworks"}. An example include looks like
8136 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
8137 the name of the framework and header.h is found in the
8138 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
8142 Emit debugging information for symbols that are used. For STABS
8143 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
8144 This is by default ON@.
8148 Emit debugging information for all symbols and types.
8150 @item -mmacosx-version-min=@var{version}
8151 The earliest version of MacOS X that this executable will run on
8152 is @var{version}. Typical values of @var{version} include @code{10.1},
8153 @code{10.2}, and @code{10.3.9}.
8155 The default for this option is to make choices that seem to be most
8160 Enable kernel development mode. The @option{-mkernel} option sets
8161 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
8162 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
8163 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
8164 applicable. This mode also sets @option{-mno-altivec},
8165 @option{-msoft-float}, @option{-fno-builtin} and
8166 @option{-mlong-branch} for PowerPC targets.
8168 @item -mone-byte-bool
8169 @opindex -mone-byte-bool
8170 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
8171 By default @samp{sizeof(bool)} is @samp{4} when compiling for
8172 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
8173 option has no effect on x86.
8175 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
8176 to generate code that is not binary compatible with code generated
8177 without that switch. Using this switch may require recompiling all
8178 other modules in a program, including system libraries. Use this
8179 switch to conform to a non-default data model.
8181 @item -mfix-and-continue
8182 @itemx -ffix-and-continue
8183 @itemx -findirect-data
8184 @opindex mfix-and-continue
8185 @opindex ffix-and-continue
8186 @opindex findirect-data
8187 Generate code suitable for fast turn around development. Needed to
8188 enable gdb to dynamically load @code{.o} files into already running
8189 programs. @option{-findirect-data} and @option{-ffix-and-continue}
8190 are provided for backwards compatibility.
8194 Loads all members of static archive libraries.
8195 See man ld(1) for more information.
8197 @item -arch_errors_fatal
8198 @opindex arch_errors_fatal
8199 Cause the errors having to do with files that have the wrong architecture
8203 @opindex bind_at_load
8204 Causes the output file to be marked such that the dynamic linker will
8205 bind all undefined references when the file is loaded or launched.
8209 Produce a Mach-o bundle format file.
8210 See man ld(1) for more information.
8212 @item -bundle_loader @var{executable}
8213 @opindex bundle_loader
8214 This option specifies the @var{executable} that will be loading the build
8215 output file being linked. See man ld(1) for more information.
8218 @opindex -dynamiclib
8219 When passed this option, GCC will produce a dynamic library instead of
8220 an executable when linking, using the Darwin @file{libtool} command.
8222 @item -force_cpusubtype_ALL
8223 @opindex -force_cpusubtype_ALL
8224 This causes GCC's output file to have the @var{ALL} subtype, instead of
8225 one controlled by the @option{-mcpu} or @option{-march} option.
8227 @item -allowable_client @var{client_name}
8229 @itemx -compatibility_version
8230 @itemx -current_version
8232 @itemx -dependency-file
8234 @itemx -dylinker_install_name
8236 @itemx -exported_symbols_list
8238 @itemx -flat_namespace
8239 @itemx -force_flat_namespace
8240 @itemx -headerpad_max_install_names
8243 @itemx -install_name
8244 @itemx -keep_private_externs
8245 @itemx -multi_module
8246 @itemx -multiply_defined
8247 @itemx -multiply_defined_unused
8249 @itemx -no_dead_strip_inits_and_terms
8250 @itemx -nofixprebinding
8253 @itemx -noseglinkedit
8254 @itemx -pagezero_size
8256 @itemx -prebind_all_twolevel_modules
8257 @itemx -private_bundle
8258 @itemx -read_only_relocs
8260 @itemx -sectobjectsymbols
8264 @itemx -sectobjectsymbols
8267 @itemx -segs_read_only_addr
8268 @itemx -segs_read_write_addr
8269 @itemx -seg_addr_table
8270 @itemx -seg_addr_table_filename
8273 @itemx -segs_read_only_addr
8274 @itemx -segs_read_write_addr
8275 @itemx -single_module
8278 @itemx -sub_umbrella
8279 @itemx -twolevel_namespace
8282 @itemx -unexported_symbols_list
8283 @itemx -weak_reference_mismatches
8286 @opindex allowable_client
8287 @opindex client_name
8288 @opindex compatibility_version
8289 @opindex current_version
8291 @opindex dependency-file
8293 @opindex dylinker_install_name
8295 @opindex exported_symbols_list
8297 @opindex flat_namespace
8298 @opindex force_flat_namespace
8299 @opindex headerpad_max_install_names
8302 @opindex install_name
8303 @opindex keep_private_externs
8304 @opindex multi_module
8305 @opindex multiply_defined
8306 @opindex multiply_defined_unused
8308 @opindex no_dead_strip_inits_and_terms
8309 @opindex nofixprebinding
8310 @opindex nomultidefs
8312 @opindex noseglinkedit
8313 @opindex pagezero_size
8315 @opindex prebind_all_twolevel_modules
8316 @opindex private_bundle
8317 @opindex read_only_relocs
8319 @opindex sectobjectsymbols
8323 @opindex sectobjectsymbols
8326 @opindex segs_read_only_addr
8327 @opindex segs_read_write_addr
8328 @opindex seg_addr_table
8329 @opindex seg_addr_table_filename
8330 @opindex seglinkedit
8332 @opindex segs_read_only_addr
8333 @opindex segs_read_write_addr
8334 @opindex single_module
8336 @opindex sub_library
8337 @opindex sub_umbrella
8338 @opindex twolevel_namespace
8341 @opindex unexported_symbols_list
8342 @opindex weak_reference_mismatches
8343 @opindex whatsloaded
8345 These options are passed to the Darwin linker. The Darwin linker man page
8346 describes them in detail.
8349 @node DEC Alpha Options
8350 @subsection DEC Alpha Options
8352 These @samp{-m} options are defined for the DEC Alpha implementations:
8355 @item -mno-soft-float
8357 @opindex mno-soft-float
8358 @opindex msoft-float
8359 Use (do not use) the hardware floating-point instructions for
8360 floating-point operations. When @option{-msoft-float} is specified,
8361 functions in @file{libgcc.a} will be used to perform floating-point
8362 operations. Unless they are replaced by routines that emulate the
8363 floating-point operations, or compiled in such a way as to call such
8364 emulations routines, these routines will issue floating-point
8365 operations. If you are compiling for an Alpha without floating-point
8366 operations, you must ensure that the library is built so as not to call
8369 Note that Alpha implementations without floating-point operations are
8370 required to have floating-point registers.
8375 @opindex mno-fp-regs
8376 Generate code that uses (does not use) the floating-point register set.
8377 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8378 register set is not used, floating point operands are passed in integer
8379 registers as if they were integers and floating-point results are passed
8380 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8381 so any function with a floating-point argument or return value called by code
8382 compiled with @option{-mno-fp-regs} must also be compiled with that
8385 A typical use of this option is building a kernel that does not use,
8386 and hence need not save and restore, any floating-point registers.
8390 The Alpha architecture implements floating-point hardware optimized for
8391 maximum performance. It is mostly compliant with the IEEE floating
8392 point standard. However, for full compliance, software assistance is
8393 required. This option generates code fully IEEE compliant code
8394 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8395 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8396 defined during compilation. The resulting code is less efficient but is
8397 able to correctly support denormalized numbers and exceptional IEEE
8398 values such as not-a-number and plus/minus infinity. Other Alpha
8399 compilers call this option @option{-ieee_with_no_inexact}.
8401 @item -mieee-with-inexact
8402 @opindex mieee-with-inexact
8403 This is like @option{-mieee} except the generated code also maintains
8404 the IEEE @var{inexact-flag}. Turning on this option causes the
8405 generated code to implement fully-compliant IEEE math. In addition to
8406 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8407 macro. On some Alpha implementations the resulting code may execute
8408 significantly slower than the code generated by default. Since there is
8409 very little code that depends on the @var{inexact-flag}, you should
8410 normally not specify this option. Other Alpha compilers call this
8411 option @option{-ieee_with_inexact}.
8413 @item -mfp-trap-mode=@var{trap-mode}
8414 @opindex mfp-trap-mode
8415 This option controls what floating-point related traps are enabled.
8416 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8417 The trap mode can be set to one of four values:
8421 This is the default (normal) setting. The only traps that are enabled
8422 are the ones that cannot be disabled in software (e.g., division by zero
8426 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8430 Like @samp{u}, but the instructions are marked to be safe for software
8431 completion (see Alpha architecture manual for details).
8434 Like @samp{su}, but inexact traps are enabled as well.
8437 @item -mfp-rounding-mode=@var{rounding-mode}
8438 @opindex mfp-rounding-mode
8439 Selects the IEEE rounding mode. Other Alpha compilers call this option
8440 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8445 Normal IEEE rounding mode. Floating point numbers are rounded towards
8446 the nearest machine number or towards the even machine number in case
8450 Round towards minus infinity.
8453 Chopped rounding mode. Floating point numbers are rounded towards zero.
8456 Dynamic rounding mode. A field in the floating point control register
8457 (@var{fpcr}, see Alpha architecture reference manual) controls the
8458 rounding mode in effect. The C library initializes this register for
8459 rounding towards plus infinity. Thus, unless your program modifies the
8460 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8463 @item -mtrap-precision=@var{trap-precision}
8464 @opindex mtrap-precision
8465 In the Alpha architecture, floating point traps are imprecise. This
8466 means without software assistance it is impossible to recover from a
8467 floating trap and program execution normally needs to be terminated.
8468 GCC can generate code that can assist operating system trap handlers
8469 in determining the exact location that caused a floating point trap.
8470 Depending on the requirements of an application, different levels of
8471 precisions can be selected:
8475 Program precision. This option is the default and means a trap handler
8476 can only identify which program caused a floating point exception.
8479 Function precision. The trap handler can determine the function that
8480 caused a floating point exception.
8483 Instruction precision. The trap handler can determine the exact
8484 instruction that caused a floating point exception.
8487 Other Alpha compilers provide the equivalent options called
8488 @option{-scope_safe} and @option{-resumption_safe}.
8490 @item -mieee-conformant
8491 @opindex mieee-conformant
8492 This option marks the generated code as IEEE conformant. You must not
8493 use this option unless you also specify @option{-mtrap-precision=i} and either
8494 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8495 is to emit the line @samp{.eflag 48} in the function prologue of the
8496 generated assembly file. Under DEC Unix, this has the effect that
8497 IEEE-conformant math library routines will be linked in.
8499 @item -mbuild-constants
8500 @opindex mbuild-constants
8501 Normally GCC examines a 32- or 64-bit integer constant to
8502 see if it can construct it from smaller constants in two or three
8503 instructions. If it cannot, it will output the constant as a literal and
8504 generate code to load it from the data segment at runtime.
8506 Use this option to require GCC to construct @emph{all} integer constants
8507 using code, even if it takes more instructions (the maximum is six).
8509 You would typically use this option to build a shared library dynamic
8510 loader. Itself a shared library, it must relocate itself in memory
8511 before it can find the variables and constants in its own data segment.
8517 Select whether to generate code to be assembled by the vendor-supplied
8518 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8536 Indicate whether GCC should generate code to use the optional BWX,
8537 CIX, FIX and MAX instruction sets. The default is to use the instruction
8538 sets supported by the CPU type specified via @option{-mcpu=} option or that
8539 of the CPU on which GCC was built if none was specified.
8544 @opindex mfloat-ieee
8545 Generate code that uses (does not use) VAX F and G floating point
8546 arithmetic instead of IEEE single and double precision.
8548 @item -mexplicit-relocs
8549 @itemx -mno-explicit-relocs
8550 @opindex mexplicit-relocs
8551 @opindex mno-explicit-relocs
8552 Older Alpha assemblers provided no way to generate symbol relocations
8553 except via assembler macros. Use of these macros does not allow
8554 optimal instruction scheduling. GNU binutils as of version 2.12
8555 supports a new syntax that allows the compiler to explicitly mark
8556 which relocations should apply to which instructions. This option
8557 is mostly useful for debugging, as GCC detects the capabilities of
8558 the assembler when it is built and sets the default accordingly.
8562 @opindex msmall-data
8563 @opindex mlarge-data
8564 When @option{-mexplicit-relocs} is in effect, static data is
8565 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8566 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8567 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8568 16-bit relocations off of the @code{$gp} register. This limits the
8569 size of the small data area to 64KB, but allows the variables to be
8570 directly accessed via a single instruction.
8572 The default is @option{-mlarge-data}. With this option the data area
8573 is limited to just below 2GB@. Programs that require more than 2GB of
8574 data must use @code{malloc} or @code{mmap} to allocate the data in the
8575 heap instead of in the program's data segment.
8577 When generating code for shared libraries, @option{-fpic} implies
8578 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8582 @opindex msmall-text
8583 @opindex mlarge-text
8584 When @option{-msmall-text} is used, the compiler assumes that the
8585 code of the entire program (or shared library) fits in 4MB, and is
8586 thus reachable with a branch instruction. When @option{-msmall-data}
8587 is used, the compiler can assume that all local symbols share the
8588 same @code{$gp} value, and thus reduce the number of instructions
8589 required for a function call from 4 to 1.
8591 The default is @option{-mlarge-text}.
8593 @item -mcpu=@var{cpu_type}
8595 Set the instruction set and instruction scheduling parameters for
8596 machine type @var{cpu_type}. You can specify either the @samp{EV}
8597 style name or the corresponding chip number. GCC supports scheduling
8598 parameters for the EV4, EV5 and EV6 family of processors and will
8599 choose the default values for the instruction set from the processor
8600 you specify. If you do not specify a processor type, GCC will default
8601 to the processor on which the compiler was built.
8603 Supported values for @var{cpu_type} are
8609 Schedules as an EV4 and has no instruction set extensions.
8613 Schedules as an EV5 and has no instruction set extensions.
8617 Schedules as an EV5 and supports the BWX extension.
8622 Schedules as an EV5 and supports the BWX and MAX extensions.
8626 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8630 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8633 @item -mtune=@var{cpu_type}
8635 Set only the instruction scheduling parameters for machine type
8636 @var{cpu_type}. The instruction set is not changed.
8638 @item -mmemory-latency=@var{time}
8639 @opindex mmemory-latency
8640 Sets the latency the scheduler should assume for typical memory
8641 references as seen by the application. This number is highly
8642 dependent on the memory access patterns used by the application
8643 and the size of the external cache on the machine.
8645 Valid options for @var{time} are
8649 A decimal number representing clock cycles.
8655 The compiler contains estimates of the number of clock cycles for
8656 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8657 (also called Dcache, Scache, and Bcache), as well as to main memory.
8658 Note that L3 is only valid for EV5.
8663 @node DEC Alpha/VMS Options
8664 @subsection DEC Alpha/VMS Options
8666 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8669 @item -mvms-return-codes
8670 @opindex mvms-return-codes
8671 Return VMS condition codes from main. The default is to return POSIX
8672 style condition (e.g.@ error) codes.
8676 @subsection FRV Options
8683 Only use the first 32 general purpose registers.
8688 Use all 64 general purpose registers.
8693 Use only the first 32 floating point registers.
8698 Use all 64 floating point registers
8701 @opindex mhard-float
8703 Use hardware instructions for floating point operations.
8706 @opindex msoft-float
8708 Use library routines for floating point operations.
8713 Dynamically allocate condition code registers.
8718 Do not try to dynamically allocate condition code registers, only
8719 use @code{icc0} and @code{fcc0}.
8724 Change ABI to use double word insns.
8729 Do not use double word instructions.
8734 Use floating point double instructions.
8739 Do not use floating point double instructions.
8744 Use media instructions.
8749 Do not use media instructions.
8754 Use multiply and add/subtract instructions.
8759 Do not use multiply and add/subtract instructions.
8764 Select the FDPIC ABI, that uses function descriptors to represent
8765 pointers to functions. Without any PIC/PIE-related options, it
8766 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
8767 assumes GOT entries and small data are within a 12-bit range from the
8768 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
8769 are computed with 32 bits.
8772 @opindex minline-plt
8774 Enable inlining of PLT entries in function calls to functions that are
8775 not known to bind locally. It has no effect without @option{-mfdpic}.
8776 It's enabled by default if optimizing for speed and compiling for
8777 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
8778 optimization option such as @option{-O3} or above is present in the
8784 Assume a large TLS segment when generating thread-local code.
8789 Do not assume a large TLS segment when generating thread-local code.
8794 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
8795 that is known to be in read-only sections. It's enabled by default,
8796 except for @option{-fpic} or @option{-fpie}: even though it may help
8797 make the global offset table smaller, it trades 1 instruction for 4.
8798 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
8799 one of which may be shared by multiple symbols, and it avoids the need
8800 for a GOT entry for the referenced symbol, so it's more likely to be a
8801 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
8803 @item -multilib-library-pic
8804 @opindex multilib-library-pic
8806 Link with the (library, not FD) pic libraries. It's implied by
8807 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
8808 @option{-fpic} without @option{-mfdpic}. You should never have to use
8814 Follow the EABI requirement of always creating a frame pointer whenever
8815 a stack frame is allocated. This option is enabled by default and can
8816 be disabled with @option{-mno-linked-fp}.
8819 @opindex mlong-calls
8821 Use indirect addressing to call functions outside the current
8822 compilation unit. This allows the functions to be placed anywhere
8823 within the 32-bit address space.
8825 @item -malign-labels
8826 @opindex malign-labels
8828 Try to align labels to an 8-byte boundary by inserting nops into the
8829 previous packet. This option only has an effect when VLIW packing
8830 is enabled. It doesn't create new packets; it merely adds nops to
8834 @opindex mlibrary-pic
8836 Generate position-independent EABI code.
8841 Use only the first four media accumulator registers.
8846 Use all eight media accumulator registers.
8851 Pack VLIW instructions.
8856 Do not pack VLIW instructions.
8861 Do not mark ABI switches in e_flags.
8866 Enable the use of conditional-move instructions (default).
8868 This switch is mainly for debugging the compiler and will likely be removed
8869 in a future version.
8871 @item -mno-cond-move
8872 @opindex mno-cond-move
8874 Disable the use of conditional-move instructions.
8876 This switch is mainly for debugging the compiler and will likely be removed
8877 in a future version.
8882 Enable the use of conditional set instructions (default).
8884 This switch is mainly for debugging the compiler and will likely be removed
8885 in a future version.
8890 Disable the use of conditional set instructions.
8892 This switch is mainly for debugging the compiler and will likely be removed
8893 in a future version.
8898 Enable the use of conditional execution (default).
8900 This switch is mainly for debugging the compiler and will likely be removed
8901 in a future version.
8903 @item -mno-cond-exec
8904 @opindex mno-cond-exec
8906 Disable the use of conditional execution.
8908 This switch is mainly for debugging the compiler and will likely be removed
8909 in a future version.
8912 @opindex mvliw-branch
8914 Run a pass to pack branches into VLIW instructions (default).
8916 This switch is mainly for debugging the compiler and will likely be removed
8917 in a future version.
8919 @item -mno-vliw-branch
8920 @opindex mno-vliw-branch
8922 Do not run a pass to pack branches into VLIW instructions.
8924 This switch is mainly for debugging the compiler and will likely be removed
8925 in a future version.
8927 @item -mmulti-cond-exec
8928 @opindex mmulti-cond-exec
8930 Enable optimization of @code{&&} and @code{||} in conditional execution
8933 This switch is mainly for debugging the compiler and will likely be removed
8934 in a future version.
8936 @item -mno-multi-cond-exec
8937 @opindex mno-multi-cond-exec
8939 Disable optimization of @code{&&} and @code{||} in conditional execution.
8941 This switch is mainly for debugging the compiler and will likely be removed
8942 in a future version.
8944 @item -mnested-cond-exec
8945 @opindex mnested-cond-exec
8947 Enable nested conditional execution optimizations (default).
8949 This switch is mainly for debugging the compiler and will likely be removed
8950 in a future version.
8952 @item -mno-nested-cond-exec
8953 @opindex mno-nested-cond-exec
8955 Disable nested conditional execution optimizations.
8957 This switch is mainly for debugging the compiler and will likely be removed
8958 in a future version.
8960 @item -moptimize-membar
8961 @opindex moptimize-membar
8963 This switch removes redundant @code{membar} instructions from the
8964 compiler generated code. It is enabled by default.
8966 @item -mno-optimize-membar
8967 @opindex mno-optimize-membar
8969 This switch disables the automatic removal of redundant @code{membar}
8970 instructions from the generated code.
8972 @item -mtomcat-stats
8973 @opindex mtomcat-stats
8975 Cause gas to print out tomcat statistics.
8977 @item -mcpu=@var{cpu}
8980 Select the processor type for which to generate code. Possible values are
8981 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
8982 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
8986 @node GNU/Linux Options
8987 @subsection GNU/Linux Options
8989 These @samp{-m} options are defined for GNU/Linux targets:
8994 Use the GNU C library instead of uClibc. This is the default except
8995 on @samp{*-*-linux-*uclibc*} targets.
8999 Use uClibc instead of the GNU C library. This is the default on
9000 @samp{*-*-linux-*uclibc*} targets.
9003 @node H8/300 Options
9004 @subsection H8/300 Options
9006 These @samp{-m} options are defined for the H8/300 implementations:
9011 Shorten some address references at link time, when possible; uses the
9012 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9013 ld, Using ld}, for a fuller description.
9017 Generate code for the H8/300H@.
9021 Generate code for the H8S@.
9025 Generate code for the H8S and H8/300H in the normal mode. This switch
9026 must be used either with @option{-mh} or @option{-ms}.
9030 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9034 Make @code{int} data 32 bits by default.
9038 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9039 The default for the H8/300H and H8S is to align longs and floats on 4
9041 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9042 This option has no effect on the H8/300.
9046 @subsection HPPA Options
9047 @cindex HPPA Options
9049 These @samp{-m} options are defined for the HPPA family of computers:
9052 @item -march=@var{architecture-type}
9054 Generate code for the specified architecture. The choices for
9055 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
9056 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
9057 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
9058 architecture option for your machine. Code compiled for lower numbered
9059 architectures will run on higher numbered architectures, but not the
9063 @itemx -mpa-risc-1-1
9064 @itemx -mpa-risc-2-0
9065 @opindex mpa-risc-1-0
9066 @opindex mpa-risc-1-1
9067 @opindex mpa-risc-2-0
9068 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
9071 @opindex mbig-switch
9072 Generate code suitable for big switch tables. Use this option only if
9073 the assembler/linker complain about out of range branches within a switch
9076 @item -mjump-in-delay
9077 @opindex mjump-in-delay
9078 Fill delay slots of function calls with unconditional jump instructions
9079 by modifying the return pointer for the function call to be the target
9080 of the conditional jump.
9082 @item -mdisable-fpregs
9083 @opindex mdisable-fpregs
9084 Prevent floating point registers from being used in any manner. This is
9085 necessary for compiling kernels which perform lazy context switching of
9086 floating point registers. If you use this option and attempt to perform
9087 floating point operations, the compiler will abort.
9089 @item -mdisable-indexing
9090 @opindex mdisable-indexing
9091 Prevent the compiler from using indexing address modes. This avoids some
9092 rather obscure problems when compiling MIG generated code under MACH@.
