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 -fms-extensions @gol
170 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
171 -fallow-single-precision -fcond-mismatch @gol
172 -fsigned-bitfields -fsigned-char @gol
173 -funsigned-bitfields -funsigned-char}
175 @item C++ Language Options
176 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
177 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
178 -fconserve-space -ffriend-injection @gol
179 -fno-elide-constructors @gol
180 -fno-enforce-eh-specs @gol
181 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
182 -fno-implicit-templates @gol
183 -fno-implicit-inline-templates @gol
184 -fno-implement-inlines -fms-extensions @gol
185 -fno-nonansi-builtins -fno-operator-names @gol
186 -fno-optional-diags -fpermissive @gol
187 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
188 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
189 -fno-default-inline -fvisibility-inlines-hidden @gol
190 -Wabi -Wctor-dtor-privacy @gol
191 -Wnon-virtual-dtor -Wreorder @gol
192 -Weffc++ -Wno-deprecated -Wstrict-null-sentinel @gol
193 -Wno-non-template-friend -Wold-style-cast @gol
194 -Woverloaded-virtual -Wno-pmf-conversions @gol
197 @item Objective-C and Objective-C++ Language Options
198 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
199 Objective-C and Objective-C++ Dialects}.
201 -fconstant-string-class=@var{class-name} @gol
202 -fgnu-runtime -fnext-runtime @gol
203 -fno-nil-receivers @gol
204 -fobjc-call-cxx-cdtors @gol
205 -fobjc-direct-dispatch @gol
206 -fobjc-exceptions @gol
208 -freplace-objc-classes @gol
211 -Wassign-intercept @gol
212 -Wno-protocol -Wselector @gol
213 -Wstrict-selector-match @gol
214 -Wundeclared-selector}
216 @item Language Independent Options
217 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
218 @gccoptlist{-fmessage-length=@var{n} @gol
219 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}} @gol
220 -fdiagnostics-show-options
222 @item Warning Options
223 @xref{Warning Options,,Options to Request or Suppress Warnings}.
224 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
225 -w -Wextra -Wall -Waggregate-return -Walways-true -Wno-attributes @gol
226 -Wc++-compat -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
227 -Wconversion -Wno-deprecated-declarations @gol
228 -Wdisabled-optimization -Wno-div-by-zero -Wno-endif-labels @gol
229 -Werror -Werror-* -Werror-implicit-function-declaration @gol
230 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
231 -Wno-format-extra-args -Wformat-nonliteral @gol
232 -Wformat-security -Wformat-y2k @gol
233 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
234 -Wimport -Wno-import -Winit-self -Winline @gol
235 -Wno-int-to-pointer-cast @gol
236 -Wno-invalid-offsetof -Winvalid-pch @gol
237 -Wlarger-than-@var{len} -Wunsafe-loop-optimizations -Wlong-long @gol
238 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
239 -Wmissing-format-attribute -Wmissing-include-dirs @gol
240 -Wmissing-noreturn @gol
241 -Wno-multichar -Wnonnull -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 @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-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 -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 -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 -mid-shared-library @gol
433 -mno-id-shared-library -mshared-library-id=@var{n} @gol
434 -mlong-calls -mno-long-calls}
437 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
438 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
439 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
440 -mstack-align -mdata-align -mconst-align @gol
441 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
442 -melf -maout -melinux -mlinux -sim -sim2 @gol
443 -mmul-bug-workaround -mno-mul-bug-workaround}
446 @gccoptlist{-mmac -mpush-args}
448 @emph{Darwin Options}
449 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
450 -arch_only -bind_at_load -bundle -bundle_loader @gol
451 -client_name -compatibility_version -current_version @gol
453 -dependency-file -dylib_file -dylinker_install_name @gol
454 -dynamic -dynamiclib -exported_symbols_list @gol
455 -filelist -flat_namespace -force_cpusubtype_ALL @gol
456 -force_flat_namespace -headerpad_max_install_names @gol
457 -image_base -init -install_name -keep_private_externs @gol
458 -multi_module -multiply_defined -multiply_defined_unused @gol
459 -noall_load -no_dead_strip_inits_and_terms @gol
460 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
461 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
462 -private_bundle -read_only_relocs -sectalign @gol
463 -sectobjectsymbols -whyload -seg1addr @gol
464 -sectcreate -sectobjectsymbols -sectorder @gol
465 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
466 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
467 -segprot -segs_read_only_addr -segs_read_write_addr @gol
468 -single_module -static -sub_library -sub_umbrella @gol
469 -twolevel_namespace -umbrella -undefined @gol
470 -unexported_symbols_list -weak_reference_mismatches @gol
471 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
474 @emph{DEC Alpha Options}
475 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
476 -mieee -mieee-with-inexact -mieee-conformant @gol
477 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
478 -mtrap-precision=@var{mode} -mbuild-constants @gol
479 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
480 -mbwx -mmax -mfix -mcix @gol
481 -mfloat-vax -mfloat-ieee @gol
482 -mexplicit-relocs -msmall-data -mlarge-data @gol
483 -msmall-text -mlarge-text @gol
484 -mmemory-latency=@var{time}}
486 @emph{DEC Alpha/VMS Options}
487 @gccoptlist{-mvms-return-codes}
490 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
491 -mhard-float -msoft-float @gol
492 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
493 -mdouble -mno-double @gol
494 -mmedia -mno-media -mmuladd -mno-muladd @gol
495 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
496 -mlinked-fp -mlong-calls -malign-labels @gol
497 -mlibrary-pic -macc-4 -macc-8 @gol
498 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
499 -moptimize-membar -mno-optimize-membar @gol
500 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
501 -mvliw-branch -mno-vliw-branch @gol
502 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
503 -mno-nested-cond-exec -mtomcat-stats @gol
507 @emph{GNU/Linux Options}
508 @gccoptlist{-muclibc}
510 @emph{H8/300 Options}
511 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
514 @gccoptlist{-march=@var{architecture-type} @gol
515 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
516 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
517 -mfixed-range=@var{register-range} @gol
518 -mjump-in-delay -mlinker-opt -mlong-calls @gol
519 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
520 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
521 -mno-jump-in-delay -mno-long-load-store @gol
522 -mno-portable-runtime -mno-soft-float @gol
523 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
524 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
525 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
526 -munix=@var{unix-std} -nolibdld -static -threads}
528 @emph{i386 and x86-64 Options}
529 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
530 -mfpmath=@var{unit} @gol
531 -masm=@var{dialect} -mno-fancy-math-387 @gol
532 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
533 -mno-wide-multiply -mrtd -malign-double @gol
534 -mpreferred-stack-boundary=@var{num} @gol
535 -mmmx -msse -msse2 -msse3 -m3dnow -msselibm @gol
536 -mthreads -mno-align-stringops -minline-all-stringops @gol
537 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
538 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
539 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
540 -mcmodel=@var{code-model} @gol
541 -m32 -m64 -mlarge-data-threshold=@var{num}}
544 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
545 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
546 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
547 -minline-float-divide-max-throughput @gol
548 -minline-int-divide-min-latency @gol
549 -minline-int-divide-max-throughput @gol
550 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
551 -mno-dwarf2-asm -mearly-stop-bits @gol
552 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
553 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
554 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
555 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
556 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
557 -mno-sched-prefer-non-data-spec-insns @gol
558 -mno-sched-prefer-non-control-spec-insns @gol
559 -mno-sched-count-spec-in-critical-path}
561 @emph{M32R/D Options}
562 @gccoptlist{-m32r2 -m32rx -m32r @gol
564 -malign-loops -mno-align-loops @gol
565 -missue-rate=@var{number} @gol
566 -mbranch-cost=@var{number} @gol
567 -mmodel=@var{code-size-model-type} @gol
568 -msdata=@var{sdata-type} @gol
569 -mno-flush-func -mflush-func=@var{name} @gol
570 -mno-flush-trap -mflush-trap=@var{number} @gol
574 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
576 @emph{M680x0 Options}
577 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
578 -m68060 -mcpu32 -m5200 -mcfv4e -m68881 -mbitfield @gol
579 -mc68000 -mc68020 @gol
580 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
581 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
582 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
584 @emph{M68hc1x Options}
585 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
586 -mauto-incdec -minmax -mlong-calls -mshort @gol
587 -msoft-reg-count=@var{count}}
590 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
591 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
592 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
593 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
594 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
597 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
598 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
599 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
600 -mshared -mno-shared -mxgot -mno-xgot -mgp32 -mgp64 @gol
601 -mfp32 -mfp64 -mhard-float -msoft-float @gol
602 -msingle-float -mdouble-float -mdsp -mpaired-single -mips3d @gol
603 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
604 -G@var{num} -membedded-data -mno-embedded-data @gol
605 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
606 -msplit-addresses -mno-split-addresses @gol
607 -mexplicit-relocs -mno-explicit-relocs @gol
608 -mcheck-zero-division -mno-check-zero-division @gol
609 -mdivide-traps -mdivide-breaks @gol
610 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
611 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
612 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
613 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 @gol
614 -mfix-sb1 -mno-fix-sb1 @gol
615 -mflush-func=@var{func} -mno-flush-func @gol
616 -mbranch-likely -mno-branch-likely @gol
617 -mfp-exceptions -mno-fp-exceptions @gol
618 -mvr4130-align -mno-vr4130-align}
621 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
622 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
623 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
624 -mno-base-addresses -msingle-exit -mno-single-exit}
626 @emph{MN10300 Options}
627 @gccoptlist{-mmult-bug -mno-mult-bug @gol
628 -mam33 -mno-am33 @gol
629 -mam33-2 -mno-am33-2 @gol
630 -mreturn-pointer-on-d0 @gol
634 @gccoptlist{-mno-crt0 -mbacc -msim @gol
635 -march=@var{cpu-type} }
637 @emph{PDP-11 Options}
638 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
639 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
640 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
641 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
642 -mbranch-expensive -mbranch-cheap @gol
643 -msplit -mno-split -munix-asm -mdec-asm}
645 @emph{PowerPC Options}
646 See RS/6000 and PowerPC Options.
648 @emph{RS/6000 and PowerPC Options}
649 @gccoptlist{-mcpu=@var{cpu-type} @gol
650 -mtune=@var{cpu-type} @gol
651 -mpower -mno-power -mpower2 -mno-power2 @gol
652 -mpowerpc -mpowerpc64 -mno-powerpc @gol
653 -maltivec -mno-altivec @gol
654 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
655 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
656 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
657 -mnew-mnemonics -mold-mnemonics @gol
658 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
659 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
660 -malign-power -malign-natural @gol
661 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
662 -mstring -mno-string -mupdate -mno-update @gol
663 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
664 -mstrict-align -mno-strict-align -mrelocatable @gol
665 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
666 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
667 -mdynamic-no-pic -maltivec -mswdiv @gol
668 -mprioritize-restricted-insns=@var{priority} @gol
669 -msched-costly-dep=@var{dependence_type} @gol
670 -minsert-sched-nops=@var{scheme} @gol
671 -mcall-sysv -mcall-netbsd @gol
672 -maix-struct-return -msvr4-struct-return @gol
673 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
674 -misel -mno-isel @gol
675 -misel=yes -misel=no @gol
677 -mspe=yes -mspe=no @gol
678 -mvrsave -mno-vrsave @gol
679 -mmulhw -mno-mulhw @gol
680 -mdlmzb -mno-dlmzb @gol
681 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
682 -mprototype -mno-prototype @gol
683 -msim -mmvme -mads -myellowknife -memb -msdata @gol
684 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
686 @emph{S/390 and zSeries Options}
687 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
688 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
689 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
690 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
691 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
692 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
693 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
696 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
697 -m4-nofpu -m4-single-only -m4-single -m4 @gol
698 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
699 -m5-64media -m5-64media-nofpu @gol
700 -m5-32media -m5-32media-nofpu @gol
701 -m5-compact -m5-compact-nofpu @gol
702 -mb -ml -mdalign -mrelax @gol
703 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
704 -mieee -misize -mpadstruct -mspace @gol
705 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
706 -mdivsi3_libfunc=@var{name} @gol
707 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
711 @gccoptlist{-mcpu=@var{cpu-type} @gol
712 -mtune=@var{cpu-type} @gol
713 -mcmodel=@var{code-model} @gol
714 -m32 -m64 -mapp-regs -mno-app-regs @gol
715 -mfaster-structs -mno-faster-structs @gol
716 -mfpu -mno-fpu -mhard-float -msoft-float @gol
717 -mhard-quad-float -msoft-quad-float @gol
718 -mimpure-text -mno-impure-text -mlittle-endian @gol
719 -mstack-bias -mno-stack-bias @gol
720 -munaligned-doubles -mno-unaligned-doubles @gol
721 -mv8plus -mno-v8plus -mvis -mno-vis
722 -threads -pthreads -pthread}
724 @emph{System V Options}
725 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
727 @emph{TMS320C3x/C4x Options}
728 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
729 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
730 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
731 -mparallel-insns -mparallel-mpy -mpreserve-float}
734 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
735 -mprolog-function -mno-prolog-function -mspace @gol
736 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
737 -mapp-regs -mno-app-regs @gol
738 -mdisable-callt -mno-disable-callt @gol
744 @gccoptlist{-mg -mgnu -munix}
746 @emph{x86-64 Options}
747 See i386 and x86-64 Options.
749 @emph{Xstormy16 Options}
752 @emph{Xtensa Options}
753 @gccoptlist{-mconst16 -mno-const16 @gol
754 -mfused-madd -mno-fused-madd @gol
755 -mtext-section-literals -mno-text-section-literals @gol
756 -mtarget-align -mno-target-align @gol
757 -mlongcalls -mno-longcalls}
759 @emph{zSeries Options}
760 See S/390 and zSeries Options.
762 @item Code Generation Options
763 @xref{Code Gen Options,,Options for Code Generation Conventions}.
764 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
765 -ffixed-@var{reg} -fexceptions @gol
766 -fnon-call-exceptions -funwind-tables @gol
767 -fasynchronous-unwind-tables @gol
768 -finhibit-size-directive -finstrument-functions @gol
769 -fno-common -fno-ident @gol
770 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
771 -fno-jump-tables @gol
772 -freg-struct-return -fshort-enums @gol
773 -fshort-double -fshort-wchar @gol
774 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
775 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
776 -fargument-alias -fargument-noalias @gol
777 -fargument-noalias-global -fargument-noalias-anything
778 -fleading-underscore -ftls-model=@var{model} @gol
779 -ftrapv -fwrapv -fbounds-check @gol
780 -fvisibility -fopenmp}
784 * Overall Options:: Controlling the kind of output:
785 an executable, object files, assembler files,
786 or preprocessed source.
787 * C Dialect Options:: Controlling the variant of C language compiled.
788 * C++ Dialect Options:: Variations on C++.
789 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
791 * Language Independent Options:: Controlling how diagnostics should be
793 * Warning Options:: How picky should the compiler be?
794 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
795 * Optimize Options:: How much optimization?
796 * Preprocessor Options:: Controlling header files and macro definitions.
797 Also, getting dependency information for Make.
798 * Assembler Options:: Passing options to the assembler.
799 * Link Options:: Specifying libraries and so on.
800 * Directory Options:: Where to find header files and libraries.
801 Where to find the compiler executable files.
802 * Spec Files:: How to pass switches to sub-processes.
803 * Target Options:: Running a cross-compiler, or an old version of GCC.
806 @node Overall Options
807 @section Options Controlling the Kind of Output
809 Compilation can involve up to four stages: preprocessing, compilation
810 proper, assembly and linking, always in that order. GCC is capable of
811 preprocessing and compiling several files either into several
812 assembler input files, or into one assembler input file; then each
813 assembler input file produces an object file, and linking combines all
814 the object files (those newly compiled, and those specified as input)
815 into an executable file.
817 @cindex file name suffix
818 For any given input file, the file name suffix determines what kind of
823 C source code which must be preprocessed.
826 C source code which should not be preprocessed.
829 C++ source code which should not be preprocessed.
832 Objective-C source code. Note that you must link with the @file{libobjc}
833 library to make an Objective-C program work.
836 Objective-C source code which should not be preprocessed.
840 Objective-C++ source code. Note that you must link with the @file{libobjc}
841 library to make an Objective-C++ program work. Note that @samp{.M} refers
842 to a literal capital M@.
845 Objective-C++ source code which should not be preprocessed.
848 C, C++, Objective-C or Objective-C++ header file to be turned into a
853 @itemx @var{file}.cxx
854 @itemx @var{file}.cpp
855 @itemx @var{file}.CPP
856 @itemx @var{file}.c++
858 C++ source code which must be preprocessed. Note that in @samp{.cxx},
859 the last two letters must both be literally @samp{x}. Likewise,
860 @samp{.C} refers to a literal capital C@.
864 Objective-C++ source code which must be preprocessed.
867 Objective-C++ source code which should not be preprocessed.
871 C++ header file to be turned into a precompiled header.
874 @itemx @var{file}.for
875 @itemx @var{file}.FOR
876 Fixed form Fortran source code which should not be preprocessed.
879 @itemx @var{file}.fpp
880 @itemx @var{file}.FPP
881 Fixed form Fortran source code which must be preprocessed (with the traditional
885 @itemx @var{file}.f95
886 Free form Fortran source code which should not be preprocessed.
889 @itemx @var{file}.F95
890 Free form Fortran source code which must be preprocessed (with the
891 traditional preprocessor).
893 @c FIXME: Descriptions of Java file types.
900 Ada source code file which contains a library unit declaration (a
901 declaration of a package, subprogram, or generic, or a generic
902 instantiation), or a library unit renaming declaration (a package,
903 generic, or subprogram renaming declaration). Such files are also
906 @itemx @var{file}.adb
907 Ada source code file containing a library unit body (a subprogram or
908 package body). Such files are also called @dfn{bodies}.
910 @c GCC also knows about some suffixes for languages not yet included:
921 Assembler code which must be preprocessed.
924 An object file to be fed straight into linking.
925 Any file name with no recognized suffix is treated this way.
929 You can specify the input language explicitly with the @option{-x} option:
932 @item -x @var{language}
933 Specify explicitly the @var{language} for the following input files
934 (rather than letting the compiler choose a default based on the file
935 name suffix). This option applies to all following input files until
936 the next @option{-x} option. Possible values for @var{language} are:
938 c c-header c-cpp-output
939 c++ c++-header c++-cpp-output
940 objective-c objective-c-header objective-c-cpp-output
941 objective-c++ objective-c++-header objective-c++-cpp-output
942 assembler assembler-with-cpp
951 Turn off any specification of a language, so that subsequent files are
952 handled according to their file name suffixes (as they are if @option{-x}
953 has not been used at all).
955 @item -pass-exit-codes
956 @opindex pass-exit-codes
957 Normally the @command{gcc} program will exit with the code of 1 if any
958 phase of the compiler returns a non-success return code. If you specify
959 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
960 numerically highest error produced by any phase that returned an error
961 indication. The C, C++, and Fortran frontends return 4, if an internal
962 compiler error is encountered.
965 If you only want some of the stages of compilation, you can use
966 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
967 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
968 @command{gcc} is to stop. Note that some combinations (for example,
969 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
974 Compile or assemble the source files, but do not link. The linking
975 stage simply is not done. The ultimate output is in the form of an
976 object file for each source file.
978 By default, the object file name for a source file is made by replacing
979 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
981 Unrecognized input files, not requiring compilation or assembly, are
986 Stop after the stage of compilation proper; do not assemble. The output
987 is in the form of an assembler code file for each non-assembler input
990 By default, the assembler file name for a source file is made by
991 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
993 Input files that don't require compilation are ignored.
997 Stop after the preprocessing stage; do not run the compiler proper. The
998 output is in the form of preprocessed source code, which is sent to the
1001 Input files which don't require preprocessing are ignored.
1003 @cindex output file option
1006 Place output in file @var{file}. This applies regardless to whatever
1007 sort of output is being produced, whether it be an executable file,
1008 an object file, an assembler file or preprocessed C code.
1010 If @option{-o} is not specified, the default is to put an executable
1011 file in @file{a.out}, the object file for
1012 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1013 assembler file in @file{@var{source}.s}, a precompiled header file in
1014 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1019 Print (on standard error output) the commands executed to run the stages
1020 of compilation. Also print the version number of the compiler driver
1021 program and of the preprocessor and the compiler proper.
1025 Like @option{-v} except the commands are not executed and all command
1026 arguments are quoted. This is useful for shell scripts to capture the
1027 driver-generated command lines.
1031 Use pipes rather than temporary files for communication between the
1032 various stages of compilation. This fails to work on some systems where
1033 the assembler is unable to read from a pipe; but the GNU assembler has
1038 If you are compiling multiple source files, this option tells the driver
1039 to pass all the source files to the compiler at once (for those
1040 languages for which the compiler can handle this). This will allow
1041 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1042 language for which this is supported is C@. If you pass source files for
1043 multiple languages to the driver, using this option, the driver will invoke
1044 the compiler(s) that support IMA once each, passing each compiler all the
1045 source files appropriate for it. For those languages that do not support
1046 IMA this option will be ignored, and the compiler will be invoked once for
1047 each source file in that language. If you use this option in conjunction
1048 with @option{-save-temps}, the compiler will generate multiple
1050 (one for each source file), but only one (combined) @file{.o} or
1055 Print (on the standard output) a description of the command line options
1056 understood by @command{gcc}. If the @option{-v} option is also specified
1057 then @option{--help} will also be passed on to the various processes
1058 invoked by @command{gcc}, so that they can display the command line options
1059 they accept. If the @option{-Wextra} option is also specified then command
1060 line options which have no documentation associated with them will also
1064 @opindex target-help
1065 Print (on the standard output) a description of target specific command
1066 line options for each tool.
1070 Display the version number and copyrights of the invoked GCC@.
1072 @include @value{srcdir}/../libiberty/at-file.texi
1076 @section Compiling C++ Programs
1078 @cindex suffixes for C++ source
1079 @cindex C++ source file suffixes
1080 C++ source files conventionally use one of the suffixes @samp{.C},
1081 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1082 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1083 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1084 files with these names and compiles them as C++ programs even if you
1085 call the compiler the same way as for compiling C programs (usually
1086 with the name @command{gcc}).
1090 However, C++ programs often require class libraries as well as a
1091 compiler that understands the C++ language---and under some
1092 circumstances, you might want to compile programs or header files from
1093 standard input, or otherwise without a suffix that flags them as C++
1094 programs. You might also like to precompile a C header file with a
1095 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
1096 program that calls GCC with the default language set to C++, and
1097 automatically specifies linking against the C++ library. On many
1098 systems, @command{g++} is also installed with the name @command{c++}.
1100 @cindex invoking @command{g++}
1101 When you compile C++ programs, you may specify many of the same
1102 command-line options that you use for compiling programs in any
1103 language; or command-line options meaningful for C and related
1104 languages; or options that are meaningful only for C++ programs.
1105 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1106 explanations of options for languages related to C@.
1107 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1108 explanations of options that are meaningful only for C++ programs.
1110 @node C Dialect Options
1111 @section Options Controlling C Dialect
1112 @cindex dialect options
1113 @cindex language dialect options
1114 @cindex options, dialect
1116 The following options control the dialect of C (or languages derived
1117 from C, such as C++, Objective-C and Objective-C++) that the compiler
1121 @cindex ANSI support
1125 In C mode, support all ISO C90 programs. In C++ mode,
1126 remove GNU extensions that conflict with ISO C++.
1128 This turns off certain features of GCC that are incompatible with ISO
1129 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1130 such as the @code{asm} and @code{typeof} keywords, and
1131 predefined macros such as @code{unix} and @code{vax} that identify the
1132 type of system you are using. It also enables the undesirable and
1133 rarely used ISO trigraph feature. For the C compiler,
1134 it disables recognition of C++ style @samp{//} comments as well as
1135 the @code{inline} keyword.
1137 The alternate keywords @code{__asm__}, @code{__extension__},
1138 @code{__inline__} and @code{__typeof__} continue to work despite
1139 @option{-ansi}. You would not want to use them in an ISO C program, of
1140 course, but it is useful to put them in header files that might be included
1141 in compilations done with @option{-ansi}. Alternate predefined macros
1142 such as @code{__unix__} and @code{__vax__} are also available, with or
1143 without @option{-ansi}.
1145 The @option{-ansi} option does not cause non-ISO programs to be
1146 rejected gratuitously. For that, @option{-pedantic} is required in
1147 addition to @option{-ansi}. @xref{Warning Options}.
1149 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1150 option is used. Some header files may notice this macro and refrain
1151 from declaring certain functions or defining certain macros that the
1152 ISO standard doesn't call for; this is to avoid interfering with any
1153 programs that might use these names for other things.
1155 Functions which would normally be built in but do not have semantics
1156 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1157 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1158 built-in functions provided by GCC}, for details of the functions
1163 Determine the language standard. This option is currently only
1164 supported when compiling C or C++. A value for this option must be
1165 provided; possible values are
1170 ISO C90 (same as @option{-ansi}).
1172 @item iso9899:199409
1173 ISO C90 as modified in amendment 1.
1179 ISO C99. Note that this standard is not yet fully supported; see
1180 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1181 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1184 Default, ISO C90 plus GNU extensions (including some C99 features).
1188 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1189 this will become the default. The name @samp{gnu9x} is deprecated.
1192 The 1998 ISO C++ standard plus amendments.
1195 The same as @option{-std=c++98} plus GNU extensions. This is the
1196 default for C++ code.
1199 Even when this option is not specified, you can still use some of the
1200 features of newer standards in so far as they do not conflict with
1201 previous C standards. For example, you may use @code{__restrict__} even
1202 when @option{-std=c99} is not specified.
1204 The @option{-std} options specifying some version of ISO C have the same
1205 effects as @option{-ansi}, except that features that were not in ISO C90
1206 but are in the specified version (for example, @samp{//} comments and
1207 the @code{inline} keyword in ISO C99) are not disabled.
1209 @xref{Standards,,Language Standards Supported by GCC}, for details of
1210 these standard versions.
1212 @item -aux-info @var{filename}
1214 Output to the given filename prototyped declarations for all functions
1215 declared and/or defined in a translation unit, including those in header
1216 files. This option is silently ignored in any language other than C@.
1218 Besides declarations, the file indicates, in comments, the origin of
1219 each declaration (source file and line), whether the declaration was
1220 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1221 @samp{O} for old, respectively, in the first character after the line
1222 number and the colon), and whether it came from a declaration or a
1223 definition (@samp{C} or @samp{F}, respectively, in the following
1224 character). In the case of function definitions, a K&R-style list of
1225 arguments followed by their declarations is also provided, inside
1226 comments, after the declaration.
1230 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1231 keyword, so that code can use these words as identifiers. You can use
1232 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1233 instead. @option{-ansi} implies @option{-fno-asm}.
1235 In C++, this switch only affects the @code{typeof} keyword, since
1236 @code{asm} and @code{inline} are standard keywords. You may want to
1237 use the @option{-fno-gnu-keywords} flag instead, which has the same
1238 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1239 switch only affects the @code{asm} and @code{typeof} keywords, since
1240 @code{inline} is a standard keyword in ISO C99.
1243 @itemx -fno-builtin-@var{function}
1244 @opindex fno-builtin
1245 @cindex built-in functions
1246 Don't recognize built-in functions that do not begin with
1247 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1248 functions provided by GCC}, for details of the functions affected,
1249 including those which are not built-in functions when @option{-ansi} or
1250 @option{-std} options for strict ISO C conformance are used because they
1251 do not have an ISO standard meaning.
1253 GCC normally generates special code to handle certain built-in functions
1254 more efficiently; for instance, calls to @code{alloca} may become single
1255 instructions that adjust the stack directly, and calls to @code{memcpy}
1256 may become inline copy loops. The resulting code is often both smaller
1257 and faster, but since the function calls no longer appear as such, you
1258 cannot set a breakpoint on those calls, nor can you change the behavior
1259 of the functions by linking with a different library. In addition,
1260 when a function is recognized as a built-in function, GCC may use
1261 information about that function to warn about problems with calls to
1262 that function, or to generate more efficient code, even if the
1263 resulting code still contains calls to that function. For example,
1264 warnings are given with @option{-Wformat} for bad calls to
1265 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1266 known not to modify global memory.
1268 With the @option{-fno-builtin-@var{function}} option
1269 only the built-in function @var{function} is
1270 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1271 function is named this is not built-in in this version of GCC, this
1272 option is ignored. There is no corresponding
1273 @option{-fbuiltin-@var{function}} option; if you wish to enable
1274 built-in functions selectively when using @option{-fno-builtin} or
1275 @option{-ffreestanding}, you may define macros such as:
1278 #define abs(n) __builtin_abs ((n))
1279 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1284 @cindex hosted environment
1286 Assert that compilation takes place in a hosted environment. This implies
1287 @option{-fbuiltin}. A hosted environment is one in which the
1288 entire standard library is available, and in which @code{main} has a return
1289 type of @code{int}. Examples are nearly everything except a kernel.
1290 This is equivalent to @option{-fno-freestanding}.
1292 @item -ffreestanding
1293 @opindex ffreestanding
1294 @cindex hosted environment
1296 Assert that compilation takes place in a freestanding environment. This
1297 implies @option{-fno-builtin}. A freestanding environment
1298 is one in which the standard library may not exist, and program startup may
1299 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1300 This is equivalent to @option{-fno-hosted}.
1302 @xref{Standards,,Language Standards Supported by GCC}, for details of
1303 freestanding and hosted environments.
1305 @item -fms-extensions
1306 @opindex fms-extensions
1307 Accept some non-standard constructs used in Microsoft header files.
1309 Some cases of unnamed fields in structures and unions are only
1310 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1311 fields within structs/unions}, for details.
1315 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1316 options for strict ISO C conformance) implies @option{-trigraphs}.
1318 @item -no-integrated-cpp
1319 @opindex no-integrated-cpp
1320 Performs a compilation in two passes: preprocessing and compiling. This
1321 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1322 @option{-B} option. The user supplied compilation step can then add in
1323 an additional preprocessing step after normal preprocessing but before
1324 compiling. The default is to use the integrated cpp (internal cpp)
1326 The semantics of this option will change if "cc1", "cc1plus", and
1327 "cc1obj" are merged.
1329 @cindex traditional C language
1330 @cindex C language, traditional
1332 @itemx -traditional-cpp
1333 @opindex traditional-cpp
1334 @opindex traditional
1335 Formerly, these options caused GCC to attempt to emulate a pre-standard
1336 C compiler. They are now only supported with the @option{-E} switch.
1337 The preprocessor continues to support a pre-standard mode. See the GNU
1338 CPP manual for details.
1340 @item -fcond-mismatch
1341 @opindex fcond-mismatch
1342 Allow conditional expressions with mismatched types in the second and
1343 third arguments. The value of such an expression is void. This option
1344 is not supported for C++.
1346 @item -funsigned-char
1347 @opindex funsigned-char
1348 Let the type @code{char} be unsigned, like @code{unsigned char}.
1350 Each kind of machine has a default for what @code{char} should
1351 be. It is either like @code{unsigned char} by default or like
1352 @code{signed char} by default.
1354 Ideally, a portable program should always use @code{signed char} or
1355 @code{unsigned char} when it depends on the signedness of an object.
1356 But many programs have been written to use plain @code{char} and
1357 expect it to be signed, or expect it to be unsigned, depending on the
1358 machines they were written for. This option, and its inverse, let you
1359 make such a program work with the opposite default.
1361 The type @code{char} is always a distinct type from each of
1362 @code{signed char} or @code{unsigned char}, even though its behavior
1363 is always just like one of those two.
1366 @opindex fsigned-char
1367 Let the type @code{char} be signed, like @code{signed char}.
1369 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1370 the negative form of @option{-funsigned-char}. Likewise, the option
1371 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1373 @item -fsigned-bitfields
1374 @itemx -funsigned-bitfields
1375 @itemx -fno-signed-bitfields
1376 @itemx -fno-unsigned-bitfields
1377 @opindex fsigned-bitfields
1378 @opindex funsigned-bitfields
1379 @opindex fno-signed-bitfields
1380 @opindex fno-unsigned-bitfields
1381 These options control whether a bit-field is signed or unsigned, when the
1382 declaration does not use either @code{signed} or @code{unsigned}. By
1383 default, such a bit-field is signed, because this is consistent: the
1384 basic integer types such as @code{int} are signed types.
1387 @node C++ Dialect Options
1388 @section Options Controlling C++ Dialect
1390 @cindex compiler options, C++
1391 @cindex C++ options, command line
1392 @cindex options, C++
1393 This section describes the command-line options that are only meaningful
1394 for C++ programs; but you can also use most of the GNU compiler options
1395 regardless of what language your program is in. For example, you
1396 might compile a file @code{firstClass.C} like this:
1399 g++ -g -frepo -O -c firstClass.C
1403 In this example, only @option{-frepo} is an option meant
1404 only for C++ programs; you can use the other options with any
1405 language supported by GCC@.
1407 Here is a list of options that are @emph{only} for compiling C++ programs:
1411 @item -fabi-version=@var{n}
1412 @opindex fabi-version
1413 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1414 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1415 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1416 the version that conforms most closely to the C++ ABI specification.
1417 Therefore, the ABI obtained using version 0 will change as ABI bugs
1420 The default is version 2.
1422 @item -fno-access-control
1423 @opindex fno-access-control
1424 Turn off all access checking. This switch is mainly useful for working
1425 around bugs in the access control code.
1429 Check that the pointer returned by @code{operator new} is non-null
1430 before attempting to modify the storage allocated. This check is
1431 normally unnecessary because the C++ standard specifies that
1432 @code{operator new} will only return @code{0} if it is declared
1433 @samp{throw()}, in which case the compiler will always check the
1434 return value even without this option. In all other cases, when
1435 @code{operator new} has a non-empty exception specification, memory
1436 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1437 @samp{new (nothrow)}.
1439 @item -fconserve-space
1440 @opindex fconserve-space
1441 Put uninitialized or runtime-initialized global variables into the
1442 common segment, as C does. This saves space in the executable at the
1443 cost of not diagnosing duplicate definitions. If you compile with this
1444 flag and your program mysteriously crashes after @code{main()} has
1445 completed, you may have an object that is being destroyed twice because
1446 two definitions were merged.
1448 This option is no longer useful on most targets, now that support has
1449 been added for putting variables into BSS without making them common.
1451 @item -ffriend-injection
1452 @opindex ffriend-injection
1453 Inject friend functions into the enclosing namespace, so that they are
1454 visible outside the scope of the class in which they are declared.
1455 Friend functions were documented to work this way in the old Annotated
1456 C++ Reference Manual, and versions of G++ before 4.1 always worked
1457 that way. However, in ISO C++ a friend function which is not declared
1458 in an enclosing scope can only be found using argument dependent
1459 lookup. This option causes friends to be injected as they were in
1462 This option is for compatibility, and may be removed in a future
1465 @item -fno-elide-constructors
1466 @opindex fno-elide-constructors
1467 The C++ standard allows an implementation to omit creating a temporary
1468 which is only used to initialize another object of the same type.
1469 Specifying this option disables that optimization, and forces G++ to
1470 call the copy constructor in all cases.
1472 @item -fno-enforce-eh-specs
1473 @opindex fno-enforce-eh-specs
1474 Don't generate code to check for violation of exception specifications
1475 at runtime. This option violates the C++ standard, but may be useful
1476 for reducing code size in production builds, much like defining
1477 @samp{NDEBUG}. This does not give user code permission to throw
1478 exceptions in violation of the exception specifications; the compiler
1479 will still optimize based on the specifications, so throwing an
1480 unexpected exception will result in undefined behavior.
1483 @itemx -fno-for-scope
1485 @opindex fno-for-scope
1486 If @option{-ffor-scope} is specified, the scope of variables declared in
1487 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1488 as specified by the C++ standard.
1489 If @option{-fno-for-scope} is specified, the scope of variables declared in
1490 a @i{for-init-statement} extends to the end of the enclosing scope,
1491 as was the case in old versions of G++, and other (traditional)
1492 implementations of C++.
1494 The default if neither flag is given to follow the standard,
1495 but to allow and give a warning for old-style code that would
1496 otherwise be invalid, or have different behavior.
1498 @item -fno-gnu-keywords
1499 @opindex fno-gnu-keywords
1500 Do not recognize @code{typeof} as a keyword, so that code can use this
1501 word as an identifier. You can use the keyword @code{__typeof__} instead.
1502 @option{-ansi} implies @option{-fno-gnu-keywords}.
1504 @item -fno-implicit-templates
1505 @opindex fno-implicit-templates
1506 Never emit code for non-inline templates which are instantiated
1507 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1508 @xref{Template Instantiation}, for more information.
1510 @item -fno-implicit-inline-templates
1511 @opindex fno-implicit-inline-templates
1512 Don't emit code for implicit instantiations of inline templates, either.
1513 The default is to handle inlines differently so that compiles with and
1514 without optimization will need the same set of explicit instantiations.
1516 @item -fno-implement-inlines
1517 @opindex fno-implement-inlines
1518 To save space, do not emit out-of-line copies of inline functions
1519 controlled by @samp{#pragma implementation}. This will cause linker
1520 errors if these functions are not inlined everywhere they are called.
1522 @item -fms-extensions
1523 @opindex fms-extensions
1524 Disable pedantic warnings about constructs used in MFC, such as implicit
1525 int and getting a pointer to member function via non-standard syntax.
1527 @item -fno-nonansi-builtins
1528 @opindex fno-nonansi-builtins
1529 Disable built-in declarations of functions that are not mandated by
1530 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1531 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1533 @item -fno-operator-names
1534 @opindex fno-operator-names
1535 Do not treat the operator name keywords @code{and}, @code{bitand},
1536 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1537 synonyms as keywords.
1539 @item -fno-optional-diags
1540 @opindex fno-optional-diags
1541 Disable diagnostics that the standard says a compiler does not need to
1542 issue. Currently, the only such diagnostic issued by G++ is the one for
1543 a name having multiple meanings within a class.
1546 @opindex fpermissive
1547 Downgrade some diagnostics about nonconformant code from errors to
1548 warnings. Thus, using @option{-fpermissive} will allow some
1549 nonconforming code to compile.
1553 Enable automatic template instantiation at link time. This option also
1554 implies @option{-fno-implicit-templates}. @xref{Template
1555 Instantiation}, for more information.
1559 Disable generation of information about every class with virtual
1560 functions for use by the C++ runtime type identification features
1561 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1562 of the language, you can save some space by using this flag. Note that
1563 exception handling uses the same information, but it will generate it as
1568 Emit statistics about front-end processing at the end of the compilation.
1569 This information is generally only useful to the G++ development team.
1571 @item -ftemplate-depth-@var{n}
1572 @opindex ftemplate-depth
1573 Set the maximum instantiation depth for template classes to @var{n}.
1574 A limit on the template instantiation depth is needed to detect
1575 endless recursions during template class instantiation. ANSI/ISO C++
1576 conforming programs must not rely on a maximum depth greater than 17.
1578 @item -fno-threadsafe-statics
1579 @opindex fno-threadsafe-statics
1580 Do not emit the extra code to use the routines specified in the C++
1581 ABI for thread-safe initialization of local statics. You can use this
1582 option to reduce code size slightly in code that doesn't need to be
1585 @item -fuse-cxa-atexit
1586 @opindex fuse-cxa-atexit
1587 Register destructors for objects with static storage duration with the
1588 @code{__cxa_atexit} function rather than the @code{atexit} function.
1589 This option is required for fully standards-compliant handling of static
1590 destructors, but will only work if your C library supports
1591 @code{__cxa_atexit}.
1593 @item -fno-use-cxa-get-exception-ptr
1594 @opindex fno-use-cxa-get-exception-ptr
1595 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1596 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1597 if the runtime routine is not available.
1599 @item -fvisibility-inlines-hidden
1600 @opindex fvisibility-inlines-hidden
1601 This switch declares that the user does not attempt to compare
1602 pointers to inline methods where the addresses of the two functions
1603 were taken in different shared objects.
1605 The effect of this is that GCC may, effectively, mark inline methods with
1606 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1607 appear in the export table of a DSO and do not require a PLT indirection
1608 when used within the DSO@. Enabling this option can have a dramatic effect
1609 on load and link times of a DSO as it massively reduces the size of the
1610 dynamic export table when the library makes heavy use of templates.
1612 The behaviour of this switch is not quite the same as marking the
1613 methods as hidden directly. Normally if there is a class with default
1614 visibility which has a hidden method, the effect of this is that the
1615 method must be defined in only one shared object. This switch does
1616 not have this restriction.
1618 You may mark a method as having a visibility explicitly to negate the
1619 effect of the switch for that method. For example, if you do want to
1620 compare pointers to a particular inline method, you might mark it as
1621 having default visibility.
1625 Do not use weak symbol support, even if it is provided by the linker.
1626 By default, G++ will use weak symbols if they are available. This
1627 option exists only for testing, and should not be used by end-users;
1628 it will result in inferior code and has no benefits. This option may
1629 be removed in a future release of G++.
1633 Do not search for header files in the standard directories specific to
1634 C++, but do still search the other standard directories. (This option
1635 is used when building the C++ library.)
1638 In addition, these optimization, warning, and code generation options
1639 have meanings only for C++ programs:
1642 @item -fno-default-inline
1643 @opindex fno-default-inline
1644 Do not assume @samp{inline} for functions defined inside a class scope.
1645 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1646 functions will have linkage like inline functions; they just won't be
1649 @item -Wabi @r{(C++ only)}
1651 Warn when G++ generates code that is probably not compatible with the
1652 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1653 all such cases, there are probably some cases that are not warned about,
1654 even though G++ is generating incompatible code. There may also be
1655 cases where warnings are emitted even though the code that is generated
1658 You should rewrite your code to avoid these warnings if you are
1659 concerned about the fact that code generated by G++ may not be binary
1660 compatible with code generated by other compilers.
1662 The known incompatibilities at this point include:
1667 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1668 pack data into the same byte as a base class. For example:
1671 struct A @{ virtual void f(); int f1 : 1; @};
1672 struct B : public A @{ int f2 : 1; @};
1676 In this case, G++ will place @code{B::f2} into the same byte
1677 as@code{A::f1}; other compilers will not. You can avoid this problem
1678 by explicitly padding @code{A} so that its size is a multiple of the
1679 byte size on your platform; that will cause G++ and other compilers to
1680 layout @code{B} identically.
1683 Incorrect handling of tail-padding for virtual bases. G++ does not use
1684 tail padding when laying out virtual bases. For example:
1687 struct A @{ virtual void f(); char c1; @};
1688 struct B @{ B(); char c2; @};
1689 struct C : public A, public virtual B @{@};
1693 In this case, G++ will not place @code{B} into the tail-padding for
1694 @code{A}; other compilers will. You can avoid this problem by
1695 explicitly padding @code{A} so that its size is a multiple of its
1696 alignment (ignoring virtual base classes); that will cause G++ and other
1697 compilers to layout @code{C} identically.
1700 Incorrect handling of bit-fields with declared widths greater than that
1701 of their underlying types, when the bit-fields appear in a union. For
1705 union U @{ int i : 4096; @};
1709 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1710 union too small by the number of bits in an @code{int}.
1713 Empty classes can be placed at incorrect offsets. For example:
1723 struct C : public B, public A @{@};
1727 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1728 it should be placed at offset zero. G++ mistakenly believes that the
1729 @code{A} data member of @code{B} is already at offset zero.
1732 Names of template functions whose types involve @code{typename} or
1733 template template parameters can be mangled incorrectly.
1736 template <typename Q>
1737 void f(typename Q::X) @{@}
1739 template <template <typename> class Q>
1740 void f(typename Q<int>::X) @{@}
1744 Instantiations of these templates may be mangled incorrectly.
1748 @item -Wctor-dtor-privacy @r{(C++ only)}
1749 @opindex Wctor-dtor-privacy
1750 Warn when a class seems unusable because all the constructors or
1751 destructors in that class are private, and it has neither friends nor
1752 public static member functions.
1754 @item -Wnon-virtual-dtor @r{(C++ only)}
1755 @opindex Wnon-virtual-dtor
1756 Warn when a class appears to be polymorphic, thereby requiring a virtual
1757 destructor, yet it declares a non-virtual one. This warning is also
1758 enabled if -Weffc++ is specified.
1760 @item -Wreorder @r{(C++ only)}
1762 @cindex reordering, warning
1763 @cindex warning for reordering of member initializers
1764 Warn when the order of member initializers given in the code does not
1765 match the order in which they must be executed. For instance:
1771 A(): j (0), i (1) @{ @}
1775 The compiler will rearrange the member initializers for @samp{i}
1776 and @samp{j} to match the declaration order of the members, emitting
1777 a warning to that effect. This warning is enabled by @option{-Wall}.
1780 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1783 @item -Weffc++ @r{(C++ only)}
1785 Warn about violations of the following style guidelines from Scott Meyers'
1786 @cite{Effective C++} book:
1790 Item 11: Define a copy constructor and an assignment operator for classes
1791 with dynamically allocated memory.
1794 Item 12: Prefer initialization to assignment in constructors.
1797 Item 14: Make destructors virtual in base classes.
1800 Item 15: Have @code{operator=} return a reference to @code{*this}.
1803 Item 23: Don't try to return a reference when you must return an object.
1807 Also warn about violations of the following style guidelines from
1808 Scott Meyers' @cite{More Effective C++} book:
1812 Item 6: Distinguish between prefix and postfix forms of increment and
1813 decrement operators.
1816 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1820 When selecting this option, be aware that the standard library
1821 headers do not obey all of these guidelines; use @samp{grep -v}
1822 to filter out those warnings.
1824 @item -Wno-deprecated @r{(C++ only)}
1825 @opindex Wno-deprecated
1826 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1828 @item -Wstrict-null-sentinel @r{(C++ only)}
1829 @opindex Wstrict-null-sentinel
1830 Warn also about the use of an uncasted @code{NULL} as sentinel. When
1831 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
1832 to @code{__null}. Although it is a null pointer constant not a null pointer,
1833 it is guaranteed to of the same size as a pointer. But this use is
1834 not portable across different compilers.
1836 @item -Wno-non-template-friend @r{(C++ only)}
1837 @opindex Wno-non-template-friend
1838 Disable warnings when non-templatized friend functions are declared
1839 within a template. Since the advent of explicit template specification
1840 support in G++, if the name of the friend is an unqualified-id (i.e.,
1841 @samp{friend foo(int)}), the C++ language specification demands that the
1842 friend declare or define an ordinary, nontemplate function. (Section
1843 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1844 could be interpreted as a particular specialization of a templatized
1845 function. Because this non-conforming behavior is no longer the default
1846 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1847 check existing code for potential trouble spots and is on by default.
1848 This new compiler behavior can be turned off with
1849 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1850 but disables the helpful warning.
1852 @item -Wold-style-cast @r{(C++ only)}
1853 @opindex Wold-style-cast
1854 Warn if an old-style (C-style) cast to a non-void type is used within
1855 a C++ program. The new-style casts (@samp{dynamic_cast},
1856 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
1857 less vulnerable to unintended effects and much easier to search for.
1859 @item -Woverloaded-virtual @r{(C++ only)}
1860 @opindex Woverloaded-virtual
1861 @cindex overloaded virtual fn, warning
1862 @cindex warning for overloaded virtual fn
1863 Warn when a function declaration hides virtual functions from a
1864 base class. For example, in:
1871 struct B: public A @{
1876 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1884 will fail to compile.
1886 @item -Wno-pmf-conversions @r{(C++ only)}
1887 @opindex Wno-pmf-conversions
1888 Disable the diagnostic for converting a bound pointer to member function
1891 @item -Wsign-promo @r{(C++ only)}
1892 @opindex Wsign-promo
1893 Warn when overload resolution chooses a promotion from unsigned or
1894 enumerated type to a signed type, over a conversion to an unsigned type of
1895 the same size. Previous versions of G++ would try to preserve
1896 unsignedness, but the standard mandates the current behavior.
1901 A& operator = (int);
1911 In this example, G++ will synthesize a default @samp{A& operator =
1912 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1915 @node Objective-C and Objective-C++ Dialect Options
1916 @section Options Controlling Objective-C and Objective-C++ Dialects
1918 @cindex compiler options, Objective-C and Objective-C++
1919 @cindex Objective-C and Objective-C++ options, command line
1920 @cindex options, Objective-C and Objective-C++
1921 (NOTE: This manual does not describe the Objective-C and Objective-C++
1922 languages themselves. See @xref{Standards,,Language Standards
1923 Supported by GCC}, for references.)
1925 This section describes the command-line options that are only meaningful
1926 for Objective-C and Objective-C++ programs, but you can also use most of
1927 the language-independent GNU compiler options.
1928 For example, you might compile a file @code{some_class.m} like this:
1931 gcc -g -fgnu-runtime -O -c some_class.m
1935 In this example, @option{-fgnu-runtime} is an option meant only for
1936 Objective-C and Objective-C++ programs; you can use the other options with
1937 any language supported by GCC@.
1939 Note that since Objective-C is an extension of the C language, Objective-C
1940 compilations may also use options specific to the C front-end (e.g.,
1941 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1942 C++-specific options (e.g., @option{-Wabi}).
1944 Here is a list of options that are @emph{only} for compiling Objective-C
1945 and Objective-C++ programs:
1948 @item -fconstant-string-class=@var{class-name}
1949 @opindex fconstant-string-class
1950 Use @var{class-name} as the name of the class to instantiate for each
1951 literal string specified with the syntax @code{@@"@dots{}"}. The default
1952 class name is @code{NXConstantString} if the GNU runtime is being used, and
1953 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1954 @option{-fconstant-cfstrings} option, if also present, will override the
1955 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1956 to be laid out as constant CoreFoundation strings.
1959 @opindex fgnu-runtime
1960 Generate object code compatible with the standard GNU Objective-C
1961 runtime. This is the default for most types of systems.
1963 @item -fnext-runtime
1964 @opindex fnext-runtime
1965 Generate output compatible with the NeXT runtime. This is the default
1966 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1967 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1970 @item -fno-nil-receivers
1971 @opindex fno-nil-receivers
1972 Assume that all Objective-C message dispatches (e.g.,
1973 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1974 is not @code{nil}. This allows for more efficient entry points in the runtime
1975 to be used. Currently, this option is only available in conjunction with
1976 the NeXT runtime on Mac OS X 10.3 and later.
1978 @item -fobjc-call-cxx-cdtors
1979 @opindex fobjc-call-cxx-cdtors
1980 For each Objective-C class, check if any of its instance variables is a
1981 C++ object with a non-trivial default constructor. If so, synthesize a
1982 special @code{- (id) .cxx_construct} instance method that will run
1983 non-trivial default constructors on any such instance variables, in order,
1984 and then return @code{self}. Similarly, check if any instance variable
1985 is a C++ object with a non-trivial destructor, and if so, synthesize a
1986 special @code{- (void) .cxx_destruct} method that will run
1987 all such default destructors, in reverse order.
1989 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
1990 thusly generated will only operate on instance variables declared in the
1991 current Objective-C class, and not those inherited from superclasses. It
1992 is the responsibility of the Objective-C runtime to invoke all such methods
1993 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
1994 will be invoked by the runtime immediately after a new object
1995 instance is allocated; the @code{- (void) .cxx_destruct} methods will
1996 be invoked immediately before the runtime deallocates an object instance.
1998 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
1999 support for invoking the @code{- (id) .cxx_construct} and
2000 @code{- (void) .cxx_destruct} methods.
2002 @item -fobjc-direct-dispatch
2003 @opindex fobjc-direct-dispatch
2004 Allow fast jumps to the message dispatcher. On Darwin this is
2005 accomplished via the comm page.
2007 @item -fobjc-exceptions
2008 @opindex fobjc-exceptions
2009 Enable syntactic support for structured exception handling in Objective-C,
2010 similar to what is offered by C++ and Java. This option is
2011 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2020 @@catch (AnObjCClass *exc) @{
2027 @@catch (AnotherClass *exc) @{
2030 @@catch (id allOthers) @{
2040 The @code{@@throw} statement may appear anywhere in an Objective-C or
2041 Objective-C++ program; when used inside of a @code{@@catch} block, the
2042 @code{@@throw} may appear without an argument (as shown above), in which case
2043 the object caught by the @code{@@catch} will be rethrown.
2045 Note that only (pointers to) Objective-C objects may be thrown and
2046 caught using this scheme. When an object is thrown, it will be caught
2047 by the nearest @code{@@catch} clause capable of handling objects of that type,
2048 analogously to how @code{catch} blocks work in C++ and Java. A
2049 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2050 any and all Objective-C exceptions not caught by previous @code{@@catch}
2053 The @code{@@finally} clause, if present, will be executed upon exit from the
2054 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2055 regardless of whether any exceptions are thrown, caught or rethrown
2056 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2057 of the @code{finally} clause in Java.
