1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007
3 @c Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
12 @c man begin COPYRIGHT
13 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
14 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006 Free Software Foundation, Inc.
16 Permission is granted to copy, distribute and/or modify this document
17 under the terms of the GNU Free Documentation License, Version 1.2 or
18 any later version published by the Free Software Foundation; with the
19 Invariant Sections being ``GNU General Public License'' and ``Funding
20 Free Software'', the Front-Cover texts being (a) (see below), and with
21 the Back-Cover Texts being (b) (see below). A copy of the license is
22 included in the gfdl(7) man page.
24 (a) The FSF's Front-Cover Text is:
28 (b) The FSF's Back-Cover Text is:
30 You have freedom to copy and modify this GNU Manual, like GNU
31 software. Copies published by the Free Software Foundation raise
32 funds for GNU development.
34 @c Set file name and title for the man page.
36 @settitle GNU project C and C++ compiler
38 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
39 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
40 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
41 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
42 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
43 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
44 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
46 Only the most useful options are listed here; see below for the
47 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
50 gpl(7), gfdl(7), fsf-funding(7),
51 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
52 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
53 @file{ld}, @file{binutils} and @file{gdb}.
56 For instructions on reporting bugs, see
57 @w{@uref{http://gcc.gnu.org/bugs.html}}.
60 See the Info entry for @command{gcc}, or
61 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
62 for contributors to GCC@.
67 @chapter GCC Command Options
68 @cindex GCC command options
69 @cindex command options
70 @cindex options, GCC command
72 @c man begin DESCRIPTION
73 When you invoke GCC, it normally does preprocessing, compilation,
74 assembly and linking. The ``overall options'' allow you to stop this
75 process at an intermediate stage. For example, the @option{-c} option
76 says not to run the linker. Then the output consists of object files
77 output by the assembler.
79 Other options are passed on to one stage of processing. Some options
80 control the preprocessor and others the compiler itself. Yet other
81 options control the assembler and linker; most of these are not
82 documented here, since you rarely need to use any of them.
84 @cindex C compilation options
85 Most of the command line options that you can use with GCC are useful
86 for C programs; when an option is only useful with another language
87 (usually C++), the explanation says so explicitly. If the description
88 for a particular option does not mention a source language, you can use
89 that option with all supported languages.
91 @cindex C++ compilation options
92 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
93 options for compiling C++ programs.
95 @cindex grouping options
96 @cindex options, grouping
97 The @command{gcc} program accepts options and file names as operands. Many
98 options have multi-letter names; therefore multiple single-letter options
99 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
102 @cindex order of options
103 @cindex options, order
104 You can mix options and other arguments. For the most part, the order
105 you use doesn't matter. Order does matter when you use several options
106 of the same kind; for example, if you specify @option{-L} more than once,
107 the directories are searched in the order specified.
109 Many options have long names starting with @samp{-f} or with
110 @samp{-W}---for example,
111 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
112 these have both positive and negative forms; the negative form of
113 @option{-ffoo} would be @option{-fno-foo}. This manual documents
114 only one of these two forms, whichever one is not the default.
118 @xref{Option Index}, for an index to GCC's options.
121 * Option Summary:: Brief list of all options, without explanations.
122 * Overall Options:: Controlling the kind of output:
123 an executable, object files, assembler files,
124 or preprocessed source.
125 * Invoking G++:: Compiling C++ programs.
126 * C Dialect Options:: Controlling the variant of C language compiled.
127 * C++ Dialect Options:: Variations on C++.
128 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
130 * Language Independent Options:: Controlling how diagnostics should be
132 * Warning Options:: How picky should the compiler be?
133 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
134 * Optimize Options:: How much optimization?
135 * Preprocessor Options:: Controlling header files and macro definitions.
136 Also, getting dependency information for Make.
137 * Assembler Options:: Passing options to the assembler.
138 * Link Options:: Specifying libraries and so on.
139 * Directory Options:: Where to find header files and libraries.
140 Where to find the compiler executable files.
141 * Spec Files:: How to pass switches to sub-processes.
142 * Target Options:: Running a cross-compiler, or an old version of GCC.
143 * Submodel Options:: Specifying minor hardware or convention variations,
144 such as 68010 vs 68020.
145 * Code Gen Options:: Specifying conventions for function calls, data layout
147 * Environment Variables:: Env vars that affect GCC.
148 * Precompiled Headers:: Compiling a header once, and using it many times.
149 * Running Protoize:: Automatically adding or removing function prototypes.
155 @section Option Summary
157 Here is a summary of all the options, grouped by type. Explanations are
158 in the following sections.
161 @item Overall Options
162 @xref{Overall Options,,Options Controlling the Kind of Output}.
163 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
164 -x @var{language} -v -### --help --target-help --version @@@var{file}}
166 @item C Language Options
167 @xref{C Dialect Options,,Options Controlling C Dialect}.
168 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
169 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
170 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
171 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
172 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
173 -fsigned-bitfields -fsigned-char @gol
174 -funsigned-bitfields -funsigned-char}
176 @item C++ Language Options
177 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
178 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
179 -fconserve-space -ffriend-injection @gol
180 -fno-elide-constructors @gol
181 -fno-enforce-eh-specs @gol
182 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
183 -fno-implicit-templates @gol
184 -fno-implicit-inline-templates @gol
185 -fno-implement-inlines -fms-extensions @gol
186 -fno-nonansi-builtins -fno-operator-names @gol
187 -fno-optional-diags -fpermissive @gol
188 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
189 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
190 -fno-default-inline -fvisibility-inlines-hidden @gol
191 -Wabi -Wctor-dtor-privacy @gol
192 -Wnon-virtual-dtor -Wreorder @gol
193 -Weffc++ -Wno-deprecated -Wstrict-null-sentinel @gol
194 -Wno-non-template-friend -Wold-style-cast @gol
195 -Woverloaded-virtual -Wno-pmf-conversions @gol
198 @item Objective-C and Objective-C++ Language Options
199 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
200 Objective-C and Objective-C++ Dialects}.
201 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
202 -fgnu-runtime -fnext-runtime @gol
203 -fno-nil-receivers @gol
204 -fobjc-call-cxx-cdtors @gol
205 -fobjc-direct-dispatch @gol
206 -fobjc-exceptions @gol
208 -freplace-objc-classes @gol
211 -Wassign-intercept @gol
212 -Wno-protocol -Wselector @gol
213 -Wstrict-selector-match @gol
214 -Wundeclared-selector}
216 @item Language Independent Options
217 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
218 @gccoptlist{-fmessage-length=@var{n} @gol
219 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
220 -fdiagnostics-show-option}
222 @item Warning Options
223 @xref{Warning Options,,Options to Request or Suppress Warnings}.
224 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
225 -w -Wextra -Wall -Waggregate-return -Walways-true -Warray-bounds @gol
226 -Wno-attributes -Wc++-compat -Wcast-align -Wcast-qual -Wchar-subscripts @gol
227 -Wclobbered -Wcomment @gol
228 -Wconversion -Wno-deprecated-declarations @gol
229 -Wdisabled-optimization -Wno-div-by-zero @gol
230 -Wempty-body -Wno-endif-labels @gol
231 -Werror -Werror-* -Werror-implicit-function-declaration @gol
232 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
233 -Wno-format-extra-args -Wformat-nonliteral @gol
234 -Wformat-security -Wformat-y2k @gol
235 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
236 -Wimport -Wno-import -Winit-self -Winline @gol
237 -Wno-int-to-pointer-cast @gol
238 -Wno-invalid-offsetof -Winvalid-pch @gol
239 -Wlarger-than-@var{len} -Wunsafe-loop-optimizations -Wlong-long @gol
240 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
241 -Wmissing-format-attribute -Wmissing-include-dirs @gol
242 -Wmissing-noreturn @gol
243 -Wno-multichar -Wnonnull -Wno-overflow @gol
244 -Woverlength-strings -Wpacked -Wpadded @gol
245 -Wparentheses -Wpointer-arith -Wno-pointer-to-int-cast @gol
246 -Wredundant-decls @gol
247 -Wreturn-type -Wsequence-point -Wshadow @gol
248 -Wsign-compare -Wstack-protector @gol
249 -Wstrict-aliasing -Wstrict-aliasing=2 @gol
250 -Wstring-literal-comparison @gol
251 -Wswitch -Wswitch-default -Wswitch-enum @gol
252 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
253 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
254 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
255 -Wunused-value -Wunused-variable -Wvariadic-macros @gol
256 -Wvolatile-register-var -Wwrite-strings}
258 @item C-only Warning Options
259 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
260 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
261 -Wold-style-declaration -Wold-style-definition @gol
262 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
263 -Wdeclaration-after-statement -Wpointer-sign}
265 @item Debugging Options
266 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
267 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
268 -fdump-noaddr -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
269 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
270 -fdump-ipa-all -fdump-ipa-cgraph @gol
272 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
273 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
274 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
275 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
277 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
278 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
279 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
280 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
281 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
282 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
283 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
284 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
285 -fdump-tree-nrv -fdump-tree-vect @gol
286 -fdump-tree-sink @gol
287 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
288 -fdump-tree-salias @gol
289 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
290 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
291 -ftree-vectorizer-verbose=@var{n} @gol
292 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
293 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
294 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
295 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
296 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
297 -ftest-coverage -ftime-report -fvar-tracking @gol
298 -g -g@var{level} -gcoff -gdwarf-2 @gol
299 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
300 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
301 -print-multi-directory -print-multi-lib @gol
302 -print-prog-name=@var{program} -print-search-dirs -Q @gol
305 @item Optimization Options
306 @xref{Optimize Options,,Options that Control Optimization}.
307 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
308 -falign-labels=@var{n} -falign-loops=@var{n} @gol
309 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
310 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
311 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
312 -fcaller-saves -fcprop-registers -fcse-follow-jumps @gol
313 -fcse-skip-blocks -fcx-limited-range -fdata-sections @gol
314 -fdelayed-branch -fdelete-null-pointer-checks -fearly-inlining @gol
315 -fexpensive-optimizations -ffast-math -ffloat-store @gol
316 -fforce-addr -fforward-propagate -ffunction-sections @gol
317 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
318 -fcrossjumping -fif-conversion -fif-conversion2 @gol
319 -finline-functions -finline-functions-called-once @gol
320 -finline-limit=@var{n} -fkeep-inline-functions @gol
321 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
322 -fmodulo-sched -fno-branch-count-reg @gol
323 -fno-default-inline -fno-defer-pop -fmove-loop-invariants @gol
324 -fno-function-cse -fno-guess-branch-probability @gol
325 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
326 -funsafe-math-optimizations -funsafe-loop-optimizations @gol
327 -ffinite-math-only -fno-signed-zeros @gol
328 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
329 -fomit-frame-pointer -foptimize-register-move @gol
330 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
331 -fprofile-generate -fprofile-use @gol
332 -fregmove -frename-registers @gol
333 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
334 -frerun-cse-after-loop @gol
335 -frounding-math -frtl-abstract-sequences @gol
336 -fschedule-insns -fschedule-insns2 @gol
337 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
338 -fsched-spec-load-dangerous @gol
339 -fsched-stalled-insns=@var{n} -fsched-stalled-insns-dep=@var{n} @gol
340 -fsched2-use-superblocks @gol
341 -fsched2-use-traces -fsee -freschedule-modulo-scheduled-loops @gol
342 -fsection-anchors -fsignaling-nans -fsingle-precision-constant @gol
343 -fstack-protector -fstack-protector-all @gol
344 -fstrict-aliasing -ftracer -fthread-jumps @gol
345 -funroll-all-loops -funroll-loops -fpeel-loops @gol
346 -fsplit-ivs-in-unroller -funswitch-loops @gol
347 -fvariable-expansion-in-unroller @gol
348 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
349 -ftree-loop-linear -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
350 -ftree-dominator-opts -ftree-dse -ftree-copyrename -ftree-sink @gol
351 -ftree-ch -ftree-sra -ftree-ter -ftree-lrs -ftree-fre -ftree-vectorize @gol
352 -ftree-vect-loop-version -ftree-salias -fipa-pta -fweb @gol
353 -ftree-copy-prop -ftree-store-ccp -ftree-store-copy-prop -fwhole-program @gol
354 --param @var{name}=@var{value}
355 -O -O0 -O1 -O2 -O3 -Os}
357 @item Preprocessor Options
358 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
359 @gccoptlist{-A@var{question}=@var{answer} @gol
360 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
361 -C -dD -dI -dM -dN @gol
362 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
363 -idirafter @var{dir} @gol
364 -include @var{file} -imacros @var{file} @gol
365 -iprefix @var{file} -iwithprefix @var{dir} @gol
366 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
367 -imultilib @var{dir} -isysroot @var{dir} @gol
368 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
369 -P -fworking-directory -remap @gol
370 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
371 -Xpreprocessor @var{option}}
373 @item Assembler Option
374 @xref{Assembler Options,,Passing Options to the Assembler}.
375 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
378 @xref{Link Options,,Options for Linking}.
379 @gccoptlist{@var{object-file-name} -l@var{library} @gol
380 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
381 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
382 -Wl,@var{option} -Xlinker @var{option} @gol
385 @item Directory Options
386 @xref{Directory Options,,Options for Directory Search}.
387 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
388 -specs=@var{file} -I- --sysroot=@var{dir}}
391 @c I wrote this xref this way to avoid overfull hbox. -- rms
392 @xref{Target Options}.
393 @gccoptlist{-V @var{version} -b @var{machine}}
395 @item Machine Dependent Options
396 @xref{Submodel Options,,Hardware Models and Configurations}.
397 @c This list is ordered alphanumerically by subsection name.
398 @c Try and put the significant identifier (CPU or system) first,
399 @c so users have a clue at guessing where the ones they want will be.