9094 @item -mno-space-regs
9095 @opindex mno-space-regs
9096 Generate code that assumes the target has no space registers. This allows
9097 GCC to generate faster indirect calls and use unscaled index address modes.
9099 Such code is suitable for level 0 PA systems and kernels.
9101 @item -mfast-indirect-calls
9102 @opindex mfast-indirect-calls
9103 Generate code that assumes calls never cross space boundaries. This
9104 allows GCC to emit code which performs faster indirect calls.
9106 This option will not work in the presence of shared libraries or nested
9109 @item -mfixed-range=@var{register-range}
9110 @opindex mfixed-range
9111 Generate code treating the given register range as fixed registers.
9112 A fixed register is one that the register allocator can not use. This is
9113 useful when compiling kernel code. A register range is specified as
9114 two registers separated by a dash. Multiple register ranges can be
9115 specified separated by a comma.
9117 @item -mlong-load-store
9118 @opindex mlong-load-store
9119 Generate 3-instruction load and store sequences as sometimes required by
9120 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
9123 @item -mportable-runtime
9124 @opindex mportable-runtime
9125 Use the portable calling conventions proposed by HP for ELF systems.
9129 Enable the use of assembler directives only GAS understands.
9131 @item -mschedule=@var{cpu-type}
9133 Schedule code according to the constraints for the machine type
9134 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
9135 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
9136 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
9137 proper scheduling option for your machine. The default scheduling is
9141 @opindex mlinker-opt
9142 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
9143 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
9144 linkers in which they give bogus error messages when linking some programs.
9147 @opindex msoft-float
9148 Generate output containing library calls for floating point.
9149 @strong{Warning:} the requisite libraries are not available for all HPPA
9150 targets. Normally the facilities of the machine's usual C compiler are
9151 used, but this cannot be done directly in cross-compilation. You must make
9152 your own arrangements to provide suitable library functions for
9153 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
9154 does provide software floating point support.
9156 @option{-msoft-float} changes the calling convention in the output file;
9157 therefore, it is only useful if you compile @emph{all} of a program with
9158 this option. In particular, you need to compile @file{libgcc.a}, the
9159 library that comes with GCC, with @option{-msoft-float} in order for
9164 Generate the predefine, @code{_SIO}, for server IO@. The default is
9165 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
9166 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
9167 options are available under HP-UX and HI-UX@.
9171 Use GNU ld specific options. This passes @option{-shared} to ld when
9172 building a shared library. It is the default when GCC is configured,
9173 explicitly or implicitly, with the GNU linker. This option does not
9174 have any affect on which ld is called, it only changes what parameters
9175 are passed to that ld. The ld that is called is determined by the
9176 @option{--with-ld} configure option, GCC's program search path, and
9177 finally by the user's @env{PATH}. The linker used by GCC can be printed
9178 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
9179 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9183 Use HP ld specific options. This passes @option{-b} to ld when building
9184 a shared library and passes @option{+Accept TypeMismatch} to ld on all
9185 links. It is the default when GCC is configured, explicitly or
9186 implicitly, with the HP linker. This option does not have any affect on
9187 which ld is called, it only changes what parameters are passed to that
9188 ld. The ld that is called is determined by the @option{--with-ld}
9189 configure option, GCC's program search path, and finally by the user's
9190 @env{PATH}. The linker used by GCC can be printed using @samp{which
9191 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
9192 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9195 @opindex mno-long-calls
9196 Generate code that uses long call sequences. This ensures that a call
9197 is always able to reach linker generated stubs. The default is to generate
9198 long calls only when the distance from the call site to the beginning
9199 of the function or translation unit, as the case may be, exceeds a
9200 predefined limit set by the branch type being used. The limits for
9201 normal calls are 7,600,000 and 240,000 bytes, respectively for the
9202 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
9205 Distances are measured from the beginning of functions when using the
9206 @option{-ffunction-sections} option, or when using the @option{-mgas}
9207 and @option{-mno-portable-runtime} options together under HP-UX with
9210 It is normally not desirable to use this option as it will degrade
9211 performance. However, it may be useful in large applications,
9212 particularly when partial linking is used to build the application.
9214 The types of long calls used depends on the capabilities of the
9215 assembler and linker, and the type of code being generated. The
9216 impact on systems that support long absolute calls, and long pic
9217 symbol-difference or pc-relative calls should be relatively small.
9218 However, an indirect call is used on 32-bit ELF systems in pic code
9219 and it is quite long.
9221 @item -munix=@var{unix-std}
9223 Generate compiler predefines and select a startfile for the specified
9224 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
9225 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
9226 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
9227 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
9228 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
9231 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
9232 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
9233 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
9234 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
9235 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
9236 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
9238 It is @emph{important} to note that this option changes the interfaces
9239 for various library routines. It also affects the operational behavior
9240 of the C library. Thus, @emph{extreme} care is needed in using this
9243 Library code that is intended to operate with more than one UNIX
9244 standard must test, set and restore the variable @var{__xpg4_extended_mask}
9245 as appropriate. Most GNU software doesn't provide this capability.
9249 Suppress the generation of link options to search libdld.sl when the
9250 @option{-static} option is specified on HP-UX 10 and later.
9254 The HP-UX implementation of setlocale in libc has a dependency on
9255 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9256 when the @option{-static} option is specified, special link options
9257 are needed to resolve this dependency.
9259 On HP-UX 10 and later, the GCC driver adds the necessary options to
9260 link with libdld.sl when the @option{-static} option is specified.
9261 This causes the resulting binary to be dynamic. On the 64-bit port,
9262 the linkers generate dynamic binaries by default in any case. The
9263 @option{-nolibdld} option can be used to prevent the GCC driver from
9264 adding these link options.
9268 Add support for multithreading with the @dfn{dce thread} library
9269 under HP-UX@. This option sets flags for both the preprocessor and
9273 @node i386 and x86-64 Options
9274 @subsection Intel 386 and AMD x86-64 Options
9275 @cindex i386 Options
9276 @cindex x86-64 Options
9277 @cindex Intel 386 Options
9278 @cindex AMD x86-64 Options
9280 These @samp{-m} options are defined for the i386 and x86-64 family of
9284 @item -mtune=@var{cpu-type}
9286 Tune to @var{cpu-type} everything applicable about the generated code, except
9287 for the ABI and the set of available instructions. The choices for
9291 Produce code optimized for the most common IA32/AMD64/EM64T processors.
9292 If you know the CPU on which your code will run, then you should use
9293 the corresponding @option{-mtune} option instead of
9294 @option{-mtune=generic}. But, if you do not know exactly what CPU users
9295 of your application will have, then you should use this option.
9297 As new processors are deployed in the marketplace, the behavior of this
9298 option will change. Therefore, if you upgrade to a newer version of
9299 GCC, the code generated option will change to reflect the processors
9300 that were most common when that version of GCC was released.
9302 There is no @option{-march=generic} option because @option{-march}
9303 indicates the instruction set the compiler can use, and there is no
9304 generic instruction set applicable to all processors. In contrast,
9305 @option{-mtune} indicates the processor (or, in this case, collection of
9306 processors) for which the code is optimized.
9308 This selects the CPU to tune for at compilation time by determining
9309 the processor type of the compiling machine. Using @option{-mtune=native}
9310 will produce code optimized for the local machine under the constraints
9311 of the selected instruction set. Using @option{-march=native} will
9312 enable all instruction subsets supported by the local machine (hence
9313 the result might not run on different machines).
9315 Original Intel's i386 CPU@.
9317 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
9319 Intel Pentium CPU with no MMX support.
9321 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
9323 Intel PentiumPro CPU@.
9325 Same as @code{generic}, but when used as @code{march} option, PentiumPro
9326 instruction set will be used, so the code will run on all i686 family chips.
9328 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
9329 @item pentium3, pentium3m
9330 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
9333 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
9334 support. Used by Centrino notebooks.
9335 @item pentium4, pentium4m
9336 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
9338 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
9341 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
9342 SSE2 and SSE3 instruction set support.
9344 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
9345 instruction set support.
9347 AMD K6 CPU with MMX instruction set support.
9349 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
9350 @item athlon, athlon-tbird
9351 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
9353 @item athlon-4, athlon-xp, athlon-mp
9354 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
9355 instruction set support.
9356 @item k8, opteron, athlon64, athlon-fx
9357 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
9358 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
9360 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
9363 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
9364 instruction set support.
9366 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
9367 implemented for this chip.)
9369 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
9370 implemented for this chip.)
9372 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
9375 While picking a specific @var{cpu-type} will schedule things appropriately
9376 for that particular chip, the compiler will not generate any code that
9377 does not run on the i386 without the @option{-march=@var{cpu-type}} option
9380 @item -march=@var{cpu-type}
9382 Generate instructions for the machine type @var{cpu-type}. The choices
9383 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
9384 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
9386 @item -mcpu=@var{cpu-type}
9388 A deprecated synonym for @option{-mtune}.
9397 @opindex mpentiumpro
9398 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
9399 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
9400 These synonyms are deprecated.
9402 @item -mfpmath=@var{unit}
9404 Generate floating point arithmetics for selected unit @var{unit}. The choices
9409 Use the standard 387 floating point coprocessor present majority of chips and
9410 emulated otherwise. Code compiled with this option will run almost everywhere.
9411 The temporary results are computed in 80bit precision instead of precision
9412 specified by the type resulting in slightly different results compared to most
9413 of other chips. See @option{-ffloat-store} for more detailed description.
9415 This is the default choice for i386 compiler.
9418 Use scalar floating point instructions present in the SSE instruction set.
9419 This instruction set is supported by Pentium3 and newer chips, in the AMD line
9420 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
9421 instruction set supports only single precision arithmetics, thus the double and
9422 extended precision arithmetics is still done using 387. Later version, present
9423 only in Pentium4 and the future AMD x86-64 chips supports double precision
9426 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
9427 or @option{-msse2} switches to enable SSE extensions and make this option
9428 effective. For the x86-64 compiler, these extensions are enabled by default.
9430 The resulting code should be considerably faster in the majority of cases and avoid
9431 the numerical instability problems of 387 code, but may break some existing
9432 code that expects temporaries to be 80bit.
9434 This is the default choice for the x86-64 compiler.
9437 Attempt to utilize both instruction sets at once. This effectively double the
9438 amount of available registers and on chips with separate execution units for
9439 387 and SSE the execution resources too. Use this option with care, as it is
9440 still experimental, because the GCC register allocator does not model separate
9441 functional units well resulting in instable performance.
9444 @item -masm=@var{dialect}
9445 @opindex masm=@var{dialect}
9446 Output asm instructions using selected @var{dialect}. Supported
9447 choices are @samp{intel} or @samp{att} (the default one). Darwin does
9448 not support @samp{intel}.
9453 @opindex mno-ieee-fp
9454 Control whether or not the compiler uses IEEE floating point
9455 comparisons. These handle correctly the case where the result of a
9456 comparison is unordered.
9459 @opindex msoft-float
9460 Generate output containing library calls for floating point.
9461 @strong{Warning:} the requisite libraries are not part of GCC@.
9462 Normally the facilities of the machine's usual C compiler are used, but
9463 this can't be done directly in cross-compilation. You must make your
9464 own arrangements to provide suitable library functions for
9467 On machines where a function returns floating point results in the 80387
9468 register stack, some floating point opcodes may be emitted even if
9469 @option{-msoft-float} is used.
9471 @item -mno-fp-ret-in-387
9472 @opindex mno-fp-ret-in-387
9473 Do not use the FPU registers for return values of functions.
9475 The usual calling convention has functions return values of types
9476 @code{float} and @code{double} in an FPU register, even if there
9477 is no FPU@. The idea is that the operating system should emulate
9480 The option @option{-mno-fp-ret-in-387} causes such values to be returned
9481 in ordinary CPU registers instead.
9483 @item -mno-fancy-math-387
9484 @opindex mno-fancy-math-387
9485 Some 387 emulators do not support the @code{sin}, @code{cos} and
9486 @code{sqrt} instructions for the 387. Specify this option to avoid
9487 generating those instructions. This option is the default on FreeBSD,
9488 OpenBSD and NetBSD@. This option is overridden when @option{-march}
9489 indicates that the target cpu will always have an FPU and so the
9490 instruction will not need emulation. As of revision 2.6.1, these
9491 instructions are not generated unless you also use the
9492 @option{-funsafe-math-optimizations} switch.
9494 @item -malign-double
9495 @itemx -mno-align-double
9496 @opindex malign-double
9497 @opindex mno-align-double
9498 Control whether GCC aligns @code{double}, @code{long double}, and
9499 @code{long long} variables on a two word boundary or a one word
9500 boundary. Aligning @code{double} variables on a two word boundary will
9501 produce code that runs somewhat faster on a @samp{Pentium} at the
9502 expense of more memory.
9504 On x86-64, @option{-malign-double} is enabled by default.
9506 @strong{Warning:} if you use the @option{-malign-double} switch,
9507 structures containing the above types will be aligned differently than
9508 the published application binary interface specifications for the 386
9509 and will not be binary compatible with structures in code compiled
9510 without that switch.
9512 @item -m96bit-long-double
9513 @itemx -m128bit-long-double
9514 @opindex m96bit-long-double
9515 @opindex m128bit-long-double
9516 These switches control the size of @code{long double} type. The i386
9517 application binary interface specifies the size to be 96 bits,
9518 so @option{-m96bit-long-double} is the default in 32 bit mode.
9520 Modern architectures (Pentium and newer) would prefer @code{long double}
9521 to be aligned to an 8 or 16 byte boundary. In arrays or structures
9522 conforming to the ABI, this would not be possible. So specifying a
9523 @option{-m128bit-long-double} will align @code{long double}
9524 to a 16 byte boundary by padding the @code{long double} with an additional
9527 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
9528 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
9530 Notice that neither of these options enable any extra precision over the x87
9531 standard of 80 bits for a @code{long double}.
9533 @strong{Warning:} if you override the default value for your target ABI, the
9534 structures and arrays containing @code{long double} variables will change
9535 their size as well as function calling convention for function taking
9536 @code{long double} will be modified. Hence they will not be binary
9537 compatible with arrays or structures in code compiled without that switch.
9539 @item -mmlarge-data-threshold=@var{number}
9540 @opindex mlarge-data-threshold=@var{number}
9541 When @option{-mcmodel=medium} is specified, the data greater than
9542 @var{threshold} are placed in large data section. This value must be the
9543 same across all object linked into the binary and defaults to 65535.
9546 @itemx -mno-svr3-shlib
9547 @opindex msvr3-shlib
9548 @opindex mno-svr3-shlib
9549 Control whether GCC places uninitialized local variables into the
9550 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
9551 into @code{bss}. These options are meaningful only on System V Release 3.
9555 Use a different function-calling convention, in which functions that
9556 take a fixed number of arguments return with the @code{ret} @var{num}
9557 instruction, which pops their arguments while returning. This saves one
9558 instruction in the caller since there is no need to pop the arguments
9561 You can specify that an individual function is called with this calling
9562 sequence with the function attribute @samp{stdcall}. You can also
9563 override the @option{-mrtd} option by using the function attribute
9564 @samp{cdecl}. @xref{Function Attributes}.
9566 @strong{Warning:} this calling convention is incompatible with the one
9567 normally used on Unix, so you cannot use it if you need to call
9568 libraries compiled with the Unix compiler.
9570 Also, you must provide function prototypes for all functions that
9571 take variable numbers of arguments (including @code{printf});
9572 otherwise incorrect code will be generated for calls to those
9575 In addition, seriously incorrect code will result if you call a
9576 function with too many arguments. (Normally, extra arguments are
9577 harmlessly ignored.)
9579 @item -mregparm=@var{num}
9581 Control how many registers are used to pass integer arguments. By
9582 default, no registers are used to pass arguments, and at most 3
9583 registers can be used. You can control this behavior for a specific
9584 function by using the function attribute @samp{regparm}.
9585 @xref{Function Attributes}.
9587 @strong{Warning:} if you use this switch, and
9588 @var{num} is nonzero, then you must build all modules with the same
9589 value, including any libraries. This includes the system libraries and
9593 @opindex msseregparm
9594 Use SSE register passing conventions for float and double arguments
9595 and return values. You can control this behavior for a specific
9596 function by using the function attribute @samp{sseregparm}.
9597 @xref{Function Attributes}.
9599 @strong{Warning:} if you use this switch then you must build all
9600 modules with the same value, including any libraries. This includes
9601 the system libraries and startup modules.
9603 @item -mstackrealign
9604 @opindex mstackrealign
9605 Realign the stack at entry. On the Intel x86, the
9606 @option{-mstackrealign} option will generate an alternate prologue and
9607 epilogue that realigns the runtime stack. This supports mixing legacy
9608 codes that keep a 4-byte aligned stack with modern codes that keep a
9609 16-byte stack for SSE compatibility. The alternate prologue and
9610 epilogue are slower and bigger than the regular ones, and the
9611 alternate prologue requires an extra scratch register; this lowers the
9612 number of registers available if used in conjunction with the
9613 @code{regparm} attribute. The @option{-mstackrealign} option is
9614 incompatible with the nested function prologue; this is considered a
9615 hard error. See also the attribute @code{force_align_arg_pointer},
9616 applicable to individual functions.
9618 @item -mpreferred-stack-boundary=@var{num}
9619 @opindex mpreferred-stack-boundary
9620 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
9621 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
9622 the default is 4 (16 bytes or 128 bits).
9624 On Pentium and PentiumPro, @code{double} and @code{long double} values
9625 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
9626 suffer significant run time performance penalties. On Pentium III, the
9627 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
9628 properly if it is not 16 byte aligned.
9630 To ensure proper alignment of this values on the stack, the stack boundary
9631 must be as aligned as that required by any value stored on the stack.
9632 Further, every function must be generated such that it keeps the stack
9633 aligned. Thus calling a function compiled with a higher preferred
9634 stack boundary from a function compiled with a lower preferred stack
9635 boundary will most likely misalign the stack. It is recommended that
9636 libraries that use callbacks always use the default setting.
9638 This extra alignment does consume extra stack space, and generally
9639 increases code size. Code that is sensitive to stack space usage, such
9640 as embedded systems and operating system kernels, may want to reduce the
9641 preferred alignment to @option{-mpreferred-stack-boundary=2}.
9661 These switches enable or disable the use of instructions in the MMX,
9662 SSE, SSE2, SSE3, SSSE3 or 3DNow! extended instruction sets.
9663 These extensions are also available as built-in functions: see
9664 @ref{X86 Built-in Functions}, for details of the functions enabled and
9665 disabled by these switches.
9667 To have SSE/SSE2 instructions generated automatically from floating-point
9668 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
9670 These options will enable GCC to use these extended instructions in
9671 generated code, even without @option{-mfpmath=sse}. Applications which
9672 perform runtime CPU detection must compile separate files for each
9673 supported architecture, using the appropriate flags. In particular,
9674 the file containing the CPU detection code should be compiled without
9678 @itemx -mno-push-args
9680 @opindex mno-push-args
9681 Use PUSH operations to store outgoing parameters. This method is shorter
9682 and usually equally fast as method using SUB/MOV operations and is enabled
9683 by default. In some cases disabling it may improve performance because of
9684 improved scheduling and reduced dependencies.
9686 @item -maccumulate-outgoing-args
9687 @opindex maccumulate-outgoing-args
9688 If enabled, the maximum amount of space required for outgoing arguments will be
9689 computed in the function prologue. This is faster on most modern CPUs
9690 because of reduced dependencies, improved scheduling and reduced stack usage
9691 when preferred stack boundary is not equal to 2. The drawback is a notable
9692 increase in code size. This switch implies @option{-mno-push-args}.
9696 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
9697 on thread-safe exception handling must compile and link all code with the
9698 @option{-mthreads} option. When compiling, @option{-mthreads} defines
9699 @option{-D_MT}; when linking, it links in a special thread helper library
9700 @option{-lmingwthrd} which cleans up per thread exception handling data.
9702 @item -mno-align-stringops
9703 @opindex mno-align-stringops
9704 Do not align destination of inlined string operations. This switch reduces
9705 code size and improves performance in case the destination is already aligned,
9706 but GCC doesn't know about it.
9708 @item -minline-all-stringops
9709 @opindex minline-all-stringops
9710 By default GCC inlines string operations only when destination is known to be
9711 aligned at least to 4 byte boundary. This enables more inlining, increase code
9712 size, but may improve performance of code that depends on fast memcpy, strlen
9713 and memset for short lengths.
9715 @item -minline-stringops-dynamically
9716 @opindex minline-stringops-dynamically
9717 For string operation of unknown size, inline runtime checks so for small
9718 blocks inline code is used, while for large blocks library call is used.
9720 @item -mstringop-strategy=@var{alg}
9721 @opindex mstringop-strategy=@var{alg}
9722 Overwrite internal decision heuristic about particular algorithm to inline
9723 string operation with. The allowed values are @code{rep_byte},
9724 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
9725 of specified size, @code{loop}, @code{unrolled_loop} for expanding inline loop,
9726 @code{libcall} for always expanding library call.
9728 @item -momit-leaf-frame-pointer
9729 @opindex momit-leaf-frame-pointer
9730 Don't keep the frame pointer in a register for leaf functions. This
9731 avoids the instructions to save, set up and restore frame pointers and
9732 makes an extra register available in leaf functions. The option
9733 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9734 which might make debugging harder.
9736 @item -mtls-direct-seg-refs
9737 @itemx -mno-tls-direct-seg-refs
9738 @opindex mtls-direct-seg-refs
9739 Controls whether TLS variables may be accessed with offsets from the
9740 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
9741 or whether the thread base pointer must be added. Whether or not this
9742 is legal depends on the operating system, and whether it maps the
9743 segment to cover the entire TLS area.
9745 For systems that use GNU libc, the default is on.
9748 These @samp{-m} switches are supported in addition to the above
9749 on AMD x86-64 processors in 64-bit environments.
9756 Generate code for a 32-bit or 64-bit environment.
9757 The 32-bit environment sets int, long and pointer to 32 bits and
9758 generates code that runs on any i386 system.
9759 The 64-bit environment sets int to 32 bits and long and pointer
9760 to 64 bits and generates code for AMD's x86-64 architecture.
9763 @opindex no-red-zone
9764 Do not use a so called red zone for x86-64 code. The red zone is mandated
9765 by the x86-64 ABI, it is a 128-byte area beyond the location of the
9766 stack pointer that will not be modified by signal or interrupt handlers
9767 and therefore can be used for temporary data without adjusting the stack
9768 pointer. The flag @option{-mno-red-zone} disables this red zone.
9770 @item -mcmodel=small
9771 @opindex mcmodel=small
9772 Generate code for the small code model: the program and its symbols must
9773 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
9774 Programs can be statically or dynamically linked. This is the default
9777 @item -mcmodel=kernel
9778 @opindex mcmodel=kernel
9779 Generate code for the kernel code model. The kernel runs in the
9780 negative 2 GB of the address space.
9781 This model has to be used for Linux kernel code.
9783 @item -mcmodel=medium
9784 @opindex mcmodel=medium
9785 Generate code for the medium model: The program is linked in the lower 2
9786 GB of the address space but symbols can be located anywhere in the
9787 address space. Programs can be statically or dynamically linked, but
9788 building of shared libraries are not supported with the medium model.