2059 There are several caveats to using the new exception mechanism:
2063 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2064 idioms provided by the @code{NSException} class, the new
2065 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2066 systems, due to additional functionality needed in the (NeXT) Objective-C
2070 As mentioned above, the new exceptions do not support handling
2071 types other than Objective-C objects. Furthermore, when used from
2072 Objective-C++, the Objective-C exception model does not interoperate with C++
2073 exceptions at this time. This means you cannot @code{@@throw} an exception
2074 from Objective-C and @code{catch} it in C++, or vice versa
2075 (i.e., @code{throw @dots{} @@catch}).
2078 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2079 blocks for thread-safe execution:
2082 @@synchronized (ObjCClass *guard) @{
2087 Upon entering the @code{@@synchronized} block, a thread of execution shall
2088 first check whether a lock has been placed on the corresponding @code{guard}
2089 object by another thread. If it has, the current thread shall wait until
2090 the other thread relinquishes its lock. Once @code{guard} becomes available,
2091 the current thread will place its own lock on it, execute the code contained in
2092 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2093 making @code{guard} available to other threads).
2095 Unlike Java, Objective-C does not allow for entire methods to be marked
2096 @code{@@synchronized}. Note that throwing exceptions out of
2097 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2098 to be unlocked properly.
2102 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2104 @item -freplace-objc-classes
2105 @opindex freplace-objc-classes
2106 Emit a special marker instructing @command{ld(1)} not to statically link in
2107 the resulting object file, and allow @command{dyld(1)} to load it in at
2108 run time instead. This is used in conjunction with the Fix-and-Continue
2109 debugging mode, where the object file in question may be recompiled and
2110 dynamically reloaded in the course of program execution, without the need
2111 to restart the program itself. Currently, Fix-and-Continue functionality
2112 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2117 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2118 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2119 compile time) with static class references that get initialized at load time,
2120 which improves run-time performance. Specifying the @option{-fzero-link} flag
2121 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2122 to be retained. This is useful in Zero-Link debugging mode, since it allows
2123 for individual class implementations to be modified during program execution.
2127 Dump interface declarations for all classes seen in the source file to a
2128 file named @file{@var{sourcename}.decl}.
2130 @item -Wassign-intercept
2131 @opindex Wassign-intercept
2132 Warn whenever an Objective-C assignment is being intercepted by the
2136 @opindex Wno-protocol
2137 If a class is declared to implement a protocol, a warning is issued for
2138 every method in the protocol that is not implemented by the class. The
2139 default behavior is to issue a warning for every method not explicitly
2140 implemented in the class, even if a method implementation is inherited
2141 from the superclass. If you use the @option{-Wno-protocol} option, then
2142 methods inherited from the superclass are considered to be implemented,
2143 and no warning is issued for them.
2147 Warn if multiple methods of different types for the same selector are
2148 found during compilation. The check is performed on the list of methods
2149 in the final stage of compilation. Additionally, a check is performed
2150 for each selector appearing in a @code{@@selector(@dots{})}
2151 expression, and a corresponding method for that selector has been found
2152 during compilation. Because these checks scan the method table only at
2153 the end of compilation, these warnings are not produced if the final
2154 stage of compilation is not reached, for example because an error is
2155 found during compilation, or because the @option{-fsyntax-only} option is
2158 @item -Wstrict-selector-match
2159 @opindex Wstrict-selector-match
2160 Warn if multiple methods with differing argument and/or return types are
2161 found for a given selector when attempting to send a message using this
2162 selector to a receiver of type @code{id} or @code{Class}. When this flag
2163 is off (which is the default behavior), the compiler will omit such warnings
2164 if any differences found are confined to types which share the same size
2167 @item -Wundeclared-selector
2168 @opindex Wundeclared-selector
2169 Warn if a @code{@@selector(@dots{})} expression referring to an
2170 undeclared selector is found. A selector is considered undeclared if no
2171 method with that name has been declared before the
2172 @code{@@selector(@dots{})} expression, either explicitly in an
2173 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2174 an @code{@@implementation} section. This option always performs its
2175 checks as soon as a @code{@@selector(@dots{})} expression is found,
2176 while @option{-Wselector} only performs its checks in the final stage of
2177 compilation. This also enforces the coding style convention
2178 that methods and selectors must be declared before being used.
2180 @item -print-objc-runtime-info
2181 @opindex print-objc-runtime-info
2182 Generate C header describing the largest structure that is passed by
2187 @node Language Independent Options
2188 @section Options to Control Diagnostic Messages Formatting
2189 @cindex options to control diagnostics formatting
2190 @cindex diagnostic messages
2191 @cindex message formatting
2193 Traditionally, diagnostic messages have been formatted irrespective of
2194 the output device's aspect (e.g.@: its width, @dots{}). The options described
2195 below can be used to control the diagnostic messages formatting
2196 algorithm, e.g.@: how many characters per line, how often source location
2197 information should be reported. Right now, only the C++ front end can
2198 honor these options. However it is expected, in the near future, that
2199 the remaining front ends would be able to digest them correctly.
2202 @item -fmessage-length=@var{n}
2203 @opindex fmessage-length
2204 Try to format error messages so that they fit on lines of about @var{n}
2205 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2206 the front ends supported by GCC@. If @var{n} is zero, then no
2207 line-wrapping will be done; each error message will appear on a single
2210 @opindex fdiagnostics-show-location
2211 @item -fdiagnostics-show-location=once
2212 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2213 reporter to emit @emph{once} source location information; that is, in
2214 case the message is too long to fit on a single physical line and has to
2215 be wrapped, the source location won't be emitted (as prefix) again,
2216 over and over, in subsequent continuation lines. This is the default
2219 @item -fdiagnostics-show-location=every-line
2220 Only meaningful in line-wrapping mode. Instructs the diagnostic
2221 messages reporter to emit the same source location information (as
2222 prefix) for physical lines that result from the process of breaking
2223 a message which is too long to fit on a single line.
2225 @item -fdiagnostics-show-options
2226 @opindex fdiagnostics-show-options
2227 This option instructs the diagnostic machinery to add text to each
2228 diagnostic emitted, which indicates which command line option directly
2229 controls that diagnostic, when such an option is known to the
2230 diagnostic machinery.
2234 @node Warning Options
2235 @section Options to Request or Suppress Warnings
2236 @cindex options to control warnings
2237 @cindex warning messages
2238 @cindex messages, warning
2239 @cindex suppressing warnings
2241 Warnings are diagnostic messages that report constructions which
2242 are not inherently erroneous but which are risky or suggest there
2243 may have been an error.
2245 You can request many specific warnings with options beginning @samp{-W},
2246 for example @option{-Wimplicit} to request warnings on implicit
2247 declarations. Each of these specific warning options also has a
2248 negative form beginning @samp{-Wno-} to turn off warnings;
2249 for example, @option{-Wno-implicit}. This manual lists only one of the
2250 two forms, whichever is not the default.
2252 The following options control the amount and kinds of warnings produced
2253 by GCC; for further, language-specific options also refer to
2254 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2258 @cindex syntax checking
2260 @opindex fsyntax-only
2261 Check the code for syntax errors, but don't do anything beyond that.
2265 Issue all the warnings demanded by strict ISO C and ISO C++;
2266 reject all programs that use forbidden extensions, and some other
2267 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2268 version of the ISO C standard specified by any @option{-std} option used.
2270 Valid ISO C and ISO C++ programs should compile properly with or without
2271 this option (though a rare few will require @option{-ansi} or a
2272 @option{-std} option specifying the required version of ISO C)@. However,
2273 without this option, certain GNU extensions and traditional C and C++
2274 features are supported as well. With this option, they are rejected.
2276 @option{-pedantic} does not cause warning messages for use of the
2277 alternate keywords whose names begin and end with @samp{__}. Pedantic
2278 warnings are also disabled in the expression that follows
2279 @code{__extension__}. However, only system header files should use
2280 these escape routes; application programs should avoid them.
2281 @xref{Alternate Keywords}.
2283 Some users try to use @option{-pedantic} to check programs for strict ISO
2284 C conformance. They soon find that it does not do quite what they want:
2285 it finds some non-ISO practices, but not all---only those for which
2286 ISO C @emph{requires} a diagnostic, and some others for which
2287 diagnostics have been added.
2289 A feature to report any failure to conform to ISO C might be useful in
2290 some instances, but would require considerable additional work and would
2291 be quite different from @option{-pedantic}. We don't have plans to
2292 support such a feature in the near future.
2294 Where the standard specified with @option{-std} represents a GNU
2295 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2296 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2297 extended dialect is based. Warnings from @option{-pedantic} are given
2298 where they are required by the base standard. (It would not make sense
2299 for such warnings to be given only for features not in the specified GNU
2300 C dialect, since by definition the GNU dialects of C include all
2301 features the compiler supports with the given option, and there would be
2302 nothing to warn about.)
2304 @item -pedantic-errors
2305 @opindex pedantic-errors
2306 Like @option{-pedantic}, except that errors are produced rather than
2311 Inhibit all warning messages.
2315 Inhibit warning messages about the use of @samp{#import}.
2317 @item -Wchar-subscripts
2318 @opindex Wchar-subscripts
2319 Warn if an array subscript has type @code{char}. This is a common cause
2320 of error, as programmers often forget that this type is signed on some
2322 This warning is enabled by @option{-Wall}.
2326 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2327 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2328 This warning is enabled by @option{-Wall}.
2330 @item -Wfatal-errors
2331 @opindex Wfatal-errors
2332 This option causes the compiler to abort compilation on the first error
2333 occurred rather than trying to keep going and printing further error
2338 @opindex ffreestanding
2339 @opindex fno-builtin
2340 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2341 the arguments supplied have types appropriate to the format string
2342 specified, and that the conversions specified in the format string make
2343 sense. This includes standard functions, and others specified by format
2344 attributes (@pxref{Function Attributes}), in the @code{printf},
2345 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2346 not in the C standard) families (or other target-specific families).
2347 Which functions are checked without format attributes having been
2348 specified depends on the standard version selected, and such checks of
2349 functions without the attribute specified are disabled by
2350 @option{-ffreestanding} or @option{-fno-builtin}.
2352 The formats are checked against the format features supported by GNU
2353 libc version 2.2. These include all ISO C90 and C99 features, as well
2354 as features from the Single Unix Specification and some BSD and GNU
2355 extensions. Other library implementations may not support all these
2356 features; GCC does not support warning about features that go beyond a
2357 particular library's limitations. However, if @option{-pedantic} is used
2358 with @option{-Wformat}, warnings will be given about format features not
2359 in the selected standard version (but not for @code{strfmon} formats,
2360 since those are not in any version of the C standard). @xref{C Dialect
2361 Options,,Options Controlling C Dialect}.
2363 Since @option{-Wformat} also checks for null format arguments for
2364 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2366 @option{-Wformat} is included in @option{-Wall}. For more control over some
2367 aspects of format checking, the options @option{-Wformat-y2k},
2368 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2369 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2370 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2373 @opindex Wformat-y2k
2374 If @option{-Wformat} is specified, also warn about @code{strftime}
2375 formats which may yield only a two-digit year.
2377 @item -Wno-format-extra-args
2378 @opindex Wno-format-extra-args
2379 If @option{-Wformat} is specified, do not warn about excess arguments to a
2380 @code{printf} or @code{scanf} format function. The C standard specifies
2381 that such arguments are ignored.
2383 Where the unused arguments lie between used arguments that are
2384 specified with @samp{$} operand number specifications, normally
2385 warnings are still given, since the implementation could not know what
2386 type to pass to @code{va_arg} to skip the unused arguments. However,
2387 in the case of @code{scanf} formats, this option will suppress the
2388 warning if the unused arguments are all pointers, since the Single
2389 Unix Specification says that such unused arguments are allowed.
2391 @item -Wno-format-zero-length
2392 @opindex Wno-format-zero-length
2393 If @option{-Wformat} is specified, do not warn about zero-length formats.
2394 The C standard specifies that zero-length formats are allowed.
2396 @item -Wformat-nonliteral
2397 @opindex Wformat-nonliteral
2398 If @option{-Wformat} is specified, also warn if the format string is not a
2399 string literal and so cannot be checked, unless the format function
2400 takes its format arguments as a @code{va_list}.
2402 @item -Wformat-security
2403 @opindex Wformat-security
2404 If @option{-Wformat} is specified, also warn about uses of format
2405 functions that represent possible security problems. At present, this
2406 warns about calls to @code{printf} and @code{scanf} functions where the
2407 format string is not a string literal and there are no format arguments,
2408 as in @code{printf (foo);}. This may be a security hole if the format
2409 string came from untrusted input and contains @samp{%n}. (This is
2410 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2411 in future warnings may be added to @option{-Wformat-security} that are not
2412 included in @option{-Wformat-nonliteral}.)
2416 Enable @option{-Wformat} plus format checks not included in
2417 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2418 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2422 Warn about passing a null pointer for arguments marked as
2423 requiring a non-null value by the @code{nonnull} function attribute.
2425 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2426 can be disabled with the @option{-Wno-nonnull} option.
2428 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2430 Warn about uninitialized variables which are initialized with themselves.
2431 Note this option can only be used with the @option{-Wuninitialized} option,
2432 which in turn only works with @option{-O1} and above.
2434 For example, GCC will warn about @code{i} being uninitialized in the
2435 following snippet only when @option{-Winit-self} has been specified:
2446 @item -Wimplicit-int
2447 @opindex Wimplicit-int
2448 Warn when a declaration does not specify a type.
2449 This warning is enabled by @option{-Wall}.
2451 @item -Wimplicit-function-declaration
2452 @itemx -Werror-implicit-function-declaration
2453 @opindex Wimplicit-function-declaration
2454 @opindex Werror-implicit-function-declaration
2455 Give a warning (or error) whenever a function is used before being
2456 declared. The form @option{-Wno-error-implicit-function-declaration}
2458 This warning is enabled by @option{-Wall} (as a warning, not an error).
2462 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2463 This warning is enabled by @option{-Wall}.
2467 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2468 function with external linkage, returning int, taking either zero
2469 arguments, two, or three arguments of appropriate types.
2470 This warning is enabled by @option{-Wall}.
2472 @item -Wmissing-braces
2473 @opindex Wmissing-braces
2474 Warn if an aggregate or union initializer is not fully bracketed. In
2475 the following example, the initializer for @samp{a} is not fully
2476 bracketed, but that for @samp{b} is fully bracketed.
2479 int a[2][2] = @{ 0, 1, 2, 3 @};
2480 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2483 This warning is enabled by @option{-Wall}.
2485 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2486 @opindex Wmissing-include-dirs
2487 Warn if a user-supplied include directory does not exist.
2490 @opindex Wparentheses
2491 Warn if parentheses are omitted in certain contexts, such
2492 as when there is an assignment in a context where a truth value
2493 is expected, or when operators are nested whose precedence people
2494 often get confused about. Only the warning for an assignment used as
2495 a truth value is supported when compiling C++; the other warnings are
2496 only supported when compiling C@.
2498 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2499 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2500 interpretation from that of ordinary mathematical notation.
2502 Also warn about constructions where there may be confusion to which
2503 @code{if} statement an @code{else} branch belongs. Here is an example of
2518 In C, every @code{else} branch belongs to the innermost possible @code{if}
2519 statement, which in this example is @code{if (b)}. This is often not
2520 what the programmer expected, as illustrated in the above example by
2521 indentation the programmer chose. When there is the potential for this
2522 confusion, GCC will issue a warning when this flag is specified.
2523 To eliminate the warning, add explicit braces around the innermost
2524 @code{if} statement so there is no way the @code{else} could belong to
2525 the enclosing @code{if}. The resulting code would look like this:
2541 This warning is enabled by @option{-Wall}.
2543 @item -Wsequence-point
2544 @opindex Wsequence-point
2545 Warn about code that may have undefined semantics because of violations
2546 of sequence point rules in the C and C++ standards.
2548 The C and C++ standards defines the order in which expressions in a C/C++
2549 program are evaluated in terms of @dfn{sequence points}, which represent
2550 a partial ordering between the execution of parts of the program: those
2551 executed before the sequence point, and those executed after it. These
2552 occur after the evaluation of a full expression (one which is not part
2553 of a larger expression), after the evaluation of the first operand of a
2554 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2555 function is called (but after the evaluation of its arguments and the
2556 expression denoting the called function), and in certain other places.
2557 Other than as expressed by the sequence point rules, the order of
2558 evaluation of subexpressions of an expression is not specified. All
2559 these rules describe only a partial order rather than a total order,
2560 since, for example, if two functions are called within one expression
2561 with no sequence point between them, the order in which the functions
2562 are called is not specified. However, the standards committee have
2563 ruled that function calls do not overlap.
2565 It is not specified when between sequence points modifications to the
2566 values of objects take effect. Programs whose behavior depends on this
2567 have undefined behavior; the C and C++ standards specify that ``Between
2568 the previous and next sequence point an object shall have its stored
2569 value modified at most once by the evaluation of an expression.
2570 Furthermore, the prior value shall be read only to determine the value
2571 to be stored.''. If a program breaks these rules, the results on any
2572 particular implementation are entirely unpredictable.
2574 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2575 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2576 diagnosed by this option, and it may give an occasional false positive
2577 result, but in general it has been found fairly effective at detecting
2578 this sort of problem in programs.
2580 The standard is worded confusingly, therefore there is some debate
2581 over the precise meaning of the sequence point rules in subtle cases.
2582 Links to discussions of the problem, including proposed formal
2583 definitions, may be found on the GCC readings page, at
2584 @w{@uref{http://gcc.gnu.org/readings.html}}.
2586 This warning is enabled by @option{-Wall} for C and C++.
2589 @opindex Wreturn-type
2590 Warn whenever a function is defined with a return-type that defaults to
2591 @code{int}. Also warn about any @code{return} statement with no
2592 return-value in a function whose return-type is not @code{void}.
2594 For C, also warn if the return type of a function has a type qualifier
2595 such as @code{const}. Such a type qualifier has no effect, since the
2596 value returned by a function is not an lvalue. ISO C prohibits
2597 qualified @code{void} return types on function definitions, so such
2598 return types always receive a warning even without this option.
2600 For C++, a function without return type always produces a diagnostic
2601 message, even when @option{-Wno-return-type} is specified. The only
2602 exceptions are @samp{main} and functions defined in system headers.
2604 This warning is enabled by @option{-Wall}.
2608 Warn whenever a @code{switch} statement has an index of enumerated type
2609 and lacks a @code{case} for one or more of the named codes of that
2610 enumeration. (The presence of a @code{default} label prevents this
2611 warning.) @code{case} labels outside the enumeration range also
2612 provoke warnings when this option is used.
2613 This warning is enabled by @option{-Wall}.
2615 @item -Wswitch-default
2616 @opindex Wswitch-switch
2617 Warn whenever a @code{switch} statement does not have a @code{default}
2621 @opindex Wswitch-enum
2622 Warn whenever a @code{switch} statement has an index of enumerated type
2623 and lacks a @code{case} for one or more of the named codes of that
2624 enumeration. @code{case} labels outside the enumeration range also
2625 provoke warnings when this option is used.
2629 Warn if any trigraphs are encountered that might change the meaning of
2630 the program (trigraphs within comments are not warned about).
2631 This warning is enabled by @option{-Wall}.
2633 @item -Wunused-function
2634 @opindex Wunused-function
2635 Warn whenever a static function is declared but not defined or a
2636 non-inline static function is unused.
2637 This warning is enabled by @option{-Wall}.
2639 @item -Wunused-label
2640 @opindex Wunused-label
2641 Warn whenever a label is declared but not used.
2642 This warning is enabled by @option{-Wall}.
2644 To suppress this warning use the @samp{unused} attribute
2645 (@pxref{Variable Attributes}).
2647 @item -Wunused-parameter
2648 @opindex Wunused-parameter
2649 Warn whenever a function parameter is unused aside from its declaration.
2651 To suppress this warning use the @samp{unused} attribute
2652 (@pxref{Variable Attributes}).
2654 @item -Wunused-variable
2655 @opindex Wunused-variable
2656 Warn whenever a local variable or non-constant static variable is unused
2657 aside from its declaration
2658 This warning is enabled by @option{-Wall}.
2660 To suppress this warning use the @samp{unused} attribute
2661 (@pxref{Variable Attributes}).
2663 @item -Wunused-value
2664 @opindex Wunused-value
2665 Warn whenever a statement computes a result that is explicitly not used.
2666 This warning is enabled by @option{-Wall}.
2668 To suppress this warning cast the expression to @samp{void}.
2672 All the above @option{-Wunused} options combined.
2674 In order to get a warning about an unused function parameter, you must
2675 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2676 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2678 @item -Wuninitialized
2679 @opindex Wuninitialized
2680 Warn if an automatic variable is used without first being initialized or
2681 if a variable may be clobbered by a @code{setjmp} call.
2683 These warnings are possible only in optimizing compilation,
2684 because they require data flow information that is computed only
2685 when optimizing. If you don't specify @option{-O}, you simply won't
2688 If you want to warn about code which uses the uninitialized value of the
2689 variable in its own initializer, use the @option{-Winit-self} option.
2691 These warnings occur for individual uninitialized or clobbered
2692 elements of structure, union or array variables as well as for
2693 variables which are uninitialized or clobbered as a whole. They do
2694 not occur for variables or elements declared @code{volatile}. Because
2695 these warnings depend on optimization, the exact variables or elements
2696 for which there are warnings will depend on the precise optimization
2697 options and version of GCC used.
2699 Note that there may be no warning about a variable that is used only
2700 to compute a value that itself is never used, because such
2701 computations may be deleted by data flow analysis before the warnings
2704 These warnings are made optional because GCC is not smart
2705 enough to see all the reasons why the code might be correct
2706 despite appearing to have an error. Here is one example of how
2727 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2728 always initialized, but GCC doesn't know this. Here is
2729 another common case:
2734 if (change_y) save_y = y, y = new_y;
2736 if (change_y) y = save_y;
2741 This has no bug because @code{save_y} is used only if it is set.
2743 @cindex @code{longjmp} warnings
2744 This option also warns when a non-volatile automatic variable might be
2745 changed by a call to @code{longjmp}. These warnings as well are possible
2746 only in optimizing compilation.
2748 The compiler sees only the calls to @code{setjmp}. It cannot know
2749 where @code{longjmp} will be called; in fact, a signal handler could
2750 call it at any point in the code. As a result, you may get a warning
2751 even when there is in fact no problem because @code{longjmp} cannot
2752 in fact be called at the place which would cause a problem.
2754 Some spurious warnings can be avoided if you declare all the functions
2755 you use that never return as @code{noreturn}. @xref{Function
2758 This warning is enabled by @option{-Wall}.
2760 @item -Wunknown-pragmas
2761 @opindex Wunknown-pragmas
2762 @cindex warning for unknown pragmas
2763 @cindex unknown pragmas, warning
2764 @cindex pragmas, warning of unknown
2765 Warn when a #pragma directive is encountered which is not understood by
2766 GCC@. If this command line option is used, warnings will even be issued
2767 for unknown pragmas in system header files. This is not the case if
2768 the warnings were only enabled by the @option{-Wall} command line option.
2771 @opindex Wno-pragmas
2773 Do not warn about misuses of pragmas, such as incorrect parameters,
2774 invalid syntax, or conflicts between pragmas. See also
2775 @samp{-Wunknown-pragmas}.
2777 @item -Wstrict-aliasing
2778 @opindex Wstrict-aliasing
2779 This option is only active when @option{-fstrict-aliasing} is active.
2780 It warns about code which might break the strict aliasing rules that the
2781 compiler is using for optimization. The warning does not catch all
2782 cases, but does attempt to catch the more common pitfalls. It is
2783 included in @option{-Wall}.
2785 @item -Wstrict-aliasing=2
2786 @opindex Wstrict-aliasing=2
2787 This option is only active when @option{-fstrict-aliasing} is active.
2788 It warns about code which might break the strict aliasing rules that the
2789 compiler is using for optimization. This warning catches more cases than
2790 @option{-Wstrict-aliasing}, but it will also give a warning for some ambiguous
2791 cases that are safe.
2795 All of the above @samp{-W} options combined. This enables all the
2796 warnings about constructions that some users consider questionable, and
2797 that are easy to avoid (or modify to prevent the warning), even in
2798 conjunction with macros. This also enables some language-specific
2799 warnings described in @ref{C++ Dialect Options} and
2800 @ref{Objective-C and Objective-C++ Dialect Options}.
2803 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2804 Some of them warn about constructions that users generally do not
2805 consider questionable, but which occasionally you might wish to check
2806 for; others warn about constructions that are necessary or hard to avoid
2807 in some cases, and there is no simple way to modify the code to suppress
2814 (This option used to be called @option{-W}. The older name is still
2815 supported, but the newer name is more descriptive.) Print extra warning
2816 messages for these events:
2820 A function can return either with or without a value. (Falling
2821 off the end of the function body is considered returning without
2822 a value.) For example, this function would evoke such a
2836 An expression-statement or the left-hand side of a comma expression
2837 contains no side effects.
2838 To suppress the warning, cast the unused expression to void.
2839 For example, an expression such as @samp{x[i,j]} will cause a warning,
2840 but @samp{x[(void)i,j]} will not.
2843 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2846 Storage-class specifiers like @code{static} are not the first things in
2847 a declaration. According to the C Standard, this usage is obsolescent.
2850 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2854 A comparison between signed and unsigned values could produce an
2855 incorrect result when the signed value is converted to unsigned.
2856 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2859 An aggregate has an initializer which does not initialize all members.
2860 This warning can be independently controlled by
2861 @option{-Wmissing-field-initializers}.
2864 A function parameter is declared without a type specifier in K&R-style
2872 An empty body occurs in an @samp{if} or @samp{else} statement.
2875 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2876 @samp{>}, or @samp{>=}.
2879 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2882 Any of several floating-point events that often indicate errors, such as
2883 overflow, underflow, loss of precision, etc.
2885 @item @r{(C++ only)}
2886 An enumerator and a non-enumerator both appear in a conditional expression.
2888 @item @r{(C++ only)}
2889 A non-static reference or non-static @samp{const} member appears in a
2890 class without constructors.
2892 @item @r{(C++ only)}
2893 Ambiguous virtual bases.
2895 @item @r{(C++ only)}
2896 Subscripting an array which has been declared @samp{register}.
2898 @item @r{(C++ only)}
2899 Taking the address of a variable which has been declared @samp{register}.
2901 @item @r{(C++ only)}
2902 A base class is not initialized in a derived class' copy constructor.
2905 @item -Wno-div-by-zero
2906 @opindex Wno-div-by-zero
2907 @opindex Wdiv-by-zero
2908 Do not warn about compile-time integer division by zero. Floating point
2909 division by zero is not warned about, as it can be a legitimate way of
2910 obtaining infinities and NaNs.
2912 @item -Wsystem-headers
2913 @opindex Wsystem-headers
2914 @cindex warnings from system headers
2915 @cindex system headers, warnings from
2916 Print warning messages for constructs found in system header files.
2917 Warnings from system headers are normally suppressed, on the assumption
2918 that they usually do not indicate real problems and would only make the
2919 compiler output harder to read. Using this command line option tells
2920 GCC to emit warnings from system headers as if they occurred in user
2921 code. However, note that using @option{-Wall} in conjunction with this
2922 option will @emph{not} warn about unknown pragmas in system
2923 headers---for that, @option{-Wunknown-pragmas} must also be used.
2926 @opindex Wfloat-equal
2927 Warn if floating point values are used in equality comparisons.
2929 The idea behind this is that sometimes it is convenient (for the
2930 programmer) to consider floating-point values as approximations to
2931 infinitely precise real numbers. If you are doing this, then you need
2932 to compute (by analyzing the code, or in some other way) the maximum or
2933 likely maximum error that the computation introduces, and allow for it
2934 when performing comparisons (and when producing output, but that's a
2935 different problem). In particular, instead of testing for equality, you
2936 would check to see whether the two values have ranges that overlap; and
2937 this is done with the relational operators, so equality comparisons are
2940 @item -Wtraditional @r{(C only)}
2941 @opindex Wtraditional
2942 Warn about certain constructs that behave differently in traditional and
2943 ISO C@. Also warn about ISO C constructs that have no traditional C
2944 equivalent, and/or problematic constructs which should be avoided.
2948 Macro parameters that appear within string literals in the macro body.
2949 In traditional C macro replacement takes place within string literals,
2950 but does not in ISO C@.
2953 In traditional C, some preprocessor directives did not exist.
2954 Traditional preprocessors would only consider a line to be a directive
2955 if the @samp{#} appeared in column 1 on the line. Therefore
2956 @option{-Wtraditional} warns about directives that traditional C
2957 understands but would ignore because the @samp{#} does not appear as the
2958 first character on the line. It also suggests you hide directives like
2959 @samp{#pragma} not understood by traditional C by indenting them. Some
2960 traditional implementations would not recognize @samp{#elif}, so it
2961 suggests avoiding it altogether.
2964 A function-like macro that appears without arguments.
2967 The unary plus operator.
2970 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2971 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2972 constants.) Note, these suffixes appear in macros defined in the system
2973 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2974 Use of these macros in user code might normally lead to spurious
2975 warnings, however GCC's integrated preprocessor has enough context to
2976 avoid warning in these cases.
2979 A function declared external in one block and then used after the end of
2983 A @code{switch} statement has an operand of type @code{long}.
2986 A non-@code{static} function declaration follows a @code{static} one.
2987 This construct is not accepted by some traditional C compilers.
2990 The ISO type of an integer constant has a different width or
2991 signedness from its traditional type. This warning is only issued if
2992 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2993 typically represent bit patterns, are not warned about.
2996 Usage of ISO string concatenation is detected.
2999 Initialization of automatic aggregates.
3002 Identifier conflicts with labels. Traditional C lacks a separate
3003 namespace for labels.
3006 Initialization of unions. If the initializer is zero, the warning is
3007 omitted. This is done under the assumption that the zero initializer in
3008 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3009 initializer warnings and relies on default initialization to zero in the
3013 Conversions by prototypes between fixed/floating point values and vice
3014 versa. The absence of these prototypes when compiling with traditional
3015 C would cause serious problems. This is a subset of the possible
3016 conversion warnings, for the full set use @option{-Wconversion}.
3019 Use of ISO C style function definitions. This warning intentionally is
3020 @emph{not} issued for prototype declarations or variadic functions
3021 because these ISO C features will appear in your code when using
3022 libiberty's traditional C compatibility macros, @code{PARAMS} and
3023 @code{VPARAMS}. This warning is also bypassed for nested functions
3024 because that feature is already a GCC extension and thus not relevant to
3025 traditional C compatibility.
3028 @item -Wdeclaration-after-statement @r{(C only)}
3029 @opindex Wdeclaration-after-statement
3030 Warn when a declaration is found after a statement in a block. This
3031 construct, known from C++, was introduced with ISO C99 and is by default
3032 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3033 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3037 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3039 @item -Wno-endif-labels
3040 @opindex Wno-endif-labels
3041 @opindex Wendif-labels
3042 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3046 Warn whenever a local variable shadows another local variable, parameter or
3047 global variable or whenever a built-in function is shadowed.
3049 @item -Wlarger-than-@var{len}
3050 @opindex Wlarger-than
3051 Warn whenever an object of larger than @var{len} bytes is defined.
3053 @item -Wunsafe-loop-optimizations
3054 @opindex Wunsafe-loop-optimizations
3055 Warn if the loop cannot be optimized because the compiler could not
3056 assume anything on the bounds of the loop indices. With
3057 @option{-funsafe-loop-optimizations} warn if the compiler made
3060 @item -Wpointer-arith
3061 @opindex Wpointer-arith
3062 Warn about anything that depends on the ``size of'' a function type or
3063 of @code{void}. GNU C assigns these types a size of 1, for
3064 convenience in calculations with @code{void *} pointers and pointers
3067 @item -Wbad-function-cast @r{(C only)}
3068 @opindex Wbad-function-cast
3069 Warn whenever a function call is cast to a non-matching type.
3070 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3073 Warn about ISO C constructs that are outside of the common subset of
3074 ISO C and ISO C++, e.g.@: request for implicit conversion from
3075 @code{void *} to a pointer to non-@code{void} type.
3079 Warn whenever a pointer is cast so as to remove a type qualifier from
3080 the target type. For example, warn if a @code{const char *} is cast
3081 to an ordinary @code{char *}.
3084 @opindex Wcast-align
3085 Warn whenever a pointer is cast such that the required alignment of the
3086 target is increased. For example, warn if a @code{char *} is cast to
3087 an @code{int *} on machines where integers can only be accessed at
3088 two- or four-byte boundaries.
3090 @item -Wwrite-strings
3091 @opindex Wwrite-strings
3092 When compiling C, give string constants the type @code{const
3093 char[@var{length}]} so that
3094 copying the address of one into a non-@code{const} @code{char *}
3095 pointer will get a warning; when compiling C++, warn about the
3096 deprecated conversion from string literals to @code{char *}. This
3097 warning, by default, is enabled for C++ programs.
3098 These warnings will help you find at
3099 compile time code that can try to write into a string constant, but
3100 only if you have been very careful about using @code{const} in
3101 declarations and prototypes. Otherwise, it will just be a nuisance;
3102 this is why we did not make @option{-Wall} request these warnings.
3105 @opindex Wconversion
3106 Warn if a prototype causes a type conversion that is different from what
3107 would happen to the same argument in the absence of a prototype. This
3108 includes conversions of fixed point to floating and vice versa, and
3109 conversions changing the width or signedness of a fixed point argument
3110 except when the same as the default promotion.
3112 Also, warn if a negative integer constant expression is implicitly
3113 converted to an unsigned type. For example, warn about the assignment
3114 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
3115 casts like @code{(unsigned) -1}.
3117 @item -Wsign-compare
3118 @opindex Wsign-compare
3119 @cindex warning for comparison of signed and unsigned values
3120 @cindex comparison of signed and unsigned values, warning
3121 @cindex signed and unsigned values, comparison warning
3122 Warn when a comparison between signed and unsigned values could produce
3123 an incorrect result when the signed value is converted to unsigned.
3124 This warning is also enabled by @option{-Wextra}; to get the other warnings
3125 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3127 @item -Waggregate-return
3128 @opindex Waggregate-return
3129 Warn if any functions that return structures or unions are defined or
3130 called. (In languages where you can return an array, this also elicits
3134 @opindex Walways-true
3135 Warn about comparisons which are always true such as testing if
3136 unsigned values are greater than or equal to zero. This warning is
3137 enabled by @option{-Wall}.
3139 @item -Wno-attributes
3140 @opindex Wno-attributes
3141 @opindex Wattributes
3142 Do not warn if an unexpected @code{__attribute__} is used, such as
3143 unrecognized attributes, function attributes applied to variables,
3144 etc. This will not stop errors for incorrect use of supported
3147 @item -Wstrict-prototypes @r{(C only)}
3148 @opindex Wstrict-prototypes
3149 Warn if a function is declared or defined without specifying the
3150 argument types. (An old-style function definition is permitted without
3151 a warning if preceded by a declaration which specifies the argument
3154 @item -Wold-style-definition @r{(C only)}
3155 @opindex Wold-style-definition
3156 Warn if an old-style function definition is used. A warning is given
3157 even if there is a previous prototype.
3159 @item -Wmissing-prototypes @r{(C only)}
3160 @opindex Wmissing-prototypes
3161 Warn if a global function is defined without a previous prototype
3162 declaration. This warning is issued even if the definition itself
3163 provides a prototype. The aim is to detect global functions that fail
3164 to be declared in header files.
3166 @item -Wmissing-declarations @r{(C only)}
3167 @opindex Wmissing-declarations
3168 Warn if a global function is defined without a previous declaration.
3169 Do so even if the definition itself provides a prototype.
3170 Use this option to detect global functions that are not declared in
3173 @item -Wmissing-field-initializers
3174 @opindex Wmissing-field-initializers
3177 Warn if a structure's initializer has some fields missing. For
3178 example, the following code would cause such a warning, because
3179 @code{x.h} is implicitly zero:
3182 struct s @{ int f, g, h; @};
3183 struct s x = @{ 3, 4 @};
3186 This option does not warn about designated initializers, so the following
3187 modification would not trigger a warning:
3190 struct s @{ int f, g, h; @};
3191 struct s x = @{ .f = 3, .g = 4 @};
3194 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3195 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3197 @item -Wmissing-noreturn
3198 @opindex Wmissing-noreturn
3199 Warn about functions which might be candidates for attribute @code{noreturn}.
3200 Note these are only possible candidates, not absolute ones. Care should
3201 be taken to manually verify functions actually do not ever return before
3202 adding the @code{noreturn} attribute, otherwise subtle code generation
3203 bugs could be introduced. You will not get a warning for @code{main} in
3204 hosted C environments.
3206 @item -Wmissing-format-attribute
3207 @opindex Wmissing-format-attribute
3209 Warn about function pointers which might be candidates for @code{format}
3210 attributes. Note these are only possible candidates, not absolute ones.
3211 GCC will guess that function pointers with @code{format} attributes that
3212 are used in assignment, initialization, parameter passing or return
3213 statements should have a corresponding @code{format} attribute in the
3214 resulting type. I.e.@: the left-hand side of the assignment or
3215 initialization, the type of the parameter variable, or the return type
3216 of the containing function respectively should also have a @code{format}
3217 attribute to avoid the warning.
3219 GCC will also warn about function definitions which might be
3220 candidates for @code{format} attributes. Again, these are only
3221 possible candidates. GCC will guess that @code{format} attributes
3222 might be appropriate for any function that calls a function like
3223 @code{vprintf} or @code{vscanf}, but this might not always be the
3224 case, and some functions for which @code{format} attributes are
3225 appropriate may not be detected.
3227 @item -Wno-multichar
3228 @opindex Wno-multichar
3230 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3231 Usually they indicate a typo in the user's code, as they have
3232 implementation-defined values, and should not be used in portable code.
3234 @item -Wnormalized=<none|id|nfc|nfkc>
3235 @opindex Wnormalized
3238 @cindex character set, input normalization
3239 In ISO C and ISO C++, two identifiers are different if they are
3240 different sequences of characters. However, sometimes when characters
3241 outside the basic ASCII character set are used, you can have two
3242 different character sequences that look the same. To avoid confusion,
3243 the ISO 10646 standard sets out some @dfn{normalization rules} which
3244 when applied ensure that two sequences that look the same are turned into
3245 the same sequence. GCC can warn you if you are using identifiers which
3246 have not been normalized; this option controls that warning.
3248 There are four levels of warning that GCC supports. The default is
3249 @option{-Wnormalized=nfc}, which warns about any identifier which is
3250 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3251 recommended form for most uses.
3253 Unfortunately, there are some characters which ISO C and ISO C++ allow
3254 in identifiers that when turned into NFC aren't allowable as
3255 identifiers. That is, there's no way to use these symbols in portable
3256 ISO C or C++ and have all your identifiers in NFC.
3257 @option{-Wnormalized=id} suppresses the warning for these characters.
3258 It is hoped that future versions of the standards involved will correct
3259 this, which is why this option is not the default.
3261 You can switch the warning off for all characters by writing
3262 @option{-Wnormalized=none}. You would only want to do this if you
3263 were using some other normalization scheme (like ``D''), because
3264 otherwise you can easily create bugs that are literally impossible to see.
3266 Some characters in ISO 10646 have distinct meanings but look identical
3267 in some fonts or display methodologies, especially once formatting has
3268 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3269 LETTER N'', will display just like a regular @code{n} which has been
3270 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3271 normalisation scheme to convert all these into a standard form as
3272 well, and GCC will warn if your code is not in NFKC if you use
3273 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3274 about every identifier that contains the letter O because it might be
3275 confused with the digit 0, and so is not the default, but may be
3276 useful as a local coding convention if the programming environment is
3277 unable to be fixed to display these characters distinctly.
3279 @item -Wno-deprecated-declarations
3280 @opindex Wno-deprecated-declarations
3281 Do not warn about uses of functions, variables, and types marked as
3282 deprecated by using the @code{deprecated} attribute.
3283 (@pxref{Function Attributes}, @pxref{Variable Attributes},
3284 @pxref{Type Attributes}.)
3288 Warn if a structure is given the packed attribute, but the packed
3289 attribute has no effect on the layout or size of the structure.
3290 Such structures may be mis-aligned for little benefit. For
3291 instance, in this code, the variable @code{f.x} in @code{struct bar}
3292 will be misaligned even though @code{struct bar} does not itself
3293 have the packed attribute:
3300 @} __attribute__((packed));
3310 Warn if padding is included in a structure, either to align an element
3311 of the structure or to align the whole structure. Sometimes when this
3312 happens it is possible to rearrange the fields of the structure to
3313 reduce the padding and so make the structure smaller.
3315 @item -Wredundant-decls
3316 @opindex Wredundant-decls
3317 Warn if anything is declared more than once in the same scope, even in
3318 cases where multiple declaration is valid and changes nothing.
3320 @item -Wnested-externs @r{(C only)}
3321 @opindex Wnested-externs
3322 Warn if an @code{extern} declaration is encountered within a function.
3324 @item -Wunreachable-code
3325 @opindex Wunreachable-code
3326 Warn if the compiler detects that code will never be executed.
3328 This option is intended to warn when the compiler detects that at
3329 least a whole line of source code will never be executed, because
3330 some condition is never satisfied or because it is after a
3331 procedure that never returns.
3333 It is possible for this option to produce a warning even though there
3334 are circumstances under which part of the affected line can be executed,
3335 so care should be taken when removing apparently-unreachable code.
3337 For instance, when a function is inlined, a warning may mean that the
3338 line is unreachable in only one inlined copy of the function.
3340 This option is not made part of @option{-Wall} because in a debugging
3341 version of a program there is often substantial code which checks
3342 correct functioning of the program and is, hopefully, unreachable
3343 because the program does work. Another common use of unreachable
3344 code is to provide behavior which is selectable at compile-time.
3348 Warn if a function can not be inlined and it was declared as inline.
3349 Even with this option, the compiler will not warn about failures to
3350 inline functions declared in system headers.
3352 The compiler uses a variety of heuristics to determine whether or not
3353 to inline a function. For example, the compiler takes into account
3354 the size of the function being inlined and the amount of inlining
3355 that has already been done in the current function. Therefore,
3356 seemingly insignificant changes in the source program can cause the
3357 warnings produced by @option{-Winline} to appear or disappear.
3359 @item -Wno-invalid-offsetof @r{(C++ only)}
3360 @opindex Wno-invalid-offsetof
3361 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3362 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3363 to a non-POD type is undefined. In existing C++ implementations,
3364 however, @samp{offsetof} typically gives meaningful results even when
3365 applied to certain kinds of non-POD types. (Such as a simple
3366 @samp{struct} that fails to be a POD type only by virtue of having a
3367 constructor.) This flag is for users who are aware that they are
3368 writing nonportable code and who have deliberately chosen to ignore the
3371 The restrictions on @samp{offsetof} may be relaxed in a future version
3372 of the C++ standard.
3374 @item -Wno-int-to-pointer-cast @r{(C only)}
3375 @opindex Wno-int-to-pointer-cast
3376 Suppress warnings from casts to pointer type of an integer of a
3379 @item -Wno-pointer-to-int-cast @r{(C only)}
3380 @opindex Wno-pointer-to-int-cast
3381 Suppress warnings from casts from a pointer to an integer type of a
3385 @opindex Winvalid-pch
3386 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3387 the search path but can't be used.
3391 @opindex Wno-long-long
3392 Warn if @samp{long long} type is used. This is default. To inhibit
3393 the warning messages, use @option{-Wno-long-long}. Flags
3394 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3395 only when @option{-pedantic} flag is used.
3397 @item -Wvariadic-macros
3398 @opindex Wvariadic-macros
3399 @opindex Wno-variadic-macros
3400 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3401 alternate syntax when in pedantic ISO C99 mode. This is default.
3402 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3404 @item -Wvolatile-register-var
3405 @opindex Wvolatile-register-var
3406 @opindex Wno-volatile-register-var
3407 Warn if a register variable is declared volatile. The volatile
3408 modifier does not inhibit all optimizations that may eliminate reads
3409 and/or writes to register variables.
3411 @item -Wdisabled-optimization
3412 @opindex Wdisabled-optimization
3413 Warn if a requested optimization pass is disabled. This warning does
3414 not generally indicate that there is anything wrong with your code; it
3415 merely indicates that GCC's optimizers were unable to handle the code
3416 effectively. Often, the problem is that your code is too big or too
3417 complex; GCC will refuse to optimize programs when the optimization
3418 itself is likely to take inordinate amounts of time.
3420 @item -Wpointer-sign
3421 @opindex Wpointer-sign
3422 @opindex Wno-pointer-sign
3423 Warn for pointer argument passing or assignment with different signedness.
3424 This option is only supported for C and Objective-C@. It is implied by
3425 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3426 @option{-Wno-pointer-sign}.
3430 Make all warnings into errors.
3434 Make the specified warning into an errors. The specifier for a
3435 warning is appended, for example @option{-Werror=switch} turns the
3436 warnings controlled by @option{-Wswitch} into errors. This switch
3437 takes a negative form, to be used to negate @option{-Werror} for
3438 specific warnings, for example @option{-Wno-error=switch} makes
3439 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3440 is in effect. You can use the @option{-fdiagnostics-show-option}
3441 option to have each controllable warning amended with the option which
3442 controls it, to determine what to use with this option.
3444 Note that specifying @option{-Werror=}@var{foo} automatically implies
3445 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3448 @item -Wstack-protector
3449 @opindex Wstack-protector
3450 This option is only active when @option{-fstack-protector} is active. It
3451 warns about functions that will not be protected against stack smashing.
3453 @item -Wstring-literal-comparison
3454 @opindex Wstring-literal-comparison
3455 Warn about suspicious comparisons to string literal constants. In C,
3456 direct comparisons against the memory address of a string literal, such
3457 as @code{if (x == "abc")}, typically indicate a programmer error, and
3458 even when intentional, result in unspecified behavior and are not portable.
3459 Usually these warnings alert that the programmer intended to use
3460 @code{strcmp}. This warning is enabled by @option{-Wall}.
3462 @item -Woverlength-strings
3463 @opindex Woverlength-strings
3464 Warn about string constants which are longer than the ``minimum
3465 maximum'' length specified in the C standard. Modern compilers
3466 generally allow string constants which are much longer than the
3467 standard's minimum limit, but very portable programs should avoid
3468 using longer strings.
3470 The limit applies @emph{after} string constant concatenation, and does
3471 not count the trailing NUL@. In C89, the limit was 509 characters; in
3472 C99, it was raised to 4095. C++98 does not specify a normative
3473 minimum maximum, so we do not diagnose overlength strings in C++@.
3475 This option is implied by @option{-pedantic}, and can be disabled with
3476 @option{-Wno-overlength-strings}.
3479 @node Debugging Options
3480 @section Options for Debugging Your Program or GCC
3481 @cindex options, debugging
3482 @cindex debugging information options
3484 GCC has various special options that are used for debugging
3485 either your program or GCC:
3490 Produce debugging information in the operating system's native format
3491 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3494 On most systems that use stabs format, @option{-g} enables use of extra
3495 debugging information that only GDB can use; this extra information
3496 makes debugging work better in GDB but will probably make other debuggers
3498 refuse to read the program. If you want to control for certain whether
3499 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3500 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3502 GCC allows you to use @option{-g} with
3503 @option{-O}. The shortcuts taken by optimized code may occasionally
3504 produce surprising results: some variables you declared may not exist
3505 at all; flow of control may briefly move where you did not expect it;
3506 some statements may not be executed because they compute constant
3507 results or their values were already at hand; some statements may
3508 execute in different places because they were moved out of loops.
3510 Nevertheless it proves possible to debug optimized output. This makes
3511 it reasonable to use the optimizer for programs that might have bugs.
3513 The following options are useful when GCC is generated with the
3514 capability for more than one debugging format.
3518 Produce debugging information for use by GDB@. This means to use the
3519 most expressive format available (DWARF 2, stabs, or the native format
3520 if neither of those are supported), including GDB extensions if at all
3525 Produce debugging information in stabs format (if that is supported),
3526 without GDB extensions. This is the format used by DBX on most BSD
3527 systems. On MIPS, Alpha and System V Release 4 systems this option
3528 produces stabs debugging output which is not understood by DBX or SDB@.
3529 On System V Release 4 systems this option requires the GNU assembler.
3531 @item -feliminate-unused-debug-symbols
3532 @opindex feliminate-unused-debug-symbols
3533 Produce debugging information in stabs format (if that is supported),
3534 for only symbols that are actually used.
3536 @item -femit-class-debug-always
3537 Instead of emitting debugging information for a C++ class in only one
3538 object file, emit it in all object files using the class. This option
3539 should be used only with debuggers that are unable to handle the way GCC
3540 normally emits debugging information for classes because using this
3541 option will increase the size of debugging information by as much as a
3546 Produce debugging information in stabs format (if that is supported),
3547 using GNU extensions understood only by the GNU debugger (GDB)@. The
3548 use of these extensions is likely to make other debuggers crash or
3549 refuse to read the program.
3553 Produce debugging information in COFF format (if that is supported).
3554 This is the format used by SDB on most System V systems prior to
3559 Produce debugging information in XCOFF format (if that is supported).
3560 This is the format used by the DBX debugger on IBM RS/6000 systems.
3564 Produce debugging information in XCOFF format (if that is supported),
3565 using GNU extensions understood only by the GNU debugger (GDB)@. The
3566 use of these extensions is likely to make other debuggers crash or
3567 refuse to read the program, and may cause assemblers other than the GNU
3568 assembler (GAS) to fail with an error.
3572 Produce debugging information in DWARF version 2 format (if that is
3573 supported). This is the format used by DBX on IRIX 6. With this
3574 option, GCC uses features of DWARF version 3 when they are useful;
3575 version 3 is upward compatible with version 2, but may still cause
3576 problems for older debuggers.
3580 Produce debugging information in VMS debug format (if that is
3581 supported). This is the format used by DEBUG on VMS systems.
3584 @itemx -ggdb@var{level}
3585 @itemx -gstabs@var{level}
3586 @itemx -gcoff@var{level}
3587 @itemx -gxcoff@var{level}
3588 @itemx -gvms@var{level}
3589 Request debugging information and also use @var{level} to specify how
3590 much information. The default level is 2.
3592 Level 1 produces minimal information, enough for making backtraces in
3593 parts of the program that you don't plan to debug. This includes
3594 descriptions of functions and external variables, but no information
3595 about local variables and no line numbers.
3597 Level 3 includes extra information, such as all the macro definitions
3598 present in the program. Some debuggers support macro expansion when
3599 you use @option{-g3}.
3601 @option{-gdwarf-2} does not accept a concatenated debug level, because
3602 GCC used to support an option @option{-gdwarf} that meant to generate
3603 debug information in version 1 of the DWARF format (which is very
3604 different from version 2), and it would have been too confusing. That
3605 debug format is long obsolete, but the option cannot be changed now.
3606 Instead use an additional @option{-g@var{level}} option to change the
3607 debug level for DWARF2.
3609 @item -feliminate-dwarf2-dups
3610 @opindex feliminate-dwarf2-dups
3611 Compress DWARF2 debugging information by eliminating duplicated
3612 information about each symbol. This option only makes sense when
3613 generating DWARF2 debugging information with @option{-gdwarf-2}.
3615 @cindex @command{prof}
3618 Generate extra code to write profile information suitable for the
3619 analysis program @command{prof}. You must use this option when compiling
3620 the source files you want data about, and you must also use it when
3623 @cindex @command{gprof}
3626 Generate extra code to write profile information suitable for the
3627 analysis program @command{gprof}. You must use this option when compiling
3628 the source files you want data about, and you must also use it when
3633 Makes the compiler print out each function name as it is compiled, and
3634 print some statistics about each pass when it finishes.
3637 @opindex ftime-report
3638 Makes the compiler print some statistics about the time consumed by each
3639 pass when it finishes.
3642 @opindex fmem-report
3643 Makes the compiler print some statistics about permanent memory
3644 allocation when it finishes.
3646 @item -fprofile-arcs
3647 @opindex fprofile-arcs
3648 Add code so that program flow @dfn{arcs} are instrumented. During
3649 execution the program records how many times each branch and call is
3650 executed and how many times it is taken or returns. When the compiled
3651 program exits it saves this data to a file called
3652 @file{@var{auxname}.gcda} for each source file. The data may be used for
3653 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3654 test coverage analysis (@option{-ftest-coverage}). Each object file's
3655 @var{auxname} is generated from the name of the output file, if
3656 explicitly specified and it is not the final executable, otherwise it is
3657 the basename of the source file. In both cases any suffix is removed
3658 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3659 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3660 @xref{Cross-profiling}.
3662 @cindex @command{gcov}
3666 This option is used to compile and link code instrumented for coverage
3667 analysis. The option is a synonym for @option{-fprofile-arcs}
3668 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
3669 linking). See the documentation for those options for more details.