402 @gccoptlist{-EB -EL @gol
403 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
404 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
407 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
408 -mabi=@var{name} @gol
409 -mapcs-stack-check -mno-apcs-stack-check @gol
410 -mapcs-float -mno-apcs-float @gol
411 -mapcs-reentrant -mno-apcs-reentrant @gol
412 -msched-prolog -mno-sched-prolog @gol
413 -mlittle-endian -mbig-endian -mwords-little-endian @gol
414 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
415 -mthumb-interwork -mno-thumb-interwork @gol
416 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
417 -mstructure-size-boundary=@var{n} @gol
418 -mabort-on-noreturn @gol
419 -mlong-calls -mno-long-calls @gol
420 -msingle-pic-base -mno-single-pic-base @gol
421 -mpic-register=@var{reg} @gol
422 -mnop-fun-dllimport @gol
423 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
424 -mpoke-function-name @gol
426 -mtpcs-frame -mtpcs-leaf-frame @gol
427 -mcaller-super-interworking -mcallee-super-interworking @gol
431 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
432 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
434 @emph{Blackfin Options}
435 @gccoptlist{-momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
436 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
437 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
438 -mno-id-shared-library -mshared-library-id=@var{n} @gol
439 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
440 -msep-data -mno-sep-data -mlong-calls -mno-long-calls}
443 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
444 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
445 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
446 -mstack-align -mdata-align -mconst-align @gol
447 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
448 -melf -maout -melinux -mlinux -sim -sim2 @gol
449 -mmul-bug-workaround -mno-mul-bug-workaround}
452 @gccoptlist{-mmac -mpush-args}
454 @emph{Darwin Options}
455 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
456 -arch_only -bind_at_load -bundle -bundle_loader @gol
457 -client_name -compatibility_version -current_version @gol
459 -dependency-file -dylib_file -dylinker_install_name @gol
460 -dynamic -dynamiclib -exported_symbols_list @gol
461 -filelist -flat_namespace -force_cpusubtype_ALL @gol
462 -force_flat_namespace -headerpad_max_install_names @gol
463 -image_base -init -install_name -keep_private_externs @gol
464 -multi_module -multiply_defined -multiply_defined_unused @gol
465 -noall_load -no_dead_strip_inits_and_terms @gol
466 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
467 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
468 -private_bundle -read_only_relocs -sectalign @gol
469 -sectobjectsymbols -whyload -seg1addr @gol
470 -sectcreate -sectobjectsymbols -sectorder @gol
471 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
472 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
473 -segprot -segs_read_only_addr -segs_read_write_addr @gol
474 -single_module -static -sub_library -sub_umbrella @gol
475 -twolevel_namespace -umbrella -undefined @gol
476 -unexported_symbols_list -weak_reference_mismatches @gol
477 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
478 -mkernel -mone-byte-bool}
480 @emph{DEC Alpha Options}
481 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
482 -mieee -mieee-with-inexact -mieee-conformant @gol
483 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
484 -mtrap-precision=@var{mode} -mbuild-constants @gol
485 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
486 -mbwx -mmax -mfix -mcix @gol
487 -mfloat-vax -mfloat-ieee @gol
488 -mexplicit-relocs -msmall-data -mlarge-data @gol
489 -msmall-text -mlarge-text @gol
490 -mmemory-latency=@var{time}}
492 @emph{DEC Alpha/VMS Options}
493 @gccoptlist{-mvms-return-codes}
496 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
497 -mhard-float -msoft-float @gol
498 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
499 -mdouble -mno-double @gol
500 -mmedia -mno-media -mmuladd -mno-muladd @gol
501 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
502 -mlinked-fp -mlong-calls -malign-labels @gol
503 -mlibrary-pic -macc-4 -macc-8 @gol
504 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
505 -moptimize-membar -mno-optimize-membar @gol
506 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
507 -mvliw-branch -mno-vliw-branch @gol
508 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
509 -mno-nested-cond-exec -mtomcat-stats @gol
513 @emph{GNU/Linux Options}
514 @gccoptlist{-muclibc}
516 @emph{H8/300 Options}
517 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
520 @gccoptlist{-march=@var{architecture-type} @gol
521 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
522 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
523 -mfixed-range=@var{register-range} @gol
524 -mjump-in-delay -mlinker-opt -mlong-calls @gol
525 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
526 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
527 -mno-jump-in-delay -mno-long-load-store @gol
528 -mno-portable-runtime -mno-soft-float @gol
529 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
530 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
531 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
532 -munix=@var{unix-std} -nolibdld -static -threads}
534 @emph{i386 and x86-64 Options}
535 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
536 -mfpmath=@var{unit} @gol
537 -masm=@var{dialect} -mno-fancy-math-387 @gol
538 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
539 -mno-wide-multiply -mrtd -malign-double @gol
540 -mpreferred-stack-boundary=@var{num} @gol
541 -mmmx -msse -msse2 -msse3 -mssse3 -m3dnow @gol
542 -mthreads -mno-align-stringops -minline-all-stringops @gol
543 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
544 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
546 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
547 -mcmodel=@var{code-model} @gol
548 -m32 -m64 -mlarge-data-threshold=@var{num}}
551 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
552 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
553 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
554 -minline-float-divide-max-throughput @gol
555 -minline-int-divide-min-latency @gol
556 -minline-int-divide-max-throughput @gol
557 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
558 -mno-dwarf2-asm -mearly-stop-bits @gol
559 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
560 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
561 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
562 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
563 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
564 -mno-sched-prefer-non-data-spec-insns @gol
565 -mno-sched-prefer-non-control-spec-insns @gol
566 -mno-sched-count-spec-in-critical-path}
568 @emph{M32R/D Options}
569 @gccoptlist{-m32r2 -m32rx -m32r @gol
571 -malign-loops -mno-align-loops @gol
572 -missue-rate=@var{number} @gol
573 -mbranch-cost=@var{number} @gol
574 -mmodel=@var{code-size-model-type} @gol
575 -msdata=@var{sdata-type} @gol
576 -mno-flush-func -mflush-func=@var{name} @gol
577 -mno-flush-trap -mflush-trap=@var{number} @gol
581 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
583 @emph{M680x0 Options}
584 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
585 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
586 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
587 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
588 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
589 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
590 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
591 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
593 @emph{M68hc1x Options}
594 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
595 -mauto-incdec -minmax -mlong-calls -mshort @gol
596 -msoft-reg-count=@var{count}}
599 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
600 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
601 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
602 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
603 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
606 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
607 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
608 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
609 -mshared -mno-shared -mxgot -mno-xgot -mgp32 -mgp64 @gol
610 -mfp32 -mfp64 -mhard-float -msoft-float @gol
611 -msingle-float -mdouble-float -mdsp -mpaired-single -mips3d @gol
612 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
613 -G@var{num} -membedded-data -mno-embedded-data @gol
614 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
615 -msplit-addresses -mno-split-addresses @gol
616 -mexplicit-relocs -mno-explicit-relocs @gol
617 -mcheck-zero-division -mno-check-zero-division @gol
618 -mdivide-traps -mdivide-breaks @gol
619 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
620 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
621 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
622 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 @gol
623 -mfix-sb1 -mno-fix-sb1 @gol
624 -mflush-func=@var{func} -mno-flush-func @gol
625 -mbranch-likely -mno-branch-likely @gol
626 -mfp-exceptions -mno-fp-exceptions @gol
627 -mvr4130-align -mno-vr4130-align}
630 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
631 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
632 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
633 -mno-base-addresses -msingle-exit -mno-single-exit}
635 @emph{MN10300 Options}
636 @gccoptlist{-mmult-bug -mno-mult-bug @gol
637 -mam33 -mno-am33 @gol
638 -mam33-2 -mno-am33-2 @gol
639 -mreturn-pointer-on-d0 @gol
643 @gccoptlist{-mno-crt0 -mbacc -msim @gol
644 -march=@var{cpu-type} }
646 @emph{PDP-11 Options}
647 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
648 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
649 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
650 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
651 -mbranch-expensive -mbranch-cheap @gol
652 -msplit -mno-split -munix-asm -mdec-asm}
654 @emph{PowerPC Options}
655 See RS/6000 and PowerPC Options.
657 @emph{RS/6000 and PowerPC Options}
658 @gccoptlist{-mcpu=@var{cpu-type} @gol
659 -mtune=@var{cpu-type} @gol
660 -mpower -mno-power -mpower2 -mno-power2 @gol
661 -mpowerpc -mpowerpc64 -mno-powerpc @gol
662 -maltivec -mno-altivec @gol
663 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
664 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
665 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
666 -mmfpgpr -mno-mfpgpr @gol
667 -mnew-mnemonics -mold-mnemonics @gol
668 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
669 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
670 -malign-power -malign-natural @gol
671 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
672 -mstring -mno-string -mupdate -mno-update @gol
673 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
674 -mstrict-align -mno-strict-align -mrelocatable @gol
675 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
676 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
677 -mdynamic-no-pic -maltivec -mswdiv @gol
678 -mprioritize-restricted-insns=@var{priority} @gol
679 -msched-costly-dep=@var{dependence_type} @gol
680 -minsert-sched-nops=@var{scheme} @gol
681 -mcall-sysv -mcall-netbsd @gol
682 -maix-struct-return -msvr4-struct-return @gol
683 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
684 -misel -mno-isel @gol
685 -misel=yes -misel=no @gol
687 -mspe=yes -mspe=no @gol
688 -mvrsave -mno-vrsave @gol
689 -mmulhw -mno-mulhw @gol
690 -mdlmzb -mno-dlmzb @gol
691 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
692 -mprototype -mno-prototype @gol
693 -msim -mmvme -mads -myellowknife -memb -msdata @gol
694 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
696 @emph{S/390 and zSeries Options}
697 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
698 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
699 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
700 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
701 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
702 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
703 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
706 @gccoptlist{-mel -mel @gol
711 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
712 -m4-nofpu -m4-single-only -m4-single -m4 @gol
713 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
714 -m5-64media -m5-64media-nofpu @gol
715 -m5-32media -m5-32media-nofpu @gol
716 -m5-compact -m5-compact-nofpu @gol
717 -mb -ml -mdalign -mrelax @gol
718 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
719 -mieee -misize -minline-ic_invalidate -mpadstruct -mspace @gol
720 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
721 -mdivsi3_libfunc=@var{name} @gol
722 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
726 @gccoptlist{-mcpu=@var{cpu-type} @gol
727 -mtune=@var{cpu-type} @gol
728 -mcmodel=@var{code-model} @gol
729 -m32 -m64 -mapp-regs -mno-app-regs @gol
730 -mfaster-structs -mno-faster-structs @gol
731 -mfpu -mno-fpu -mhard-float -msoft-float @gol
732 -mhard-quad-float -msoft-quad-float @gol
733 -mimpure-text -mno-impure-text -mlittle-endian @gol
734 -mstack-bias -mno-stack-bias @gol
735 -munaligned-doubles -mno-unaligned-doubles @gol
736 -mv8plus -mno-v8plus -mvis -mno-vis
737 -threads -pthreads -pthread}
740 @gccoptlist{-mwarn-reloc -merror-reloc @gol
741 -msafe-dma -munsafe-dma @gol
743 -msmall-mem -mlarge-mem -mstdmain @gol
744 -mfixed-range=@var{register-range}}
746 @emph{System V Options}
747 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
749 @emph{TMS320C3x/C4x Options}
750 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
751 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
752 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
753 -mparallel-insns -mparallel-mpy -mpreserve-float}
756 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
757 -mprolog-function -mno-prolog-function -mspace @gol
758 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
759 -mapp-regs -mno-app-regs @gol
760 -mdisable-callt -mno-disable-callt @gol
766 @gccoptlist{-mg -mgnu -munix}
768 @emph{x86-64 Options}
769 See i386 and x86-64 Options.
771 @emph{Xstormy16 Options}
774 @emph{Xtensa Options}
775 @gccoptlist{-mconst16 -mno-const16 @gol
776 -mfused-madd -mno-fused-madd @gol
777 -mtext-section-literals -mno-text-section-literals @gol
778 -mtarget-align -mno-target-align @gol
779 -mlongcalls -mno-longcalls}
781 @emph{zSeries Options}
782 See S/390 and zSeries Options.
784 @item Code Generation Options
785 @xref{Code Gen Options,,Options for Code Generation Conventions}.
786 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
787 -ffixed-@var{reg} -fexceptions @gol
788 -fnon-call-exceptions -funwind-tables @gol
789 -fasynchronous-unwind-tables @gol
790 -finhibit-size-directive -finstrument-functions @gol
791 -fno-common -fno-ident @gol
792 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
793 -fno-jump-tables @gol
794 -frecord-gcc-switches @gol
795 -freg-struct-return -fshort-enums @gol
796 -fshort-double -fshort-wchar @gol
797 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
798 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
799 -fargument-alias -fargument-noalias @gol
800 -fargument-noalias-global -fargument-noalias-anything
801 -fleading-underscore -ftls-model=@var{model} @gol
802 -ftrapv -fwrapv -fbounds-check @gol
807 * Overall Options:: Controlling the kind of output:
808 an executable, object files, assembler files,
809 or preprocessed source.
810 * C Dialect Options:: Controlling the variant of C language compiled.
811 * C++ Dialect Options:: Variations on C++.
812 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
814 * Language Independent Options:: Controlling how diagnostics should be
816 * Warning Options:: How picky should the compiler be?
817 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
818 * Optimize Options:: How much optimization?
819 * Preprocessor Options:: Controlling header files and macro definitions.
820 Also, getting dependency information for Make.
821 * Assembler Options:: Passing options to the assembler.
822 * Link Options:: Specifying libraries and so on.
823 * Directory Options:: Where to find header files and libraries.
824 Where to find the compiler executable files.
825 * Spec Files:: How to pass switches to sub-processes.
826 * Target Options:: Running a cross-compiler, or an old version of GCC.
829 @node Overall Options
830 @section Options Controlling the Kind of Output
832 Compilation can involve up to four stages: preprocessing, compilation
833 proper, assembly and linking, always in that order. GCC is capable of
834 preprocessing and compiling several files either into several
835 assembler input files, or into one assembler input file; then each
836 assembler input file produces an object file, and linking combines all
837 the object files (those newly compiled, and those specified as input)
838 into an executable file.
840 @cindex file name suffix
841 For any given input file, the file name suffix determines what kind of
846 C source code which must be preprocessed.
849 C source code which should not be preprocessed.
852 C++ source code which should not be preprocessed.
855 Objective-C source code. Note that you must link with the @file{libobjc}
856 library to make an Objective-C program work.
859 Objective-C source code which should not be preprocessed.
863 Objective-C++ source code. Note that you must link with the @file{libobjc}
864 library to make an Objective-C++ program work. Note that @samp{.M} refers
865 to a literal capital M@.
868 Objective-C++ source code which should not be preprocessed.
871 C, C++, Objective-C or Objective-C++ header file to be turned into a
876 @itemx @var{file}.cxx
877 @itemx @var{file}.cpp
878 @itemx @var{file}.CPP
879 @itemx @var{file}.c++
881 C++ source code which must be preprocessed. Note that in @samp{.cxx},
882 the last two letters must both be literally @samp{x}. Likewise,
883 @samp{.C} refers to a literal capital C@.
887 Objective-C++ source code which must be preprocessed.
890 Objective-C++ source code which should not be preprocessed.
894 C++ header file to be turned into a precompiled header.
897 @itemx @var{file}.for
898 @itemx @var{file}.FOR
899 Fixed form Fortran source code which should not be preprocessed.
902 @itemx @var{file}.fpp
903 @itemx @var{file}.FPP
904 Fixed form Fortran source code which must be preprocessed (with the traditional
908 @itemx @var{file}.f95
909 Free form Fortran source code which should not be preprocessed.
912 @itemx @var{file}.F95
913 Free form Fortran source code which must be preprocessed (with the
914 traditional preprocessor).
916 @c FIXME: Descriptions of Java file types.
923 Ada source code file which contains a library unit declaration (a
924 declaration of a package, subprogram, or generic, or a generic
925 instantiation), or a library unit renaming declaration (a package,
926 generic, or subprogram renaming declaration). Such files are also
929 @itemx @var{file}.adb
930 Ada source code file containing a library unit body (a subprogram or
931 package body). Such files are also called @dfn{bodies}.
933 @c GCC also knows about some suffixes for languages not yet included:
944 Assembler code which must be preprocessed.
947 An object file to be fed straight into linking.
948 Any file name with no recognized suffix is treated this way.
952 You can specify the input language explicitly with the @option{-x} option:
955 @item -x @var{language}
956 Specify explicitly the @var{language} for the following input files
957 (rather than letting the compiler choose a default based on the file
958 name suffix). This option applies to all following input files until
959 the next @option{-x} option. Possible values for @var{language} are:
961 c c-header c-cpp-output
962 c++ c++-header c++-cpp-output
963 objective-c objective-c-header objective-c-cpp-output
964 objective-c++ objective-c++-header objective-c++-cpp-output
965 assembler assembler-with-cpp
973 Turn off any specification of a language, so that subsequent files are
974 handled according to their file name suffixes (as they are if @option{-x}
975 has not been used at all).
977 @item -pass-exit-codes
978 @opindex pass-exit-codes
979 Normally the @command{gcc} program will exit with the code of 1 if any
980 phase of the compiler returns a non-success return code. If you specify
981 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
982 numerically highest error produced by any phase that returned an error
983 indication. The C, C++, and Fortran frontends return 4, if an internal
984 compiler error is encountered.
987 If you only want some of the stages of compilation, you can use
988 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
989 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
990 @command{gcc} is to stop. Note that some combinations (for example,
991 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
996 Compile or assemble the source files, but do not link. The linking
997 stage simply is not done. The ultimate output is in the form of an
998 object file for each source file.
1000 By default, the object file name for a source file is made by replacing
1001 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1003 Unrecognized input files, not requiring compilation or assembly, are
1008 Stop after the stage of compilation proper; do not assemble. The output
1009 is in the form of an assembler code file for each non-assembler input
1012 By default, the assembler file name for a source file is made by
1013 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1015 Input files that don't require compilation are ignored.
1019 Stop after the preprocessing stage; do not run the compiler proper. The
1020 output is in the form of preprocessed source code, which is sent to the
1023 Input files which don't require preprocessing are ignored.
1025 @cindex output file option
1028 Place output in file @var{file}. This applies regardless to whatever
1029 sort of output is being produced, whether it be an executable file,
1030 an object file, an assembler file or preprocessed C code.
1032 If @option{-o} is not specified, the default is to put an executable
1033 file in @file{a.out}, the object file for
1034 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1035 assembler file in @file{@var{source}.s}, a precompiled header file in
1036 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1041 Print (on standard error output) the commands executed to run the stages
1042 of compilation. Also print the version number of the compiler driver
1043 program and of the preprocessor and the compiler proper.
1047 Like @option{-v} except the commands are not executed and all command
1048 arguments are quoted. This is useful for shell scripts to capture the
1049 driver-generated command lines.
1053 Use pipes rather than temporary files for communication between the
1054 various stages of compilation. This fails to work on some systems where
1055 the assembler is unable to read from a pipe; but the GNU assembler has
1060 If you are compiling multiple source files, this option tells the driver
1061 to pass all the source files to the compiler at once (for those
1062 languages for which the compiler can handle this). This will allow
1063 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1064 language for which this is supported is C@. If you pass source files for
1065 multiple languages to the driver, using this option, the driver will invoke
1066 the compiler(s) that support IMA once each, passing each compiler all the
1067 source files appropriate for it. For those languages that do not support
1068 IMA this option will be ignored, and the compiler will be invoked once for
1069 each source file in that language. If you use this option in conjunction
1070 with @option{-save-temps}, the compiler will generate multiple
1072 (one for each source file), but only one (combined) @file{.o} or
1077 Print (on the standard output) a description of the command line options
1078 understood by @command{gcc}. If the @option{-v} option is also specified
1079 then @option{--help} will also be passed on to the various processes
1080 invoked by @command{gcc}, so that they can display the command line options
1081 they accept. If the @option{-Wextra} option is also specified then command
1082 line options which have no documentation associated with them will also
1086 @opindex target-help
1087 Print (on the standard output) a description of target specific command
1088 line options for each tool.
1092 Display the version number and copyrights of the invoked GCC@.
1094 @include @value{srcdir}/../libiberty/at-file.texi
1098 @section Compiling C++ Programs
1100 @cindex suffixes for C++ source
1101 @cindex C++ source file suffixes
1102 C++ source files conventionally use one of the suffixes @samp{.C},
1103 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1104 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1105 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1106 files with these names and compiles them as C++ programs even if you
1107 call the compiler the same way as for compiling C programs (usually
1108 with the name @command{gcc}).
1112 However, the use of @command{gcc} does not add the C++ library.
1113 @command{g++} is a program that calls GCC and treats @samp{.c},
1114 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1115 files unless @option{-x} is used, and automatically specifies linking
1116 against the C++ library. This program is also useful when
1117 precompiling a C header file with a @samp{.h} extension for use in C++
1118 compilations. On many systems, @command{g++} is also installed with
1119 the name @command{c++}.
1121 @cindex invoking @command{g++}
1122 When you compile C++ programs, you may specify many of the same
1123 command-line options that you use for compiling programs in any
1124 language; or command-line options meaningful for C and related
1125 languages; or options that are meaningful only for C++ programs.
1126 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1127 explanations of options for languages related to C@.
1128 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1129 explanations of options that are meaningful only for C++ programs.
1131 @node C Dialect Options
1132 @section Options Controlling C Dialect
1133 @cindex dialect options
1134 @cindex language dialect options
1135 @cindex options, dialect
1137 The following options control the dialect of C (or languages derived
1138 from C, such as C++, Objective-C and Objective-C++) that the compiler
1142 @cindex ANSI support
1146 In C mode, support all ISO C90 programs. In C++ mode,
1147 remove GNU extensions that conflict with ISO C++.
1149 This turns off certain features of GCC that are incompatible with ISO
1150 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1151 such as the @code{asm} and @code{typeof} keywords, and
1152 predefined macros such as @code{unix} and @code{vax} that identify the
1153 type of system you are using. It also enables the undesirable and
1154 rarely used ISO trigraph feature. For the C compiler,
1155 it disables recognition of C++ style @samp{//} comments as well as
1156 the @code{inline} keyword.
1158 The alternate keywords @code{__asm__}, @code{__extension__},
1159 @code{__inline__} and @code{__typeof__} continue to work despite
1160 @option{-ansi}. You would not want to use them in an ISO C program, of
1161 course, but it is useful to put them in header files that might be included
1162 in compilations done with @option{-ansi}. Alternate predefined macros
1163 such as @code{__unix__} and @code{__vax__} are also available, with or
1164 without @option{-ansi}.
1166 The @option{-ansi} option does not cause non-ISO programs to be
1167 rejected gratuitously. For that, @option{-pedantic} is required in
1168 addition to @option{-ansi}. @xref{Warning Options}.
1170 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1171 option is used. Some header files may notice this macro and refrain
1172 from declaring certain functions or defining certain macros that the
1173 ISO standard doesn't call for; this is to avoid interfering with any
1174 programs that might use these names for other things.
1176 Functions which would normally be built in but do not have semantics
1177 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1178 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1179 built-in functions provided by GCC}, for details of the functions
1184 Determine the language standard. This option is currently only
1185 supported when compiling C or C++. A value for this option must be
1186 provided; possible values are
1191 ISO C90 (same as @option{-ansi}).
1193 @item iso9899:199409
1194 ISO C90 as modified in amendment 1.
1200 ISO C99. Note that this standard is not yet fully supported; see
1201 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1202 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1205 Default, ISO C90 plus GNU extensions (including some C99 features).
1209 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1210 this will become the default. The name @samp{gnu9x} is deprecated.
1213 The 1998 ISO C++ standard plus amendments.
1216 The same as @option{-std=c++98} plus GNU extensions. This is the
1217 default for C++ code.
1220 The working draft of the upcoming ISO C++0x standard. This option
1221 enables experimental features that are likely to be included in
1222 C++0x. The working draft is constantly changing, and any feature that is
1223 enabled by this flag may be removed from future versions of GCC if it is
1224 not part of the C++0x standard.
1227 The same as @option{-std=c++0x} plus GNU extensions. As with
1228 @option{-std=c++0x}, this option enables experimental features that may
1229 be removed in future versions of GCC.
1232 Even when this option is not specified, you can still use some of the
1233 features of newer standards in so far as they do not conflict with
1234 previous C standards. For example, you may use @code{__restrict__} even
1235 when @option{-std=c99} is not specified.
1237 The @option{-std} options specifying some version of ISO C have the same
1238 effects as @option{-ansi}, except that features that were not in ISO C90
1239 but are in the specified version (for example, @samp{//} comments and
1240 the @code{inline} keyword in ISO C99) are not disabled.
1242 @xref{Standards,,Language Standards Supported by GCC}, for details of
1243 these standard versions.
1245 @item -aux-info @var{filename}
1247 Output to the given filename prototyped declarations for all functions
1248 declared and/or defined in a translation unit, including those in header
1249 files. This option is silently ignored in any language other than C@.
1251 Besides declarations, the file indicates, in comments, the origin of
1252 each declaration (source file and line), whether the declaration was
1253 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1254 @samp{O} for old, respectively, in the first character after the line
1255 number and the colon), and whether it came from a declaration or a
1256 definition (@samp{C} or @samp{F}, respectively, in the following
1257 character). In the case of function definitions, a K&R-style list of
1258 arguments followed by their declarations is also provided, inside
1259 comments, after the declaration.
1263 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1264 keyword, so that code can use these words as identifiers. You can use
1265 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1266 instead. @option{-ansi} implies @option{-fno-asm}.
1268 In C++, this switch only affects the @code{typeof} keyword, since
1269 @code{asm} and @code{inline} are standard keywords. You may want to
1270 use the @option{-fno-gnu-keywords} flag instead, which has the same
1271 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1272 switch only affects the @code{asm} and @code{typeof} keywords, since
1273 @code{inline} is a standard keyword in ISO C99.
1276 @itemx -fno-builtin-@var{function}
1277 @opindex fno-builtin
1278 @cindex built-in functions
1279 Don't recognize built-in functions that do not begin with
1280 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1281 functions provided by GCC}, for details of the functions affected,
1282 including those which are not built-in functions when @option{-ansi} or
1283 @option{-std} options for strict ISO C conformance are used because they
1284 do not have an ISO standard meaning.