9790 @item -mcmodel=large
9791 @opindex mcmodel=large
9792 Generate code for the large model: This model makes no assumptions
9793 about addresses and sizes of sections. Currently GCC does not implement
9798 @subsection IA-64 Options
9799 @cindex IA-64 Options
9801 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9805 @opindex mbig-endian
9806 Generate code for a big endian target. This is the default for HP-UX@.
9808 @item -mlittle-endian
9809 @opindex mlittle-endian
9810 Generate code for a little endian target. This is the default for AIX5
9817 Generate (or don't) code for the GNU assembler. This is the default.
9818 @c Also, this is the default if the configure option @option{--with-gnu-as}
9825 Generate (or don't) code for the GNU linker. This is the default.
9826 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9831 Generate code that does not use a global pointer register. The result
9832 is not position independent code, and violates the IA-64 ABI@.
9834 @item -mvolatile-asm-stop
9835 @itemx -mno-volatile-asm-stop
9836 @opindex mvolatile-asm-stop
9837 @opindex mno-volatile-asm-stop
9838 Generate (or don't) a stop bit immediately before and after volatile asm
9841 @item -mregister-names
9842 @itemx -mno-register-names
9843 @opindex mregister-names
9844 @opindex mno-register-names
9845 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9846 the stacked registers. This may make assembler output more readable.
9852 Disable (or enable) optimizations that use the small data section. This may
9853 be useful for working around optimizer bugs.
9856 @opindex mconstant-gp
9857 Generate code that uses a single constant global pointer value. This is
9858 useful when compiling kernel code.
9862 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9863 This is useful when compiling firmware code.
9865 @item -minline-float-divide-min-latency
9866 @opindex minline-float-divide-min-latency
9867 Generate code for inline divides of floating point values
9868 using the minimum latency algorithm.
9870 @item -minline-float-divide-max-throughput
9871 @opindex minline-float-divide-max-throughput
9872 Generate code for inline divides of floating point values
9873 using the maximum throughput algorithm.
9875 @item -minline-int-divide-min-latency
9876 @opindex minline-int-divide-min-latency
9877 Generate code for inline divides of integer values
9878 using the minimum latency algorithm.
9880 @item -minline-int-divide-max-throughput
9881 @opindex minline-int-divide-max-throughput
9882 Generate code for inline divides of integer values
9883 using the maximum throughput algorithm.
9885 @item -minline-sqrt-min-latency
9886 @opindex minline-sqrt-min-latency
9887 Generate code for inline square roots
9888 using the minimum latency algorithm.
9890 @item -minline-sqrt-max-throughput
9891 @opindex minline-sqrt-max-throughput
9892 Generate code for inline square roots
9893 using the maximum throughput algorithm.
9895 @item -mno-dwarf2-asm
9897 @opindex mno-dwarf2-asm
9898 @opindex mdwarf2-asm
9899 Don't (or do) generate assembler code for the DWARF2 line number debugging
9900 info. This may be useful when not using the GNU assembler.
9902 @item -mearly-stop-bits
9903 @itemx -mno-early-stop-bits
9904 @opindex mearly-stop-bits
9905 @opindex mno-early-stop-bits
9906 Allow stop bits to be placed earlier than immediately preceding the
9907 instruction that triggered the stop bit. This can improve instruction
9908 scheduling, but does not always do so.
9910 @item -mfixed-range=@var{register-range}
9911 @opindex mfixed-range
9912 Generate code treating the given register range as fixed registers.
9913 A fixed register is one that the register allocator can not use. This is
9914 useful when compiling kernel code. A register range is specified as
9915 two registers separated by a dash. Multiple register ranges can be
9916 specified separated by a comma.
9918 @item -mtls-size=@var{tls-size}
9920 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
9923 @item -mtune=@var{cpu-type}
9925 Tune the instruction scheduling for a particular CPU, Valid values are
9926 itanium, itanium1, merced, itanium2, and mckinley.
9932 Add support for multithreading using the POSIX threads library. This
9933 option sets flags for both the preprocessor and linker. It does
9934 not affect the thread safety of object code produced by the compiler or
9935 that of libraries supplied with it. These are HP-UX specific flags.
9941 Generate code for a 32-bit or 64-bit environment.
9942 The 32-bit environment sets int, long and pointer to 32 bits.
9943 The 64-bit environment sets int to 32 bits and long and pointer
9944 to 64 bits. These are HP-UX specific flags.
9946 @item -mno-sched-br-data-spec
9947 @itemx -msched-br-data-spec
9948 @opindex -mno-sched-br-data-spec
9949 @opindex -msched-br-data-spec
9950 (Dis/En)able data speculative scheduling before reload.
9951 This will result in generation of the ld.a instructions and
9952 the corresponding check instructions (ld.c / chk.a).
9953 The default is 'disable'.
9955 @item -msched-ar-data-spec
9956 @itemx -mno-sched-ar-data-spec
9957 @opindex -msched-ar-data-spec
9958 @opindex -mno-sched-ar-data-spec
9959 (En/Dis)able data speculative scheduling after reload.
9960 This will result in generation of the ld.a instructions and
9961 the corresponding check instructions (ld.c / chk.a).
9962 The default is 'enable'.
9964 @item -mno-sched-control-spec
9965 @itemx -msched-control-spec
9966 @opindex -mno-sched-control-spec
9967 @opindex -msched-control-spec
9968 (Dis/En)able control speculative scheduling. This feature is
9969 available only during region scheduling (i.e. before reload).
9970 This will result in generation of the ld.s instructions and
9971 the corresponding check instructions chk.s .
9972 The default is 'disable'.
9974 @item -msched-br-in-data-spec
9975 @itemx -mno-sched-br-in-data-spec
9976 @opindex -msched-br-in-data-spec
9977 @opindex -mno-sched-br-in-data-spec
9978 (En/Dis)able speculative scheduling of the instructions that
9979 are dependent on the data speculative loads before reload.
9980 This is effective only with @option{-msched-br-data-spec} enabled.
9981 The default is 'enable'.
9983 @item -msched-ar-in-data-spec
9984 @itemx -mno-sched-ar-in-data-spec
9985 @opindex -msched-ar-in-data-spec
9986 @opindex -mno-sched-ar-in-data-spec
9987 (En/Dis)able speculative scheduling of the instructions that
9988 are dependent on the data speculative loads after reload.
9989 This is effective only with @option{-msched-ar-data-spec} enabled.
9990 The default is 'enable'.
9992 @item -msched-in-control-spec
9993 @itemx -mno-sched-in-control-spec
9994 @opindex -msched-in-control-spec
9995 @opindex -mno-sched-in-control-spec
9996 (En/Dis)able speculative scheduling of the instructions that
9997 are dependent on the control speculative loads.
9998 This is effective only with @option{-msched-control-spec} enabled.
9999 The default is 'enable'.
10002 @itemx -mno-sched-ldc
10003 @opindex -msched-ldc
10004 @opindex -mno-sched-ldc
10005 (En/Dis)able use of simple data speculation checks ld.c .
10006 If disabled, only chk.a instructions will be emitted to check
10007 data speculative loads.
10008 The default is 'enable'.
10010 @item -mno-sched-control-ldc
10011 @itemx -msched-control-ldc
10012 @opindex -mno-sched-control-ldc
10013 @opindex -msched-control-ldc
10014 (Dis/En)able use of ld.c instructions to check control speculative loads.
10015 If enabled, in case of control speculative load with no speculatively
10016 scheduled dependent instructions this load will be emitted as ld.sa and
10017 ld.c will be used to check it.
10018 The default is 'disable'.
10020 @item -mno-sched-spec-verbose
10021 @itemx -msched-spec-verbose
10022 @opindex -mno-sched-spec-verbose
10023 @opindex -msched-spec-verbose
10024 (Dis/En)able printing of the information about speculative motions.
10026 @item -mno-sched-prefer-non-data-spec-insns
10027 @itemx -msched-prefer-non-data-spec-insns
10028 @opindex -mno-sched-prefer-non-data-spec-insns
10029 @opindex -msched-prefer-non-data-spec-insns
10030 If enabled, data speculative instructions will be chosen for schedule
10031 only if there are no other choices at the moment. This will make
10032 the use of the data speculation much more conservative.
10033 The default is 'disable'.
10035 @item -mno-sched-prefer-non-control-spec-insns
10036 @itemx -msched-prefer-non-control-spec-insns
10037 @opindex -mno-sched-prefer-non-control-spec-insns
10038 @opindex -msched-prefer-non-control-spec-insns
10039 If enabled, control speculative instructions will be chosen for schedule
10040 only if there are no other choices at the moment. This will make
10041 the use of the control speculation much more conservative.
10042 The default is 'disable'.
10044 @item -mno-sched-count-spec-in-critical-path
10045 @itemx -msched-count-spec-in-critical-path
10046 @opindex -mno-sched-count-spec-in-critical-path
10047 @opindex -msched-count-spec-in-critical-path
10048 If enabled, speculative dependencies will be considered during
10049 computation of the instructions priorities. This will make the use of the
10050 speculation a bit more conservative.
10051 The default is 'disable'.
10056 @subsection M32C Options
10057 @cindex M32C options
10060 @item -mcpu=@var{name}
10062 Select the CPU for which code is generated. @var{name} may be one of
10063 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
10064 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
10065 the M32C/80 series.
10069 Specifies that the program will be run on the simulator. This causes
10070 an alternate runtime library to be linked in which supports, for
10071 example, file I/O. You must not use this option when generating
10072 programs that will run on real hardware; you must provide your own
10073 runtime library for whatever I/O functions are needed.
10075 @item -memregs=@var{number}
10077 Specifies the number of memory-based pseudo-registers GCC will use
10078 during code generation. These pseudo-registers will be used like real
10079 registers, so there is a tradeoff between GCC's ability to fit the
10080 code into available registers, and the performance penalty of using
10081 memory instead of registers. Note that all modules in a program must
10082 be compiled with the same value for this option. Because of that, you
10083 must not use this option with the default runtime libraries gcc
10088 @node M32R/D Options
10089 @subsection M32R/D Options
10090 @cindex M32R/D options
10092 These @option{-m} options are defined for Renesas M32R/D architectures:
10097 Generate code for the M32R/2@.
10101 Generate code for the M32R/X@.
10105 Generate code for the M32R@. This is the default.
10107 @item -mmodel=small
10108 @opindex mmodel=small
10109 Assume all objects live in the lower 16MB of memory (so that their addresses
10110 can be loaded with the @code{ld24} instruction), and assume all subroutines
10111 are reachable with the @code{bl} instruction.
10112 This is the default.
10114 The addressability of a particular object can be set with the
10115 @code{model} attribute.
10117 @item -mmodel=medium
10118 @opindex mmodel=medium
10119 Assume objects may be anywhere in the 32-bit address space (the compiler
10120 will generate @code{seth/add3} instructions to load their addresses), and
10121 assume all subroutines are reachable with the @code{bl} instruction.
10123 @item -mmodel=large
10124 @opindex mmodel=large
10125 Assume objects may be anywhere in the 32-bit address space (the compiler
10126 will generate @code{seth/add3} instructions to load their addresses), and
10127 assume subroutines may not be reachable with the @code{bl} instruction
10128 (the compiler will generate the much slower @code{seth/add3/jl}
10129 instruction sequence).
10132 @opindex msdata=none
10133 Disable use of the small data area. Variables will be put into
10134 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
10135 @code{section} attribute has been specified).
10136 This is the default.
10138 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
10139 Objects may be explicitly put in the small data area with the
10140 @code{section} attribute using one of these sections.
10142 @item -msdata=sdata
10143 @opindex msdata=sdata
10144 Put small global and static data in the small data area, but do not
10145 generate special code to reference them.
10148 @opindex msdata=use
10149 Put small global and static data in the small data area, and generate
10150 special instructions to reference them.
10154 @cindex smaller data references
10155 Put global and static objects less than or equal to @var{num} bytes
10156 into the small data or bss sections instead of the normal data or bss
10157 sections. The default value of @var{num} is 8.
10158 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
10159 for this option to have any effect.
10161 All modules should be compiled with the same @option{-G @var{num}} value.
10162 Compiling with different values of @var{num} may or may not work; if it
10163 doesn't the linker will give an error message---incorrect code will not be
10168 Makes the M32R specific code in the compiler display some statistics
10169 that might help in debugging programs.
10171 @item -malign-loops
10172 @opindex malign-loops
10173 Align all loops to a 32-byte boundary.
10175 @item -mno-align-loops
10176 @opindex mno-align-loops
10177 Do not enforce a 32-byte alignment for loops. This is the default.
10179 @item -missue-rate=@var{number}
10180 @opindex missue-rate=@var{number}
10181 Issue @var{number} instructions per cycle. @var{number} can only be 1
10184 @item -mbranch-cost=@var{number}
10185 @opindex mbranch-cost=@var{number}
10186 @var{number} can only be 1 or 2. If it is 1 then branches will be
10187 preferred over conditional code, if it is 2, then the opposite will
10190 @item -mflush-trap=@var{number}
10191 @opindex mflush-trap=@var{number}
10192 Specifies the trap number to use to flush the cache. The default is
10193 12. Valid numbers are between 0 and 15 inclusive.
10195 @item -mno-flush-trap
10196 @opindex mno-flush-trap
10197 Specifies that the cache cannot be flushed by using a trap.
10199 @item -mflush-func=@var{name}
10200 @opindex mflush-func=@var{name}
10201 Specifies the name of the operating system function to call to flush
10202 the cache. The default is @emph{_flush_cache}, but a function call
10203 will only be used if a trap is not available.
10205 @item -mno-flush-func
10206 @opindex mno-flush-func
10207 Indicates that there is no OS function for flushing the cache.
10211 @node M680x0 Options
10212 @subsection M680x0 Options
10213 @cindex M680x0 options
10215 These are the @samp{-m} options defined for the 68000 series. The default
10216 values for these options depends on which style of 68000 was selected when
10217 the compiler was configured; the defaults for the most common choices are
10225 Generate output for a 68000. This is the default
10226 when the compiler is configured for 68000-based systems.
10228 Use this option for microcontrollers with a 68000 or EC000 core,
10229 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
10235 Generate output for a 68020. This is the default
10236 when the compiler is configured for 68020-based systems.
10240 Generate output containing 68881 instructions for floating point.
10241 This is the default for most 68020 systems unless @option{--nfp} was
10242 specified when the compiler was configured.
10246 Generate output for a 68030. This is the default when the compiler is
10247 configured for 68030-based systems.
10251 Generate output for a 68040. This is the default when the compiler is
10252 configured for 68040-based systems.
10254 This option inhibits the use of 68881/68882 instructions that have to be
10255 emulated by software on the 68040. Use this option if your 68040 does not
10256 have code to emulate those instructions.
10260 Generate output for a 68060. This is the default when the compiler is
10261 configured for 68060-based systems.
10263 This option inhibits the use of 68020 and 68881/68882 instructions that
10264 have to be emulated by software on the 68060. Use this option if your 68060
10265 does not have code to emulate those instructions.
10269 Generate output for a CPU32. This is the default
10270 when the compiler is configured for CPU32-based systems.
10272 Use this option for microcontrollers with a
10273 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
10274 68336, 68340, 68341, 68349 and 68360.
10278 Generate output for a 520X ``coldfire'' family cpu. This is the default
10279 when the compiler is configured for 520X-based systems.
10281 Use this option for microcontroller with a 5200 core, including
10282 the MCF5202, MCF5203, MCF5204 and MCF5202.
10286 Generate output for a ColdFire V4e family cpu (e.g.@: 547x/548x).
10287 This includes use of hardware floating point instructions.
10291 Generate output for a 68040, without using any of the new instructions.
10292 This results in code which can run relatively efficiently on either a
10293 68020/68881 or a 68030 or a 68040. The generated code does use the
10294 68881 instructions that are emulated on the 68040.
10298 Generate output for a 68060, without using any of the new instructions.
10299 This results in code which can run relatively efficiently on either a
10300 68020/68881 or a 68030 or a 68040. The generated code does use the
10301 68881 instructions that are emulated on the 68060.
10304 @opindex msoft-float
10305 Generate output containing library calls for floating point.
10306 @strong{Warning:} the requisite libraries are not available for all m68k
10307 targets. Normally the facilities of the machine's usual C compiler are
10308 used, but this can't be done directly in cross-compilation. You must
10309 make your own arrangements to provide suitable library functions for
10310 cross-compilation. The embedded targets @samp{m68k-*-aout} and
10311 @samp{m68k-*-coff} do provide software floating point support.
10315 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10316 Additionally, parameters passed on the stack are also aligned to a
10317 16-bit boundary even on targets whose API mandates promotion to 32-bit.
10320 @opindex mnobitfield
10321 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
10322 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
10326 Do use the bit-field instructions. The @option{-m68020} option implies
10327 @option{-mbitfield}. This is the default if you use a configuration
10328 designed for a 68020.
10332 Use a different function-calling convention, in which functions
10333 that take a fixed number of arguments return with the @code{rtd}
10334 instruction, which pops their arguments while returning. This
10335 saves one instruction in the caller since there is no need to pop
10336 the arguments there.
10338 This calling convention is incompatible with the one normally
10339 used on Unix, so you cannot use it if you need to call libraries
10340 compiled with the Unix compiler.
10342 Also, you must provide function prototypes for all functions that
10343 take variable numbers of arguments (including @code{printf});
10344 otherwise incorrect code will be generated for calls to those
10347 In addition, seriously incorrect code will result if you call a
10348 function with too many arguments. (Normally, extra arguments are
10349 harmlessly ignored.)
10351 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
10352 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
10355 @itemx -mno-align-int
10356 @opindex malign-int
10357 @opindex mno-align-int
10358 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
10359 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
10360 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
10361 Aligning variables on 32-bit boundaries produces code that runs somewhat
10362 faster on processors with 32-bit busses at the expense of more memory.
10364 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
10365 align structures containing the above types differently than
10366 most published application binary interface specifications for the m68k.
10370 Use the pc-relative addressing mode of the 68000 directly, instead of
10371 using a global offset table. At present, this option implies @option{-fpic},
10372 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
10373 not presently supported with @option{-mpcrel}, though this could be supported for
10374 68020 and higher processors.
10376 @item -mno-strict-align
10377 @itemx -mstrict-align
10378 @opindex mno-strict-align
10379 @opindex mstrict-align
10380 Do not (do) assume that unaligned memory references will be handled by
10384 Generate code that allows the data segment to be located in a different
10385 area of memory from the text segment. This allows for execute in place in
10386 an environment without virtual memory management. This option implies
10389 @item -mno-sep-data
10390 Generate code that assumes that the data segment follows the text segment.
10391 This is the default.
10393 @item -mid-shared-library
10394 Generate code that supports shared libraries via the library ID method.
10395 This allows for execute in place and shared libraries in an environment
10396 without virtual memory management. This option implies @option{-fPIC}.
10398 @item -mno-id-shared-library
10399 Generate code that doesn't assume ID based shared libraries are being used.
10400 This is the default.
10402 @item -mshared-library-id=n
10403 Specified the identification number of the ID based shared library being
10404 compiled. Specifying a value of 0 will generate more compact code, specifying
10405 other values will force the allocation of that number to the current
10406 library but is no more space or time efficient than omitting this option.
10410 @node M68hc1x Options
10411 @subsection M68hc1x Options
10412 @cindex M68hc1x options
10414 These are the @samp{-m} options defined for the 68hc11 and 68hc12
10415 microcontrollers. The default values for these options depends on
10416 which style of microcontroller was selected when the compiler was configured;
10417 the defaults for the most common choices are given below.
10424 Generate output for a 68HC11. This is the default
10425 when the compiler is configured for 68HC11-based systems.
10431 Generate output for a 68HC12. This is the default
10432 when the compiler is configured for 68HC12-based systems.
10438 Generate output for a 68HCS12.
10440 @item -mauto-incdec
10441 @opindex mauto-incdec
10442 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
10449 Enable the use of 68HC12 min and max instructions.
10452 @itemx -mno-long-calls
10453 @opindex mlong-calls
10454 @opindex mno-long-calls
10455 Treat all calls as being far away (near). If calls are assumed to be
10456 far away, the compiler will use the @code{call} instruction to
10457 call a function and the @code{rtc} instruction for returning.
10461 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10463 @item -msoft-reg-count=@var{count}
10464 @opindex msoft-reg-count
10465 Specify the number of pseudo-soft registers which are used for the
10466 code generation. The maximum number is 32. Using more pseudo-soft
10467 register may or may not result in better code depending on the program.
10468 The default is 4 for 68HC11 and 2 for 68HC12.
10472 @node MCore Options
10473 @subsection MCore Options
10474 @cindex MCore options
10476 These are the @samp{-m} options defined for the Motorola M*Core
10482 @itemx -mno-hardlit
10484 @opindex mno-hardlit
10485 Inline constants into the code stream if it can be done in two
10486 instructions or less.
10492 Use the divide instruction. (Enabled by default).
10494 @item -mrelax-immediate
10495 @itemx -mno-relax-immediate
10496 @opindex mrelax-immediate
10497 @opindex mno-relax-immediate
10498 Allow arbitrary sized immediates in bit operations.
10500 @item -mwide-bitfields
10501 @itemx -mno-wide-bitfields
10502 @opindex mwide-bitfields
10503 @opindex mno-wide-bitfields
10504 Always treat bit-fields as int-sized.
10506 @item -m4byte-functions
10507 @itemx -mno-4byte-functions
10508 @opindex m4byte-functions
10509 @opindex mno-4byte-functions
10510 Force all functions to be aligned to a four byte boundary.
10512 @item -mcallgraph-data
10513 @itemx -mno-callgraph-data
10514 @opindex mcallgraph-data
10515 @opindex mno-callgraph-data
10516 Emit callgraph information.
10519 @itemx -mno-slow-bytes
10520 @opindex mslow-bytes
10521 @opindex mno-slow-bytes
10522 Prefer word access when reading byte quantities.
10524 @item -mlittle-endian
10525 @itemx -mbig-endian
10526 @opindex mlittle-endian
10527 @opindex mbig-endian
10528 Generate code for a little endian target.
10534 Generate code for the 210 processor.
10538 @subsection MIPS Options
10539 @cindex MIPS options
10545 Generate big-endian code.
10549 Generate little-endian code. This is the default for @samp{mips*el-*-*}
10552 @item -march=@var{arch}
10554 Generate code that will run on @var{arch}, which can be the name of a
10555 generic MIPS ISA, or the name of a particular processor.
10557 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
10558 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
10559 The processor names are:
10560 @samp{4kc}, @samp{4km}, @samp{4kp},
10561 @samp{4kec}, @samp{4kem}, @samp{4kep},
10562 @samp{5kc}, @samp{5kf},
10564 @samp{24kc}, @samp{24kf}, @samp{24kx},
10565 @samp{24kec}, @samp{24kef}, @samp{24kex},
10566 @samp{34kc}, @samp{34kf}, @samp{34kx},
10569 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
10570 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
10571 @samp{rm7000}, @samp{rm9000},
10574 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
10575 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
10576 The special value @samp{from-abi} selects the
10577 most compatible architecture for the selected ABI (that is,
10578 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
10580 In processor names, a final @samp{000} can be abbreviated as @samp{k}
10581 (for example, @samp{-march=r2k}). Prefixes are optional, and
10582 @samp{vr} may be written @samp{r}.