3674 Compile the source files with @option{-fprofile-arcs} plus optimization
3675 and code generation options. For test coverage analysis, use the
3676 additional @option{-ftest-coverage} option. You do not need to profile
3677 every source file in a program.
3680 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3681 (the latter implies the former).
3684 Run the program on a representative workload to generate the arc profile
3685 information. This may be repeated any number of times. You can run
3686 concurrent instances of your program, and provided that the file system
3687 supports locking, the data files will be correctly updated. Also
3688 @code{fork} calls are detected and correctly handled (double counting
3692 For profile-directed optimizations, compile the source files again with
3693 the same optimization and code generation options plus
3694 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3695 Control Optimization}).
3698 For test coverage analysis, use @command{gcov} to produce human readable
3699 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3700 @command{gcov} documentation for further information.
3704 With @option{-fprofile-arcs}, for each function of your program GCC
3705 creates a program flow graph, then finds a spanning tree for the graph.
3706 Only arcs that are not on the spanning tree have to be instrumented: the
3707 compiler adds code to count the number of times that these arcs are
3708 executed. When an arc is the only exit or only entrance to a block, the
3709 instrumentation code can be added to the block; otherwise, a new basic
3710 block must be created to hold the instrumentation code.
3713 @item -ftest-coverage
3714 @opindex ftest-coverage
3715 Produce a notes file that the @command{gcov} code-coverage utility
3716 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3717 show program coverage. Each source file's note file is called
3718 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3719 above for a description of @var{auxname} and instructions on how to
3720 generate test coverage data. Coverage data will match the source files
3721 more closely, if you do not optimize.
3723 @item -d@var{letters}
3724 @item -fdump-rtl-@var{pass}
3726 Says to make debugging dumps during compilation at times specified by
3727 @var{letters}. This is used for debugging the RTL-based passes of the
3728 compiler. The file names for most of the dumps are made by appending a
3729 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3730 from the name of the output file, if explicitly specified and it is not
3731 an executable, otherwise it is the basename of the source file.
3733 Most debug dumps can be enabled either passing a letter to the @option{-d}
3734 option, or with a long @option{-fdump-rtl} switch; here are the possible
3735 letters for use in @var{letters} and @var{pass}, and their meanings:
3740 Annotate the assembler output with miscellaneous debugging information.
3743 @itemx -fdump-rtl-bbro
3745 @opindex fdump-rtl-bbro
3746 Dump after block reordering, to @file{@var{file}.148r.bbro}.
3749 @itemx -fdump-rtl-combine
3751 @opindex fdump-rtl-combine
3752 Dump after instruction combination, to the file @file{@var{file}.129r.combine}.
3755 @itemx -fdump-rtl-ce1
3756 @itemx -fdump-rtl-ce2
3758 @opindex fdump-rtl-ce1
3759 @opindex fdump-rtl-ce2
3760 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3761 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
3762 and @option{-fdump-rtl-ce2} enable dumping after the second if
3763 conversion, to the file @file{@var{file}.130r.ce2}.
3766 @itemx -fdump-rtl-btl
3767 @itemx -fdump-rtl-dbr
3769 @opindex fdump-rtl-btl
3770 @opindex fdump-rtl-dbr
3771 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3772 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
3773 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3774 scheduling, to @file{@var{file}.36.dbr}.
3778 Dump all macro definitions, at the end of preprocessing, in addition to
3782 @itemx -fdump-rtl-ce3
3784 @opindex fdump-rtl-ce3
3785 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
3788 @itemx -fdump-rtl-cfg
3789 @itemx -fdump-rtl-life
3791 @opindex fdump-rtl-cfg
3792 @opindex fdump-rtl-life
3793 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3794 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
3795 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3796 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
3799 @itemx -fdump-rtl-greg
3801 @opindex fdump-rtl-greg
3802 Dump after global register allocation, to @file{@var{file}.139r.greg}.
3805 @itemx -fdump-rtl-gcse
3806 @itemx -fdump-rtl-bypass
3808 @opindex fdump-rtl-gcse
3809 @opindex fdump-rtl-bypass
3810 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3811 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3812 enable dumping after jump bypassing and control flow optimizations, to
3813 @file{@var{file}.115r.bypass}.
3816 @itemx -fdump-rtl-eh
3818 @opindex fdump-rtl-eh
3819 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3822 @itemx -fdump-rtl-sibling
3824 @opindex fdump-rtl-sibling
3825 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
3828 @itemx -fdump-rtl-jump
3830 @opindex fdump-rtl-jump
3831 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
3834 @itemx -fdump-rtl-stack
3836 @opindex fdump-rtl-stack
3837 Dump after conversion from registers to stack, to @file{@var{file}.152r.stack}.
3840 @itemx -fdump-rtl-lreg
3842 @opindex fdump-rtl-lreg
3843 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
3846 @itemx -fdump-rtl-loop2
3848 @opindex fdump-rtl-loop2
3849 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
3850 loop optimization pass, to @file{@var{file}.119r.loop2},
3851 @file{@var{file}.120r.loop2_init},
3852 @file{@var{file}.121r.loop2_invariant}, and
3853 @file{@var{file}.125r.loop2_done}.
3856 @itemx -fdump-rtl-sms
3858 @opindex fdump-rtl-sms
3859 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
3862 @itemx -fdump-rtl-mach
3864 @opindex fdump-rtl-mach
3865 Dump after performing the machine dependent reorganization pass, to
3866 @file{@var{file}.155r.mach}.
3869 @itemx -fdump-rtl-rnreg
3871 @opindex fdump-rtl-rnreg
3872 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
3875 @itemx -fdump-rtl-regmove
3877 @opindex fdump-rtl-regmove
3878 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
3881 @itemx -fdump-rtl-postreload
3883 @opindex fdump-rtl-postreload
3884 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
3887 @itemx -fdump-rtl-expand
3889 @opindex fdump-rtl-expand
3890 Dump after RTL generation, to @file{@var{file}.104r.expand}.
3893 @itemx -fdump-rtl-sched2
3895 @opindex fdump-rtl-sched2
3896 Dump after the second scheduling pass, to @file{@var{file}.150r.sched2}.
3899 @itemx -fdump-rtl-cse
3901 @opindex fdump-rtl-cse
3902 Dump after CSE (including the jump optimization that sometimes follows
3903 CSE), to @file{@var{file}.113r.cse}.
3906 @itemx -fdump-rtl-sched
3908 @opindex fdump-rtl-sched
3909 Dump after the first scheduling pass, to @file{@var{file}.21.sched}.
3912 @itemx -fdump-rtl-cse2
3914 @opindex fdump-rtl-cse2
3915 Dump after the second CSE pass (including the jump optimization that
3916 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
3919 @itemx -fdump-rtl-tracer
3921 @opindex fdump-rtl-tracer
3922 Dump after running tracer, to @file{@var{file}.118r.tracer}.
3925 @itemx -fdump-rtl-vpt
3926 @itemx -fdump-rtl-vartrack
3928 @opindex fdump-rtl-vpt
3929 @opindex fdump-rtl-vartrack
3930 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
3931 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
3932 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
3933 to @file{@var{file}.154r.vartrack}.
3936 @itemx -fdump-rtl-flow2
3938 @opindex fdump-rtl-flow2
3939 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
3942 @itemx -fdump-rtl-peephole2
3944 @opindex fdump-rtl-peephole2
3945 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
3948 @itemx -fdump-rtl-web
3950 @opindex fdump-rtl-web
3951 Dump after live range splitting, to @file{@var{file}.126r.web}.
3954 @itemx -fdump-rtl-all
3956 @opindex fdump-rtl-all
3957 Produce all the dumps listed above.
3961 Produce a core dump whenever an error occurs.
3965 Print statistics on memory usage, at the end of the run, to
3970 Annotate the assembler output with a comment indicating which
3971 pattern and alternative was used. The length of each instruction is
3976 Dump the RTL in the assembler output as a comment before each instruction.
3977 Also turns on @option{-dp} annotation.
3981 For each of the other indicated dump files (either with @option{-d} or
3982 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
3983 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3987 Just generate RTL for a function instead of compiling it. Usually used
3988 with @samp{r} (@option{-fdump-rtl-expand}).
3992 Dump debugging information during parsing, to standard error.
3995 @item -fdump-unnumbered
3996 @opindex fdump-unnumbered
3997 When doing debugging dumps (see @option{-d} option above), suppress instruction
3998 numbers and line number note output. This makes it more feasible to
3999 use diff on debugging dumps for compiler invocations with different
4000 options, in particular with and without @option{-g}.
4002 @item -fdump-translation-unit @r{(C++ only)}
4003 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4004 @opindex fdump-translation-unit
4005 Dump a representation of the tree structure for the entire translation
4006 unit to a file. The file name is made by appending @file{.tu} to the
4007 source file name. If the @samp{-@var{options}} form is used, @var{options}
4008 controls the details of the dump as described for the
4009 @option{-fdump-tree} options.
4011 @item -fdump-class-hierarchy @r{(C++ only)}
4012 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4013 @opindex fdump-class-hierarchy
4014 Dump a representation of each class's hierarchy and virtual function
4015 table layout to a file. The file name is made by appending @file{.class}
4016 to the source file name. If the @samp{-@var{options}} form is used,
4017 @var{options} controls the details of the dump as described for the
4018 @option{-fdump-tree} options.
4020 @item -fdump-ipa-@var{switch}
4022 Control the dumping at various stages of inter-procedural analysis
4023 language tree to a file. The file name is generated by appending a switch
4024 specific suffix to the source file name. The following dumps are possible:
4028 Enables all inter-procedural analysis dumps; currently the only produced
4029 dump is the @samp{cgraph} dump.
4032 Dumps information about call-graph optimization, unused function removal,
4033 and inlining decisions.
4036 @item -fdump-tree-@var{switch}
4037 @itemx -fdump-tree-@var{switch}-@var{options}
4039 Control the dumping at various stages of processing the intermediate
4040 language tree to a file. The file name is generated by appending a switch
4041 specific suffix to the source file name. If the @samp{-@var{options}}
4042 form is used, @var{options} is a list of @samp{-} separated options that
4043 control the details of the dump. Not all options are applicable to all
4044 dumps, those which are not meaningful will be ignored. The following
4045 options are available
4049 Print the address of each node. Usually this is not meaningful as it
4050 changes according to the environment and source file. Its primary use
4051 is for tying up a dump file with a debug environment.
4053 Inhibit dumping of members of a scope or body of a function merely
4054 because that scope has been reached. Only dump such items when they
4055 are directly reachable by some other path. When dumping pretty-printed
4056 trees, this option inhibits dumping the bodies of control structures.
4058 Print a raw representation of the tree. By default, trees are
4059 pretty-printed into a C-like representation.
4061 Enable more detailed dumps (not honored by every dump option).
4063 Enable dumping various statistics about the pass (not honored by every dump
4066 Enable showing basic block boundaries (disabled in raw dumps).
4068 Enable showing virtual operands for every statement.
4070 Enable showing line numbers for statements.
4072 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4074 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4077 The following tree dumps are possible:
4081 Dump before any tree based optimization, to @file{@var{file}.original}.
4084 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4087 Dump after function inlining, to @file{@var{file}.inlined}.
4090 @opindex fdump-tree-gimple
4091 Dump each function before and after the gimplification pass to a file. The
4092 file name is made by appending @file{.gimple} to the source file name.
4095 @opindex fdump-tree-cfg
4096 Dump the control flow graph of each function to a file. The file name is
4097 made by appending @file{.cfg} to the source file name.
4100 @opindex fdump-tree-vcg
4101 Dump the control flow graph of each function to a file in VCG format. The
4102 file name is made by appending @file{.vcg} to the source file name. Note
4103 that if the file contains more than one function, the generated file cannot
4104 be used directly by VCG@. You will need to cut and paste each function's
4105 graph into its own separate file first.
4108 @opindex fdump-tree-ch
4109 Dump each function after copying loop headers. The file name is made by
4110 appending @file{.ch} to the source file name.
4113 @opindex fdump-tree-ssa
4114 Dump SSA related information to a file. The file name is made by appending
4115 @file{.ssa} to the source file name.
4118 @opindex fdump-tree-salias
4119 Dump structure aliasing variable information to a file. This file name
4120 is made by appending @file{.salias} to the source file name.
4123 @opindex fdump-tree-alias
4124 Dump aliasing information for each function. The file name is made by
4125 appending @file{.alias} to the source file name.
4128 @opindex fdump-tree-ccp
4129 Dump each function after CCP@. The file name is made by appending
4130 @file{.ccp} to the source file name.
4133 @opindex fdump-tree-storeccp
4134 Dump each function after STORE-CCP. The file name is made by appending
4135 @file{.storeccp} to the source file name.
4138 @opindex fdump-tree-pre
4139 Dump trees after partial redundancy elimination. The file name is made
4140 by appending @file{.pre} to the source file name.
4143 @opindex fdump-tree-fre
4144 Dump trees after full redundancy elimination. The file name is made
4145 by appending @file{.fre} to the source file name.
4148 @opindex fdump-tree-copyprop
4149 Dump trees after copy propagation. The file name is made
4150 by appending @file{.copyprop} to the source file name.
4152 @item store_copyprop
4153 @opindex fdump-tree-store_copyprop
4154 Dump trees after store copy-propagation. The file name is made
4155 by appending @file{.store_copyprop} to the source file name.
4158 @opindex fdump-tree-dce
4159 Dump each function after dead code elimination. The file name is made by
4160 appending @file{.dce} to the source file name.
4163 @opindex fdump-tree-mudflap
4164 Dump each function after adding mudflap instrumentation. The file name is
4165 made by appending @file{.mudflap} to the source file name.
4168 @opindex fdump-tree-sra
4169 Dump each function after performing scalar replacement of aggregates. The
4170 file name is made by appending @file{.sra} to the source file name.
4173 @opindex fdump-tree-sink
4174 Dump each function after performing code sinking. The file name is made
4175 by appending @file{.sink} to the source file name.
4178 @opindex fdump-tree-dom
4179 Dump each function after applying dominator tree optimizations. The file
4180 name is made by appending @file{.dom} to the source file name.
4183 @opindex fdump-tree-dse
4184 Dump each function after applying dead store elimination. The file
4185 name is made by appending @file{.dse} to the source file name.
4188 @opindex fdump-tree-phiopt
4189 Dump each function after optimizing PHI nodes into straightline code. The file
4190 name is made by appending @file{.phiopt} to the source file name.
4193 @opindex fdump-tree-forwprop
4194 Dump each function after forward propagating single use variables. The file
4195 name is made by appending @file{.forwprop} to the source file name.
4198 @opindex fdump-tree-copyrename
4199 Dump each function after applying the copy rename optimization. The file
4200 name is made by appending @file{.copyrename} to the source file name.
4203 @opindex fdump-tree-nrv
4204 Dump each function after applying the named return value optimization on
4205 generic trees. The file name is made by appending @file{.nrv} to the source
4209 @opindex fdump-tree-vect
4210 Dump each function after applying vectorization of loops. The file name is
4211 made by appending @file{.vect} to the source file name.
4214 @opindex fdump-tree-vrp
4215 Dump each function after Value Range Propagation (VRP). The file name
4216 is made by appending @file{.vrp} to the source file name.
4219 @opindex fdump-tree-all
4220 Enable all the available tree dumps with the flags provided in this option.
4223 @item -ftree-vectorizer-verbose=@var{n}
4224 @opindex ftree-vectorizer-verbose
4225 This option controls the amount of debugging output the vectorizer prints.
4226 This information is written to standard error, unless
4227 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4228 in which case it is output to the usual dump listing file, @file{.vect}.
4229 For @var{n}=0 no diagnostic information is reported.
4230 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4231 and the total number of loops that got vectorized.
4232 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4233 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4234 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4235 level that @option{-fdump-tree-vect-stats} uses.
4236 Higher verbosity levels mean either more information dumped for each
4237 reported loop, or same amount of information reported for more loops:
4238 If @var{n}=3, alignment related information is added to the reports.
4239 If @var{n}=4, data-references related information (e.g. memory dependences,
4240 memory access-patterns) is added to the reports.
4241 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4242 that did not pass the first analysis phase (i.e. may not be countable, or
4243 may have complicated control-flow).
4244 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4245 For @var{n}=7, all the information the vectorizer generates during its
4246 analysis and transformation is reported. This is the same verbosity level
4247 that @option{-fdump-tree-vect-details} uses.
4249 @item -frandom-seed=@var{string}
4250 @opindex frandom-string
4251 This option provides a seed that GCC uses when it would otherwise use
4252 random numbers. It is used to generate certain symbol names
4253 that have to be different in every compiled file. It is also used to
4254 place unique stamps in coverage data files and the object files that
4255 produce them. You can use the @option{-frandom-seed} option to produce
4256 reproducibly identical object files.
4258 The @var{string} should be different for every file you compile.
4260 @item -fsched-verbose=@var{n}
4261 @opindex fsched-verbose
4262 On targets that use instruction scheduling, this option controls the
4263 amount of debugging output the scheduler prints. This information is
4264 written to standard error, unless @option{-dS} or @option{-dR} is
4265 specified, in which case it is output to the usual dump
4266 listing file, @file{.sched} or @file{.sched2} respectively. However
4267 for @var{n} greater than nine, the output is always printed to standard
4270 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4271 same information as @option{-dRS}. For @var{n} greater than one, it
4272 also output basic block probabilities, detailed ready list information
4273 and unit/insn info. For @var{n} greater than two, it includes RTL
4274 at abort point, control-flow and regions info. And for @var{n} over
4275 four, @option{-fsched-verbose} also includes dependence info.
4279 Store the usual ``temporary'' intermediate files permanently; place them
4280 in the current directory and name them based on the source file. Thus,
4281 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4282 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4283 preprocessed @file{foo.i} output file even though the compiler now
4284 normally uses an integrated preprocessor.
4286 When used in combination with the @option{-x} command line option,
4287 @option{-save-temps} is sensible enough to avoid over writing an
4288 input source file with the same extension as an intermediate file.
4289 The corresponding intermediate file may be obtained by renaming the
4290 source file before using @option{-save-temps}.
4294 Report the CPU time taken by each subprocess in the compilation
4295 sequence. For C source files, this is the compiler proper and assembler
4296 (plus the linker if linking is done). The output looks like this:
4303 The first number on each line is the ``user time'', that is time spent
4304 executing the program itself. The second number is ``system time'',
4305 time spent executing operating system routines on behalf of the program.
4306 Both numbers are in seconds.
4308 @item -fvar-tracking
4309 @opindex fvar-tracking
4310 Run variable tracking pass. It computes where variables are stored at each
4311 position in code. Better debugging information is then generated
4312 (if the debugging information format supports this information).
4314 It is enabled by default when compiling with optimization (@option{-Os},
4315 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4316 the debug info format supports it.
4318 @item -print-file-name=@var{library}
4319 @opindex print-file-name
4320 Print the full absolute name of the library file @var{library} that
4321 would be used when linking---and don't do anything else. With this
4322 option, GCC does not compile or link anything; it just prints the
4325 @item -print-multi-directory
4326 @opindex print-multi-directory
4327 Print the directory name corresponding to the multilib selected by any
4328 other switches present in the command line. This directory is supposed
4329 to exist in @env{GCC_EXEC_PREFIX}.
4331 @item -print-multi-lib
4332 @opindex print-multi-lib
4333 Print the mapping from multilib directory names to compiler switches
4334 that enable them. The directory name is separated from the switches by
4335 @samp{;}, and each switch starts with an @samp{@@} instead of the
4336 @samp{-}, without spaces between multiple switches. This is supposed to
4337 ease shell-processing.
4339 @item -print-prog-name=@var{program}
4340 @opindex print-prog-name
4341 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4343 @item -print-libgcc-file-name
4344 @opindex print-libgcc-file-name
4345 Same as @option{-print-file-name=libgcc.a}.
4347 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4348 but you do want to link with @file{libgcc.a}. You can do
4351 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4354 @item -print-search-dirs
4355 @opindex print-search-dirs
4356 Print the name of the configured installation directory and a list of
4357 program and library directories @command{gcc} will search---and don't do anything else.
4359 This is useful when @command{gcc} prints the error message
4360 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4361 To resolve this you either need to put @file{cpp0} and the other compiler
4362 components where @command{gcc} expects to find them, or you can set the environment
4363 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4364 Don't forget the trailing @samp{/}.
4365 @xref{Environment Variables}.
4368 @opindex dumpmachine
4369 Print the compiler's target machine (for example,
4370 @samp{i686-pc-linux-gnu})---and don't do anything else.
4373 @opindex dumpversion
4374 Print the compiler version (for example, @samp{3.0})---and don't do
4379 Print the compiler's built-in specs---and don't do anything else. (This
4380 is used when GCC itself is being built.) @xref{Spec Files}.
4382 @item -feliminate-unused-debug-types
4383 @opindex feliminate-unused-debug-types
4384 Normally, when producing DWARF2 output, GCC will emit debugging
4385 information for all types declared in a compilation
4386 unit, regardless of whether or not they are actually used
4387 in that compilation unit. Sometimes this is useful, such as
4388 if, in the debugger, you want to cast a value to a type that is
4389 not actually used in your program (but is declared). More often,
4390 however, this results in a significant amount of wasted space.
4391 With this option, GCC will avoid producing debug symbol output
4392 for types that are nowhere used in the source file being compiled.
4395 @node Optimize Options
4396 @section Options That Control Optimization
4397 @cindex optimize options
4398 @cindex options, optimization
4400 These options control various sorts of optimizations.
4402 Without any optimization option, the compiler's goal is to reduce the
4403 cost of compilation and to make debugging produce the expected
4404 results. Statements are independent: if you stop the program with a
4405 breakpoint between statements, you can then assign a new value to any
4406 variable or change the program counter to any other statement in the
4407 function and get exactly the results you would expect from the source
4410 Turning on optimization flags makes the compiler attempt to improve
4411 the performance and/or code size at the expense of compilation time
4412 and possibly the ability to debug the program.
4414 The compiler performs optimization based on the knowledge it has of
4415 the program. Optimization levels @option{-O2} and above, in
4416 particular, enable @emph{unit-at-a-time} mode, which allows the
4417 compiler to consider information gained from later functions in
4418 the file when compiling a function. Compiling multiple files at
4419 once to a single output file in @emph{unit-at-a-time} mode allows
4420 the compiler to use information gained from all of the files when
4421 compiling each of them.
4423 Not all optimizations are controlled directly by a flag. Only
4424 optimizations that have a flag are listed.
4431 Optimize. Optimizing compilation takes somewhat more time, and a lot
4432 more memory for a large function.
4434 With @option{-O}, the compiler tries to reduce code size and execution
4435 time, without performing any optimizations that take a great deal of
4438 @option{-O} turns on the following optimization flags:
4439 @gccoptlist{-fdefer-pop @gol
4440 -fdelayed-branch @gol
4441 -fguess-branch-probability @gol
4442 -fcprop-registers @gol
4443 -fif-conversion @gol
4444 -fif-conversion2 @gol
4447 -ftree-dominator-opts @gol
4452 -ftree-copyrename @gol
4457 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4458 where doing so does not interfere with debugging.
4462 Optimize even more. GCC performs nearly all supported optimizations
4463 that do not involve a space-speed tradeoff. The compiler does not
4464 perform loop unrolling or function inlining when you specify @option{-O2}.
4465 As compared to @option{-O}, this option increases both compilation time
4466 and the performance of the generated code.
4468 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4469 also turns on the following optimization flags:
4470 @gccoptlist{-fthread-jumps @gol
4472 -foptimize-sibling-calls @gol
4473 -fcse-follow-jumps -fcse-skip-blocks @gol
4474 -fgcse -fgcse-lm @gol
4475 -fexpensive-optimizations @gol
4476 -frerun-cse-after-loop @gol
4479 -fschedule-insns -fschedule-insns2 @gol
4480 -fsched-interblock -fsched-spec @gol
4482 -fstrict-aliasing @gol
4483 -fdelete-null-pointer-checks @gol
4484 -freorder-blocks -freorder-functions @gol
4485 -funit-at-a-time @gol
4486 -falign-functions -falign-jumps @gol
4487 -falign-loops -falign-labels @gol
4491 Please note the warning under @option{-fgcse} about
4492 invoking @option{-O2} on programs that use computed gotos.
4496 Optimize yet more. @option{-O3} turns on all optimizations specified by
4497 @option{-O2} and also turns on the @option{-finline-functions},
4498 @option{-funswitch-loops} and @option{-fgcse-after-reload} options.
4502 Do not optimize. This is the default.
4506 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4507 do not typically increase code size. It also performs further
4508 optimizations designed to reduce code size.
4510 @option{-Os} disables the following optimization flags:
4511 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4512 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
4513 -fprefetch-loop-arrays -ftree-vect-loop-version}
4515 If you use multiple @option{-O} options, with or without level numbers,
4516 the last such option is the one that is effective.
4519 Options of the form @option{-f@var{flag}} specify machine-independent
4520 flags. Most flags have both positive and negative forms; the negative
4521 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4522 below, only one of the forms is listed---the one you typically will
4523 use. You can figure out the other form by either removing @samp{no-}
4526 The following options control specific optimizations. They are either
4527 activated by @option{-O} options or are related to ones that are. You
4528 can use the following flags in the rare cases when ``fine-tuning'' of
4529 optimizations to be performed is desired.
4532 @item -fno-default-inline
4533 @opindex fno-default-inline
4534 Do not make member functions inline by default merely because they are
4535 defined inside the class scope (C++ only). Otherwise, when you specify
4536 @w{@option{-O}}, member functions defined inside class scope are compiled
4537 inline by default; i.e., you don't need to add @samp{inline} in front of
4538 the member function name.
4540 @item -fno-defer-pop
4541 @opindex fno-defer-pop
4542 Always pop the arguments to each function call as soon as that function
4543 returns. For machines which must pop arguments after a function call,
4544 the compiler normally lets arguments accumulate on the stack for several
4545 function calls and pops them all at once.
4547 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4551 Force memory operands to be copied into registers before doing
4552 arithmetic on them. This produces better code by making all memory
4553 references potential common subexpressions. When they are not common
4554 subexpressions, instruction combination should eliminate the separate
4555 register-load. This option is now a nop and will be removed in 4.2.
4558 @opindex fforce-addr
4559 Force memory address constants to be copied into registers before
4560 doing arithmetic on them.
4562 @item -fomit-frame-pointer
4563 @opindex fomit-frame-pointer
4564 Don't keep the frame pointer in a register for functions that
4565 don't need one. This avoids the instructions to save, set up and
4566 restore frame pointers; it also makes an extra register available
4567 in many functions. @strong{It also makes debugging impossible on
4570 On some machines, such as the VAX, this flag has no effect, because
4571 the standard calling sequence automatically handles the frame pointer
4572 and nothing is saved by pretending it doesn't exist. The
4573 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4574 whether a target machine supports this flag. @xref{Registers,,Register
4575 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4577 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4579 @item -foptimize-sibling-calls
4580 @opindex foptimize-sibling-calls
4581 Optimize sibling and tail recursive calls.
4583 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4587 Don't pay attention to the @code{inline} keyword. Normally this option
4588 is used to keep the compiler from expanding any functions inline.
4589 Note that if you are not optimizing, no functions can be expanded inline.
4591 @item -finline-functions
4592 @opindex finline-functions
4593 Integrate all simple functions into their callers. The compiler
4594 heuristically decides which functions are simple enough to be worth
4595 integrating in this way.
4597 If all calls to a given function are integrated, and the function is
4598 declared @code{static}, then the function is normally not output as
4599 assembler code in its own right.
4601 Enabled at level @option{-O3}.
4603 @item -finline-functions-called-once
4604 @opindex finline-functions-called-once
4605 Consider all @code{static} functions called once for inlining into their
4606 caller even if they are not marked @code{inline}. If a call to a given
4607 function is integrated, then the function is not output as assembler code
4610 Enabled if @option{-funit-at-a-time} is enabled.
4612 @item -fearly-inlining
4613 @opindex fearly-inlining
4614 Inline functions marked by @code{always_inline} and functions whose body seems
4615 smaller than the function call overhead early before doing
4616 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
4617 makes profiling significantly cheaper and usually inlining faster on programs
4618 having large chains of nested wrapper functions.
4622 @item -finline-limit=@var{n}
4623 @opindex finline-limit
4624 By default, GCC limits the size of functions that can be inlined. This flag
4625 allows the control of this limit for functions that are explicitly marked as
4626 inline (i.e., marked with the inline keyword or defined within the class
4627 definition in c++). @var{n} is the size of functions that can be inlined in
4628 number of pseudo instructions (not counting parameter handling). The default
4629 value of @var{n} is 600.
4630 Increasing this value can result in more inlined code at
4631 the cost of compilation time and memory consumption. Decreasing usually makes
4632 the compilation faster and less code will be inlined (which presumably
4633 means slower programs). This option is particularly useful for programs that
4634 use inlining heavily such as those based on recursive templates with C++.
4636 Inlining is actually controlled by a number of parameters, which may be
4637 specified individually by using @option{--param @var{name}=@var{value}}.
4638 The @option{-finline-limit=@var{n}} option sets some of these parameters
4642 @item max-inline-insns-single
4643 is set to @var{n}/2.
4644 @item max-inline-insns-auto
4645 is set to @var{n}/2.
4646 @item min-inline-insns
4647 is set to 130 or @var{n}/4, whichever is smaller.
4648 @item max-inline-insns-rtl
4652 See below for a documentation of the individual
4653 parameters controlling inlining.
4655 @emph{Note:} pseudo instruction represents, in this particular context, an
4656 abstract measurement of function's size. In no way does it represent a count
4657 of assembly instructions and as such its exact meaning might change from one
4658 release to an another.
4660 @item -fkeep-inline-functions
4661 @opindex fkeep-inline-functions
4662 In C, emit @code{static} functions that are declared @code{inline}
4663 into the object file, even if the function has been inlined into all
4664 of its callers. This switch does not affect functions using the
4665 @code{extern inline} extension in GNU C@. In C++, emit any and all
4666 inline functions into the object file.
4668 @item -fkeep-static-consts
4669 @opindex fkeep-static-consts
4670 Emit variables declared @code{static const} when optimization isn't turned
4671 on, even if the variables aren't referenced.
4673 GCC enables this option by default. If you want to force the compiler to
4674 check if the variable was referenced, regardless of whether or not
4675 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4677 @item -fmerge-constants
4678 Attempt to merge identical constants (string constants and floating point
4679 constants) across compilation units.
4681 This option is the default for optimized compilation if the assembler and
4682 linker support it. Use @option{-fno-merge-constants} to inhibit this
4685 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4687 @item -fmerge-all-constants
4688 Attempt to merge identical constants and identical variables.
4690 This option implies @option{-fmerge-constants}. In addition to
4691 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4692 arrays or initialized constant variables with integral or floating point
4693 types. Languages like C or C++ require each non-automatic variable to
4694 have distinct location, so using this option will result in non-conforming
4697 @item -fmodulo-sched
4698 @opindex fmodulo-sched
4699 Perform swing modulo scheduling immediately before the first scheduling
4700 pass. This pass looks at innermost loops and reorders their
4701 instructions by overlapping different iterations.
4703 @item -fno-branch-count-reg
4704 @opindex fno-branch-count-reg
4705 Do not use ``decrement and branch'' instructions on a count register,
4706 but instead generate a sequence of instructions that decrement a
4707 register, compare it against zero, then branch based upon the result.
4708 This option is only meaningful on architectures that support such
4709 instructions, which include x86, PowerPC, IA-64 and S/390.
4711 The default is @option{-fbranch-count-reg}.
4713 @item -fno-function-cse
4714 @opindex fno-function-cse
4715 Do not put function addresses in registers; make each instruction that
4716 calls a constant function contain the function's address explicitly.
4718 This option results in less efficient code, but some strange hacks
4719 that alter the assembler output may be confused by the optimizations
4720 performed when this option is not used.
4722 The default is @option{-ffunction-cse}
4724 @item -fno-zero-initialized-in-bss
4725 @opindex fno-zero-initialized-in-bss
4726 If the target supports a BSS section, GCC by default puts variables that
4727 are initialized to zero into BSS@. This can save space in the resulting
4730 This option turns off this behavior because some programs explicitly
4731 rely on variables going to the data section. E.g., so that the
4732 resulting executable can find the beginning of that section and/or make
4733 assumptions based on that.
4735 The default is @option{-fzero-initialized-in-bss}.
4737 @item -fbounds-check
4738 @opindex fbounds-check
4739 For front-ends that support it, generate additional code to check that
4740 indices used to access arrays are within the declared range. This is
4741 currently only supported by the Java and Fortran front-ends, where
4742 this option defaults to true and false respectively.
4744 @item -fmudflap -fmudflapth -fmudflapir
4748 @cindex bounds checking
4750 For front-ends that support it (C and C++), instrument all risky
4751 pointer/array dereferencing operations, some standard library
4752 string/heap functions, and some other associated constructs with
4753 range/validity tests. Modules so instrumented should be immune to
4754 buffer overflows, invalid heap use, and some other classes of C/C++
4755 programming errors. The instrumentation relies on a separate runtime
4756 library (@file{libmudflap}), which will be linked into a program if
4757 @option{-fmudflap} is given at link time. Run-time behavior of the
4758 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4759 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4762 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4763 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4764 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4765 instrumentation should ignore pointer reads. This produces less
4766 instrumentation (and therefore faster execution) and still provides
4767 some protection against outright memory corrupting writes, but allows
4768 erroneously read data to propagate within a program.
4770 @item -fthread-jumps
4771 @opindex fthread-jumps
4772 Perform optimizations where we check to see if a jump branches to a
4773 location where another comparison subsumed by the first is found. If
4774 so, the first branch is redirected to either the destination of the
4775 second branch or a point immediately following it, depending on whether
4776 the condition is known to be true or false.
4778 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4780 @item -fcse-follow-jumps
4781 @opindex fcse-follow-jumps
4782 In common subexpression elimination, scan through jump instructions
4783 when the target of the jump is not reached by any other path. For
4784 example, when CSE encounters an @code{if} statement with an
4785 @code{else} clause, CSE will follow the jump when the condition
4788 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4790 @item -fcse-skip-blocks
4791 @opindex fcse-skip-blocks
4792 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4793 follow jumps which conditionally skip over blocks. When CSE
4794 encounters a simple @code{if} statement with no else clause,
4795 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4796 body of the @code{if}.
4798 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4800 @item -frerun-cse-after-loop
4801 @opindex frerun-cse-after-loop
4802 Re-run common subexpression elimination after loop optimizations has been
4805 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4809 Perform a global common subexpression elimination pass.
4810 This pass also performs global constant and copy propagation.
4812 @emph{Note:} When compiling a program using computed gotos, a GCC
4813 extension, you may get better runtime performance if you disable
4814 the global common subexpression elimination pass by adding
4815 @option{-fno-gcse} to the command line.
4817 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4821 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4822 attempt to move loads which are only killed by stores into themselves. This
4823 allows a loop containing a load/store sequence to be changed to a load outside
4824 the loop, and a copy/store within the loop.
4826 Enabled by default when gcse is enabled.
4830 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4831 global common subexpression elimination. This pass will attempt to move
4832 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4833 loops containing a load/store sequence can be changed to a load before
4834 the loop and a store after the loop.
4836 Not enabled at any optimization level.
4840 When @option{-fgcse-las} is enabled, the global common subexpression
4841 elimination pass eliminates redundant loads that come after stores to the
4842 same memory location (both partial and full redundancies).
4844 Not enabled at any optimization level.
4846 @item -fgcse-after-reload
4847 @opindex fgcse-after-reload
4848 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4849 pass is performed after reload. The purpose of this pass is to cleanup
4852 @item -funsafe-loop-optimizations
4853 @opindex funsafe-loop-optimizations
4854 If given, the loop optimizer will assume that loop indices do not
4855 overflow, and that the loops with nontrivial exit condition are not
4856 infinite. This enables a wider range of loop optimizations even if
4857 the loop optimizer itself cannot prove that these assumptions are valid.
4858 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
4859 if it finds this kind of loop.
4861 @item -fcrossjumping
4862 @opindex crossjumping
4863 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4864 resulting code may or may not perform better than without cross-jumping.
4866 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4868 @item -fif-conversion
4869 @opindex if-conversion
4870 Attempt to transform conditional jumps into branch-less equivalents. This
4871 include use of conditional moves, min, max, set flags and abs instructions, and
4872 some tricks doable by standard arithmetics. The use of conditional execution
4873 on chips where it is available is controlled by @code{if-conversion2}.
4875 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4877 @item -fif-conversion2
4878 @opindex if-conversion2
4879 Use conditional execution (where available) to transform conditional jumps into
4880 branch-less equivalents.
4882 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4884 @item -fdelete-null-pointer-checks
4885 @opindex fdelete-null-pointer-checks
4886 Use global dataflow analysis to identify and eliminate useless checks
4887 for null pointers. The compiler assumes that dereferencing a null
4888 pointer would have halted the program. If a pointer is checked after
4889 it has already been dereferenced, it cannot be null.
4891 In some environments, this assumption is not true, and programs can
4892 safely dereference null pointers. Use
4893 @option{-fno-delete-null-pointer-checks} to disable this optimization
4894 for programs which depend on that behavior.
4896 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4898 @item -fexpensive-optimizations
4899 @opindex fexpensive-optimizations
4900 Perform a number of minor optimizations that are relatively expensive.
4902 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4904 @item -foptimize-register-move
4906 @opindex foptimize-register-move
4908 Attempt to reassign register numbers in move instructions and as
4909 operands of other simple instructions in order to maximize the amount of
4910 register tying. This is especially helpful on machines with two-operand
4913 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4916 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4918 @item -fdelayed-branch
4919 @opindex fdelayed-branch
4920 If supported for the target machine, attempt to reorder instructions
4921 to exploit instruction slots available after delayed branch
4924 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4926 @item -fschedule-insns
4927 @opindex fschedule-insns
4928 If supported for the target machine, attempt to reorder instructions to
4929 eliminate execution stalls due to required data being unavailable. This
4930 helps machines that have slow floating point or memory load instructions
4931 by allowing other instructions to be issued until the result of the load
4932 or floating point instruction is required.
4934 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4936 @item -fschedule-insns2
4937 @opindex fschedule-insns2
4938 Similar to @option{-fschedule-insns}, but requests an additional pass of
4939 instruction scheduling after register allocation has been done. This is
4940 especially useful on machines with a relatively small number of
4941 registers and where memory load instructions take more than one cycle.
4943 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4945 @item -fno-sched-interblock
4946 @opindex fno-sched-interblock
4947 Don't schedule instructions across basic blocks. This is normally
4948 enabled by default when scheduling before register allocation, i.e.@:
4949 with @option{-fschedule-insns} or at @option{-O2} or higher.
4951 @item -fno-sched-spec
4952 @opindex fno-sched-spec
4953 Don't allow speculative motion of non-load instructions. This is normally
4954 enabled by default when scheduling before register allocation, i.e.@:
4955 with @option{-fschedule-insns} or at @option{-O2} or higher.
4957 @item -fsched-spec-load
4958 @opindex fsched-spec-load
4959 Allow speculative motion of some load instructions. This only makes
4960 sense when scheduling before register allocation, i.e.@: with
4961 @option{-fschedule-insns} or at @option{-O2} or higher.
4963 @item -fsched-spec-load-dangerous
4964 @opindex fsched-spec-load-dangerous
4965 Allow speculative motion of more load instructions. This only makes
4966 sense when scheduling before register allocation, i.e.@: with
4967 @option{-fschedule-insns} or at @option{-O2} or higher.
4969 @item -fsched-stalled-insns=@var{n}
4970 @opindex fsched-stalled-insns
4971 Define how many insns (if any) can be moved prematurely from the queue
4972 of stalled insns into the ready list, during the second scheduling pass.
4974 @item -fsched-stalled-insns-dep=@var{n}
4975 @opindex fsched-stalled-insns-dep
4976 Define how many insn groups (cycles) will be examined for a dependency
4977 on a stalled insn that is candidate for premature removal from the queue
4978 of stalled insns. Has an effect only during the second scheduling pass,
4979 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4981 @item -fsched2-use-superblocks
4982 @opindex fsched2-use-superblocks
4983 When scheduling after register allocation, do use superblock scheduling
4984 algorithm. Superblock scheduling allows motion across basic block boundaries
4985 resulting on faster schedules. This option is experimental, as not all machine
4986 descriptions used by GCC model the CPU closely enough to avoid unreliable
4987 results from the algorithm.
4989 This only makes sense when scheduling after register allocation, i.e.@: with
4990 @option{-fschedule-insns2} or at @option{-O2} or higher.
4992 @item -fsched2-use-traces
4993 @opindex fsched2-use-traces
4994 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4995 allocation and additionally perform code duplication in order to increase the
4996 size of superblocks using tracer pass. See @option{-ftracer} for details on
4999 This mode should produce faster but significantly longer programs. Also
5000 without @option{-fbranch-probabilities} the traces constructed may not
5001 match the reality and hurt the performance. This only makes
5002 sense when scheduling after register allocation, i.e.@: with
5003 @option{-fschedule-insns2} or at @option{-O2} or higher.
5005 @item -freschedule-modulo-scheduled-loops
5006 @opindex fscheduling-in-modulo-scheduled-loops
5007 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
5008 we may want to prevent the later scheduling passes from changing its schedule, we use this
5009 option to control that.
5011 @item -fcaller-saves
5012 @opindex fcaller-saves
5013 Enable values to be allocated in registers that will be clobbered by
5014 function calls, by emitting extra instructions to save and restore the
5015 registers around such calls. Such allocation is done only when it
5016 seems to result in better code than would otherwise be produced.
5018 This option is always enabled by default on certain machines, usually
5019 those which have no call-preserved registers to use instead.
5021 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5024 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
5025 enabled by default at @option{-O2} and @option{-O3}.
5028 Perform Full Redundancy Elimination (FRE) on trees. The difference
5029 between FRE and PRE is that FRE only considers expressions
5030 that are computed on all paths leading to the redundant computation.
5031 This analysis faster than PRE, though it exposes fewer redundancies.
5032 This flag is enabled by default at @option{-O} and higher.
5034 @item -ftree-copy-prop
5035 Perform copy propagation on trees. This pass eliminates unnecessary
5036 copy operations. This flag is enabled by default at @option{-O} and
5039 @item -ftree-store-copy-prop
5040 Perform copy propagation of memory loads and stores. This pass
5041 eliminates unnecessary copy operations in memory references
5042 (structures, global variables, arrays, etc). This flag is enabled by
5043 default at @option{-O2} and higher.
5046 Perform structural alias analysis on trees. This flag
5047 is enabled by default at @option{-O} and higher.
5050 Perform interprocedural pointer analysis.
5053 Perform forward store motion on trees. This flag is
5054 enabled by default at @option{-O} and higher.
5057 Perform sparse conditional constant propagation (CCP) on trees. This
5058 pass only operates on local scalar variables and is enabled by default
5059 at @option{-O} and higher.
5061 @item -ftree-store-ccp
5062 Perform sparse conditional constant propagation (CCP) on trees. This
5063 pass operates on both local scalar variables and memory stores and
5064 loads (global variables, structures, arrays, etc). This flag is
5065 enabled by default at @option{-O2} and higher.
5068 Perform dead code elimination (DCE) on trees. This flag is enabled by
5069 default at @option{-O} and higher.
5071 @item -ftree-dominator-opts
5072 Perform a variety of simple scalar cleanups (constant/copy
5073 propagation, redundancy elimination, range propagation and expression
5074 simplification) based on a dominator tree traversal. This also
5075 performs jump threading (to reduce jumps to jumps). This flag is
5076 enabled by default at @option{-O} and higher.
5079 Perform loop header copying on trees. This is beneficial since it increases
5080 effectiveness of code motion optimizations. It also saves one jump. This flag
5081 is enabled by default at @option{-O} and higher. It is not enabled
5082 for @option{-Os}, since it usually increases code size.
5084 @item -ftree-loop-optimize
5085 Perform loop optimizations on trees. This flag is enabled by default
5086 at @option{-O} and higher.
5088 @item -ftree-loop-linear
5089 Perform linear loop transformations on tree. This flag can improve cache
5090 performance and allow further loop optimizations to take place.
5092 @item -ftree-loop-im
5093 Perform loop invariant motion on trees. This pass moves only invariants that
5094 would be hard to handle at RTL level (function calls, operations that expand to
5095 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5096 operands of conditions that are invariant out of the loop, so that we can use
5097 just trivial invariantness analysis in loop unswitching. The pass also includes
5100 @item -ftree-loop-ivcanon
5101 Create a canonical counter for number of iterations in the loop for that
5102 determining number of iterations requires complicated analysis. Later
5103 optimizations then may determine the number easily. Useful especially
5104 in connection with unrolling.
5107 Perform induction variable optimizations (strength reduction, induction
5108 variable merging and induction variable elimination) on trees.
5111 Perform scalar replacement of aggregates. This pass replaces structure
5112 references with scalars to prevent committing structures to memory too
5113 early. This flag is enabled by default at @option{-O} and higher.
5115 @item -ftree-copyrename
5116 Perform copy renaming on trees. This pass attempts to rename compiler
5117 temporaries to other variables at copy locations, usually resulting in
5118 variable names which more closely resemble the original variables. This flag
5119 is enabled by default at @option{-O} and higher.
5122 Perform temporary expression replacement during the SSA->normal phase. Single
5123 use/single def temporaries are replaced at their use location with their
5124 defining expression. This results in non-GIMPLE code, but gives the expanders
5125 much more complex trees to work on resulting in better RTL generation. This is
5126 enabled by default at @option{-O} and higher.
5129 Perform live range splitting during the SSA->normal phase. Distinct live
5130 ranges of a variable are split into unique variables, allowing for better
5131 optimization later. This is enabled by default at @option{-O} and higher.
5133 @item -ftree-vectorize
5134 Perform loop vectorization on trees.
5136 @item -ftree-vect-loop-version
5137 @opindex ftree-vect-loop-version
5138 Perform loop versioning when doing loop vectorization on trees. When a loop
5139 appears to be vectorizable except that data alignment or data dependence cannot
5140 be determined at compile time then vectorized and non-vectorized versions of
5141 the loop are generated along with runtime checks for alignment or dependence
5142 to control which version is executed. This option is enabled by default
5143 except at level @option{-Os} where it is disabled.
5146 Perform Value Range Propagation on trees. This is similar to the
5147 constant propagation pass, but instead of values, ranges of values are
5148 propagated. This allows the optimizers to remove unnecessary range
5149 checks like array bound checks and null pointer checks. This is
5150 enabled by default at @option{-O2} and higher. Null pointer check
5151 elimination is only done if @option{-fdelete-null-pointer-checks} is
5156 Perform tail duplication to enlarge superblock size. This transformation
5157 simplifies the control flow of the function allowing other optimizations to do
5160 @item -funroll-loops
5161 @opindex funroll-loops
5162 Unroll loops whose number of iterations can be determined at compile
5163 time or upon entry to the loop. @option{-funroll-loops} implies
5164 @option{-frerun-cse-after-loop}. This option makes code larger,
5165 and may or may not make it run faster.
5167 @item -funroll-all-loops
5168 @opindex funroll-all-loops
5169 Unroll all loops, even if their number of iterations is uncertain when
5170 the loop is entered. This usually makes programs run more slowly.
5171 @option{-funroll-all-loops} implies the same options as
5172 @option{-funroll-loops},
5174 @item -fsplit-ivs-in-unroller
5175 @opindex -fsplit-ivs-in-unroller
5176 Enables expressing of values of induction variables in later iterations
5177 of the unrolled loop using the value in the first iteration. This breaks
5178 long dependency chains, thus improving efficiency of the scheduling passes.
5180 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5181 same effect. However in cases the loop body is more complicated than
5182 a single basic block, this is not reliable. It also does not work at all
5183 on some of the architectures due to restrictions in the CSE pass.
5185 This optimization is enabled by default.
5187 @item -fvariable-expansion-in-unroller
5188 @opindex -fvariable-expansion-in-unroller
5189 With this option, the compiler will create multiple copies of some
5190 local variables when unrolling a loop which can result in superior code.
5192 @item -fprefetch-loop-arrays
5193 @opindex fprefetch-loop-arrays
5194 If supported by the target machine, generate instructions to prefetch
5195 memory to improve the performance of loops that access large arrays.
5197 This option may generate better or worse code; results are highly
5198 dependent on the structure of loops within the source code.
5200 Disabled at level @option{-Os}.
5203 @itemx -fno-peephole2
5204 @opindex fno-peephole
5205 @opindex fno-peephole2
5206 Disable any machine-specific peephole optimizations. The difference
5207 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5208 are implemented in the compiler; some targets use one, some use the
5209 other, a few use both.
5211 @option{-fpeephole} is enabled by default.
5212 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5214 @item -fno-guess-branch-probability
5215 @opindex fno-guess-branch-probability
5216 Do not guess branch probabilities using heuristics.
5218 GCC will use heuristics to guess branch probabilities if they are
5219 not provided by profiling feedback (@option{-fprofile-arcs}). These
5220 heuristics are based on the control flow graph. If some branch probabilities
5221 are specified by @samp{__builtin_expect}, then the heuristics will be
5222 used to guess branch probabilities for the rest of the control flow graph,
5223 taking the @samp{__builtin_expect} info into account. The interactions
5224 between the heuristics and @samp{__builtin_expect} can be complex, and in
5225 some cases, it may be useful to disable the heuristics so that the effects
5226 of @samp{__builtin_expect} are easier to understand.
5228 The default is @option{-fguess-branch-probability} at levels
5229 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5231 @item -freorder-blocks
5232 @opindex freorder-blocks
5233 Reorder basic blocks in the compiled function in order to reduce number of
5234 taken branches and improve code locality.
5236 Enabled at levels @option{-O2}, @option{-O3}.
5238 @item -freorder-blocks-and-partition
5239 @opindex freorder-blocks-and-partition
5240 In addition to reordering basic blocks in the compiled function, in order
5241 to reduce number of taken branches, partitions hot and cold basic blocks
5242 into separate sections of the assembly and .o files, to improve
5243 paging and cache locality performance.
5245 This optimization is automatically turned off in the presence of
5246 exception handling, for linkonce sections, for functions with a user-defined
5247 section attribute and on any architecture that does not support named
5250 @item -freorder-functions
5251 @opindex freorder-functions
5252 Reorder functions in the object file in order to
5253 improve code locality. This is implemented by using special
5254 subsections @code{.text.hot} for most frequently executed functions and
5255 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5256 the linker so object file format must support named sections and linker must
5257 place them in a reasonable way.
5259 Also profile feedback must be available in to make this option effective. See
5260 @option{-fprofile-arcs} for details.
5262 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5264 @item -fstrict-aliasing
5265 @opindex fstrict-aliasing
5266 Allows the compiler to assume the strictest aliasing rules applicable to
5267 the language being compiled. For C (and C++), this activates
5268 optimizations based on the type of expressions. In particular, an
5269 object of one type is assumed never to reside at the same address as an
5270 object of a different type, unless the types are almost the same. For
5271 example, an @code{unsigned int} can alias an @code{int}, but not a
5272 @code{void*} or a @code{double}. A character type may alias any other
5275 Pay special attention to code like this:
5288 The practice of reading from a different union member than the one most
5289 recently written to (called ``type-punning'') is common. Even with
5290 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5291 is accessed through the union type. So, the code above will work as
5292 expected. However, this code might not:
5303 Every language that wishes to perform language-specific alias analysis
5304 should define a function that computes, given an @code{tree}
5305 node, an alias set for the node. Nodes in different alias sets are not
5306 allowed to alias. For an example, see the C front-end function
5307 @code{c_get_alias_set}.
5309 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5311 @item -falign-functions
5312 @itemx -falign-functions=@var{n}
5313 @opindex falign-functions
5314 Align the start of functions to the next power-of-two greater than
5315 @var{n}, skipping up to @var{n} bytes. For instance,
5316 @option{-falign-functions=32} aligns functions to the next 32-byte
5317 boundary, but @option{-falign-functions=24} would align to the next
5318 32-byte boundary only if this can be done by skipping 23 bytes or less.
5320 @option{-fno-align-functions} and @option{-falign-functions=1} are
5321 equivalent and mean that functions will not be aligned.
5323 Some assemblers only support this flag when @var{n} is a power of two;
5324 in that case, it is rounded up.
5326 If @var{n} is not specified or is zero, use a machine-dependent default.
5328 Enabled at levels @option{-O2}, @option{-O3}.
5330 @item -falign-labels
5331 @itemx -falign-labels=@var{n}
5332 @opindex falign-labels
5333 Align all branch targets to a power-of-two boundary, skipping up to
5334 @var{n} bytes like @option{-falign-functions}. This option can easily
5335 make code slower, because it must insert dummy operations for when the
5336 branch target is reached in the usual flow of the code.