1286 GCC normally generates special code to handle certain built-in functions
1287 more efficiently; for instance, calls to @code{alloca} may become single
1288 instructions that adjust the stack directly, and calls to @code{memcpy}
1289 may become inline copy loops. The resulting code is often both smaller
1290 and faster, but since the function calls no longer appear as such, you
1291 cannot set a breakpoint on those calls, nor can you change the behavior
1292 of the functions by linking with a different library. In addition,
1293 when a function is recognized as a built-in function, GCC may use
1294 information about that function to warn about problems with calls to
1295 that function, or to generate more efficient code, even if the
1296 resulting code still contains calls to that function. For example,
1297 warnings are given with @option{-Wformat} for bad calls to
1298 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1299 known not to modify global memory.
1301 With the @option{-fno-builtin-@var{function}} option
1302 only the built-in function @var{function} is
1303 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1304 function is named this is not built-in in this version of GCC, this
1305 option is ignored. There is no corresponding
1306 @option{-fbuiltin-@var{function}} option; if you wish to enable
1307 built-in functions selectively when using @option{-fno-builtin} or
1308 @option{-ffreestanding}, you may define macros such as:
1311 #define abs(n) __builtin_abs ((n))
1312 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1317 @cindex hosted environment
1319 Assert that compilation takes place in a hosted environment. This implies
1320 @option{-fbuiltin}. A hosted environment is one in which the
1321 entire standard library is available, and in which @code{main} has a return
1322 type of @code{int}. Examples are nearly everything except a kernel.
1323 This is equivalent to @option{-fno-freestanding}.
1325 @item -ffreestanding
1326 @opindex ffreestanding
1327 @cindex hosted environment
1329 Assert that compilation takes place in a freestanding environment. This
1330 implies @option{-fno-builtin}. A freestanding environment
1331 is one in which the standard library may not exist, and program startup may
1332 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1333 This is equivalent to @option{-fno-hosted}.
1335 @xref{Standards,,Language Standards Supported by GCC}, for details of
1336 freestanding and hosted environments.
1340 @cindex openmp parallel
1341 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1342 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1343 compiler generates parallel code according to the OpenMP Application
1344 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.
1346 @item -fms-extensions
1347 @opindex fms-extensions
1348 Accept some non-standard constructs used in Microsoft header files.
1350 Some cases of unnamed fields in structures and unions are only
1351 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1352 fields within structs/unions}, for details.
1356 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1357 options for strict ISO C conformance) implies @option{-trigraphs}.
1359 @item -no-integrated-cpp
1360 @opindex no-integrated-cpp
1361 Performs a compilation in two passes: preprocessing and compiling. This
1362 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1363 @option{-B} option. The user supplied compilation step can then add in
1364 an additional preprocessing step after normal preprocessing but before
1365 compiling. The default is to use the integrated cpp (internal cpp)
1367 The semantics of this option will change if "cc1", "cc1plus", and
1368 "cc1obj" are merged.
1370 @cindex traditional C language
1371 @cindex C language, traditional
1373 @itemx -traditional-cpp
1374 @opindex traditional-cpp
1375 @opindex traditional
1376 Formerly, these options caused GCC to attempt to emulate a pre-standard
1377 C compiler. They are now only supported with the @option{-E} switch.
1378 The preprocessor continues to support a pre-standard mode. See the GNU
1379 CPP manual for details.
1381 @item -fcond-mismatch
1382 @opindex fcond-mismatch
1383 Allow conditional expressions with mismatched types in the second and
1384 third arguments. The value of such an expression is void. This option
1385 is not supported for C++.
1387 @item -flax-vector-conversions
1388 @opindex flax-vector-conversions
1389 Allow implicit conversions between vectors with differing numbers of
1390 elements and/or incompatible element types. This option should not be
1393 @item -funsigned-char
1394 @opindex funsigned-char
1395 Let the type @code{char} be unsigned, like @code{unsigned char}.
1397 Each kind of machine has a default for what @code{char} should
1398 be. It is either like @code{unsigned char} by default or like
1399 @code{signed char} by default.
1401 Ideally, a portable program should always use @code{signed char} or
1402 @code{unsigned char} when it depends on the signedness of an object.
1403 But many programs have been written to use plain @code{char} and
1404 expect it to be signed, or expect it to be unsigned, depending on the
1405 machines they were written for. This option, and its inverse, let you
1406 make such a program work with the opposite default.
1408 The type @code{char} is always a distinct type from each of
1409 @code{signed char} or @code{unsigned char}, even though its behavior
1410 is always just like one of those two.
1413 @opindex fsigned-char
1414 Let the type @code{char} be signed, like @code{signed char}.
1416 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1417 the negative form of @option{-funsigned-char}. Likewise, the option
1418 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1420 @item -fsigned-bitfields
1421 @itemx -funsigned-bitfields
1422 @itemx -fno-signed-bitfields
1423 @itemx -fno-unsigned-bitfields
1424 @opindex fsigned-bitfields
1425 @opindex funsigned-bitfields
1426 @opindex fno-signed-bitfields
1427 @opindex fno-unsigned-bitfields
1428 These options control whether a bit-field is signed or unsigned, when the
1429 declaration does not use either @code{signed} or @code{unsigned}. By
1430 default, such a bit-field is signed, because this is consistent: the
1431 basic integer types such as @code{int} are signed types.
1434 @node C++ Dialect Options
1435 @section Options Controlling C++ Dialect
1437 @cindex compiler options, C++
1438 @cindex C++ options, command line
1439 @cindex options, C++
1440 This section describes the command-line options that are only meaningful
1441 for C++ programs; but you can also use most of the GNU compiler options
1442 regardless of what language your program is in. For example, you
1443 might compile a file @code{firstClass.C} like this:
1446 g++ -g -frepo -O -c firstClass.C
1450 In this example, only @option{-frepo} is an option meant
1451 only for C++ programs; you can use the other options with any
1452 language supported by GCC@.
1454 Here is a list of options that are @emph{only} for compiling C++ programs:
1458 @item -fabi-version=@var{n}
1459 @opindex fabi-version
1460 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1461 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1462 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1463 the version that conforms most closely to the C++ ABI specification.
1464 Therefore, the ABI obtained using version 0 will change as ABI bugs
1467 The default is version 2.
1469 @item -fno-access-control
1470 @opindex fno-access-control
1471 Turn off all access checking. This switch is mainly useful for working
1472 around bugs in the access control code.
1476 Check that the pointer returned by @code{operator new} is non-null
1477 before attempting to modify the storage allocated. This check is
1478 normally unnecessary because the C++ standard specifies that
1479 @code{operator new} will only return @code{0} if it is declared
1480 @samp{throw()}, in which case the compiler will always check the
1481 return value even without this option. In all other cases, when
1482 @code{operator new} has a non-empty exception specification, memory
1483 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1484 @samp{new (nothrow)}.
1486 @item -fconserve-space
1487 @opindex fconserve-space
1488 Put uninitialized or runtime-initialized global variables into the
1489 common segment, as C does. This saves space in the executable at the
1490 cost of not diagnosing duplicate definitions. If you compile with this
1491 flag and your program mysteriously crashes after @code{main()} has
1492 completed, you may have an object that is being destroyed twice because
1493 two definitions were merged.
1495 This option is no longer useful on most targets, now that support has
1496 been added for putting variables into BSS without making them common.
1498 @item -ffriend-injection
1499 @opindex ffriend-injection
1500 Inject friend functions into the enclosing namespace, so that they are
1501 visible outside the scope of the class in which they are declared.
1502 Friend functions were documented to work this way in the old Annotated
1503 C++ Reference Manual, and versions of G++ before 4.1 always worked
1504 that way. However, in ISO C++ a friend function which is not declared
1505 in an enclosing scope can only be found using argument dependent
1506 lookup. This option causes friends to be injected as they were in
1509 This option is for compatibility, and may be removed in a future
1512 @item -fno-elide-constructors
1513 @opindex fno-elide-constructors
1514 The C++ standard allows an implementation to omit creating a temporary
1515 which is only used to initialize another object of the same type.
1516 Specifying this option disables that optimization, and forces G++ to
1517 call the copy constructor in all cases.
1519 @item -fno-enforce-eh-specs
1520 @opindex fno-enforce-eh-specs
1521 Don't generate code to check for violation of exception specifications
1522 at runtime. This option violates the C++ standard, but may be useful
1523 for reducing code size in production builds, much like defining
1524 @samp{NDEBUG}. This does not give user code permission to throw
1525 exceptions in violation of the exception specifications; the compiler
1526 will still optimize based on the specifications, so throwing an
1527 unexpected exception will result in undefined behavior.
1530 @itemx -fno-for-scope
1532 @opindex fno-for-scope
1533 If @option{-ffor-scope} is specified, the scope of variables declared in
1534 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1535 as specified by the C++ standard.
1536 If @option{-fno-for-scope} is specified, the scope of variables declared in
1537 a @i{for-init-statement} extends to the end of the enclosing scope,
1538 as was the case in old versions of G++, and other (traditional)
1539 implementations of C++.
1541 The default if neither flag is given to follow the standard,
1542 but to allow and give a warning for old-style code that would
1543 otherwise be invalid, or have different behavior.
1545 @item -fno-gnu-keywords
1546 @opindex fno-gnu-keywords
1547 Do not recognize @code{typeof} as a keyword, so that code can use this
1548 word as an identifier. You can use the keyword @code{__typeof__} instead.
1549 @option{-ansi} implies @option{-fno-gnu-keywords}.
1551 @item -fno-implicit-templates
1552 @opindex fno-implicit-templates
1553 Never emit code for non-inline templates which are instantiated
1554 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1555 @xref{Template Instantiation}, for more information.
1557 @item -fno-implicit-inline-templates
1558 @opindex fno-implicit-inline-templates
1559 Don't emit code for implicit instantiations of inline templates, either.
1560 The default is to handle inlines differently so that compiles with and
1561 without optimization will need the same set of explicit instantiations.
1563 @item -fno-implement-inlines
1564 @opindex fno-implement-inlines
1565 To save space, do not emit out-of-line copies of inline functions
1566 controlled by @samp{#pragma implementation}. This will cause linker
1567 errors if these functions are not inlined everywhere they are called.
1569 @item -fms-extensions
1570 @opindex fms-extensions
1571 Disable pedantic warnings about constructs used in MFC, such as implicit
1572 int and getting a pointer to member function via non-standard syntax.
1574 @item -fno-nonansi-builtins
1575 @opindex fno-nonansi-builtins
1576 Disable built-in declarations of functions that are not mandated by
1577 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1578 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1580 @item -fno-operator-names
1581 @opindex fno-operator-names
1582 Do not treat the operator name keywords @code{and}, @code{bitand},
1583 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1584 synonyms as keywords.
1586 @item -fno-optional-diags
1587 @opindex fno-optional-diags
1588 Disable diagnostics that the standard says a compiler does not need to
1589 issue. Currently, the only such diagnostic issued by G++ is the one for
1590 a name having multiple meanings within a class.
1593 @opindex fpermissive
1594 Downgrade some diagnostics about nonconformant code from errors to
1595 warnings. Thus, using @option{-fpermissive} will allow some
1596 nonconforming code to compile.
1600 Enable automatic template instantiation at link time. This option also
1601 implies @option{-fno-implicit-templates}. @xref{Template
1602 Instantiation}, for more information.
1606 Disable generation of information about every class with virtual
1607 functions for use by the C++ runtime type identification features
1608 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1609 of the language, you can save some space by using this flag. Note that
1610 exception handling uses the same information, but it will generate it as
1611 needed. The @samp{dynamic_cast} operator can still be used for casts that
1612 do not require runtime type information, i.e. casts to @code{void *} or to
1613 unambiguous base classes.
1617 Emit statistics about front-end processing at the end of the compilation.
1618 This information is generally only useful to the G++ development team.
1620 @item -ftemplate-depth-@var{n}
1621 @opindex ftemplate-depth
1622 Set the maximum instantiation depth for template classes to @var{n}.
1623 A limit on the template instantiation depth is needed to detect
1624 endless recursions during template class instantiation. ANSI/ISO C++
1625 conforming programs must not rely on a maximum depth greater than 17.
1627 @item -fno-threadsafe-statics
1628 @opindex fno-threadsafe-statics
1629 Do not emit the extra code to use the routines specified in the C++
1630 ABI for thread-safe initialization of local statics. You can use this
1631 option to reduce code size slightly in code that doesn't need to be
1634 @item -fuse-cxa-atexit
1635 @opindex fuse-cxa-atexit
1636 Register destructors for objects with static storage duration with the
1637 @code{__cxa_atexit} function rather than the @code{atexit} function.
1638 This option is required for fully standards-compliant handling of static
1639 destructors, but will only work if your C library supports
1640 @code{__cxa_atexit}.
1642 @item -fno-use-cxa-get-exception-ptr
1643 @opindex fno-use-cxa-get-exception-ptr
1644 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1645 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1646 if the runtime routine is not available.
1648 @item -fvisibility-inlines-hidden
1649 @opindex fvisibility-inlines-hidden
1650 This switch declares that the user does not attempt to compare
1651 pointers to inline methods where the addresses of the two functions
1652 were taken in different shared objects.
1654 The effect of this is that GCC may, effectively, mark inline methods with
1655 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1656 appear in the export table of a DSO and do not require a PLT indirection
1657 when used within the DSO@. Enabling this option can have a dramatic effect
1658 on load and link times of a DSO as it massively reduces the size of the
1659 dynamic export table when the library makes heavy use of templates.
1661 The behaviour of this switch is not quite the same as marking the
1662 methods as hidden directly, because it does not affect static variables
1663 local to the function or cause the compiler to deduce that
1664 the function is defined in only one shared object.
1666 You may mark a method as having a visibility explicitly to negate the
1667 effect of the switch for that method. For example, if you do want to
1668 compare pointers to a particular inline method, you might mark it as
1669 having default visibility. Marking the enclosing class with explicit
1670 visibility will have no effect.
1672 Explicitly instantiated inline methods are unaffected by this option
1673 as their linkage might otherwise cross a shared library boundary.
1674 @xref{Template Instantiation}.
1678 Do not use weak symbol support, even if it is provided by the linker.
1679 By default, G++ will use weak symbols if they are available. This
1680 option exists only for testing, and should not be used by end-users;
1681 it will result in inferior code and has no benefits. This option may
1682 be removed in a future release of G++.
1686 Do not search for header files in the standard directories specific to
1687 C++, but do still search the other standard directories. (This option
1688 is used when building the C++ library.)
1691 In addition, these optimization, warning, and code generation options
1692 have meanings only for C++ programs:
1695 @item -fno-default-inline
1696 @opindex fno-default-inline
1697 Do not assume @samp{inline} for functions defined inside a class scope.
1698 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1699 functions will have linkage like inline functions; they just won't be
1702 @item -Wabi @r{(C++ only)}
1704 Warn when G++ generates code that is probably not compatible with the
1705 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1706 all such cases, there are probably some cases that are not warned about,
1707 even though G++ is generating incompatible code. There may also be
1708 cases where warnings are emitted even though the code that is generated
1711 You should rewrite your code to avoid these warnings if you are
1712 concerned about the fact that code generated by G++ may not be binary
1713 compatible with code generated by other compilers.
1715 The known incompatibilities at this point include:
1720 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1721 pack data into the same byte as a base class. For example:
1724 struct A @{ virtual void f(); int f1 : 1; @};
1725 struct B : public A @{ int f2 : 1; @};
1729 In this case, G++ will place @code{B::f2} into the same byte
1730 as@code{A::f1}; other compilers will not. You can avoid this problem
1731 by explicitly padding @code{A} so that its size is a multiple of the
1732 byte size on your platform; that will cause G++ and other compilers to
1733 layout @code{B} identically.
1736 Incorrect handling of tail-padding for virtual bases. G++ does not use
1737 tail padding when laying out virtual bases. For example:
1740 struct A @{ virtual void f(); char c1; @};
1741 struct B @{ B(); char c2; @};
1742 struct C : public A, public virtual B @{@};
1746 In this case, G++ will not place @code{B} into the tail-padding for
1747 @code{A}; other compilers will. You can avoid this problem by
1748 explicitly padding @code{A} so that its size is a multiple of its
1749 alignment (ignoring virtual base classes); that will cause G++ and other
1750 compilers to layout @code{C} identically.
1753 Incorrect handling of bit-fields with declared widths greater than that
1754 of their underlying types, when the bit-fields appear in a union. For
1758 union U @{ int i : 4096; @};
1762 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1763 union too small by the number of bits in an @code{int}.
1766 Empty classes can be placed at incorrect offsets. For example:
1776 struct C : public B, public A @{@};
1780 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1781 it should be placed at offset zero. G++ mistakenly believes that the
1782 @code{A} data member of @code{B} is already at offset zero.
1785 Names of template functions whose types involve @code{typename} or
1786 template template parameters can be mangled incorrectly.
1789 template <typename Q>
1790 void f(typename Q::X) @{@}
1792 template <template <typename> class Q>
1793 void f(typename Q<int>::X) @{@}
1797 Instantiations of these templates may be mangled incorrectly.
1801 @item -Wctor-dtor-privacy @r{(C++ only)}
1802 @opindex Wctor-dtor-privacy
1803 Warn when a class seems unusable because all the constructors or
1804 destructors in that class are private, and it has neither friends nor
1805 public static member functions.
1807 @item -Wnon-virtual-dtor @r{(C++ only)}
1808 @opindex Wnon-virtual-dtor
1809 Warn when a class appears to be polymorphic, thereby requiring a virtual
1810 destructor, yet it declares a non-virtual one. This warning is also
1811 enabled if -Weffc++ is specified.
1813 @item -Wreorder @r{(C++ only)}
1815 @cindex reordering, warning
1816 @cindex warning for reordering of member initializers
1817 Warn when the order of member initializers given in the code does not
1818 match the order in which they must be executed. For instance:
1824 A(): j (0), i (1) @{ @}
1828 The compiler will rearrange the member initializers for @samp{i}
1829 and @samp{j} to match the declaration order of the members, emitting
1830 a warning to that effect. This warning is enabled by @option{-Wall}.
1833 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1836 @item -Weffc++ @r{(C++ only)}
1838 Warn about violations of the following style guidelines from Scott Meyers'
1839 @cite{Effective C++} book:
1843 Item 11: Define a copy constructor and an assignment operator for classes
1844 with dynamically allocated memory.
1847 Item 12: Prefer initialization to assignment in constructors.
1850 Item 14: Make destructors virtual in base classes.
1853 Item 15: Have @code{operator=} return a reference to @code{*this}.
1856 Item 23: Don't try to return a reference when you must return an object.
1860 Also warn about violations of the following style guidelines from
1861 Scott Meyers' @cite{More Effective C++} book:
1865 Item 6: Distinguish between prefix and postfix forms of increment and
1866 decrement operators.
1869 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1873 When selecting this option, be aware that the standard library
1874 headers do not obey all of these guidelines; use @samp{grep -v}
1875 to filter out those warnings.
1877 @item -Wno-deprecated @r{(C++ only)}
1878 @opindex Wno-deprecated
1879 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1881 @item -Wstrict-null-sentinel @r{(C++ only)}
1882 @opindex Wstrict-null-sentinel
1883 Warn also about the use of an uncasted @code{NULL} as sentinel. When
1884 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
1885 to @code{__null}. Although it is a null pointer constant not a null pointer,
1886 it is guaranteed to of the same size as a pointer. But this use is
1887 not portable across different compilers.
1889 @item -Wno-non-template-friend @r{(C++ only)}
1890 @opindex Wno-non-template-friend
1891 Disable warnings when non-templatized friend functions are declared
1892 within a template. Since the advent of explicit template specification
1893 support in G++, if the name of the friend is an unqualified-id (i.e.,
1894 @samp{friend foo(int)}), the C++ language specification demands that the
1895 friend declare or define an ordinary, nontemplate function. (Section
1896 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1897 could be interpreted as a particular specialization of a templatized
1898 function. Because this non-conforming behavior is no longer the default
1899 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1900 check existing code for potential trouble spots and is on by default.