10584 GCC defines two macros based on the value of this option. The first
10585 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
10586 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
10587 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
10588 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
10589 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
10591 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
10592 above. In other words, it will have the full prefix and will not
10593 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
10594 the macro names the resolved architecture (either @samp{"mips1"} or
10595 @samp{"mips3"}). It names the default architecture when no
10596 @option{-march} option is given.
10598 @item -mtune=@var{arch}
10600 Optimize for @var{arch}. Among other things, this option controls
10601 the way instructions are scheduled, and the perceived cost of arithmetic
10602 operations. The list of @var{arch} values is the same as for
10605 When this option is not used, GCC will optimize for the processor
10606 specified by @option{-march}. By using @option{-march} and
10607 @option{-mtune} together, it is possible to generate code that will
10608 run on a family of processors, but optimize the code for one
10609 particular member of that family.
10611 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
10612 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
10613 @samp{-march} ones described above.
10617 Equivalent to @samp{-march=mips1}.
10621 Equivalent to @samp{-march=mips2}.
10625 Equivalent to @samp{-march=mips3}.
10629 Equivalent to @samp{-march=mips4}.
10633 Equivalent to @samp{-march=mips32}.
10637 Equivalent to @samp{-march=mips32r2}.
10641 Equivalent to @samp{-march=mips64}.
10646 @opindex mno-mips16
10647 Generate (do not generate) MIPS16 code. If GCC is targetting a
10648 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
10660 Generate code for the given ABI@.
10662 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
10663 generates 64-bit code when you select a 64-bit architecture, but you
10664 can use @option{-mgp32} to get 32-bit code instead.
10666 For information about the O64 ABI, see
10667 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
10669 GCC supports a variant of the o32 ABI in which floating-point registers
10670 are 64 rather than 32 bits wide. You can select this combination with
10671 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
10672 and @samp{mfhc1} instructions and is therefore only supported for
10673 MIPS32R2 processors.
10675 The register assignments for arguments and return values remain the
10676 same, but each scalar value is passed in a single 64-bit register
10677 rather than a pair of 32-bit registers. For example, scalar
10678 floating-point values are returned in @samp{$f0} only, not a
10679 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
10680 remains the same, but all 64 bits are saved.
10683 @itemx -mno-abicalls
10685 @opindex mno-abicalls
10686 Generate (do not generate) code that is suitable for SVR4-style
10687 dynamic objects. @option{-mabicalls} is the default for SVR4-based
10692 Generate (do not generate) code that is fully position-independent,
10693 and that can therefore be linked into shared libraries. This option
10694 only affects @option{-mabicalls}.
10696 All @option{-mabicalls} code has traditionally been position-independent,
10697 regardless of options like @option{-fPIC} and @option{-fpic}. However,
10698 as an extension, the GNU toolchain allows executables to use absolute
10699 accesses for locally-binding symbols. It can also use shorter GP
10700 initialization sequences and generate direct calls to locally-defined
10701 functions. This mode is selected by @option{-mno-shared}.
10703 @option{-mno-shared} depends on binutils 2.16 or higher and generates
10704 objects that can only be linked by the GNU linker. However, the option
10705 does not affect the ABI of the final executable; it only affects the ABI
10706 of relocatable objects. Using @option{-mno-shared} will generally make
10707 executables both smaller and quicker.
10709 @option{-mshared} is the default.
10715 Lift (do not lift) the usual restrictions on the size of the global
10718 GCC normally uses a single instruction to load values from the GOT@.
10719 While this is relatively efficient, it will only work if the GOT
10720 is smaller than about 64k. Anything larger will cause the linker
10721 to report an error such as:
10723 @cindex relocation truncated to fit (MIPS)
10725 relocation truncated to fit: R_MIPS_GOT16 foobar
10728 If this happens, you should recompile your code with @option{-mxgot}.
10729 It should then work with very large GOTs, although it will also be
10730 less efficient, since it will take three instructions to fetch the
10731 value of a global symbol.
10733 Note that some linkers can create multiple GOTs. If you have such a
10734 linker, you should only need to use @option{-mxgot} when a single object
10735 file accesses more than 64k's worth of GOT entries. Very few do.
10737 These options have no effect unless GCC is generating position
10742 Assume that general-purpose registers are 32 bits wide.
10746 Assume that general-purpose registers are 64 bits wide.
10750 Assume that floating-point registers are 32 bits wide.
10754 Assume that floating-point registers are 64 bits wide.
10757 @opindex mhard-float
10758 Use floating-point coprocessor instructions.
10761 @opindex msoft-float
10762 Do not use floating-point coprocessor instructions. Implement
10763 floating-point calculations using library calls instead.
10765 @item -msingle-float
10766 @opindex msingle-float
10767 Assume that the floating-point coprocessor only supports single-precision
10770 @itemx -mdouble-float
10771 @opindex mdouble-float
10772 Assume that the floating-point coprocessor supports double-precision
10773 operations. This is the default.
10779 Use (do not use) the MIPS DSP ASE. @xref{MIPS DSP Built-in Functions}.
10781 @itemx -mpaired-single
10782 @itemx -mno-paired-single
10783 @opindex mpaired-single
10784 @opindex mno-paired-single
10785 Use (do not use) paired-single floating-point instructions.
10786 @xref{MIPS Paired-Single Support}. This option can only be used
10787 when generating 64-bit code and requires hardware floating-point
10788 support to be enabled.
10793 @opindex mno-mips3d
10794 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
10795 The option @option{-mips3d} implies @option{-mpaired-single}.
10799 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
10800 an explanation of the default and the way that the pointer size is
10805 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
10807 The default size of @code{int}s, @code{long}s and pointers depends on
10808 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
10809 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
10810 32-bit @code{long}s. Pointers are the same size as @code{long}s,
10811 or the same size as integer registers, whichever is smaller.
10817 Assume (do not assume) that all symbols have 32-bit values, regardless
10818 of the selected ABI@. This option is useful in combination with
10819 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
10820 to generate shorter and faster references to symbolic addresses.
10824 @cindex smaller data references (MIPS)
10825 @cindex gp-relative references (MIPS)
10826 Put global and static items less than or equal to @var{num} bytes into
10827 the small data or bss section instead of the normal data or bss section.
10828 This allows the data to be accessed using a single instruction.
10830 All modules should be compiled with the same @option{-G @var{num}}
10833 @item -membedded-data
10834 @itemx -mno-embedded-data
10835 @opindex membedded-data
10836 @opindex mno-embedded-data
10837 Allocate variables to the read-only data section first if possible, then
10838 next in the small data section if possible, otherwise in data. This gives
10839 slightly slower code than the default, but reduces the amount of RAM required
10840 when executing, and thus may be preferred for some embedded systems.
10842 @item -muninit-const-in-rodata
10843 @itemx -mno-uninit-const-in-rodata
10844 @opindex muninit-const-in-rodata
10845 @opindex mno-uninit-const-in-rodata
10846 Put uninitialized @code{const} variables in the read-only data section.
10847 This option is only meaningful in conjunction with @option{-membedded-data}.
10849 @item -msplit-addresses
10850 @itemx -mno-split-addresses
10851 @opindex msplit-addresses
10852 @opindex mno-split-addresses
10853 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
10854 relocation operators. This option has been superseded by
10855 @option{-mexplicit-relocs} but is retained for backwards compatibility.
10857 @item -mexplicit-relocs
10858 @itemx -mno-explicit-relocs
10859 @opindex mexplicit-relocs
10860 @opindex mno-explicit-relocs
10861 Use (do not use) assembler relocation operators when dealing with symbolic
10862 addresses. The alternative, selected by @option{-mno-explicit-relocs},
10863 is to use assembler macros instead.
10865 @option{-mexplicit-relocs} is the default if GCC was configured
10866 to use an assembler that supports relocation operators.
10868 @item -mcheck-zero-division
10869 @itemx -mno-check-zero-division
10870 @opindex mcheck-zero-division
10871 @opindex mno-check-zero-division
10872 Trap (do not trap) on integer division by zero. The default is
10873 @option{-mcheck-zero-division}.
10875 @item -mdivide-traps
10876 @itemx -mdivide-breaks
10877 @opindex mdivide-traps
10878 @opindex mdivide-breaks
10879 MIPS systems check for division by zero by generating either a
10880 conditional trap or a break instruction. Using traps results in
10881 smaller code, but is only supported on MIPS II and later. Also, some
10882 versions of the Linux kernel have a bug that prevents trap from
10883 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
10884 allow conditional traps on architectures that support them and
10885 @option{-mdivide-breaks} to force the use of breaks.
10887 The default is usually @option{-mdivide-traps}, but this can be
10888 overridden at configure time using @option{--with-divide=breaks}.
10889 Divide-by-zero checks can be completely disabled using
10890 @option{-mno-check-zero-division}.
10895 @opindex mno-memcpy
10896 Force (do not force) the use of @code{memcpy()} for non-trivial block
10897 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
10898 most constant-sized copies.
10901 @itemx -mno-long-calls
10902 @opindex mlong-calls
10903 @opindex mno-long-calls
10904 Disable (do not disable) use of the @code{jal} instruction. Calling
10905 functions using @code{jal} is more efficient but requires the caller
10906 and callee to be in the same 256 megabyte segment.
10908 This option has no effect on abicalls code. The default is
10909 @option{-mno-long-calls}.
10915 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
10916 instructions, as provided by the R4650 ISA@.
10919 @itemx -mno-fused-madd
10920 @opindex mfused-madd
10921 @opindex mno-fused-madd
10922 Enable (disable) use of the floating point multiply-accumulate
10923 instructions, when they are available. The default is
10924 @option{-mfused-madd}.
10926 When multiply-accumulate instructions are used, the intermediate
10927 product is calculated to infinite precision and is not subject to
10928 the FCSR Flush to Zero bit. This may be undesirable in some
10933 Tell the MIPS assembler to not run its preprocessor over user
10934 assembler files (with a @samp{.s} suffix) when assembling them.
10937 @itemx -mno-fix-r4000
10938 @opindex mfix-r4000
10939 @opindex mno-fix-r4000
10940 Work around certain R4000 CPU errata:
10943 A double-word or a variable shift may give an incorrect result if executed
10944 immediately after starting an integer division.
10946 A double-word or a variable shift may give an incorrect result if executed
10947 while an integer multiplication is in progress.
10949 An integer division may give an incorrect result if started in a delay slot
10950 of a taken branch or a jump.
10954 @itemx -mno-fix-r4400
10955 @opindex mfix-r4400
10956 @opindex mno-fix-r4400
10957 Work around certain R4400 CPU errata:
10960 A double-word or a variable shift may give an incorrect result if executed
10961 immediately after starting an integer division.
10965 @itemx -mno-fix-vr4120
10966 @opindex mfix-vr4120
10967 Work around certain VR4120 errata:
10970 @code{dmultu} does not always produce the correct result.
10972 @code{div} and @code{ddiv} do not always produce the correct result if one
10973 of the operands is negative.
10975 The workarounds for the division errata rely on special functions in
10976 @file{libgcc.a}. At present, these functions are only provided by
10977 the @code{mips64vr*-elf} configurations.
10979 Other VR4120 errata require a nop to be inserted between certain pairs of
10980 instructions. These errata are handled by the assembler, not by GCC itself.
10983 @opindex mfix-vr4130
10984 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
10985 workarounds are implemented by the assembler rather than by GCC,
10986 although GCC will avoid using @code{mflo} and @code{mfhi} if the
10987 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
10988 instructions are available instead.
10991 @itemx -mno-fix-sb1
10993 Work around certain SB-1 CPU core errata.
10994 (This flag currently works around the SB-1 revision 2
10995 ``F1'' and ``F2'' floating point errata.)
10997 @item -mflush-func=@var{func}
10998 @itemx -mno-flush-func
10999 @opindex mflush-func
11000 Specifies the function to call to flush the I and D caches, or to not
11001 call any such function. If called, the function must take the same
11002 arguments as the common @code{_flush_func()}, that is, the address of the
11003 memory range for which the cache is being flushed, the size of the
11004 memory range, and the number 3 (to flush both caches). The default
11005 depends on the target GCC was configured for, but commonly is either
11006 @samp{_flush_func} or @samp{__cpu_flush}.
11008 @item -mbranch-likely
11009 @itemx -mno-branch-likely
11010 @opindex mbranch-likely
11011 @opindex mno-branch-likely
11012 Enable or disable use of Branch Likely instructions, regardless of the
11013 default for the selected architecture. By default, Branch Likely
11014 instructions may be generated if they are supported by the selected
11015 architecture. An exception is for the MIPS32 and MIPS64 architectures
11016 and processors which implement those architectures; for those, Branch
11017 Likely instructions will not be generated by default because the MIPS32
11018 and MIPS64 architectures specifically deprecate their use.
11020 @item -mfp-exceptions
11021 @itemx -mno-fp-exceptions
11022 @opindex mfp-exceptions
11023 Specifies whether FP exceptions are enabled. This affects how we schedule
11024 FP instructions for some processors. The default is that FP exceptions are
11027 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
11028 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
11031 @item -mvr4130-align
11032 @itemx -mno-vr4130-align
11033 @opindex mvr4130-align
11034 The VR4130 pipeline is two-way superscalar, but can only issue two
11035 instructions together if the first one is 8-byte aligned. When this
11036 option is enabled, GCC will align pairs of instructions that it
11037 thinks should execute in parallel.
11039 This option only has an effect when optimizing for the VR4130.
11040 It normally makes code faster, but at the expense of making it bigger.
11041 It is enabled by default at optimization level @option{-O3}.
11045 @subsection MMIX Options
11046 @cindex MMIX Options
11048 These options are defined for the MMIX:
11052 @itemx -mno-libfuncs
11054 @opindex mno-libfuncs
11055 Specify that intrinsic library functions are being compiled, passing all
11056 values in registers, no matter the size.
11059 @itemx -mno-epsilon
11061 @opindex mno-epsilon
11062 Generate floating-point comparison instructions that compare with respect
11063 to the @code{rE} epsilon register.
11065 @item -mabi=mmixware
11067 @opindex mabi-mmixware
11069 Generate code that passes function parameters and return values that (in
11070 the called function) are seen as registers @code{$0} and up, as opposed to
11071 the GNU ABI which uses global registers @code{$231} and up.
11073 @item -mzero-extend
11074 @itemx -mno-zero-extend
11075 @opindex mzero-extend
11076 @opindex mno-zero-extend
11077 When reading data from memory in sizes shorter than 64 bits, use (do not
11078 use) zero-extending load instructions by default, rather than
11079 sign-extending ones.
11082 @itemx -mno-knuthdiv
11084 @opindex mno-knuthdiv
11085 Make the result of a division yielding a remainder have the same sign as
11086 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
11087 remainder follows the sign of the dividend. Both methods are
11088 arithmetically valid, the latter being almost exclusively used.
11090 @item -mtoplevel-symbols
11091 @itemx -mno-toplevel-symbols
11092 @opindex mtoplevel-symbols
11093 @opindex mno-toplevel-symbols
11094 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
11095 code can be used with the @code{PREFIX} assembly directive.
11099 Generate an executable in the ELF format, rather than the default
11100 @samp{mmo} format used by the @command{mmix} simulator.
11102 @item -mbranch-predict
11103 @itemx -mno-branch-predict
11104 @opindex mbranch-predict
11105 @opindex mno-branch-predict
11106 Use (do not use) the probable-branch instructions, when static branch
11107 prediction indicates a probable branch.
11109 @item -mbase-addresses
11110 @itemx -mno-base-addresses
11111 @opindex mbase-addresses
11112 @opindex mno-base-addresses
11113 Generate (do not generate) code that uses @emph{base addresses}. Using a
11114 base address automatically generates a request (handled by the assembler
11115 and the linker) for a constant to be set up in a global register. The
11116 register is used for one or more base address requests within the range 0
11117 to 255 from the value held in the register. The generally leads to short
11118 and fast code, but the number of different data items that can be
11119 addressed is limited. This means that a program that uses lots of static
11120 data may require @option{-mno-base-addresses}.
11122 @item -msingle-exit
11123 @itemx -mno-single-exit
11124 @opindex msingle-exit
11125 @opindex mno-single-exit
11126 Force (do not force) generated code to have a single exit point in each
11130 @node MN10300 Options
11131 @subsection MN10300 Options
11132 @cindex MN10300 options
11134 These @option{-m} options are defined for Matsushita MN10300 architectures:
11139 Generate code to avoid bugs in the multiply instructions for the MN10300
11140 processors. This is the default.
11142 @item -mno-mult-bug
11143 @opindex mno-mult-bug
11144 Do not generate code to avoid bugs in the multiply instructions for the
11145 MN10300 processors.
11149 Generate code which uses features specific to the AM33 processor.
11153 Do not generate code which uses features specific to the AM33 processor. This
11156 @item -mreturn-pointer-on-d0
11157 @opindex mreturn-pointer-on-d0
11158 When generating a function which returns a pointer, return the pointer
11159 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
11160 only in a0, and attempts to call such functions without a prototype
11161 would result in errors. Note that this option is on by default; use
11162 @option{-mno-return-pointer-on-d0} to disable it.
11166 Do not link in the C run-time initialization object file.
11170 Indicate to the linker that it should perform a relaxation optimization pass
11171 to shorten branches, calls and absolute memory addresses. This option only
11172 has an effect when used on the command line for the final link step.
11174 This option makes symbolic debugging impossible.
11178 @subsection MT Options
11181 These @option{-m} options are defined for Morpho MT architectures:
11185 @item -march=@var{cpu-type}
11187 Generate code that will run on @var{cpu-type}, which is the name of a system
11188 representing a certain processor type. Possible values for
11189 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
11190 @samp{ms1-16-003} and @samp{ms2}.
11192 When this option is not used, the default is @option{-march=ms1-16-002}.
11196 Use byte loads and stores when generating code.
11200 Do not use byte loads and stores when generating code.
11204 Use simulator runtime
11208 Do not link in the C run-time initialization object file
11209 @file{crti.o}. Other run-time initialization and termination files
11210 such as @file{startup.o} and @file{exit.o} are still included on the
11211 linker command line.
11215 @node PDP-11 Options
11216 @subsection PDP-11 Options
11217 @cindex PDP-11 Options
11219 These options are defined for the PDP-11:
11224 Use hardware FPP floating point. This is the default. (FIS floating
11225 point on the PDP-11/40 is not supported.)
11228 @opindex msoft-float
11229 Do not use hardware floating point.
11233 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
11237 Return floating-point results in memory. This is the default.
11241 Generate code for a PDP-11/40.
11245 Generate code for a PDP-11/45. This is the default.
11249 Generate code for a PDP-11/10.
11251 @item -mbcopy-builtin
11252 @opindex bcopy-builtin
11253 Use inline @code{movmemhi} patterns for copying memory. This is the
11258 Do not use inline @code{movmemhi} patterns for copying memory.
11264 Use 16-bit @code{int}. This is the default.
11270 Use 32-bit @code{int}.
11273 @itemx -mno-float32
11275 @opindex mno-float32
11276 Use 64-bit @code{float}. This is the default.
11279 @itemx -mno-float64
11281 @opindex mno-float64
11282 Use 32-bit @code{float}.
11286 Use @code{abshi2} pattern. This is the default.
11290 Do not use @code{abshi2} pattern.
11292 @item -mbranch-expensive
11293 @opindex mbranch-expensive
11294 Pretend that branches are expensive. This is for experimenting with
11295 code generation only.
11297 @item -mbranch-cheap
11298 @opindex mbranch-cheap
11299 Do not pretend that branches are expensive. This is the default.
11303 Generate code for a system with split I&D@.
11307 Generate code for a system without split I&D@. This is the default.
11311 Use Unix assembler syntax. This is the default when configured for
11312 @samp{pdp11-*-bsd}.
11316 Use DEC assembler syntax. This is the default when configured for any
11317 PDP-11 target other than @samp{pdp11-*-bsd}.
11320 @node PowerPC Options
11321 @subsection PowerPC Options
11322 @cindex PowerPC options
11324 These are listed under @xref{RS/6000 and PowerPC Options}.
11326 @node RS/6000 and PowerPC Options
11327 @subsection IBM RS/6000 and PowerPC Options
11328 @cindex RS/6000 and PowerPC Options
11329 @cindex IBM RS/6000 and PowerPC Options
11331 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
11338 @itemx -mno-powerpc
11339 @itemx -mpowerpc-gpopt
11340 @itemx -mno-powerpc-gpopt
11341 @itemx -mpowerpc-gfxopt
11342 @itemx -mno-powerpc-gfxopt
11344 @itemx -mno-powerpc64
11348 @itemx -mno-popcntb
11356 @opindex mno-power2
11358 @opindex mno-powerpc
11359 @opindex mpowerpc-gpopt
11360 @opindex mno-powerpc-gpopt
11361 @opindex mpowerpc-gfxopt
11362 @opindex mno-powerpc-gfxopt
11363 @opindex mpowerpc64
11364 @opindex mno-powerpc64
11368 @opindex mno-popcntb
11372 @opindex mno-mfpgpr
11373 GCC supports two related instruction set architectures for the
11374 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
11375 instructions supported by the @samp{rios} chip set used in the original
11376 RS/6000 systems and the @dfn{PowerPC} instruction set is the
11377 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
11378 the IBM 4xx, 6xx, and follow-on microprocessors.
11380 Neither architecture is a subset of the other. However there is a
11381 large common subset of instructions supported by both. An MQ
11382 register is included in processors supporting the POWER architecture.
11384 You use these options to specify which instructions are available on the
11385 processor you are using. The default value of these options is
11386 determined when configuring GCC@. Specifying the
11387 @option{-mcpu=@var{cpu_type}} overrides the specification of these
11388 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
11389 rather than the options listed above.
11391 The @option{-mpower} option allows GCC to generate instructions that
11392 are found only in the POWER architecture and to use the MQ register.
11393 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
11394 to generate instructions that are present in the POWER2 architecture but
11395 not the original POWER architecture.
11397 The @option{-mpowerpc} option allows GCC to generate instructions that
11398 are found only in the 32-bit subset of the PowerPC architecture.
11399 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
11400 GCC to use the optional PowerPC architecture instructions in the
11401 General Purpose group, including floating-point square root. Specifying
11402 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
11403 use the optional PowerPC architecture instructions in the Graphics
11404 group, including floating-point select.
11406 The @option{-mmfcrf} option allows GCC to generate the move from
11407 condition register field instruction implemented on the POWER4
11408 processor and other processors that support the PowerPC V2.01
11410 The @option{-mpopcntb} option allows GCC to generate the popcount and
11411 double precision FP reciprocal estimate instruction implemented on the
11412 POWER5 processor and other processors that support the PowerPC V2.02
11414 The @option{-mfprnd} option allows GCC to generate the FP round to
11415 integer instructions implemented on the POWER5+ processor and other
11416 processors that support the PowerPC V2.03 architecture.
11417 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
11418 general purpose register instructions implemented on the POWER6X
11419 processor and other processors that support the extended PowerPC V2.05
11422 The @option{-mpowerpc64} option allows GCC to generate the additional
11423 64-bit instructions that are found in the full PowerPC64 architecture
11424 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
11425 @option{-mno-powerpc64}.
11427 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
11428 will use only the instructions in the common subset of both
11429 architectures plus some special AIX common-mode calls, and will not use
11430 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
11431 permits GCC to use any instruction from either architecture and to
11432 allow use of the MQ register; specify this for the Motorola MPC601.