5338 @option{-fno-align-labels} and @option{-falign-labels=1} are
5339 equivalent and mean that labels will not be aligned.
5341 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
5342 are greater than this value, then their values are used instead.
5344 If @var{n} is not specified or is zero, use a machine-dependent default
5345 which is very likely to be @samp{1}, meaning no alignment.
5347 Enabled at levels @option{-O2}, @option{-O3}.
5350 @itemx -falign-loops=@var{n}
5351 @opindex falign-loops
5352 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
5353 like @option{-falign-functions}. The hope is that the loop will be
5354 executed many times, which will make up for any execution of the dummy
5357 @option{-fno-align-loops} and @option{-falign-loops=1} are
5358 equivalent and mean that loops will not be aligned.
5360 If @var{n} is not specified or is zero, use a machine-dependent default.
5362 Enabled at levels @option{-O2}, @option{-O3}.
5365 @itemx -falign-jumps=@var{n}
5366 @opindex falign-jumps
5367 Align branch targets to a power-of-two boundary, for branch targets
5368 where the targets can only be reached by jumping, skipping up to @var{n}
5369 bytes like @option{-falign-functions}. In this case, no dummy operations
5372 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
5373 equivalent and mean that loops will not be aligned.
5375 If @var{n} is not specified or is zero, use a machine-dependent default.
5377 Enabled at levels @option{-O2}, @option{-O3}.
5379 @item -funit-at-a-time
5380 @opindex funit-at-a-time
5381 Parse the whole compilation unit before starting to produce code.
5382 This allows some extra optimizations to take place but consumes
5383 more memory (in general). There are some compatibility issues
5384 with @emph{unit-at-a-time} mode:
5387 enabling @emph{unit-at-a-time} mode may change the order
5388 in which functions, variables, and top-level @code{asm} statements
5389 are emitted, and will likely break code relying on some particular
5390 ordering. The majority of such top-level @code{asm} statements,
5391 though, can be replaced by @code{section} attributes. The
5392 @option{fno-toplevel-reorder} option may be used to keep the ordering
5393 used in the input file, at the cost of some optimizations.
5396 @emph{unit-at-a-time} mode removes unreferenced static variables
5397 and functions. This may result in undefined references
5398 when an @code{asm} statement refers directly to variables or functions
5399 that are otherwise unused. In that case either the variable/function
5400 shall be listed as an operand of the @code{asm} statement operand or,
5401 in the case of top-level @code{asm} statements the attribute @code{used}
5402 shall be used on the declaration.
5405 Static functions now can use non-standard passing conventions that
5406 may break @code{asm} statements calling functions directly. Again,
5407 attribute @code{used} will prevent this behavior.
5410 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
5411 but this scheme may not be supported by future releases of GCC@.
5413 Enabled at levels @option{-O2}, @option{-O3}.
5415 @item -fno-toplevel-reorder
5416 Do not reorder top-level functions, variables, and @code{asm}
5417 statements. Output them in the same order that they appear in the
5418 input file. When this option is used, unreferenced static variables
5419 will not be removed. This option is intended to support existing code
5420 which relies on a particular ordering. For new code, it is better to
5425 Constructs webs as commonly used for register allocation purposes and assign
5426 each web individual pseudo register. This allows the register allocation pass
5427 to operate on pseudos directly, but also strengthens several other optimization
5428 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
5429 however, make debugging impossible, since variables will no longer stay in a
5432 Enabled by default with @option{-funroll-loops}.
5434 @item -fwhole-program
5435 @opindex fwhole-program
5436 Assume that the current compilation unit represents whole program being
5437 compiled. All public functions and variables with the exception of @code{main}
5438 and those merged by attribute @code{externally_visible} become static functions
5439 and in a affect gets more aggressively optimized by interprocedural optimizers.
5440 While this option is equivalent to proper use of @code{static} keyword for
5441 programs consisting of single file, in combination with option
5442 @option{--combine} this flag can be used to compile most of smaller scale C
5443 programs since the functions and variables become local for the whole combined
5444 compilation unit, not for the single source file itself.
5447 @item -fno-cprop-registers
5448 @opindex fno-cprop-registers
5449 After register allocation and post-register allocation instruction splitting,
5450 we perform a copy-propagation pass to try to reduce scheduling dependencies
5451 and occasionally eliminate the copy.
5453 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5455 @item -fprofile-generate
5456 @opindex fprofile-generate
5458 Enable options usually used for instrumenting application to produce
5459 profile useful for later recompilation with profile feedback based
5460 optimization. You must use @option{-fprofile-generate} both when
5461 compiling and when linking your program.
5463 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
5466 @opindex fprofile-use
5467 Enable profile feedback directed optimizations, and optimizations
5468 generally profitable only with profile feedback available.
5470 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
5471 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
5475 The following options control compiler behavior regarding floating
5476 point arithmetic. These options trade off between speed and
5477 correctness. All must be specifically enabled.
5481 @opindex ffloat-store
5482 Do not store floating point variables in registers, and inhibit other
5483 options that might change whether a floating point value is taken from a
5486 @cindex floating point precision
5487 This option prevents undesirable excess precision on machines such as
5488 the 68000 where the floating registers (of the 68881) keep more
5489 precision than a @code{double} is supposed to have. Similarly for the
5490 x86 architecture. For most programs, the excess precision does only
5491 good, but a few programs rely on the precise definition of IEEE floating
5492 point. Use @option{-ffloat-store} for such programs, after modifying
5493 them to store all pertinent intermediate computations into variables.
5497 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5498 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5499 @option{-fno-rounding-math}, @option{-fno-signaling-nans}
5500 and @option{fcx-limited-range}.
5502 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5504 This option should never be turned on by any @option{-O} option since
5505 it can result in incorrect output for programs which depend on
5506 an exact implementation of IEEE or ISO rules/specifications for
5509 @item -fno-math-errno
5510 @opindex fno-math-errno
5511 Do not set ERRNO after calling math functions that are executed
5512 with a single instruction, e.g., sqrt. A program that relies on
5513 IEEE exceptions for math error handling may want to use this flag
5514 for speed while maintaining IEEE arithmetic compatibility.
5516 This option should never be turned on by any @option{-O} option since
5517 it can result in incorrect output for programs which depend on
5518 an exact implementation of IEEE or ISO rules/specifications for
5521 The default is @option{-fmath-errno}.
5523 On Darwin systems, the math library never sets @code{errno}. There is therefore
5524 no reason for the compiler to consider the possibility that it might,
5525 and @option{-fno-math-errno} is the default.
5527 @item -funsafe-math-optimizations
5528 @opindex funsafe-math-optimizations
5529 Allow optimizations for floating-point arithmetic that (a) assume
5530 that arguments and results are valid and (b) may violate IEEE or
5531 ANSI standards. When used at link-time, it may include libraries
5532 or startup files that change the default FPU control word or other
5533 similar optimizations.
5535 This option should never be turned on by any @option{-O} option since
5536 it can result in incorrect output for programs which depend on
5537 an exact implementation of IEEE or ISO rules/specifications for
5540 The default is @option{-fno-unsafe-math-optimizations}.
5542 @item -ffinite-math-only
5543 @opindex ffinite-math-only
5544 Allow optimizations for floating-point arithmetic that assume
5545 that arguments and results are not NaNs or +-Infs.
5547 This option should never be turned on by any @option{-O} option since
5548 it can result in incorrect output for programs which depend on
5549 an exact implementation of IEEE or ISO rules/specifications.
5551 The default is @option{-fno-finite-math-only}.
5553 @item -fno-trapping-math
5554 @opindex fno-trapping-math
5555 Compile code assuming that floating-point operations cannot generate
5556 user-visible traps. These traps include division by zero, overflow,
5557 underflow, inexact result and invalid operation. This option implies
5558 @option{-fno-signaling-nans}. Setting this option may allow faster
5559 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5561 This option should never be turned on by any @option{-O} option since
5562 it can result in incorrect output for programs which depend on
5563 an exact implementation of IEEE or ISO rules/specifications for
5566 The default is @option{-ftrapping-math}.
5568 @item -frounding-math
5569 @opindex frounding-math
5570 Disable transformations and optimizations that assume default floating
5571 point rounding behavior. This is round-to-zero for all floating point
5572 to integer conversions, and round-to-nearest for all other arithmetic
5573 truncations. This option should be specified for programs that change
5574 the FP rounding mode dynamically, or that may be executed with a
5575 non-default rounding mode. This option disables constant folding of
5576 floating point expressions at compile-time (which may be affected by
5577 rounding mode) and arithmetic transformations that are unsafe in the
5578 presence of sign-dependent rounding modes.
5580 The default is @option{-fno-rounding-math}.
5582 This option is experimental and does not currently guarantee to
5583 disable all GCC optimizations that are affected by rounding mode.
5584 Future versions of GCC may provide finer control of this setting
5585 using C99's @code{FENV_ACCESS} pragma. This command line option
5586 will be used to specify the default state for @code{FENV_ACCESS}.
5588 @item -frtl-abstract-sequences
5589 @opindex frtl-abstract-sequences
5590 It is a size optimization method. This option is to find identical
5591 sequences of code, which can be turned into pseudo-procedures and
5592 then replace all occurrences with calls to the newly created
5593 subroutine. It is kind of an opposite of @option{-finline-functions}.
5594 This optimization runs at RTL level.
5596 @item -fsignaling-nans
5597 @opindex fsignaling-nans
5598 Compile code assuming that IEEE signaling NaNs may generate user-visible
5599 traps during floating-point operations. Setting this option disables
5600 optimizations that may change the number of exceptions visible with
5601 signaling NaNs. This option implies @option{-ftrapping-math}.
5603 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5606 The default is @option{-fno-signaling-nans}.
5608 This option is experimental and does not currently guarantee to
5609 disable all GCC optimizations that affect signaling NaN behavior.
5611 @item -fsingle-precision-constant
5612 @opindex fsingle-precision-constant
5613 Treat floating point constant as single precision constant instead of
5614 implicitly converting it to double precision constant.
5616 @item -fcx-limited-range
5617 @itemx -fno-cx-limited-range
5618 @opindex fcx-limited-range
5619 @opindex fno-cx-limited-range
5620 When enabled, this option states that a range reduction step is not
5621 needed when performing complex division. The default is
5622 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
5624 This option controls the default setting of the ISO C99
5625 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
5630 The following options control optimizations that may improve
5631 performance, but are not enabled by any @option{-O} options. This
5632 section includes experimental options that may produce broken code.
5635 @item -fbranch-probabilities
5636 @opindex fbranch-probabilities
5637 After running a program compiled with @option{-fprofile-arcs}
5638 (@pxref{Debugging Options,, Options for Debugging Your Program or
5639 @command{gcc}}), you can compile it a second time using
5640 @option{-fbranch-probabilities}, to improve optimizations based on
5641 the number of times each branch was taken. When the program
5642 compiled with @option{-fprofile-arcs} exits it saves arc execution
5643 counts to a file called @file{@var{sourcename}.gcda} for each source
5644 file The information in this data file is very dependent on the
5645 structure of the generated code, so you must use the same source code
5646 and the same optimization options for both compilations.
5648 With @option{-fbranch-probabilities}, GCC puts a
5649 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5650 These can be used to improve optimization. Currently, they are only
5651 used in one place: in @file{reorg.c}, instead of guessing which path a
5652 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5653 exactly determine which path is taken more often.
5655 @item -fprofile-values
5656 @opindex fprofile-values
5657 If combined with @option{-fprofile-arcs}, it adds code so that some
5658 data about values of expressions in the program is gathered.
5660 With @option{-fbranch-probabilities}, it reads back the data gathered
5661 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5662 notes to instructions for their later usage in optimizations.
5664 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5668 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5669 a code to gather information about values of expressions.
5671 With @option{-fbranch-probabilities}, it reads back the data gathered
5672 and actually performs the optimizations based on them.
5673 Currently the optimizations include specialization of division operation
5674 using the knowledge about the value of the denominator.
5676 @item -frename-registers
5677 @opindex frename-registers
5678 Attempt to avoid false dependencies in scheduled code by making use
5679 of registers left over after register allocation. This optimization
5680 will most benefit processors with lots of registers. Depending on the
5681 debug information format adopted by the target, however, it can
5682 make debugging impossible, since variables will no longer stay in
5683 a ``home register''.
5685 Enabled by default with @option{-funroll-loops}.
5689 Perform tail duplication to enlarge superblock size. This transformation
5690 simplifies the control flow of the function allowing other optimizations to do
5693 Enabled with @option{-fprofile-use}.
5695 @item -funroll-loops
5696 @opindex funroll-loops
5697 Unroll loops whose number of iterations can be determined at compile time or
5698 upon entry to the loop. @option{-funroll-loops} implies
5699 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
5700 It also turns on complete loop peeling (i.e.@: complete removal of loops with
5701 small constant number of iterations). This option makes code larger, and may
5702 or may not make it run faster.
5704 Enabled with @option{-fprofile-use}.
5706 @item -funroll-all-loops
5707 @opindex funroll-all-loops
5708 Unroll all loops, even if their number of iterations is uncertain when
5709 the loop is entered. This usually makes programs run more slowly.
5710 @option{-funroll-all-loops} implies the same options as
5711 @option{-funroll-loops}.
5714 @opindex fpeel-loops
5715 Peels the loops for that there is enough information that they do not
5716 roll much (from profile feedback). It also turns on complete loop peeling
5717 (i.e.@: complete removal of loops with small constant number of iterations).
5719 Enabled with @option{-fprofile-use}.
5721 @item -fmove-loop-invariants
5722 @opindex fmove-loop-invariants
5723 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
5724 at level @option{-O1}
5726 @item -funswitch-loops
5727 @opindex funswitch-loops
5728 Move branches with loop invariant conditions out of the loop, with duplicates
5729 of the loop on both branches (modified according to result of the condition).
5731 @item -ffunction-sections
5732 @itemx -fdata-sections
5733 @opindex ffunction-sections
5734 @opindex fdata-sections
5735 Place each function or data item into its own section in the output
5736 file if the target supports arbitrary sections. The name of the
5737 function or the name of the data item determines the section's name
5740 Use these options on systems where the linker can perform optimizations
5741 to improve locality of reference in the instruction space. Most systems
5742 using the ELF object format and SPARC processors running Solaris 2 have
5743 linkers with such optimizations. AIX may have these optimizations in
5746 Only use these options when there are significant benefits from doing
5747 so. When you specify these options, the assembler and linker will
5748 create larger object and executable files and will also be slower.
5749 You will not be able to use @code{gprof} on all systems if you
5750 specify this option and you may have problems with debugging if
5751 you specify both this option and @option{-g}.
5753 @item -fbranch-target-load-optimize
5754 @opindex fbranch-target-load-optimize
5755 Perform branch target register load optimization before prologue / epilogue
5757 The use of target registers can typically be exposed only during reload,
5758 thus hoisting loads out of loops and doing inter-block scheduling needs
5759 a separate optimization pass.
5761 @item -fbranch-target-load-optimize2
5762 @opindex fbranch-target-load-optimize2
5763 Perform branch target register load optimization after prologue / epilogue
5766 @item -fbtr-bb-exclusive
5767 @opindex fbtr-bb-exclusive
5768 When performing branch target register load optimization, don't reuse
5769 branch target registers in within any basic block.
5771 @item -fstack-protector
5772 Emit extra code to check for buffer overflows, such as stack smashing
5773 attacks. This is done by adding a guard variable to functions with
5774 vulnerable objects. This includes functions that call alloca, and
5775 functions with buffers larger than 8 bytes. The guards are initialized
5776 when a function is entered and then checked when the function exits.
5777 If a guard check fails, an error message is printed and the program exits.
5779 @item -fstack-protector-all
5780 Like @option{-fstack-protector} except that all functions are protected.
5782 @item -fsection-anchors
5783 @opindex fsection-anchors
5784 Try to reduce the number of symbolic address calculations by using
5785 shared ``anchor'' symbols to address nearby objects. This transformation
5786 can help to reduce the number of GOT entries and GOT accesses on some
5789 For example, the implementation of the following function @code{foo}:
5793 int foo (void) @{ return a + b + c; @}
5796 would usually calculate the addresses of all three variables, but if you
5797 compile it with @option{-fsection-anchors}, it will access the variables
5798 from a common anchor point instead. The effect is similar to the
5799 following pseudocode (which isn't valid C):
5804 register int *xr = &x;
5805 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5809 Not all targets support this option.
5811 @item --param @var{name}=@var{value}
5813 In some places, GCC uses various constants to control the amount of
5814 optimization that is done. For example, GCC will not inline functions
5815 that contain more that a certain number of instructions. You can
5816 control some of these constants on the command-line using the
5817 @option{--param} option.
5819 The names of specific parameters, and the meaning of the values, are
5820 tied to the internals of the compiler, and are subject to change
5821 without notice in future releases.
5823 In each case, the @var{value} is an integer. The allowable choices for
5824 @var{name} are given in the following table:
5827 @item salias-max-implicit-fields
5828 The maximum number of fields in a variable without direct
5829 structure accesses for which structure aliasing will consider trying
5830 to track each field. The default is 5
5832 @item salias-max-array-elements
5833 The maximum number of elements an array can have and its elements
5834 still be tracked individually by structure aliasing. The default is 4
5836 @item sra-max-structure-size
5837 The maximum structure size, in bytes, at which the scalar replacement
5838 of aggregates (SRA) optimization will perform block copies. The
5839 default value, 0, implies that GCC will select the most appropriate
5842 @item sra-field-structure-ratio
5843 The threshold ratio (as a percentage) between instantiated fields and
5844 the complete structure size. We say that if the ratio of the number
5845 of bytes in instantiated fields to the number of bytes in the complete
5846 structure exceeds this parameter, then block copies are not used. The
5849 @item max-crossjump-edges
5850 The maximum number of incoming edges to consider for crossjumping.
5851 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5852 the number of edges incoming to each block. Increasing values mean
5853 more aggressive optimization, making the compile time increase with
5854 probably small improvement in executable size.
5856 @item min-crossjump-insns
5857 The minimum number of instructions which must be matched at the end
5858 of two blocks before crossjumping will be performed on them. This
5859 value is ignored in the case where all instructions in the block being
5860 crossjumped from are matched. The default value is 5.
5862 @item max-grow-copy-bb-insns
5863 The maximum code size expansion factor when copying basic blocks
5864 instead of jumping. The expansion is relative to a jump instruction.
5865 The default value is 8.
5867 @item max-goto-duplication-insns
5868 The maximum number of instructions to duplicate to a block that jumps
5869 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
5870 passes, GCC factors computed gotos early in the compilation process,
5871 and unfactors them as late as possible. Only computed jumps at the
5872 end of a basic blocks with no more than max-goto-duplication-insns are
5873 unfactored. The default value is 8.
5875 @item max-delay-slot-insn-search
5876 The maximum number of instructions to consider when looking for an
5877 instruction to fill a delay slot. If more than this arbitrary number of
5878 instructions is searched, the time savings from filling the delay slot
5879 will be minimal so stop searching. Increasing values mean more
5880 aggressive optimization, making the compile time increase with probably
5881 small improvement in executable run time.
5883 @item max-delay-slot-live-search
5884 When trying to fill delay slots, the maximum number of instructions to
5885 consider when searching for a block with valid live register
5886 information. Increasing this arbitrarily chosen value means more
5887 aggressive optimization, increasing the compile time. This parameter
5888 should be removed when the delay slot code is rewritten to maintain the
5891 @item max-gcse-memory
5892 The approximate maximum amount of memory that will be allocated in
5893 order to perform the global common subexpression elimination
5894 optimization. If more memory than specified is required, the
5895 optimization will not be done.
5897 @item max-gcse-passes
5898 The maximum number of passes of GCSE to run. The default is 1.
5900 @item max-pending-list-length
5901 The maximum number of pending dependencies scheduling will allow
5902 before flushing the current state and starting over. Large functions
5903 with few branches or calls can create excessively large lists which
5904 needlessly consume memory and resources.
5906 @item max-inline-insns-single
5907 Several parameters control the tree inliner used in gcc.
5908 This number sets the maximum number of instructions (counted in GCC's
5909 internal representation) in a single function that the tree inliner
5910 will consider for inlining. This only affects functions declared
5911 inline and methods implemented in a class declaration (C++).
5912 The default value is 450.
5914 @item max-inline-insns-auto
5915 When you use @option{-finline-functions} (included in @option{-O3}),
5916 a lot of functions that would otherwise not be considered for inlining
5917 by the compiler will be investigated. To those functions, a different
5918 (more restrictive) limit compared to functions declared inline can
5920 The default value is 90.
5922 @item large-function-insns
5923 The limit specifying really large functions. For functions larger than this
5924 limit after inlining inlining is constrained by
5925 @option{--param large-function-growth}. This parameter is useful primarily
5926 to avoid extreme compilation time caused by non-linear algorithms used by the
5928 This parameter is ignored when @option{-funit-at-a-time} is not used.
5929 The default value is 2700.
5931 @item large-function-growth
5932 Specifies maximal growth of large function caused by inlining in percents.
5933 This parameter is ignored when @option{-funit-at-a-time} is not used.
5934 The default value is 100 which limits large function growth to 2.0 times
5937 @item large-unit-insns
5938 The limit specifying large translation unit. Growth caused by inlining of
5939 units larger than this limit is limited by @option{--param inline-unit-growth}.
5940 For small units this might be too tight (consider unit consisting of function A
5941 that is inline and B that just calls A three time. If B is small relative to
5942 A, the growth of unit is 300\% and yet such inlining is very sane. For very
5943 large units consisting of small inlininable functions however the overall unit
5944 growth limit is needed to avoid exponential explosion of code size. Thus for
5945 smaller units, the size is increased to @option{--param large-unit-insns}
5946 before applying @option{--param inline-unit-growth}. The default is 10000
5948 @item inline-unit-growth
5949 Specifies maximal overall growth of the compilation unit caused by inlining.
5950 This parameter is ignored when @option{-funit-at-a-time} is not used.
5951 The default value is 50 which limits unit growth to 1.5 times the original
5954 @item max-inline-insns-recursive
5955 @itemx max-inline-insns-recursive-auto
5956 Specifies maximum number of instructions out-of-line copy of self recursive inline
5957 function can grow into by performing recursive inlining.
5959 For functions declared inline @option{--param max-inline-insns-recursive} is
5960 taken into acount. For function not declared inline, recursive inlining
5961 happens only when @option{-finline-functions} (included in @option{-O3}) is
5962 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
5963 default value is 450.
5965 @item max-inline-recursive-depth
5966 @itemx max-inline-recursive-depth-auto
5967 Specifies maximum recursion depth used by the recursive inlining.
5969 For functions declared inline @option{--param max-inline-recursive-depth} is
5970 taken into acount. For function not declared inline, recursive inlining
5971 happens only when @option{-finline-functions} (included in @option{-O3}) is
5972 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
5973 default value is 450.
5975 @item min-inline-recursive-probability
5976 Recursive inlining is profitable only for function having deep recursion
5977 in average and can hurt for function having little recursion depth by
5978 increasing the prologue size or complexity of function body to other
5981 When profile feedback is available (see @option{-fprofile-generate}) the actual
5982 recursion depth can be guessed from probability that function will recurse via
5983 given call expression. This parameter limits inlining only to call expression
5984 whose probability exceeds given threshold (in percents). The default value is
5987 @item inline-call-cost
5988 Specify cost of call instruction relative to simple arithmetics operations
5989 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
5990 functions and at the same time increases size of leaf function that is believed to
5991 reduce function size by being inlined. In effect it increases amount of
5992 inlining for code having large abstraction penalty (many functions that just
5993 pass the arguments to other functions) and decrease inlining for code with low
5994 abstraction penalty. The default value is 16.
5996 @item max-unrolled-insns
5997 The maximum number of instructions that a loop should have if that loop
5998 is unrolled, and if the loop is unrolled, it determines how many times
5999 the loop code is unrolled.
6001 @item max-average-unrolled-insns
6002 The maximum number of instructions biased by probabilities of their execution
6003 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6004 it determines how many times the loop code is unrolled.
6006 @item max-unroll-times
6007 The maximum number of unrollings of a single loop.
6009 @item max-peeled-insns
6010 The maximum number of instructions that a loop should have if that loop
6011 is peeled, and if the loop is peeled, it determines how many times
6012 the loop code is peeled.
6014 @item max-peel-times
6015 The maximum number of peelings of a single loop.
6017 @item max-completely-peeled-insns
6018 The maximum number of insns of a completely peeled loop.
6020 @item max-completely-peel-times
6021 The maximum number of iterations of a loop to be suitable for complete peeling.
6023 @item max-unswitch-insns
6024 The maximum number of insns of an unswitched loop.
6026 @item max-unswitch-level
6027 The maximum number of branches unswitched in a single loop.
6030 The minimum cost of an expensive expression in the loop invariant motion.
6032 @item iv-consider-all-candidates-bound
6033 Bound on number of candidates for induction variables below that
6034 all candidates are considered for each use in induction variable
6035 optimizations. Only the most relevant candidates are considered
6036 if there are more candidates, to avoid quadratic time complexity.
6038 @item iv-max-considered-uses
6039 The induction variable optimizations give up on loops that contain more
6040 induction variable uses.
6042 @item iv-always-prune-cand-set-bound
6043 If number of candidates in the set is smaller than this value,
6044 we always try to remove unnecessary ivs from the set during its
6045 optimization when a new iv is added to the set.
6047 @item scev-max-expr-size
6048 Bound on size of expressions used in the scalar evolutions analyzer.
6049 Large expressions slow the analyzer.
6051 @item vect-max-version-checks
6052 The maximum number of runtime checks that can be performed when doing
6053 loop versioning in the vectorizer. See option ftree-vect-loop-version
6054 for more information.
6056 @item max-iterations-to-track
6058 The maximum number of iterations of a loop the brute force algorithm
6059 for analysis of # of iterations of the loop tries to evaluate.
6061 @item hot-bb-count-fraction
6062 Select fraction of the maximal count of repetitions of basic block in program
6063 given basic block needs to have to be considered hot.
6065 @item hot-bb-frequency-fraction
6066 Select fraction of the maximal frequency of executions of basic block in
6067 function given basic block needs to have to be considered hot
6069 @item max-predicted-iterations
6070 The maximum number of loop iterations we predict statically. This is useful
6071 in cases where function contain single loop with known bound and other loop
6072 with unknown. We predict the known number of iterations correctly, while
6073 the unknown number of iterations average to roughly 10. This means that the
6074 loop without bounds would appear artificially cold relative to the other one.
6076 @item tracer-dynamic-coverage
6077 @itemx tracer-dynamic-coverage-feedback
6079 This value is used to limit superblock formation once the given percentage of
6080 executed instructions is covered. This limits unnecessary code size
6083 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6084 feedback is available. The real profiles (as opposed to statically estimated
6085 ones) are much less balanced allowing the threshold to be larger value.
6087 @item tracer-max-code-growth
6088 Stop tail duplication once code growth has reached given percentage. This is
6089 rather hokey argument, as most of the duplicates will be eliminated later in
6090 cross jumping, so it may be set to much higher values than is the desired code
6093 @item tracer-min-branch-ratio
6095 Stop reverse growth when the reverse probability of best edge is less than this
6096 threshold (in percent).
6098 @item tracer-min-branch-ratio
6099 @itemx tracer-min-branch-ratio-feedback
6101 Stop forward growth if the best edge do have probability lower than this
6104 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6105 compilation for profile feedback and one for compilation without. The value
6106 for compilation with profile feedback needs to be more conservative (higher) in
6107 order to make tracer effective.
6109 @item max-cse-path-length
6111 Maximum number of basic blocks on path that cse considers. The default is 10.
6114 The maximum instructions CSE process before flushing. The default is 1000.
6116 @item global-var-threshold
6118 Counts the number of function calls (@var{n}) and the number of
6119 call-clobbered variables (@var{v}). If @var{n}x@var{v} is larger than this limit, a
6120 single artificial variable will be created to represent all the
6121 call-clobbered variables at function call sites. This artificial
6122 variable will then be made to alias every call-clobbered variable.
6123 (done as @code{int * size_t} on the host machine; beware overflow).
6125 @item max-aliased-vops
6127 Maximum number of virtual operands allowed to represent aliases
6128 before triggering the alias grouping heuristic. Alias grouping
6129 reduces compile times and memory consumption needed for aliasing at
6130 the expense of precision loss in alias information.
6132 @item ggc-min-expand
6134 GCC uses a garbage collector to manage its own memory allocation. This
6135 parameter specifies the minimum percentage by which the garbage
6136 collector's heap should be allowed to expand between collections.
6137 Tuning this may improve compilation speed; it has no effect on code
6140 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6141 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6142 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6143 GCC is not able to calculate RAM on a particular platform, the lower
6144 bound of 30% is used. Setting this parameter and
6145 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6146 every opportunity. This is extremely slow, but can be useful for
6149 @item ggc-min-heapsize
6151 Minimum size of the garbage collector's heap before it begins bothering
6152 to collect garbage. The first collection occurs after the heap expands
6153 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6154 tuning this may improve compilation speed, and has no effect on code
6157 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6158 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6159 with a lower bound of 4096 (four megabytes) and an upper bound of
6160 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6161 particular platform, the lower bound is used. Setting this parameter
6162 very large effectively disables garbage collection. Setting this
6163 parameter and @option{ggc-min-expand} to zero causes a full collection
6164 to occur at every opportunity.
6166 @item max-reload-search-insns
6167 The maximum number of instruction reload should look backward for equivalent
6168 register. Increasing values mean more aggressive optimization, making the
6169 compile time increase with probably slightly better performance. The default
6172 @item max-cselib-memory-location
6173 The maximum number of memory locations cselib should take into acount.
6174 Increasing values mean more aggressive optimization, making the compile time
6175 increase with probably slightly better performance. The default value is 500.
6177 @item max-flow-memory-location
6178 Similar as @option{max-cselib-memory-location} but for dataflow liveness.
6179 The default value is 100.
6181 @item reorder-blocks-duplicate
6182 @itemx reorder-blocks-duplicate-feedback
6184 Used by basic block reordering pass to decide whether to use unconditional
6185 branch or duplicate the code on its destination. Code is duplicated when its
6186 estimated size is smaller than this value multiplied by the estimated size of
6187 unconditional jump in the hot spots of the program.
6189 The @option{reorder-block-duplicate-feedback} is used only when profile
6190 feedback is available and may be set to higher values than
6191 @option{reorder-block-duplicate} since information about the hot spots is more
6194 @item max-sched-region-blocks
6195 The maximum number of blocks in a region to be considered for
6196 interblock scheduling. The default value is 10.
6198 @item max-sched-region-insns
6199 The maximum number of insns in a region to be considered for
6200 interblock scheduling. The default value is 100.
6203 The minimum probability (in percents) of reaching a source block
6204 for interblock speculative scheduling. The default value is 40.
6206 @item max-sched-extend-regions-iters
6207 The maximum number of iterations through CFG to extend regions.
6208 0 - disable region extension,
6209 N - do at most N iterations.
6210 The default value is 2.
6212 @item max-sched-insn-conflict-delay
6213 The maximum conflict delay for an insn to be considered for speculative motion.
6214 The default value is 3.
6216 @item sched-spec-prob-cutoff
6217 The minimal probability of speculation success (in percents), so that
6218 speculative insn will be scheduled.
6219 The default value is 40.
6221 @item max-last-value-rtl
6223 The maximum size measured as number of RTLs that can be recorded in an expression
6224 in combiner for a pseudo register as last known value of that register. The default
6227 @item integer-share-limit
6228 Small integer constants can use a shared data structure, reducing the
6229 compiler's memory usage and increasing its speed. This sets the maximum
6230 value of a shared integer constant's. The default value is 256.
6232 @item min-virtual-mappings
6233 Specifies the minimum number of virtual mappings in the incremental
6234 SSA updater that should be registered to trigger the virtual mappings
6235 heuristic defined by virtual-mappings-ratio. The default value is
6238 @item virtual-mappings-ratio
6239 If the number of virtual mappings is virtual-mappings-ratio bigger
6240 than the number of virtual symbols to be updated, then the incremental
6241 SSA updater switches to a full update for those symbols. The default
6244 @item ssp-buffer-size
6245 The minimum size of buffers (i.e. arrays) that will receive stack smashing
6246 protection when @option{-fstack-protection} is used.
6248 @item max-jump-thread-duplication-stmts
6249 Maximum number of statements allowed in a block that needs to be
6250 duplicated when threading jumps.
6252 @item max-fields-for-field-sensitive
6253 Maximum number of fields in a structure we will treat in
6254 a field sensitive manner during pointer analysis.
6259 @node Preprocessor Options
6260 @section Options Controlling the Preprocessor
6261 @cindex preprocessor options
6262 @cindex options, preprocessor
6264 These options control the C preprocessor, which is run on each C source
6265 file before actual compilation.
6267 If you use the @option{-E} option, nothing is done except preprocessing.
6268 Some of these options make sense only together with @option{-E} because
6269 they cause the preprocessor output to be unsuitable for actual
6274 You can use @option{-Wp,@var{option}} to bypass the compiler driver
6275 and pass @var{option} directly through to the preprocessor. If
6276 @var{option} contains commas, it is split into multiple options at the
6277 commas. However, many options are modified, translated or interpreted
6278 by the compiler driver before being passed to the preprocessor, and
6279 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
6280 interface is undocumented and subject to change, so whenever possible
6281 you should avoid using @option{-Wp} and let the driver handle the
6284 @item -Xpreprocessor @var{option}
6285 @opindex preprocessor
6286 Pass @var{option} as an option to the preprocessor. You can use this to
6287 supply system-specific preprocessor options which GCC does not know how to
6290 If you want to pass an option that takes an argument, you must use
6291 @option{-Xpreprocessor} twice, once for the option and once for the argument.
6294 @include cppopts.texi
6296 @node Assembler Options
6297 @section Passing Options to the Assembler
6299 @c prevent bad page break with this line
6300 You can pass options to the assembler.
6303 @item -Wa,@var{option}
6305 Pass @var{option} as an option to the assembler. If @var{option}
6306 contains commas, it is split into multiple options at the commas.
6308 @item -Xassembler @var{option}
6310 Pass @var{option} as an option to the assembler. You can use this to
6311 supply system-specific assembler options which GCC does not know how to
6314 If you want to pass an option that takes an argument, you must use
6315 @option{-Xassembler} twice, once for the option and once for the argument.
6320 @section Options for Linking
6321 @cindex link options
6322 @cindex options, linking
6324 These options come into play when the compiler links object files into
6325 an executable output file. They are meaningless if the compiler is
6326 not doing a link step.
6330 @item @var{object-file-name}
6331 A file name that does not end in a special recognized suffix is
6332 considered to name an object file or library. (Object files are
6333 distinguished from libraries by the linker according to the file
6334 contents.) If linking is done, these object files are used as input
6343 If any of these options is used, then the linker is not run, and
6344 object file names should not be used as arguments. @xref{Overall
6348 @item -l@var{library}
6349 @itemx -l @var{library}
6351 Search the library named @var{library} when linking. (The second
6352 alternative with the library as a separate argument is only for
6353 POSIX compliance and is not recommended.)
6355 It makes a difference where in the command you write this option; the
6356 linker searches and processes libraries and object files in the order they
6357 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
6358 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
6359 to functions in @samp{z}, those functions may not be loaded.
6361 The linker searches a standard list of directories for the library,
6362 which is actually a file named @file{lib@var{library}.a}. The linker
6363 then uses this file as if it had been specified precisely by name.
6365 The directories searched include several standard system directories
6366 plus any that you specify with @option{-L}.
6368 Normally the files found this way are library files---archive files
6369 whose members are object files. The linker handles an archive file by
6370 scanning through it for members which define symbols that have so far
6371 been referenced but not defined. But if the file that is found is an
6372 ordinary object file, it is linked in the usual fashion. The only
6373 difference between using an @option{-l} option and specifying a file name
6374 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
6375 and searches several directories.
6379 You need this special case of the @option{-l} option in order to
6380 link an Objective-C or Objective-C++ program.
6383 @opindex nostartfiles
6384 Do not use the standard system startup files when linking.
6385 The standard system libraries are used normally, unless @option{-nostdlib}
6386 or @option{-nodefaultlibs} is used.
6388 @item -nodefaultlibs
6389 @opindex nodefaultlibs
6390 Do not use the standard system libraries when linking.
6391 Only the libraries you specify will be passed to the linker.
6392 The standard startup files are used normally, unless @option{-nostartfiles}
6393 is used. The compiler may generate calls to @code{memcmp},
6394 @code{memset}, @code{memcpy} and @code{memmove}.
6395 These entries are usually resolved by entries in
6396 libc. These entry points should be supplied through some other
6397 mechanism when this option is specified.
6401 Do not use the standard system startup files or libraries when linking.
6402 No startup files and only the libraries you specify will be passed to
6403 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
6404 @code{memcpy} and @code{memmove}.
6405 These entries are usually resolved by entries in
6406 libc. These entry points should be supplied through some other
6407 mechanism when this option is specified.
6409 @cindex @option{-lgcc}, use with @option{-nostdlib}
6410 @cindex @option{-nostdlib} and unresolved references
6411 @cindex unresolved references and @option{-nostdlib}
6412 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
6413 @cindex @option{-nodefaultlibs} and unresolved references
6414 @cindex unresolved references and @option{-nodefaultlibs}
6415 One of the standard libraries bypassed by @option{-nostdlib} and
6416 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
6417 that GCC uses to overcome shortcomings of particular machines, or special
6418 needs for some languages.
6419 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
6420 Collection (GCC) Internals},
6421 for more discussion of @file{libgcc.a}.)
6422 In most cases, you need @file{libgcc.a} even when you want to avoid
6423 other standard libraries. In other words, when you specify @option{-nostdlib}
6424 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
6425 This ensures that you have no unresolved references to internal GCC
6426 library subroutines. (For example, @samp{__main}, used to ensure C++
6427 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
6428 GNU Compiler Collection (GCC) Internals}.)
6432 Produce a position independent executable on targets which support it.
6433 For predictable results, you must also specify the same set of options
6434 that were used to generate code (@option{-fpie}, @option{-fPIE},
6435 or model suboptions) when you specify this option.
6439 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
6440 that support it. This instructs the linker to add all symbols, not
6441 only used ones, to the dynamic symbol table. This option is needed
6442 for some uses of @code{dlopen} or to allow obtaining backtraces
6443 from within a program.
6447 Remove all symbol table and relocation information from the executable.
6451 On systems that support dynamic linking, this prevents linking with the shared
6452 libraries. On other systems, this option has no effect.
6456 Produce a shared object which can then be linked with other objects to
6457 form an executable. Not all systems support this option. For predictable
6458 results, you must also specify the same set of options that were used to
6459 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
6460 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
6461 needs to build supplementary stub code for constructors to work. On
6462 multi-libbed systems, @samp{gcc -shared} must select the correct support
6463 libraries to link against. Failing to supply the correct flags may lead
6464 to subtle defects. Supplying them in cases where they are not necessary
6467 @item -shared-libgcc
6468 @itemx -static-libgcc
6469 @opindex shared-libgcc
6470 @opindex static-libgcc
6471 On systems that provide @file{libgcc} as a shared library, these options
6472 force the use of either the shared or static version respectively.
6473 If no shared version of @file{libgcc} was built when the compiler was
6474 configured, these options have no effect.
6476 There are several situations in which an application should use the
6477 shared @file{libgcc} instead of the static version. The most common
6478 of these is when the application wishes to throw and catch exceptions
6479 across different shared libraries. In that case, each of the libraries
6480 as well as the application itself should use the shared @file{libgcc}.
6482 Therefore, the G++ and GCJ drivers automatically add
6483 @option{-shared-libgcc} whenever you build a shared library or a main
6484 executable, because C++ and Java programs typically use exceptions, so
6485 this is the right thing to do.
6487 If, instead, you use the GCC driver to create shared libraries, you may
6488 find that they will not always be linked with the shared @file{libgcc}.
6489 If GCC finds, at its configuration time, that you have a non-GNU linker
6490 or a GNU linker that does not support option @option{--eh-frame-hdr},
6491 it will link the shared version of @file{libgcc} into shared libraries
6492 by default. Otherwise, it will take advantage of the linker and optimize
6493 away the linking with the shared version of @file{libgcc}, linking with
6494 the static version of libgcc by default. This allows exceptions to
6495 propagate through such shared libraries, without incurring relocation
6496 costs at library load time.
6498 However, if a library or main executable is supposed to throw or catch
6499 exceptions, you must link it using the G++ or GCJ driver, as appropriate
6500 for the languages used in the program, or using the option
6501 @option{-shared-libgcc}, such that it is linked with the shared
6506 Bind references to global symbols when building a shared object. Warn
6507 about any unresolved references (unless overridden by the link editor
6508 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
6511 @item -Xlinker @var{option}
6513 Pass @var{option} as an option to the linker. You can use this to
6514 supply system-specific linker options which GCC does not know how to
6517 If you want to pass an option that takes an argument, you must use
6518 @option{-Xlinker} twice, once for the option and once for the argument.
6519 For example, to pass @option{-assert definitions}, you must write
6520 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
6521 @option{-Xlinker "-assert definitions"}, because this passes the entire
6522 string as a single argument, which is not what the linker expects.
6524 @item -Wl,@var{option}
6526 Pass @var{option} as an option to the linker. If @var{option} contains
6527 commas, it is split into multiple options at the commas.
6529 @item -u @var{symbol}
6531 Pretend the symbol @var{symbol} is undefined, to force linking of
6532 library modules to define it. You can use @option{-u} multiple times with
6533 different symbols to force loading of additional library modules.
6536 @node Directory Options
6537 @section Options for Directory Search
6538 @cindex directory options
6539 @cindex options, directory search
6542 These options specify directories to search for header files, for
6543 libraries and for parts of the compiler:
6548 Add the directory @var{dir} to the head of the list of directories to be
6549 searched for header files. This can be used to override a system header
6550 file, substituting your own version, since these directories are
6551 searched before the system header file directories. However, you should
6552 not use this option to add directories that contain vendor-supplied
6553 system header files (use @option{-isystem} for that). If you use more than
6554 one @option{-I} option, the directories are scanned in left-to-right
6555 order; the standard system directories come after.
6557 If a standard system include directory, or a directory specified with
6558 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
6559 option will be ignored. The directory will still be searched but as a
6560 system directory at its normal position in the system include chain.
6561 This is to ensure that GCC's procedure to fix buggy system headers and
6562 the ordering for the include_next directive are not inadvertently changed.
6563 If you really need to change the search order for system directories,
6564 use the @option{-nostdinc} and/or @option{-isystem} options.
6566 @item -iquote@var{dir}
6568 Add the directory @var{dir} to the head of the list of directories to
6569 be searched for header files only for the case of @samp{#include
6570 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
6571 otherwise just like @option{-I}.
6575 Add directory @var{dir} to the list of directories to be searched
6578 @item -B@var{prefix}
6580 This option specifies where to find the executables, libraries,
6581 include files, and data files of the compiler itself.
6583 The compiler driver program runs one or more of the subprograms
6584 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
6585 @var{prefix} as a prefix for each program it tries to run, both with and
6586 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
6588 For each subprogram to be run, the compiler driver first tries the
6589 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
6590 was not specified, the driver tries two standard prefixes, which are
6591 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
6592 those results in a file name that is found, the unmodified program
6593 name is searched for using the directories specified in your
6594 @env{PATH} environment variable.
6596 The compiler will check to see if the path provided by the @option{-B}
6597 refers to a directory, and if necessary it will add a directory
6598 separator character at the end of the path.
6600 @option{-B} prefixes that effectively specify directory names also apply
6601 to libraries in the linker, because the compiler translates these
6602 options into @option{-L} options for the linker. They also apply to
6603 includes files in the preprocessor, because the compiler translates these
6604 options into @option{-isystem} options for the preprocessor. In this case,
6605 the compiler appends @samp{include} to the prefix.
6607 The run-time support file @file{libgcc.a} can also be searched for using
6608 the @option{-B} prefix, if needed. If it is not found there, the two
6609 standard prefixes above are tried, and that is all. The file is left
6610 out of the link if it is not found by those means.
6612 Another way to specify a prefix much like the @option{-B} prefix is to use
6613 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
6616 As a special kludge, if the path provided by @option{-B} is
6617 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
6618 9, then it will be replaced by @file{[dir/]include}. This is to help
6619 with boot-strapping the compiler.
6621 @item -specs=@var{file}
6623 Process @var{file} after the compiler reads in the standard @file{specs}
6624 file, in order to override the defaults that the @file{gcc} driver
6625 program uses when determining what switches to pass to @file{cc1},
6626 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
6627 @option{-specs=@var{file}} can be specified on the command line, and they
6628 are processed in order, from left to right.
6630 @item --sysroot=@var{dir}
6632 Use @var{dir} as the logical root directory for headers and libraries.
6633 For example, if the compiler would normally search for headers in
6634 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
6635 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
6637 If you use both this option and the @option{-isysroot} option, then
6638 the @option{--sysroot} option will apply to libraries, but the
6639 @option{-isysroot} option will apply to header files.
6641 The GNU linker (beginning with version 2.16) has the necessary support
6642 for this option. If your linker does not support this option, the
6643 header file aspect of @option{--sysroot} will still work, but the
6644 library aspect will not.
6648 This option has been deprecated. Please use @option{-iquote} instead for
6649 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
6650 Any directories you specify with @option{-I} options before the @option{-I-}
6651 option are searched only for the case of @samp{#include "@var{file}"};
6652 they are not searched for @samp{#include <@var{file}>}.
6654 If additional directories are specified with @option{-I} options after
6655 the @option{-I-}, these directories are searched for all @samp{#include}
6656 directives. (Ordinarily @emph{all} @option{-I} directories are used
6659 In addition, the @option{-I-} option inhibits the use of the current
6660 directory (where the current input file came from) as the first search
6661 directory for @samp{#include "@var{file}"}. There is no way to
6662 override this effect of @option{-I-}. With @option{-I.} you can specify
6663 searching the directory which was current when the compiler was
6664 invoked. That is not exactly the same as what the preprocessor does
6665 by default, but it is often satisfactory.
6667 @option{-I-} does not inhibit the use of the standard system directories
6668 for header files. Thus, @option{-I-} and @option{-nostdinc} are
6675 @section Specifying subprocesses and the switches to pass to them
6678 @command{gcc} is a driver program. It performs its job by invoking a
6679 sequence of other programs to do the work of compiling, assembling and
6680 linking. GCC interprets its command-line parameters and uses these to
6681 deduce which programs it should invoke, and which command-line options
6682 it ought to place on their command lines. This behavior is controlled
6683 by @dfn{spec strings}. In most cases there is one spec string for each
6684 program that GCC can invoke, but a few programs have multiple spec
6685 strings to control their behavior. The spec strings built into GCC can
6686 be overridden by using the @option{-specs=} command-line switch to specify
6689 @dfn{Spec files} are plaintext files that are used to construct spec
6690 strings. They consist of a sequence of directives separated by blank
6691 lines. The type of directive is determined by the first non-whitespace
6692 character on the line and it can be one of the following:
6695 @item %@var{command}
6696 Issues a @var{command} to the spec file processor. The commands that can
6700 @item %include <@var{file}>
6702 Search for @var{file} and insert its text at the current point in the
6705 @item %include_noerr <@var{file}>
6706 @cindex %include_noerr
6707 Just like @samp{%include}, but do not generate an error message if the include
6708 file cannot be found.
6710 @item %rename @var{old_name} @var{new_name}
6712 Rename the spec string @var{old_name} to @var{new_name}.
6716 @item *[@var{spec_name}]:
6717 This tells the compiler to create, override or delete the named spec
6718 string. All lines after this directive up to the next directive or
6719 blank line are considered to be the text for the spec string. If this
6720 results in an empty string then the spec will be deleted. (Or, if the
6721 spec did not exist, then nothing will happened.) Otherwise, if the spec
6722 does not currently exist a new spec will be created. If the spec does
6723 exist then its contents will be overridden by the text of this
6724 directive, unless the first character of that text is the @samp{+}
6725 character, in which case the text will be appended to the spec.
6727 @item [@var{suffix}]:
6728 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6729 and up to the next directive or blank line are considered to make up the
6730 spec string for the indicated suffix. When the compiler encounters an
6731 input file with the named suffix, it will processes the spec string in
6732 order to work out how to compile that file. For example:
6739 This says that any input file whose name ends in @samp{.ZZ} should be
6740 passed to the program @samp{z-compile}, which should be invoked with the
6741 command-line switch @option{-input} and with the result of performing the
6742 @samp{%i} substitution. (See below.)
6744 As an alternative to providing a spec string, the text that follows a
6745 suffix directive can be one of the following:
6748 @item @@@var{language}
6749 This says that the suffix is an alias for a known @var{language}. This is
6750 similar to using the @option{-x} command-line switch to GCC to specify a
6751 language explicitly. For example:
6758 Says that .ZZ files are, in fact, C++ source files.
6761 This causes an error messages saying:
6764 @var{name} compiler not installed on this system.
6768 GCC already has an extensive list of suffixes built into it.
6769 This directive will add an entry to the end of the list of suffixes, but
6770 since the list is searched from the end backwards, it is effectively
6771 possible to override earlier entries using this technique.
6775 GCC has the following spec strings built into it. Spec files can
6776 override these strings or create their own. Note that individual
6777 targets can also add their own spec strings to this list.
6780 asm Options to pass to the assembler
6781 asm_final Options to pass to the assembler post-processor
6782 cpp Options to pass to the C preprocessor
6783 cc1 Options to pass to the C compiler
6784 cc1plus Options to pass to the C++ compiler
6785 endfile Object files to include at the end of the link
6786 link Options to pass to the linker
6787 lib Libraries to include on the command line to the linker
6788 libgcc Decides which GCC support library to pass to the linker
6789 linker Sets the name of the linker
6790 predefines Defines to be passed to the C preprocessor
6791 signed_char Defines to pass to CPP to say whether @code{char} is signed
6793 startfile Object files to include at the start of the link
6796 Here is a small example of a spec file:
6802 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6805 This example renames the spec called @samp{lib} to @samp{old_lib} and
6806 then overrides the previous definition of @samp{lib} with a new one.
6807 The new definition adds in some extra command-line options before
6808 including the text of the old definition.
6810 @dfn{Spec strings} are a list of command-line options to be passed to their
6811 corresponding program. In addition, the spec strings can contain
6812 @samp{%}-prefixed sequences to substitute variable text or to
6813 conditionally insert text into the command line. Using these constructs
6814 it is possible to generate quite complex command lines.
6816 Here is a table of all defined @samp{%}-sequences for spec
6817 strings. Note that spaces are not generated automatically around the
6818 results of expanding these sequences. Therefore you can concatenate them
6819 together or combine them with constant text in a single argument.
6823 Substitute one @samp{%} into the program name or argument.
6826 Substitute the name of the input file being processed.
6829 Substitute the basename of the input file being processed.
6830 This is the substring up to (and not including) the last period
6831 and not including the directory.
6834 This is the same as @samp{%b}, but include the file suffix (text after
6838 Marks the argument containing or following the @samp{%d} as a
6839 temporary file name, so that that file will be deleted if GCC exits
6840 successfully. Unlike @samp{%g}, this contributes no text to the
6843 @item %g@var{suffix}
6844 Substitute a file name that has suffix @var{suffix} and is chosen
6845 once per compilation, and mark the argument in the same way as
6846 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
6847 name is now chosen in a way that is hard to predict even when previously
6848 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
6849 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
6850 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
6851 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
6852 was simply substituted with a file name chosen once per compilation,
6853 without regard to any appended suffix (which was therefore treated
6854 just like ordinary text), making such attacks more likely to succeed.
6856 @item %u@var{suffix}
6857 Like @samp{%g}, but generates a new temporary file name even if
6858 @samp{%u@var{suffix}} was already seen.
6860 @item %U@var{suffix}
6861 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6862 new one if there is no such last file name. In the absence of any
6863 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6864 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6865 would involve the generation of two distinct file names, one
6866 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6867 simply substituted with a file name chosen for the previous @samp{%u},
6868 without regard to any appended suffix.
6870 @item %j@var{suffix}
6871 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6872 writable, and if save-temps is off; otherwise, substitute the name
6873 of a temporary file, just like @samp{%u}. This temporary file is not
6874 meant for communication between processes, but rather as a junk
6877 @item %|@var{suffix}
6878 @itemx %m@var{suffix}
6879 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6880 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6881 all. These are the two most common ways to instruct a program that it
6882 should read from standard input or write to standard output. If you
6883 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6884 construct: see for example @file{f/lang-specs.h}.
6886 @item %.@var{SUFFIX}
6887 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6888 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6889 terminated by the next space or %.
6892 Marks the argument containing or following the @samp{%w} as the
6893 designated output file of this compilation. This puts the argument
6894 into the sequence of arguments that @samp{%o} will substitute later.