1901 This new compiler behavior can be turned off with
1902 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1903 but disables the helpful warning.
1905 @item -Wold-style-cast @r{(C++ only)}
1906 @opindex Wold-style-cast
1907 Warn if an old-style (C-style) cast to a non-void type is used within
1908 a C++ program. The new-style casts (@samp{dynamic_cast},
1909 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
1910 less vulnerable to unintended effects and much easier to search for.
1912 @item -Woverloaded-virtual @r{(C++ only)}
1913 @opindex Woverloaded-virtual
1914 @cindex overloaded virtual fn, warning
1915 @cindex warning for overloaded virtual fn
1916 Warn when a function declaration hides virtual functions from a
1917 base class. For example, in:
1924 struct B: public A @{
1929 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1937 will fail to compile.
1939 @item -Wno-pmf-conversions @r{(C++ only)}
1940 @opindex Wno-pmf-conversions
1941 Disable the diagnostic for converting a bound pointer to member function
1944 @item -Wsign-promo @r{(C++ only)}
1945 @opindex Wsign-promo
1946 Warn when overload resolution chooses a promotion from unsigned or
1947 enumerated type to a signed type, over a conversion to an unsigned type of
1948 the same size. Previous versions of G++ would try to preserve
1949 unsignedness, but the standard mandates the current behavior.
1954 A& operator = (int);
1964 In this example, G++ will synthesize a default @samp{A& operator =
1965 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1968 @node Objective-C and Objective-C++ Dialect Options
1969 @section Options Controlling Objective-C and Objective-C++ Dialects
1971 @cindex compiler options, Objective-C and Objective-C++
1972 @cindex Objective-C and Objective-C++ options, command line
1973 @cindex options, Objective-C and Objective-C++
1974 (NOTE: This manual does not describe the Objective-C and Objective-C++
1975 languages themselves. See @xref{Standards,,Language Standards
1976 Supported by GCC}, for references.)
1978 This section describes the command-line options that are only meaningful
1979 for Objective-C and Objective-C++ programs, but you can also use most of
1980 the language-independent GNU compiler options.
1981 For example, you might compile a file @code{some_class.m} like this:
1984 gcc -g -fgnu-runtime -O -c some_class.m
1988 In this example, @option{-fgnu-runtime} is an option meant only for
1989 Objective-C and Objective-C++ programs; you can use the other options with
1990 any language supported by GCC@.
1992 Note that since Objective-C is an extension of the C language, Objective-C
1993 compilations may also use options specific to the C front-end (e.g.,
1994 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1995 C++-specific options (e.g., @option{-Wabi}).
1997 Here is a list of options that are @emph{only} for compiling Objective-C
1998 and Objective-C++ programs:
2001 @item -fconstant-string-class=@var{class-name}
2002 @opindex fconstant-string-class
2003 Use @var{class-name} as the name of the class to instantiate for each
2004 literal string specified with the syntax @code{@@"@dots{}"}. The default
2005 class name is @code{NXConstantString} if the GNU runtime is being used, and
2006 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2007 @option{-fconstant-cfstrings} option, if also present, will override the
2008 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2009 to be laid out as constant CoreFoundation strings.
2012 @opindex fgnu-runtime
2013 Generate object code compatible with the standard GNU Objective-C
2014 runtime. This is the default for most types of systems.
2016 @item -fnext-runtime
2017 @opindex fnext-runtime
2018 Generate output compatible with the NeXT runtime. This is the default
2019 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2020 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2023 @item -fno-nil-receivers
2024 @opindex fno-nil-receivers
2025 Assume that all Objective-C message dispatches (e.g.,
2026 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2027 is not @code{nil}. This allows for more efficient entry points in the runtime
2028 to be used. Currently, this option is only available in conjunction with
2029 the NeXT runtime on Mac OS X 10.3 and later.
2031 @item -fobjc-call-cxx-cdtors
2032 @opindex fobjc-call-cxx-cdtors
2033 For each Objective-C class, check if any of its instance variables is a
2034 C++ object with a non-trivial default constructor. If so, synthesize a
2035 special @code{- (id) .cxx_construct} instance method that will run
2036 non-trivial default constructors on any such instance variables, in order,
2037 and then return @code{self}. Similarly, check if any instance variable
2038 is a C++ object with a non-trivial destructor, and if so, synthesize a
2039 special @code{- (void) .cxx_destruct} method that will run
2040 all such default destructors, in reverse order.
2042 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2043 thusly generated will only operate on instance variables declared in the
2044 current Objective-C class, and not those inherited from superclasses. It
2045 is the responsibility of the Objective-C runtime to invoke all such methods
2046 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2047 will be invoked by the runtime immediately after a new object
2048 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2049 be invoked immediately before the runtime deallocates an object instance.
2051 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2052 support for invoking the @code{- (id) .cxx_construct} and
2053 @code{- (void) .cxx_destruct} methods.
2055 @item -fobjc-direct-dispatch
2056 @opindex fobjc-direct-dispatch
2057 Allow fast jumps to the message dispatcher. On Darwin this is
2058 accomplished via the comm page.
2060 @item -fobjc-exceptions
2061 @opindex fobjc-exceptions
2062 Enable syntactic support for structured exception handling in Objective-C,
2063 similar to what is offered by C++ and Java. This option is
2064 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2073 @@catch (AnObjCClass *exc) @{
2080 @@catch (AnotherClass *exc) @{
2083 @@catch (id allOthers) @{
2093 The @code{@@throw} statement may appear anywhere in an Objective-C or
2094 Objective-C++ program; when used inside of a @code{@@catch} block, the
2095 @code{@@throw} may appear without an argument (as shown above), in which case
2096 the object caught by the @code{@@catch} will be rethrown.
2098 Note that only (pointers to) Objective-C objects may be thrown and
2099 caught using this scheme. When an object is thrown, it will be caught
2100 by the nearest @code{@@catch} clause capable of handling objects of that type,
2101 analogously to how @code{catch} blocks work in C++ and Java. A
2102 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2103 any and all Objective-C exceptions not caught by previous @code{@@catch}
2106 The @code{@@finally} clause, if present, will be executed upon exit from the
2107 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2108 regardless of whether any exceptions are thrown, caught or rethrown
2109 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2110 of the @code{finally} clause in Java.
2112 There are several caveats to using the new exception mechanism:
2116 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2117 idioms provided by the @code{NSException} class, the new
2118 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2119 systems, due to additional functionality needed in the (NeXT) Objective-C
2123 As mentioned above, the new exceptions do not support handling
2124 types other than Objective-C objects. Furthermore, when used from
2125 Objective-C++, the Objective-C exception model does not interoperate with C++
2126 exceptions at this time. This means you cannot @code{@@throw} an exception
2127 from Objective-C and @code{catch} it in C++, or vice versa
2128 (i.e., @code{throw @dots{} @@catch}).
2131 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2132 blocks for thread-safe execution:
2135 @@synchronized (ObjCClass *guard) @{
2140 Upon entering the @code{@@synchronized} block, a thread of execution shall
2141 first check whether a lock has been placed on the corresponding @code{guard}
2142 object by another thread. If it has, the current thread shall wait until
2143 the other thread relinquishes its lock. Once @code{guard} becomes available,
2144 the current thread will place its own lock on it, execute the code contained in
2145 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2146 making @code{guard} available to other threads).
2148 Unlike Java, Objective-C does not allow for entire methods to be marked
2149 @code{@@synchronized}. Note that throwing exceptions out of
2150 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2151 to be unlocked properly.
2155 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2157 @item -freplace-objc-classes
2158 @opindex freplace-objc-classes
2159 Emit a special marker instructing @command{ld(1)} not to statically link in
2160 the resulting object file, and allow @command{dyld(1)} to load it in at
2161 run time instead. This is used in conjunction with the Fix-and-Continue
2162 debugging mode, where the object file in question may be recompiled and
2163 dynamically reloaded in the course of program execution, without the need
2164 to restart the program itself. Currently, Fix-and-Continue functionality
2165 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2170 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2171 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2172 compile time) with static class references that get initialized at load time,
2173 which improves run-time performance. Specifying the @option{-fzero-link} flag
2174 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2175 to be retained. This is useful in Zero-Link debugging mode, since it allows
2176 for individual class implementations to be modified during program execution.
2180 Dump interface declarations for all classes seen in the source file to a
2181 file named @file{@var{sourcename}.decl}.
2183 @item -Wassign-intercept
2184 @opindex Wassign-intercept
2185 Warn whenever an Objective-C assignment is being intercepted by the
2189 @opindex Wno-protocol
2190 If a class is declared to implement a protocol, a warning is issued for
2191 every method in the protocol that is not implemented by the class. The
2192 default behavior is to issue a warning for every method not explicitly
2193 implemented in the class, even if a method implementation is inherited
2194 from the superclass. If you use the @option{-Wno-protocol} option, then
2195 methods inherited from the superclass are considered to be implemented,
2196 and no warning is issued for them.
2200 Warn if multiple methods of different types for the same selector are
2201 found during compilation. The check is performed on the list of methods
2202 in the final stage of compilation. Additionally, a check is performed
2203 for each selector appearing in a @code{@@selector(@dots{})}
2204 expression, and a corresponding method for that selector has been found
2205 during compilation. Because these checks scan the method table only at
2206 the end of compilation, these warnings are not produced if the final
2207 stage of compilation is not reached, for example because an error is
2208 found during compilation, or because the @option{-fsyntax-only} option is
2211 @item -Wstrict-selector-match
2212 @opindex Wstrict-selector-match
2213 Warn if multiple methods with differing argument and/or return types are
2214 found for a given selector when attempting to send a message using this
2215 selector to a receiver of type @code{id} or @code{Class}. When this flag
2216 is off (which is the default behavior), the compiler will omit such warnings
2217 if any differences found are confined to types which share the same size
2220 @item -Wundeclared-selector
2221 @opindex Wundeclared-selector
2222 Warn if a @code{@@selector(@dots{})} expression referring to an
2223 undeclared selector is found. A selector is considered undeclared if no
2224 method with that name has been declared before the
2225 @code{@@selector(@dots{})} expression, either explicitly in an
2226 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2227 an @code{@@implementation} section. This option always performs its
2228 checks as soon as a @code{@@selector(@dots{})} expression is found,
2229 while @option{-Wselector} only performs its checks in the final stage of
2230 compilation. This also enforces the coding style convention
2231 that methods and selectors must be declared before being used.
2233 @item -print-objc-runtime-info
2234 @opindex print-objc-runtime-info
2235 Generate C header describing the largest structure that is passed by
2240 @node Language Independent Options
2241 @section Options to Control Diagnostic Messages Formatting
2242 @cindex options to control diagnostics formatting
2243 @cindex diagnostic messages
2244 @cindex message formatting
2246 Traditionally, diagnostic messages have been formatted irrespective of
2247 the output device's aspect (e.g.@: its width, @dots{}). The options described
2248 below can be used to control the diagnostic messages formatting
2249 algorithm, e.g.@: how many characters per line, how often source location
2250 information should be reported. Right now, only the C++ front end can
2251 honor these options. However it is expected, in the near future, that
2252 the remaining front ends would be able to digest them correctly.
2255 @item -fmessage-length=@var{n}
2256 @opindex fmessage-length
2257 Try to format error messages so that they fit on lines of about @var{n}
2258 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2259 the front ends supported by GCC@. If @var{n} is zero, then no
2260 line-wrapping will be done; each error message will appear on a single
2263 @opindex fdiagnostics-show-location
2264 @item -fdiagnostics-show-location=once
2265 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2266 reporter to emit @emph{once} source location information; that is, in
2267 case the message is too long to fit on a single physical line and has to
2268 be wrapped, the source location won't be emitted (as prefix) again,
2269 over and over, in subsequent continuation lines. This is the default
2272 @item -fdiagnostics-show-location=every-line
2273 Only meaningful in line-wrapping mode. Instructs the diagnostic
2274 messages reporter to emit the same source location information (as
2275 prefix) for physical lines that result from the process of breaking
2276 a message which is too long to fit on a single line.
2278 @item -fdiagnostics-show-option
2279 @opindex fdiagnostics-show-option
2280 This option instructs the diagnostic machinery to add text to each
2281 diagnostic emitted, which indicates which command line option directly
2282 controls that diagnostic, when such an option is known to the
2283 diagnostic machinery.
2287 @node Warning Options
2288 @section Options to Request or Suppress Warnings
2289 @cindex options to control warnings
2290 @cindex warning messages
2291 @cindex messages, warning
2292 @cindex suppressing warnings
2294 Warnings are diagnostic messages that report constructions which
2295 are not inherently erroneous but which are risky or suggest there
2296 may have been an error.
2298 You can request many specific warnings with options beginning @samp{-W},
2299 for example @option{-Wimplicit} to request warnings on implicit
2300 declarations. Each of these specific warning options also has a
2301 negative form beginning @samp{-Wno-} to turn off warnings;
2302 for example, @option{-Wno-implicit}. This manual lists only one of the
2303 two forms, whichever is not the default.
2305 The following options control the amount and kinds of warnings produced
2306 by GCC; for further, language-specific options also refer to
2307 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2311 @cindex syntax checking
2313 @opindex fsyntax-only
2314 Check the code for syntax errors, but don't do anything beyond that.
2318 Issue all the warnings demanded by strict ISO C and ISO C++;
2319 reject all programs that use forbidden extensions, and some other
2320 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2321 version of the ISO C standard specified by any @option{-std} option used.
2323 Valid ISO C and ISO C++ programs should compile properly with or without
2324 this option (though a rare few will require @option{-ansi} or a
2325 @option{-std} option specifying the required version of ISO C)@. However,
2326 without this option, certain GNU extensions and traditional C and C++
2327 features are supported as well. With this option, they are rejected.
2329 @option{-pedantic} does not cause warning messages for use of the
2330 alternate keywords whose names begin and end with @samp{__}. Pedantic
2331 warnings are also disabled in the expression that follows
2332 @code{__extension__}. However, only system header files should use
2333 these escape routes; application programs should avoid them.
2334 @xref{Alternate Keywords}.
2336 Some users try to use @option{-pedantic} to check programs for strict ISO
2337 C conformance. They soon find that it does not do quite what they want:
2338 it finds some non-ISO practices, but not all---only those for which
2339 ISO C @emph{requires} a diagnostic, and some others for which
2340 diagnostics have been added.
2342 A feature to report any failure to conform to ISO C might be useful in
2343 some instances, but would require considerable additional work and would
2344 be quite different from @option{-pedantic}. We don't have plans to
2345 support such a feature in the near future.
2347 Where the standard specified with @option{-std} represents a GNU
2348 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2349 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2350 extended dialect is based. Warnings from @option{-pedantic} are given
2351 where they are required by the base standard. (It would not make sense
2352 for such warnings to be given only for features not in the specified GNU
2353 C dialect, since by definition the GNU dialects of C include all
2354 features the compiler supports with the given option, and there would be
2355 nothing to warn about.)
2357 @item -pedantic-errors
2358 @opindex pedantic-errors
2359 Like @option{-pedantic}, except that errors are produced rather than
2364 Inhibit all warning messages.
2368 Inhibit warning messages about the use of @samp{#import}.
2370 @item -Wchar-subscripts
2371 @opindex Wchar-subscripts
2372 Warn if an array subscript has type @code{char}. This is a common cause
2373 of error, as programmers often forget that this type is signed on some
2375 This warning is enabled by @option{-Wall}.
2379 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2380 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2381 This warning is enabled by @option{-Wall}.
2383 @item -Wfatal-errors
2384 @opindex Wfatal-errors
2385 This option causes the compiler to abort compilation on the first error
2386 occurred rather than trying to keep going and printing further error
2391 @opindex ffreestanding
2392 @opindex fno-builtin
2393 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2394 the arguments supplied have types appropriate to the format string
2395 specified, and that the conversions specified in the format string make
2396 sense. This includes standard functions, and others specified by format
2397 attributes (@pxref{Function Attributes}), in the @code{printf},
2398 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2399 not in the C standard) families (or other target-specific families).
2400 Which functions are checked without format attributes having been
2401 specified depends on the standard version selected, and such checks of
2402 functions without the attribute specified are disabled by
2403 @option{-ffreestanding} or @option{-fno-builtin}.
2405 The formats are checked against the format features supported by GNU
2406 libc version 2.2. These include all ISO C90 and C99 features, as well
2407 as features from the Single Unix Specification and some BSD and GNU
2408 extensions. Other library implementations may not support all these
2409 features; GCC does not support warning about features that go beyond a
2410 particular library's limitations. However, if @option{-pedantic} is used
2411 with @option{-Wformat}, warnings will be given about format features not
2412 in the selected standard version (but not for @code{strfmon} formats,
2413 since those are not in any version of the C standard). @xref{C Dialect
2414 Options,,Options Controlling C Dialect}.
2416 Since @option{-Wformat} also checks for null format arguments for
2417 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2419 @option{-Wformat} is included in @option{-Wall}. For more control over some
2420 aspects of format checking, the options @option{-Wformat-y2k},
2421 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2422 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2423 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2426 @opindex Wformat-y2k
2427 If @option{-Wformat} is specified, also warn about @code{strftime}
2428 formats which may yield only a two-digit year.
2430 @item -Wno-format-extra-args
2431 @opindex Wno-format-extra-args
2432 If @option{-Wformat} is specified, do not warn about excess arguments to a
2433 @code{printf} or @code{scanf} format function. The C standard specifies
2434 that such arguments are ignored.
2436 Where the unused arguments lie between used arguments that are
2437 specified with @samp{$} operand number specifications, normally
2438 warnings are still given, since the implementation could not know what
2439 type to pass to @code{va_arg} to skip the unused arguments. However,
2440 in the case of @code{scanf} formats, this option will suppress the
2441 warning if the unused arguments are all pointers, since the Single
2442 Unix Specification says that such unused arguments are allowed.
2444 @item -Wno-format-zero-length
2445 @opindex Wno-format-zero-length
2446 If @option{-Wformat} is specified, do not warn about zero-length formats.
2447 The C standard specifies that zero-length formats are allowed.
2449 @item -Wformat-nonliteral
2450 @opindex Wformat-nonliteral
2451 If @option{-Wformat} is specified, also warn if the format string is not a
2452 string literal and so cannot be checked, unless the format function
2453 takes its format arguments as a @code{va_list}.
2455 @item -Wformat-security
2456 @opindex Wformat-security
2457 If @option{-Wformat} is specified, also warn about uses of format
2458 functions that represent possible security problems. At present, this
2459 warns about calls to @code{printf} and @code{scanf} functions where the
2460 format string is not a string literal and there are no format arguments,
2461 as in @code{printf (foo);}. This may be a security hole if the format
2462 string came from untrusted input and contains @samp{%n}. (This is
2463 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2464 in future warnings may be added to @option{-Wformat-security} that are not
2465 included in @option{-Wformat-nonliteral}.)
2469 Enable @option{-Wformat} plus format checks not included in
2470 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2471 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2475 Warn about passing a null pointer for arguments marked as
2476 requiring a non-null value by the @code{nonnull} function attribute.
2478 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2479 can be disabled with the @option{-Wno-nonnull} option.
2481 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2483 Warn about uninitialized variables which are initialized with themselves.
2484 Note this option can only be used with the @option{-Wuninitialized} option,
2485 which in turn only works with @option{-O1} and above.
2487 For example, GCC will warn about @code{i} being uninitialized in the
2488 following snippet only when @option{-Winit-self} has been specified:
2499 @item -Wimplicit-int
2500 @opindex Wimplicit-int
2501 Warn when a declaration does not specify a type.
2502 This warning is enabled by @option{-Wall}.
2504 @item -Wimplicit-function-declaration
2505 @itemx -Werror-implicit-function-declaration
2506 @opindex Wimplicit-function-declaration
2507 @opindex Werror-implicit-function-declaration
2508 Give a warning (or error) whenever a function is used before being
2509 declared. The form @option{-Wno-error-implicit-function-declaration}
2511 This warning is enabled by @option{-Wall} (as a warning, not an error).
2515 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2516 This warning is enabled by @option{-Wall}.