11434 @item -mnew-mnemonics
11435 @itemx -mold-mnemonics
11436 @opindex mnew-mnemonics
11437 @opindex mold-mnemonics
11438 Select which mnemonics to use in the generated assembler code. With
11439 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
11440 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
11441 assembler mnemonics defined for the POWER architecture. Instructions
11442 defined in only one architecture have only one mnemonic; GCC uses that
11443 mnemonic irrespective of which of these options is specified.
11445 GCC defaults to the mnemonics appropriate for the architecture in
11446 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
11447 value of these option. Unless you are building a cross-compiler, you
11448 should normally not specify either @option{-mnew-mnemonics} or
11449 @option{-mold-mnemonics}, but should instead accept the default.
11451 @item -mcpu=@var{cpu_type}
11453 Set architecture type, register usage, choice of mnemonics, and
11454 instruction scheduling parameters for machine type @var{cpu_type}.
11455 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
11456 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
11457 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
11458 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
11459 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
11460 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
11461 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
11462 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6},
11463 @samp{power6x}, @samp{common}, @samp{powerpc}, @samp{powerpc64},
11464 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
11466 @option{-mcpu=common} selects a completely generic processor. Code
11467 generated under this option will run on any POWER or PowerPC processor.
11468 GCC will use only the instructions in the common subset of both
11469 architectures, and will not use the MQ register. GCC assumes a generic
11470 processor model for scheduling purposes.
11472 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
11473 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
11474 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
11475 types, with an appropriate, generic processor model assumed for
11476 scheduling purposes.
11478 The other options specify a specific processor. Code generated under
11479 those options will run best on that processor, and may not run at all on
11482 The @option{-mcpu} options automatically enable or disable the
11483 following options: @option{-maltivec}, @option{-mfprnd},
11484 @option{-mhard-float}, @option{-mmfcrf}, @option{-mmultiple},
11485 @option{-mnew-mnemonics}, @option{-mpopcntb}, @option{-mpower},
11486 @option{-mpower2}, @option{-mpowerpc64}, @option{-mpowerpc-gpopt},
11487 @option{-mpowerpc-gfxopt}, @option{-mstring}, @option{-mmulhw},
11488 @option{-mdlmzb}, @option{-mmfpgpr}.
11489 The particular options set for any particular CPU will vary between
11490 compiler versions, depending on what setting seems to produce optimal
11491 code for that CPU; it doesn't necessarily reflect the actual hardware's
11492 capabilities. If you wish to set an individual option to a particular
11493 value, you may specify it after the @option{-mcpu} option, like
11494 @samp{-mcpu=970 -mno-altivec}.
11496 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
11497 not enabled or disabled by the @option{-mcpu} option at present because
11498 AIX does not have full support for these options. You may still
11499 enable or disable them individually if you're sure it'll work in your
11502 @item -mtune=@var{cpu_type}
11504 Set the instruction scheduling parameters for machine type
11505 @var{cpu_type}, but do not set the architecture type, register usage, or
11506 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
11507 values for @var{cpu_type} are used for @option{-mtune} as for
11508 @option{-mcpu}. If both are specified, the code generated will use the
11509 architecture, registers, and mnemonics set by @option{-mcpu}, but the
11510 scheduling parameters set by @option{-mtune}.
11516 Generate code to compute division as reciprocal estimate and iterative
11517 refinement, creating opportunities for increased throughput. This
11518 feature requires: optional PowerPC Graphics instruction set for single
11519 precision and FRE instruction for double precision, assuming divides
11520 cannot generate user-visible traps, and the domain values not include
11521 Infinities, denormals or zero denominator.
11524 @itemx -mno-altivec
11526 @opindex mno-altivec
11527 Generate code that uses (does not use) AltiVec instructions, and also
11528 enable the use of built-in functions that allow more direct access to
11529 the AltiVec instruction set. You may also need to set
11530 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
11536 @opindex mno-vrsave
11537 Generate VRSAVE instructions when generating AltiVec code.
11540 @opindex msecure-plt
11541 Generate code that allows ld and ld.so to build executables and shared
11542 libraries with non-exec .plt and .got sections. This is a PowerPC
11543 32-bit SYSV ABI option.
11547 Generate code that uses a BSS .plt section that ld.so fills in, and
11548 requires .plt and .got sections that are both writable and executable.
11549 This is a PowerPC 32-bit SYSV ABI option.
11555 This switch enables or disables the generation of ISEL instructions.
11557 @item -misel=@var{yes/no}
11558 This switch has been deprecated. Use @option{-misel} and
11559 @option{-mno-isel} instead.
11565 This switch enables or disables the generation of SPE simd
11568 @item -mspe=@var{yes/no}
11569 This option has been deprecated. Use @option{-mspe} and
11570 @option{-mno-spe} instead.
11572 @item -mfloat-gprs=@var{yes/single/double/no}
11573 @itemx -mfloat-gprs
11574 @opindex mfloat-gprs
11575 This switch enables or disables the generation of floating point
11576 operations on the general purpose registers for architectures that
11579 The argument @var{yes} or @var{single} enables the use of
11580 single-precision floating point operations.
11582 The argument @var{double} enables the use of single and
11583 double-precision floating point operations.
11585 The argument @var{no} disables floating point operations on the
11586 general purpose registers.
11588 This option is currently only available on the MPC854x.
11594 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
11595 targets (including GNU/Linux). The 32-bit environment sets int, long
11596 and pointer to 32 bits and generates code that runs on any PowerPC
11597 variant. The 64-bit environment sets int to 32 bits and long and
11598 pointer to 64 bits, and generates code for PowerPC64, as for
11599 @option{-mpowerpc64}.
11602 @itemx -mno-fp-in-toc
11603 @itemx -mno-sum-in-toc
11604 @itemx -mminimal-toc
11606 @opindex mno-fp-in-toc
11607 @opindex mno-sum-in-toc
11608 @opindex mminimal-toc
11609 Modify generation of the TOC (Table Of Contents), which is created for
11610 every executable file. The @option{-mfull-toc} option is selected by
11611 default. In that case, GCC will allocate at least one TOC entry for
11612 each unique non-automatic variable reference in your program. GCC
11613 will also place floating-point constants in the TOC@. However, only
11614 16,384 entries are available in the TOC@.
11616 If you receive a linker error message that saying you have overflowed
11617 the available TOC space, you can reduce the amount of TOC space used
11618 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
11619 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
11620 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
11621 generate code to calculate the sum of an address and a constant at
11622 run-time instead of putting that sum into the TOC@. You may specify one
11623 or both of these options. Each causes GCC to produce very slightly
11624 slower and larger code at the expense of conserving TOC space.
11626 If you still run out of space in the TOC even when you specify both of
11627 these options, specify @option{-mminimal-toc} instead. This option causes
11628 GCC to make only one TOC entry for every file. When you specify this
11629 option, GCC will produce code that is slower and larger but which
11630 uses extremely little TOC space. You may wish to use this option
11631 only on files that contain less frequently executed code.
11637 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
11638 @code{long} type, and the infrastructure needed to support them.
11639 Specifying @option{-maix64} implies @option{-mpowerpc64} and
11640 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
11641 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
11644 @itemx -mno-xl-compat
11645 @opindex mxl-compat
11646 @opindex mno-xl-compat
11647 Produce code that conforms more closely to IBM XL compiler semantics
11648 when using AIX-compatible ABI. Pass floating-point arguments to
11649 prototyped functions beyond the register save area (RSA) on the stack
11650 in addition to argument FPRs. Do not assume that most significant
11651 double in 128-bit long double value is properly rounded when comparing
11652 values and converting to double. Use XL symbol names for long double
11655 The AIX calling convention was extended but not initially documented to
11656 handle an obscure K&R C case of calling a function that takes the
11657 address of its arguments with fewer arguments than declared. IBM XL
11658 compilers access floating point arguments which do not fit in the
11659 RSA from the stack when a subroutine is compiled without
11660 optimization. Because always storing floating-point arguments on the
11661 stack is inefficient and rarely needed, this option is not enabled by
11662 default and only is necessary when calling subroutines compiled by IBM
11663 XL compilers without optimization.
11667 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
11668 application written to use message passing with special startup code to
11669 enable the application to run. The system must have PE installed in the
11670 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
11671 must be overridden with the @option{-specs=} option to specify the
11672 appropriate directory location. The Parallel Environment does not
11673 support threads, so the @option{-mpe} option and the @option{-pthread}
11674 option are incompatible.
11676 @item -malign-natural
11677 @itemx -malign-power
11678 @opindex malign-natural
11679 @opindex malign-power
11680 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
11681 @option{-malign-natural} overrides the ABI-defined alignment of larger
11682 types, such as floating-point doubles, on their natural size-based boundary.
11683 The option @option{-malign-power} instructs GCC to follow the ABI-specified
11684 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
11686 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
11690 @itemx -mhard-float
11691 @opindex msoft-float
11692 @opindex mhard-float
11693 Generate code that does not use (uses) the floating-point register set.
11694 Software floating point emulation is provided if you use the
11695 @option{-msoft-float} option, and pass the option to GCC when linking.
11698 @itemx -mno-multiple
11700 @opindex mno-multiple
11701 Generate code that uses (does not use) the load multiple word
11702 instructions and the store multiple word instructions. These
11703 instructions are generated by default on POWER systems, and not
11704 generated on PowerPC systems. Do not use @option{-mmultiple} on little
11705 endian PowerPC systems, since those instructions do not work when the
11706 processor is in little endian mode. The exceptions are PPC740 and
11707 PPC750 which permit the instructions usage in little endian mode.
11712 @opindex mno-string
11713 Generate code that uses (does not use) the load string instructions
11714 and the store string word instructions to save multiple registers and
11715 do small block moves. These instructions are generated by default on
11716 POWER systems, and not generated on PowerPC systems. Do not use
11717 @option{-mstring} on little endian PowerPC systems, since those
11718 instructions do not work when the processor is in little endian mode.
11719 The exceptions are PPC740 and PPC750 which permit the instructions
11720 usage in little endian mode.
11725 @opindex mno-update
11726 Generate code that uses (does not use) the load or store instructions
11727 that update the base register to the address of the calculated memory
11728 location. These instructions are generated by default. If you use
11729 @option{-mno-update}, there is a small window between the time that the
11730 stack pointer is updated and the address of the previous frame is
11731 stored, which means code that walks the stack frame across interrupts or
11732 signals may get corrupted data.
11735 @itemx -mno-fused-madd
11736 @opindex mfused-madd
11737 @opindex mno-fused-madd
11738 Generate code that uses (does not use) the floating point multiply and
11739 accumulate instructions. These instructions are generated by default if
11740 hardware floating is used.
11746 Generate code that uses (does not use) the half-word multiply and
11747 multiply-accumulate instructions on the IBM 405 and 440 processors.
11748 These instructions are generated by default when targetting those
11755 Generate code that uses (does not use) the string-search @samp{dlmzb}
11756 instruction on the IBM 405 and 440 processors. This instruction is
11757 generated by default when targetting those processors.
11759 @item -mno-bit-align
11761 @opindex mno-bit-align
11762 @opindex mbit-align
11763 On System V.4 and embedded PowerPC systems do not (do) force structures
11764 and unions that contain bit-fields to be aligned to the base type of the
11767 For example, by default a structure containing nothing but 8
11768 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
11769 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
11770 the structure would be aligned to a 1 byte boundary and be one byte in
11773 @item -mno-strict-align
11774 @itemx -mstrict-align
11775 @opindex mno-strict-align
11776 @opindex mstrict-align
11777 On System V.4 and embedded PowerPC systems do not (do) assume that
11778 unaligned memory references will be handled by the system.
11780 @item -mrelocatable
11781 @itemx -mno-relocatable
11782 @opindex mrelocatable
11783 @opindex mno-relocatable
11784 On embedded PowerPC systems generate code that allows (does not allow)
11785 the program to be relocated to a different address at runtime. If you
11786 use @option{-mrelocatable} on any module, all objects linked together must
11787 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
11789 @item -mrelocatable-lib
11790 @itemx -mno-relocatable-lib
11791 @opindex mrelocatable-lib
11792 @opindex mno-relocatable-lib
11793 On embedded PowerPC systems generate code that allows (does not allow)
11794 the program to be relocated to a different address at runtime. Modules
11795 compiled with @option{-mrelocatable-lib} can be linked with either modules
11796 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
11797 with modules compiled with the @option{-mrelocatable} options.
11803 On System V.4 and embedded PowerPC systems do not (do) assume that
11804 register 2 contains a pointer to a global area pointing to the addresses
11805 used in the program.
11808 @itemx -mlittle-endian
11810 @opindex mlittle-endian
11811 On System V.4 and embedded PowerPC systems compile code for the
11812 processor in little endian mode. The @option{-mlittle-endian} option is
11813 the same as @option{-mlittle}.
11816 @itemx -mbig-endian
11818 @opindex mbig-endian
11819 On System V.4 and embedded PowerPC systems compile code for the
11820 processor in big endian mode. The @option{-mbig-endian} option is
11821 the same as @option{-mbig}.
11823 @item -mdynamic-no-pic
11824 @opindex mdynamic-no-pic
11825 On Darwin and Mac OS X systems, compile code so that it is not
11826 relocatable, but that its external references are relocatable. The
11827 resulting code is suitable for applications, but not shared
11830 @item -mprioritize-restricted-insns=@var{priority}
11831 @opindex mprioritize-restricted-insns
11832 This option controls the priority that is assigned to
11833 dispatch-slot restricted instructions during the second scheduling
11834 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
11835 @var{no/highest/second-highest} priority to dispatch slot restricted
11838 @item -msched-costly-dep=@var{dependence_type}
11839 @opindex msched-costly-dep
11840 This option controls which dependences are considered costly
11841 by the target during instruction scheduling. The argument
11842 @var{dependence_type} takes one of the following values:
11843 @var{no}: no dependence is costly,
11844 @var{all}: all dependences are costly,
11845 @var{true_store_to_load}: a true dependence from store to load is costly,
11846 @var{store_to_load}: any dependence from store to load is costly,
11847 @var{number}: any dependence which latency >= @var{number} is costly.
11849 @item -minsert-sched-nops=@var{scheme}
11850 @opindex minsert-sched-nops
11851 This option controls which nop insertion scheme will be used during
11852 the second scheduling pass. The argument @var{scheme} takes one of the
11854 @var{no}: Don't insert nops.
11855 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
11856 according to the scheduler's grouping.
11857 @var{regroup_exact}: Insert nops to force costly dependent insns into
11858 separate groups. Insert exactly as many nops as needed to force an insn
11859 to a new group, according to the estimated processor grouping.
11860 @var{number}: Insert nops to force costly dependent insns into
11861 separate groups. Insert @var{number} nops to force an insn to a new group.
11864 @opindex mcall-sysv
11865 On System V.4 and embedded PowerPC systems compile code using calling
11866 conventions that adheres to the March 1995 draft of the System V
11867 Application Binary Interface, PowerPC processor supplement. This is the
11868 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
11870 @item -mcall-sysv-eabi
11871 @opindex mcall-sysv-eabi
11872 Specify both @option{-mcall-sysv} and @option{-meabi} options.
11874 @item -mcall-sysv-noeabi
11875 @opindex mcall-sysv-noeabi
11876 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
11878 @item -mcall-solaris
11879 @opindex mcall-solaris
11880 On System V.4 and embedded PowerPC systems compile code for the Solaris
11884 @opindex mcall-linux
11885 On System V.4 and embedded PowerPC systems compile code for the
11886 Linux-based GNU system.
11890 On System V.4 and embedded PowerPC systems compile code for the
11891 Hurd-based GNU system.
11893 @item -mcall-netbsd
11894 @opindex mcall-netbsd
11895 On System V.4 and embedded PowerPC systems compile code for the
11896 NetBSD operating system.
11898 @item -maix-struct-return
11899 @opindex maix-struct-return
11900 Return all structures in memory (as specified by the AIX ABI)@.
11902 @item -msvr4-struct-return
11903 @opindex msvr4-struct-return
11904 Return structures smaller than 8 bytes in registers (as specified by the
11907 @item -mabi=@var{abi-type}
11909 Extend the current ABI with a particular extension, or remove such extension.
11910 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
11911 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
11915 Extend the current ABI with SPE ABI extensions. This does not change
11916 the default ABI, instead it adds the SPE ABI extensions to the current
11920 @opindex mabi=no-spe
11921 Disable Booke SPE ABI extensions for the current ABI@.
11923 @item -mabi=ibmlongdouble
11924 @opindex mabi=ibmlongdouble
11925 Change the current ABI to use IBM extended precision long double.
11926 This is a PowerPC 32-bit SYSV ABI option.
11928 @item -mabi=ieeelongdouble
11929 @opindex mabi=ieeelongdouble
11930 Change the current ABI to use IEEE extended precision long double.
11931 This is a PowerPC 32-bit Linux ABI option.
11934 @itemx -mno-prototype
11935 @opindex mprototype
11936 @opindex mno-prototype
11937 On System V.4 and embedded PowerPC systems assume that all calls to
11938 variable argument functions are properly prototyped. Otherwise, the
11939 compiler must insert an instruction before every non prototyped call to
11940 set or clear bit 6 of the condition code register (@var{CR}) to
11941 indicate whether floating point values were passed in the floating point
11942 registers in case the function takes a variable arguments. With
11943 @option{-mprototype}, only calls to prototyped variable argument functions
11944 will set or clear the bit.
11948 On embedded PowerPC systems, assume that the startup module is called
11949 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
11950 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
11955 On embedded PowerPC systems, assume that the startup module is called
11956 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
11961 On embedded PowerPC systems, assume that the startup module is called
11962 @file{crt0.o} and the standard C libraries are @file{libads.a} and
11965 @item -myellowknife
11966 @opindex myellowknife
11967 On embedded PowerPC systems, assume that the startup module is called
11968 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
11973 On System V.4 and embedded PowerPC systems, specify that you are
11974 compiling for a VxWorks system.
11978 Specify that you are compiling for the WindISS simulation environment.
11982 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
11983 header to indicate that @samp{eabi} extended relocations are used.
11989 On System V.4 and embedded PowerPC systems do (do not) adhere to the
11990 Embedded Applications Binary Interface (eabi) which is a set of
11991 modifications to the System V.4 specifications. Selecting @option{-meabi}
11992 means that the stack is aligned to an 8 byte boundary, a function
11993 @code{__eabi} is called to from @code{main} to set up the eabi
11994 environment, and the @option{-msdata} option can use both @code{r2} and
11995 @code{r13} to point to two separate small data areas. Selecting
11996 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
11997 do not call an initialization function from @code{main}, and the
11998 @option{-msdata} option will only use @code{r13} to point to a single
11999 small data area. The @option{-meabi} option is on by default if you
12000 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
12003 @opindex msdata=eabi
12004 On System V.4 and embedded PowerPC systems, put small initialized
12005 @code{const} global and static data in the @samp{.sdata2} section, which
12006 is pointed to by register @code{r2}. Put small initialized
12007 non-@code{const} global and static data in the @samp{.sdata} section,
12008 which is pointed to by register @code{r13}. Put small uninitialized
12009 global and static data in the @samp{.sbss} section, which is adjacent to
12010 the @samp{.sdata} section. The @option{-msdata=eabi} option is
12011 incompatible with the @option{-mrelocatable} option. The
12012 @option{-msdata=eabi} option also sets the @option{-memb} option.
12015 @opindex msdata=sysv
12016 On System V.4 and embedded PowerPC systems, put small global and static
12017 data in the @samp{.sdata} section, which is pointed to by register
12018 @code{r13}. Put small uninitialized global and static data in the
12019 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
12020 The @option{-msdata=sysv} option is incompatible with the
12021 @option{-mrelocatable} option.
12023 @item -msdata=default
12025 @opindex msdata=default
12027 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
12028 compile code the same as @option{-msdata=eabi}, otherwise compile code the
12029 same as @option{-msdata=sysv}.
12032 @opindex msdata-data
12033 On System V.4 and embedded PowerPC systems, put small global
12034 data in the @samp{.sdata} section. Put small uninitialized global
12035 data in the @samp{.sbss} section. Do not use register @code{r13}
12036 to address small data however. This is the default behavior unless
12037 other @option{-msdata} options are used.
12041 @opindex msdata=none
12043 On embedded PowerPC systems, put all initialized global and static data
12044 in the @samp{.data} section, and all uninitialized data in the
12045 @samp{.bss} section.
12049 @cindex smaller data references (PowerPC)
12050 @cindex .sdata/.sdata2 references (PowerPC)
12051 On embedded PowerPC systems, put global and static items less than or
12052 equal to @var{num} bytes into the small data or bss sections instead of
12053 the normal data or bss section. By default, @var{num} is 8. The
12054 @option{-G @var{num}} switch is also passed to the linker.
12055 All modules should be compiled with the same @option{-G @var{num}} value.
12058 @itemx -mno-regnames
12060 @opindex mno-regnames
12061 On System V.4 and embedded PowerPC systems do (do not) emit register
12062 names in the assembly language output using symbolic forms.
12065 @itemx -mno-longcall
12067 @opindex mno-longcall
12068 By default assume that all calls are far away so that a longer more
12069 expensive calling sequence is required. This is required for calls
12070 further than 32 megabytes (33,554,432 bytes) from the current location.
12071 A short call will be generated if the compiler knows
12072 the call cannot be that far away. This setting can be overridden by
12073 the @code{shortcall} function attribute, or by @code{#pragma
12076 Some linkers are capable of detecting out-of-range calls and generating
12077 glue code on the fly. On these systems, long calls are unnecessary and
12078 generate slower code. As of this writing, the AIX linker can do this,
12079 as can the GNU linker for PowerPC/64. It is planned to add this feature
12080 to the GNU linker for 32-bit PowerPC systems as well.
12082 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
12083 callee, L42'', plus a ``branch island'' (glue code). The two target
12084 addresses represent the callee and the ``branch island''. The
12085 Darwin/PPC linker will prefer the first address and generate a ``bl
12086 callee'' if the PPC ``bl'' instruction will reach the callee directly;
12087 otherwise, the linker will generate ``bl L42'' to call the ``branch
12088 island''. The ``branch island'' is appended to the body of the
12089 calling function; it computes the full 32-bit address of the callee
12092 On Mach-O (Darwin) systems, this option directs the compiler emit to
12093 the glue for every direct call, and the Darwin linker decides whether
12094 to use or discard it.
12096 In the future, we may cause GCC to ignore all longcall specifications
12097 when the linker is known to generate glue.
12101 Adds support for multithreading with the @dfn{pthreads} library.
12102 This option sets flags for both the preprocessor and linker.
12106 @node S/390 and zSeries Options
12107 @subsection S/390 and zSeries Options
12108 @cindex S/390 and zSeries Options
12110 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
12114 @itemx -msoft-float
12115 @opindex mhard-float
12116 @opindex msoft-float
12117 Use (do not use) the hardware floating-point instructions and registers
12118 for floating-point operations. When @option{-msoft-float} is specified,
12119 functions in @file{libgcc.a} will be used to perform floating-point
12120 operations. When @option{-mhard-float} is specified, the compiler
12121 generates IEEE floating-point instructions. This is the default.