6897 Substitutes the names of all the output files, with spaces
6898 automatically placed around them. You should write spaces
6899 around the @samp{%o} as well or the results are undefined.
6900 @samp{%o} is for use in the specs for running the linker.
6901 Input files whose names have no recognized suffix are not compiled
6902 at all, but they are included among the output files, so they will
6906 Substitutes the suffix for object files. Note that this is
6907 handled specially when it immediately follows @samp{%g, %u, or %U},
6908 because of the need for those to form complete file names. The
6909 handling is such that @samp{%O} is treated exactly as if it had already
6910 been substituted, except that @samp{%g, %u, and %U} do not currently
6911 support additional @var{suffix} characters following @samp{%O} as they would
6912 following, for example, @samp{.o}.
6915 Substitutes the standard macro predefinitions for the
6916 current target machine. Use this when running @code{cpp}.
6919 Like @samp{%p}, but puts @samp{__} before and after the name of each
6920 predefined macro, except for macros that start with @samp{__} or with
6921 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
6925 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
6926 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
6927 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
6928 and @option{-imultilib} as necessary.
6931 Current argument is the name of a library or startup file of some sort.
6932 Search for that file in a standard list of directories and substitute
6933 the full name found.
6936 Print @var{str} as an error message. @var{str} is terminated by a newline.
6937 Use this when inconsistent options are detected.
6940 Substitute the contents of spec string @var{name} at this point.
6943 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
6945 @item %x@{@var{option}@}
6946 Accumulate an option for @samp{%X}.
6949 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
6953 Output the accumulated assembler options specified by @option{-Wa}.
6956 Output the accumulated preprocessor options specified by @option{-Wp}.
6959 Process the @code{asm} spec. This is used to compute the
6960 switches to be passed to the assembler.
6963 Process the @code{asm_final} spec. This is a spec string for
6964 passing switches to an assembler post-processor, if such a program is
6968 Process the @code{link} spec. This is the spec for computing the
6969 command line passed to the linker. Typically it will make use of the
6970 @samp{%L %G %S %D and %E} sequences.
6973 Dump out a @option{-L} option for each directory that GCC believes might
6974 contain startup files. If the target supports multilibs then the
6975 current multilib directory will be prepended to each of these paths.
6978 Process the @code{lib} spec. This is a spec string for deciding which
6979 libraries should be included on the command line to the linker.
6982 Process the @code{libgcc} spec. This is a spec string for deciding
6983 which GCC support library should be included on the command line to the linker.
6986 Process the @code{startfile} spec. This is a spec for deciding which
6987 object files should be the first ones passed to the linker. Typically
6988 this might be a file named @file{crt0.o}.
6991 Process the @code{endfile} spec. This is a spec string that specifies
6992 the last object files that will be passed to the linker.
6995 Process the @code{cpp} spec. This is used to construct the arguments
6996 to be passed to the C preprocessor.
6999 Process the @code{cc1} spec. This is used to construct the options to be
7000 passed to the actual C compiler (@samp{cc1}).
7003 Process the @code{cc1plus} spec. This is used to construct the options to be
7004 passed to the actual C++ compiler (@samp{cc1plus}).
7007 Substitute the variable part of a matched option. See below.
7008 Note that each comma in the substituted string is replaced by
7012 Remove all occurrences of @code{-S} from the command line. Note---this
7013 command is position dependent. @samp{%} commands in the spec string
7014 before this one will see @code{-S}, @samp{%} commands in the spec string
7015 after this one will not.
7017 @item %:@var{function}(@var{args})
7018 Call the named function @var{function}, passing it @var{args}.
7019 @var{args} is first processed as a nested spec string, then split
7020 into an argument vector in the usual fashion. The function returns
7021 a string which is processed as if it had appeared literally as part
7022 of the current spec.
7024 The following built-in spec functions are provided:
7027 @item @code{if-exists}
7028 The @code{if-exists} spec function takes one argument, an absolute
7029 pathname to a file. If the file exists, @code{if-exists} returns the
7030 pathname. Here is a small example of its usage:
7034 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7037 @item @code{if-exists-else}
7038 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7039 spec function, except that it takes two arguments. The first argument is
7040 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7041 returns the pathname. If it does not exist, it returns the second argument.
7042 This way, @code{if-exists-else} can be used to select one file or another,
7043 based on the existence of the first. Here is a small example of its usage:
7047 crt0%O%s %:if-exists(crti%O%s) \
7048 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
7051 @item @code{replace-outfile}
7052 The @code{replace-outfile} spec function takes two arguments. It looks for the
7053 first argument in the outfiles array and replaces it with the second argument. Here
7054 is a small example of its usage:
7057 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7063 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7064 If that switch was not specified, this substitutes nothing. Note that
7065 the leading dash is omitted when specifying this option, and it is
7066 automatically inserted if the substitution is performed. Thus the spec
7067 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7068 and would output the command line option @option{-foo}.
7070 @item %W@{@code{S}@}
7071 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7074 @item %@{@code{S}*@}
7075 Substitutes all the switches specified to GCC whose names start
7076 with @code{-S}, but which also take an argument. This is used for
7077 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7078 GCC considers @option{-o foo} as being
7079 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7080 text, including the space. Thus two arguments would be generated.
7082 @item %@{@code{S}*&@code{T}*@}
7083 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7084 (the order of @code{S} and @code{T} in the spec is not significant).
7085 There can be any number of ampersand-separated variables; for each the
7086 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
7088 @item %@{@code{S}:@code{X}@}
7089 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
7091 @item %@{!@code{S}:@code{X}@}
7092 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
7094 @item %@{@code{S}*:@code{X}@}
7095 Substitutes @code{X} if one or more switches whose names start with
7096 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
7097 once, no matter how many such switches appeared. However, if @code{%*}
7098 appears somewhere in @code{X}, then @code{X} will be substituted once
7099 for each matching switch, with the @code{%*} replaced by the part of
7100 that switch that matched the @code{*}.
7102 @item %@{.@code{S}:@code{X}@}
7103 Substitutes @code{X}, if processing a file with suffix @code{S}.
7105 @item %@{!.@code{S}:@code{X}@}
7106 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
7108 @item %@{@code{S}|@code{P}:@code{X}@}
7109 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
7110 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
7111 although they have a stronger binding than the @samp{|}. If @code{%*}
7112 appears in @code{X}, all of the alternatives must be starred, and only
7113 the first matching alternative is substituted.
7115 For example, a spec string like this:
7118 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
7121 will output the following command-line options from the following input
7122 command-line options:
7127 -d fred.c -foo -baz -boggle
7128 -d jim.d -bar -baz -boggle
7131 @item %@{S:X; T:Y; :D@}
7133 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
7134 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
7135 be as many clauses as you need. This may be combined with @code{.},
7136 @code{!}, @code{|}, and @code{*} as needed.
7141 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
7142 construct may contain other nested @samp{%} constructs or spaces, or
7143 even newlines. They are processed as usual, as described above.
7144 Trailing white space in @code{X} is ignored. White space may also
7145 appear anywhere on the left side of the colon in these constructs,
7146 except between @code{.} or @code{*} and the corresponding word.
7148 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
7149 handled specifically in these constructs. If another value of
7150 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7151 @option{-W} switch is found later in the command line, the earlier
7152 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7153 just one letter, which passes all matching options.
7155 The character @samp{|} at the beginning of the predicate text is used to
7156 indicate that a command should be piped to the following command, but
7157 only if @option{-pipe} is specified.
7159 It is built into GCC which switches take arguments and which do not.
7160 (You might think it would be useful to generalize this to allow each
7161 compiler's spec to say which switches take arguments. But this cannot
7162 be done in a consistent fashion. GCC cannot even decide which input
7163 files have been specified without knowing which switches take arguments,
7164 and it must know which input files to compile in order to tell which
7167 GCC also knows implicitly that arguments starting in @option{-l} are to be
7168 treated as compiler output files, and passed to the linker in their
7169 proper position among the other output files.
7171 @c man begin OPTIONS
7173 @node Target Options
7174 @section Specifying Target Machine and Compiler Version
7175 @cindex target options
7176 @cindex cross compiling
7177 @cindex specifying machine version
7178 @cindex specifying compiler version and target machine
7179 @cindex compiler version, specifying
7180 @cindex target machine, specifying
7182 The usual way to run GCC is to run the executable called @file{gcc}, or
7183 @file{<machine>-gcc} when cross-compiling, or
7184 @file{<machine>-gcc-<version>} to run a version other than the one that
7185 was installed last. Sometimes this is inconvenient, so GCC provides
7186 options that will switch to another cross-compiler or version.
7189 @item -b @var{machine}
7191 The argument @var{machine} specifies the target machine for compilation.
7193 The value to use for @var{machine} is the same as was specified as the
7194 machine type when configuring GCC as a cross-compiler. For
7195 example, if a cross-compiler was configured with @samp{configure
7196 arm-elf}, meaning to compile for an arm processor with elf binaries,
7197 then you would specify @option{-b arm-elf} to run that cross compiler.
7198 Because there are other options beginning with @option{-b}, the
7199 configuration must contain a hyphen.
7201 @item -V @var{version}
7203 The argument @var{version} specifies which version of GCC to run.
7204 This is useful when multiple versions are installed. For example,
7205 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
7208 The @option{-V} and @option{-b} options work by running the
7209 @file{<machine>-gcc-<version>} executable, so there's no real reason to
7210 use them if you can just run that directly.
7212 @node Submodel Options
7213 @section Hardware Models and Configurations
7214 @cindex submodel options
7215 @cindex specifying hardware config
7216 @cindex hardware models and configurations, specifying
7217 @cindex machine dependent options
7219 Earlier we discussed the standard option @option{-b} which chooses among
7220 different installed compilers for completely different target
7221 machines, such as VAX vs.@: 68000 vs.@: 80386.
7223 In addition, each of these target machine types can have its own
7224 special options, starting with @samp{-m}, to choose among various
7225 hardware models or configurations---for example, 68010 vs 68020,
7226 floating coprocessor or none. A single installed version of the
7227 compiler can compile for any model or configuration, according to the
7230 Some configurations of the compiler also support additional special
7231 options, usually for compatibility with other compilers on the same
7234 @c This list is ordered alphanumerically by subsection name.
7235 @c It should be the same order and spelling as these options are listed
7236 @c in Machine Dependent Options
7242 * Blackfin Options::
7246 * DEC Alpha Options::
7247 * DEC Alpha/VMS Options::
7249 * GNU/Linux Options::
7252 * i386 and x86-64 Options::
7265 * RS/6000 and PowerPC Options::
7266 * S/390 and zSeries Options::
7269 * System V Options::
7270 * TMS320C3x/C4x Options::
7274 * Xstormy16 Options::
7280 @subsection ARC Options
7283 These options are defined for ARC implementations:
7288 Compile code for little endian mode. This is the default.
7292 Compile code for big endian mode.
7295 @opindex mmangle-cpu
7296 Prepend the name of the cpu to all public symbol names.
7297 In multiple-processor systems, there are many ARC variants with different
7298 instruction and register set characteristics. This flag prevents code
7299 compiled for one cpu to be linked with code compiled for another.
7300 No facility exists for handling variants that are ``almost identical''.
7301 This is an all or nothing option.
7303 @item -mcpu=@var{cpu}
7305 Compile code for ARC variant @var{cpu}.
7306 Which variants are supported depend on the configuration.
7307 All variants support @option{-mcpu=base}, this is the default.
7309 @item -mtext=@var{text-section}
7310 @itemx -mdata=@var{data-section}
7311 @itemx -mrodata=@var{readonly-data-section}
7315 Put functions, data, and readonly data in @var{text-section},
7316 @var{data-section}, and @var{readonly-data-section} respectively
7317 by default. This can be overridden with the @code{section} attribute.
7318 @xref{Variable Attributes}.
7323 @subsection ARM Options
7326 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
7330 @item -mabi=@var{name}
7332 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
7333 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
7336 @opindex mapcs-frame
7337 Generate a stack frame that is compliant with the ARM Procedure Call
7338 Standard for all functions, even if this is not strictly necessary for
7339 correct execution of the code. Specifying @option{-fomit-frame-pointer}
7340 with this option will cause the stack frames not to be generated for
7341 leaf functions. The default is @option{-mno-apcs-frame}.
7345 This is a synonym for @option{-mapcs-frame}.
7348 @c not currently implemented
7349 @item -mapcs-stack-check
7350 @opindex mapcs-stack-check
7351 Generate code to check the amount of stack space available upon entry to
7352 every function (that actually uses some stack space). If there is
7353 insufficient space available then either the function
7354 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
7355 called, depending upon the amount of stack space required. The run time
7356 system is required to provide these functions. The default is
7357 @option{-mno-apcs-stack-check}, since this produces smaller code.
7359 @c not currently implemented
7361 @opindex mapcs-float
7362 Pass floating point arguments using the float point registers. This is
7363 one of the variants of the APCS@. This option is recommended if the
7364 target hardware has a floating point unit or if a lot of floating point
7365 arithmetic is going to be performed by the code. The default is
7366 @option{-mno-apcs-float}, since integer only code is slightly increased in
7367 size if @option{-mapcs-float} is used.
7369 @c not currently implemented
7370 @item -mapcs-reentrant
7371 @opindex mapcs-reentrant
7372 Generate reentrant, position independent code. The default is
7373 @option{-mno-apcs-reentrant}.
7376 @item -mthumb-interwork
7377 @opindex mthumb-interwork
7378 Generate code which supports calling between the ARM and Thumb
7379 instruction sets. Without this option the two instruction sets cannot
7380 be reliably used inside one program. The default is
7381 @option{-mno-thumb-interwork}, since slightly larger code is generated
7382 when @option{-mthumb-interwork} is specified.
7384 @item -mno-sched-prolog
7385 @opindex mno-sched-prolog
7386 Prevent the reordering of instructions in the function prolog, or the
7387 merging of those instruction with the instructions in the function's
7388 body. This means that all functions will start with a recognizable set
7389 of instructions (or in fact one of a choice from a small set of
7390 different function prologues), and this information can be used to
7391 locate the start if functions inside an executable piece of code. The
7392 default is @option{-msched-prolog}.
7395 @opindex mhard-float
7396 Generate output containing floating point instructions. This is the
7400 @opindex msoft-float
7401 Generate output containing library calls for floating point.
7402 @strong{Warning:} the requisite libraries are not available for all ARM
7403 targets. Normally the facilities of the machine's usual C compiler are
7404 used, but this cannot be done directly in cross-compilation. You must make
7405 your own arrangements to provide suitable library functions for
7408 @option{-msoft-float} changes the calling convention in the output file;
7409 therefore, it is only useful if you compile @emph{all} of a program with
7410 this option. In particular, you need to compile @file{libgcc.a}, the
7411 library that comes with GCC, with @option{-msoft-float} in order for
7414 @item -mfloat-abi=@var{name}
7416 Specifies which ABI to use for floating point values. Permissible values
7417 are: @samp{soft}, @samp{softfp} and @samp{hard}.
7419 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
7420 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
7421 of floating point instructions, but still uses the soft-float calling
7424 @item -mlittle-endian
7425 @opindex mlittle-endian
7426 Generate code for a processor running in little-endian mode. This is
7427 the default for all standard configurations.
7430 @opindex mbig-endian
7431 Generate code for a processor running in big-endian mode; the default is
7432 to compile code for a little-endian processor.
7434 @item -mwords-little-endian
7435 @opindex mwords-little-endian
7436 This option only applies when generating code for big-endian processors.
7437 Generate code for a little-endian word order but a big-endian byte
7438 order. That is, a byte order of the form @samp{32107654}. Note: this
7439 option should only be used if you require compatibility with code for
7440 big-endian ARM processors generated by versions of the compiler prior to
7443 @item -mcpu=@var{name}
7445 This specifies the name of the target ARM processor. GCC uses this name
7446 to determine what kind of instructions it can emit when generating
7447 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
7448 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
7449 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
7450 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
7451 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
7452 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
7453 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
7454 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
7455 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
7456 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
7457 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
7458 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
7459 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
7460 @samp{arm1176jz-s}, @samp{arm1176jzf-s}, @samp{xscale}, @samp{iwmmxt},
7463 @itemx -mtune=@var{name}
7465 This option is very similar to the @option{-mcpu=} option, except that
7466 instead of specifying the actual target processor type, and hence
7467 restricting which instructions can be used, it specifies that GCC should
7468 tune the performance of the code as if the target were of the type
7469 specified in this option, but still choosing the instructions that it
7470 will generate based on the cpu specified by a @option{-mcpu=} option.
7471 For some ARM implementations better performance can be obtained by using
7474 @item -march=@var{name}
7476 This specifies the name of the target ARM architecture. GCC uses this
7477 name to determine what kind of instructions it can emit when generating
7478 assembly code. This option can be used in conjunction with or instead
7479 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
7480 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
7481 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
7482 @samp{iwmmxt}, @samp{ep9312}.
7484 @item -mfpu=@var{name}
7485 @itemx -mfpe=@var{number}
7486 @itemx -mfp=@var{number}
7490 This specifies what floating point hardware (or hardware emulation) is
7491 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
7492 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
7493 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
7494 with older versions of GCC@.
7496 If @option{-msoft-float} is specified this specifies the format of
7497 floating point values.
7499 @item -mstructure-size-boundary=@var{n}
7500 @opindex mstructure-size-boundary
7501 The size of all structures and unions will be rounded up to a multiple
7502 of the number of bits set by this option. Permissible values are 8, 32
7503 and 64. The default value varies for different toolchains. For the COFF
7504 targeted toolchain the default value is 8. A value of 64 is only allowed
7505 if the underlying ABI supports it.
7507 Specifying the larger number can produce faster, more efficient code, but
7508 can also increase the size of the program. Different values are potentially
7509 incompatible. Code compiled with one value cannot necessarily expect to
7510 work with code or libraries compiled with another value, if they exchange
7511 information using structures or unions.
7513 @item -mabort-on-noreturn
7514 @opindex mabort-on-noreturn
7515 Generate a call to the function @code{abort} at the end of a
7516 @code{noreturn} function. It will be executed if the function tries to
7520 @itemx -mno-long-calls
7521 @opindex mlong-calls
7522 @opindex mno-long-calls
7523 Tells the compiler to perform function calls by first loading the
7524 address of the function into a register and then performing a subroutine
7525 call on this register. This switch is needed if the target function
7526 will lie outside of the 64 megabyte addressing range of the offset based
7527 version of subroutine call instruction.
7529 Even if this switch is enabled, not all function calls will be turned
7530 into long calls. The heuristic is that static functions, functions
7531 which have the @samp{short-call} attribute, functions that are inside
7532 the scope of a @samp{#pragma no_long_calls} directive and functions whose
7533 definitions have already been compiled within the current compilation
7534 unit, will not be turned into long calls. The exception to this rule is
7535 that weak function definitions, functions with the @samp{long-call}
7536 attribute or the @samp{section} attribute, and functions that are within
7537 the scope of a @samp{#pragma long_calls} directive, will always be
7538 turned into long calls.
7540 This feature is not enabled by default. Specifying
7541 @option{-mno-long-calls} will restore the default behavior, as will
7542 placing the function calls within the scope of a @samp{#pragma
7543 long_calls_off} directive. Note these switches have no effect on how
7544 the compiler generates code to handle function calls via function
7547 @item -mnop-fun-dllimport
7548 @opindex mnop-fun-dllimport
7549 Disable support for the @code{dllimport} attribute.
7551 @item -msingle-pic-base
7552 @opindex msingle-pic-base
7553 Treat the register used for PIC addressing as read-only, rather than
7554 loading it in the prologue for each function. The run-time system is
7555 responsible for initializing this register with an appropriate value
7556 before execution begins.
7558 @item -mpic-register=@var{reg}
7559 @opindex mpic-register
7560 Specify the register to be used for PIC addressing. The default is R10
7561 unless stack-checking is enabled, when R9 is used.
7563 @item -mcirrus-fix-invalid-insns
7564 @opindex mcirrus-fix-invalid-insns
7565 @opindex mno-cirrus-fix-invalid-insns
7566 Insert NOPs into the instruction stream to in order to work around
7567 problems with invalid Maverick instruction combinations. This option
7568 is only valid if the @option{-mcpu=ep9312} option has been used to
7569 enable generation of instructions for the Cirrus Maverick floating
7570 point co-processor. This option is not enabled by default, since the
7571 problem is only present in older Maverick implementations. The default
7572 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
7575 @item -mpoke-function-name
7576 @opindex mpoke-function-name
7577 Write the name of each function into the text section, directly
7578 preceding the function prologue. The generated code is similar to this:
7582 .ascii "arm_poke_function_name", 0
7585 .word 0xff000000 + (t1 - t0)
7586 arm_poke_function_name
7588 stmfd sp!, @{fp, ip, lr, pc@}
7592 When performing a stack backtrace, code can inspect the value of
7593 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
7594 location @code{pc - 12} and the top 8 bits are set, then we know that
7595 there is a function name embedded immediately preceding this location
7596 and has length @code{((pc[-3]) & 0xff000000)}.
7600 Generate code for the 16-bit Thumb instruction set. The default is to
7601 use the 32-bit ARM instruction set.
7604 @opindex mtpcs-frame
7605 Generate a stack frame that is compliant with the Thumb Procedure Call
7606 Standard for all non-leaf functions. (A leaf function is one that does
7607 not call any other functions.) The default is @option{-mno-tpcs-frame}.
7609 @item -mtpcs-leaf-frame
7610 @opindex mtpcs-leaf-frame
7611 Generate a stack frame that is compliant with the Thumb Procedure Call
7612 Standard for all leaf functions. (A leaf function is one that does
7613 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
7615 @item -mcallee-super-interworking
7616 @opindex mcallee-super-interworking
7617 Gives all externally visible functions in the file being compiled an ARM
7618 instruction set header which switches to Thumb mode before executing the
7619 rest of the function. This allows these functions to be called from
7620 non-interworking code.
7622 @item -mcaller-super-interworking
7623 @opindex mcaller-super-interworking
7624 Allows calls via function pointers (including virtual functions) to
7625 execute correctly regardless of whether the target code has been
7626 compiled for interworking or not. There is a small overhead in the cost
7627 of executing a function pointer if this option is enabled.
7629 @item -mtp=@var{name}
7631 Specify the access model for the thread local storage pointer. The valid
7632 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
7633 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
7634 (supported in the arm6k architecture), and @option{auto}, which uses the
7635 best available method for the selected processor. The default setting is
7641 @subsection AVR Options
7644 These options are defined for AVR implementations:
7647 @item -mmcu=@var{mcu}
7649 Specify ATMEL AVR instruction set or MCU type.
7651 Instruction set avr1 is for the minimal AVR core, not supported by the C
7652 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7653 attiny11, attiny12, attiny15, attiny28).
7655 Instruction set avr2 (default) is for the classic AVR core with up to
7656 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7657 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7658 at90c8534, at90s8535).
7660 Instruction set avr3 is for the classic AVR core with up to 128K program
7661 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
7663 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7664 memory space (MCU types: atmega8, atmega83, atmega85).
7666 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7667 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
7668 atmega64, atmega128, at43usb355, at94k).
7672 Output instruction sizes to the asm file.
7674 @item -minit-stack=@var{N}
7675 @opindex minit-stack
7676 Specify the initial stack address, which may be a symbol or numeric value,
7677 @samp{__stack} is the default.
7679 @item -mno-interrupts
7680 @opindex mno-interrupts
7681 Generated code is not compatible with hardware interrupts.
7682 Code size will be smaller.
7684 @item -mcall-prologues
7685 @opindex mcall-prologues
7686 Functions prologues/epilogues expanded as call to appropriate
7687 subroutines. Code size will be smaller.
7689 @item -mno-tablejump
7690 @opindex mno-tablejump
7691 Do not generate tablejump insns which sometimes increase code size.
7694 @opindex mtiny-stack
7695 Change only the low 8 bits of the stack pointer.
7699 Assume int to be 8 bit integer. This affects the sizes of all types: A
7700 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7701 and long long will be 4 bytes. Please note that this option does not
7702 comply to the C standards, but it will provide you with smaller code
7706 @node Blackfin Options
7707 @subsection Blackfin Options
7708 @cindex Blackfin Options
7711 @item -momit-leaf-frame-pointer
7712 @opindex momit-leaf-frame-pointer
7713 Don't keep the frame pointer in a register for leaf functions. This
7714 avoids the instructions to save, set up and restore frame pointers and
7715 makes an extra register available in leaf functions. The option
7716 @option{-fomit-frame-pointer} removes the frame pointer for all functions
7717 which might make debugging harder.
7719 @item -mspecld-anomaly
7720 @opindex mspecld-anomaly
7721 When enabled, the compiler will ensure that the generated code does not
7722 contain speculative loads after jump instructions. This option is enabled
7725 @item -mno-specld-anomaly
7726 @opindex mno-specld-anomaly
7727 Don't generate extra code to prevent speculative loads from occurring.
7729 @item -mcsync-anomaly
7730 @opindex mcsync-anomaly
7731 When enabled, the compiler will ensure that the generated code does not
7732 contain CSYNC or SSYNC instructions too soon after conditional branches.
7733 This option is enabled by default.
7735 @item -mno-csync-anomaly
7736 @opindex mno-csync-anomaly
7737 Don't generate extra code to prevent CSYNC or SSYNC instructions from
7738 occurring too soon after a conditional branch.
7742 When enabled, the compiler is free to take advantage of the knowledge that
7743 the entire program fits into the low 64k of memory.
7746 @opindex mno-low-64k
7747 Assume that the program is arbitrarily large. This is the default.
7749 @item -mid-shared-library
7750 @opindex mid-shared-library
7751 Generate code that supports shared libraries via the library ID method.
7752 This allows for execute in place and shared libraries in an environment
7753 without virtual memory management. This option implies @option{-fPIC}.
7755 @item -mno-id-shared-library
7756 @opindex mno-id-shared-library
7757 Generate code that doesn't assume ID based shared libraries are being used.
7758 This is the default.
7760 @item -mshared-library-id=n
7761 @opindex mshared-library-id
7762 Specified the identification number of the ID based shared library being
7763 compiled. Specifying a value of 0 will generate more compact code, specifying
7764 other values will force the allocation of that number to the current
7765 library but is no more space or time efficient than omitting this option.
7768 @itemx -mno-long-calls
7769 @opindex mlong-calls
7770 @opindex mno-long-calls
7771 Tells the compiler to perform function calls by first loading the
7772 address of the function into a register and then performing a subroutine
7773 call on this register. This switch is needed if the target function
7774 will lie outside of the 24 bit addressing range of the offset based
7775 version of subroutine call instruction.
7777 This feature is not enabled by default. Specifying
7778 @option{-mno-long-calls} will restore the default behavior. Note these
7779 switches have no effect on how the compiler generates code to handle
7780 function calls via function pointers.
7784 @subsection CRIS Options
7785 @cindex CRIS Options
7787 These options are defined specifically for the CRIS ports.
7790 @item -march=@var{architecture-type}
7791 @itemx -mcpu=@var{architecture-type}
7794 Generate code for the specified architecture. The choices for
7795 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7796 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
7797 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7800 @item -mtune=@var{architecture-type}
7802 Tune to @var{architecture-type} everything applicable about the generated
7803 code, except for the ABI and the set of available instructions. The
7804 choices for @var{architecture-type} are the same as for
7805 @option{-march=@var{architecture-type}}.
7807 @item -mmax-stack-frame=@var{n}
7808 @opindex mmax-stack-frame
7809 Warn when the stack frame of a function exceeds @var{n} bytes.
7811 @item -melinux-stacksize=@var{n}
7812 @opindex melinux-stacksize
7813 Only available with the @samp{cris-axis-aout} target. Arranges for
7814 indications in the program to the kernel loader that the stack of the
7815 program should be set to @var{n} bytes.
7821 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7822 @option{-march=v3} and @option{-march=v8} respectively.
7824 @item -mmul-bug-workaround
7825 @itemx -mno-mul-bug-workaround
7826 @opindex mmul-bug-workaround
7827 @opindex mno-mul-bug-workaround
7828 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7829 models where it applies. This option is active by default.
7833 Enable CRIS-specific verbose debug-related information in the assembly
7834 code. This option also has the effect to turn off the @samp{#NO_APP}
7835 formatted-code indicator to the assembler at the beginning of the
7840 Do not use condition-code results from previous instruction; always emit
7841 compare and test instructions before use of condition codes.
7843 @item -mno-side-effects
7844 @opindex mno-side-effects
7845 Do not emit instructions with side-effects in addressing modes other than
7849 @itemx -mno-stack-align
7851 @itemx -mno-data-align
7852 @itemx -mconst-align
7853 @itemx -mno-const-align
7854 @opindex mstack-align
7855 @opindex mno-stack-align
7856 @opindex mdata-align
7857 @opindex mno-data-align
7858 @opindex mconst-align
7859 @opindex mno-const-align
7860 These options (no-options) arranges (eliminate arrangements) for the
7861 stack-frame, individual data and constants to be aligned for the maximum
7862 single data access size for the chosen CPU model. The default is to
7863 arrange for 32-bit alignment. ABI details such as structure layout are
7864 not affected by these options.
7872 Similar to the stack- data- and const-align options above, these options
7873 arrange for stack-frame, writable data and constants to all be 32-bit,
7874 16-bit or 8-bit aligned. The default is 32-bit alignment.
7876 @item -mno-prologue-epilogue
7877 @itemx -mprologue-epilogue
7878 @opindex mno-prologue-epilogue
7879 @opindex mprologue-epilogue
7880 With @option{-mno-prologue-epilogue}, the normal function prologue and
7881 epilogue that sets up the stack-frame are omitted and no return
7882 instructions or return sequences are generated in the code. Use this
7883 option only together with visual inspection of the compiled code: no
7884 warnings or errors are generated when call-saved registers must be saved,
7885 or storage for local variable needs to be allocated.
7891 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
7892 instruction sequences that load addresses for functions from the PLT part
7893 of the GOT rather than (traditional on other architectures) calls to the
7894 PLT@. The default is @option{-mgotplt}.
7898 Legacy no-op option only recognized with the cris-axis-aout target.
7902 Legacy no-op option only recognized with the cris-axis-elf and
7903 cris-axis-linux-gnu targets.
7907 Only recognized with the cris-axis-aout target, where it selects a
7908 GNU/linux-like multilib, include files and instruction set for
7913 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
7917 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
7918 to link with input-output functions from a simulator library. Code,
7919 initialized data and zero-initialized data are allocated consecutively.
7923 Like @option{-sim}, but pass linker options to locate initialized data at
7924 0x40000000 and zero-initialized data at 0x80000000.
7928 @subsection CRX Options
7931 These options are defined specifically for the CRX ports.
7937 Enable the use of multiply-accumulate instructions. Disabled by default.
7941 Push instructions will be used to pass outgoing arguments when functions
7942 are called. Enabled by default.
7945 @node Darwin Options
7946 @subsection Darwin Options
7947 @cindex Darwin options
7949 These options are defined for all architectures running the Darwin operating
7952 FSF GCC on Darwin does not create ``fat'' object files; it will create
7953 an object file for the single architecture that it was built to
7954 target. Apple's GCC on Darwin does create ``fat'' files if multiple
7955 @option{-arch} options are used; it does so by running the compiler or
7956 linker multiple times and joining the results together with
7959 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
7960 @samp{i686}) is determined by the flags that specify the ISA
7961 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
7962 @option{-force_cpusubtype_ALL} option can be used to override this.
7964 The Darwin tools vary in their behavior when presented with an ISA
7965 mismatch. The assembler, @file{as}, will only permit instructions to
7966 be used that are valid for the subtype of the file it is generating,
7967 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
7968 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
7969 and print an error if asked to create a shared library with a less
7970 restrictive subtype than its input files (for instance, trying to put
7971 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
7972 for executables, @file{ld}, will quietly give the executable the most
7973 restrictive subtype of any of its input files.
7978 Add the framework directory @var{dir} to the head of the list of
7979 directories to be searched for header files. These directories are
7980 interleaved with those specified by @option{-I} options and are
7981 scanned in a left-to-right order.
7983 A framework directory is a directory with frameworks in it. A
7984 framework is a directory with a @samp{"Headers"} and/or
7985 @samp{"PrivateHeaders"} directory contained directly in it that ends
7986 in @samp{".framework"}. The name of a framework is the name of this
7987 directory excluding the @samp{".framework"}. Headers associated with
7988 the framework are found in one of those two directories, with
7989 @samp{"Headers"} being searched first. A subframework is a framework
7990 directory that is in a framework's @samp{"Frameworks"} directory.
7991 Includes of subframework headers can only appear in a header of a
7992 framework that contains the subframework, or in a sibling subframework
7993 header. Two subframeworks are siblings if they occur in the same
7994 framework. A subframework should not have the same name as a
7995 framework, a warning will be issued if this is violated. Currently a
7996 subframework cannot have subframeworks, in the future, the mechanism
7997 may be extended to support this. The standard frameworks can be found
7998 in @samp{"/System/Library/Frameworks"} and
7999 @samp{"/Library/Frameworks"}. An example include looks like
8000 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
8001 the name of the framework and header.h is found in the
8002 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
8006 Emit debugging information for symbols that are used. For STABS
8007 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
8008 This is by default ON@.
8012 Emit debugging information for all symbols and types.
8014 @item -mmacosx-version-min=@var{version}
8015 The earliest version of MacOS X that this executable will run on
8016 is @var{version}. Typical values of @var{version} include @code{10.1},
8017 @code{10.2}, and @code{10.3.9}.
8019 The default for this option is to make choices that seem to be most
8022 @item -mone-byte-bool
8023 @opindex -mone-byte-bool
8024 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
8025 By default @samp{sizeof(bool)} is @samp{4} when compiling for
8026 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
8027 option has no effect on x86.
8029 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
8030 to generate code that is not binary compatible with code generated
8031 without that switch. Using this switch may require recompiling all
8032 other modules in a program, including system libraries. Use this
8033 switch to conform to a non-default data model.
8035 @item -mfix-and-continue
8036 @itemx -ffix-and-continue
8037 @itemx -findirect-data
8038 @opindex mfix-and-continue
8039 @opindex ffix-and-continue
8040 @opindex findirect-data
8041 Generate code suitable for fast turn around development. Needed to
8042 enable gdb to dynamically load @code{.o} files into already running
8043 programs. @option{-findirect-data} and @option{-ffix-and-continue}
8044 are provided for backwards compatibility.
8048 Loads all members of static archive libraries.
8049 See man ld(1) for more information.
8051 @item -arch_errors_fatal
8052 @opindex arch_errors_fatal
8053 Cause the errors having to do with files that have the wrong architecture
8057 @opindex bind_at_load
8058 Causes the output file to be marked such that the dynamic linker will
8059 bind all undefined references when the file is loaded or launched.
8063 Produce a Mach-o bundle format file.
8064 See man ld(1) for more information.
8066 @item -bundle_loader @var{executable}
8067 @opindex bundle_loader
8068 This option specifies the @var{executable} that will be loading the build
8069 output file being linked. See man ld(1) for more information.
8072 @opindex -dynamiclib
8073 When passed this option, GCC will produce a dynamic library instead of
8074 an executable when linking, using the Darwin @file{libtool} command.
8076 @item -force_cpusubtype_ALL
8077 @opindex -force_cpusubtype_ALL
8078 This causes GCC's output file to have the @var{ALL} subtype, instead of
8079 one controlled by the @option{-mcpu} or @option{-march} option.
8081 @item -allowable_client @var{client_name}
8083 @itemx -compatibility_version
8084 @itemx -current_version
8086 @itemx -dependency-file
8088 @itemx -dylinker_install_name
8090 @itemx -exported_symbols_list
8092 @itemx -flat_namespace
8093 @itemx -force_flat_namespace
8094 @itemx -headerpad_max_install_names
8097 @itemx -install_name
8098 @itemx -keep_private_externs
8099 @itemx -multi_module
8100 @itemx -multiply_defined
8101 @itemx -multiply_defined_unused
8103 @itemx -no_dead_strip_inits_and_terms
8104 @itemx -nofixprebinding
8107 @itemx -noseglinkedit
8108 @itemx -pagezero_size
8110 @itemx -prebind_all_twolevel_modules
8111 @itemx -private_bundle
8112 @itemx -read_only_relocs
8114 @itemx -sectobjectsymbols
8118 @itemx -sectobjectsymbols
8121 @itemx -segs_read_only_addr
8122 @itemx -segs_read_write_addr
8123 @itemx -seg_addr_table
8124 @itemx -seg_addr_table_filename
8127 @itemx -segs_read_only_addr
8128 @itemx -segs_read_write_addr
8129 @itemx -single_module
8132 @itemx -sub_umbrella
8133 @itemx -twolevel_namespace
8136 @itemx -unexported_symbols_list
8137 @itemx -weak_reference_mismatches
8140 @opindex allowable_client
8141 @opindex client_name
8142 @opindex compatibility_version
8143 @opindex current_version
8145 @opindex dependency-file
8147 @opindex dylinker_install_name
8149 @opindex exported_symbols_list
8151 @opindex flat_namespace
8152 @opindex force_flat_namespace
8153 @opindex headerpad_max_install_names
8156 @opindex install_name
8157 @opindex keep_private_externs
8158 @opindex multi_module
8159 @opindex multiply_defined
8160 @opindex multiply_defined_unused
8162 @opindex no_dead_strip_inits_and_terms
8163 @opindex nofixprebinding
8164 @opindex nomultidefs
8166 @opindex noseglinkedit
8167 @opindex pagezero_size
8169 @opindex prebind_all_twolevel_modules
8170 @opindex private_bundle
8171 @opindex read_only_relocs
8173 @opindex sectobjectsymbols
8177 @opindex sectobjectsymbols
8180 @opindex segs_read_only_addr
8181 @opindex segs_read_write_addr
8182 @opindex seg_addr_table
8183 @opindex seg_addr_table_filename
8184 @opindex seglinkedit
8186 @opindex segs_read_only_addr
8187 @opindex segs_read_write_addr
8188 @opindex single_module
8190 @opindex sub_library
8191 @opindex sub_umbrella
8192 @opindex twolevel_namespace
8195 @opindex unexported_symbols_list
8196 @opindex weak_reference_mismatches
8197 @opindex whatsloaded
8199 These options are passed to the Darwin linker. The Darwin linker man page
8200 describes them in detail.
8203 @node DEC Alpha Options
8204 @subsection DEC Alpha Options
8206 These @samp{-m} options are defined for the DEC Alpha implementations:
8209 @item -mno-soft-float
8211 @opindex mno-soft-float
8212 @opindex msoft-float
8213 Use (do not use) the hardware floating-point instructions for
8214 floating-point operations. When @option{-msoft-float} is specified,
8215 functions in @file{libgcc.a} will be used to perform floating-point
8216 operations. Unless they are replaced by routines that emulate the
8217 floating-point operations, or compiled in such a way as to call such
8218 emulations routines, these routines will issue floating-point
8219 operations. If you are compiling for an Alpha without floating-point
8220 operations, you must ensure that the library is built so as not to call
8223 Note that Alpha implementations without floating-point operations are
8224 required to have floating-point registers.
8229 @opindex mno-fp-regs
8230 Generate code that uses (does not use) the floating-point register set.
8231 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8232 register set is not used, floating point operands are passed in integer
8233 registers as if they were integers and floating-point results are passed
8234 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8235 so any function with a floating-point argument or return value called by code
8236 compiled with @option{-mno-fp-regs} must also be compiled with that
8239 A typical use of this option is building a kernel that does not use,
8240 and hence need not save and restore, any floating-point registers.
8244 The Alpha architecture implements floating-point hardware optimized for
8245 maximum performance. It is mostly compliant with the IEEE floating
8246 point standard. However, for full compliance, software assistance is
8247 required. This option generates code fully IEEE compliant code
8248 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8249 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8250 defined during compilation. The resulting code is less efficient but is
8251 able to correctly support denormalized numbers and exceptional IEEE
8252 values such as not-a-number and plus/minus infinity. Other Alpha
8253 compilers call this option @option{-ieee_with_no_inexact}.
8255 @item -mieee-with-inexact
8256 @opindex mieee-with-inexact
8257 This is like @option{-mieee} except the generated code also maintains
8258 the IEEE @var{inexact-flag}. Turning on this option causes the
8259 generated code to implement fully-compliant IEEE math. In addition to
8260 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8261 macro. On some Alpha implementations the resulting code may execute
8262 significantly slower than the code generated by default. Since there is
8263 very little code that depends on the @var{inexact-flag}, you should
8264 normally not specify this option. Other Alpha compilers call this
8265 option @option{-ieee_with_inexact}.
8267 @item -mfp-trap-mode=@var{trap-mode}
8268 @opindex mfp-trap-mode
8269 This option controls what floating-point related traps are enabled.
8270 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8271 The trap mode can be set to one of four values:
8275 This is the default (normal) setting. The only traps that are enabled
8276 are the ones that cannot be disabled in software (e.g., division by zero
8280 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8284 Like @samp{su}, but the instructions are marked to be safe for software
8285 completion (see Alpha architecture manual for details).
8288 Like @samp{su}, but inexact traps are enabled as well.
8291 @item -mfp-rounding-mode=@var{rounding-mode}
8292 @opindex mfp-rounding-mode
8293 Selects the IEEE rounding mode. Other Alpha compilers call this option
8294 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8299 Normal IEEE rounding mode. Floating point numbers are rounded towards
8300 the nearest machine number or towards the even machine number in case
8304 Round towards minus infinity.
8307 Chopped rounding mode. Floating point numbers are rounded towards zero.
8310 Dynamic rounding mode. A field in the floating point control register
8311 (@var{fpcr}, see Alpha architecture reference manual) controls the
8312 rounding mode in effect. The C library initializes this register for
8313 rounding towards plus infinity. Thus, unless your program modifies the
8314 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8317 @item -mtrap-precision=@var{trap-precision}
8318 @opindex mtrap-precision
8319 In the Alpha architecture, floating point traps are imprecise. This
8320 means without software assistance it is impossible to recover from a
8321 floating trap and program execution normally needs to be terminated.
8322 GCC can generate code that can assist operating system trap handlers
8323 in determining the exact location that caused a floating point trap.
8324 Depending on the requirements of an application, different levels of
8325 precisions can be selected:
8329 Program precision. This option is the default and means a trap handler
8330 can only identify which program caused a floating point exception.
8333 Function precision. The trap handler can determine the function that
8334 caused a floating point exception.
8337 Instruction precision. The trap handler can determine the exact
8338 instruction that caused a floating point exception.
8341 Other Alpha compilers provide the equivalent options called
8342 @option{-scope_safe} and @option{-resumption_safe}.
8344 @item -mieee-conformant
8345 @opindex mieee-conformant
8346 This option marks the generated code as IEEE conformant. You must not
8347 use this option unless you also specify @option{-mtrap-precision=i} and either
8348 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8349 is to emit the line @samp{.eflag 48} in the function prologue of the
8350 generated assembly file. Under DEC Unix, this has the effect that
8351 IEEE-conformant math library routines will be linked in.
8353 @item -mbuild-constants
8354 @opindex mbuild-constants
8355 Normally GCC examines a 32- or 64-bit integer constant to
8356 see if it can construct it from smaller constants in two or three
8357 instructions. If it cannot, it will output the constant as a literal and
8358 generate code to load it from the data segment at runtime.
8360 Use this option to require GCC to construct @emph{all} integer constants
8361 using code, even if it takes more instructions (the maximum is six).
8363 You would typically use this option to build a shared library dynamic
8364 loader. Itself a shared library, it must relocate itself in memory
8365 before it can find the variables and constants in its own data segment.
8371 Select whether to generate code to be assembled by the vendor-supplied
8372 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8390 Indicate whether GCC should generate code to use the optional BWX,
8391 CIX, FIX and MAX instruction sets. The default is to use the instruction
8392 sets supported by the CPU type specified via @option{-mcpu=} option or that
8393 of the CPU on which GCC was built if none was specified.
8398 @opindex mfloat-ieee
8399 Generate code that uses (does not use) VAX F and G floating point
8400 arithmetic instead of IEEE single and double precision.
8402 @item -mexplicit-relocs
8403 @itemx -mno-explicit-relocs
8404 @opindex mexplicit-relocs
8405 @opindex mno-explicit-relocs
8406 Older Alpha assemblers provided no way to generate symbol relocations
8407 except via assembler macros. Use of these macros does not allow
8408 optimal instruction scheduling. GNU binutils as of version 2.12
8409 supports a new syntax that allows the compiler to explicitly mark
8410 which relocations should apply to which instructions. This option
8411 is mostly useful for debugging, as GCC detects the capabilities of
8412 the assembler when it is built and sets the default accordingly.
8416 @opindex msmall-data
8417 @opindex mlarge-data
8418 When @option{-mexplicit-relocs} is in effect, static data is
8419 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8420 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8421 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8422 16-bit relocations off of the @code{$gp} register. This limits the
8423 size of the small data area to 64KB, but allows the variables to be
8424 directly accessed via a single instruction.
8426 The default is @option{-mlarge-data}. With this option the data area
8427 is limited to just below 2GB@. Programs that require more than 2GB of
8428 data must use @code{malloc} or @code{mmap} to allocate the data in the
8429 heap instead of in the program's data segment.
8431 When generating code for shared libraries, @option{-fpic} implies
8432 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8436 @opindex msmall-text
8437 @opindex mlarge-text
8438 When @option{-msmall-text} is used, the compiler assumes that the
8439 code of the entire program (or shared library) fits in 4MB, and is
8440 thus reachable with a branch instruction. When @option{-msmall-data}
8441 is used, the compiler can assume that all local symbols share the
8442 same @code{$gp} value, and thus reduce the number of instructions
8443 required for a function call from 4 to 1.
8445 The default is @option{-mlarge-text}.
8447 @item -mcpu=@var{cpu_type}
8449 Set the instruction set and instruction scheduling parameters for
8450 machine type @var{cpu_type}. You can specify either the @samp{EV}
8451 style name or the corresponding chip number. GCC supports scheduling
8452 parameters for the EV4, EV5 and EV6 family of processors and will
8453 choose the default values for the instruction set from the processor
8454 you specify. If you do not specify a processor type, GCC will default
8455 to the processor on which the compiler was built.
8457 Supported values for @var{cpu_type} are
8463 Schedules as an EV4 and has no instruction set extensions.
8467 Schedules as an EV5 and has no instruction set extensions.
8471 Schedules as an EV5 and supports the BWX extension.
8476 Schedules as an EV5 and supports the BWX and MAX extensions.
8480 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8484 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8487 @item -mtune=@var{cpu_type}
8489 Set only the instruction scheduling parameters for machine type
8490 @var{cpu_type}. The instruction set is not changed.
8492 @item -mmemory-latency=@var{time}
8493 @opindex mmemory-latency
8494 Sets the latency the scheduler should assume for typical memory
8495 references as seen by the application. This number is highly
8496 dependent on the memory access patterns used by the application
8497 and the size of the external cache on the machine.
8499 Valid options for @var{time} are
8503 A decimal number representing clock cycles.
8509 The compiler contains estimates of the number of clock cycles for
8510 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8511 (also called Dcache, Scache, and Bcache), as well as to main memory.
8512 Note that L3 is only valid for EV5.
8517 @node DEC Alpha/VMS Options
8518 @subsection DEC Alpha/VMS Options
8520 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8523 @item -mvms-return-codes
8524 @opindex mvms-return-codes
8525 Return VMS condition codes from main. The default is to return POSIX
8526 style condition (e.g.@ error) codes.
8530 @subsection FRV Options
8537 Only use the first 32 general purpose registers.
8542 Use all 64 general purpose registers.
8547 Use only the first 32 floating point registers.
8552 Use all 64 floating point registers
8555 @opindex mhard-float
8557 Use hardware instructions for floating point operations.
8560 @opindex msoft-float
8562 Use library routines for floating point operations.
8567 Dynamically allocate condition code registers.
8572 Do not try to dynamically allocate condition code registers, only
8573 use @code{icc0} and @code{fcc0}.
8578 Change ABI to use double word insns.
8583 Do not use double word instructions.
8588 Use floating point double instructions.
8593 Do not use floating point double instructions.
8598 Use media instructions.
8603 Do not use media instructions.
8608 Use multiply and add/subtract instructions.
8613 Do not use multiply and add/subtract instructions.
8618 Select the FDPIC ABI, that uses function descriptors to represent
8619 pointers to functions. Without any PIC/PIE-related options, it
8620 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
8621 assumes GOT entries and small data are within a 12-bit range from the
8622 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
8623 are computed with 32 bits.
8626 @opindex minline-plt
8628 Enable inlining of PLT entries in function calls to functions that are
8629 not known to bind locally. It has no effect without @option{-mfdpic}.
8630 It's enabled by default if optimizing for speed and compiling for
8631 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
8632 optimization option such as @option{-O3} or above is present in the
8638 Assume a large TLS segment when generating thread-local code.
8643 Do not assume a large TLS segment when generating thread-local code.
8648 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
8649 that is known to be in read-only sections. It's enabled by default,
8650 except for @option{-fpic} or @option{-fpie}: even though it may help
8651 make the global offset table smaller, it trades 1 instruction for 4.
8652 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
8653 one of which may be shared by multiple symbols, and it avoids the need
8654 for a GOT entry for the referenced symbol, so it's more likely to be a
8655 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
8657 @item -multilib-library-pic
8658 @opindex multilib-library-pic
8660 Link with the (library, not FD) pic libraries. It's implied by
8661 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
8662 @option{-fpic} without @option{-mfdpic}. You should never have to use
8668 Follow the EABI requirement of always creating a frame pointer whenever
8669 a stack frame is allocated. This option is enabled by default and can
8670 be disabled with @option{-mno-linked-fp}.
8673 @opindex mlong-calls
8675 Use indirect addressing to call functions outside the current
8676 compilation unit. This allows the functions to be placed anywhere
8677 within the 32-bit address space.
8679 @item -malign-labels
8680 @opindex malign-labels
8682 Try to align labels to an 8-byte boundary by inserting nops into the
8683 previous packet. This option only has an effect when VLIW packing
8684 is enabled. It doesn't create new packets; it merely adds nops to
8688 @opindex mlibrary-pic
8690 Generate position-independent EABI code.
8695 Use only the first four media accumulator registers.
8700 Use all eight media accumulator registers.
8705 Pack VLIW instructions.
8710 Do not pack VLIW instructions.
8715 Do not mark ABI switches in e_flags.
8720 Enable the use of conditional-move instructions (default).
8722 This switch is mainly for debugging the compiler and will likely be removed
8723 in a future version.
8725 @item -mno-cond-move
8726 @opindex mno-cond-move
8728 Disable the use of conditional-move instructions.
8730 This switch is mainly for debugging the compiler and will likely be removed
8731 in a future version.
8736 Enable the use of conditional set instructions (default).
8738 This switch is mainly for debugging the compiler and will likely be removed
8739 in a future version.
8744 Disable the use of conditional set instructions.
8746 This switch is mainly for debugging the compiler and will likely be removed
8747 in a future version.
8752 Enable the use of conditional execution (default).
8754 This switch is mainly for debugging the compiler and will likely be removed
8755 in a future version.
8757 @item -mno-cond-exec
8758 @opindex mno-cond-exec
8760 Disable the use of conditional execution.
8762 This switch is mainly for debugging the compiler and will likely be removed
8763 in a future version.
8766 @opindex mvliw-branch
8768 Run a pass to pack branches into VLIW instructions (default).
8770 This switch is mainly for debugging the compiler and will likely be removed
8771 in a future version.
8773 @item -mno-vliw-branch
8774 @opindex mno-vliw-branch
8776 Do not run a pass to pack branches into VLIW instructions.
8778 This switch is mainly for debugging the compiler and will likely be removed
8779 in a future version.
8781 @item -mmulti-cond-exec
8782 @opindex mmulti-cond-exec
8784 Enable optimization of @code{&&} and @code{||} in conditional execution
8787 This switch is mainly for debugging the compiler and will likely be removed
8788 in a future version.
8790 @item -mno-multi-cond-exec
8791 @opindex mno-multi-cond-exec
8793 Disable optimization of @code{&&} and @code{||} in conditional execution.
8795 This switch is mainly for debugging the compiler and will likely be removed
8796 in a future version.
8798 @item -mnested-cond-exec
8799 @opindex mnested-cond-exec
8801 Enable nested conditional execution optimizations (default).
8803 This switch is mainly for debugging the compiler and will likely be removed
8804 in a future version.
8806 @item -mno-nested-cond-exec
8807 @opindex mno-nested-cond-exec
8809 Disable nested conditional execution optimizations.
8811 This switch is mainly for debugging the compiler and will likely be removed
8812 in a future version.
8814 @item -moptimize-membar
8815 @opindex moptimize-membar
8817 This switch removes redundant @code{membar} instructions from the
8818 compiler generated code. It is enabled by default.