2520 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2521 function with external linkage, returning int, taking either zero
2522 arguments, two, or three arguments of appropriate types.
2523 This warning is enabled by @option{-Wall}.
2525 @item -Wmissing-braces
2526 @opindex Wmissing-braces
2527 Warn if an aggregate or union initializer is not fully bracketed. In
2528 the following example, the initializer for @samp{a} is not fully
2529 bracketed, but that for @samp{b} is fully bracketed.
2532 int a[2][2] = @{ 0, 1, 2, 3 @};
2533 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2536 This warning is enabled by @option{-Wall}.
2538 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2539 @opindex Wmissing-include-dirs
2540 Warn if a user-supplied include directory does not exist.
2543 @opindex Wparentheses
2544 Warn if parentheses are omitted in certain contexts, such
2545 as when there is an assignment in a context where a truth value
2546 is expected, or when operators are nested whose precedence people
2547 often get confused about.
2549 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2550 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2551 interpretation from that of ordinary mathematical notation.
2553 Also warn about constructions where there may be confusion to which
2554 @code{if} statement an @code{else} branch belongs. Here is an example of
2569 In C/C++, every @code{else} branch belongs to the innermost possible
2570 @code{if} statement, which in this example is @code{if (b)}. This is
2571 often not what the programmer expected, as illustrated in the above
2572 example by indentation the programmer chose. When there is the
2573 potential for this confusion, GCC will issue a warning when this flag
2574 is specified. To eliminate the warning, add explicit braces around
2575 the innermost @code{if} statement so there is no way the @code{else}
2576 could belong to the enclosing @code{if}. The resulting code would
2593 This warning is enabled by @option{-Wall}.
2595 @item -Wsequence-point
2596 @opindex Wsequence-point
2597 Warn about code that may have undefined semantics because of violations
2598 of sequence point rules in the C and C++ standards.
2600 The C and C++ standards defines the order in which expressions in a C/C++
2601 program are evaluated in terms of @dfn{sequence points}, which represent
2602 a partial ordering between the execution of parts of the program: those
2603 executed before the sequence point, and those executed after it. These
2604 occur after the evaluation of a full expression (one which is not part
2605 of a larger expression), after the evaluation of the first operand of a
2606 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2607 function is called (but after the evaluation of its arguments and the
2608 expression denoting the called function), and in certain other places.
2609 Other than as expressed by the sequence point rules, the order of
2610 evaluation of subexpressions of an expression is not specified. All
2611 these rules describe only a partial order rather than a total order,
2612 since, for example, if two functions are called within one expression
2613 with no sequence point between them, the order in which the functions
2614 are called is not specified. However, the standards committee have
2615 ruled that function calls do not overlap.
2617 It is not specified when between sequence points modifications to the
2618 values of objects take effect. Programs whose behavior depends on this
2619 have undefined behavior; the C and C++ standards specify that ``Between
2620 the previous and next sequence point an object shall have its stored
2621 value modified at most once by the evaluation of an expression.
2622 Furthermore, the prior value shall be read only to determine the value
2623 to be stored.''. If a program breaks these rules, the results on any
2624 particular implementation are entirely unpredictable.
2626 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2627 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2628 diagnosed by this option, and it may give an occasional false positive
2629 result, but in general it has been found fairly effective at detecting
2630 this sort of problem in programs.
2632 The standard is worded confusingly, therefore there is some debate
2633 over the precise meaning of the sequence point rules in subtle cases.
2634 Links to discussions of the problem, including proposed formal
2635 definitions, may be found on the GCC readings page, at
2636 @w{@uref{http://gcc.gnu.org/readings.html}}.
2638 This warning is enabled by @option{-Wall} for C and C++.
2641 @opindex Wreturn-type
2642 Warn whenever a function is defined with a return-type that defaults to
2643 @code{int}. Also warn about any @code{return} statement with no
2644 return-value in a function whose return-type is not @code{void}.
2646 For C, also warn if the return type of a function has a type qualifier
2647 such as @code{const}. Such a type qualifier has no effect, since the
2648 value returned by a function is not an lvalue. ISO C prohibits
2649 qualified @code{void} return types on function definitions, so such
2650 return types always receive a warning even without this option.
2652 For C++, a function without return type always produces a diagnostic
2653 message, even when @option{-Wno-return-type} is specified. The only
2654 exceptions are @samp{main} and functions defined in system headers.
2656 This warning is enabled by @option{-Wall}.
2660 Warn whenever a @code{switch} statement has an index of enumerated type
2661 and lacks a @code{case} for one or more of the named codes of that
2662 enumeration. (The presence of a @code{default} label prevents this
2663 warning.) @code{case} labels outside the enumeration range also
2664 provoke warnings when this option is used.
2665 This warning is enabled by @option{-Wall}.
2667 @item -Wswitch-default
2668 @opindex Wswitch-switch
2669 Warn whenever a @code{switch} statement does not have a @code{default}
2673 @opindex Wswitch-enum
2674 Warn whenever a @code{switch} statement has an index of enumerated type
2675 and lacks a @code{case} for one or more of the named codes of that
2676 enumeration. @code{case} labels outside the enumeration range also
2677 provoke warnings when this option is used.
2681 Warn if any trigraphs are encountered that might change the meaning of
2682 the program (trigraphs within comments are not warned about).
2683 This warning is enabled by @option{-Wall}.
2685 @item -Wunused-function
2686 @opindex Wunused-function
2687 Warn whenever a static function is declared but not defined or a
2688 non-inline static function is unused.
2689 This warning is enabled by @option{-Wall}.
2691 @item -Wunused-label
2692 @opindex Wunused-label
2693 Warn whenever a label is declared but not used.
2694 This warning is enabled by @option{-Wall}.
2696 To suppress this warning use the @samp{unused} attribute
2697 (@pxref{Variable Attributes}).
2699 @item -Wunused-parameter
2700 @opindex Wunused-parameter
2701 Warn whenever a function parameter is unused aside from its declaration.
2703 To suppress this warning use the @samp{unused} attribute
2704 (@pxref{Variable Attributes}).
2706 @item -Wunused-variable
2707 @opindex Wunused-variable
2708 Warn whenever a local variable or non-constant static variable is unused
2709 aside from its declaration.
2710 This warning is enabled by @option{-Wall}.
2712 To suppress this warning use the @samp{unused} attribute
2713 (@pxref{Variable Attributes}).
2715 @item -Wunused-value
2716 @opindex Wunused-value
2717 Warn whenever a statement computes a result that is explicitly not used.
2718 This warning is enabled by @option{-Wall}.
2720 To suppress this warning cast the expression to @samp{void}.
2724 All the above @option{-Wunused} options combined.
2726 In order to get a warning about an unused function parameter, you must
2727 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2728 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2730 @item -Wuninitialized
2731 @opindex Wuninitialized
2732 Warn if an automatic variable is used without first being initialized or
2733 if a variable may be clobbered by a @code{setjmp} call.
2735 These warnings are possible only in optimizing compilation,
2736 because they require data flow information that is computed only
2737 when optimizing. If you do not specify @option{-O}, you will not get
2738 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
2739 requiring @option{-O}.
2741 If you want to warn about code which uses the uninitialized value of the
2742 variable in its own initializer, use the @option{-Winit-self} option.
2744 These warnings occur for individual uninitialized or clobbered
2745 elements of structure, union or array variables as well as for
2746 variables which are uninitialized or clobbered as a whole. They do
2747 not occur for variables or elements declared @code{volatile}. Because
2748 these warnings depend on optimization, the exact variables or elements
2749 for which there are warnings will depend on the precise optimization
2750 options and version of GCC used.
2752 Note that there may be no warning about a variable that is used only
2753 to compute a value that itself is never used, because such
2754 computations may be deleted by data flow analysis before the warnings
2757 These warnings are made optional because GCC is not smart
2758 enough to see all the reasons why the code might be correct
2759 despite appearing to have an error. Here is one example of how
2780 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2781 always initialized, but GCC doesn't know this. Here is
2782 another common case:
2787 if (change_y) save_y = y, y = new_y;
2789 if (change_y) y = save_y;
2794 This has no bug because @code{save_y} is used only if it is set.
2796 @cindex @code{longjmp} warnings
2797 This option also warns when a non-volatile automatic variable might be
2798 changed by a call to @code{longjmp}. These warnings as well are possible
2799 only in optimizing compilation.
2801 The compiler sees only the calls to @code{setjmp}. It cannot know
2802 where @code{longjmp} will be called; in fact, a signal handler could
2803 call it at any point in the code. As a result, you may get a warning
2804 even when there is in fact no problem because @code{longjmp} cannot
2805 in fact be called at the place which would cause a problem.
2807 Some spurious warnings can be avoided if you declare all the functions
2808 you use that never return as @code{noreturn}. @xref{Function
2811 This warning is enabled by @option{-Wall}.
2813 @item -Wunknown-pragmas
2814 @opindex Wunknown-pragmas
2815 @cindex warning for unknown pragmas
2816 @cindex unknown pragmas, warning
2817 @cindex pragmas, warning of unknown
2818 Warn when a #pragma directive is encountered which is not understood by
2819 GCC@. If this command line option is used, warnings will even be issued
2820 for unknown pragmas in system header files. This is not the case if
2821 the warnings were only enabled by the @option{-Wall} command line option.
2824 @opindex Wno-pragmas
2826 Do not warn about misuses of pragmas, such as incorrect parameters,
2827 invalid syntax, or conflicts between pragmas. See also
2828 @samp{-Wunknown-pragmas}.
2830 @item -Wstrict-aliasing
2831 @opindex Wstrict-aliasing
2832 This option is only active when @option{-fstrict-aliasing} is active.
2833 It warns about code which might break the strict aliasing rules that the
2834 compiler is using for optimization. The warning does not catch all
2835 cases, but does attempt to catch the more common pitfalls. It is
2836 included in @option{-Wall}.
2838 @item -Wstrict-aliasing=2
2839 @opindex Wstrict-aliasing=2
2840 This option is only active when @option{-fstrict-aliasing} is active.
2841 It warns about code which might break the strict aliasing rules that the
2842 compiler is using for optimization. This warning catches more cases than
2843 @option{-Wstrict-aliasing}, but it will also give a warning for some ambiguous
2844 cases that are safe.
2846 @item -Warray-bounds
2847 @opindex Wno-array-bounds
2848 @opindex Warray-bounds
2849 This option is only active when @option{-ftree-vrp} is active
2850 (default for -O2 and above). It warns about subscripts to arrays
2851 that are always out of bounds. This warning is enabled by @option{-Wall}.
2855 All of the above @samp{-W} options combined. This enables all the
2856 warnings about constructions that some users consider questionable, and
2857 that are easy to avoid (or modify to prevent the warning), even in
2858 conjunction with macros. This also enables some language-specific
2859 warnings described in @ref{C++ Dialect Options} and
2860 @ref{Objective-C and Objective-C++ Dialect Options}.
2863 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2864 Some of them warn about constructions that users generally do not
2865 consider questionable, but which occasionally you might wish to check
2866 for; others warn about constructions that are necessary or hard to avoid
2867 in some cases, and there is no simple way to modify the code to suppress
2874 (This option used to be called @option{-W}. The older name is still
2875 supported, but the newer name is more descriptive.) Print extra warning
2876 messages for these events:
2880 A function can return either with or without a value. (Falling
2881 off the end of the function body is considered returning without
2882 a value.) For example, this function would evoke such a
2896 An expression-statement or the left-hand side of a comma expression
2897 contains no side effects.
2898 To suppress the warning, cast the unused expression to void.
2899 For example, an expression such as @samp{x[i,j]} will cause a warning,
2900 but @samp{x[(void)i,j]} will not.
2903 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2906 Storage-class specifiers like @code{static} are not the first things
2907 in a declaration. According to the C Standard, this usage is
2908 obsolescent. This warning can be independently controlled by
2909 @option{-Wold-style-declaration}.
2912 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2916 A comparison between signed and unsigned values could produce an
2917 incorrect result when the signed value is converted to unsigned.
2918 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2921 An aggregate has an initializer which does not initialize all members.
2922 This warning can be independently controlled by
2923 @option{-Wmissing-field-initializers}.
2926 An initialized field without side effects is overridden when using
2927 designated initializers (@pxref{Designated Inits, , Designated
2928 Initializers}). This warning can be independently controlled by
2929 @option{-Woverride-init}.
2932 A function parameter is declared without a type specifier in K&R-style
2933 functions. This warning can be independently controlled by
2934 @option{-Wmissing-parameter-type}.
2937 An empty body occurs in an @samp{if} or @samp{else} statement. This
2938 warning can be independently controlled by @option{-Wempty-body}.
2941 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2942 @samp{>}, or @samp{>=}.
2945 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2946 This warning can be independently controlled by @option{-Wclobbered}.
2949 Any of several floating-point events that often indicate errors, such as
2950 overflow, underflow, loss of precision, etc.
2952 @item @r{(C++ only)}
2953 An enumerator and a non-enumerator both appear in a conditional expression.
2955 @item @r{(C++ only)}
2956 A non-static reference or non-static @samp{const} member appears in a
2957 class without constructors.
2959 @item @r{(C++ only)}
2960 Ambiguous virtual bases.
2962 @item @r{(C++ only)}
2963 Subscripting an array which has been declared @samp{register}.
2965 @item @r{(C++ only)}
2966 Taking the address of a variable which has been declared @samp{register}.
2968 @item @r{(C++ only)}
2969 A base class is not initialized in a derived class' copy constructor.
2972 @item -Wno-div-by-zero
2973 @opindex Wno-div-by-zero
2974 @opindex Wdiv-by-zero
2975 Do not warn about compile-time integer division by zero. Floating point
2976 division by zero is not warned about, as it can be a legitimate way of
2977 obtaining infinities and NaNs.
2979 @item -Wsystem-headers
2980 @opindex Wsystem-headers
2981 @cindex warnings from system headers
2982 @cindex system headers, warnings from
2983 Print warning messages for constructs found in system header files.
2984 Warnings from system headers are normally suppressed, on the assumption
2985 that they usually do not indicate real problems and would only make the
2986 compiler output harder to read. Using this command line option tells
2987 GCC to emit warnings from system headers as if they occurred in user
2988 code. However, note that using @option{-Wall} in conjunction with this
2989 option will @emph{not} warn about unknown pragmas in system
2990 headers---for that, @option{-Wunknown-pragmas} must also be used.
2993 @opindex Wfloat-equal
2994 Warn if floating point values are used in equality comparisons.
2996 The idea behind this is that sometimes it is convenient (for the
2997 programmer) to consider floating-point values as approximations to
2998 infinitely precise real numbers. If you are doing this, then you need
2999 to compute (by analyzing the code, or in some other way) the maximum or
3000 likely maximum error that the computation introduces, and allow for it
3001 when performing comparisons (and when producing output, but that's a
3002 different problem). In particular, instead of testing for equality, you
3003 would check to see whether the two values have ranges that overlap; and
3004 this is done with the relational operators, so equality comparisons are
3007 @item -Wtraditional @r{(C only)}
3008 @opindex Wtraditional
3009 Warn about certain constructs that behave differently in traditional and
3010 ISO C@. Also warn about ISO C constructs that have no traditional C
3011 equivalent, and/or problematic constructs which should be avoided.
3015 Macro parameters that appear within string literals in the macro body.
3016 In traditional C macro replacement takes place within string literals,
3017 but does not in ISO C@.
3020 In traditional C, some preprocessor directives did not exist.
3021 Traditional preprocessors would only consider a line to be a directive
3022 if the @samp{#} appeared in column 1 on the line. Therefore
3023 @option{-Wtraditional} warns about directives that traditional C
3024 understands but would ignore because the @samp{#} does not appear as the
3025 first character on the line. It also suggests you hide directives like
3026 @samp{#pragma} not understood by traditional C by indenting them. Some
3027 traditional implementations would not recognize @samp{#elif}, so it
3028 suggests avoiding it altogether.
3031 A function-like macro that appears without arguments.
3034 The unary plus operator.
3037 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3038 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3039 constants.) Note, these suffixes appear in macros defined in the system
3040 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3041 Use of these macros in user code might normally lead to spurious
3042 warnings, however GCC's integrated preprocessor has enough context to
3043 avoid warning in these cases.
3046 A function declared external in one block and then used after the end of
3050 A @code{switch} statement has an operand of type @code{long}.
3053 A non-@code{static} function declaration follows a @code{static} one.
3054 This construct is not accepted by some traditional C compilers.
3057 The ISO type of an integer constant has a different width or
3058 signedness from its traditional type. This warning is only issued if
3059 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3060 typically represent bit patterns, are not warned about.
3063 Usage of ISO string concatenation is detected.
3066 Initialization of automatic aggregates.
3069 Identifier conflicts with labels. Traditional C lacks a separate
3070 namespace for labels.
3073 Initialization of unions. If the initializer is zero, the warning is
3074 omitted. This is done under the assumption that the zero initializer in
3075 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3076 initializer warnings and relies on default initialization to zero in the
3080 Conversions by prototypes between fixed/floating point values and vice
3081 versa. The absence of these prototypes when compiling with traditional
3082 C would cause serious problems. This is a subset of the possible
3083 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3086 Use of ISO C style function definitions. This warning intentionally is
3087 @emph{not} issued for prototype declarations or variadic functions
3088 because these ISO C features will appear in your code when using
3089 libiberty's traditional C compatibility macros, @code{PARAMS} and
3090 @code{VPARAMS}. This warning is also bypassed for nested functions
3091 because that feature is already a GCC extension and thus not relevant to
3092 traditional C compatibility.
3095 @item -Wtraditional-conversion @r{(C only)}
3096 @opindex Wtraditional-conversion
3097 Warn if a prototype causes a type conversion that is different from what
3098 would happen to the same argument in the absence of a prototype. This
3099 includes conversions of fixed point to floating and vice versa, and
3100 conversions changing the width or signedness of a fixed point argument
3101 except when the same as the default promotion.
3103 @item -Wdeclaration-after-statement @r{(C only)}
3104 @opindex Wdeclaration-after-statement
3105 Warn when a declaration is found after a statement in a block. This
3106 construct, known from C++, was introduced with ISO C99 and is by default
3107 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3108 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3112 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3114 @item -Wno-endif-labels
3115 @opindex Wno-endif-labels
3116 @opindex Wendif-labels
3117 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3121 Warn whenever a local variable shadows another local variable, parameter or
3122 global variable or whenever a built-in function is shadowed.
3124 @item -Wlarger-than-@var{len}
3125 @opindex Wlarger-than
3126 Warn whenever an object of larger than @var{len} bytes is defined.
3128 @item -Wunsafe-loop-optimizations
3129 @opindex Wunsafe-loop-optimizations
3130 Warn if the loop cannot be optimized because the compiler could not
3131 assume anything on the bounds of the loop indices. With
3132 @option{-funsafe-loop-optimizations} warn if the compiler made
3135 @item -Wpointer-arith
3136 @opindex Wpointer-arith
3137 Warn about anything that depends on the ``size of'' a function type or
3138 of @code{void}. GNU C assigns these types a size of 1, for
3139 convenience in calculations with @code{void *} pointers and pointers
3142 @item -Wbad-function-cast @r{(C only)}
3143 @opindex Wbad-function-cast
3144 Warn whenever a function call is cast to a non-matching type.
3145 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3148 Warn about ISO C constructs that are outside of the common subset of
3149 ISO C and ISO C++, e.g.@: request for implicit conversion from
3150 @code{void *} to a pointer to non-@code{void} type.
3154 Warn whenever a pointer is cast so as to remove a type qualifier from
3155 the target type. For example, warn if a @code{const char *} is cast
3156 to an ordinary @code{char *}.
3159 @opindex Wcast-align
3160 Warn whenever a pointer is cast such that the required alignment of the
3161 target is increased. For example, warn if a @code{char *} is cast to
3162 an @code{int *} on machines where integers can only be accessed at
3163 two- or four-byte boundaries.