12123 @item -mlong-double-64
12124 @itemx -mlong-double-128
12125 @opindex mlong-double-64
12126 @opindex mlong-double-128
12127 These switches control the size of @code{long double} type. A size
12128 of 64bit makes the @code{long double} type equivalent to the @code{double}
12129 type. This is the default.
12132 @itemx -mno-backchain
12133 @opindex mbackchain
12134 @opindex mno-backchain
12135 Store (do not store) the address of the caller's frame as backchain pointer
12136 into the callee's stack frame.
12137 A backchain may be needed to allow debugging using tools that do not understand
12138 DWARF-2 call frame information.
12139 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
12140 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
12141 the backchain is placed into the topmost word of the 96/160 byte register
12144 In general, code compiled with @option{-mbackchain} is call-compatible with
12145 code compiled with @option{-mmo-backchain}; however, use of the backchain
12146 for debugging purposes usually requires that the whole binary is built with
12147 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
12148 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
12149 to build a linux kernel use @option{-msoft-float}.
12151 The default is to not maintain the backchain.
12153 @item -mpacked-stack
12154 @item -mno-packed-stack
12155 @opindex mpacked-stack
12156 @opindex mno-packed-stack
12157 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
12158 specified, the compiler uses the all fields of the 96/160 byte register save
12159 area only for their default purpose; unused fields still take up stack space.
12160 When @option{-mpacked-stack} is specified, register save slots are densely
12161 packed at the top of the register save area; unused space is reused for other
12162 purposes, allowing for more efficient use of the available stack space.
12163 However, when @option{-mbackchain} is also in effect, the topmost word of
12164 the save area is always used to store the backchain, and the return address
12165 register is always saved two words below the backchain.
12167 As long as the stack frame backchain is not used, code generated with
12168 @option{-mpacked-stack} is call-compatible with code generated with
12169 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
12170 S/390 or zSeries generated code that uses the stack frame backchain at run
12171 time, not just for debugging purposes. Such code is not call-compatible
12172 with code compiled with @option{-mpacked-stack}. Also, note that the
12173 combination of @option{-mbackchain},
12174 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
12175 to build a linux kernel use @option{-msoft-float}.
12177 The default is to not use the packed stack layout.
12180 @itemx -mno-small-exec
12181 @opindex msmall-exec
12182 @opindex mno-small-exec
12183 Generate (or do not generate) code using the @code{bras} instruction
12184 to do subroutine calls.
12185 This only works reliably if the total executable size does not
12186 exceed 64k. The default is to use the @code{basr} instruction instead,
12187 which does not have this limitation.
12193 When @option{-m31} is specified, generate code compliant to the
12194 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
12195 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
12196 particular to generate 64-bit instructions. For the @samp{s390}
12197 targets, the default is @option{-m31}, while the @samp{s390x}
12198 targets default to @option{-m64}.
12204 When @option{-mzarch} is specified, generate code using the
12205 instructions available on z/Architecture.
12206 When @option{-mesa} is specified, generate code using the
12207 instructions available on ESA/390. Note that @option{-mesa} is
12208 not possible with @option{-m64}.
12209 When generating code compliant to the GNU/Linux for S/390 ABI,
12210 the default is @option{-mesa}. When generating code compliant
12211 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
12217 Generate (or do not generate) code using the @code{mvcle} instruction
12218 to perform block moves. When @option{-mno-mvcle} is specified,
12219 use a @code{mvc} loop instead. This is the default unless optimizing for
12226 Print (or do not print) additional debug information when compiling.
12227 The default is to not print debug information.
12229 @item -march=@var{cpu-type}
12231 Generate code that will run on @var{cpu-type}, which is the name of a system
12232 representing a certain processor type. Possible values for
12233 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
12234 When generating code using the instructions available on z/Architecture,
12235 the default is @option{-march=z900}. Otherwise, the default is
12236 @option{-march=g5}.
12238 @item -mtune=@var{cpu-type}
12240 Tune to @var{cpu-type} everything applicable about the generated code,
12241 except for the ABI and the set of available instructions.
12242 The list of @var{cpu-type} values is the same as for @option{-march}.
12243 The default is the value used for @option{-march}.
12246 @itemx -mno-tpf-trace
12247 @opindex mtpf-trace
12248 @opindex mno-tpf-trace
12249 Generate code that adds (does not add) in TPF OS specific branches to trace
12250 routines in the operating system. This option is off by default, even
12251 when compiling for the TPF OS@.
12254 @itemx -mno-fused-madd
12255 @opindex mfused-madd
12256 @opindex mno-fused-madd
12257 Generate code that uses (does not use) the floating point multiply and
12258 accumulate instructions. These instructions are generated by default if
12259 hardware floating point is used.
12261 @item -mwarn-framesize=@var{framesize}
12262 @opindex mwarn-framesize
12263 Emit a warning if the current function exceeds the given frame size. Because
12264 this is a compile time check it doesn't need to be a real problem when the program
12265 runs. It is intended to identify functions which most probably cause
12266 a stack overflow. It is useful to be used in an environment with limited stack
12267 size e.g.@: the linux kernel.
12269 @item -mwarn-dynamicstack
12270 @opindex mwarn-dynamicstack
12271 Emit a warning if the function calls alloca or uses dynamically
12272 sized arrays. This is generally a bad idea with a limited stack size.
12274 @item -mstack-guard=@var{stack-guard}
12275 @item -mstack-size=@var{stack-size}
12276 @opindex mstack-guard
12277 @opindex mstack-size
12278 These arguments always have to be used in conjunction. If they are present the s390
12279 back end emits additional instructions in the function prologue which trigger a trap
12280 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
12281 (remember that the stack on s390 grows downward). These options are intended to
12282 be used to help debugging stack overflow problems. The additionally emitted code
12283 causes only little overhead and hence can also be used in production like systems
12284 without greater performance degradation. The given values have to be exact
12285 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard} without
12287 In order to be efficient the extra code makes the assumption that the stack starts
12288 at an address aligned to the value given by @var{stack-size}.
12291 @node Score Options
12292 @subsection Score Options
12293 @cindex Score Options
12295 These options are defined for Score implementations:
12300 Compile code for little endian mode.
12304 Compile code for big endian mode. This is the default.
12308 Enable the use of multiply-accumulate instructions. Disabled by default.
12312 Specify the SCORE5U of the target architecture.
12316 Specify the SCORE7 of the target architecture. This is the default.
12320 @subsection SH Options
12322 These @samp{-m} options are defined for the SH implementations:
12327 Generate code for the SH1.
12331 Generate code for the SH2.
12334 Generate code for the SH2e.
12338 Generate code for the SH3.
12342 Generate code for the SH3e.
12346 Generate code for the SH4 without a floating-point unit.
12348 @item -m4-single-only
12349 @opindex m4-single-only
12350 Generate code for the SH4 with a floating-point unit that only
12351 supports single-precision arithmetic.
12355 Generate code for the SH4 assuming the floating-point unit is in
12356 single-precision mode by default.
12360 Generate code for the SH4.
12364 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
12365 floating-point unit is not used.
12367 @item -m4a-single-only
12368 @opindex m4a-single-only
12369 Generate code for the SH4a, in such a way that no double-precision
12370 floating point operations are used.
12373 @opindex m4a-single
12374 Generate code for the SH4a assuming the floating-point unit is in
12375 single-precision mode by default.
12379 Generate code for the SH4a.
12383 Same as @option{-m4a-nofpu}, except that it implicitly passes
12384 @option{-dsp} to the assembler. GCC doesn't generate any DSP
12385 instructions at the moment.
12389 Compile code for the processor in big endian mode.
12393 Compile code for the processor in little endian mode.
12397 Align doubles at 64-bit boundaries. Note that this changes the calling
12398 conventions, and thus some functions from the standard C library will
12399 not work unless you recompile it first with @option{-mdalign}.
12403 Shorten some address references at link time, when possible; uses the
12404 linker option @option{-relax}.
12408 Use 32-bit offsets in @code{switch} tables. The default is to use
12413 Enable the use of the instruction @code{fmovd}.
12417 Comply with the calling conventions defined by Renesas.
12421 Comply with the calling conventions defined by Renesas.
12425 Comply with the calling conventions defined for GCC before the Renesas
12426 conventions were available. This option is the default for all
12427 targets of the SH toolchain except for @samp{sh-symbianelf}.
12430 @opindex mnomacsave
12431 Mark the @code{MAC} register as call-clobbered, even if
12432 @option{-mhitachi} is given.
12436 Increase IEEE-compliance of floating-point code.
12437 At the moment, this is equivalent to @option{-fno-finite-math-only}.
12438 When generating 16 bit SH opcodes, getting IEEE-conforming results for
12439 comparisons of NANs / infinities incurs extra overhead in every
12440 floating point comparison, therefore the default is set to
12441 @option{-ffinite-math-only}.
12443 @item -minline-ic_invalidate
12444 @opindex minline-ic_invalidate
12445 Inline code to invalidate instruction cache entries after setting up
12446 nested function trampolines.
12447 This option has no effect if -musermode is in effect and the selected
12448 code generation option (e.g. -m4) does not allow the use of the icbi
12450 If the selected code generation option does not allow the use of the icbi
12451 instruction, and -musermode is not in effect, the inlined code will
12452 manipulate the instruction cache address array directly with an associative
12453 write. This not only requires privileged mode, but it will also
12454 fail if the cache line had been mapped via the TLB and has become unmapped.
12458 Dump instruction size and location in the assembly code.
12461 @opindex mpadstruct
12462 This option is deprecated. It pads structures to multiple of 4 bytes,
12463 which is incompatible with the SH ABI@.
12467 Optimize for space instead of speed. Implied by @option{-Os}.
12470 @opindex mprefergot
12471 When generating position-independent code, emit function calls using
12472 the Global Offset Table instead of the Procedure Linkage Table.
12476 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
12477 if the inlined code would not work in user mode.
12478 This is the default when the target is @code{sh-*-linux*}.
12480 @item -multcost=@var{number}
12481 @opindex multcost=@var{number}
12482 Set the cost to assume for a multiply insn.
12484 @item -mdiv=@var{strategy}
12485 @opindex mdiv=@var{strategy}
12486 Set the division strategy to use for SHmedia code. @var{strategy} must be
12487 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
12488 inv:call2, inv:fp .
12489 "fp" performs the operation in floating point. This has a very high latency,
12490 but needs only a few instructions, so it might be a good choice if
12491 your code has enough easily exploitable ILP to allow the compiler to
12492 schedule the floating point instructions together with other instructions.
12493 Division by zero causes a floating point exception.
12494 "inv" uses integer operations to calculate the inverse of the divisor,
12495 and then multiplies the dividend with the inverse. This strategy allows
12496 cse and hoisting of the inverse calculation. Division by zero calculates
12497 an unspecified result, but does not trap.
12498 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
12499 have been found, or if the entire operation has been hoisted to the same
12500 place, the last stages of the inverse calculation are intertwined with the
12501 final multiply to reduce the overall latency, at the expense of using a few
12502 more instructions, and thus offering fewer scheduling opportunities with
12504 "call" calls a library function that usually implements the inv:minlat
12506 This gives high code density for m5-*media-nofpu compilations.
12507 "call2" uses a different entry point of the same library function, where it
12508 assumes that a pointer to a lookup table has already been set up, which
12509 exposes the pointer load to cse / code hoisting optimizations.
12510 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
12511 code generation, but if the code stays unoptimized, revert to the "call",
12512 "call2", or "fp" strategies, respectively. Note that the
12513 potentially-trapping side effect of division by zero is carried by a
12514 separate instruction, so it is possible that all the integer instructions
12515 are hoisted out, but the marker for the side effect stays where it is.
12516 A recombination to fp operations or a call is not possible in that case.
12517 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
12518 that the inverse calculation was nor separated from the multiply, they speed
12519 up division where the dividend fits into 20 bits (plus sign where applicable),
12520 by inserting a test to skip a number of operations in this case; this test
12521 slows down the case of larger dividends. inv20u assumes the case of a such
12522 a small dividend to be unlikely, and inv20l assumes it to be likely.
12524 @item -mdivsi3_libfunc=@var{name}
12525 @opindex mdivsi3_libfunc=@var{name}
12526 Set the name of the library function used for 32 bit signed division to
12527 @var{name}. This only affect the name used in the call and inv:call
12528 division strategies, and the compiler will still expect the same
12529 sets of input/output/clobbered registers as if this option was not present.
12531 @item -madjust-unroll
12532 @opindex madjust-unroll
12533 Throttle unrolling to avoid thrashing target registers.
12534 This option only has an effect if the gcc code base supports the
12535 TARGET_ADJUST_UNROLL_MAX target hook.
12537 @item -mindexed-addressing
12538 @opindex mindexed-addressing
12539 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
12540 This is only safe if the hardware and/or OS implement 32 bit wrap-around
12541 semantics for the indexed addressing mode. The architecture allows the
12542 implementation of processors with 64 bit MMU, which the OS could use to
12543 get 32 bit addressing, but since no current hardware implementation supports
12544 this or any other way to make the indexed addressing mode safe to use in
12545 the 32 bit ABI, the default is -mno-indexed-addressing.
12547 @item -mgettrcost=@var{number}
12548 @opindex mgettrcost=@var{number}
12549 Set the cost assumed for the gettr instruction to @var{number}.
12550 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
12554 Assume pt* instructions won't trap. This will generally generate better
12555 scheduled code, but is unsafe on current hardware. The current architecture
12556 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
12557 This has the unintentional effect of making it unsafe to schedule ptabs /
12558 ptrel before a branch, or hoist it out of a loop. For example,
12559 __do_global_ctors, a part of libgcc that runs constructors at program
12560 startup, calls functions in a list which is delimited by -1. With the
12561 -mpt-fixed option, the ptabs will be done before testing against -1.
12562 That means that all the constructors will be run a bit quicker, but when
12563 the loop comes to the end of the list, the program crashes because ptabs
12564 loads -1 into a target register. Since this option is unsafe for any
12565 hardware implementing the current architecture specification, the default
12566 is -mno-pt-fixed. Unless the user specifies a specific cost with
12567 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
12568 this deters register allocation using target registers for storing
12571 @item -minvalid-symbols
12572 @opindex minvalid-symbols
12573 Assume symbols might be invalid. Ordinary function symbols generated by
12574 the compiler will always be valid to load with movi/shori/ptabs or
12575 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
12576 to generate symbols that will cause ptabs / ptrel to trap.
12577 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
12578 It will then prevent cross-basic-block cse, hoisting and most scheduling
12579 of symbol loads. The default is @option{-mno-invalid-symbols}.
12582 @node SPARC Options
12583 @subsection SPARC Options
12584 @cindex SPARC options
12586 These @samp{-m} options are supported on the SPARC:
12589 @item -mno-app-regs
12591 @opindex mno-app-regs
12593 Specify @option{-mapp-regs} to generate output using the global registers
12594 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
12597 To be fully SVR4 ABI compliant at the cost of some performance loss,
12598 specify @option{-mno-app-regs}. You should compile libraries and system
12599 software with this option.
12602 @itemx -mhard-float
12604 @opindex mhard-float
12605 Generate output containing floating point instructions. This is the
12609 @itemx -msoft-float
12611 @opindex msoft-float
12612 Generate output containing library calls for floating point.
12613 @strong{Warning:} the requisite libraries are not available for all SPARC
12614 targets. Normally the facilities of the machine's usual C compiler are
12615 used, but this cannot be done directly in cross-compilation. You must make
12616 your own arrangements to provide suitable library functions for
12617 cross-compilation. The embedded targets @samp{sparc-*-aout} and
12618 @samp{sparclite-*-*} do provide software floating point support.
12620 @option{-msoft-float} changes the calling convention in the output file;
12621 therefore, it is only useful if you compile @emph{all} of a program with
12622 this option. In particular, you need to compile @file{libgcc.a}, the
12623 library that comes with GCC, with @option{-msoft-float} in order for
12626 @item -mhard-quad-float
12627 @opindex mhard-quad-float
12628 Generate output containing quad-word (long double) floating point
12631 @item -msoft-quad-float
12632 @opindex msoft-quad-float
12633 Generate output containing library calls for quad-word (long double)
12634 floating point instructions. The functions called are those specified
12635 in the SPARC ABI@. This is the default.
12637 As of this writing, there are no SPARC implementations that have hardware
12638 support for the quad-word floating point instructions. They all invoke
12639 a trap handler for one of these instructions, and then the trap handler
12640 emulates the effect of the instruction. Because of the trap handler overhead,
12641 this is much slower than calling the ABI library routines. Thus the
12642 @option{-msoft-quad-float} option is the default.
12644 @item -mno-unaligned-doubles
12645 @itemx -munaligned-doubles
12646 @opindex mno-unaligned-doubles
12647 @opindex munaligned-doubles
12648 Assume that doubles have 8 byte alignment. This is the default.
12650 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
12651 alignment only if they are contained in another type, or if they have an
12652 absolute address. Otherwise, it assumes they have 4 byte alignment.
12653 Specifying this option avoids some rare compatibility problems with code
12654 generated by other compilers. It is not the default because it results
12655 in a performance loss, especially for floating point code.
12657 @item -mno-faster-structs
12658 @itemx -mfaster-structs
12659 @opindex mno-faster-structs
12660 @opindex mfaster-structs
12661 With @option{-mfaster-structs}, the compiler assumes that structures
12662 should have 8 byte alignment. This enables the use of pairs of
12663 @code{ldd} and @code{std} instructions for copies in structure
12664 assignment, in place of twice as many @code{ld} and @code{st} pairs.
12665 However, the use of this changed alignment directly violates the SPARC
12666 ABI@. Thus, it's intended only for use on targets where the developer
12667 acknowledges that their resulting code will not be directly in line with
12668 the rules of the ABI@.
12670 @item -mimpure-text
12671 @opindex mimpure-text
12672 @option{-mimpure-text}, used in addition to @option{-shared}, tells
12673 the compiler to not pass @option{-z text} to the linker when linking a
12674 shared object. Using this option, you can link position-dependent
12675 code into a shared object.
12677 @option{-mimpure-text} suppresses the ``relocations remain against
12678 allocatable but non-writable sections'' linker error message.
12679 However, the necessary relocations will trigger copy-on-write, and the
12680 shared object is not actually shared across processes. Instead of
12681 using @option{-mimpure-text}, you should compile all source code with
12682 @option{-fpic} or @option{-fPIC}.
12684 This option is only available on SunOS and Solaris.
12686 @item -mcpu=@var{cpu_type}
12688 Set the instruction set, register set, and instruction scheduling parameters
12689 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
12690 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
12691 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
12692 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
12693 @samp{ultrasparc3}, and @samp{niagara}.
12695 Default instruction scheduling parameters are used for values that select
12696 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
12697 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
12699 Here is a list of each supported architecture and their supported
12704 v8: supersparc, hypersparc
12705 sparclite: f930, f934, sparclite86x
12707 v9: ultrasparc, ultrasparc3, niagara
12710 By default (unless configured otherwise), GCC generates code for the V7
12711 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
12712 additionally optimizes it for the Cypress CY7C602 chip, as used in the
12713 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
12714 SPARCStation 1, 2, IPX etc.
12716 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
12717 architecture. The only difference from V7 code is that the compiler emits
12718 the integer multiply and integer divide instructions which exist in SPARC-V8
12719 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
12720 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
12723 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
12724 the SPARC architecture. This adds the integer multiply, integer divide step
12725 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
12726 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
12727 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
12728 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
12729 MB86934 chip, which is the more recent SPARClite with FPU@.
12731 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
12732 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
12733 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
12734 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
12735 optimizes it for the TEMIC SPARClet chip.
12737 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
12738 architecture. This adds 64-bit integer and floating-point move instructions,
12739 3 additional floating-point condition code registers and conditional move
12740 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
12741 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
12742 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
12743 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
12744 @option{-mcpu=niagara}, the compiler additionally optimizes it for
12745 Sun UltraSPARC T1 chips.
12747 @item -mtune=@var{cpu_type}
12749 Set the instruction scheduling parameters for machine type
12750 @var{cpu_type}, but do not set the instruction set or register set that the
12751 option @option{-mcpu=@var{cpu_type}} would.
12753 The same values for @option{-mcpu=@var{cpu_type}} can be used for
12754 @option{-mtune=@var{cpu_type}}, but the only useful values are those
12755 that select a particular cpu implementation. Those are @samp{cypress},
12756 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
12757 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
12758 @samp{ultrasparc3}, and @samp{niagara}.
12763 @opindex mno-v8plus
12764 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
12765 difference from the V8 ABI is that the global and out registers are
12766 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
12767 mode for all SPARC-V9 processors.
12773 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
12774 Visual Instruction Set extensions. The default is @option{-mno-vis}.
12777 These @samp{-m} options are supported in addition to the above
12778 on SPARC-V9 processors in 64-bit environments:
12781 @item -mlittle-endian
12782 @opindex mlittle-endian
12783 Generate code for a processor running in little-endian mode. It is only
12784 available for a few configurations and most notably not on Solaris and Linux.
12790 Generate code for a 32-bit or 64-bit environment.
12791 The 32-bit environment sets int, long and pointer to 32 bits.
12792 The 64-bit environment sets int to 32 bits and long and pointer
12795 @item -mcmodel=medlow
12796 @opindex mcmodel=medlow
12797 Generate code for the Medium/Low code model: 64-bit addresses, programs
12798 must be linked in the low 32 bits of memory. Programs can be statically
12799 or dynamically linked.
12801 @item -mcmodel=medmid
12802 @opindex mcmodel=medmid
12803 Generate code for the Medium/Middle code model: 64-bit addresses, programs
12804 must be linked in the low 44 bits of memory, the text and data segments must
12805 be less than 2GB in size and the data segment must be located within 2GB of
12808 @item -mcmodel=medany
12809 @opindex mcmodel=medany
12810 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
12811 may be linked anywhere in memory, the text and data segments must be less
12812 than 2GB in size and the data segment must be located within 2GB of the
12815 @item -mcmodel=embmedany
12816 @opindex mcmodel=embmedany
12817 Generate code for the Medium/Anywhere code model for embedded systems:
12818 64-bit addresses, the text and data segments must be less than 2GB in
12819 size, both starting anywhere in memory (determined at link time). The
12820 global register %g4 points to the base of the data segment. Programs
12821 are statically linked and PIC is not supported.
12824 @itemx -mno-stack-bias
12825 @opindex mstack-bias
12826 @opindex mno-stack-bias
12827 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
12828 frame pointer if present, are offset by @minus{}2047 which must be added back
12829 when making stack frame references. This is the default in 64-bit mode.
12830 Otherwise, assume no such offset is present.
12833 These switches are supported in addition to the above on Solaris:
12838 Add support for multithreading using the Solaris threads library. This
12839 option sets flags for both the preprocessor and linker. This option does
12840 not affect the thread safety of object code produced by the compiler or
12841 that of libraries supplied with it.
12845 Add support for multithreading using the POSIX threads library. This
12846 option sets flags for both the preprocessor and linker. This option does
12847 not affect the thread safety of object code produced by the compiler or
12848 that of libraries supplied with it.
12852 This is a synonym for @option{-pthreads}.
12856 @subsection SPU Options
12857 @cindex SPU options
12859 These @samp{-m} options are supported on the SPU:
12863 @itemx -merror-reloc
12864 @opindex mwarn-reloc
12865 @opindex merror-reloc
12867 The loader for SPU does not handle dynamic relocations. By default, GCC
12868 will give an error when it generates code that requires a dynamic
12869 relocation. @option{-mno-error-reloc} disables the error,
12870 @option{-mwarn-reloc} will generate a warning instead.