8820 @item -mno-optimize-membar
8821 @opindex mno-optimize-membar
8823 This switch disables the automatic removal of redundant @code{membar}
8824 instructions from the generated code.
8826 @item -mtomcat-stats
8827 @opindex mtomcat-stats
8829 Cause gas to print out tomcat statistics.
8831 @item -mcpu=@var{cpu}
8834 Select the processor type for which to generate code. Possible values are
8835 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
8836 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
8840 @node GNU/Linux Options
8841 @subsection GNU/Linux Options
8843 These @samp{-m} options are defined for GNU/Linux targets:
8848 Use the GNU C library instead of uClibc. This is the default except
8849 on @samp{*-*-linux-*uclibc*} targets.
8853 Use uClibc instead of the GNU C library. This is the default on
8854 @samp{*-*-linux-*uclibc*} targets.
8857 @node H8/300 Options
8858 @subsection H8/300 Options
8860 These @samp{-m} options are defined for the H8/300 implementations:
8865 Shorten some address references at link time, when possible; uses the
8866 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8867 ld, Using ld}, for a fuller description.
8871 Generate code for the H8/300H@.
8875 Generate code for the H8S@.
8879 Generate code for the H8S and H8/300H in the normal mode. This switch
8880 must be used either with @option{-mh} or @option{-ms}.
8884 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
8888 Make @code{int} data 32 bits by default.
8892 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
8893 The default for the H8/300H and H8S is to align longs and floats on 4
8895 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8896 This option has no effect on the H8/300.
8900 @subsection HPPA Options
8901 @cindex HPPA Options
8903 These @samp{-m} options are defined for the HPPA family of computers:
8906 @item -march=@var{architecture-type}
8908 Generate code for the specified architecture. The choices for
8909 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8910 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8911 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8912 architecture option for your machine. Code compiled for lower numbered
8913 architectures will run on higher numbered architectures, but not the
8917 @itemx -mpa-risc-1-1
8918 @itemx -mpa-risc-2-0
8919 @opindex mpa-risc-1-0
8920 @opindex mpa-risc-1-1
8921 @opindex mpa-risc-2-0
8922 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8925 @opindex mbig-switch
8926 Generate code suitable for big switch tables. Use this option only if
8927 the assembler/linker complain about out of range branches within a switch
8930 @item -mjump-in-delay
8931 @opindex mjump-in-delay
8932 Fill delay slots of function calls with unconditional jump instructions
8933 by modifying the return pointer for the function call to be the target
8934 of the conditional jump.
8936 @item -mdisable-fpregs
8937 @opindex mdisable-fpregs
8938 Prevent floating point registers from being used in any manner. This is
8939 necessary for compiling kernels which perform lazy context switching of
8940 floating point registers. If you use this option and attempt to perform
8941 floating point operations, the compiler will abort.
8943 @item -mdisable-indexing
8944 @opindex mdisable-indexing
8945 Prevent the compiler from using indexing address modes. This avoids some
8946 rather obscure problems when compiling MIG generated code under MACH@.
8948 @item -mno-space-regs
8949 @opindex mno-space-regs
8950 Generate code that assumes the target has no space registers. This allows
8951 GCC to generate faster indirect calls and use unscaled index address modes.
8953 Such code is suitable for level 0 PA systems and kernels.
8955 @item -mfast-indirect-calls
8956 @opindex mfast-indirect-calls
8957 Generate code that assumes calls never cross space boundaries. This
8958 allows GCC to emit code which performs faster indirect calls.
8960 This option will not work in the presence of shared libraries or nested
8963 @item -mfixed-range=@var{register-range}
8964 @opindex mfixed-range
8965 Generate code treating the given register range as fixed registers.
8966 A fixed register is one that the register allocator can not use. This is
8967 useful when compiling kernel code. A register range is specified as
8968 two registers separated by a dash. Multiple register ranges can be
8969 specified separated by a comma.
8971 @item -mlong-load-store
8972 @opindex mlong-load-store
8973 Generate 3-instruction load and store sequences as sometimes required by
8974 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8977 @item -mportable-runtime
8978 @opindex mportable-runtime
8979 Use the portable calling conventions proposed by HP for ELF systems.
8983 Enable the use of assembler directives only GAS understands.
8985 @item -mschedule=@var{cpu-type}
8987 Schedule code according to the constraints for the machine type
8988 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8989 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8990 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8991 proper scheduling option for your machine. The default scheduling is
8995 @opindex mlinker-opt
8996 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8997 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8998 linkers in which they give bogus error messages when linking some programs.
9001 @opindex msoft-float
9002 Generate output containing library calls for floating point.
9003 @strong{Warning:} the requisite libraries are not available for all HPPA
9004 targets. Normally the facilities of the machine's usual C compiler are
9005 used, but this cannot be done directly in cross-compilation. You must make
9006 your own arrangements to provide suitable library functions for
9007 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
9008 does provide software floating point support.
9010 @option{-msoft-float} changes the calling convention in the output file;
9011 therefore, it is only useful if you compile @emph{all} of a program with
9012 this option. In particular, you need to compile @file{libgcc.a}, the
9013 library that comes with GCC, with @option{-msoft-float} in order for
9018 Generate the predefine, @code{_SIO}, for server IO@. The default is
9019 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
9020 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
9021 options are available under HP-UX and HI-UX@.
9025 Use GNU ld specific options. This passes @option{-shared} to ld when
9026 building a shared library. It is the default when GCC is configured,
9027 explicitly or implicitly, with the GNU linker. This option does not
9028 have any affect on which ld is called, it only changes what parameters
9029 are passed to that ld. The ld that is called is determined by the
9030 @option{--with-ld} configure option, GCC's program search path, and
9031 finally by the user's @env{PATH}. The linker used by GCC can be printed
9032 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
9033 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9037 Use HP ld specific options. This passes @option{-b} to ld when building
9038 a shared library and passes @option{+Accept TypeMismatch} to ld on all
9039 links. It is the default when GCC is configured, explicitly or
9040 implicitly, with the HP linker. This option does not have any affect on
9041 which ld is called, it only changes what parameters are passed to that
9042 ld. The ld that is called is determined by the @option{--with-ld}
9043 configure option, GCC's program search path, and finally by the user's
9044 @env{PATH}. The linker used by GCC can be printed using @samp{which
9045 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
9046 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9049 @opindex mno-long-calls
9050 Generate code that uses long call sequences. This ensures that a call
9051 is always able to reach linker generated stubs. The default is to generate
9052 long calls only when the distance from the call site to the beginning
9053 of the function or translation unit, as the case may be, exceeds a
9054 predefined limit set by the branch type being used. The limits for
9055 normal calls are 7,600,000 and 240,000 bytes, respectively for the
9056 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
9059 Distances are measured from the beginning of functions when using the
9060 @option{-ffunction-sections} option, or when using the @option{-mgas}
9061 and @option{-mno-portable-runtime} options together under HP-UX with
9064 It is normally not desirable to use this option as it will degrade
9065 performance. However, it may be useful in large applications,
9066 particularly when partial linking is used to build the application.
9068 The types of long calls used depends on the capabilities of the
9069 assembler and linker, and the type of code being generated. The
9070 impact on systems that support long absolute calls, and long pic
9071 symbol-difference or pc-relative calls should be relatively small.
9072 However, an indirect call is used on 32-bit ELF systems in pic code
9073 and it is quite long.
9075 @item -munix=@var{unix-std}
9077 Generate compiler predefines and select a startfile for the specified
9078 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
9079 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
9080 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
9081 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
9082 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
9085 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
9086 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
9087 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
9088 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
9089 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
9090 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
9092 It is @emph{important} to note that this option changes the interfaces
9093 for various library routines. It also affects the operational behavior
9094 of the C library. Thus, @emph{extreme} care is needed in using this
9097 Library code that is intended to operate with more than one UNIX
9098 standard must test, set and restore the variable @var{__xpg4_extended_mask}
9099 as appropriate. Most GNU software doesn't provide this capability.
9103 Suppress the generation of link options to search libdld.sl when the
9104 @option{-static} option is specified on HP-UX 10 and later.
9108 The HP-UX implementation of setlocale in libc has a dependency on
9109 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9110 when the @option{-static} option is specified, special link options
9111 are needed to resolve this dependency.
9113 On HP-UX 10 and later, the GCC driver adds the necessary options to
9114 link with libdld.sl when the @option{-static} option is specified.
9115 This causes the resulting binary to be dynamic. On the 64-bit port,
9116 the linkers generate dynamic binaries by default in any case. The
9117 @option{-nolibdld} option can be used to prevent the GCC driver from
9118 adding these link options.
9122 Add support for multithreading with the @dfn{dce thread} library
9123 under HP-UX@. This option sets flags for both the preprocessor and
9127 @node i386 and x86-64 Options
9128 @subsection Intel 386 and AMD x86-64 Options
9129 @cindex i386 Options
9130 @cindex x86-64 Options
9131 @cindex Intel 386 Options
9132 @cindex AMD x86-64 Options
9134 These @samp{-m} options are defined for the i386 and x86-64 family of
9138 @item -mtune=@var{cpu-type}
9140 Tune to @var{cpu-type} everything applicable about the generated code, except
9141 for the ABI and the set of available instructions. The choices for
9145 Produce code optimized for the most common IA32/AMD64/EM64T processors.
9146 If you know the CPU on which your code will run, then you should use
9147 the corresponding @option{-mtune} option instead of
9148 @option{-mtune=generic}. But, if you do not know exactly what CPU users
9149 of your application will have, then you should use this option.
9151 As new processors are deployed in the marketplace, the behavior of this
9152 option will change. Therefore, if you upgrade to a newer version of
9153 GCC, the code generated option will change to reflect the processors
9154 that were most common when that version of GCC was released.
9156 There is no @option{-march=generic} option because @option{-march}
9157 indicates the instruction set the compiler can use, and there is no
9158 generic instruction set applicable to all processors. In contrast,
9159 @option{-mtune} indicates the processor (or, in this case, collection of
9160 processors) for which the code is optimized.
9162 Original Intel's i386 CPU@.
9164 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
9166 Intel Pentium CPU with no MMX support.
9168 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
9170 Intel PentiumPro CPU@.
9172 Same as @code{generic}, but when used as @code{march} option, PentiumPro
9173 instruction set will be used, so the code will run on all i686 family chips.
9175 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
9176 @item pentium3, pentium3m
9177 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
9180 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
9181 support. Used by Centrino notebooks.
9182 @item pentium4, pentium4m
9183 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
9185 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
9188 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
9189 SSE2 and SSE3 instruction set support.
9191 AMD K6 CPU with MMX instruction set support.
9193 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
9194 @item athlon, athlon-tbird
9195 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
9197 @item athlon-4, athlon-xp, athlon-mp
9198 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
9199 instruction set support.
9200 @item k8, opteron, athlon64, athlon-fx
9201 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
9202 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
9204 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
9207 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
9208 instruction set support.
9210 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
9211 implemented for this chip.)
9213 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
9214 implemented for this chip.)
9217 While picking a specific @var{cpu-type} will schedule things appropriately
9218 for that particular chip, the compiler will not generate any code that
9219 does not run on the i386 without the @option{-march=@var{cpu-type}} option
9222 @item -march=@var{cpu-type}
9224 Generate instructions for the machine type @var{cpu-type}. The choices
9225 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
9226 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
9228 @item -mcpu=@var{cpu-type}
9230 A deprecated synonym for @option{-mtune}.
9239 @opindex mpentiumpro
9240 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
9241 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
9242 These synonyms are deprecated.
9244 @item -mfpmath=@var{unit}
9246 Generate floating point arithmetics for selected unit @var{unit}. The choices
9251 Use the standard 387 floating point coprocessor present majority of chips and
9252 emulated otherwise. Code compiled with this option will run almost everywhere.
9253 The temporary results are computed in 80bit precision instead of precision
9254 specified by the type resulting in slightly different results compared to most
9255 of other chips. See @option{-ffloat-store} for more detailed description.
9257 This is the default choice for i386 compiler.
9260 Use scalar floating point instructions present in the SSE instruction set.
9261 This instruction set is supported by Pentium3 and newer chips, in the AMD line
9262 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
9263 instruction set supports only single precision arithmetics, thus the double and
9264 extended precision arithmetics is still done using 387. Later version, present
9265 only in Pentium4 and the future AMD x86-64 chips supports double precision
9268 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
9269 or @option{-msse2} switches to enable SSE extensions and make this option
9270 effective. For the x86-64 compiler, these extensions are enabled by default.
9272 The resulting code should be considerably faster in the majority of cases and avoid
9273 the numerical instability problems of 387 code, but may break some existing
9274 code that expects temporaries to be 80bit.
9276 This is the default choice for the x86-64 compiler.
9279 Attempt to utilize both instruction sets at once. This effectively double the
9280 amount of available registers and on chips with separate execution units for
9281 387 and SSE the execution resources too. Use this option with care, as it is
9282 still experimental, because the GCC register allocator does not model separate
9283 functional units well resulting in instable performance.
9286 @item -masm=@var{dialect}
9287 @opindex masm=@var{dialect}
9288 Output asm instructions using selected @var{dialect}. Supported
9289 choices are @samp{intel} or @samp{att} (the default one). Darwin does
9290 not support @samp{intel}.
9295 @opindex mno-ieee-fp
9296 Control whether or not the compiler uses IEEE floating point
9297 comparisons. These handle correctly the case where the result of a
9298 comparison is unordered.
9301 @opindex msoft-float
9302 Generate output containing library calls for floating point.
9303 @strong{Warning:} the requisite libraries are not part of GCC@.
9304 Normally the facilities of the machine's usual C compiler are used, but
9305 this can't be done directly in cross-compilation. You must make your
9306 own arrangements to provide suitable library functions for
9309 On machines where a function returns floating point results in the 80387
9310 register stack, some floating point opcodes may be emitted even if
9311 @option{-msoft-float} is used.
9313 @item -mno-fp-ret-in-387
9314 @opindex mno-fp-ret-in-387
9315 Do not use the FPU registers for return values of functions.
9317 The usual calling convention has functions return values of types
9318 @code{float} and @code{double} in an FPU register, even if there
9319 is no FPU@. The idea is that the operating system should emulate
9322 The option @option{-mno-fp-ret-in-387} causes such values to be returned
9323 in ordinary CPU registers instead.
9325 @item -mno-fancy-math-387
9326 @opindex mno-fancy-math-387
9327 Some 387 emulators do not support the @code{sin}, @code{cos} and
9328 @code{sqrt} instructions for the 387. Specify this option to avoid
9329 generating those instructions. This option is the default on FreeBSD,
9330 OpenBSD and NetBSD@. This option is overridden when @option{-march}
9331 indicates that the target cpu will always have an FPU and so the
9332 instruction will not need emulation. As of revision 2.6.1, these
9333 instructions are not generated unless you also use the
9334 @option{-funsafe-math-optimizations} switch.
9336 @item -malign-double
9337 @itemx -mno-align-double
9338 @opindex malign-double
9339 @opindex mno-align-double
9340 Control whether GCC aligns @code{double}, @code{long double}, and
9341 @code{long long} variables on a two word boundary or a one word
9342 boundary. Aligning @code{double} variables on a two word boundary will
9343 produce code that runs somewhat faster on a @samp{Pentium} at the
9344 expense of more memory.
9346 @strong{Warning:} if you use the @option{-malign-double} switch,
9347 structures containing the above types will be aligned differently than
9348 the published application binary interface specifications for the 386
9349 and will not be binary compatible with structures in code compiled
9350 without that switch.
9352 @item -m96bit-long-double
9353 @itemx -m128bit-long-double
9354 @opindex m96bit-long-double
9355 @opindex m128bit-long-double
9356 These switches control the size of @code{long double} type. The i386
9357 application binary interface specifies the size to be 96 bits,
9358 so @option{-m96bit-long-double} is the default in 32 bit mode.
9360 Modern architectures (Pentium and newer) would prefer @code{long double}
9361 to be aligned to an 8 or 16 byte boundary. In arrays or structures
9362 conforming to the ABI, this would not be possible. So specifying a
9363 @option{-m128bit-long-double} will align @code{long double}
9364 to a 16 byte boundary by padding the @code{long double} with an additional
9367 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
9368 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
9370 Notice that neither of these options enable any extra precision over the x87
9371 standard of 80 bits for a @code{long double}.
9373 @strong{Warning:} if you override the default value for your target ABI, the
9374 structures and arrays containing @code{long double} variables will change
9375 their size as well as function calling convention for function taking
9376 @code{long double} will be modified. Hence they will not be binary
9377 compatible with arrays or structures in code compiled without that switch.
9379 @item -mmlarge-data-threshold=@var{number}
9380 @opindex mlarge-data-threshold=@var{number}
9381 When @option{-mcmodel=medium} is specified, the data greater than
9382 @var{threshold} are placed in large data section. This value must be the
9383 same across all object linked into the binary and defaults to 65535.
9386 @itemx -mno-svr3-shlib
9387 @opindex msvr3-shlib
9388 @opindex mno-svr3-shlib
9389 Control whether GCC places uninitialized local variables into the
9390 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
9391 into @code{bss}. These options are meaningful only on System V Release 3.
9395 Use a different function-calling convention, in which functions that
9396 take a fixed number of arguments return with the @code{ret} @var{num}
9397 instruction, which pops their arguments while returning. This saves one
9398 instruction in the caller since there is no need to pop the arguments
9401 You can specify that an individual function is called with this calling
9402 sequence with the function attribute @samp{stdcall}. You can also
9403 override the @option{-mrtd} option by using the function attribute
9404 @samp{cdecl}. @xref{Function Attributes}.
9406 @strong{Warning:} this calling convention is incompatible with the one
9407 normally used on Unix, so you cannot use it if you need to call
9408 libraries compiled with the Unix compiler.
9410 Also, you must provide function prototypes for all functions that
9411 take variable numbers of arguments (including @code{printf});
9412 otherwise incorrect code will be generated for calls to those
9415 In addition, seriously incorrect code will result if you call a
9416 function with too many arguments. (Normally, extra arguments are
9417 harmlessly ignored.)
9419 @item -mregparm=@var{num}
9421 Control how many registers are used to pass integer arguments. By
9422 default, no registers are used to pass arguments, and at most 3
9423 registers can be used. You can control this behavior for a specific
9424 function by using the function attribute @samp{regparm}.
9425 @xref{Function Attributes}.
9427 @strong{Warning:} if you use this switch, and
9428 @var{num} is nonzero, then you must build all modules with the same
9429 value, including any libraries. This includes the system libraries and
9433 @opindex msseregparm
9434 Use SSE register passing conventions for float and double arguments
9435 and return values. You can control this behavior for a specific
9436 function by using the function attribute @samp{sseregparm}.
9437 @xref{Function Attributes}.
9439 @strong{Warning:} if you use this switch then you must build all
9440 modules with the same value, including any libraries. This includes
9441 the system libraries and startup modules.
9443 @item -mpreferred-stack-boundary=@var{num}
9444 @opindex mpreferred-stack-boundary
9445 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
9446 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
9447 the default is 4 (16 bytes or 128 bits), except when optimizing for code
9448 size (@option{-Os}), in which case the default is the minimum correct
9449 alignment (4 bytes for x86, and 8 bytes for x86-64).
9451 On Pentium and PentiumPro, @code{double} and @code{long double} values
9452 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
9453 suffer significant run time performance penalties. On Pentium III, the
9454 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
9455 penalties if it is not 16 byte aligned.
9457 To ensure proper alignment of this values on the stack, the stack boundary
9458 must be as aligned as that required by any value stored on the stack.
9459 Further, every function must be generated such that it keeps the stack
9460 aligned. Thus calling a function compiled with a higher preferred
9461 stack boundary from a function compiled with a lower preferred stack
9462 boundary will most likely misalign the stack. It is recommended that
9463 libraries that use callbacks always use the default setting.
9465 This extra alignment does consume extra stack space, and generally
9466 increases code size. Code that is sensitive to stack space usage, such
9467 as embedded systems and operating system kernels, may want to reduce the
9468 preferred alignment to @option{-mpreferred-stack-boundary=2}.
9486 These switches enable or disable the use of instructions in the MMX,
9487 SSE, SSE2 or 3DNow! extended instruction sets. These extensions are
9488 also available as built-in functions: see @ref{X86 Built-in Functions},
9489 for details of the functions enabled and disabled by these switches.
9491 To have SSE/SSE2 instructions generated automatically from floating-point
9492 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
9494 These options will enable GCC to use these extended instructions in
9495 generated code, even without @option{-mfpmath=sse}. Applications which
9496 perform runtime CPU detection must compile separate files for each
9497 supported architecture, using the appropriate flags. In particular,
9498 the file containing the CPU detection code should be compiled without
9503 Use special versions of certain libm routines that come with an SSE
9504 ABI and an SSE implementation. Useful together with @option{-mfpmath=sse}
9505 to avoid moving values between SSE registers and the x87 FP stack.
9508 @itemx -mno-push-args
9510 @opindex mno-push-args
9511 Use PUSH operations to store outgoing parameters. This method is shorter
9512 and usually equally fast as method using SUB/MOV operations and is enabled
9513 by default. In some cases disabling it may improve performance because of
9514 improved scheduling and reduced dependencies.
9516 @item -maccumulate-outgoing-args
9517 @opindex maccumulate-outgoing-args
9518 If enabled, the maximum amount of space required for outgoing arguments will be
9519 computed in the function prologue. This is faster on most modern CPUs
9520 because of reduced dependencies, improved scheduling and reduced stack usage
9521 when preferred stack boundary is not equal to 2. The drawback is a notable
9522 increase in code size. This switch implies @option{-mno-push-args}.
9526 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
9527 on thread-safe exception handling must compile and link all code with the
9528 @option{-mthreads} option. When compiling, @option{-mthreads} defines
9529 @option{-D_MT}; when linking, it links in a special thread helper library
9530 @option{-lmingwthrd} which cleans up per thread exception handling data.
9532 @item -mno-align-stringops
9533 @opindex mno-align-stringops
9534 Do not align destination of inlined string operations. This switch reduces
9535 code size and improves performance in case the destination is already aligned,
9536 but GCC doesn't know about it.
9538 @item -minline-all-stringops
9539 @opindex minline-all-stringops
9540 By default GCC inlines string operations only when destination is known to be
9541 aligned at least to 4 byte boundary. This enables more inlining, increase code
9542 size, but may improve performance of code that depends on fast memcpy, strlen
9543 and memset for short lengths.
9545 @item -momit-leaf-frame-pointer
9546 @opindex momit-leaf-frame-pointer
9547 Don't keep the frame pointer in a register for leaf functions. This
9548 avoids the instructions to save, set up and restore frame pointers and
9549 makes an extra register available in leaf functions. The option
9550 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9551 which might make debugging harder.
9553 @item -mtls-direct-seg-refs
9554 @itemx -mno-tls-direct-seg-refs
9555 @opindex mtls-direct-seg-refs
9556 Controls whether TLS variables may be accessed with offsets from the
9557 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
9558 or whether the thread base pointer must be added. Whether or not this
9559 is legal depends on the operating system, and whether it maps the
9560 segment to cover the entire TLS area.
9562 For systems that use GNU libc, the default is on.
9565 These @samp{-m} switches are supported in addition to the above
9566 on AMD x86-64 processors in 64-bit environments.
9573 Generate code for a 32-bit or 64-bit environment.
9574 The 32-bit environment sets int, long and pointer to 32 bits and
9575 generates code that runs on any i386 system.
9576 The 64-bit environment sets int to 32 bits and long and pointer
9577 to 64 bits and generates code for AMD's x86-64 architecture.
9580 @opindex no-red-zone
9581 Do not use a so called red zone for x86-64 code. The red zone is mandated
9582 by the x86-64 ABI, it is a 128-byte area beyond the location of the
9583 stack pointer that will not be modified by signal or interrupt handlers
9584 and therefore can be used for temporary data without adjusting the stack
9585 pointer. The flag @option{-mno-red-zone} disables this red zone.
9587 @item -mcmodel=small
9588 @opindex mcmodel=small
9589 Generate code for the small code model: the program and its symbols must
9590 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
9591 Programs can be statically or dynamically linked. This is the default
9594 @item -mcmodel=kernel
9595 @opindex mcmodel=kernel
9596 Generate code for the kernel code model. The kernel runs in the
9597 negative 2 GB of the address space.
9598 This model has to be used for Linux kernel code.
9600 @item -mcmodel=medium
9601 @opindex mcmodel=medium
9602 Generate code for the medium model: The program is linked in the lower 2
9603 GB of the address space but symbols can be located anywhere in the
9604 address space. Programs can be statically or dynamically linked, but
9605 building of shared libraries are not supported with the medium model.
9607 @item -mcmodel=large
9608 @opindex mcmodel=large
9609 Generate code for the large model: This model makes no assumptions
9610 about addresses and sizes of sections. Currently GCC does not implement
9615 @subsection IA-64 Options
9616 @cindex IA-64 Options
9618 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9622 @opindex mbig-endian
9623 Generate code for a big endian target. This is the default for HP-UX@.
9625 @item -mlittle-endian
9626 @opindex mlittle-endian
9627 Generate code for a little endian target. This is the default for AIX5
9634 Generate (or don't) code for the GNU assembler. This is the default.
9635 @c Also, this is the default if the configure option @option{--with-gnu-as}
9642 Generate (or don't) code for the GNU linker. This is the default.
9643 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9648 Generate code that does not use a global pointer register. The result
9649 is not position independent code, and violates the IA-64 ABI@.
9651 @item -mvolatile-asm-stop
9652 @itemx -mno-volatile-asm-stop
9653 @opindex mvolatile-asm-stop
9654 @opindex mno-volatile-asm-stop
9655 Generate (or don't) a stop bit immediately before and after volatile asm
9658 @item -mregister-names
9659 @itemx -mno-register-names
9660 @opindex mregister-names
9661 @opindex mno-register-names
9662 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9663 the stacked registers. This may make assembler output more readable.
9669 Disable (or enable) optimizations that use the small data section. This may
9670 be useful for working around optimizer bugs.
9673 @opindex mconstant-gp
9674 Generate code that uses a single constant global pointer value. This is
9675 useful when compiling kernel code.
9679 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9680 This is useful when compiling firmware code.
9682 @item -minline-float-divide-min-latency
9683 @opindex minline-float-divide-min-latency
9684 Generate code for inline divides of floating point values
9685 using the minimum latency algorithm.
9687 @item -minline-float-divide-max-throughput
9688 @opindex minline-float-divide-max-throughput
9689 Generate code for inline divides of floating point values
9690 using the maximum throughput algorithm.
9692 @item -minline-int-divide-min-latency
9693 @opindex minline-int-divide-min-latency
9694 Generate code for inline divides of integer values
9695 using the minimum latency algorithm.
9697 @item -minline-int-divide-max-throughput
9698 @opindex minline-int-divide-max-throughput
9699 Generate code for inline divides of integer values
9700 using the maximum throughput algorithm.
9702 @item -minline-sqrt-min-latency
9703 @opindex minline-sqrt-min-latency
9704 Generate code for inline square roots
9705 using the minimum latency algorithm.
9707 @item -minline-sqrt-max-throughput
9708 @opindex minline-sqrt-max-throughput
9709 Generate code for inline square roots
9710 using the maximum throughput algorithm.
9712 @item -mno-dwarf2-asm
9714 @opindex mno-dwarf2-asm
9715 @opindex mdwarf2-asm
9716 Don't (or do) generate assembler code for the DWARF2 line number debugging
9717 info. This may be useful when not using the GNU assembler.
9719 @item -mearly-stop-bits
9720 @itemx -mno-early-stop-bits
9721 @opindex mearly-stop-bits
9722 @opindex mno-early-stop-bits
9723 Allow stop bits to be placed earlier than immediately preceding the
9724 instruction that triggered the stop bit. This can improve instruction
9725 scheduling, but does not always do so.
9727 @item -mfixed-range=@var{register-range}
9728 @opindex mfixed-range
9729 Generate code treating the given register range as fixed registers.
9730 A fixed register is one that the register allocator can not use. This is
9731 useful when compiling kernel code. A register range is specified as
9732 two registers separated by a dash. Multiple register ranges can be
9733 specified separated by a comma.
9735 @item -mtls-size=@var{tls-size}
9737 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
9740 @item -mtune=@var{cpu-type}
9742 Tune the instruction scheduling for a particular CPU, Valid values are
9743 itanium, itanium1, merced, itanium2, and mckinley.
9749 Add support for multithreading using the POSIX threads library. This
9750 option sets flags for both the preprocessor and linker. It does
9751 not affect the thread safety of object code produced by the compiler or
9752 that of libraries supplied with it. These are HP-UX specific flags.
9758 Generate code for a 32-bit or 64-bit environment.
9759 The 32-bit environment sets int, long and pointer to 32 bits.
9760 The 64-bit environment sets int to 32 bits and long and pointer
9761 to 64 bits. These are HP-UX specific flags.
9763 @item -mno-sched-br-data-spec
9764 @itemx -msched-br-data-spec
9765 @opindex -mno-sched-br-data-spec
9766 @opindex -msched-br-data-spec
9767 (Dis/En)able data speculative scheduling before reload.
9768 This will result in generation of the ld.a instructions and
9769 the corresponding check instructions (ld.c / chk.a).
9770 The default is 'disable'.
9772 @item -msched-ar-data-spec
9773 @itemx -mno-sched-ar-data-spec
9774 @opindex -msched-ar-data-spec
9775 @opindex -mno-sched-ar-data-spec
9776 (En/Dis)able data speculative scheduling after reload.
9777 This will result in generation of the ld.a instructions and
9778 the corresponding check instructions (ld.c / chk.a).
9779 The default is 'enable'.
9781 @item -mno-sched-control-spec
9782 @itemx -msched-control-spec
9783 @opindex -mno-sched-control-spec
9784 @opindex -msched-control-spec
9785 (Dis/En)able control speculative scheduling. This feature is
9786 available only during region scheduling (i.e. before reload).
9787 This will result in generation of the ld.s instructions and
9788 the corresponding check instructions chk.s .
9789 The default is 'disable'.
9791 @item -msched-br-in-data-spec
9792 @itemx -mno-sched-br-in-data-spec
9793 @opindex -msched-br-in-data-spec
9794 @opindex -mno-sched-br-in-data-spec
9795 (En/Dis)able speculative scheduling of the instructions that
9796 are dependent on the data speculative loads before reload.
9797 This is effective only with @option{-msched-br-data-spec} enabled.
9798 The default is 'enable'.
9800 @item -msched-ar-in-data-spec
9801 @itemx -mno-sched-ar-in-data-spec
9802 @opindex -msched-ar-in-data-spec
9803 @opindex -mno-sched-ar-in-data-spec
9804 (En/Dis)able speculative scheduling of the instructions that
9805 are dependent on the data speculative loads after reload.
9806 This is effective only with @option{-msched-ar-data-spec} enabled.
9807 The default is 'enable'.
9809 @item -msched-in-control-spec
9810 @itemx -mno-sched-in-control-spec
9811 @opindex -msched-in-control-spec
9812 @opindex -mno-sched-in-control-spec
9813 (En/Dis)able speculative scheduling of the instructions that
9814 are dependent on the control speculative loads.
9815 This is effective only with @option{-msched-control-spec} enabled.
9816 The default is 'enable'.
9819 @itemx -mno-sched-ldc
9820 @opindex -msched-ldc
9821 @opindex -mno-sched-ldc
9822 (En/Dis)able use of simple data speculation checks ld.c .
9823 If disabled, only chk.a instructions will be emitted to check
9824 data speculative loads.
9825 The default is 'enable'.
9827 @item -mno-sched-control-ldc
9828 @itemx -msched-control-ldc
9829 @opindex -mno-sched-control-ldc
9830 @opindex -msched-control-ldc
9831 (Dis/En)able use of ld.c instructions to check control speculative loads.
9832 If enabled, in case of control speculative load with no speculatively
9833 scheduled dependent instructions this load will be emitted as ld.sa and
9834 ld.c will be used to check it.
9835 The default is 'disable'.
9837 @item -mno-sched-spec-verbose
9838 @itemx -msched-spec-verbose
9839 @opindex -mno-sched-spec-verbose
9840 @opindex -msched-spec-verbose
9841 (Dis/En)able printing of the information about speculative motions.
9843 @item -mno-sched-prefer-non-data-spec-insns
9844 @itemx -msched-prefer-non-data-spec-insns
9845 @opindex -mno-sched-prefer-non-data-spec-insns
9846 @opindex -msched-prefer-non-data-spec-insns
9847 If enabled, data speculative instructions will be chosen for schedule
9848 only if there are no other choices at the moment. This will make
9849 the use of the data speculation much more conservative.
9850 The default is 'disable'.
9852 @item -mno-sched-prefer-non-control-spec-insns
9853 @itemx -msched-prefer-non-control-spec-insns
9854 @opindex -mno-sched-prefer-non-control-spec-insns
9855 @opindex -msched-prefer-non-control-spec-insns
9856 If enabled, control speculative instructions will be chosen for schedule
9857 only if there are no other choices at the moment. This will make
9858 the use of the control speculation much more conservative.
9859 The default is 'disable'.
9861 @item -mno-sched-count-spec-in-critical-path
9862 @itemx -msched-count-spec-in-critical-path
9863 @opindex -mno-sched-count-spec-in-critical-path
9864 @opindex -msched-count-spec-in-critical-path
9865 If enabled, speculative dependencies will be considered during
9866 computation of the instructions priorities. This will make the use of the
9867 speculation a bit more conservative.
9868 The default is 'disable'.
9873 @subsection M32C Options
9874 @cindex M32C options
9877 @item -mcpu=@var{name}
9879 Select the CPU for which code is generated. @var{name} may be one of
9880 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
9881 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
9886 Specifies that the program will be run on the simulator. This causes
9887 an alternate runtime library to be linked in which supports, for
9888 example, file I/O. You must not use this option when generating
9889 programs that will run on real hardware; you must provide your own
9890 runtime library for whatever I/O functions are needed.
9892 @item -memregs=@var{number}
9894 Specifies the number of memory-based pseudo-registers GCC will use
9895 during code generation. These pseudo-registers will be used like real
9896 registers, so there is a tradeoff between GCC's ability to fit the
9897 code into available registers, and the performance penalty of using
9898 memory instead of registers. Note that all modules in a program must
9899 be compiled with the same value for this option. Because of that, you
9900 must not use this option with the default runtime libraries gcc
9905 @node M32R/D Options
9906 @subsection M32R/D Options
9907 @cindex M32R/D options
9909 These @option{-m} options are defined for Renesas M32R/D architectures:
9914 Generate code for the M32R/2@.
9918 Generate code for the M32R/X@.
9922 Generate code for the M32R@. This is the default.
9925 @opindex mmodel=small
9926 Assume all objects live in the lower 16MB of memory (so that their addresses
9927 can be loaded with the @code{ld24} instruction), and assume all subroutines
9928 are reachable with the @code{bl} instruction.
9929 This is the default.
9931 The addressability of a particular object can be set with the
9932 @code{model} attribute.
9934 @item -mmodel=medium
9935 @opindex mmodel=medium
9936 Assume objects may be anywhere in the 32-bit address space (the compiler
9937 will generate @code{seth/add3} instructions to load their addresses), and
9938 assume all subroutines are reachable with the @code{bl} instruction.
9941 @opindex mmodel=large
9942 Assume objects may be anywhere in the 32-bit address space (the compiler
9943 will generate @code{seth/add3} instructions to load their addresses), and
9944 assume subroutines may not be reachable with the @code{bl} instruction
9945 (the compiler will generate the much slower @code{seth/add3/jl}
9946 instruction sequence).
9949 @opindex msdata=none
9950 Disable use of the small data area. Variables will be put into
9951 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
9952 @code{section} attribute has been specified).
9953 This is the default.
9955 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
9956 Objects may be explicitly put in the small data area with the
9957 @code{section} attribute using one of these sections.
9960 @opindex msdata=sdata
9961 Put small global and static data in the small data area, but do not
9962 generate special code to reference them.
9966 Put small global and static data in the small data area, and generate
9967 special instructions to reference them.
9971 @cindex smaller data references
9972 Put global and static objects less than or equal to @var{num} bytes
9973 into the small data or bss sections instead of the normal data or bss
9974 sections. The default value of @var{num} is 8.
9975 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
9976 for this option to have any effect.
9978 All modules should be compiled with the same @option{-G @var{num}} value.
9979 Compiling with different values of @var{num} may or may not work; if it
9980 doesn't the linker will give an error message---incorrect code will not be
9985 Makes the M32R specific code in the compiler display some statistics
9986 that might help in debugging programs.
9989 @opindex malign-loops
9990 Align all loops to a 32-byte boundary.
9992 @item -mno-align-loops
9993 @opindex mno-align-loops
9994 Do not enforce a 32-byte alignment for loops. This is the default.
9996 @item -missue-rate=@var{number}
9997 @opindex missue-rate=@var{number}
9998 Issue @var{number} instructions per cycle. @var{number} can only be 1
10001 @item -mbranch-cost=@var{number}
10002 @opindex mbranch-cost=@var{number}
10003 @var{number} can only be 1 or 2. If it is 1 then branches will be
10004 preferred over conditional code, if it is 2, then the opposite will
10007 @item -mflush-trap=@var{number}
10008 @opindex mflush-trap=@var{number}
10009 Specifies the trap number to use to flush the cache. The default is
10010 12. Valid numbers are between 0 and 15 inclusive.
10012 @item -mno-flush-trap
10013 @opindex mno-flush-trap
10014 Specifies that the cache cannot be flushed by using a trap.
10016 @item -mflush-func=@var{name}
10017 @opindex mflush-func=@var{name}
10018 Specifies the name of the operating system function to call to flush
10019 the cache. The default is @emph{_flush_cache}, but a function call
10020 will only be used if a trap is not available.
10022 @item -mno-flush-func
10023 @opindex mno-flush-func
10024 Indicates that there is no OS function for flushing the cache.
10028 @node M680x0 Options
10029 @subsection M680x0 Options
10030 @cindex M680x0 options
10032 These are the @samp{-m} options defined for the 68000 series. The default
10033 values for these options depends on which style of 68000 was selected when
10034 the compiler was configured; the defaults for the most common choices are
10042 Generate output for a 68000. This is the default
10043 when the compiler is configured for 68000-based systems.
10045 Use this option for microcontrollers with a 68000 or EC000 core,
10046 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
10052 Generate output for a 68020. This is the default
10053 when the compiler is configured for 68020-based systems.
10057 Generate output containing 68881 instructions for floating point.
10058 This is the default for most 68020 systems unless @option{--nfp} was
10059 specified when the compiler was configured.
10063 Generate output for a 68030. This is the default when the compiler is
10064 configured for 68030-based systems.
10068 Generate output for a 68040. This is the default when the compiler is
10069 configured for 68040-based systems.
10071 This option inhibits the use of 68881/68882 instructions that have to be
10072 emulated by software on the 68040. Use this option if your 68040 does not
10073 have code to emulate those instructions.
10077 Generate output for a 68060. This is the default when the compiler is
10078 configured for 68060-based systems.
10080 This option inhibits the use of 68020 and 68881/68882 instructions that
10081 have to be emulated by software on the 68060. Use this option if your 68060
10082 does not have code to emulate those instructions.
10086 Generate output for a CPU32. This is the default
10087 when the compiler is configured for CPU32-based systems.
10089 Use this option for microcontrollers with a
10090 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
10091 68336, 68340, 68341, 68349 and 68360.
10095 Generate output for a 520X ``coldfire'' family cpu. This is the default
10096 when the compiler is configured for 520X-based systems.
10098 Use this option for microcontroller with a 5200 core, including
10099 the MCF5202, MCF5203, MCF5204 and MCF5202.
10103 Generate output for a ColdFire V4e family cpu (e.g.@: 547x/548x).
10104 This includes use of hardware floating point instructions.
10108 Generate output for a 68040, without using any of the new instructions.
10109 This results in code which can run relatively efficiently on either a
10110 68020/68881 or a 68030 or a 68040. The generated code does use the
10111 68881 instructions that are emulated on the 68040.
10115 Generate output for a 68060, without using any of the new instructions.
10116 This results in code which can run relatively efficiently on either a
10117 68020/68881 or a 68030 or a 68040. The generated code does use the
10118 68881 instructions that are emulated on the 68060.
10121 @opindex msoft-float
10122 Generate output containing library calls for floating point.
10123 @strong{Warning:} the requisite libraries are not available for all m68k
10124 targets. Normally the facilities of the machine's usual C compiler are
10125 used, but this can't be done directly in cross-compilation. You must
10126 make your own arrangements to provide suitable library functions for
10127 cross-compilation. The embedded targets @samp{m68k-*-aout} and
10128 @samp{m68k-*-coff} do provide software floating point support.
10132 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10133 Additionally, parameters passed on the stack are also aligned to a
10134 16-bit boundary even on targets whose API mandates promotion to 32-bit.
10137 @opindex mnobitfield
10138 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
10139 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
10143 Do use the bit-field instructions. The @option{-m68020} option implies
10144 @option{-mbitfield}. This is the default if you use a configuration
10145 designed for a 68020.
10149 Use a different function-calling convention, in which functions
10150 that take a fixed number of arguments return with the @code{rtd}
10151 instruction, which pops their arguments while returning. This
10152 saves one instruction in the caller since there is no need to pop
10153 the arguments there.
10155 This calling convention is incompatible with the one normally
10156 used on Unix, so you cannot use it if you need to call libraries
10157 compiled with the Unix compiler.
10159 Also, you must provide function prototypes for all functions that
10160 take variable numbers of arguments (including @code{printf});
10161 otherwise incorrect code will be generated for calls to those
10164 In addition, seriously incorrect code will result if you call a
10165 function with too many arguments. (Normally, extra arguments are
10166 harmlessly ignored.)
10168 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
10169 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
10172 @itemx -mno-align-int
10173 @opindex malign-int
10174 @opindex mno-align-int
10175 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
10176 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
10177 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
10178 Aligning variables on 32-bit boundaries produces code that runs somewhat
10179 faster on processors with 32-bit busses at the expense of more memory.
10181 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
10182 align structures containing the above types differently than
10183 most published application binary interface specifications for the m68k.
10187 Use the pc-relative addressing mode of the 68000 directly, instead of
10188 using a global offset table. At present, this option implies @option{-fpic},
10189 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
10190 not presently supported with @option{-mpcrel}, though this could be supported for
10191 68020 and higher processors.
10193 @item -mno-strict-align
10194 @itemx -mstrict-align
10195 @opindex mno-strict-align
10196 @opindex mstrict-align
10197 Do not (do) assume that unaligned memory references will be handled by
10201 Generate code that allows the data segment to be located in a different
10202 area of memory from the text segment. This allows for execute in place in
10203 an environment without virtual memory management. This option implies
10206 @item -mno-sep-data
10207 Generate code that assumes that the data segment follows the text segment.
10208 This is the default.
10210 @item -mid-shared-library
10211 Generate code that supports shared libraries via the library ID method.
10212 This allows for execute in place and shared libraries in an environment
10213 without virtual memory management. This option implies @option{-fPIC}.
10215 @item -mno-id-shared-library
10216 Generate code that doesn't assume ID based shared libraries are being used.
10217 This is the default.
10219 @item -mshared-library-id=n
10220 Specified the identification number of the ID based shared library being
10221 compiled. Specifying a value of 0 will generate more compact code, specifying
10222 other values will force the allocation of that number to the current
10223 library but is no more space or time efficient than omitting this option.
10227 @node M68hc1x Options
10228 @subsection M68hc1x Options
10229 @cindex M68hc1x options
10231 These are the @samp{-m} options defined for the 68hc11 and 68hc12
10232 microcontrollers. The default values for these options depends on
10233 which style of microcontroller was selected when the compiler was configured;
10234 the defaults for the most common choices are given below.
10241 Generate output for a 68HC11. This is the default
10242 when the compiler is configured for 68HC11-based systems.
10248 Generate output for a 68HC12. This is the default
10249 when the compiler is configured for 68HC12-based systems.
10255 Generate output for a 68HCS12.
10257 @item -mauto-incdec
10258 @opindex mauto-incdec
10259 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
10266 Enable the use of 68HC12 min and max instructions.
10269 @itemx -mno-long-calls
10270 @opindex mlong-calls
10271 @opindex mno-long-calls
10272 Treat all calls as being far away (near). If calls are assumed to be
10273 far away, the compiler will use the @code{call} instruction to
10274 call a function and the @code{rtc} instruction for returning.
10278 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10280 @item -msoft-reg-count=@var{count}
10281 @opindex msoft-reg-count
10282 Specify the number of pseudo-soft registers which are used for the
10283 code generation. The maximum number is 32. Using more pseudo-soft
10284 register may or may not result in better code depending on the program.
10285 The default is 4 for 68HC11 and 2 for 68HC12.
10289 @node MCore Options
10290 @subsection MCore Options
10291 @cindex MCore options
10293 These are the @samp{-m} options defined for the Motorola M*Core
10299 @itemx -mno-hardlit
10301 @opindex mno-hardlit
10302 Inline constants into the code stream if it can be done in two
10303 instructions or less.
10309 Use the divide instruction. (Enabled by default).
10311 @item -mrelax-immediate
10312 @itemx -mno-relax-immediate
10313 @opindex mrelax-immediate
10314 @opindex mno-relax-immediate
10315 Allow arbitrary sized immediates in bit operations.
10317 @item -mwide-bitfields
10318 @itemx -mno-wide-bitfields
10319 @opindex mwide-bitfields
10320 @opindex mno-wide-bitfields
10321 Always treat bit-fields as int-sized.
10323 @item -m4byte-functions
10324 @itemx -mno-4byte-functions
10325 @opindex m4byte-functions
10326 @opindex mno-4byte-functions
10327 Force all functions to be aligned to a four byte boundary.
10329 @item -mcallgraph-data
10330 @itemx -mno-callgraph-data
10331 @opindex mcallgraph-data
10332 @opindex mno-callgraph-data
10333 Emit callgraph information.
10336 @itemx -mno-slow-bytes
10337 @opindex mslow-bytes
10338 @opindex mno-slow-bytes
10339 Prefer word access when reading byte quantities.
10341 @item -mlittle-endian
10342 @itemx -mbig-endian
10343 @opindex mlittle-endian
10344 @opindex mbig-endian
10345 Generate code for a little endian target.
10351 Generate code for the 210 processor.
10355 @subsection MIPS Options
10356 @cindex MIPS options
10362 Generate big-endian code.
10366 Generate little-endian code. This is the default for @samp{mips*el-*-*}
10369 @item -march=@var{arch}
10371 Generate code that will run on @var{arch}, which can be the name of a
10372 generic MIPS ISA, or the name of a particular processor.
10374 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
10375 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
10376 The processor names are:
10377 @samp{4kc}, @samp{4km}, @samp{4kp},
10378 @samp{5kc}, @samp{5kf},
10380 @samp{24k}, @samp{24kc}, @samp{24kf}, @samp{24kx},
10383 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
10384 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
10385 @samp{rm7000}, @samp{rm9000},
10388 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
10389 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
10390 The special value @samp{from-abi} selects the
10391 most compatible architecture for the selected ABI (that is,
10392 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
10394 In processor names, a final @samp{000} can be abbreviated as @samp{k}
10395 (for example, @samp{-march=r2k}). Prefixes are optional, and
10396 @samp{vr} may be written @samp{r}.
10398 GCC defines two macros based on the value of this option. The first
10399 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
10400 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
10401 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
10402 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
10403 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
10405 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
10406 above. In other words, it will have the full prefix and will not
10407 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
10408 the macro names the resolved architecture (either @samp{"mips1"} or
10409 @samp{"mips3"}). It names the default architecture when no
10410 @option{-march} option is given.
10412 @item -mtune=@var{arch}
10414 Optimize for @var{arch}. Among other things, this option controls
10415 the way instructions are scheduled, and the perceived cost of arithmetic
10416 operations. The list of @var{arch} values is the same as for
10419 When this option is not used, GCC will optimize for the processor
10420 specified by @option{-march}. By using @option{-march} and
10421 @option{-mtune} together, it is possible to generate code that will
10422 run on a family of processors, but optimize the code for one
10423 particular member of that family.
10425 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
10426 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
10427 @samp{-march} ones described above.
10431 Equivalent to @samp{-march=mips1}.
10435 Equivalent to @samp{-march=mips2}.
10439 Equivalent to @samp{-march=mips3}.
10443 Equivalent to @samp{-march=mips4}.
10447 Equivalent to @samp{-march=mips32}.
10451 Equivalent to @samp{-march=mips32r2}.
10455 Equivalent to @samp{-march=mips64}.
10460 @opindex mno-mips16
10461 Generate (do not generate) MIPS16 code. If GCC is targetting a
10462 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
10474 Generate code for the given ABI@.
10476 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
10477 generates 64-bit code when you select a 64-bit architecture, but you
10478 can use @option{-mgp32} to get 32-bit code instead.
10480 For information about the O64 ABI, see
10481 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
10484 @itemx -mno-abicalls
10486 @opindex mno-abicalls
10487 Generate (do not generate) code that is suitable for SVR4-style
10488 dynamic objects. @option{-mabicalls} is the default for SVR4-based
10493 Generate (do not generate) code that is fully position-independent,
10494 and that can therefore be linked into shared libraries. This option
10495 only affects @option{-mabicalls}.
10497 All @option{-mabicalls} code has traditionally been position-independent,
10498 regardless of options like @option{-fPIC} and @option{-fpic}. However,
10499 as an extension, the GNU toolchain allows executables to use absolute
10500 accesses for locally-binding symbols. It can also use shorter GP
10501 initialization sequences and generate direct calls to locally-defined
10502 functions. This mode is selected by @option{-mno-shared}.
10504 @option{-mno-shared} depends on binutils 2.16 or higher and generates
10505 objects that can only be linked by the GNU linker. However, the option
10506 does not affect the ABI of the final executable; it only affects the ABI
10507 of relocatable objects. Using @option{-mno-shared} will generally make
10508 executables both smaller and quicker.
10510 @option{-mshared} is the default.
10516 Lift (do not lift) the usual restrictions on the size of the global
10519 GCC normally uses a single instruction to load values from the GOT@.
10520 While this is relatively efficient, it will only work if the GOT
10521 is smaller than about 64k. Anything larger will cause the linker
10522 to report an error such as:
10524 @cindex relocation truncated to fit (MIPS)
10526 relocation truncated to fit: R_MIPS_GOT16 foobar
10529 If this happens, you should recompile your code with @option{-mxgot}.
10530 It should then work with very large GOTs, although it will also be
10531 less efficient, since it will take three instructions to fetch the
10532 value of a global symbol.
10534 Note that some linkers can create multiple GOTs. If you have such a
10535 linker, you should only need to use @option{-mxgot} when a single object
10536 file accesses more than 64k's worth of GOT entries. Very few do.
10538 These options have no effect unless GCC is generating position
10543 Assume that general-purpose registers are 32 bits wide.
10547 Assume that general-purpose registers are 64 bits wide.
10551 Assume that floating-point registers are 32 bits wide.
10555 Assume that floating-point registers are 64 bits wide.
10558 @opindex mhard-float
10559 Use floating-point coprocessor instructions.
10562 @opindex msoft-float
10563 Do not use floating-point coprocessor instructions. Implement
10564 floating-point calculations using library calls instead.
10566 @item -msingle-float
10567 @opindex msingle-float
10568 Assume that the floating-point coprocessor only supports single-precision
10571 @itemx -mdouble-float
10572 @opindex mdouble-float
10573 Assume that the floating-point coprocessor supports double-precision
10574 operations. This is the default.
10580 Use (do not use) the MIPS DSP ASE. @xref{MIPS DSP Built-in Functions}.
10582 @itemx -mpaired-single
10583 @itemx -mno-paired-single
10584 @opindex mpaired-single
10585 @opindex mno-paired-single
10586 Use (do not use) paired-single floating-point instructions.
10587 @xref{MIPS Paired-Single Support}. This option can only be used
10588 when generating 64-bit code and requires hardware floating-point
10589 support to be enabled.
10594 @opindex mno-mips3d
10595 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
10596 The option @option{-mips3d} implies @option{-mpaired-single}.
10600 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
10601 an explanation of the default and the way that the pointer size is
10606 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
10608 The default size of @code{int}s, @code{long}s and pointers depends on
10609 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
10610 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
10611 32-bit @code{long}s. Pointers are the same size as @code{long}s,
10612 or the same size as integer registers, whichever is smaller.
10618 Assume (do not assume) that all symbols have 32-bit values, regardless
10619 of the selected ABI@. This option is useful in combination with
10620 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
10621 to generate shorter and faster references to symbolic addresses.
10625 @cindex smaller data references (MIPS)
10626 @cindex gp-relative references (MIPS)
10627 Put global and static items less than or equal to @var{num} bytes into
10628 the small data or bss section instead of the normal data or bss section.