3165 @item -Wwrite-strings
3166 @opindex Wwrite-strings
3167 When compiling C, give string constants the type @code{const
3168 char[@var{length}]} so that
3169 copying the address of one into a non-@code{const} @code{char *}
3170 pointer will get a warning; when compiling C++, warn about the
3171 deprecated conversion from string literals to @code{char *}. This
3172 warning, by default, is enabled for C++ programs.
3173 These warnings will help you find at
3174 compile time code that can try to write into a string constant, but
3175 only if you have been very careful about using @code{const} in
3176 declarations and prototypes. Otherwise, it will just be a nuisance;
3177 this is why we did not make @option{-Wall} request these warnings.
3181 Warn for variables that might be changed by @samp{longjmp} or
3182 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3185 @opindex Wconversion
3186 Warn for implicit conversions that may alter a value. This includes
3187 conversions between real and integer, like @code{abs (x)} when
3188 @code{x} is @code{double}; conversions between signed and unsigned,
3189 like @code{unsigned ui = -1}; and conversions to smaller types, like
3190 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3191 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3192 changed by the conversion like in @code{abs (2.0)}.
3195 @opindex Wempty-body
3196 An empty body occurs in an @samp{if} or @samp{else} statement.
3197 This warning is also enabled by @option{-Wextra}.
3199 @item -Wsign-compare
3200 @opindex Wsign-compare
3201 @cindex warning for comparison of signed and unsigned values
3202 @cindex comparison of signed and unsigned values, warning
3203 @cindex signed and unsigned values, comparison warning
3204 Warn when a comparison between signed and unsigned values could produce
3205 an incorrect result when the signed value is converted to unsigned.
3206 This warning is also enabled by @option{-Wextra}; to get the other warnings
3207 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3209 @item -Waggregate-return
3210 @opindex Waggregate-return
3211 Warn if any functions that return structures or unions are defined or
3212 called. (In languages where you can return an array, this also elicits
3216 @opindex Walways-true
3217 Warn about comparisons which are always true such as testing if
3218 unsigned values are greater than or equal to zero. This warning is
3219 enabled by @option{-Wall}.
3221 @item -Wno-attributes
3222 @opindex Wno-attributes
3223 @opindex Wattributes
3224 Do not warn if an unexpected @code{__attribute__} is used, such as
3225 unrecognized attributes, function attributes applied to variables,
3226 etc. This will not stop errors for incorrect use of supported
3229 @item -Wstrict-prototypes @r{(C only)}
3230 @opindex Wstrict-prototypes
3231 Warn if a function is declared or defined without specifying the
3232 argument types. (An old-style function definition is permitted without
3233 a warning if preceded by a declaration which specifies the argument
3236 @item -Wold-style-declaration @r{(C only)}
3237 @opindex Wold-style-declaration
3238 Warn for obsolescent usages, according to the C Standard, in a
3239 declaration. For example, warn if storage-class specifiers like
3240 @code{static} are not the first things in a declaration. This warning
3241 is also enabled by @option{-Wextra}.
3243 @item -Wold-style-definition @r{(C only)}
3244 @opindex Wold-style-definition
3245 Warn if an old-style function definition is used. A warning is given
3246 even if there is a previous prototype.
3248 @item -Wmissing-parameter-type @r{(C only)}
3249 @opindex Wmissing-parameter-type
3250 A function parameter is declared without a type specifier in K&R-style
3257 This warning is also enabled by @option{-Wextra}.
3259 @item -Wmissing-prototypes @r{(C only)}
3260 @opindex Wmissing-prototypes
3261 Warn if a global function is defined without a previous prototype
3262 declaration. This warning is issued even if the definition itself
3263 provides a prototype. The aim is to detect global functions that fail
3264 to be declared in header files.
3266 @item -Wmissing-declarations @r{(C and C++ only)}
3267 @opindex Wmissing-declarations
3268 Warn if a global function is defined without a previous declaration.
3269 Do so even if the definition itself provides a prototype.
3270 Use this option to detect global functions that are not declared in
3271 header files. In C++, no warnings are issued for function templates,
3272 or for inline functions, or for functions in anonymous namespaces.
3274 @item -Wmissing-field-initializers
3275 @opindex Wmissing-field-initializers
3278 Warn if a structure's initializer has some fields missing. For
3279 example, the following code would cause such a warning, because
3280 @code{x.h} is implicitly zero:
3283 struct s @{ int f, g, h; @};
3284 struct s x = @{ 3, 4 @};
3287 This option does not warn about designated initializers, so the following
3288 modification would not trigger a warning:
3291 struct s @{ int f, g, h; @};
3292 struct s x = @{ .f = 3, .g = 4 @};
3295 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3296 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3298 @item -Wmissing-noreturn
3299 @opindex Wmissing-noreturn
3300 Warn about functions which might be candidates for attribute @code{noreturn}.
3301 Note these are only possible candidates, not absolute ones. Care should
3302 be taken to manually verify functions actually do not ever return before
3303 adding the @code{noreturn} attribute, otherwise subtle code generation
3304 bugs could be introduced. You will not get a warning for @code{main} in
3305 hosted C environments.
3307 @item -Wmissing-format-attribute
3308 @opindex Wmissing-format-attribute
3310 Warn about function pointers which might be candidates for @code{format}
3311 attributes. Note these are only possible candidates, not absolute ones.
3312 GCC will guess that function pointers with @code{format} attributes that
3313 are used in assignment, initialization, parameter passing or return
3314 statements should have a corresponding @code{format} attribute in the
3315 resulting type. I.e.@: the left-hand side of the assignment or
3316 initialization, the type of the parameter variable, or the return type
3317 of the containing function respectively should also have a @code{format}
3318 attribute to avoid the warning.
3320 GCC will also warn about function definitions which might be
3321 candidates for @code{format} attributes. Again, these are only
3322 possible candidates. GCC will guess that @code{format} attributes
3323 might be appropriate for any function that calls a function like
3324 @code{vprintf} or @code{vscanf}, but this might not always be the
3325 case, and some functions for which @code{format} attributes are
3326 appropriate may not be detected.
3328 @item -Wno-multichar
3329 @opindex Wno-multichar
3331 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3332 Usually they indicate a typo in the user's code, as they have
3333 implementation-defined values, and should not be used in portable code.
3335 @item -Wnormalized=<none|id|nfc|nfkc>
3336 @opindex Wnormalized
3339 @cindex character set, input normalization
3340 In ISO C and ISO C++, two identifiers are different if they are
3341 different sequences of characters. However, sometimes when characters
3342 outside the basic ASCII character set are used, you can have two
3343 different character sequences that look the same. To avoid confusion,
3344 the ISO 10646 standard sets out some @dfn{normalization rules} which
3345 when applied ensure that two sequences that look the same are turned into
3346 the same sequence. GCC can warn you if you are using identifiers which
3347 have not been normalized; this option controls that warning.
3349 There are four levels of warning that GCC supports. The default is
3350 @option{-Wnormalized=nfc}, which warns about any identifier which is
3351 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3352 recommended form for most uses.
3354 Unfortunately, there are some characters which ISO C and ISO C++ allow
3355 in identifiers that when turned into NFC aren't allowable as
3356 identifiers. That is, there's no way to use these symbols in portable
3357 ISO C or C++ and have all your identifiers in NFC.
3358 @option{-Wnormalized=id} suppresses the warning for these characters.
3359 It is hoped that future versions of the standards involved will correct
3360 this, which is why this option is not the default.
3362 You can switch the warning off for all characters by writing
3363 @option{-Wnormalized=none}. You would only want to do this if you
3364 were using some other normalization scheme (like ``D''), because
3365 otherwise you can easily create bugs that are literally impossible to see.
3367 Some characters in ISO 10646 have distinct meanings but look identical
3368 in some fonts or display methodologies, especially once formatting has
3369 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3370 LETTER N'', will display just like a regular @code{n} which has been
3371 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3372 normalization scheme to convert all these into a standard form as
3373 well, and GCC will warn if your code is not in NFKC if you use
3374 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3375 about every identifier that contains the letter O because it might be
3376 confused with the digit 0, and so is not the default, but may be
3377 useful as a local coding convention if the programming environment is
3378 unable to be fixed to display these characters distinctly.
3380 @item -Wno-deprecated-declarations
3381 @opindex Wno-deprecated-declarations
3382 Do not warn about uses of functions (@pxref{Function Attributes}),
3383 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3384 Attributes}) marked as deprecated by using the @code{deprecated}
3388 @opindex Wno-overflow
3389 Do not warn about compile-time overflow in constant expressions.
3391 @item -Woverride-init
3392 @opindex Woverride-init
3395 Warn if an initialized field without side effects is overridden when
3396 using designated initializers (@pxref{Designated Inits, , Designated
3399 This warning is included in @option{-Wextra}. To get other
3400 @option{-Wextra} warnings without this one, use @samp{-Wextra
3401 -Wno-override-init}.
3405 Warn if a structure is given the packed attribute, but the packed
3406 attribute has no effect on the layout or size of the structure.
3407 Such structures may be mis-aligned for little benefit. For
3408 instance, in this code, the variable @code{f.x} in @code{struct bar}
3409 will be misaligned even though @code{struct bar} does not itself
3410 have the packed attribute:
3417 @} __attribute__((packed));
3427 Warn if padding is included in a structure, either to align an element
3428 of the structure or to align the whole structure. Sometimes when this
3429 happens it is possible to rearrange the fields of the structure to
3430 reduce the padding and so make the structure smaller.
3432 @item -Wredundant-decls
3433 @opindex Wredundant-decls
3434 Warn if anything is declared more than once in the same scope, even in
3435 cases where multiple declaration is valid and changes nothing.
3437 @item -Wnested-externs @r{(C only)}
3438 @opindex Wnested-externs
3439 Warn if an @code{extern} declaration is encountered within a function.
3441 @item -Wunreachable-code
3442 @opindex Wunreachable-code
3443 Warn if the compiler detects that code will never be executed.
3445 This option is intended to warn when the compiler detects that at
3446 least a whole line of source code will never be executed, because
3447 some condition is never satisfied or because it is after a
3448 procedure that never returns.
3450 It is possible for this option to produce a warning even though there
3451 are circumstances under which part of the affected line can be executed,
3452 so care should be taken when removing apparently-unreachable code.
3454 For instance, when a function is inlined, a warning may mean that the
3455 line is unreachable in only one inlined copy of the function.
3457 This option is not made part of @option{-Wall} because in a debugging
3458 version of a program there is often substantial code which checks
3459 correct functioning of the program and is, hopefully, unreachable
3460 because the program does work. Another common use of unreachable
3461 code is to provide behavior which is selectable at compile-time.
3465 Warn if a function can not be inlined and it was declared as inline.
3466 Even with this option, the compiler will not warn about failures to
3467 inline functions declared in system headers.
3469 The compiler uses a variety of heuristics to determine whether or not
3470 to inline a function. For example, the compiler takes into account
3471 the size of the function being inlined and the amount of inlining
3472 that has already been done in the current function. Therefore,
3473 seemingly insignificant changes in the source program can cause the
3474 warnings produced by @option{-Winline} to appear or disappear.
3476 @item -Wno-invalid-offsetof @r{(C++ only)}
3477 @opindex Wno-invalid-offsetof
3478 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3479 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3480 to a non-POD type is undefined. In existing C++ implementations,
3481 however, @samp{offsetof} typically gives meaningful results even when
3482 applied to certain kinds of non-POD types. (Such as a simple
3483 @samp{struct} that fails to be a POD type only by virtue of having a
3484 constructor.) This flag is for users who are aware that they are
3485 writing nonportable code and who have deliberately chosen to ignore the
3488 The restrictions on @samp{offsetof} may be relaxed in a future version
3489 of the C++ standard.
3491 @item -Wno-int-to-pointer-cast @r{(C only)}
3492 @opindex Wno-int-to-pointer-cast
3493 Suppress warnings from casts to pointer type of an integer of a
3496 @item -Wno-pointer-to-int-cast @r{(C only)}
3497 @opindex Wno-pointer-to-int-cast
3498 Suppress warnings from casts from a pointer to an integer type of a
3502 @opindex Winvalid-pch
3503 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3504 the search path but can't be used.
3508 @opindex Wno-long-long
3509 Warn if @samp{long long} type is used. This is default. To inhibit
3510 the warning messages, use @option{-Wno-long-long}. Flags
3511 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3512 only when @option{-pedantic} flag is used.
3514 @item -Wvariadic-macros
3515 @opindex Wvariadic-macros
3516 @opindex Wno-variadic-macros
3517 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3518 alternate syntax when in pedantic ISO C99 mode. This is default.
3519 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3521 @item -Wvolatile-register-var
3522 @opindex Wvolatile-register-var
3523 @opindex Wno-volatile-register-var
3524 Warn if a register variable is declared volatile. The volatile
3525 modifier does not inhibit all optimizations that may eliminate reads
3526 and/or writes to register variables.
3528 @item -Wdisabled-optimization
3529 @opindex Wdisabled-optimization
3530 Warn if a requested optimization pass is disabled. This warning does
3531 not generally indicate that there is anything wrong with your code; it
3532 merely indicates that GCC's optimizers were unable to handle the code
3533 effectively. Often, the problem is that your code is too big or too
3534 complex; GCC will refuse to optimize programs when the optimization
3535 itself is likely to take inordinate amounts of time.
3537 @item -Wpointer-sign
3538 @opindex Wpointer-sign
3539 @opindex Wno-pointer-sign
3540 Warn for pointer argument passing or assignment with different signedness.
3541 This option is only supported for C and Objective-C@. It is implied by
3542 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3543 @option{-Wno-pointer-sign}.
3547 Make all warnings into errors.
3551 Make the specified warning into an errors. The specifier for a
3552 warning is appended, for example @option{-Werror=switch} turns the
3553 warnings controlled by @option{-Wswitch} into errors. This switch
3554 takes a negative form, to be used to negate @option{-Werror} for
3555 specific warnings, for example @option{-Wno-error=switch} makes
3556 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3557 is in effect. You can use the @option{-fdiagnostics-show-option}
3558 option to have each controllable warning amended with the option which
3559 controls it, to determine what to use with this option.
3561 Note that specifying @option{-Werror=}@var{foo} automatically implies
3562 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3565 @item -Wstack-protector
3566 @opindex Wstack-protector
3567 This option is only active when @option{-fstack-protector} is active. It
3568 warns about functions that will not be protected against stack smashing.
3570 @item -Wstring-literal-comparison
3571 @opindex Wstring-literal-comparison
3572 Warn about suspicious comparisons to string literal constants. In C,
3573 direct comparisons against the memory address of a string literal, such
3574 as @code{if (x == "abc")}, typically indicate a programmer error, and
3575 even when intentional, result in unspecified behavior and are not portable.
3576 Usually these warnings alert that the programmer intended to use
3577 @code{strcmp}. This warning is enabled by @option{-Wall}.
3579 @item -Woverlength-strings
3580 @opindex Woverlength-strings
3581 Warn about string constants which are longer than the ``minimum
3582 maximum'' length specified in the C standard. Modern compilers
3583 generally allow string constants which are much longer than the
3584 standard's minimum limit, but very portable programs should avoid
3585 using longer strings.
3587 The limit applies @emph{after} string constant concatenation, and does
3588 not count the trailing NUL@. In C89, the limit was 509 characters; in
3589 C99, it was raised to 4095. C++98 does not specify a normative
3590 minimum maximum, so we do not diagnose overlength strings in C++@.
3592 This option is implied by @option{-pedantic}, and can be disabled with
3593 @option{-Wno-overlength-strings}.
3596 @node Debugging Options
3597 @section Options for Debugging Your Program or GCC
3598 @cindex options, debugging
3599 @cindex debugging information options
3601 GCC has various special options that are used for debugging
3602 either your program or GCC:
3607 Produce debugging information in the operating system's native format
3608 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3611 On most systems that use stabs format, @option{-g} enables use of extra
3612 debugging information that only GDB can use; this extra information
3613 makes debugging work better in GDB but will probably make other debuggers
3615 refuse to read the program. If you want to control for certain whether
3616 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3617 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3619 GCC allows you to use @option{-g} with
3620 @option{-O}. The shortcuts taken by optimized code may occasionally
3621 produce surprising results: some variables you declared may not exist
3622 at all; flow of control may briefly move where you did not expect it;
3623 some statements may not be executed because they compute constant
3624 results or their values were already at hand; some statements may
3625 execute in different places because they were moved out of loops.
3627 Nevertheless it proves possible to debug optimized output. This makes
3628 it reasonable to use the optimizer for programs that might have bugs.
3630 The following options are useful when GCC is generated with the
3631 capability for more than one debugging format.
3635 Produce debugging information for use by GDB@. This means to use the
3636 most expressive format available (DWARF 2, stabs, or the native format
3637 if neither of those are supported), including GDB extensions if at all
3642 Produce debugging information in stabs format (if that is supported),
3643 without GDB extensions. This is the format used by DBX on most BSD
3644 systems. On MIPS, Alpha and System V Release 4 systems this option
3645 produces stabs debugging output which is not understood by DBX or SDB@.
3646 On System V Release 4 systems this option requires the GNU assembler.
3648 @item -feliminate-unused-debug-symbols
3649 @opindex feliminate-unused-debug-symbols
3650 Produce debugging information in stabs format (if that is supported),
3651 for only symbols that are actually used.
3653 @item -femit-class-debug-always
3654 Instead of emitting debugging information for a C++ class in only one
3655 object file, emit it in all object files using the class. This option
3656 should be used only with debuggers that are unable to handle the way GCC
3657 normally emits debugging information for classes because using this
3658 option will increase the size of debugging information by as much as a
3663 Produce debugging information in stabs format (if that is supported),
3664 using GNU extensions understood only by the GNU debugger (GDB)@. The
3665 use of these extensions is likely to make other debuggers crash or
3666 refuse to read the program.
3670 Produce debugging information in COFF format (if that is supported).
3671 This is the format used by SDB on most System V systems prior to
3676 Produce debugging information in XCOFF format (if that is supported).
3677 This is the format used by the DBX debugger on IBM RS/6000 systems.
3681 Produce debugging information in XCOFF format (if that is supported),
3682 using GNU extensions understood only by the GNU debugger (GDB)@. The
3683 use of these extensions is likely to make other debuggers crash or
3684 refuse to read the program, and may cause assemblers other than the GNU
3685 assembler (GAS) to fail with an error.
3689 Produce debugging information in DWARF version 2 format (if that is
3690 supported). This is the format used by DBX on IRIX 6. With this
3691 option, GCC uses features of DWARF version 3 when they are useful;
3692 version 3 is upward compatible with version 2, but may still cause
3693 problems for older debuggers.
3697 Produce debugging information in VMS debug format (if that is
3698 supported). This is the format used by DEBUG on VMS systems.
3701 @itemx -ggdb@var{level}
3702 @itemx -gstabs@var{level}
3703 @itemx -gcoff@var{level}
3704 @itemx -gxcoff@var{level}
3705 @itemx -gvms@var{level}
3706 Request debugging information and also use @var{level} to specify how
3707 much information. The default level is 2.
3709 Level 0 produces no debug information at all. Thus, @option{-g0} negates
3712 Level 1 produces minimal information, enough for making backtraces in
3713 parts of the program that you don't plan to debug. This includes
3714 descriptions of functions and external variables, but no information
3715 about local variables and no line numbers.
3717 Level 3 includes extra information, such as all the macro definitions
3718 present in the program. Some debuggers support macro expansion when
3719 you use @option{-g3}.
3721 @option{-gdwarf-2} does not accept a concatenated debug level, because
3722 GCC used to support an option @option{-gdwarf} that meant to generate
3723 debug information in version 1 of the DWARF format (which is very
3724 different from version 2), and it would have been too confusing. That
3725 debug format is long obsolete, but the option cannot be changed now.
3726 Instead use an additional @option{-g@var{level}} option to change the
3727 debug level for DWARF2.