12873 @itemx -munsafe-dma
12875 @opindex munsafe-dma
12877 Instructions which initiate or test completion of DMA must not be
12878 reordered with respect to loads and stores of the memory which is being
12879 accessed. Users typically address this problem using the volatile
12880 keyword, but that can lead to inefficient code in places where the
12881 memory is known to not change. Rather than mark the memory as volatile
12882 we treat the DMA instructions as potentially effecting all memory. With
12883 @option{-munsafe-dma} users must use the volatile keyword to protect
12886 @item -mbranch-hints
12887 @opindex mbranch-hints
12889 By default, GCC will generate a branch hint instruction to avoid
12890 pipeline stalls for always taken or probably taken branches. A hint
12891 will not be generated closer than 8 instructions away from its branch.
12892 There is little reason to disable them, except for debugging purposes,
12893 or to make an object a little bit smaller.
12897 @opindex msmall-mem
12898 @opindex mlarge-mem
12900 By default, GCC generates code assuming that addresses are never larger
12901 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
12902 a full 32 bit address.
12907 By default, GCC links against startup code that assumes the SPU-style
12908 main function interface (which has an unconventional parameter list).
12909 With @option{-mstdmain}, GCC will link your program against startup
12910 code that assumes a C99-style interface to @code{main}, including a
12911 local copy of @code{argv} strings.
12913 @item -mfixed-range=@var{register-range}
12914 @opindex mfixed-range
12915 Generate code treating the given register range as fixed registers.
12916 A fixed register is one that the register allocator can not use. This is
12917 useful when compiling kernel code. A register range is specified as
12918 two registers separated by a dash. Multiple register ranges can be
12919 specified separated by a comma.
12923 @node System V Options
12924 @subsection Options for System V
12926 These additional options are available on System V Release 4 for
12927 compatibility with other compilers on those systems:
12932 Create a shared object.
12933 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
12937 Identify the versions of each tool used by the compiler, in a
12938 @code{.ident} assembler directive in the output.
12942 Refrain from adding @code{.ident} directives to the output file (this is
12945 @item -YP,@var{dirs}
12947 Search the directories @var{dirs}, and no others, for libraries
12948 specified with @option{-l}.
12950 @item -Ym,@var{dir}
12952 Look in the directory @var{dir} to find the M4 preprocessor.
12953 The assembler uses this option.
12954 @c This is supposed to go with a -Yd for predefined M4 macro files, but
12955 @c the generic assembler that comes with Solaris takes just -Ym.
12958 @node TMS320C3x/C4x Options
12959 @subsection TMS320C3x/C4x Options
12960 @cindex TMS320C3x/C4x Options
12962 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
12966 @item -mcpu=@var{cpu_type}
12968 Set the instruction set, register set, and instruction scheduling
12969 parameters for machine type @var{cpu_type}. Supported values for
12970 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
12971 @samp{c44}. The default is @samp{c40} to generate code for the
12976 @itemx -msmall-memory
12978 @opindex mbig-memory
12980 @opindex msmall-memory
12982 Generates code for the big or small memory model. The small memory
12983 model assumed that all data fits into one 64K word page. At run-time
12984 the data page (DP) register must be set to point to the 64K page
12985 containing the .bss and .data program sections. The big memory model is
12986 the default and requires reloading of the DP register for every direct
12993 Allow (disallow) allocation of general integer operands into the block
12994 count register BK@.
13000 Enable (disable) generation of code using decrement and branch,
13001 DBcond(D), instructions. This is enabled by default for the C4x. To be
13002 on the safe side, this is disabled for the C3x, since the maximum
13003 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
13004 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
13005 that it can utilize the decrement and branch instruction, but will give
13006 up if there is more than one memory reference in the loop. Thus a loop
13007 where the loop counter is decremented can generate slightly more
13008 efficient code, in cases where the RPTB instruction cannot be utilized.
13010 @item -mdp-isr-reload
13012 @opindex mdp-isr-reload
13014 Force the DP register to be saved on entry to an interrupt service
13015 routine (ISR), reloaded to point to the data section, and restored on
13016 exit from the ISR@. This should not be required unless someone has
13017 violated the small memory model by modifying the DP register, say within
13024 For the C3x use the 24-bit MPYI instruction for integer multiplies
13025 instead of a library call to guarantee 32-bit results. Note that if one
13026 of the operands is a constant, then the multiplication will be performed
13027 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
13028 then squaring operations are performed inline instead of a library call.
13031 @itemx -mno-fast-fix
13033 @opindex mno-fast-fix
13034 The C3x/C4x FIX instruction to convert a floating point value to an
13035 integer value chooses the nearest integer less than or equal to the
13036 floating point value rather than to the nearest integer. Thus if the
13037 floating point number is negative, the result will be incorrectly
13038 truncated an additional code is necessary to detect and correct this
13039 case. This option can be used to disable generation of the additional
13040 code required to correct the result.
13046 Enable (disable) generation of repeat block sequences using the RPTB
13047 instruction for zero overhead looping. The RPTB construct is only used
13048 for innermost loops that do not call functions or jump across the loop
13049 boundaries. There is no advantage having nested RPTB loops due to the
13050 overhead required to save and restore the RC, RS, and RE registers.
13051 This is enabled by default with @option{-O2}.
13053 @item -mrpts=@var{count}
13057 Enable (disable) the use of the single instruction repeat instruction
13058 RPTS@. If a repeat block contains a single instruction, and the loop
13059 count can be guaranteed to be less than the value @var{count}, GCC will
13060 emit a RPTS instruction instead of a RPTB@. If no value is specified,
13061 then a RPTS will be emitted even if the loop count cannot be determined
13062 at compile time. Note that the repeated instruction following RPTS does
13063 not have to be reloaded from memory each iteration, thus freeing up the
13064 CPU buses for operands. However, since interrupts are blocked by this
13065 instruction, it is disabled by default.
13067 @item -mloop-unsigned
13068 @itemx -mno-loop-unsigned
13069 @opindex mloop-unsigned
13070 @opindex mno-loop-unsigned
13071 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
13072 is @math{2^{31} + 1} since these instructions test if the iteration count is
13073 negative to terminate the loop. If the iteration count is unsigned
13074 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
13075 exceeded. This switch allows an unsigned iteration count.
13079 Try to emit an assembler syntax that the TI assembler (asm30) is happy
13080 with. This also enforces compatibility with the API employed by the TI
13081 C3x C compiler. For example, long doubles are passed as structures
13082 rather than in floating point registers.
13088 Generate code that uses registers (stack) for passing arguments to functions.
13089 By default, arguments are passed in registers where possible rather
13090 than by pushing arguments on to the stack.
13092 @item -mparallel-insns
13093 @itemx -mno-parallel-insns
13094 @opindex mparallel-insns
13095 @opindex mno-parallel-insns
13096 Allow the generation of parallel instructions. This is enabled by
13097 default with @option{-O2}.
13099 @item -mparallel-mpy
13100 @itemx -mno-parallel-mpy
13101 @opindex mparallel-mpy
13102 @opindex mno-parallel-mpy
13103 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
13104 provided @option{-mparallel-insns} is also specified. These instructions have
13105 tight register constraints which can pessimize the code generation
13106 of large functions.
13111 @subsection V850 Options
13112 @cindex V850 Options
13114 These @samp{-m} options are defined for V850 implementations:
13118 @itemx -mno-long-calls
13119 @opindex mlong-calls
13120 @opindex mno-long-calls
13121 Treat all calls as being far away (near). If calls are assumed to be
13122 far away, the compiler will always load the functions address up into a
13123 register, and call indirect through the pointer.
13129 Do not optimize (do optimize) basic blocks that use the same index
13130 pointer 4 or more times to copy pointer into the @code{ep} register, and
13131 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
13132 option is on by default if you optimize.
13134 @item -mno-prolog-function
13135 @itemx -mprolog-function
13136 @opindex mno-prolog-function
13137 @opindex mprolog-function
13138 Do not use (do use) external functions to save and restore registers
13139 at the prologue and epilogue of a function. The external functions
13140 are slower, but use less code space if more than one function saves
13141 the same number of registers. The @option{-mprolog-function} option
13142 is on by default if you optimize.
13146 Try to make the code as small as possible. At present, this just turns
13147 on the @option{-mep} and @option{-mprolog-function} options.
13149 @item -mtda=@var{n}
13151 Put static or global variables whose size is @var{n} bytes or less into
13152 the tiny data area that register @code{ep} points to. The tiny data
13153 area can hold up to 256 bytes in total (128 bytes for byte references).
13155 @item -msda=@var{n}
13157 Put static or global variables whose size is @var{n} bytes or less into
13158 the small data area that register @code{gp} points to. The small data
13159 area can hold up to 64 kilobytes.
13161 @item -mzda=@var{n}
13163 Put static or global variables whose size is @var{n} bytes or less into
13164 the first 32 kilobytes of memory.
13168 Specify that the target processor is the V850.
13171 @opindex mbig-switch
13172 Generate code suitable for big switch tables. Use this option only if
13173 the assembler/linker complain about out of range branches within a switch
13178 This option will cause r2 and r5 to be used in the code generated by
13179 the compiler. This setting is the default.
13181 @item -mno-app-regs
13182 @opindex mno-app-regs
13183 This option will cause r2 and r5 to be treated as fixed registers.
13187 Specify that the target processor is the V850E1. The preprocessor
13188 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
13189 this option is used.
13193 Specify that the target processor is the V850E@. The preprocessor
13194 constant @samp{__v850e__} will be defined if this option is used.
13196 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
13197 are defined then a default target processor will be chosen and the
13198 relevant @samp{__v850*__} preprocessor constant will be defined.
13200 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
13201 defined, regardless of which processor variant is the target.
13203 @item -mdisable-callt
13204 @opindex mdisable-callt
13205 This option will suppress generation of the CALLT instruction for the
13206 v850e and v850e1 flavors of the v850 architecture. The default is
13207 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
13212 @subsection VAX Options
13213 @cindex VAX options
13215 These @samp{-m} options are defined for the VAX:
13220 Do not output certain jump instructions (@code{aobleq} and so on)
13221 that the Unix assembler for the VAX cannot handle across long
13226 Do output those jump instructions, on the assumption that you
13227 will assemble with the GNU assembler.
13231 Output code for g-format floating point numbers instead of d-format.
13234 @node x86-64 Options
13235 @subsection x86-64 Options
13236 @cindex x86-64 options
13238 These are listed under @xref{i386 and x86-64 Options}.
13240 @node Xstormy16 Options
13241 @subsection Xstormy16 Options
13242 @cindex Xstormy16 Options
13244 These options are defined for Xstormy16:
13249 Choose startup files and linker script suitable for the simulator.
13252 @node Xtensa Options
13253 @subsection Xtensa Options
13254 @cindex Xtensa Options
13256 These options are supported for Xtensa targets:
13260 @itemx -mno-const16
13262 @opindex mno-const16
13263 Enable or disable use of @code{CONST16} instructions for loading
13264 constant values. The @code{CONST16} instruction is currently not a
13265 standard option from Tensilica. When enabled, @code{CONST16}
13266 instructions are always used in place of the standard @code{L32R}
13267 instructions. The use of @code{CONST16} is enabled by default only if
13268 the @code{L32R} instruction is not available.
13271 @itemx -mno-fused-madd
13272 @opindex mfused-madd
13273 @opindex mno-fused-madd
13274 Enable or disable use of fused multiply/add and multiply/subtract
13275 instructions in the floating-point option. This has no effect if the
13276 floating-point option is not also enabled. Disabling fused multiply/add
13277 and multiply/subtract instructions forces the compiler to use separate
13278 instructions for the multiply and add/subtract operations. This may be
13279 desirable in some cases where strict IEEE 754-compliant results are
13280 required: the fused multiply add/subtract instructions do not round the
13281 intermediate result, thereby producing results with @emph{more} bits of
13282 precision than specified by the IEEE standard. Disabling fused multiply
13283 add/subtract instructions also ensures that the program output is not
13284 sensitive to the compiler's ability to combine multiply and add/subtract
13287 @item -mtext-section-literals
13288 @itemx -mno-text-section-literals
13289 @opindex mtext-section-literals
13290 @opindex mno-text-section-literals
13291 Control the treatment of literal pools. The default is
13292 @option{-mno-text-section-literals}, which places literals in a separate
13293 section in the output file. This allows the literal pool to be placed
13294 in a data RAM/ROM, and it also allows the linker to combine literal
13295 pools from separate object files to remove redundant literals and
13296 improve code size. With @option{-mtext-section-literals}, the literals
13297 are interspersed in the text section in order to keep them as close as
13298 possible to their references. This may be necessary for large assembly
13301 @item -mtarget-align
13302 @itemx -mno-target-align
13303 @opindex mtarget-align
13304 @opindex mno-target-align
13305 When this option is enabled, GCC instructs the assembler to
13306 automatically align instructions to reduce branch penalties at the
13307 expense of some code density. The assembler attempts to widen density
13308 instructions to align branch targets and the instructions following call
13309 instructions. If there are not enough preceding safe density
13310 instructions to align a target, no widening will be performed. The
13311 default is @option{-mtarget-align}. These options do not affect the
13312 treatment of auto-aligned instructions like @code{LOOP}, which the
13313 assembler will always align, either by widening density instructions or
13314 by inserting no-op instructions.
13317 @itemx -mno-longcalls
13318 @opindex mlongcalls
13319 @opindex mno-longcalls
13320 When this option is enabled, GCC instructs the assembler to translate
13321 direct calls to indirect calls unless it can determine that the target
13322 of a direct call is in the range allowed by the call instruction. This
13323 translation typically occurs for calls to functions in other source
13324 files. Specifically, the assembler translates a direct @code{CALL}
13325 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
13326 The default is @option{-mno-longcalls}. This option should be used in
13327 programs where the call target can potentially be out of range. This
13328 option is implemented in the assembler, not the compiler, so the
13329 assembly code generated by GCC will still show direct call
13330 instructions---look at the disassembled object code to see the actual
13331 instructions. Note that the assembler will use an indirect call for
13332 every cross-file call, not just those that really will be out of range.
13335 @node zSeries Options
13336 @subsection zSeries Options
13337 @cindex zSeries options
13339 These are listed under @xref{S/390 and zSeries Options}.
13341 @node Code Gen Options
13342 @section Options for Code Generation Conventions
13343 @cindex code generation conventions
13344 @cindex options, code generation
13345 @cindex run-time options
13347 These machine-independent options control the interface conventions
13348 used in code generation.
13350 Most of them have both positive and negative forms; the negative form
13351 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
13352 one of the forms is listed---the one which is not the default. You
13353 can figure out the other form by either removing @samp{no-} or adding
13357 @item -fbounds-check
13358 @opindex fbounds-check
13359 For front-ends that support it, generate additional code to check that
13360 indices used to access arrays are within the declared range. This is
13361 currently only supported by the Java and Fortran front-ends, where
13362 this option defaults to true and false respectively.
13366 This option generates traps for signed overflow on addition, subtraction,
13367 multiplication operations.
13371 This option instructs the compiler to assume that signed arithmetic
13372 overflow of addition, subtraction and multiplication wraps around
13373 using twos-complement representation. This flag enables some optimizations
13374 and disables others. This option is enabled by default for the Java
13375 front-end, as required by the Java language specification.
13378 @opindex fexceptions
13379 Enable exception handling. Generates extra code needed to propagate
13380 exceptions. For some targets, this implies GCC will generate frame
13381 unwind information for all functions, which can produce significant data
13382 size overhead, although it does not affect execution. If you do not
13383 specify this option, GCC will enable it by default for languages like
13384 C++ which normally require exception handling, and disable it for
13385 languages like C that do not normally require it. However, you may need
13386 to enable this option when compiling C code that needs to interoperate
13387 properly with exception handlers written in C++. You may also wish to
13388 disable this option if you are compiling older C++ programs that don't
13389 use exception handling.
13391 @item -fnon-call-exceptions
13392 @opindex fnon-call-exceptions
13393 Generate code that allows trapping instructions to throw exceptions.
13394 Note that this requires platform-specific runtime support that does
13395 not exist everywhere. Moreover, it only allows @emph{trapping}
13396 instructions to throw exceptions, i.e.@: memory references or floating
13397 point instructions. It does not allow exceptions to be thrown from
13398 arbitrary signal handlers such as @code{SIGALRM}.
13400 @item -funwind-tables
13401 @opindex funwind-tables
13402 Similar to @option{-fexceptions}, except that it will just generate any needed
13403 static data, but will not affect the generated code in any other way.
13404 You will normally not enable this option; instead, a language processor
13405 that needs this handling would enable it on your behalf.
13407 @item -fasynchronous-unwind-tables
13408 @opindex fasynchronous-unwind-tables
13409 Generate unwind table in dwarf2 format, if supported by target machine. The
13410 table is exact at each instruction boundary, so it can be used for stack
13411 unwinding from asynchronous events (such as debugger or garbage collector).
13413 @item -fpcc-struct-return
13414 @opindex fpcc-struct-return
13415 Return ``short'' @code{struct} and @code{union} values in memory like
13416 longer ones, rather than in registers. This convention is less
13417 efficient, but it has the advantage of allowing intercallability between
13418 GCC-compiled files and files compiled with other compilers, particularly
13419 the Portable C Compiler (pcc).
13421 The precise convention for returning structures in memory depends
13422 on the target configuration macros.
13424 Short structures and unions are those whose size and alignment match
13425 that of some integer type.
13427 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
13428 switch is not binary compatible with code compiled with the
13429 @option{-freg-struct-return} switch.
13430 Use it to conform to a non-default application binary interface.
13432 @item -freg-struct-return
13433 @opindex freg-struct-return
13434 Return @code{struct} and @code{union} values in registers when possible.
13435 This is more efficient for small structures than
13436 @option{-fpcc-struct-return}.
13438 If you specify neither @option{-fpcc-struct-return} nor
13439 @option{-freg-struct-return}, GCC defaults to whichever convention is
13440 standard for the target. If there is no standard convention, GCC
13441 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
13442 the principal compiler. In those cases, we can choose the standard, and
13443 we chose the more efficient register return alternative.
13445 @strong{Warning:} code compiled with the @option{-freg-struct-return}
13446 switch is not binary compatible with code compiled with the
13447 @option{-fpcc-struct-return} switch.
13448 Use it to conform to a non-default application binary interface.
13450 @item -fshort-enums
13451 @opindex fshort-enums
13452 Allocate to an @code{enum} type only as many bytes as it needs for the
13453 declared range of possible values. Specifically, the @code{enum} type
13454 will be equivalent to the smallest integer type which has enough room.
13456 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
13457 code that is not binary compatible with code generated without that switch.
13458 Use it to conform to a non-default application binary interface.
13460 @item -fshort-double
13461 @opindex fshort-double
13462 Use the same size for @code{double} as for @code{float}.
13464 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
13465 code that is not binary compatible with code generated without that switch.
13466 Use it to conform to a non-default application binary interface.
13468 @item -fshort-wchar
13469 @opindex fshort-wchar
13470 Override the underlying type for @samp{wchar_t} to be @samp{short
13471 unsigned int} instead of the default for the target. This option is
13472 useful for building programs to run under WINE@.
13474 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
13475 code that is not binary compatible with code generated without that switch.
13476 Use it to conform to a non-default application binary interface.
13479 @opindex fno-common
13480 In C, allocate even uninitialized global variables in the data section of the
13481 object file, rather than generating them as common blocks. This has the
13482 effect that if the same variable is declared (without @code{extern}) in
13483 two different compilations, you will get an error when you link them.
13484 The only reason this might be useful is if you wish to verify that the
13485 program will work on other systems which always work this way.
13489 Ignore the @samp{#ident} directive.
13491 @item -finhibit-size-directive
13492 @opindex finhibit-size-directive
13493 Don't output a @code{.size} assembler directive, or anything else that
13494 would cause trouble if the function is split in the middle, and the
13495 two halves are placed at locations far apart in memory. This option is
13496 used when compiling @file{crtstuff.c}; you should not need to use it
13499 @item -fverbose-asm
13500 @opindex fverbose-asm
13501 Put extra commentary information in the generated assembly code to
13502 make it more readable. This option is generally only of use to those
13503 who actually need to read the generated assembly code (perhaps while
13504 debugging the compiler itself).
13506 @option{-fno-verbose-asm}, the default, causes the
13507 extra information to be omitted and is useful when comparing two assembler
13510 @item -frecord-gcc-switches
13511 @opindex frecord-gcc-switches
13512 This switch causes the command line that was used to invoke the
13513 compiler to be recorded into the object file that is being created.
13514 This switch is only implemented on some targets and the exact format
13515 of the recording is target and binary file format dependent, but it
13516 usually takes the form of a section containing ASCII text. This
13517 switch is related to the @option{-fverbose-asm} switch, but that
13518 switch only records information in the assembler output file as
13519 comments, so it never reaches the object file.
13523 @cindex global offset table
13525 Generate position-independent code (PIC) suitable for use in a shared
13526 library, if supported for the target machine. Such code accesses all
13527 constant addresses through a global offset table (GOT)@. The dynamic
13528 loader resolves the GOT entries when the program starts (the dynamic
13529 loader is not part of GCC; it is part of the operating system). If
13530 the GOT size for the linked executable exceeds a machine-specific
13531 maximum size, you get an error message from the linker indicating that
13532 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
13533 instead. (These maximums are 8k on the SPARC and 32k
13534 on the m68k and RS/6000. The 386 has no such limit.)
13536 Position-independent code requires special support, and therefore works
13537 only on certain machines. For the 386, GCC supports PIC for System V
13538 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
13539 position-independent.
13541 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13546 If supported for the target machine, emit position-independent code,
13547 suitable for dynamic linking and avoiding any limit on the size of the
13548 global offset table. This option makes a difference on the m68k,
13549 PowerPC and SPARC@.
13551 Position-independent code requires special support, and therefore works
13552 only on certain machines.
13554 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13561 These options are similar to @option{-fpic} and @option{-fPIC}, but
13562 generated position independent code can be only linked into executables.
13563 Usually these options are used when @option{-pie} GCC option will be
13564 used during linking.
13566 @item -fno-jump-tables
13567 @opindex fno-jump-tables
13568 Do not use jump tables for switch statements even where it would be
13569 more efficient than other code generation strategies. This option is
13570 of use in conjunction with @option{-fpic} or @option{-fPIC} for
13571 building code which forms part of a dynamic linker and cannot
13572 reference the address of a jump table. On some targets, jump tables
13573 do not require a GOT and this option is not needed.
13575 @item -ffixed-@var{reg}
13577 Treat the register named @var{reg} as a fixed register; generated code
13578 should never refer to it (except perhaps as a stack pointer, frame
13579 pointer or in some other fixed role).
13581 @var{reg} must be the name of a register. The register names accepted
13582 are machine-specific and are defined in the @code{REGISTER_NAMES}
13583 macro in the machine description macro file.
13585 This flag does not have a negative form, because it specifies a
13588 @item -fcall-used-@var{reg}
13589 @opindex fcall-used
13590 Treat the register named @var{reg} as an allocable register that is
13591 clobbered by function calls. It may be allocated for temporaries or
13592 variables that do not live across a call. Functions compiled this way
13593 will not save and restore the register @var{reg}.