10629 This allows the data to be accessed using a single instruction.
10631 All modules should be compiled with the same @option{-G @var{num}}
10634 @item -membedded-data
10635 @itemx -mno-embedded-data
10636 @opindex membedded-data
10637 @opindex mno-embedded-data
10638 Allocate variables to the read-only data section first if possible, then
10639 next in the small data section if possible, otherwise in data. This gives
10640 slightly slower code than the default, but reduces the amount of RAM required
10641 when executing, and thus may be preferred for some embedded systems.
10643 @item -muninit-const-in-rodata
10644 @itemx -mno-uninit-const-in-rodata
10645 @opindex muninit-const-in-rodata
10646 @opindex mno-uninit-const-in-rodata
10647 Put uninitialized @code{const} variables in the read-only data section.
10648 This option is only meaningful in conjunction with @option{-membedded-data}.
10650 @item -msplit-addresses
10651 @itemx -mno-split-addresses
10652 @opindex msplit-addresses
10653 @opindex mno-split-addresses
10654 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
10655 relocation operators. This option has been superseded by
10656 @option{-mexplicit-relocs} but is retained for backwards compatibility.
10658 @item -mexplicit-relocs
10659 @itemx -mno-explicit-relocs
10660 @opindex mexplicit-relocs
10661 @opindex mno-explicit-relocs
10662 Use (do not use) assembler relocation operators when dealing with symbolic
10663 addresses. The alternative, selected by @option{-mno-explicit-relocs},
10664 is to use assembler macros instead.
10666 @option{-mexplicit-relocs} is the default if GCC was configured
10667 to use an assembler that supports relocation operators.
10669 @item -mcheck-zero-division
10670 @itemx -mno-check-zero-division
10671 @opindex mcheck-zero-division
10672 @opindex mno-check-zero-division
10673 Trap (do not trap) on integer division by zero. The default is
10674 @option{-mcheck-zero-division}.
10676 @item -mdivide-traps
10677 @itemx -mdivide-breaks
10678 @opindex mdivide-traps
10679 @opindex mdivide-breaks
10680 MIPS systems check for division by zero by generating either a
10681 conditional trap or a break instruction. Using traps results in
10682 smaller code, but is only supported on MIPS II and later. Also, some
10683 versions of the Linux kernel have a bug that prevents trap from
10684 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
10685 allow conditional traps on architectures that support them and
10686 @option{-mdivide-breaks} to force the use of breaks.
10688 The default is usually @option{-mdivide-traps}, but this can be
10689 overridden at configure time using @option{--with-divide=breaks}.
10690 Divide-by-zero checks can be completely disabled using
10691 @option{-mno-check-zero-division}.
10696 @opindex mno-memcpy
10697 Force (do not force) the use of @code{memcpy()} for non-trivial block
10698 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
10699 most constant-sized copies.
10702 @itemx -mno-long-calls
10703 @opindex mlong-calls
10704 @opindex mno-long-calls
10705 Disable (do not disable) use of the @code{jal} instruction. Calling
10706 functions using @code{jal} is more efficient but requires the caller
10707 and callee to be in the same 256 megabyte segment.
10709 This option has no effect on abicalls code. The default is
10710 @option{-mno-long-calls}.
10716 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
10717 instructions, as provided by the R4650 ISA@.
10720 @itemx -mno-fused-madd
10721 @opindex mfused-madd
10722 @opindex mno-fused-madd
10723 Enable (disable) use of the floating point multiply-accumulate
10724 instructions, when they are available. The default is
10725 @option{-mfused-madd}.
10727 When multiply-accumulate instructions are used, the intermediate
10728 product is calculated to infinite precision and is not subject to
10729 the FCSR Flush to Zero bit. This may be undesirable in some
10734 Tell the MIPS assembler to not run its preprocessor over user
10735 assembler files (with a @samp{.s} suffix) when assembling them.
10738 @itemx -mno-fix-r4000
10739 @opindex mfix-r4000
10740 @opindex mno-fix-r4000
10741 Work around certain R4000 CPU errata:
10744 A double-word or a variable shift may give an incorrect result if executed
10745 immediately after starting an integer division.
10747 A double-word or a variable shift may give an incorrect result if executed
10748 while an integer multiplication is in progress.
10750 An integer division may give an incorrect result if started in a delay slot
10751 of a taken branch or a jump.
10755 @itemx -mno-fix-r4400
10756 @opindex mfix-r4400
10757 @opindex mno-fix-r4400
10758 Work around certain R4400 CPU errata:
10761 A double-word or a variable shift may give an incorrect result if executed
10762 immediately after starting an integer division.
10766 @itemx -mno-fix-vr4120
10767 @opindex mfix-vr4120
10768 Work around certain VR4120 errata:
10771 @code{dmultu} does not always produce the correct result.
10773 @code{div} and @code{ddiv} do not always produce the correct result if one
10774 of the operands is negative.
10776 The workarounds for the division errata rely on special functions in
10777 @file{libgcc.a}. At present, these functions are only provided by
10778 the @code{mips64vr*-elf} configurations.
10780 Other VR4120 errata require a nop to be inserted between certain pairs of
10781 instructions. These errata are handled by the assembler, not by GCC itself.
10784 @opindex mfix-vr4130
10785 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
10786 workarounds are implemented by the assembler rather than by GCC,
10787 although GCC will avoid using @code{mflo} and @code{mfhi} if the
10788 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
10789 instructions are available instead.
10792 @itemx -mno-fix-sb1
10794 Work around certain SB-1 CPU core errata.
10795 (This flag currently works around the SB-1 revision 2
10796 ``F1'' and ``F2'' floating point errata.)
10798 @item -mflush-func=@var{func}
10799 @itemx -mno-flush-func
10800 @opindex mflush-func
10801 Specifies the function to call to flush the I and D caches, or to not
10802 call any such function. If called, the function must take the same
10803 arguments as the common @code{_flush_func()}, that is, the address of the
10804 memory range for which the cache is being flushed, the size of the
10805 memory range, and the number 3 (to flush both caches). The default
10806 depends on the target GCC was configured for, but commonly is either
10807 @samp{_flush_func} or @samp{__cpu_flush}.
10809 @item -mbranch-likely
10810 @itemx -mno-branch-likely
10811 @opindex mbranch-likely
10812 @opindex mno-branch-likely
10813 Enable or disable use of Branch Likely instructions, regardless of the
10814 default for the selected architecture. By default, Branch Likely
10815 instructions may be generated if they are supported by the selected
10816 architecture. An exception is for the MIPS32 and MIPS64 architectures
10817 and processors which implement those architectures; for those, Branch
10818 Likely instructions will not be generated by default because the MIPS32
10819 and MIPS64 architectures specifically deprecate their use.
10821 @item -mfp-exceptions
10822 @itemx -mno-fp-exceptions
10823 @opindex mfp-exceptions
10824 Specifies whether FP exceptions are enabled. This affects how we schedule
10825 FP instructions for some processors. The default is that FP exceptions are
10828 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
10829 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
10832 @item -mvr4130-align
10833 @itemx -mno-vr4130-align
10834 @opindex mvr4130-align
10835 The VR4130 pipeline is two-way superscalar, but can only issue two
10836 instructions together if the first one is 8-byte aligned. When this
10837 option is enabled, GCC will align pairs of instructions that it
10838 thinks should execute in parallel.
10840 This option only has an effect when optimizing for the VR4130.
10841 It normally makes code faster, but at the expense of making it bigger.
10842 It is enabled by default at optimization level @option{-O3}.
10846 @subsection MMIX Options
10847 @cindex MMIX Options
10849 These options are defined for the MMIX:
10853 @itemx -mno-libfuncs
10855 @opindex mno-libfuncs
10856 Specify that intrinsic library functions are being compiled, passing all
10857 values in registers, no matter the size.
10860 @itemx -mno-epsilon
10862 @opindex mno-epsilon
10863 Generate floating-point comparison instructions that compare with respect
10864 to the @code{rE} epsilon register.
10866 @item -mabi=mmixware
10868 @opindex mabi-mmixware
10870 Generate code that passes function parameters and return values that (in
10871 the called function) are seen as registers @code{$0} and up, as opposed to
10872 the GNU ABI which uses global registers @code{$231} and up.
10874 @item -mzero-extend
10875 @itemx -mno-zero-extend
10876 @opindex mzero-extend
10877 @opindex mno-zero-extend
10878 When reading data from memory in sizes shorter than 64 bits, use (do not
10879 use) zero-extending load instructions by default, rather than
10880 sign-extending ones.
10883 @itemx -mno-knuthdiv
10885 @opindex mno-knuthdiv
10886 Make the result of a division yielding a remainder have the same sign as
10887 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10888 remainder follows the sign of the dividend. Both methods are
10889 arithmetically valid, the latter being almost exclusively used.
10891 @item -mtoplevel-symbols
10892 @itemx -mno-toplevel-symbols
10893 @opindex mtoplevel-symbols
10894 @opindex mno-toplevel-symbols
10895 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10896 code can be used with the @code{PREFIX} assembly directive.
10900 Generate an executable in the ELF format, rather than the default
10901 @samp{mmo} format used by the @command{mmix} simulator.
10903 @item -mbranch-predict
10904 @itemx -mno-branch-predict
10905 @opindex mbranch-predict
10906 @opindex mno-branch-predict
10907 Use (do not use) the probable-branch instructions, when static branch
10908 prediction indicates a probable branch.
10910 @item -mbase-addresses
10911 @itemx -mno-base-addresses
10912 @opindex mbase-addresses
10913 @opindex mno-base-addresses
10914 Generate (do not generate) code that uses @emph{base addresses}. Using a
10915 base address automatically generates a request (handled by the assembler
10916 and the linker) for a constant to be set up in a global register. The
10917 register is used for one or more base address requests within the range 0
10918 to 255 from the value held in the register. The generally leads to short
10919 and fast code, but the number of different data items that can be
10920 addressed is limited. This means that a program that uses lots of static
10921 data may require @option{-mno-base-addresses}.
10923 @item -msingle-exit
10924 @itemx -mno-single-exit
10925 @opindex msingle-exit
10926 @opindex mno-single-exit
10927 Force (do not force) generated code to have a single exit point in each
10931 @node MN10300 Options
10932 @subsection MN10300 Options
10933 @cindex MN10300 options
10935 These @option{-m} options are defined for Matsushita MN10300 architectures:
10940 Generate code to avoid bugs in the multiply instructions for the MN10300
10941 processors. This is the default.
10943 @item -mno-mult-bug
10944 @opindex mno-mult-bug
10945 Do not generate code to avoid bugs in the multiply instructions for the
10946 MN10300 processors.
10950 Generate code which uses features specific to the AM33 processor.
10954 Do not generate code which uses features specific to the AM33 processor. This
10957 @item -mreturn-pointer-on-d0
10958 @opindex mreturn-pointer-on-d0
10959 When generating a function which returns a pointer, return the pointer
10960 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
10961 only in a0, and attempts to call such functions without a prototype
10962 would result in errors. Note that this option is on by default; use
10963 @option{-mno-return-pointer-on-d0} to disable it.
10967 Do not link in the C run-time initialization object file.
10971 Indicate to the linker that it should perform a relaxation optimization pass
10972 to shorten branches, calls and absolute memory addresses. This option only
10973 has an effect when used on the command line for the final link step.
10975 This option makes symbolic debugging impossible.
10979 @subsection MT Options
10982 These @option{-m} options are defined for Morpho MT architectures:
10986 @item -march=@var{cpu-type}
10988 Generate code that will run on @var{cpu-type}, which is the name of a system
10989 representing a certain processor type. Possible values for
10990 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
10991 @samp{ms1-16-003} and @samp{ms2}.
10993 When this option is not used, the default is @option{-march=ms1-16-002}.
10997 Use byte loads and stores when generating code.
11001 Do not use byte loads and stores when generating code.
11005 Use simulator runtime
11009 Do not link in the C run-time initialization object file
11010 @file{crti.o}. Other run-time initialization and termination files
11011 such as @file{startup.o} and @file{exit.o} are still included on the
11012 linker command line.
11016 @node PDP-11 Options
11017 @subsection PDP-11 Options
11018 @cindex PDP-11 Options
11020 These options are defined for the PDP-11:
11025 Use hardware FPP floating point. This is the default. (FIS floating
11026 point on the PDP-11/40 is not supported.)
11029 @opindex msoft-float
11030 Do not use hardware floating point.
11034 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
11038 Return floating-point results in memory. This is the default.
11042 Generate code for a PDP-11/40.
11046 Generate code for a PDP-11/45. This is the default.
11050 Generate code for a PDP-11/10.
11052 @item -mbcopy-builtin
11053 @opindex bcopy-builtin
11054 Use inline @code{movmemhi} patterns for copying memory. This is the
11059 Do not use inline @code{movmemhi} patterns for copying memory.
11065 Use 16-bit @code{int}. This is the default.
11071 Use 32-bit @code{int}.
11074 @itemx -mno-float32
11076 @opindex mno-float32
11077 Use 64-bit @code{float}. This is the default.
11080 @itemx -mno-float64
11082 @opindex mno-float64
11083 Use 32-bit @code{float}.
11087 Use @code{abshi2} pattern. This is the default.
11091 Do not use @code{abshi2} pattern.
11093 @item -mbranch-expensive
11094 @opindex mbranch-expensive
11095 Pretend that branches are expensive. This is for experimenting with
11096 code generation only.
11098 @item -mbranch-cheap
11099 @opindex mbranch-cheap
11100 Do not pretend that branches are expensive. This is the default.
11104 Generate code for a system with split I&D@.
11108 Generate code for a system without split I&D@. This is the default.
11112 Use Unix assembler syntax. This is the default when configured for
11113 @samp{pdp11-*-bsd}.
11117 Use DEC assembler syntax. This is the default when configured for any
11118 PDP-11 target other than @samp{pdp11-*-bsd}.
11121 @node PowerPC Options
11122 @subsection PowerPC Options
11123 @cindex PowerPC options
11125 These are listed under @xref{RS/6000 and PowerPC Options}.
11127 @node RS/6000 and PowerPC Options
11128 @subsection IBM RS/6000 and PowerPC Options
11129 @cindex RS/6000 and PowerPC Options
11130 @cindex IBM RS/6000 and PowerPC Options
11132 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
11139 @itemx -mno-powerpc
11140 @itemx -mpowerpc-gpopt
11141 @itemx -mno-powerpc-gpopt
11142 @itemx -mpowerpc-gfxopt
11143 @itemx -mno-powerpc-gfxopt
11145 @itemx -mno-powerpc64
11149 @itemx -mno-popcntb
11155 @opindex mno-power2
11157 @opindex mno-powerpc
11158 @opindex mpowerpc-gpopt
11159 @opindex mno-powerpc-gpopt
11160 @opindex mpowerpc-gfxopt
11161 @opindex mno-powerpc-gfxopt
11162 @opindex mpowerpc64
11163 @opindex mno-powerpc64
11167 @opindex mno-popcntb
11170 GCC supports two related instruction set architectures for the
11171 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
11172 instructions supported by the @samp{rios} chip set used in the original
11173 RS/6000 systems and the @dfn{PowerPC} instruction set is the
11174 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
11175 the IBM 4xx, 6xx, and follow-on microprocessors.
11177 Neither architecture is a subset of the other. However there is a
11178 large common subset of instructions supported by both. An MQ
11179 register is included in processors supporting the POWER architecture.
11181 You use these options to specify which instructions are available on the
11182 processor you are using. The default value of these options is
11183 determined when configuring GCC@. Specifying the
11184 @option{-mcpu=@var{cpu_type}} overrides the specification of these
11185 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
11186 rather than the options listed above.
11188 The @option{-mpower} option allows GCC to generate instructions that
11189 are found only in the POWER architecture and to use the MQ register.
11190 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
11191 to generate instructions that are present in the POWER2 architecture but
11192 not the original POWER architecture.
11194 The @option{-mpowerpc} option allows GCC to generate instructions that
11195 are found only in the 32-bit subset of the PowerPC architecture.
11196 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
11197 GCC to use the optional PowerPC architecture instructions in the
11198 General Purpose group, including floating-point square root. Specifying
11199 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
11200 use the optional PowerPC architecture instructions in the Graphics
11201 group, including floating-point select.
11203 The @option{-mmfcrf} option allows GCC to generate the move from
11204 condition register field instruction implemented on the POWER4
11205 processor and other processors that support the PowerPC V2.01
11207 The @option{-mpopcntb} option allows GCC to generate the popcount and
11208 double precision FP reciprocal estimate instruction implemented on the
11209 POWER5 processor and other processors that support the PowerPC V2.02
11211 The @option{-mfprnd} option allows GCC to generate the FP round to
11212 integer instructions implemented on the POWER5+ processor and other
11213 processors that support the PowerPC V2.03 architecture.
11215 The @option{-mpowerpc64} option allows GCC to generate the additional
11216 64-bit instructions that are found in the full PowerPC64 architecture
11217 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
11218 @option{-mno-powerpc64}.
11220 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
11221 will use only the instructions in the common subset of both
11222 architectures plus some special AIX common-mode calls, and will not use
11223 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
11224 permits GCC to use any instruction from either architecture and to
11225 allow use of the MQ register; specify this for the Motorola MPC601.
11227 @item -mnew-mnemonics
11228 @itemx -mold-mnemonics
11229 @opindex mnew-mnemonics
11230 @opindex mold-mnemonics
11231 Select which mnemonics to use in the generated assembler code. With
11232 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
11233 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
11234 assembler mnemonics defined for the POWER architecture. Instructions
11235 defined in only one architecture have only one mnemonic; GCC uses that
11236 mnemonic irrespective of which of these options is specified.
11238 GCC defaults to the mnemonics appropriate for the architecture in
11239 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
11240 value of these option. Unless you are building a cross-compiler, you
11241 should normally not specify either @option{-mnew-mnemonics} or
11242 @option{-mold-mnemonics}, but should instead accept the default.
11244 @item -mcpu=@var{cpu_type}
11246 Set architecture type, register usage, choice of mnemonics, and
11247 instruction scheduling parameters for machine type @var{cpu_type}.
11248 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
11249 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
11250 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
11251 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
11252 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
11253 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
11254 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
11255 @samp{power4}, @samp{power5}, @samp{power5+},
11256 @samp{common}, @samp{powerpc}, @samp{powerpc64},
11257 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
11259 @option{-mcpu=common} selects a completely generic processor. Code
11260 generated under this option will run on any POWER or PowerPC processor.
11261 GCC will use only the instructions in the common subset of both
11262 architectures, and will not use the MQ register. GCC assumes a generic
11263 processor model for scheduling purposes.
11265 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
11266 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
11267 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
11268 types, with an appropriate, generic processor model assumed for
11269 scheduling purposes.
11271 The other options specify a specific processor. Code generated under
11272 those options will run best on that processor, and may not run at all on
11275 The @option{-mcpu} options automatically enable or disable the
11276 following options: @option{-maltivec}, @option{-mfprnd},
11277 @option{-mhard-float}, @option{-mmfcrf}, @option{-mmultiple},
11278 @option{-mnew-mnemonics}, @option{-mpopcntb}, @option{-mpower},
11279 @option{-mpower2}, @option{-mpowerpc64}, @option{-mpowerpc-gpopt},
11280 @option{-mpowerpc-gfxopt}, @option{-mstring}, @option{-mmulhw}, @option{dlmzb}.
11281 The particular options
11282 set for any particular CPU will vary between compiler versions,
11283 depending on what setting seems to produce optimal code for that CPU;
11284 it doesn't necessarily reflect the actual hardware's capabilities. If
11285 you wish to set an individual option to a particular value, you may
11286 specify it after the @option{-mcpu} option, like @samp{-mcpu=970
11289 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
11290 not enabled or disabled by the @option{-mcpu} option at present because
11291 AIX does not have full support for these options. You may still
11292 enable or disable them individually if you're sure it'll work in your
11295 @item -mtune=@var{cpu_type}
11297 Set the instruction scheduling parameters for machine type
11298 @var{cpu_type}, but do not set the architecture type, register usage, or
11299 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
11300 values for @var{cpu_type} are used for @option{-mtune} as for
11301 @option{-mcpu}. If both are specified, the code generated will use the
11302 architecture, registers, and mnemonics set by @option{-mcpu}, but the
11303 scheduling parameters set by @option{-mtune}.
11309 Generate code to compute division as reciprocal estimate and iterative
11310 refinement, creating opportunities for increased throughput. This
11311 feature requires: optional PowerPC Graphics instruction set for single
11312 precision and FRE instruction for double precision, assuming divides
11313 cannot generate user-visible traps, and the domain values not include
11314 Infinities, denormals or zero denominator.
11317 @itemx -mno-altivec
11319 @opindex mno-altivec
11320 Generate code that uses (does not use) AltiVec instructions, and also
11321 enable the use of built-in functions that allow more direct access to
11322 the AltiVec instruction set. You may also need to set
11323 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
11329 @opindex mno-vrsave
11330 Generate VRSAVE instructions when generating AltiVec code.
11333 @opindex msecure-plt
11334 Generate code that allows ld and ld.so to build executables and shared
11335 libraries with non-exec .plt and .got sections. This is a PowerPC
11336 32-bit SYSV ABI option.
11340 Generate code that uses a BSS .plt section that ld.so fills in, and
11341 requires .plt and .got sections that are both writable and executable.
11342 This is a PowerPC 32-bit SYSV ABI option.
11348 This switch enables or disables the generation of ISEL instructions.
11350 @item -misel=@var{yes/no}
11351 This switch has been deprecated. Use @option{-misel} and
11352 @option{-mno-isel} instead.
11358 This switch enables or disables the generation of SPE simd
11361 @item -mspe=@var{yes/no}
11362 This option has been deprecated. Use @option{-mspe} and
11363 @option{-mno-spe} instead.
11365 @item -mfloat-gprs=@var{yes/single/double/no}
11366 @itemx -mfloat-gprs
11367 @opindex mfloat-gprs
11368 This switch enables or disables the generation of floating point
11369 operations on the general purpose registers for architectures that
11372 The argument @var{yes} or @var{single} enables the use of
11373 single-precision floating point operations.
11375 The argument @var{double} enables the use of single and
11376 double-precision floating point operations.
11378 The argument @var{no} disables floating point operations on the
11379 general purpose registers.
11381 This option is currently only available on the MPC854x.
11387 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
11388 targets (including GNU/Linux). The 32-bit environment sets int, long
11389 and pointer to 32 bits and generates code that runs on any PowerPC
11390 variant. The 64-bit environment sets int to 32 bits and long and
11391 pointer to 64 bits, and generates code for PowerPC64, as for
11392 @option{-mpowerpc64}.
11395 @itemx -mno-fp-in-toc
11396 @itemx -mno-sum-in-toc
11397 @itemx -mminimal-toc
11399 @opindex mno-fp-in-toc
11400 @opindex mno-sum-in-toc
11401 @opindex mminimal-toc
11402 Modify generation of the TOC (Table Of Contents), which is created for
11403 every executable file. The @option{-mfull-toc} option is selected by
11404 default. In that case, GCC will allocate at least one TOC entry for
11405 each unique non-automatic variable reference in your program. GCC
11406 will also place floating-point constants in the TOC@. However, only
11407 16,384 entries are available in the TOC@.
11409 If you receive a linker error message that saying you have overflowed
11410 the available TOC space, you can reduce the amount of TOC space used
11411 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
11412 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
11413 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
11414 generate code to calculate the sum of an address and a constant at
11415 run-time instead of putting that sum into the TOC@. You may specify one
11416 or both of these options. Each causes GCC to produce very slightly
11417 slower and larger code at the expense of conserving TOC space.
11419 If you still run out of space in the TOC even when you specify both of
11420 these options, specify @option{-mminimal-toc} instead. This option causes
11421 GCC to make only one TOC entry for every file. When you specify this
11422 option, GCC will produce code that is slower and larger but which
11423 uses extremely little TOC space. You may wish to use this option
11424 only on files that contain less frequently executed code.
11430 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
11431 @code{long} type, and the infrastructure needed to support them.
11432 Specifying @option{-maix64} implies @option{-mpowerpc64} and
11433 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
11434 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
11437 @itemx -mno-xl-compat
11438 @opindex mxl-compat
11439 @opindex mno-xl-compat
11440 Produce code that conforms more closely to IBM XL compiler semantics
11441 when using AIX-compatible ABI. Pass floating-point arguments to
11442 prototyped functions beyond the register save area (RSA) on the stack
11443 in addition to argument FPRs. Do not assume that most significant
11444 double in 128-bit long double value is properly rounded when comparing
11445 values and converting to double. Use XL symbol names for long double
11448 The AIX calling convention was extended but not initially documented to
11449 handle an obscure K&R C case of calling a function that takes the
11450 address of its arguments with fewer arguments than declared. IBM XL
11451 compilers access floating point arguments which do not fit in the
11452 RSA from the stack when a subroutine is compiled without
11453 optimization. Because always storing floating-point arguments on the
11454 stack is inefficient and rarely needed, this option is not enabled by
11455 default and only is necessary when calling subroutines compiled by IBM
11456 XL compilers without optimization.
11460 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
11461 application written to use message passing with special startup code to
11462 enable the application to run. The system must have PE installed in the
11463 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
11464 must be overridden with the @option{-specs=} option to specify the
11465 appropriate directory location. The Parallel Environment does not
11466 support threads, so the @option{-mpe} option and the @option{-pthread}
11467 option are incompatible.
11469 @item -malign-natural
11470 @itemx -malign-power
11471 @opindex malign-natural
11472 @opindex malign-power
11473 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
11474 @option{-malign-natural} overrides the ABI-defined alignment of larger
11475 types, such as floating-point doubles, on their natural size-based boundary.
11476 The option @option{-malign-power} instructs GCC to follow the ABI-specified
11477 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
11479 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
11483 @itemx -mhard-float
11484 @opindex msoft-float
11485 @opindex mhard-float
11486 Generate code that does not use (uses) the floating-point register set.
11487 Software floating point emulation is provided if you use the
11488 @option{-msoft-float} option, and pass the option to GCC when linking.
11491 @itemx -mno-multiple
11493 @opindex mno-multiple
11494 Generate code that uses (does not use) the load multiple word
11495 instructions and the store multiple word instructions. These
11496 instructions are generated by default on POWER systems, and not
11497 generated on PowerPC systems. Do not use @option{-mmultiple} on little
11498 endian PowerPC systems, since those instructions do not work when the
11499 processor is in little endian mode. The exceptions are PPC740 and
11500 PPC750 which permit the instructions usage in little endian mode.
11505 @opindex mno-string
11506 Generate code that uses (does not use) the load string instructions
11507 and the store string word instructions to save multiple registers and
11508 do small block moves. These instructions are generated by default on
11509 POWER systems, and not generated on PowerPC systems. Do not use
11510 @option{-mstring} on little endian PowerPC systems, since those
11511 instructions do not work when the processor is in little endian mode.
11512 The exceptions are PPC740 and PPC750 which permit the instructions
11513 usage in little endian mode.
11518 @opindex mno-update
11519 Generate code that uses (does not use) the load or store instructions
11520 that update the base register to the address of the calculated memory
11521 location. These instructions are generated by default. If you use
11522 @option{-mno-update}, there is a small window between the time that the
11523 stack pointer is updated and the address of the previous frame is
11524 stored, which means code that walks the stack frame across interrupts or
11525 signals may get corrupted data.
11528 @itemx -mno-fused-madd
11529 @opindex mfused-madd
11530 @opindex mno-fused-madd
11531 Generate code that uses (does not use) the floating point multiply and
11532 accumulate instructions. These instructions are generated by default if
11533 hardware floating is used.
11539 Generate code that uses (does not use) the half-word multiply and
11540 multiply-accumulate instructions on the IBM 405 and 440 processors.
11541 These instructions are generated by default when targetting those
11548 Generate code that uses (does not use) the string-search @samp{dlmzb}
11549 instruction on the IBM 405 and 440 processors. This instruction is
11550 generated by default when targetting those processors.
11552 @item -mno-bit-align
11554 @opindex mno-bit-align
11555 @opindex mbit-align
11556 On System V.4 and embedded PowerPC systems do not (do) force structures
11557 and unions that contain bit-fields to be aligned to the base type of the
11560 For example, by default a structure containing nothing but 8
11561 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
11562 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
11563 the structure would be aligned to a 1 byte boundary and be one byte in
11566 @item -mno-strict-align
11567 @itemx -mstrict-align
11568 @opindex mno-strict-align
11569 @opindex mstrict-align
11570 On System V.4 and embedded PowerPC systems do not (do) assume that
11571 unaligned memory references will be handled by the system.
11573 @item -mrelocatable
11574 @itemx -mno-relocatable
11575 @opindex mrelocatable
11576 @opindex mno-relocatable
11577 On embedded PowerPC systems generate code that allows (does not allow)
11578 the program to be relocated to a different address at runtime. If you
11579 use @option{-mrelocatable} on any module, all objects linked together must
11580 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
11582 @item -mrelocatable-lib
11583 @itemx -mno-relocatable-lib
11584 @opindex mrelocatable-lib
11585 @opindex mno-relocatable-lib
11586 On embedded PowerPC systems generate code that allows (does not allow)
11587 the program to be relocated to a different address at runtime. Modules
11588 compiled with @option{-mrelocatable-lib} can be linked with either modules
11589 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
11590 with modules compiled with the @option{-mrelocatable} options.
11596 On System V.4 and embedded PowerPC systems do not (do) assume that
11597 register 2 contains a pointer to a global area pointing to the addresses
11598 used in the program.
11601 @itemx -mlittle-endian
11603 @opindex mlittle-endian
11604 On System V.4 and embedded PowerPC systems compile code for the
11605 processor in little endian mode. The @option{-mlittle-endian} option is
11606 the same as @option{-mlittle}.
11609 @itemx -mbig-endian
11611 @opindex mbig-endian
11612 On System V.4 and embedded PowerPC systems compile code for the
11613 processor in big endian mode. The @option{-mbig-endian} option is
11614 the same as @option{-mbig}.
11616 @item -mdynamic-no-pic
11617 @opindex mdynamic-no-pic
11618 On Darwin and Mac OS X systems, compile code so that it is not
11619 relocatable, but that its external references are relocatable. The
11620 resulting code is suitable for applications, but not shared
11623 @item -mprioritize-restricted-insns=@var{priority}
11624 @opindex mprioritize-restricted-insns
11625 This option controls the priority that is assigned to
11626 dispatch-slot restricted instructions during the second scheduling
11627 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
11628 @var{no/highest/second-highest} priority to dispatch slot restricted
11631 @item -msched-costly-dep=@var{dependence_type}
11632 @opindex msched-costly-dep
11633 This option controls which dependences are considered costly
11634 by the target during instruction scheduling. The argument
11635 @var{dependence_type} takes one of the following values:
11636 @var{no}: no dependence is costly,
11637 @var{all}: all dependences are costly,
11638 @var{true_store_to_load}: a true dependence from store to load is costly,
11639 @var{store_to_load}: any dependence from store to load is costly,
11640 @var{number}: any dependence which latency >= @var{number} is costly.
11642 @item -minsert-sched-nops=@var{scheme}
11643 @opindex minsert-sched-nops
11644 This option controls which nop insertion scheme will be used during
11645 the second scheduling pass. The argument @var{scheme} takes one of the
11647 @var{no}: Don't insert nops.
11648 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
11649 according to the scheduler's grouping.
11650 @var{regroup_exact}: Insert nops to force costly dependent insns into
11651 separate groups. Insert exactly as many nops as needed to force an insn
11652 to a new group, according to the estimated processor grouping.
11653 @var{number}: Insert nops to force costly dependent insns into
11654 separate groups. Insert @var{number} nops to force an insn to a new group.
11657 @opindex mcall-sysv
11658 On System V.4 and embedded PowerPC systems compile code using calling
11659 conventions that adheres to the March 1995 draft of the System V
11660 Application Binary Interface, PowerPC processor supplement. This is the
11661 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
11663 @item -mcall-sysv-eabi
11664 @opindex mcall-sysv-eabi
11665 Specify both @option{-mcall-sysv} and @option{-meabi} options.
11667 @item -mcall-sysv-noeabi
11668 @opindex mcall-sysv-noeabi
11669 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
11671 @item -mcall-solaris
11672 @opindex mcall-solaris
11673 On System V.4 and embedded PowerPC systems compile code for the Solaris
11677 @opindex mcall-linux
11678 On System V.4 and embedded PowerPC systems compile code for the
11679 Linux-based GNU system.
11683 On System V.4 and embedded PowerPC systems compile code for the
11684 Hurd-based GNU system.
11686 @item -mcall-netbsd
11687 @opindex mcall-netbsd
11688 On System V.4 and embedded PowerPC systems compile code for the
11689 NetBSD operating system.
11691 @item -maix-struct-return
11692 @opindex maix-struct-return
11693 Return all structures in memory (as specified by the AIX ABI)@.
11695 @item -msvr4-struct-return
11696 @opindex msvr4-struct-return
11697 Return structures smaller than 8 bytes in registers (as specified by the
11700 @item -mabi=@var{abi-type}
11702 Extend the current ABI with a particular extension, or remove such extension.
11703 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
11704 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
11708 Extend the current ABI with SPE ABI extensions. This does not change
11709 the default ABI, instead it adds the SPE ABI extensions to the current
11713 @opindex mabi=no-spe
11714 Disable Booke SPE ABI extensions for the current ABI@.
11716 @item -mabi=ibmlongdouble
11717 @opindex mabi=ibmlongdouble
11718 Change the current ABI to use IBM extended precision long double.
11719 This is a PowerPC 32-bit SYSV ABI option.
11721 @item -mabi=ieeelongdouble
11722 @opindex mabi=ieeelongdouble
11723 Change the current ABI to use IEEE extended precision long double.
11724 This is a PowerPC 32-bit Linux ABI option.
11727 @itemx -mno-prototype
11728 @opindex mprototype
11729 @opindex mno-prototype
11730 On System V.4 and embedded PowerPC systems assume that all calls to
11731 variable argument functions are properly prototyped. Otherwise, the
11732 compiler must insert an instruction before every non prototyped call to
11733 set or clear bit 6 of the condition code register (@var{CR}) to
11734 indicate whether floating point values were passed in the floating point
11735 registers in case the function takes a variable arguments. With
11736 @option{-mprototype}, only calls to prototyped variable argument functions
11737 will set or clear the bit.
11741 On embedded PowerPC systems, assume that the startup module is called
11742 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
11743 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
11748 On embedded PowerPC systems, assume that the startup module is called
11749 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
11754 On embedded PowerPC systems, assume that the startup module is called
11755 @file{crt0.o} and the standard C libraries are @file{libads.a} and
11758 @item -myellowknife
11759 @opindex myellowknife
11760 On embedded PowerPC systems, assume that the startup module is called
11761 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
11766 On System V.4 and embedded PowerPC systems, specify that you are
11767 compiling for a VxWorks system.
11771 Specify that you are compiling for the WindISS simulation environment.
11775 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
11776 header to indicate that @samp{eabi} extended relocations are used.
11782 On System V.4 and embedded PowerPC systems do (do not) adhere to the
11783 Embedded Applications Binary Interface (eabi) which is a set of
11784 modifications to the System V.4 specifications. Selecting @option{-meabi}
11785 means that the stack is aligned to an 8 byte boundary, a function
11786 @code{__eabi} is called to from @code{main} to set up the eabi
11787 environment, and the @option{-msdata} option can use both @code{r2} and
11788 @code{r13} to point to two separate small data areas. Selecting
11789 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
11790 do not call an initialization function from @code{main}, and the
11791 @option{-msdata} option will only use @code{r13} to point to a single
11792 small data area. The @option{-meabi} option is on by default if you
11793 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
11796 @opindex msdata=eabi
11797 On System V.4 and embedded PowerPC systems, put small initialized
11798 @code{const} global and static data in the @samp{.sdata2} section, which
11799 is pointed to by register @code{r2}. Put small initialized
11800 non-@code{const} global and static data in the @samp{.sdata} section,
11801 which is pointed to by register @code{r13}. Put small uninitialized
11802 global and static data in the @samp{.sbss} section, which is adjacent to
11803 the @samp{.sdata} section. The @option{-msdata=eabi} option is
11804 incompatible with the @option{-mrelocatable} option. The
11805 @option{-msdata=eabi} option also sets the @option{-memb} option.
11808 @opindex msdata=sysv
11809 On System V.4 and embedded PowerPC systems, put small global and static
11810 data in the @samp{.sdata} section, which is pointed to by register
11811 @code{r13}. Put small uninitialized global and static data in the
11812 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
11813 The @option{-msdata=sysv} option is incompatible with the
11814 @option{-mrelocatable} option.
11816 @item -msdata=default
11818 @opindex msdata=default
11820 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
11821 compile code the same as @option{-msdata=eabi}, otherwise compile code the
11822 same as @option{-msdata=sysv}.
11825 @opindex msdata-data
11826 On System V.4 and embedded PowerPC systems, put small global
11827 data in the @samp{.sdata} section. Put small uninitialized global
11828 data in the @samp{.sbss} section. Do not use register @code{r13}
11829 to address small data however. This is the default behavior unless
11830 other @option{-msdata} options are used.
11834 @opindex msdata=none
11836 On embedded PowerPC systems, put all initialized global and static data
11837 in the @samp{.data} section, and all uninitialized data in the
11838 @samp{.bss} section.
11842 @cindex smaller data references (PowerPC)
11843 @cindex .sdata/.sdata2 references (PowerPC)
11844 On embedded PowerPC systems, put global and static items less than or
11845 equal to @var{num} bytes into the small data or bss sections instead of
11846 the normal data or bss section. By default, @var{num} is 8. The
11847 @option{-G @var{num}} switch is also passed to the linker.
11848 All modules should be compiled with the same @option{-G @var{num}} value.
11851 @itemx -mno-regnames
11853 @opindex mno-regnames
11854 On System V.4 and embedded PowerPC systems do (do not) emit register
11855 names in the assembly language output using symbolic forms.
11858 @itemx -mno-longcall
11860 @opindex mno-longcall
11861 Default to making all function calls indirectly, using a register, so
11862 that functions which reside further than 32 megabytes (33,554,432
11863 bytes) from the current location can be called. This setting can be
11864 overridden by the @code{shortcall} function attribute, or by
11865 @code{#pragma longcall(0)}.
11867 Some linkers are capable of detecting out-of-range calls and generating
11868 glue code on the fly. On these systems, long calls are unnecessary and
11869 generate slower code. As of this writing, the AIX linker can do this,
11870 as can the GNU linker for PowerPC/64. It is planned to add this feature
11871 to the GNU linker for 32-bit PowerPC systems as well.
11873 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
11874 callee, L42'', plus a ``branch island'' (glue code). The two target
11875 addresses represent the callee and the ``branch island''. The
11876 Darwin/PPC linker will prefer the first address and generate a ``bl
11877 callee'' if the PPC ``bl'' instruction will reach the callee directly;
11878 otherwise, the linker will generate ``bl L42'' to call the ``branch
11879 island''. The ``branch island'' is appended to the body of the
11880 calling function; it computes the full 32-bit address of the callee
11883 On Mach-O (Darwin) systems, this option directs the compiler emit to
11884 the glue for every direct call, and the Darwin linker decides whether
11885 to use or discard it.
11887 In the future, we may cause GCC to ignore all longcall specifications
11888 when the linker is known to generate glue.
11892 Adds support for multithreading with the @dfn{pthreads} library.
11893 This option sets flags for both the preprocessor and linker.
11897 @node S/390 and zSeries Options
11898 @subsection S/390 and zSeries Options
11899 @cindex S/390 and zSeries Options
11901 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
11905 @itemx -msoft-float
11906 @opindex mhard-float
11907 @opindex msoft-float
11908 Use (do not use) the hardware floating-point instructions and registers
11909 for floating-point operations. When @option{-msoft-float} is specified,
11910 functions in @file{libgcc.a} will be used to perform floating-point
11911 operations. When @option{-mhard-float} is specified, the compiler
11912 generates IEEE floating-point instructions. This is the default.
11914 @item -mlong-double-64
11915 @itemx -mlong-double-128
11916 @opindex mlong-double-64
11917 @opindex mlong-double-128
11918 These switches control the size of @code{long double} type. A size
11919 of 64bit makes the @code{long double} type equivalent to the @code{double}
11920 type. This is the default.
11923 @itemx -mno-backchain
11924 @opindex mbackchain
11925 @opindex mno-backchain
11926 Store (do not store) the address of the caller's frame as backchain pointer
11927 into the callee's stack frame.
11928 A backchain may be needed to allow debugging using tools that do not understand
11929 DWARF-2 call frame information.
11930 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
11931 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
11932 the backchain is placed into the topmost word of the 96/160 byte register
11935 In general, code compiled with @option{-mbackchain} is call-compatible with
11936 code compiled with @option{-mmo-backchain}; however, use of the backchain
11937 for debugging purposes usually requires that the whole binary is built with
11938 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
11939 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
11940 to build a linux kernel use @option{-msoft-float}.
11942 The default is to not maintain the backchain.
11944 @item -mpacked-stack
11945 @item -mno-packed-stack
11946 @opindex mpacked-stack
11947 @opindex mno-packed-stack
11948 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
11949 specified, the compiler uses the all fields of the 96/160 byte register save
11950 area only for their default purpose; unused fields still take up stack space.
11951 When @option{-mpacked-stack} is specified, register save slots are densely
11952 packed at the top of the register save area; unused space is reused for other
11953 purposes, allowing for more efficient use of the available stack space.
11954 However, when @option{-mbackchain} is also in effect, the topmost word of
11955 the save area is always used to store the backchain, and the return address
11956 register is always saved two words below the backchain.
11958 As long as the stack frame backchain is not used, code generated with
11959 @option{-mpacked-stack} is call-compatible with code generated with
11960 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
11961 S/390 or zSeries generated code that uses the stack frame backchain at run
11962 time, not just for debugging purposes. Such code is not call-compatible
11963 with code compiled with @option{-mpacked-stack}. Also, note that the
11964 combination of @option{-mbackchain},
11965 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
11966 to build a linux kernel use @option{-msoft-float}.
11968 The default is to not use the packed stack layout.
11971 @itemx -mno-small-exec
11972 @opindex msmall-exec
11973 @opindex mno-small-exec
11974 Generate (or do not generate) code using the @code{bras} instruction
11975 to do subroutine calls.
11976 This only works reliably if the total executable size does not
11977 exceed 64k. The default is to use the @code{basr} instruction instead,
11978 which does not have this limitation.
11984 When @option{-m31} is specified, generate code compliant to the
11985 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
11986 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
11987 particular to generate 64-bit instructions. For the @samp{s390}
11988 targets, the default is @option{-m31}, while the @samp{s390x}
11989 targets default to @option{-m64}.
11995 When @option{-mzarch} is specified, generate code using the
11996 instructions available on z/Architecture.
11997 When @option{-mesa} is specified, generate code using the
11998 instructions available on ESA/390. Note that @option{-mesa} is
11999 not possible with @option{-m64}.
12000 When generating code compliant to the GNU/Linux for S/390 ABI,
12001 the default is @option{-mesa}. When generating code compliant
12002 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
12008 Generate (or do not generate) code using the @code{mvcle} instruction
12009 to perform block moves. When @option{-mno-mvcle} is specified,
12010 use a @code{mvc} loop instead. This is the default unless optimizing for
12017 Print (or do not print) additional debug information when compiling.
12018 The default is to not print debug information.
12020 @item -march=@var{cpu-type}
12022 Generate code that will run on @var{cpu-type}, which is the name of a system
12023 representing a certain processor type. Possible values for
12024 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
12025 When generating code using the instructions available on z/Architecture,
12026 the default is @option{-march=z900}. Otherwise, the default is
12027 @option{-march=g5}.
12029 @item -mtune=@var{cpu-type}
12031 Tune to @var{cpu-type} everything applicable about the generated code,
12032 except for the ABI and the set of available instructions.
12033 The list of @var{cpu-type} values is the same as for @option{-march}.
12034 The default is the value used for @option{-march}.
12037 @itemx -mno-tpf-trace
12038 @opindex mtpf-trace
12039 @opindex mno-tpf-trace
12040 Generate code that adds (does not add) in TPF OS specific branches to trace
12041 routines in the operating system. This option is off by default, even
12042 when compiling for the TPF OS@.
12045 @itemx -mno-fused-madd
12046 @opindex mfused-madd
12047 @opindex mno-fused-madd
12048 Generate code that uses (does not use) the floating point multiply and
12049 accumulate instructions. These instructions are generated by default if
12050 hardware floating point is used.
12052 @item -mwarn-framesize=@var{framesize}
12053 @opindex mwarn-framesize
12054 Emit a warning if the current function exceeds the given frame size. Because
12055 this is a compile time check it doesn't need to be a real problem when the program
12056 runs. It is intended to identify functions which most probably cause
12057 a stack overflow. It is useful to be used in an environment with limited stack
12058 size e.g.@: the linux kernel.
12060 @item -mwarn-dynamicstack
12061 @opindex mwarn-dynamicstack
12062 Emit a warning if the function calls alloca or uses dynamically
12063 sized arrays. This is generally a bad idea with a limited stack size.
12065 @item -mstack-guard=@var{stack-guard}
12066 @item -mstack-size=@var{stack-size}
12067 @opindex mstack-guard
12068 @opindex mstack-size
12069 These arguments always have to be used in conjunction. If they are present the s390
12070 back end emits additional instructions in the function prologue which trigger a trap
12071 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
12072 (remember that the stack on s390 grows downward). These options are intended to
12073 be used to help debugging stack overflow problems. The additionally emitted code
12074 causes only little overhead and hence can also be used in production like systems
12075 without greater performance degradation. The given values have to be exact
12076 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard} without
12078 In order to be efficient the extra code makes the assumption that the stack starts
12079 at an address aligned to the value given by @var{stack-size}.
12083 @subsection SH Options
12085 These @samp{-m} options are defined for the SH implementations:
12090 Generate code for the SH1.
12094 Generate code for the SH2.
12097 Generate code for the SH2e.
12101 Generate code for the SH3.
12105 Generate code for the SH3e.
12109 Generate code for the SH4 without a floating-point unit.
12111 @item -m4-single-only
12112 @opindex m4-single-only
12113 Generate code for the SH4 with a floating-point unit that only
12114 supports single-precision arithmetic.
12118 Generate code for the SH4 assuming the floating-point unit is in
12119 single-precision mode by default.
12123 Generate code for the SH4.
12127 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
12128 floating-point unit is not used.
12130 @item -m4a-single-only
12131 @opindex m4a-single-only
12132 Generate code for the SH4a, in such a way that no double-precision
12133 floating point operations are used.
12136 @opindex m4a-single
12137 Generate code for the SH4a assuming the floating-point unit is in
12138 single-precision mode by default.
12142 Generate code for the SH4a.
12146 Same as @option{-m4a-nofpu}, except that it implicitly passes
12147 @option{-dsp} to the assembler. GCC doesn't generate any DSP
12148 instructions at the moment.
12152 Compile code for the processor in big endian mode.
12156 Compile code for the processor in little endian mode.
12160 Align doubles at 64-bit boundaries. Note that this changes the calling
12161 conventions, and thus some functions from the standard C library will
12162 not work unless you recompile it first with @option{-mdalign}.
12166 Shorten some address references at link time, when possible; uses the
12167 linker option @option{-relax}.
12171 Use 32-bit offsets in @code{switch} tables. The default is to use
12176 Enable the use of the instruction @code{fmovd}.
12180 Comply with the calling conventions defined by Renesas.
12184 Comply with the calling conventions defined by Renesas.
12188 Comply with the calling conventions defined for GCC before the Renesas
12189 conventions were available. This option is the default for all
12190 targets of the SH toolchain except for @samp{sh-symbianelf}.
12193 @opindex mnomacsave
12194 Mark the @code{MAC} register as call-clobbered, even if
12195 @option{-mhitachi} is given.
12199 Increase IEEE-compliance of floating-point code.
12200 At the moment, this is equivalent to @option{-fno-finite-math-only}.
12201 When generating 16 bit SH opcodes, getting IEEE-conforming results for
12202 comparisons of NANs / infinities incurs extra overhead in every
12203 floating point comparison, therefore the default is set to
12204 @option{-ffinite-math-only}.
12208 Dump instruction size and location in the assembly code.
12211 @opindex mpadstruct
12212 This option is deprecated. It pads structures to multiple of 4 bytes,
12213 which is incompatible with the SH ABI@.
12217 Optimize for space instead of speed. Implied by @option{-Os}.
12220 @opindex mprefergot
12221 When generating position-independent code, emit function calls using
12222 the Global Offset Table instead of the Procedure Linkage Table.
12226 Generate a library function call to invalidate instruction cache
12227 entries, after fixing up a trampoline. This library function call
12228 doesn't assume it can write to the whole memory address space. This
12229 is the default when the target is @code{sh-*-linux*}.
12231 @item -multcost=@var{number}
12232 @opindex multcost=@var{number}
12233 Set the cost to assume for a multiply insn.
12235 @item -mdiv=@var{strategy}
12236 @opindex mdiv=@var{strategy}
12237 Set the division strategy to use for SHmedia code. @var{strategy} must be
12238 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
12239 inv:call2, inv:fp .
12240 "fp" performs the operation in floating point. This has a very high latency,
12241 but needs only a few instructions, so it might be a good choice if
12242 your code has enough easily exploitable ILP to allow the compiler to
12243 schedule the floating point instructions together with other instructions.
12244 Division by zero causes a floating point exception.
12245 "inv" uses integer operations to calculate the inverse of the divisor,
12246 and then multiplies the dividend with the inverse. This strategy allows
12247 cse and hoisting of the inverse calculation. Division by zero calculates
12248 an unspecified result, but does not trap.
12249 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
12250 have been found, or if the entire operation has been hoisted to the same
12251 place, the last stages of the inverse calculation are intertwined with the
12252 final multiply to reduce the overall latency, at the expense of using a few
12253 more instructions, and thus offering fewer scheduling opportunities with
12255 "call" calls a library function that usually implements the inv:minlat
12257 This gives high code density for m5-*media-nofpu compilations.
12258 "call2" uses a different entry point of the same library function, where it
12259 assumes that a pointer to a lookup table has already been set up, which
12260 exposes the pointer load to cse / code hoisting optimizations.
12261 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
12262 code generation, but if the code stays unoptimized, revert to the "call",
12263 "call2", or "fp" strategies, respectively. Note that the
12264 potentially-trapping side effect of division by zero is carried by a
12265 separate instruction, so it is possible that all the integer instructions
12266 are hoisted out, but the marker for the side effect stays where it is.
12267 A recombination to fp operations or a call is not possible in that case.
12268 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
12269 that the inverse calculation was nor separated from the multiply, they speed
12270 up division where the dividend fits into 20 bits (plus sign where applicable),
12271 by inserting a test to skip a number of operations in this case; this test
12272 slows down the case of larger dividends. inv20u assumes the case of a such
12273 a small dividend to be unlikely, and inv20l assumes it to be likely.
12275 @item -mdivsi3_libfunc=@var{name}
12276 @opindex mdivsi3_libfunc=@var{name}
12277 Set the name of the library function used for 32 bit signed division to
12278 @var{name}. This only affect the name used in the call and inv:call
12279 division strategies, and the compiler will still expect the same
12280 sets of input/output/clobbered registers as if this option was not present.
12282 @item -madjust-unroll
12283 @opindex madjust-unroll
12284 Throttle unrolling to avoid thrashing target registers.
12285 This option only has an effect if the gcc code base supports the
12286 TARGET_ADJUST_UNROLL_MAX target hook.
12288 @item -mindexed-addressing
12289 @opindex mindexed-addressing
12290 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
12291 This is only safe if the hardware and/or OS implement 32 bit wrap-around
12292 semantics for the indexed addressing mode. The architecture allows the
12293 implementation of processors with 64 bit MMU, which the OS could use to
12294 get 32 bit addressing, but since no current hardware implementation supports
12295 this or any other way to make the indexed addressing mode safe to use in
12296 the 32 bit ABI, the default is -mno-indexed-addressing.
12298 @item -mgettrcost=@var{number}
12299 @opindex mgettrcost=@var{number}
12300 Set the cost assumed for the gettr instruction to @var{number}.
12301 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
12305 Assume pt* instructions won't trap. This will generally generate better
12306 scheduled code, but is unsafe on current hardware. The current architecture
12307 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
12308 This has the unintentional effect of making it unsafe to schedule ptabs /
12309 ptrel before a branch, or hoist it out of a loop. For example,
12310 __do_global_ctors, a part of libgcc that runs constructors at program
12311 startup, calls functions in a list which is delimited by -1. With the
12312 -mpt-fixed option, the ptabs will be done before testing against -1.