3729 @item -feliminate-dwarf2-dups
3730 @opindex feliminate-dwarf2-dups
3731 Compress DWARF2 debugging information by eliminating duplicated
3732 information about each symbol. This option only makes sense when
3733 generating DWARF2 debugging information with @option{-gdwarf-2}.
3735 @cindex @command{prof}
3738 Generate extra code to write profile information suitable for the
3739 analysis program @command{prof}. You must use this option when compiling
3740 the source files you want data about, and you must also use it when
3743 @cindex @command{gprof}
3746 Generate extra code to write profile information suitable for the
3747 analysis program @command{gprof}. You must use this option when compiling
3748 the source files you want data about, and you must also use it when
3753 Makes the compiler print out each function name as it is compiled, and
3754 print some statistics about each pass when it finishes.
3757 @opindex ftime-report
3758 Makes the compiler print some statistics about the time consumed by each
3759 pass when it finishes.
3762 @opindex fmem-report
3763 Makes the compiler print some statistics about permanent memory
3764 allocation when it finishes.
3766 @item -fpre-ipa-mem-report
3767 @opindex fpre-ipa-mem-report
3768 @item -fpost-ipa-mem-report
3769 @opindex fpost-ipa-mem-report
3770 Makes the compiler print some statistics about permanent memory
3771 allocation before or after interprocedural optimization.
3773 @item -fprofile-arcs
3774 @opindex fprofile-arcs
3775 Add code so that program flow @dfn{arcs} are instrumented. During
3776 execution the program records how many times each branch and call is
3777 executed and how many times it is taken or returns. When the compiled
3778 program exits it saves this data to a file called
3779 @file{@var{auxname}.gcda} for each source file. The data may be used for
3780 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3781 test coverage analysis (@option{-ftest-coverage}). Each object file's
3782 @var{auxname} is generated from the name of the output file, if
3783 explicitly specified and it is not the final executable, otherwise it is
3784 the basename of the source file. In both cases any suffix is removed
3785 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3786 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3787 @xref{Cross-profiling}.
3789 @cindex @command{gcov}
3793 This option is used to compile and link code instrumented for coverage
3794 analysis. The option is a synonym for @option{-fprofile-arcs}
3795 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
3796 linking). See the documentation for those options for more details.
3801 Compile the source files with @option{-fprofile-arcs} plus optimization
3802 and code generation options. For test coverage analysis, use the
3803 additional @option{-ftest-coverage} option. You do not need to profile
3804 every source file in a program.
3807 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3808 (the latter implies the former).
3811 Run the program on a representative workload to generate the arc profile
3812 information. This may be repeated any number of times. You can run
3813 concurrent instances of your program, and provided that the file system
3814 supports locking, the data files will be correctly updated. Also
3815 @code{fork} calls are detected and correctly handled (double counting
3819 For profile-directed optimizations, compile the source files again with
3820 the same optimization and code generation options plus
3821 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3822 Control Optimization}).
3825 For test coverage analysis, use @command{gcov} to produce human readable
3826 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3827 @command{gcov} documentation for further information.
3831 With @option{-fprofile-arcs}, for each function of your program GCC
3832 creates a program flow graph, then finds a spanning tree for the graph.
3833 Only arcs that are not on the spanning tree have to be instrumented: the
3834 compiler adds code to count the number of times that these arcs are
3835 executed. When an arc is the only exit or only entrance to a block, the
3836 instrumentation code can be added to the block; otherwise, a new basic
3837 block must be created to hold the instrumentation code.
3840 @item -ftest-coverage
3841 @opindex ftest-coverage
3842 Produce a notes file that the @command{gcov} code-coverage utility
3843 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3844 show program coverage. Each source file's note file is called
3845 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3846 above for a description of @var{auxname} and instructions on how to
3847 generate test coverage data. Coverage data will match the source files
3848 more closely, if you do not optimize.
3850 @item -d@var{letters}
3851 @item -fdump-rtl-@var{pass}
3853 Says to make debugging dumps during compilation at times specified by
3854 @var{letters}. This is used for debugging the RTL-based passes of the
3855 compiler. The file names for most of the dumps are made by appending a
3856 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3857 from the name of the output file, if explicitly specified and it is not
3858 an executable, otherwise it is the basename of the source file.
3860 Most debug dumps can be enabled either passing a letter to the @option{-d}
3861 option, or with a long @option{-fdump-rtl} switch; here are the possible
3862 letters for use in @var{letters} and @var{pass}, and their meanings:
3867 Annotate the assembler output with miscellaneous debugging information.
3870 @itemx -fdump-rtl-bbro
3872 @opindex fdump-rtl-bbro
3873 Dump after block reordering, to @file{@var{file}.148r.bbro}.
3876 @itemx -fdump-rtl-combine
3878 @opindex fdump-rtl-combine
3879 Dump after the RTL instruction combination pass, to the file
3880 @file{@var{file}.129r.combine}.
3883 @itemx -fdump-rtl-ce1
3884 @itemx -fdump-rtl-ce2
3886 @opindex fdump-rtl-ce1
3887 @opindex fdump-rtl-ce2
3888 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3889 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
3890 and @option{-fdump-rtl-ce2} enable dumping after the second if
3891 conversion, to the file @file{@var{file}.130r.ce2}.
3894 @itemx -fdump-rtl-btl
3895 @itemx -fdump-rtl-dbr
3897 @opindex fdump-rtl-btl
3898 @opindex fdump-rtl-dbr
3899 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3900 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
3901 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3902 scheduling, to @file{@var{file}.36.dbr}.
3906 Dump all macro definitions, at the end of preprocessing, in addition to
3910 @itemx -fdump-rtl-ce3
3912 @opindex fdump-rtl-ce3
3913 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
3916 @itemx -fdump-rtl-cfg
3917 @itemx -fdump-rtl-life
3919 @opindex fdump-rtl-cfg
3920 @opindex fdump-rtl-life
3921 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3922 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
3923 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3924 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
3927 @itemx -fdump-rtl-greg
3929 @opindex fdump-rtl-greg
3930 Dump after global register allocation, to @file{@var{file}.139r.greg}.
3933 @itemx -fdump-rtl-gcse
3934 @itemx -fdump-rtl-bypass
3936 @opindex fdump-rtl-gcse
3937 @opindex fdump-rtl-bypass
3938 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3939 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3940 enable dumping after jump bypassing and control flow optimizations, to
3941 @file{@var{file}.115r.bypass}.
3944 @itemx -fdump-rtl-eh
3946 @opindex fdump-rtl-eh
3947 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3950 @itemx -fdump-rtl-sibling
3952 @opindex fdump-rtl-sibling
3953 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
3956 @itemx -fdump-rtl-jump
3958 @opindex fdump-rtl-jump
3959 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
3962 @itemx -fdump-rtl-stack
3964 @opindex fdump-rtl-stack
3965 Dump after conversion from GCC's "flat register file" registers to the
3966 x87's stack-like registers, to @file{@var{file}.152r.stack}.
3969 @itemx -fdump-rtl-lreg
3971 @opindex fdump-rtl-lreg
3972 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
3975 @itemx -fdump-rtl-loop2
3977 @opindex fdump-rtl-loop2
3978 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
3979 loop optimization pass, to @file{@var{file}.119r.loop2},
3980 @file{@var{file}.120r.loop2_init},
3981 @file{@var{file}.121r.loop2_invariant}, and
3982 @file{@var{file}.125r.loop2_done}.
3985 @itemx -fdump-rtl-sms
3987 @opindex fdump-rtl-sms
3988 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
3991 @itemx -fdump-rtl-mach
3993 @opindex fdump-rtl-mach
3994 Dump after performing the machine dependent reorganization pass, to
3995 @file{@var{file}.155r.mach}.
3998 @itemx -fdump-rtl-rnreg
4000 @opindex fdump-rtl-rnreg
4001 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
4004 @itemx -fdump-rtl-regmove
4006 @opindex fdump-rtl-regmove
4007 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
4010 @itemx -fdump-rtl-postreload
4012 @opindex fdump-rtl-postreload
4013 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
4016 @itemx -fdump-rtl-expand
4018 @opindex fdump-rtl-expand
4019 Dump after RTL generation, to @file{@var{file}.104r.expand}.
4022 @itemx -fdump-rtl-sched2
4024 @opindex fdump-rtl-sched2
4025 Dump after the second scheduling pass, to @file{@var{file}.149r.sched2}.
4028 @itemx -fdump-rtl-cse
4030 @opindex fdump-rtl-cse
4031 Dump after CSE (including the jump optimization that sometimes follows
4032 CSE), to @file{@var{file}.113r.cse}.
4035 @itemx -fdump-rtl-sched1
4037 @opindex fdump-rtl-sched1
4038 Dump after the first scheduling pass, to @file{@var{file}.136r.sched1}.
4041 @itemx -fdump-rtl-cse2
4043 @opindex fdump-rtl-cse2
4044 Dump after the second CSE pass (including the jump optimization that
4045 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
4048 @itemx -fdump-rtl-tracer
4050 @opindex fdump-rtl-tracer
4051 Dump after running tracer, to @file{@var{file}.118r.tracer}.
4054 @itemx -fdump-rtl-vpt
4055 @itemx -fdump-rtl-vartrack
4057 @opindex fdump-rtl-vpt
4058 @opindex fdump-rtl-vartrack
4059 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
4060 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
4061 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
4062 to @file{@var{file}.154r.vartrack}.
4065 @itemx -fdump-rtl-flow2
4067 @opindex fdump-rtl-flow2
4068 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
4071 @itemx -fdump-rtl-peephole2
4073 @opindex fdump-rtl-peephole2
4074 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
4077 @itemx -fdump-rtl-web
4079 @opindex fdump-rtl-web
4080 Dump after live range splitting, to @file{@var{file}.126r.web}.
4083 @itemx -fdump-rtl-all
4085 @opindex fdump-rtl-all
4086 Produce all the dumps listed above.
4090 Produce a core dump whenever an error occurs.
4094 Print statistics on memory usage, at the end of the run, to
4099 Annotate the assembler output with a comment indicating which
4100 pattern and alternative was used. The length of each instruction is
4105 Dump the RTL in the assembler output as a comment before each instruction.
4106 Also turns on @option{-dp} annotation.
4110 For each of the other indicated dump files (either with @option{-d} or
4111 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4112 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4116 Just generate RTL for a function instead of compiling it. Usually used
4117 with @samp{r} (@option{-fdump-rtl-expand}).
4121 Dump debugging information during parsing, to standard error.
4125 @opindex fdump-noaddr
4126 When doing debugging dumps (see @option{-d} option above), suppress
4127 address output. This makes it more feasible to use diff on debugging
4128 dumps for compiler invocations with different compiler binaries and/or
4129 different text / bss / data / heap / stack / dso start locations.
4131 @item -fdump-unnumbered
4132 @opindex fdump-unnumbered
4133 When doing debugging dumps (see @option{-d} option above), suppress instruction
4134 numbers, line number note and address output. This makes it more feasible to
4135 use diff on debugging dumps for compiler invocations with different
4136 options, in particular with and without @option{-g}.
4138 @item -fdump-translation-unit @r{(C++ only)}
4139 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4140 @opindex fdump-translation-unit
4141 Dump a representation of the tree structure for the entire translation
4142 unit to a file. The file name is made by appending @file{.tu} to the
4143 source file name. If the @samp{-@var{options}} form is used, @var{options}
4144 controls the details of the dump as described for the
4145 @option{-fdump-tree} options.
4147 @item -fdump-class-hierarchy @r{(C++ only)}
4148 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4149 @opindex fdump-class-hierarchy
4150 Dump a representation of each class's hierarchy and virtual function
4151 table layout to a file. The file name is made by appending @file{.class}
4152 to the source file name. If the @samp{-@var{options}} form is used,
4153 @var{options} controls the details of the dump as described for the
4154 @option{-fdump-tree} options.
4156 @item -fdump-ipa-@var{switch}
4158 Control the dumping at various stages of inter-procedural analysis
4159 language tree to a file. The file name is generated by appending a switch
4160 specific suffix to the source file name. The following dumps are possible:
4164 Enables all inter-procedural analysis dumps; currently the only produced
4165 dump is the @samp{cgraph} dump.
4168 Dumps information about call-graph optimization, unused function removal,
4169 and inlining decisions.
4172 @item -fdump-tree-@var{switch}
4173 @itemx -fdump-tree-@var{switch}-@var{options}
4175 Control the dumping at various stages of processing the intermediate
4176 language tree to a file. The file name is generated by appending a switch
4177 specific suffix to the source file name. If the @samp{-@var{options}}
4178 form is used, @var{options} is a list of @samp{-} separated options that
4179 control the details of the dump. Not all options are applicable to all
4180 dumps, those which are not meaningful will be ignored. The following
4181 options are available
4185 Print the address of each node. Usually this is not meaningful as it
4186 changes according to the environment and source file. Its primary use
4187 is for tying up a dump file with a debug environment.
4189 Inhibit dumping of members of a scope or body of a function merely
4190 because that scope has been reached. Only dump such items when they
4191 are directly reachable by some other path. When dumping pretty-printed
4192 trees, this option inhibits dumping the bodies of control structures.
4194 Print a raw representation of the tree. By default, trees are
4195 pretty-printed into a C-like representation.
4197 Enable more detailed dumps (not honored by every dump option).
4199 Enable dumping various statistics about the pass (not honored by every dump
4202 Enable showing basic block boundaries (disabled in raw dumps).
4204 Enable showing virtual operands for every statement.
4206 Enable showing line numbers for statements.
4208 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4210 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4213 The following tree dumps are possible:
4217 Dump before any tree based optimization, to @file{@var{file}.original}.
4220 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4223 Dump after function inlining, to @file{@var{file}.inlined}.
4226 @opindex fdump-tree-gimple
4227 Dump each function before and after the gimplification pass to a file. The
4228 file name is made by appending @file{.gimple} to the source file name.
4231 @opindex fdump-tree-cfg
4232 Dump the control flow graph of each function to a file. The file name is
4233 made by appending @file{.cfg} to the source file name.
4236 @opindex fdump-tree-vcg
4237 Dump the control flow graph of each function to a file in VCG format. The
4238 file name is made by appending @file{.vcg} to the source file name. Note
4239 that if the file contains more than one function, the generated file cannot
4240 be used directly by VCG@. You will need to cut and paste each function's
4241 graph into its own separate file first.
4244 @opindex fdump-tree-ch
4245 Dump each function after copying loop headers. The file name is made by
4246 appending @file{.ch} to the source file name.
4249 @opindex fdump-tree-ssa
4250 Dump SSA related information to a file. The file name is made by appending
4251 @file{.ssa} to the source file name.
4254 @opindex fdump-tree-salias
4255 Dump structure aliasing variable information to a file. This file name
4256 is made by appending @file{.salias} to the source file name.
4259 @opindex fdump-tree-alias
4260 Dump aliasing information for each function. The file name is made by
4261 appending @file{.alias} to the source file name.
4264 @opindex fdump-tree-ccp
4265 Dump each function after CCP@. The file name is made by appending
4266 @file{.ccp} to the source file name.
4269 @opindex fdump-tree-storeccp
4270 Dump each function after STORE-CCP. The file name is made by appending
4271 @file{.storeccp} to the source file name.
4274 @opindex fdump-tree-pre
4275 Dump trees after partial redundancy elimination. The file name is made
4276 by appending @file{.pre} to the source file name.
4279 @opindex fdump-tree-fre
4280 Dump trees after full redundancy elimination. The file name is made
4281 by appending @file{.fre} to the source file name.
4284 @opindex fdump-tree-copyprop
4285 Dump trees after copy propagation. The file name is made
4286 by appending @file{.copyprop} to the source file name.
4288 @item store_copyprop
4289 @opindex fdump-tree-store_copyprop
4290 Dump trees after store copy-propagation. The file name is made
4291 by appending @file{.store_copyprop} to the source file name.
4294 @opindex fdump-tree-dce
4295 Dump each function after dead code elimination. The file name is made by
4296 appending @file{.dce} to the source file name.
4299 @opindex fdump-tree-mudflap
4300 Dump each function after adding mudflap instrumentation. The file name is
4301 made by appending @file{.mudflap} to the source file name.
4304 @opindex fdump-tree-sra
4305 Dump each function after performing scalar replacement of aggregates. The
4306 file name is made by appending @file{.sra} to the source file name.
4309 @opindex fdump-tree-sink
4310 Dump each function after performing code sinking. The file name is made
4311 by appending @file{.sink} to the source file name.
4314 @opindex fdump-tree-dom
4315 Dump each function after applying dominator tree optimizations. The file
4316 name is made by appending @file{.dom} to the source file name.
4319 @opindex fdump-tree-dse
4320 Dump each function after applying dead store elimination. The file
4321 name is made by appending @file{.dse} to the source file name.
4324 @opindex fdump-tree-phiopt
4325 Dump each function after optimizing PHI nodes into straightline code. The file
4326 name is made by appending @file{.phiopt} to the source file name.
4329 @opindex fdump-tree-forwprop
4330 Dump each function after forward propagating single use variables. The file
4331 name is made by appending @file{.forwprop} to the source file name.
4334 @opindex fdump-tree-copyrename
4335 Dump each function after applying the copy rename optimization. The file
4336 name is made by appending @file{.copyrename} to the source file name.
4339 @opindex fdump-tree-nrv
4340 Dump each function after applying the named return value optimization on
4341 generic trees. The file name is made by appending @file{.nrv} to the source
4345 @opindex fdump-tree-vect
4346 Dump each function after applying vectorization of loops. The file name is
4347 made by appending @file{.vect} to the source file name.
4350 @opindex fdump-tree-vrp
4351 Dump each function after Value Range Propagation (VRP). The file name
4352 is made by appending @file{.vrp} to the source file name.
4355 @opindex fdump-tree-all
4356 Enable all the available tree dumps with the flags provided in this option.
4359 @item -ftree-vectorizer-verbose=@var{n}
4360 @opindex ftree-vectorizer-verbose
4361 This option controls the amount of debugging output the vectorizer prints.
4362 This information is written to standard error, unless
4363 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4364 in which case it is output to the usual dump listing file, @file{.vect}.
4365 For @var{n}=0 no diagnostic information is reported.
4366 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4367 and the total number of loops that got vectorized.
4368 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4369 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4370 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4371 level that @option{-fdump-tree-vect-stats} uses.
4372 Higher verbosity levels mean either more information dumped for each
4373 reported loop, or same amount of information reported for more loops:
4374 If @var{n}=3, alignment related information is added to the reports.
4375 If @var{n}=4, data-references related information (e.g. memory dependences,
4376 memory access-patterns) is added to the reports.
4377 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4378 that did not pass the first analysis phase (i.e. may not be countable, or
4379 may have complicated control-flow).
4380 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4381 For @var{n}=7, all the information the vectorizer generates during its
4382 analysis and transformation is reported. This is the same verbosity level
4383 that @option{-fdump-tree-vect-details} uses.
4385 @item -frandom-seed=@var{string}
4386 @opindex frandom-string
4387 This option provides a seed that GCC uses when it would otherwise use
4388 random numbers. It is used to generate certain symbol names
4389 that have to be different in every compiled file. It is also used to
4390 place unique stamps in coverage data files and the object files that
4391 produce them. You can use the @option{-frandom-seed} option to produce
4392 reproducibly identical object files.
4394 The @var{string} should be different for every file you compile.
4396 @item -fsched-verbose=@var{n}
4397 @opindex fsched-verbose
4398 On targets that use instruction scheduling, this option controls the
4399 amount of debugging output the scheduler prints. This information is
4400 written to standard error, unless @option{-dS} or @option{-dR} is
4401 specified, in which case it is output to the usual dump
4402 listing file, @file{.sched} or @file{.sched2} respectively. However
4403 for @var{n} greater than nine, the output is always printed to standard
4406 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4407 same information as @option{-dRS}. For @var{n} greater than one, it
4408 also output basic block probabilities, detailed ready list information
4409 and unit/insn info. For @var{n} greater than two, it includes RTL
4410 at abort point, control-flow and regions info. And for @var{n} over
4411 four, @option{-fsched-verbose} also includes dependence info.