13595 It is an error to used this flag with the frame pointer or stack pointer.
13596 Use of this flag for other registers that have fixed pervasive roles in
13597 the machine's execution model will produce disastrous results.
13599 This flag does not have a negative form, because it specifies a
13602 @item -fcall-saved-@var{reg}
13603 @opindex fcall-saved
13604 Treat the register named @var{reg} as an allocable register saved by
13605 functions. It may be allocated even for temporaries or variables that
13606 live across a call. Functions compiled this way will save and restore
13607 the register @var{reg} if they use it.
13609 It is an error to used this flag with the frame pointer or stack pointer.
13610 Use of this flag for other registers that have fixed pervasive roles in
13611 the machine's execution model will produce disastrous results.
13613 A different sort of disaster will result from the use of this flag for
13614 a register in which function values may be returned.
13616 This flag does not have a negative form, because it specifies a
13619 @item -fpack-struct[=@var{n}]
13620 @opindex fpack-struct
13621 Without a value specified, pack all structure members together without
13622 holes. When a value is specified (which must be a small power of two), pack
13623 structure members according to this value, representing the maximum
13624 alignment (that is, objects with default alignment requirements larger than
13625 this will be output potentially unaligned at the next fitting location.
13627 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
13628 code that is not binary compatible with code generated without that switch.
13629 Additionally, it makes the code suboptimal.
13630 Use it to conform to a non-default application binary interface.
13632 @item -finstrument-functions
13633 @opindex finstrument-functions
13634 Generate instrumentation calls for entry and exit to functions. Just
13635 after function entry and just before function exit, the following
13636 profiling functions will be called with the address of the current
13637 function and its call site. (On some platforms,
13638 @code{__builtin_return_address} does not work beyond the current
13639 function, so the call site information may not be available to the
13640 profiling functions otherwise.)
13643 void __cyg_profile_func_enter (void *this_fn,
13645 void __cyg_profile_func_exit (void *this_fn,
13649 The first argument is the address of the start of the current function,
13650 which may be looked up exactly in the symbol table.
13652 This instrumentation is also done for functions expanded inline in other
13653 functions. The profiling calls will indicate where, conceptually, the
13654 inline function is entered and exited. This means that addressable
13655 versions of such functions must be available. If all your uses of a
13656 function are expanded inline, this may mean an additional expansion of
13657 code size. If you use @samp{extern inline} in your C code, an
13658 addressable version of such functions must be provided. (This is
13659 normally the case anyways, but if you get lucky and the optimizer always
13660 expands the functions inline, you might have gotten away without
13661 providing static copies.)
13663 A function may be given the attribute @code{no_instrument_function}, in
13664 which case this instrumentation will not be done. This can be used, for
13665 example, for the profiling functions listed above, high-priority
13666 interrupt routines, and any functions from which the profiling functions
13667 cannot safely be called (perhaps signal handlers, if the profiling
13668 routines generate output or allocate memory).
13670 @item -fstack-check
13671 @opindex fstack-check
13672 Generate code to verify that you do not go beyond the boundary of the
13673 stack. You should specify this flag if you are running in an
13674 environment with multiple threads, but only rarely need to specify it in
13675 a single-threaded environment since stack overflow is automatically
13676 detected on nearly all systems if there is only one stack.
13678 Note that this switch does not actually cause checking to be done; the
13679 operating system must do that. The switch causes generation of code
13680 to ensure that the operating system sees the stack being extended.
13682 @item -fstack-limit-register=@var{reg}
13683 @itemx -fstack-limit-symbol=@var{sym}
13684 @itemx -fno-stack-limit
13685 @opindex fstack-limit-register
13686 @opindex fstack-limit-symbol
13687 @opindex fno-stack-limit
13688 Generate code to ensure that the stack does not grow beyond a certain value,
13689 either the value of a register or the address of a symbol. If the stack
13690 would grow beyond the value, a signal is raised. For most targets,
13691 the signal is raised before the stack overruns the boundary, so
13692 it is possible to catch the signal without taking special precautions.
13694 For instance, if the stack starts at absolute address @samp{0x80000000}
13695 and grows downwards, you can use the flags
13696 @option{-fstack-limit-symbol=__stack_limit} and
13697 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
13698 of 128KB@. Note that this may only work with the GNU linker.
13700 @cindex aliasing of parameters
13701 @cindex parameters, aliased
13702 @item -fargument-alias
13703 @itemx -fargument-noalias
13704 @itemx -fargument-noalias-global
13705 @itemx -fargument-noalias-anything
13706 @opindex fargument-alias
13707 @opindex fargument-noalias
13708 @opindex fargument-noalias-global
13709 @opindex fargument-noalias-anything
13710 Specify the possible relationships among parameters and between
13711 parameters and global data.
13713 @option{-fargument-alias} specifies that arguments (parameters) may
13714 alias each other and may alias global storage.@*
13715 @option{-fargument-noalias} specifies that arguments do not alias
13716 each other, but may alias global storage.@*
13717 @option{-fargument-noalias-global} specifies that arguments do not
13718 alias each other and do not alias global storage.
13719 @option{-fargument-noalias-anything} specifies that arguments do not
13720 alias any other storage.
13722 Each language will automatically use whatever option is required by
13723 the language standard. You should not need to use these options yourself.
13725 @item -fleading-underscore
13726 @opindex fleading-underscore
13727 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
13728 change the way C symbols are represented in the object file. One use
13729 is to help link with legacy assembly code.
13731 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
13732 generate code that is not binary compatible with code generated without that
13733 switch. Use it to conform to a non-default application binary interface.
13734 Not all targets provide complete support for this switch.
13736 @item -ftls-model=@var{model}
13737 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
13738 The @var{model} argument should be one of @code{global-dynamic},
13739 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
13741 The default without @option{-fpic} is @code{initial-exec}; with
13742 @option{-fpic} the default is @code{global-dynamic}.
13744 @item -fvisibility=@var{default|internal|hidden|protected}
13745 @opindex fvisibility
13746 Set the default ELF image symbol visibility to the specified option---all
13747 symbols will be marked with this unless overridden within the code.
13748 Using this feature can very substantially improve linking and
13749 load times of shared object libraries, produce more optimized
13750 code, provide near-perfect API export and prevent symbol clashes.
13751 It is @strong{strongly} recommended that you use this in any shared objects
13754 Despite the nomenclature, @code{default} always means public ie;
13755 available to be linked against from outside the shared object.
13756 @code{protected} and @code{internal} are pretty useless in real-world
13757 usage so the only other commonly used option will be @code{hidden}.
13758 The default if @option{-fvisibility} isn't specified is
13759 @code{default}, i.e., make every
13760 symbol public---this causes the same behavior as previous versions of
13763 A good explanation of the benefits offered by ensuring ELF
13764 symbols have the correct visibility is given by ``How To Write
13765 Shared Libraries'' by Ulrich Drepper (which can be found at
13766 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
13767 solution made possible by this option to marking things hidden when
13768 the default is public is to make the default hidden and mark things
13769 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
13770 and @code{__attribute__ ((visibility("default")))} instead of
13771 @code{__declspec(dllexport)} you get almost identical semantics with
13772 identical syntax. This is a great boon to those working with
13773 cross-platform projects.
13775 For those adding visibility support to existing code, you may find
13776 @samp{#pragma GCC visibility} of use. This works by you enclosing
13777 the declarations you wish to set visibility for with (for example)
13778 @samp{#pragma GCC visibility push(hidden)} and
13779 @samp{#pragma GCC visibility pop}.
13780 Bear in mind that symbol visibility should be viewed @strong{as
13781 part of the API interface contract} and thus all new code should
13782 always specify visibility when it is not the default ie; declarations
13783 only for use within the local DSO should @strong{always} be marked explicitly
13784 as hidden as so to avoid PLT indirection overheads---making this
13785 abundantly clear also aids readability and self-documentation of the code.
13786 Note that due to ISO C++ specification requirements, operator new and
13787 operator delete must always be of default visibility.
13789 Be aware that headers from outside your project, in particular system
13790 headers and headers from any other library you use, may not be
13791 expecting to be compiled with visibility other than the default. You
13792 may need to explicitly say @samp{#pragma GCC visibility push(default)}
13793 before including any such headers.
13795 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
13796 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
13797 no modifications. However, this means that calls to @samp{extern}
13798 functions with no explicit visibility will use the PLT, so it is more
13799 effective to use @samp{__attribute ((visibility))} and/or
13800 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
13801 declarations should be treated as hidden.
13803 Note that @samp{-fvisibility} does affect C++ vague linkage
13804 entities. This means that, for instance, an exception class that will
13805 be thrown between DSOs must be explicitly marked with default
13806 visibility so that the @samp{type_info} nodes will be unified between
13809 An overview of these techniques, their benefits and how to use them
13810 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
13816 @node Environment Variables
13817 @section Environment Variables Affecting GCC
13818 @cindex environment variables
13820 @c man begin ENVIRONMENT
13821 This section describes several environment variables that affect how GCC
13822 operates. Some of them work by specifying directories or prefixes to use
13823 when searching for various kinds of files. Some are used to specify other
13824 aspects of the compilation environment.
13826 Note that you can also specify places to search using options such as
13827 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
13828 take precedence over places specified using environment variables, which
13829 in turn take precedence over those specified by the configuration of GCC@.
13830 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
13831 GNU Compiler Collection (GCC) Internals}.
13836 @c @itemx LC_COLLATE
13838 @c @itemx LC_MONETARY
13839 @c @itemx LC_NUMERIC
13844 @c @findex LC_COLLATE
13845 @findex LC_MESSAGES
13846 @c @findex LC_MONETARY
13847 @c @findex LC_NUMERIC
13851 These environment variables control the way that GCC uses
13852 localization information that allow GCC to work with different
13853 national conventions. GCC inspects the locale categories
13854 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
13855 so. These locale categories can be set to any value supported by your
13856 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
13857 Kingdom encoded in UTF-8.
13859 The @env{LC_CTYPE} environment variable specifies character
13860 classification. GCC uses it to determine the character boundaries in
13861 a string; this is needed for some multibyte encodings that contain quote
13862 and escape characters that would otherwise be interpreted as a string
13865 The @env{LC_MESSAGES} environment variable specifies the language to
13866 use in diagnostic messages.
13868 If the @env{LC_ALL} environment variable is set, it overrides the value
13869 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
13870 and @env{LC_MESSAGES} default to the value of the @env{LANG}
13871 environment variable. If none of these variables are set, GCC
13872 defaults to traditional C English behavior.
13876 If @env{TMPDIR} is set, it specifies the directory to use for temporary
13877 files. GCC uses temporary files to hold the output of one stage of
13878 compilation which is to be used as input to the next stage: for example,
13879 the output of the preprocessor, which is the input to the compiler
13882 @item GCC_EXEC_PREFIX
13883 @findex GCC_EXEC_PREFIX
13884 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
13885 names of the subprograms executed by the compiler. No slash is added
13886 when this prefix is combined with the name of a subprogram, but you can
13887 specify a prefix that ends with a slash if you wish.
13889 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
13890 an appropriate prefix to use based on the pathname it was invoked with.
13892 If GCC cannot find the subprogram using the specified prefix, it
13893 tries looking in the usual places for the subprogram.
13895 The default value of @env{GCC_EXEC_PREFIX} is
13896 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
13897 of @code{prefix} when you ran the @file{configure} script.
13899 Other prefixes specified with @option{-B} take precedence over this prefix.
13901 This prefix is also used for finding files such as @file{crt0.o} that are
13904 In addition, the prefix is used in an unusual way in finding the
13905 directories to search for header files. For each of the standard
13906 directories whose name normally begins with @samp{/usr/local/lib/gcc}
13907 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
13908 replacing that beginning with the specified prefix to produce an
13909 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
13910 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
13911 These alternate directories are searched first; the standard directories
13914 @item COMPILER_PATH
13915 @findex COMPILER_PATH
13916 The value of @env{COMPILER_PATH} is a colon-separated list of
13917 directories, much like @env{PATH}. GCC tries the directories thus
13918 specified when searching for subprograms, if it can't find the
13919 subprograms using @env{GCC_EXEC_PREFIX}.
13922 @findex LIBRARY_PATH
13923 The value of @env{LIBRARY_PATH} is a colon-separated list of
13924 directories, much like @env{PATH}. When configured as a native compiler,
13925 GCC tries the directories thus specified when searching for special
13926 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
13927 using GCC also uses these directories when searching for ordinary
13928 libraries for the @option{-l} option (but directories specified with
13929 @option{-L} come first).
13933 @cindex locale definition
13934 This variable is used to pass locale information to the compiler. One way in
13935 which this information is used is to determine the character set to be used
13936 when character literals, string literals and comments are parsed in C and C++.
13937 When the compiler is configured to allow multibyte characters,
13938 the following values for @env{LANG} are recognized:
13942 Recognize JIS characters.
13944 Recognize SJIS characters.
13946 Recognize EUCJP characters.
13949 If @env{LANG} is not defined, or if it has some other value, then the
13950 compiler will use mblen and mbtowc as defined by the default locale to
13951 recognize and translate multibyte characters.
13955 Some additional environments variables affect the behavior of the
13958 @include cppenv.texi
13962 @node Precompiled Headers
13963 @section Using Precompiled Headers
13964 @cindex precompiled headers
13965 @cindex speed of compilation
13967 Often large projects have many header files that are included in every
13968 source file. The time the compiler takes to process these header files
13969 over and over again can account for nearly all of the time required to
13970 build the project. To make builds faster, GCC allows users to
13971 `precompile' a header file; then, if builds can use the precompiled
13972 header file they will be much faster.
13974 To create a precompiled header file, simply compile it as you would any
13975 other file, if necessary using the @option{-x} option to make the driver
13976 treat it as a C or C++ header file. You will probably want to use a
13977 tool like @command{make} to keep the precompiled header up-to-date when
13978 the headers it contains change.
13980 A precompiled header file will be searched for when @code{#include} is
13981 seen in the compilation. As it searches for the included file
13982 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
13983 compiler looks for a precompiled header in each directory just before it
13984 looks for the include file in that directory. The name searched for is
13985 the name specified in the @code{#include} with @samp{.gch} appended. If
13986 the precompiled header file can't be used, it is ignored.
13988 For instance, if you have @code{#include "all.h"}, and you have
13989 @file{all.h.gch} in the same directory as @file{all.h}, then the
13990 precompiled header file will be used if possible, and the original
13991 header will be used otherwise.
13993 Alternatively, you might decide to put the precompiled header file in a
13994 directory and use @option{-I} to ensure that directory is searched
13995 before (or instead of) the directory containing the original header.
13996 Then, if you want to check that the precompiled header file is always
13997 used, you can put a file of the same name as the original header in this
13998 directory containing an @code{#error} command.
14000 This also works with @option{-include}. So yet another way to use
14001 precompiled headers, good for projects not designed with precompiled
14002 header files in mind, is to simply take most of the header files used by
14003 a project, include them from another header file, precompile that header
14004 file, and @option{-include} the precompiled header. If the header files
14005 have guards against multiple inclusion, they will be skipped because
14006 they've already been included (in the precompiled header).
14008 If you need to precompile the same header file for different
14009 languages, targets, or compiler options, you can instead make a
14010 @emph{directory} named like @file{all.h.gch}, and put each precompiled
14011 header in the directory, perhaps using @option{-o}. It doesn't matter
14012 what you call the files in the directory, every precompiled header in
14013 the directory will be considered. The first precompiled header
14014 encountered in the directory that is valid for this compilation will
14015 be used; they're searched in no particular order.
14017 There are many other possibilities, limited only by your imagination,
14018 good sense, and the constraints of your build system.
14020 A precompiled header file can be used only when these conditions apply:
14024 Only one precompiled header can be used in a particular compilation.
14027 A precompiled header can't be used once the first C token is seen. You
14028 can have preprocessor directives before a precompiled header; you can
14029 even include a precompiled header from inside another header, so long as
14030 there are no C tokens before the @code{#include}.
14033 The precompiled header file must be produced for the same language as
14034 the current compilation. You can't use a C precompiled header for a C++
14038 The precompiled header file must have been produced by the same compiler
14039 binary as the current compilation is using.
14042 Any macros defined before the precompiled header is included must
14043 either be defined in the same way as when the precompiled header was
14044 generated, or must not affect the precompiled header, which usually
14045 means that they don't appear in the precompiled header at all.
14047 The @option{-D} option is one way to define a macro before a
14048 precompiled header is included; using a @code{#define} can also do it.
14049 There are also some options that define macros implicitly, like
14050 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
14053 @item If debugging information is output when using the precompiled
14054 header, using @option{-g} or similar, the same kind of debugging information
14055 must have been output when building the precompiled header. However,
14056 a precompiled header built using @option{-g} can be used in a compilation
14057 when no debugging information is being output.
14059 @item The same @option{-m} options must generally be used when building
14060 and using the precompiled header. @xref{Submodel Options},
14061 for any cases where this rule is relaxed.
14063 @item Each of the following options must be the same when building and using
14064 the precompiled header:
14066 @gccoptlist{-fexceptions -funit-at-a-time}
14069 Some other command-line options starting with @option{-f},
14070 @option{-p}, or @option{-O} must be defined in the same way as when
14071 the precompiled header was generated. At present, it's not clear
14072 which options are safe to change and which are not; the safest choice
14073 is to use exactly the same options when generating and using the
14074 precompiled header. The following are known to be safe:
14076 @gccoptlist{-fmessage-length= -fpreprocessed
14077 -fsched-interblock -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous
14078 -fsched-verbose=<number> -fschedule-insns -fvisibility=
14083 For all of these except the last, the compiler will automatically
14084 ignore the precompiled header if the conditions aren't met. If you
14085 find an option combination that doesn't work and doesn't cause the
14086 precompiled header to be ignored, please consider filing a bug report,
14089 If you do use differing options when generating and using the
14090 precompiled header, the actual behavior will be a mixture of the
14091 behavior for the options. For instance, if you use @option{-g} to
14092 generate the precompiled header but not when using it, you may or may
14093 not get debugging information for routines in the precompiled header.
14095 @node Running Protoize
14096 @section Running Protoize
14098 The program @code{protoize} is an optional part of GCC@. You can use
14099 it to add prototypes to a program, thus converting the program to ISO
14100 C in one respect. The companion program @code{unprotoize} does the
14101 reverse: it removes argument types from any prototypes that are found.
14103 When you run these programs, you must specify a set of source files as
14104 command line arguments. The conversion programs start out by compiling
14105 these files to see what functions they define. The information gathered
14106 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
14108 After scanning comes actual conversion. The specified files are all
14109 eligible to be converted; any files they include (whether sources or
14110 just headers) are eligible as well.
14112 But not all the eligible files are converted. By default,
14113 @code{protoize} and @code{unprotoize} convert only source and header
14114 files in the current directory. You can specify additional directories
14115 whose files should be converted with the @option{-d @var{directory}}
14116 option. You can also specify particular files to exclude with the
14117 @option{-x @var{file}} option. A file is converted if it is eligible, its
14118 directory name matches one of the specified directory names, and its
14119 name within the directory has not been excluded.
14121 Basic conversion with @code{protoize} consists of rewriting most
14122 function definitions and function declarations to specify the types of
14123 the arguments. The only ones not rewritten are those for varargs
14126 @code{protoize} optionally inserts prototype declarations at the
14127 beginning of the source file, to make them available for any calls that
14128 precede the function's definition. Or it can insert prototype
14129 declarations with block scope in the blocks where undeclared functions
14132 Basic conversion with @code{unprotoize} consists of rewriting most
14133 function declarations to remove any argument types, and rewriting
14134 function definitions to the old-style pre-ISO form.
14136 Both conversion programs print a warning for any function declaration or
14137 definition that they can't convert. You can suppress these warnings
14140 The output from @code{protoize} or @code{unprotoize} replaces the
14141 original source file. The original file is renamed to a name ending
14142 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
14143 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
14144 for DOS) file already exists, then the source file is simply discarded.
14146 @code{protoize} and @code{unprotoize} both depend on GCC itself to
14147 scan the program and collect information about the functions it uses.
14148 So neither of these programs will work until GCC is installed.
14150 Here is a table of the options you can use with @code{protoize} and
14151 @code{unprotoize}. Each option works with both programs unless
14155 @item -B @var{directory}
14156 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
14157 usual directory (normally @file{/usr/local/lib}). This file contains
14158 prototype information about standard system functions. This option
14159 applies only to @code{protoize}.
14161 @item -c @var{compilation-options}
14162 Use @var{compilation-options} as the options when running @command{gcc} to
14163 produce the @samp{.X} files. The special option @option{-aux-info} is
14164 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
14166 Note that the compilation options must be given as a single argument to
14167 @code{protoize} or @code{unprotoize}. If you want to specify several
14168 @command{gcc} options, you must quote the entire set of compilation options
14169 to make them a single word in the shell.
14171 There are certain @command{gcc} arguments that you cannot use, because they
14172 would produce the wrong kind of output. These include @option{-g},
14173 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
14174 the @var{compilation-options}, they are ignored.
14177 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
14178 systems) instead of @samp{.c}. This is convenient if you are converting
14179 a C program to C++. This option applies only to @code{protoize}.
14182 Add explicit global declarations. This means inserting explicit
14183 declarations at the beginning of each source file for each function
14184 that is called in the file and was not declared. These declarations
14185 precede the first function definition that contains a call to an
14186 undeclared function. This option applies only to @code{protoize}.
14188 @item -i @var{string}
14189 Indent old-style parameter declarations with the string @var{string}.
14190 This option applies only to @code{protoize}.
14192 @code{unprotoize} converts prototyped function definitions to old-style
14193 function definitions, where the arguments are declared between the
14194 argument list and the initial @samp{@{}. By default, @code{unprotoize}
14195 uses five spaces as the indentation. If you want to indent with just
14196 one space instead, use @option{-i " "}.
14199 Keep the @samp{.X} files. Normally, they are deleted after conversion
14203 Add explicit local declarations. @code{protoize} with @option{-l} inserts
14204 a prototype declaration for each function in each block which calls the
14205 function without any declaration. This option applies only to
14209 Make no real changes. This mode just prints information about the conversions
14210 that would have been done without @option{-n}.
14213 Make no @samp{.save} files. The original files are simply deleted.
14214 Use this option with caution.
14216 @item -p @var{program}
14217 Use the program @var{program} as the compiler. Normally, the name
14218 @file{gcc} is used.
14221 Work quietly. Most warnings are suppressed.
14224 Print the version number, just like @option{-v} for @command{gcc}.
14227 If you need special compiler options to compile one of your program's
14228 source files, then you should generate that file's @samp{.X} file
14229 specially, by running @command{gcc} on that source file with the
14230 appropriate options and the option @option{-aux-info}. Then run
14231 @code{protoize} on the entire set of files. @code{protoize} will use
14232 the existing @samp{.X} file because it is newer than the source file.
14236 gcc -Dfoo=bar file1.c -aux-info file1.X
14241 You need to include the special files along with the rest in the
14242 @code{protoize} command, even though their @samp{.X} files already
14243 exist, because otherwise they won't get converted.
14245 @xref{Protoize Caveats}, for more information on how to use
14246 @code{protoize} successfully.