12313 That means that all the constructors will be run a bit quicker, but when
12314 the loop comes to the end of the list, the program crashes because ptabs
12315 loads -1 into a target register. Since this option is unsafe for any
12316 hardware implementing the current architecture specification, the default
12317 is -mno-pt-fixed. Unless the user specifies a specific cost with
12318 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
12319 this deters register allocation using target registers for storing
12322 @item -minvalid-symbols
12323 @opindex minvalid-symbols
12324 Assume symbols might be invalid. Ordinary function symbols generated by
12325 the compiler will always be valid to load with movi/shori/ptabs or
12326 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
12327 to generate symbols that will cause ptabs / ptrel to trap.
12328 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
12329 It will then prevent cross-basic-block cse, hoisting and most scheduling
12330 of symbol loads. The default is @option{-mno-invalid-symbols}.
12333 @node SPARC Options
12334 @subsection SPARC Options
12335 @cindex SPARC options
12337 These @samp{-m} options are supported on the SPARC:
12340 @item -mno-app-regs
12342 @opindex mno-app-regs
12344 Specify @option{-mapp-regs} to generate output using the global registers
12345 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
12348 To be fully SVR4 ABI compliant at the cost of some performance loss,
12349 specify @option{-mno-app-regs}. You should compile libraries and system
12350 software with this option.
12353 @itemx -mhard-float
12355 @opindex mhard-float
12356 Generate output containing floating point instructions. This is the
12360 @itemx -msoft-float
12362 @opindex msoft-float
12363 Generate output containing library calls for floating point.
12364 @strong{Warning:} the requisite libraries are not available for all SPARC
12365 targets. Normally the facilities of the machine's usual C compiler are
12366 used, but this cannot be done directly in cross-compilation. You must make
12367 your own arrangements to provide suitable library functions for
12368 cross-compilation. The embedded targets @samp{sparc-*-aout} and
12369 @samp{sparclite-*-*} do provide software floating point support.
12371 @option{-msoft-float} changes the calling convention in the output file;
12372 therefore, it is only useful if you compile @emph{all} of a program with
12373 this option. In particular, you need to compile @file{libgcc.a}, the
12374 library that comes with GCC, with @option{-msoft-float} in order for
12377 @item -mhard-quad-float
12378 @opindex mhard-quad-float
12379 Generate output containing quad-word (long double) floating point
12382 @item -msoft-quad-float
12383 @opindex msoft-quad-float
12384 Generate output containing library calls for quad-word (long double)
12385 floating point instructions. The functions called are those specified
12386 in the SPARC ABI@. This is the default.
12388 As of this writing, there are no SPARC implementations that have hardware
12389 support for the quad-word floating point instructions. They all invoke
12390 a trap handler for one of these instructions, and then the trap handler
12391 emulates the effect of the instruction. Because of the trap handler overhead,
12392 this is much slower than calling the ABI library routines. Thus the
12393 @option{-msoft-quad-float} option is the default.
12395 @item -mno-unaligned-doubles
12396 @itemx -munaligned-doubles
12397 @opindex mno-unaligned-doubles
12398 @opindex munaligned-doubles
12399 Assume that doubles have 8 byte alignment. This is the default.
12401 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
12402 alignment only if they are contained in another type, or if they have an
12403 absolute address. Otherwise, it assumes they have 4 byte alignment.
12404 Specifying this option avoids some rare compatibility problems with code
12405 generated by other compilers. It is not the default because it results
12406 in a performance loss, especially for floating point code.
12408 @item -mno-faster-structs
12409 @itemx -mfaster-structs
12410 @opindex mno-faster-structs
12411 @opindex mfaster-structs
12412 With @option{-mfaster-structs}, the compiler assumes that structures
12413 should have 8 byte alignment. This enables the use of pairs of
12414 @code{ldd} and @code{std} instructions for copies in structure
12415 assignment, in place of twice as many @code{ld} and @code{st} pairs.
12416 However, the use of this changed alignment directly violates the SPARC
12417 ABI@. Thus, it's intended only for use on targets where the developer
12418 acknowledges that their resulting code will not be directly in line with
12419 the rules of the ABI@.
12421 @item -mimpure-text
12422 @opindex mimpure-text
12423 @option{-mimpure-text}, used in addition to @option{-shared}, tells
12424 the compiler to not pass @option{-z text} to the linker when linking a
12425 shared object. Using this option, you can link position-dependent
12426 code into a shared object.
12428 @option{-mimpure-text} suppresses the ``relocations remain against
12429 allocatable but non-writable sections'' linker error message.
12430 However, the necessary relocations will trigger copy-on-write, and the
12431 shared object is not actually shared across processes. Instead of
12432 using @option{-mimpure-text}, you should compile all source code with
12433 @option{-fpic} or @option{-fPIC}.
12435 This option is only available on SunOS and Solaris.
12437 @item -mcpu=@var{cpu_type}
12439 Set the instruction set, register set, and instruction scheduling parameters
12440 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
12441 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
12442 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
12443 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
12444 @samp{ultrasparc3}, and @samp{niagara}.
12446 Default instruction scheduling parameters are used for values that select
12447 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
12448 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
12450 Here is a list of each supported architecture and their supported
12455 v8: supersparc, hypersparc
12456 sparclite: f930, f934, sparclite86x
12458 v9: ultrasparc, ultrasparc3, niagara
12461 By default (unless configured otherwise), GCC generates code for the V7
12462 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
12463 additionally optimizes it for the Cypress CY7C602 chip, as used in the
12464 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
12465 SPARCStation 1, 2, IPX etc.
12467 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
12468 architecture. The only difference from V7 code is that the compiler emits
12469 the integer multiply and integer divide instructions which exist in SPARC-V8
12470 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
12471 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
12474 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
12475 the SPARC architecture. This adds the integer multiply, integer divide step
12476 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
12477 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
12478 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
12479 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
12480 MB86934 chip, which is the more recent SPARClite with FPU@.
12482 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
12483 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
12484 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
12485 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
12486 optimizes it for the TEMIC SPARClet chip.
12488 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
12489 architecture. This adds 64-bit integer and floating-point move instructions,
12490 3 additional floating-point condition code registers and conditional move
12491 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
12492 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
12493 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
12494 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
12495 @option{-mcpu=niagara}, the compiler additionally optimizes it for
12496 Sun UltraSPARC T1 chips.
12498 @item -mtune=@var{cpu_type}
12500 Set the instruction scheduling parameters for machine type
12501 @var{cpu_type}, but do not set the instruction set or register set that the
12502 option @option{-mcpu=@var{cpu_type}} would.
12504 The same values for @option{-mcpu=@var{cpu_type}} can be used for
12505 @option{-mtune=@var{cpu_type}}, but the only useful values are those
12506 that select a particular cpu implementation. Those are @samp{cypress},
12507 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
12508 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
12509 @samp{ultrasparc3}, and @samp{niagara}.
12514 @opindex mno-v8plus
12515 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
12516 difference from the V8 ABI is that the global and out registers are
12517 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
12518 mode for all SPARC-V9 processors.
12524 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
12525 Visual Instruction Set extensions. The default is @option{-mno-vis}.
12528 These @samp{-m} options are supported in addition to the above
12529 on SPARC-V9 processors in 64-bit environments:
12532 @item -mlittle-endian
12533 @opindex mlittle-endian
12534 Generate code for a processor running in little-endian mode. It is only
12535 available for a few configurations and most notably not on Solaris and Linux.
12541 Generate code for a 32-bit or 64-bit environment.
12542 The 32-bit environment sets int, long and pointer to 32 bits.
12543 The 64-bit environment sets int to 32 bits and long and pointer
12546 @item -mcmodel=medlow
12547 @opindex mcmodel=medlow
12548 Generate code for the Medium/Low code model: 64-bit addresses, programs
12549 must be linked in the low 32 bits of memory. Programs can be statically
12550 or dynamically linked.
12552 @item -mcmodel=medmid
12553 @opindex mcmodel=medmid
12554 Generate code for the Medium/Middle code model: 64-bit addresses, programs
12555 must be linked in the low 44 bits of memory, the text and data segments must
12556 be less than 2GB in size and the data segment must be located within 2GB of
12559 @item -mcmodel=medany
12560 @opindex mcmodel=medany
12561 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
12562 may be linked anywhere in memory, the text and data segments must be less
12563 than 2GB in size and the data segment must be located within 2GB of the
12566 @item -mcmodel=embmedany
12567 @opindex mcmodel=embmedany
12568 Generate code for the Medium/Anywhere code model for embedded systems:
12569 64-bit addresses, the text and data segments must be less than 2GB in
12570 size, both starting anywhere in memory (determined at link time). The
12571 global register %g4 points to the base of the data segment. Programs
12572 are statically linked and PIC is not supported.
12575 @itemx -mno-stack-bias
12576 @opindex mstack-bias
12577 @opindex mno-stack-bias
12578 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
12579 frame pointer if present, are offset by @minus{}2047 which must be added back
12580 when making stack frame references. This is the default in 64-bit mode.
12581 Otherwise, assume no such offset is present.
12584 These switches are supported in addition to the above on Solaris:
12589 Add support for multithreading using the Solaris threads library. This
12590 option sets flags for both the preprocessor and linker. This option does
12591 not affect the thread safety of object code produced by the compiler or
12592 that of libraries supplied with it.
12596 Add support for multithreading using the POSIX threads library. This
12597 option sets flags for both the preprocessor and linker. This option does
12598 not affect the thread safety of object code produced by the compiler or
12599 that of libraries supplied with it.
12603 This is a synonym for @option{-pthreads}.
12606 @node System V Options
12607 @subsection Options for System V
12609 These additional options are available on System V Release 4 for
12610 compatibility with other compilers on those systems:
12615 Create a shared object.
12616 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
12620 Identify the versions of each tool used by the compiler, in a
12621 @code{.ident} assembler directive in the output.
12625 Refrain from adding @code{.ident} directives to the output file (this is
12628 @item -YP,@var{dirs}
12630 Search the directories @var{dirs}, and no others, for libraries
12631 specified with @option{-l}.
12633 @item -Ym,@var{dir}
12635 Look in the directory @var{dir} to find the M4 preprocessor.
12636 The assembler uses this option.
12637 @c This is supposed to go with a -Yd for predefined M4 macro files, but
12638 @c the generic assembler that comes with Solaris takes just -Ym.
12641 @node TMS320C3x/C4x Options
12642 @subsection TMS320C3x/C4x Options
12643 @cindex TMS320C3x/C4x Options
12645 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
12649 @item -mcpu=@var{cpu_type}
12651 Set the instruction set, register set, and instruction scheduling
12652 parameters for machine type @var{cpu_type}. Supported values for
12653 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
12654 @samp{c44}. The default is @samp{c40} to generate code for the
12659 @itemx -msmall-memory
12661 @opindex mbig-memory
12663 @opindex msmall-memory
12665 Generates code for the big or small memory model. The small memory
12666 model assumed that all data fits into one 64K word page. At run-time
12667 the data page (DP) register must be set to point to the 64K page
12668 containing the .bss and .data program sections. The big memory model is
12669 the default and requires reloading of the DP register for every direct
12676 Allow (disallow) allocation of general integer operands into the block
12677 count register BK@.
12683 Enable (disable) generation of code using decrement and branch,
12684 DBcond(D), instructions. This is enabled by default for the C4x. To be
12685 on the safe side, this is disabled for the C3x, since the maximum
12686 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
12687 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
12688 that it can utilize the decrement and branch instruction, but will give
12689 up if there is more than one memory reference in the loop. Thus a loop
12690 where the loop counter is decremented can generate slightly more
12691 efficient code, in cases where the RPTB instruction cannot be utilized.
12693 @item -mdp-isr-reload
12695 @opindex mdp-isr-reload
12697 Force the DP register to be saved on entry to an interrupt service
12698 routine (ISR), reloaded to point to the data section, and restored on
12699 exit from the ISR@. This should not be required unless someone has
12700 violated the small memory model by modifying the DP register, say within
12707 For the C3x use the 24-bit MPYI instruction for integer multiplies
12708 instead of a library call to guarantee 32-bit results. Note that if one
12709 of the operands is a constant, then the multiplication will be performed
12710 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
12711 then squaring operations are performed inline instead of a library call.
12714 @itemx -mno-fast-fix
12716 @opindex mno-fast-fix
12717 The C3x/C4x FIX instruction to convert a floating point value to an
12718 integer value chooses the nearest integer less than or equal to the
12719 floating point value rather than to the nearest integer. Thus if the
12720 floating point number is negative, the result will be incorrectly
12721 truncated an additional code is necessary to detect and correct this
12722 case. This option can be used to disable generation of the additional
12723 code required to correct the result.
12729 Enable (disable) generation of repeat block sequences using the RPTB
12730 instruction for zero overhead looping. The RPTB construct is only used
12731 for innermost loops that do not call functions or jump across the loop
12732 boundaries. There is no advantage having nested RPTB loops due to the
12733 overhead required to save and restore the RC, RS, and RE registers.
12734 This is enabled by default with @option{-O2}.
12736 @item -mrpts=@var{count}
12740 Enable (disable) the use of the single instruction repeat instruction
12741 RPTS@. If a repeat block contains a single instruction, and the loop
12742 count can be guaranteed to be less than the value @var{count}, GCC will
12743 emit a RPTS instruction instead of a RPTB@. If no value is specified,
12744 then a RPTS will be emitted even if the loop count cannot be determined
12745 at compile time. Note that the repeated instruction following RPTS does
12746 not have to be reloaded from memory each iteration, thus freeing up the
12747 CPU buses for operands. However, since interrupts are blocked by this
12748 instruction, it is disabled by default.
12750 @item -mloop-unsigned
12751 @itemx -mno-loop-unsigned
12752 @opindex mloop-unsigned
12753 @opindex mno-loop-unsigned
12754 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
12755 is @math{2^{31} + 1} since these instructions test if the iteration count is
12756 negative to terminate the loop. If the iteration count is unsigned
12757 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
12758 exceeded. This switch allows an unsigned iteration count.
12762 Try to emit an assembler syntax that the TI assembler (asm30) is happy
12763 with. This also enforces compatibility with the API employed by the TI
12764 C3x C compiler. For example, long doubles are passed as structures
12765 rather than in floating point registers.
12771 Generate code that uses registers (stack) for passing arguments to functions.
12772 By default, arguments are passed in registers where possible rather
12773 than by pushing arguments on to the stack.
12775 @item -mparallel-insns
12776 @itemx -mno-parallel-insns
12777 @opindex mparallel-insns
12778 @opindex mno-parallel-insns
12779 Allow the generation of parallel instructions. This is enabled by
12780 default with @option{-O2}.
12782 @item -mparallel-mpy
12783 @itemx -mno-parallel-mpy
12784 @opindex mparallel-mpy
12785 @opindex mno-parallel-mpy
12786 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
12787 provided @option{-mparallel-insns} is also specified. These instructions have
12788 tight register constraints which can pessimize the code generation
12789 of large functions.
12794 @subsection V850 Options
12795 @cindex V850 Options
12797 These @samp{-m} options are defined for V850 implementations:
12801 @itemx -mno-long-calls
12802 @opindex mlong-calls
12803 @opindex mno-long-calls
12804 Treat all calls as being far away (near). If calls are assumed to be
12805 far away, the compiler will always load the functions address up into a
12806 register, and call indirect through the pointer.
12812 Do not optimize (do optimize) basic blocks that use the same index
12813 pointer 4 or more times to copy pointer into the @code{ep} register, and
12814 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
12815 option is on by default if you optimize.
12817 @item -mno-prolog-function
12818 @itemx -mprolog-function
12819 @opindex mno-prolog-function
12820 @opindex mprolog-function
12821 Do not use (do use) external functions to save and restore registers
12822 at the prologue and epilogue of a function. The external functions
12823 are slower, but use less code space if more than one function saves
12824 the same number of registers. The @option{-mprolog-function} option
12825 is on by default if you optimize.
12829 Try to make the code as small as possible. At present, this just turns
12830 on the @option{-mep} and @option{-mprolog-function} options.
12832 @item -mtda=@var{n}
12834 Put static or global variables whose size is @var{n} bytes or less into
12835 the tiny data area that register @code{ep} points to. The tiny data
12836 area can hold up to 256 bytes in total (128 bytes for byte references).
12838 @item -msda=@var{n}
12840 Put static or global variables whose size is @var{n} bytes or less into
12841 the small data area that register @code{gp} points to. The small data
12842 area can hold up to 64 kilobytes.
12844 @item -mzda=@var{n}
12846 Put static or global variables whose size is @var{n} bytes or less into
12847 the first 32 kilobytes of memory.
12851 Specify that the target processor is the V850.
12854 @opindex mbig-switch
12855 Generate code suitable for big switch tables. Use this option only if
12856 the assembler/linker complain about out of range branches within a switch
12861 This option will cause r2 and r5 to be used in the code generated by
12862 the compiler. This setting is the default.
12864 @item -mno-app-regs
12865 @opindex mno-app-regs
12866 This option will cause r2 and r5 to be treated as fixed registers.
12870 Specify that the target processor is the V850E1. The preprocessor
12871 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
12872 this option is used.
12876 Specify that the target processor is the V850E@. The preprocessor
12877 constant @samp{__v850e__} will be defined if this option is used.
12879 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
12880 are defined then a default target processor will be chosen and the
12881 relevant @samp{__v850*__} preprocessor constant will be defined.
12883 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
12884 defined, regardless of which processor variant is the target.
12886 @item -mdisable-callt
12887 @opindex mdisable-callt
12888 This option will suppress generation of the CALLT instruction for the
12889 v850e and v850e1 flavors of the v850 architecture. The default is
12890 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
12895 @subsection VAX Options
12896 @cindex VAX options
12898 These @samp{-m} options are defined for the VAX:
12903 Do not output certain jump instructions (@code{aobleq} and so on)
12904 that the Unix assembler for the VAX cannot handle across long
12909 Do output those jump instructions, on the assumption that you
12910 will assemble with the GNU assembler.
12914 Output code for g-format floating point numbers instead of d-format.
12917 @node x86-64 Options
12918 @subsection x86-64 Options
12919 @cindex x86-64 options
12921 These are listed under @xref{i386 and x86-64 Options}.
12923 @node Xstormy16 Options
12924 @subsection Xstormy16 Options
12925 @cindex Xstormy16 Options
12927 These options are defined for Xstormy16:
12932 Choose startup files and linker script suitable for the simulator.
12935 @node Xtensa Options
12936 @subsection Xtensa Options
12937 @cindex Xtensa Options
12939 These options are supported for Xtensa targets:
12943 @itemx -mno-const16
12945 @opindex mno-const16
12946 Enable or disable use of @code{CONST16} instructions for loading
12947 constant values. The @code{CONST16} instruction is currently not a
12948 standard option from Tensilica. When enabled, @code{CONST16}
12949 instructions are always used in place of the standard @code{L32R}
12950 instructions. The use of @code{CONST16} is enabled by default only if
12951 the @code{L32R} instruction is not available.
12954 @itemx -mno-fused-madd
12955 @opindex mfused-madd
12956 @opindex mno-fused-madd
12957 Enable or disable use of fused multiply/add and multiply/subtract
12958 instructions in the floating-point option. This has no effect if the
12959 floating-point option is not also enabled. Disabling fused multiply/add
12960 and multiply/subtract instructions forces the compiler to use separate
12961 instructions for the multiply and add/subtract operations. This may be
12962 desirable in some cases where strict IEEE 754-compliant results are
12963 required: the fused multiply add/subtract instructions do not round the
12964 intermediate result, thereby producing results with @emph{more} bits of
12965 precision than specified by the IEEE standard. Disabling fused multiply
12966 add/subtract instructions also ensures that the program output is not
12967 sensitive to the compiler's ability to combine multiply and add/subtract
12970 @item -mtext-section-literals
12971 @itemx -mno-text-section-literals
12972 @opindex mtext-section-literals
12973 @opindex mno-text-section-literals
12974 Control the treatment of literal pools. The default is
12975 @option{-mno-text-section-literals}, which places literals in a separate
12976 section in the output file. This allows the literal pool to be placed
12977 in a data RAM/ROM, and it also allows the linker to combine literal
12978 pools from separate object files to remove redundant literals and
12979 improve code size. With @option{-mtext-section-literals}, the literals
12980 are interspersed in the text section in order to keep them as close as
12981 possible to their references. This may be necessary for large assembly
12984 @item -mtarget-align
12985 @itemx -mno-target-align
12986 @opindex mtarget-align
12987 @opindex mno-target-align
12988 When this option is enabled, GCC instructs the assembler to
12989 automatically align instructions to reduce branch penalties at the
12990 expense of some code density. The assembler attempts to widen density
12991 instructions to align branch targets and the instructions following call
12992 instructions. If there are not enough preceding safe density
12993 instructions to align a target, no widening will be performed. The
12994 default is @option{-mtarget-align}. These options do not affect the
12995 treatment of auto-aligned instructions like @code{LOOP}, which the
12996 assembler will always align, either by widening density instructions or
12997 by inserting no-op instructions.
13000 @itemx -mno-longcalls
13001 @opindex mlongcalls
13002 @opindex mno-longcalls
13003 When this option is enabled, GCC instructs the assembler to translate
13004 direct calls to indirect calls unless it can determine that the target
13005 of a direct call is in the range allowed by the call instruction. This
13006 translation typically occurs for calls to functions in other source
13007 files. Specifically, the assembler translates a direct @code{CALL}
13008 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
13009 The default is @option{-mno-longcalls}. This option should be used in
13010 programs where the call target can potentially be out of range. This
13011 option is implemented in the assembler, not the compiler, so the
13012 assembly code generated by GCC will still show direct call
13013 instructions---look at the disassembled object code to see the actual
13014 instructions. Note that the assembler will use an indirect call for
13015 every cross-file call, not just those that really will be out of range.
13018 @node zSeries Options
13019 @subsection zSeries Options
13020 @cindex zSeries options
13022 These are listed under @xref{S/390 and zSeries Options}.
13024 @node Code Gen Options
13025 @section Options for Code Generation Conventions
13026 @cindex code generation conventions
13027 @cindex options, code generation
13028 @cindex run-time options
13030 These machine-independent options control the interface conventions
13031 used in code generation.
13033 Most of them have both positive and negative forms; the negative form
13034 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
13035 one of the forms is listed---the one which is not the default. You
13036 can figure out the other form by either removing @samp{no-} or adding
13040 @item -fbounds-check
13041 @opindex fbounds-check
13042 For front-ends that support it, generate additional code to check that
13043 indices used to access arrays are within the declared range. This is
13044 currently only supported by the Java and Fortran 77 front-ends, where
13045 this option defaults to true and false respectively.
13049 This option generates traps for signed overflow on addition, subtraction,
13050 multiplication operations.
13054 This option instructs the compiler to assume that signed arithmetic
13055 overflow of addition, subtraction and multiplication wraps around
13056 using twos-complement representation. This flag enables some optimizations
13057 and disables others. This option is enabled by default for the Java
13058 front-end, as required by the Java language specification.
13061 @opindex fexceptions
13062 Enable exception handling. Generates extra code needed to propagate
13063 exceptions. For some targets, this implies GCC will generate frame
13064 unwind information for all functions, which can produce significant data
13065 size overhead, although it does not affect execution. If you do not
13066 specify this option, GCC will enable it by default for languages like
13067 C++ which normally require exception handling, and disable it for
13068 languages like C that do not normally require it. However, you may need
13069 to enable this option when compiling C code that needs to interoperate
13070 properly with exception handlers written in C++. You may also wish to
13071 disable this option if you are compiling older C++ programs that don't
13072 use exception handling.
13074 @item -fnon-call-exceptions
13075 @opindex fnon-call-exceptions
13076 Generate code that allows trapping instructions to throw exceptions.
13077 Note that this requires platform-specific runtime support that does
13078 not exist everywhere. Moreover, it only allows @emph{trapping}
13079 instructions to throw exceptions, i.e.@: memory references or floating
13080 point instructions. It does not allow exceptions to be thrown from
13081 arbitrary signal handlers such as @code{SIGALRM}.
13083 @item -funwind-tables
13084 @opindex funwind-tables
13085 Similar to @option{-fexceptions}, except that it will just generate any needed
13086 static data, but will not affect the generated code in any other way.
13087 You will normally not enable this option; instead, a language processor
13088 that needs this handling would enable it on your behalf.
13090 @item -fasynchronous-unwind-tables
13091 @opindex fasynchronous-unwind-tables
13092 Generate unwind table in dwarf2 format, if supported by target machine. The
13093 table is exact at each instruction boundary, so it can be used for stack
13094 unwinding from asynchronous events (such as debugger or garbage collector).
13096 @item -fpcc-struct-return
13097 @opindex fpcc-struct-return
13098 Return ``short'' @code{struct} and @code{union} values in memory like
13099 longer ones, rather than in registers. This convention is less
13100 efficient, but it has the advantage of allowing intercallability between
13101 GCC-compiled files and files compiled with other compilers, particularly
13102 the Portable C Compiler (pcc).
13104 The precise convention for returning structures in memory depends
13105 on the target configuration macros.
13107 Short structures and unions are those whose size and alignment match
13108 that of some integer type.
13110 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
13111 switch is not binary compatible with code compiled with the
13112 @option{-freg-struct-return} switch.
13113 Use it to conform to a non-default application binary interface.
13115 @item -freg-struct-return
13116 @opindex freg-struct-return
13117 Return @code{struct} and @code{union} values in registers when possible.
13118 This is more efficient for small structures than
13119 @option{-fpcc-struct-return}.
13121 If you specify neither @option{-fpcc-struct-return} nor
13122 @option{-freg-struct-return}, GCC defaults to whichever convention is
13123 standard for the target. If there is no standard convention, GCC
13124 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
13125 the principal compiler. In those cases, we can choose the standard, and
13126 we chose the more efficient register return alternative.
13128 @strong{Warning:} code compiled with the @option{-freg-struct-return}
13129 switch is not binary compatible with code compiled with the
13130 @option{-fpcc-struct-return} switch.
13131 Use it to conform to a non-default application binary interface.
13133 @item -fshort-enums
13134 @opindex fshort-enums
13135 Allocate to an @code{enum} type only as many bytes as it needs for the
13136 declared range of possible values. Specifically, the @code{enum} type
13137 will be equivalent to the smallest integer type which has enough room.
13139 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
13140 code that is not binary compatible with code generated without that switch.
13141 Use it to conform to a non-default application binary interface.
13143 @item -fshort-double
13144 @opindex fshort-double
13145 Use the same size for @code{double} as for @code{float}.
13147 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
13148 code that is not binary compatible with code generated without that switch.
13149 Use it to conform to a non-default application binary interface.
13151 @item -fshort-wchar
13152 @opindex fshort-wchar
13153 Override the underlying type for @samp{wchar_t} to be @samp{short
13154 unsigned int} instead of the default for the target. This option is
13155 useful for building programs to run under WINE@.
13157 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
13158 code that is not binary compatible with code generated without that switch.
13159 Use it to conform to a non-default application binary interface.
13162 @opindex fno-common
13163 In C, allocate even uninitialized global variables in the data section of the
13164 object file, rather than generating them as common blocks. This has the
13165 effect that if the same variable is declared (without @code{extern}) in
13166 two different compilations, you will get an error when you link them.
13167 The only reason this might be useful is if you wish to verify that the
13168 program will work on other systems which always work this way.
13172 Ignore the @samp{#ident} directive.
13174 @item -finhibit-size-directive
13175 @opindex finhibit-size-directive
13176 Don't output a @code{.size} assembler directive, or anything else that
13177 would cause trouble if the function is split in the middle, and the
13178 two halves are placed at locations far apart in memory. This option is
13179 used when compiling @file{crtstuff.c}; you should not need to use it
13182 @item -fverbose-asm
13183 @opindex fverbose-asm
13184 Put extra commentary information in the generated assembly code to
13185 make it more readable. This option is generally only of use to those
13186 who actually need to read the generated assembly code (perhaps while
13187 debugging the compiler itself).
13189 @option{-fno-verbose-asm}, the default, causes the
13190 extra information to be omitted and is useful when comparing two assembler
13195 @cindex global offset table
13197 Generate position-independent code (PIC) suitable for use in a shared
13198 library, if supported for the target machine. Such code accesses all
13199 constant addresses through a global offset table (GOT)@. The dynamic
13200 loader resolves the GOT entries when the program starts (the dynamic
13201 loader is not part of GCC; it is part of the operating system). If
13202 the GOT size for the linked executable exceeds a machine-specific
13203 maximum size, you get an error message from the linker indicating that
13204 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
13205 instead. (These maximums are 8k on the SPARC and 32k
13206 on the m68k and RS/6000. The 386 has no such limit.)
13208 Position-independent code requires special support, and therefore works
13209 only on certain machines. For the 386, GCC supports PIC for System V
13210 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
13211 position-independent.
13213 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13218 If supported for the target machine, emit position-independent code,
13219 suitable for dynamic linking and avoiding any limit on the size of the
13220 global offset table. This option makes a difference on the m68k,
13221 PowerPC and SPARC@.
13223 Position-independent code requires special support, and therefore works
13224 only on certain machines.
13226 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13233 These options are similar to @option{-fpic} and @option{-fPIC}, but
13234 generated position independent code can be only linked into executables.
13235 Usually these options are used when @option{-pie} GCC option will be
13236 used during linking.
13238 @item -fno-jump-tables
13239 @opindex fno-jump-tables
13240 Do not use jump tables for switch statements even where it would be
13241 more efficient than other code generation strategies. This option is
13242 of use in conjunction with @option{-fpic} or @option{-fPIC} for
13243 building code which forms part of a dynamic linker and cannot
13244 reference the address of a jump table. On some targets, jump tables
13245 do not require a GOT and this option is not needed.
13247 @item -ffixed-@var{reg}
13249 Treat the register named @var{reg} as a fixed register; generated code
13250 should never refer to it (except perhaps as a stack pointer, frame
13251 pointer or in some other fixed role).
13253 @var{reg} must be the name of a register. The register names accepted
13254 are machine-specific and are defined in the @code{REGISTER_NAMES}
13255 macro in the machine description macro file.
13257 This flag does not have a negative form, because it specifies a
13260 @item -fcall-used-@var{reg}
13261 @opindex fcall-used
13262 Treat the register named @var{reg} as an allocable register that is
13263 clobbered by function calls. It may be allocated for temporaries or
13264 variables that do not live across a call. Functions compiled this way
13265 will not save and restore the register @var{reg}.
13267 It is an error to used this flag with the frame pointer or stack pointer.
13268 Use of this flag for other registers that have fixed pervasive roles in
13269 the machine's execution model will produce disastrous results.
13271 This flag does not have a negative form, because it specifies a
13274 @item -fcall-saved-@var{reg}
13275 @opindex fcall-saved
13276 Treat the register named @var{reg} as an allocable register saved by
13277 functions. It may be allocated even for temporaries or variables that
13278 live across a call. Functions compiled this way will save and restore
13279 the register @var{reg} if they use it.
13281 It is an error to used this flag with the frame pointer or stack pointer.
13282 Use of this flag for other registers that have fixed pervasive roles in
13283 the machine's execution model will produce disastrous results.
13285 A different sort of disaster will result from the use of this flag for
13286 a register in which function values may be returned.
13288 This flag does not have a negative form, because it specifies a
13291 @item -fpack-struct[=@var{n}]
13292 @opindex fpack-struct
13293 Without a value specified, pack all structure members together without
13294 holes. When a value is specified (which must be a small power of two), pack
13295 structure members according to this value, representing the maximum
13296 alignment (that is, objects with default alignment requirements larger than
13297 this will be output potentially unaligned at the next fitting location.
13299 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
13300 code that is not binary compatible with code generated without that switch.
13301 Additionally, it makes the code suboptimal.
13302 Use it to conform to a non-default application binary interface.
13304 @item -finstrument-functions
13305 @opindex finstrument-functions
13306 Generate instrumentation calls for entry and exit to functions. Just
13307 after function entry and just before function exit, the following
13308 profiling functions will be called with the address of the current
13309 function and its call site. (On some platforms,
13310 @code{__builtin_return_address} does not work beyond the current
13311 function, so the call site information may not be available to the
13312 profiling functions otherwise.)
13315 void __cyg_profile_func_enter (void *this_fn,
13317 void __cyg_profile_func_exit (void *this_fn,
13321 The first argument is the address of the start of the current function,
13322 which may be looked up exactly in the symbol table.
13324 This instrumentation is also done for functions expanded inline in other
13325 functions. The profiling calls will indicate where, conceptually, the
13326 inline function is entered and exited. This means that addressable
13327 versions of such functions must be available. If all your uses of a
13328 function are expanded inline, this may mean an additional expansion of
13329 code size. If you use @samp{extern inline} in your C code, an
13330 addressable version of such functions must be provided. (This is
13331 normally the case anyways, but if you get lucky and the optimizer always
13332 expands the functions inline, you might have gotten away without
13333 providing static copies.)
13335 A function may be given the attribute @code{no_instrument_function}, in
13336 which case this instrumentation will not be done. This can be used, for
13337 example, for the profiling functions listed above, high-priority
13338 interrupt routines, and any functions from which the profiling functions
13339 cannot safely be called (perhaps signal handlers, if the profiling
13340 routines generate output or allocate memory).
13342 @item -fstack-check
13343 @opindex fstack-check
13344 Generate code to verify that you do not go beyond the boundary of the
13345 stack. You should specify this flag if you are running in an
13346 environment with multiple threads, but only rarely need to specify it in
13347 a single-threaded environment since stack overflow is automatically
13348 detected on nearly all systems if there is only one stack.
13350 Note that this switch does not actually cause checking to be done; the
13351 operating system must do that. The switch causes generation of code
13352 to ensure that the operating system sees the stack being extended.
13354 @item -fstack-limit-register=@var{reg}
13355 @itemx -fstack-limit-symbol=@var{sym}
13356 @itemx -fno-stack-limit
13357 @opindex fstack-limit-register
13358 @opindex fstack-limit-symbol
13359 @opindex fno-stack-limit
13360 Generate code to ensure that the stack does not grow beyond a certain value,
13361 either the value of a register or the address of a symbol. If the stack
13362 would grow beyond the value, a signal is raised. For most targets,
13363 the signal is raised before the stack overruns the boundary, so
13364 it is possible to catch the signal without taking special precautions.
13366 For instance, if the stack starts at absolute address @samp{0x80000000}
13367 and grows downwards, you can use the flags
13368 @option{-fstack-limit-symbol=__stack_limit} and
13369 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
13370 of 128KB@. Note that this may only work with the GNU linker.
13372 @cindex aliasing of parameters
13373 @cindex parameters, aliased
13374 @item -fargument-alias
13375 @itemx -fargument-noalias
13376 @itemx -fargument-noalias-global
13377 @itemx -fargument-noalias-anything
13378 @opindex fargument-alias
13379 @opindex fargument-noalias
13380 @opindex fargument-noalias-global
13381 @opindex fargument-noalias-anything
13382 Specify the possible relationships among parameters and between
13383 parameters and global data.
13385 @option{-fargument-alias} specifies that arguments (parameters) may
13386 alias each other and may alias global storage.@*
13387 @option{-fargument-noalias} specifies that arguments do not alias
13388 each other, but may alias global storage.@*
13389 @option{-fargument-noalias-global} specifies that arguments do not
13390 alias each other and do not alias global storage.
13391 @option{-fargument-noalias-anything} specifies that arguments do not
13392 alias any other storage.
13394 Each language will automatically use whatever option is required by
13395 the language standard. You should not need to use these options yourself.
13397 @item -fleading-underscore
13398 @opindex fleading-underscore
13399 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
13400 change the way C symbols are represented in the object file. One use
13401 is to help link with legacy assembly code.
13403 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
13404 generate code that is not binary compatible with code generated without that
13405 switch. Use it to conform to a non-default application binary interface.
13406 Not all targets provide complete support for this switch.
13408 @item -ftls-model=@var{model}
13409 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
13410 The @var{model} argument should be one of @code{global-dynamic},
13411 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
13413 The default without @option{-fpic} is @code{initial-exec}; with
13414 @option{-fpic} the default is @code{global-dynamic}.
13416 @item -fvisibility=@var{default|internal|hidden|protected}
13417 @opindex fvisibility
13418 Set the default ELF image symbol visibility to the specified option---all
13419 symbols will be marked with this unless overridden within the code.
13420 Using this feature can very substantially improve linking and
13421 load times of shared object libraries, produce more optimized
13422 code, provide near-perfect API export and prevent symbol clashes.
13423 It is @strong{strongly} recommended that you use this in any shared objects
13426 Despite the nomenclature, @code{default} always means public ie;
13427 available to be linked against from outside the shared object.
13428 @code{protected} and @code{internal} are pretty useless in real-world
13429 usage so the only other commonly used option will be @code{hidden}.
13430 The default if @option{-fvisibility} isn't specified is
13431 @code{default}, i.e., make every
13432 symbol public---this causes the same behavior as previous versions of
13435 A good explanation of the benefits offered by ensuring ELF
13436 symbols have the correct visibility is given by ``How To Write
13437 Shared Libraries'' by Ulrich Drepper (which can be found at
13438 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
13439 solution made possible by this option to marking things hidden when
13440 the default is public is to make the default hidden and mark things
13441 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
13442 and @code{__attribute__ ((visibility("default")))} instead of
13443 @code{__declspec(dllexport)} you get almost identical semantics with
13444 identical syntax. This is a great boon to those working with
13445 cross-platform projects.
13447 For those adding visibility support to existing code, you may find
13448 @samp{#pragma GCC visibility} of use. This works by you enclosing
13449 the declarations you wish to set visibility for with (for example)
13450 @samp{#pragma GCC visibility push(hidden)} and
13451 @samp{#pragma GCC visibility pop}.
13452 Bear in mind that symbol visibility should be viewed @strong{as
13453 part of the API interface contract} and thus all new code should
13454 always specify visibility when it is not the default ie; declarations
13455 only for use within the local DSO should @strong{always} be marked explicitly
13456 as hidden as so to avoid PLT indirection overheads---making this
13457 abundantly clear also aids readability and self-documentation of the code.
13458 Note that due to ISO C++ specification requirements, operator new and
13459 operator delete must always be of default visibility.
13461 Be aware that headers from outside your project, in particular system
13462 headers and headers from any other library you use, may not be
13463 expecting to be compiled with visibility other than the default. You
13464 may need to explicitly say @samp{#pragma GCC visibility push(default)}
13465 before including any such headers.
13467 An overview of these techniques, their benefits and how to use them
13468 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
13472 @cindex openmp parallel
13473 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
13474 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
13475 compiler generates parallel code according to the OpenMP Application
13476 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.
13482 @node Environment Variables
13483 @section Environment Variables Affecting GCC
13484 @cindex environment variables
13486 @c man begin ENVIRONMENT
13487 This section describes several environment variables that affect how GCC
13488 operates. Some of them work by specifying directories or prefixes to use
13489 when searching for various kinds of files. Some are used to specify other
13490 aspects of the compilation environment.
13492 Note that you can also specify places to search using options such as
13493 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
13494 take precedence over places specified using environment variables, which
13495 in turn take precedence over those specified by the configuration of GCC@.
13496 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
13497 GNU Compiler Collection (GCC) Internals}.
13502 @c @itemx LC_COLLATE
13504 @c @itemx LC_MONETARY
13505 @c @itemx LC_NUMERIC
13510 @c @findex LC_COLLATE
13511 @findex LC_MESSAGES
13512 @c @findex LC_MONETARY
13513 @c @findex LC_NUMERIC
13517 These environment variables control the way that GCC uses
13518 localization information that allow GCC to work with different
13519 national conventions. GCC inspects the locale categories
13520 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
13521 so. These locale categories can be set to any value supported by your
13522 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
13523 Kingdom encoded in UTF-8.
13525 The @env{LC_CTYPE} environment variable specifies character
13526 classification. GCC uses it to determine the character boundaries in
13527 a string; this is needed for some multibyte encodings that contain quote
13528 and escape characters that would otherwise be interpreted as a string
13531 The @env{LC_MESSAGES} environment variable specifies the language to
13532 use in diagnostic messages.
13534 If the @env{LC_ALL} environment variable is set, it overrides the value
13535 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
13536 and @env{LC_MESSAGES} default to the value of the @env{LANG}
13537 environment variable. If none of these variables are set, GCC
13538 defaults to traditional C English behavior.
13542 If @env{TMPDIR} is set, it specifies the directory to use for temporary
13543 files. GCC uses temporary files to hold the output of one stage of
13544 compilation which is to be used as input to the next stage: for example,
13545 the output of the preprocessor, which is the input to the compiler
13548 @item GCC_EXEC_PREFIX
13549 @findex GCC_EXEC_PREFIX
13550 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
13551 names of the subprograms executed by the compiler. No slash is added
13552 when this prefix is combined with the name of a subprogram, but you can
13553 specify a prefix that ends with a slash if you wish.
13555 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
13556 an appropriate prefix to use based on the pathname it was invoked with.
13558 If GCC cannot find the subprogram using the specified prefix, it
13559 tries looking in the usual places for the subprogram.
13561 The default value of @env{GCC_EXEC_PREFIX} is
13562 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
13563 of @code{prefix} when you ran the @file{configure} script.
13565 Other prefixes specified with @option{-B} take precedence over this prefix.
13567 This prefix is also used for finding files such as @file{crt0.o} that are
13570 In addition, the prefix is used in an unusual way in finding the
13571 directories to search for header files. For each of the standard
13572 directories whose name normally begins with @samp{/usr/local/lib/gcc}
13573 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
13574 replacing that beginning with the specified prefix to produce an
13575 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
13576 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
13577 These alternate directories are searched first; the standard directories
13580 @item COMPILER_PATH
13581 @findex COMPILER_PATH
13582 The value of @env{COMPILER_PATH} is a colon-separated list of
13583 directories, much like @env{PATH}. GCC tries the directories thus
13584 specified when searching for subprograms, if it can't find the
13585 subprograms using @env{GCC_EXEC_PREFIX}.
13588 @findex LIBRARY_PATH
13589 The value of @env{LIBRARY_PATH} is a colon-separated list of
13590 directories, much like @env{PATH}. When configured as a native compiler,
13591 GCC tries the directories thus specified when searching for special
13592 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
13593 using GCC also uses these directories when searching for ordinary
13594 libraries for the @option{-l} option (but directories specified with
13595 @option{-L} come first).
13599 @cindex locale definition
13600 This variable is used to pass locale information to the compiler. One way in
13601 which this information is used is to determine the character set to be used
13602 when character literals, string literals and comments are parsed in C and C++.
13603 When the compiler is configured to allow multibyte characters,
13604 the following values for @env{LANG} are recognized:
13608 Recognize JIS characters.
13610 Recognize SJIS characters.
13612 Recognize EUCJP characters.
13615 If @env{LANG} is not defined, or if it has some other value, then the
13616 compiler will use mblen and mbtowc as defined by the default locale to
13617 recognize and translate multibyte characters.
13621 Some additional environments variables affect the behavior of the
13624 @include cppenv.texi
13628 @node Precompiled Headers
13629 @section Using Precompiled Headers
13630 @cindex precompiled headers
13631 @cindex speed of compilation
13633 Often large projects have many header files that are included in every
13634 source file. The time the compiler takes to process these header files
13635 over and over again can account for nearly all of the time required to
13636 build the project. To make builds faster, GCC allows users to
13637 `precompile' a header file; then, if builds can use the precompiled
13638 header file they will be much faster.
13640 To create a precompiled header file, simply compile it as you would any
13641 other file, if necessary using the @option{-x} option to make the driver
13642 treat it as a C or C++ header file. You will probably want to use a
13643 tool like @command{make} to keep the precompiled header up-to-date when
13644 the headers it contains change.
13646 A precompiled header file will be searched for when @code{#include} is
13647 seen in the compilation. As it searches for the included file
13648 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
13649 compiler looks for a precompiled header in each directory just before it
13650 looks for the include file in that directory. The name searched for is
13651 the name specified in the @code{#include} with @samp{.gch} appended. If
13652 the precompiled header file can't be used, it is ignored.
13654 For instance, if you have @code{#include "all.h"}, and you have
13655 @file{all.h.gch} in the same directory as @file{all.h}, then the
13656 precompiled header file will be used if possible, and the original
13657 header will be used otherwise.
13659 Alternatively, you might decide to put the precompiled header file in a
13660 directory and use @option{-I} to ensure that directory is searched
13661 before (or instead of) the directory containing the original header.
13662 Then, if you want to check that the precompiled header file is always
13663 used, you can put a file of the same name as the original header in this
13664 directory containing an @code{#error} command.
13666 This also works with @option{-include}. So yet another way to use
13667 precompiled headers, good for projects not designed with precompiled
13668 header files in mind, is to simply take most of the header files used by
13669 a project, include them from another header file, precompile that header
13670 file, and @option{-include} the precompiled header. If the header files
13671 have guards against multiple inclusion, they will be skipped because
13672 they've already been included (in the precompiled header).
13674 If you need to precompile the same header file for different
13675 languages, targets, or compiler options, you can instead make a
13676 @emph{directory} named like @file{all.h.gch}, and put each precompiled
13677 header in the directory, perhaps using @option{-o}. It doesn't matter
13678 what you call the files in the directory, every precompiled header in
13679 the directory will be considered. The first precompiled header
13680 encountered in the directory that is valid for this compilation will
13681 be used; they're searched in no particular order.
13683 There are many other possibilities, limited only by your imagination,
13684 good sense, and the constraints of your build system.
13686 A precompiled header file can be used only when these conditions apply:
13690 Only one precompiled header can be used in a particular compilation.
13693 A precompiled header can't be used once the first C token is seen. You
13694 can have preprocessor directives before a precompiled header; you can
13695 even include a precompiled header from inside another header, so long as
13696 there are no C tokens before the @code{#include}.
13699 The precompiled header file must be produced for the same language as
13700 the current compilation. You can't use a C precompiled header for a C++
13704 The precompiled header file must have been produced by the same compiler
13705 binary as the current compilation is using.
13708 Any macros defined before the precompiled header is included must
13709 either be defined in the same way as when the precompiled header was
13710 generated, or must not affect the precompiled header, which usually
13711 means that they don't appear in the precompiled header at all.
13713 The @option{-D} option is one way to define a macro before a
13714 precompiled header is included; using a @code{#define} can also do it.
13715 There are also some options that define macros implicitly, like
13716 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
13719 @item If debugging information is output when using the precompiled
13720 header, using @option{-g} or similar, the same kind of debugging information
13721 must have been output when building the precompiled header. However,
13722 a precompiled header built using @option{-g} can be used in a compilation
13723 when no debugging information is being output.
13725 @item The same @option{-m} options must generally be used when building
13726 and using the precompiled header. @xref{Submodel Options},
13727 for any cases where this rule is relaxed.
13729 @item Each of the following options must be the same when building and using
13730 the precompiled header:
13732 @gccoptlist{-fexceptions -funit-at-a-time}
13735 Some other command-line options starting with @option{-f},
13736 @option{-p}, or @option{-O} must be defined in the same way as when
13737 the precompiled header was generated. At present, it's not clear
13738 which options are safe to change and which are not; the safest choice
13739 is to use exactly the same options when generating and using the
13740 precompiled header. The following are known to be safe:
13742 @gccoptlist{-fmessage-length= -fpreprocessed
13743 -fsched-interblock -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous
13744 -fsched-verbose=<number> -fschedule-insns -fvisibility=
13749 For all of these except the last, the compiler will automatically
13750 ignore the precompiled header if the conditions aren't met. If you
13751 find an option combination that doesn't work and doesn't cause the
13752 precompiled header to be ignored, please consider filing a bug report,
13755 If you do use differing options when generating and using the
13756 precompiled header, the actual behavior will be a mixture of the
13757 behavior for the options. For instance, if you use @option{-g} to
13758 generate the precompiled header but not when using it, you may or may
13759 not get debugging information for routines in the precompiled header.
13761 @node Running Protoize
13762 @section Running Protoize
13764 The program @code{protoize} is an optional part of GCC@. You can use
13765 it to add prototypes to a program, thus converting the program to ISO
13766 C in one respect. The companion program @code{unprotoize} does the
13767 reverse: it removes argument types from any prototypes that are found.
13769 When you run these programs, you must specify a set of source files as
13770 command line arguments. The conversion programs start out by compiling
13771 these files to see what functions they define. The information gathered
13772 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
13774 After scanning comes actual conversion. The specified files are all
13775 eligible to be converted; any files they include (whether sources or
13776 just headers) are eligible as well.
13778 But not all the eligible files are converted. By default,
13779 @code{protoize} and @code{unprotoize} convert only source and header
13780 files in the current directory. You can specify additional directories
13781 whose files should be converted with the @option{-d @var{directory}}
13782 option. You can also specify particular files to exclude with the
13783 @option{-x @var{file}} option. A file is converted if it is eligible, its
13784 directory name matches one of the specified directory names, and its
13785 name within the directory has not been excluded.
13787 Basic conversion with @code{protoize} consists of rewriting most
13788 function definitions and function declarations to specify the types of
13789 the arguments. The only ones not rewritten are those for varargs
13792 @code{protoize} optionally inserts prototype declarations at the
13793 beginning of the source file, to make them available for any calls that
13794 precede the function's definition. Or it can insert prototype
13795 declarations with block scope in the blocks where undeclared functions
13798 Basic conversion with @code{unprotoize} consists of rewriting most
13799 function declarations to remove any argument types, and rewriting
13800 function definitions to the old-style pre-ISO form.
13802 Both conversion programs print a warning for any function declaration or
13803 definition that they can't convert. You can suppress these warnings
13806 The output from @code{protoize} or @code{unprotoize} replaces the
13807 original source file. The original file is renamed to a name ending
13808 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
13809 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
13810 for DOS) file already exists, then the source file is simply discarded.
13812 @code{protoize} and @code{unprotoize} both depend on GCC itself to
13813 scan the program and collect information about the functions it uses.
13814 So neither of these programs will work until GCC is installed.
13816 Here is a table of the options you can use with @code{protoize} and
13817 @code{unprotoize}. Each option works with both programs unless
13821 @item -B @var{directory}
13822 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
13823 usual directory (normally @file{/usr/local/lib}). This file contains
13824 prototype information about standard system functions. This option
13825 applies only to @code{protoize}.
13827 @item -c @var{compilation-options}
13828 Use @var{compilation-options} as the options when running @command{gcc} to
13829 produce the @samp{.X} files. The special option @option{-aux-info} is
13830 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
13832 Note that the compilation options must be given as a single argument to
13833 @code{protoize} or @code{unprotoize}. If you want to specify several
13834 @command{gcc} options, you must quote the entire set of compilation options
13835 to make them a single word in the shell.
13837 There are certain @command{gcc} arguments that you cannot use, because they
13838 would produce the wrong kind of output. These include @option{-g},
13839 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
13840 the @var{compilation-options}, they are ignored.
13843 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
13844 systems) instead of @samp{.c}. This is convenient if you are converting
13845 a C program to C++. This option applies only to @code{protoize}.
13848 Add explicit global declarations. This means inserting explicit
13849 declarations at the beginning of each source file for each function
13850 that is called in the file and was not declared. These declarations
13851 precede the first function definition that contains a call to an
13852 undeclared function. This option applies only to @code{protoize}.
13854 @item -i @var{string}
13855 Indent old-style parameter declarations with the string @var{string}.
13856 This option applies only to @code{protoize}.
13858 @code{unprotoize} converts prototyped function definitions to old-style
13859 function definitions, where the arguments are declared between the
13860 argument list and the initial @samp{@{}. By default, @code{unprotoize}
13861 uses five spaces as the indentation. If you want to indent with just
13862 one space instead, use @option{-i " "}.
13865 Keep the @samp{.X} files. Normally, they are deleted after conversion
13869 Add explicit local declarations. @code{protoize} with @option{-l} inserts
13870 a prototype declaration for each function in each block which calls the
13871 function without any declaration. This option applies only to
13875 Make no real changes. This mode just prints information about the conversions
13876 that would have been done without @option{-n}.
13879 Make no @samp{.save} files. The original files are simply deleted.
13880 Use this option with caution.
13882 @item -p @var{program}
13883 Use the program @var{program} as the compiler. Normally, the name
13884 @file{gcc} is used.
13887 Work quietly. Most warnings are suppressed.
13890 Print the version number, just like @option{-v} for @command{gcc}.
13893 If you need special compiler options to compile one of your program's
13894 source files, then you should generate that file's @samp{.X} file
13895 specially, by running @command{gcc} on that source file with the
13896 appropriate options and the option @option{-aux-info}. Then run
13897 @code{protoize} on the entire set of files. @code{protoize} will use
13898 the existing @samp{.X} file because it is newer than the source file.
13902 gcc -Dfoo=bar file1.c -aux-info file1.X
13907 You need to include the special files along with the rest in the
13908 @code{protoize} command, even though their @samp{.X} files already
13909 exist, because otherwise they won't get converted.
13911 @xref{Protoize Caveats}, for more information on how to use
13912 @code{protoize} successfully.