4415 Store the usual ``temporary'' intermediate files permanently; place them
4416 in the current directory and name them based on the source file. Thus,
4417 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4418 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4419 preprocessed @file{foo.i} output file even though the compiler now
4420 normally uses an integrated preprocessor.
4422 When used in combination with the @option{-x} command line option,
4423 @option{-save-temps} is sensible enough to avoid over writing an
4424 input source file with the same extension as an intermediate file.
4425 The corresponding intermediate file may be obtained by renaming the
4426 source file before using @option{-save-temps}.
4430 Report the CPU time taken by each subprocess in the compilation
4431 sequence. For C source files, this is the compiler proper and assembler
4432 (plus the linker if linking is done). The output looks like this:
4439 The first number on each line is the ``user time'', that is time spent
4440 executing the program itself. The second number is ``system time'',
4441 time spent executing operating system routines on behalf of the program.
4442 Both numbers are in seconds.
4444 @item -fvar-tracking
4445 @opindex fvar-tracking
4446 Run variable tracking pass. It computes where variables are stored at each
4447 position in code. Better debugging information is then generated
4448 (if the debugging information format supports this information).
4450 It is enabled by default when compiling with optimization (@option{-Os},
4451 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4452 the debug info format supports it.
4454 @item -print-file-name=@var{library}
4455 @opindex print-file-name
4456 Print the full absolute name of the library file @var{library} that
4457 would be used when linking---and don't do anything else. With this
4458 option, GCC does not compile or link anything; it just prints the
4461 @item -print-multi-directory
4462 @opindex print-multi-directory
4463 Print the directory name corresponding to the multilib selected by any
4464 other switches present in the command line. This directory is supposed
4465 to exist in @env{GCC_EXEC_PREFIX}.
4467 @item -print-multi-lib
4468 @opindex print-multi-lib
4469 Print the mapping from multilib directory names to compiler switches
4470 that enable them. The directory name is separated from the switches by
4471 @samp{;}, and each switch starts with an @samp{@@} instead of the
4472 @samp{-}, without spaces between multiple switches. This is supposed to
4473 ease shell-processing.
4475 @item -print-prog-name=@var{program}
4476 @opindex print-prog-name
4477 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4479 @item -print-libgcc-file-name
4480 @opindex print-libgcc-file-name
4481 Same as @option{-print-file-name=libgcc.a}.
4483 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4484 but you do want to link with @file{libgcc.a}. You can do
4487 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4490 @item -print-search-dirs
4491 @opindex print-search-dirs
4492 Print the name of the configured installation directory and a list of
4493 program and library directories @command{gcc} will search---and don't do anything else.
4495 This is useful when @command{gcc} prints the error message
4496 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4497 To resolve this you either need to put @file{cpp0} and the other compiler
4498 components where @command{gcc} expects to find them, or you can set the environment
4499 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4500 Don't forget the trailing @samp{/}.
4501 @xref{Environment Variables}.
4504 @opindex dumpmachine
4505 Print the compiler's target machine (for example,
4506 @samp{i686-pc-linux-gnu})---and don't do anything else.
4509 @opindex dumpversion
4510 Print the compiler version (for example, @samp{3.0})---and don't do
4515 Print the compiler's built-in specs---and don't do anything else. (This
4516 is used when GCC itself is being built.) @xref{Spec Files}.
4518 @item -feliminate-unused-debug-types
4519 @opindex feliminate-unused-debug-types
4520 Normally, when producing DWARF2 output, GCC will emit debugging
4521 information for all types declared in a compilation
4522 unit, regardless of whether or not they are actually used
4523 in that compilation unit. Sometimes this is useful, such as
4524 if, in the debugger, you want to cast a value to a type that is
4525 not actually used in your program (but is declared). More often,
4526 however, this results in a significant amount of wasted space.
4527 With this option, GCC will avoid producing debug symbol output
4528 for types that are nowhere used in the source file being compiled.
4531 @node Optimize Options
4532 @section Options That Control Optimization
4533 @cindex optimize options
4534 @cindex options, optimization
4536 These options control various sorts of optimizations.
4538 Without any optimization option, the compiler's goal is to reduce the
4539 cost of compilation and to make debugging produce the expected
4540 results. Statements are independent: if you stop the program with a
4541 breakpoint between statements, you can then assign a new value to any
4542 variable or change the program counter to any other statement in the
4543 function and get exactly the results you would expect from the source
4546 Turning on optimization flags makes the compiler attempt to improve
4547 the performance and/or code size at the expense of compilation time
4548 and possibly the ability to debug the program.
4550 The compiler performs optimization based on the knowledge it has of
4551 the program. Optimization levels @option{-O} and above, in
4552 particular, enable @emph{unit-at-a-time} mode, which allows the
4553 compiler to consider information gained from later functions in
4554 the file when compiling a function. Compiling multiple files at
4555 once to a single output file in @emph{unit-at-a-time} mode allows
4556 the compiler to use information gained from all of the files when
4557 compiling each of them.
4559 Not all optimizations are controlled directly by a flag. Only
4560 optimizations that have a flag are listed.
4567 Optimize. Optimizing compilation takes somewhat more time, and a lot
4568 more memory for a large function.
4570 With @option{-O}, the compiler tries to reduce code size and execution
4571 time, without performing any optimizations that take a great deal of
4574 @option{-O} turns on the following optimization flags:
4575 @gccoptlist{-fdefer-pop @gol
4576 -fdelayed-branch @gol
4577 -fguess-branch-probability @gol
4578 -fcprop-registers @gol
4579 -fif-conversion @gol
4580 -fif-conversion2 @gol
4583 -ftree-dominator-opts @gol
4588 -ftree-copyrename @gol
4591 -funit-at-a-time @gol
4594 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4595 where doing so does not interfere with debugging.
4599 Optimize even more. GCC performs nearly all supported optimizations
4600 that do not involve a space-speed tradeoff. The compiler does not
4601 perform loop unrolling or function inlining when you specify @option{-O2}.
4602 As compared to @option{-O}, this option increases both compilation time
4603 and the performance of the generated code.
4605 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4606 also turns on the following optimization flags:
4607 @gccoptlist{-fthread-jumps @gol
4609 -foptimize-sibling-calls @gol
4610 -fcse-follow-jumps -fcse-skip-blocks @gol
4611 -fgcse -fgcse-lm @gol
4612 -fexpensive-optimizations @gol
4613 -frerun-cse-after-loop @gol
4616 -fschedule-insns -fschedule-insns2 @gol
4617 -fsched-interblock -fsched-spec @gol
4619 -fstrict-aliasing @gol
4620 -fdelete-null-pointer-checks @gol
4621 -freorder-blocks -freorder-functions @gol
4622 -falign-functions -falign-jumps @gol
4623 -falign-loops -falign-labels @gol
4627 Please note the warning under @option{-fgcse} about
4628 invoking @option{-O2} on programs that use computed gotos.
4632 Optimize yet more. @option{-O3} turns on all optimizations specified by
4633 @option{-O2} and also turns on the @option{-finline-functions},
4634 @option{-funswitch-loops} and @option{-fgcse-after-reload} options.
4638 Reduce compilation time and make debugging produce the expected
4639 results. This is the default.
4643 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4644 do not typically increase code size. It also performs further
4645 optimizations designed to reduce code size.
4647 @option{-Os} disables the following optimization flags:
4648 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4649 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
4650 -fprefetch-loop-arrays -ftree-vect-loop-version}
4652 If you use multiple @option{-O} options, with or without level numbers,
4653 the last such option is the one that is effective.
4656 Options of the form @option{-f@var{flag}} specify machine-independent
4657 flags. Most flags have both positive and negative forms; the negative
4658 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4659 below, only one of the forms is listed---the one you typically will
4660 use. You can figure out the other form by either removing @samp{no-}
4663 The following options control specific optimizations. They are either
4664 activated by @option{-O} options or are related to ones that are. You
4665 can use the following flags in the rare cases when ``fine-tuning'' of
4666 optimizations to be performed is desired.
4669 @item -fno-default-inline
4670 @opindex fno-default-inline
4671 Do not make member functions inline by default merely because they are
4672 defined inside the class scope (C++ only). Otherwise, when you specify
4673 @w{@option{-O}}, member functions defined inside class scope are compiled
4674 inline by default; i.e., you don't need to add @samp{inline} in front of
4675 the member function name.
4677 @item -fno-defer-pop
4678 @opindex fno-defer-pop
4679 Always pop the arguments to each function call as soon as that function
4680 returns. For machines which must pop arguments after a function call,
4681 the compiler normally lets arguments accumulate on the stack for several
4682 function calls and pops them all at once.
4684 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4687 @opindex fforce-addr
4688 Force memory address constants to be copied into registers before
4689 doing arithmetic on them.
4691 @item -fforward-propagate
4692 @opindex fforward-propagate
4693 Perform a forward propagation pass on RTL. The pass tries to combine two
4694 instructions and checks if the result can be simplified. If loop unrolling
4695 is active, two passes are performed and the second is scheduled after
4698 This option is enabled by default at optimization levels @option{-O2},
4699 @option{-O3}, @option{-Os}.
4701 @item -fomit-frame-pointer
4702 @opindex fomit-frame-pointer
4703 Don't keep the frame pointer in a register for functions that
4704 don't need one. This avoids the instructions to save, set up and
4705 restore frame pointers; it also makes an extra register available
4706 in many functions. @strong{It also makes debugging impossible on
4709 On some machines, such as the VAX, this flag has no effect, because
4710 the standard calling sequence automatically handles the frame pointer
4711 and nothing is saved by pretending it doesn't exist. The
4712 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4713 whether a target machine supports this flag. @xref{Registers,,Register
4714 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4716 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4718 @item -foptimize-sibling-calls
4719 @opindex foptimize-sibling-calls
4720 Optimize sibling and tail recursive calls.
4722 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4726 Don't pay attention to the @code{inline} keyword. Normally this option
4727 is used to keep the compiler from expanding any functions inline.
4728 Note that if you are not optimizing, no functions can be expanded inline.
4730 @item -finline-functions
4731 @opindex finline-functions
4732 Integrate all simple functions into their callers. The compiler
4733 heuristically decides which functions are simple enough to be worth
4734 integrating in this way.
4736 If all calls to a given function are integrated, and the function is
4737 declared @code{static}, then the function is normally not output as
4738 assembler code in its own right.
4740 Enabled at level @option{-O3}.
4742 @item -finline-functions-called-once
4743 @opindex finline-functions-called-once
4744 Consider all @code{static} functions called once for inlining into their
4745 caller even if they are not marked @code{inline}. If a call to a given
4746 function is integrated, then the function is not output as assembler code
4749 Enabled if @option{-funit-at-a-time} is enabled.
4751 @item -fearly-inlining
4752 @opindex fearly-inlining
4753 Inline functions marked by @code{always_inline} and functions whose body seems
4754 smaller than the function call overhead early before doing
4755 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
4756 makes profiling significantly cheaper and usually inlining faster on programs
4757 having large chains of nested wrapper functions.
4761 @item -finline-limit=@var{n}
4762 @opindex finline-limit
4763 By default, GCC limits the size of functions that can be inlined. This flag
4764 allows the control of this limit for functions that are explicitly marked as
4765 inline (i.e., marked with the inline keyword or defined within the class
4766 definition in c++). @var{n} is the size of functions that can be inlined in
4767 number of pseudo instructions (not counting parameter handling). The default
4768 value of @var{n} is 600.
4769 Increasing this value can result in more inlined code at
4770 the cost of compilation time and memory consumption. Decreasing usually makes
4771 the compilation faster and less code will be inlined (which presumably
4772 means slower programs). This option is particularly useful for programs that
4773 use inlining heavily such as those based on recursive templates with C++.
4775 Inlining is actually controlled by a number of parameters, which may be
4776 specified individually by using @option{--param @var{name}=@var{value}}.
4777 The @option{-finline-limit=@var{n}} option sets some of these parameters
4781 @item max-inline-insns-single
4782 is set to @var{n}/2.
4783 @item max-inline-insns-auto
4784 is set to @var{n}/2.
4785 @item min-inline-insns
4786 is set to 130 or @var{n}/4, whichever is smaller.
4787 @item max-inline-insns-rtl
4791 See below for a documentation of the individual
4792 parameters controlling inlining.
4794 @emph{Note:} pseudo instruction represents, in this particular context, an
4795 abstract measurement of function's size. In no way does it represent a count
4796 of assembly instructions and as such its exact meaning might change from one
4797 release to an another.
4799 @item -fkeep-inline-functions
4800 @opindex fkeep-inline-functions
4801 In C, emit @code{static} functions that are declared @code{inline}
4802 into the object file, even if the function has been inlined into all
4803 of its callers. This switch does not affect functions using the
4804 @code{extern inline} extension in GNU C89@. In C++, emit any and all
4805 inline functions into the object file.
4807 @item -fkeep-static-consts
4808 @opindex fkeep-static-consts
4809 Emit variables declared @code{static const} when optimization isn't turned
4810 on, even if the variables aren't referenced.
4812 GCC enables this option by default. If you want to force the compiler to
4813 check if the variable was referenced, regardless of whether or not
4814 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4816 @item -fmerge-constants
4817 Attempt to merge identical constants (string constants and floating point
4818 constants) across compilation units.
4820 This option is the default for optimized compilation if the assembler and
4821 linker support it. Use @option{-fno-merge-constants} to inhibit this
4824 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4826 @item -fmerge-all-constants
4827 Attempt to merge identical constants and identical variables.
4829 This option implies @option{-fmerge-constants}. In addition to
4830 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4831 arrays or initialized constant variables with integral or floating point
4832 types. Languages like C or C++ require each non-automatic variable to
4833 have distinct location, so using this option will result in non-conforming
4836 @item -fmodulo-sched
4837 @opindex fmodulo-sched
4838 Perform swing modulo scheduling immediately before the first scheduling
4839 pass. This pass looks at innermost loops and reorders their
4840 instructions by overlapping different iterations.
4842 @item -fno-branch-count-reg
4843 @opindex fno-branch-count-reg
4844 Do not use ``decrement and branch'' instructions on a count register,
4845 but instead generate a sequence of instructions that decrement a
4846 register, compare it against zero, then branch based upon the result.
4847 This option is only meaningful on architectures that support such
4848 instructions, which include x86, PowerPC, IA-64 and S/390.
4850 The default is @option{-fbranch-count-reg}.
4852 @item -fno-function-cse
4853 @opindex fno-function-cse
4854 Do not put function addresses in registers; make each instruction that
4855 calls a constant function contain the function's address explicitly.
4857 This option results in less efficient code, but some strange hacks
4858 that alter the assembler output may be confused by the optimizations
4859 performed when this option is not used.
4861 The default is @option{-ffunction-cse}
4863 @item -fno-zero-initialized-in-bss
4864 @opindex fno-zero-initialized-in-bss
4865 If the target supports a BSS section, GCC by default puts variables that
4866 are initialized to zero into BSS@. This can save space in the resulting
4869 This option turns off this behavior because some programs explicitly
4870 rely on variables going to the data section. E.g., so that the
4871 resulting executable can find the beginning of that section and/or make
4872 assumptions based on that.
4874 The default is @option{-fzero-initialized-in-bss}.
4876 @item -fbounds-check
4877 @opindex fbounds-check
4878 For front-ends that support it, generate additional code to check that
4879 indices used to access arrays are within the declared range. This is
4880 currently only supported by the Java and Fortran front-ends, where
4881 this option defaults to true and false respectively.
4883 @item -fmudflap -fmudflapth -fmudflapir
4887 @cindex bounds checking
4889 For front-ends that support it (C and C++), instrument all risky
4890 pointer/array dereferencing operations, some standard library
4891 string/heap functions, and some other associated constructs with
4892 range/validity tests. Modules so instrumented should be immune to
4893 buffer overflows, invalid heap use, and some other classes of C/C++
4894 programming errors. The instrumentation relies on a separate runtime
4895 library (@file{libmudflap}), which will be linked into a program if
4896 @option{-fmudflap} is given at link time. Run-time behavior of the
4897 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4898 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4901 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4902 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4903 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4904 instrumentation should ignore pointer reads. This produces less
4905 instrumentation (and therefore faster execution) and still provides
4906 some protection against outright memory corrupting writes, but allows
4907 erroneously read data to propagate within a program.
4909 @item -fthread-jumps
4910 @opindex fthread-jumps
4911 Perform optimizations where we check to see if a jump branches to a
4912 location where another comparison subsumed by the first is found. If
4913 so, the first branch is redirected to either the destination of the
4914 second branch or a point immediately following it, depending on whether
4915 the condition is known to be true or false.
4917 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4919 @item -fcse-follow-jumps
4920 @opindex fcse-follow-jumps
4921 In common subexpression elimination, scan through jump instructions
4922 when the target of the jump is not reached by any other path. For
4923 example, when CSE encounters an @code{if} statement with an
4924 @code{else} clause, CSE will follow the jump when the condition
4927 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4929 @item -fcse-skip-blocks
4930 @opindex fcse-skip-blocks
4931 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4932 follow jumps which conditionally skip over blocks. When CSE
4933 encounters a simple @code{if} statement with no else clause,
4934 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4935 body of the @code{if}.
4937 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4939 @item -frerun-cse-after-loop
4940 @opindex frerun-cse-after-loop
4941 Re-run common subexpression elimination after loop optimizations has been
4944 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4948 Perform a global common subexpression elimination pass.
4949 This pass also performs global constant and copy propagation.
4951 @emph{Note:} When compiling a program using computed gotos, a GCC
4952 extension, you may get better runtime performance if you disable
4953 the global common subexpression elimination pass by adding
4954 @option{-fno-gcse} to the command line.
4956 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4960 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4961 attempt to move loads which are only killed by stores into themselves. This
4962 allows a loop containing a load/store sequence to be changed to a load outside
4963 the loop, and a copy/store within the loop.
4965 Enabled by default when gcse is enabled.
4969 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4970 global common subexpression elimination. This pass will attempt to move
4971 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4972 loops containing a load/store sequence can be changed to a load before
4973 the loop and a store after the loop.
4975 Not enabled at any optimization level.
4979 When @option{-fgcse-las} is enabled, the global common subexpression
4980 elimination pass eliminates redundant loads that come after stores to the
4981 same memory location (both partial and full redundancies).
4983 Not enabled at any optimization level.
4985 @item -fgcse-after-reload
4986 @opindex fgcse-after-reload
4987 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4988 pass is performed after reload. The purpose of this pass is to cleanup
4991 @item -funsafe-loop-optimizations
4992 @opindex funsafe-loop-optimizations
4993 If given, the loop optimizer will assume that loop indices do not
4994 overflow, and that the loops with nontrivial exit condition are not
4995 infinite. This enables a wider range of loop optimizations even if
4996 the loop optimizer itself cannot prove that these assumptions are valid.
4997 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
4998 if it finds this kind of loop.
5000 @item -fcrossjumping
5001 @opindex crossjumping
5002 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
5003 resulting code may or may not perform better than without cross-jumping.
5005 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5007 @item -fif-conversion
5008 @opindex if-conversion
5009 Attempt to transform conditional jumps into branch-less equivalents. This
5010 include use of conditional moves, min, max, set flags and abs instructions, and
5011 some tricks doable by standard arithmetics. The use of conditional execution
5012 on chips where it is available is controlled by @code{if-conversion2}.
5014 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5016 @item -fif-conversion2
5017 @opindex if-conversion2
5018 Use conditional execution (where available) to transform conditional jumps into
5019 branch-less equivalents.
5021 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5023 @item -fdelete-null-pointer-checks
5024 @opindex fdelete-null-pointer-checks
5025 Use global dataflow analysis to identify and eliminate useless checks
5026 for null pointers. The compiler assumes that dereferencing a null
5027 pointer would have halted the program. If a pointer is checked after
5028 it has already been dereferenced, it cannot be null.
5030 In some environments, this assumption is not true, and programs can
5031 safely dereference null pointers. Use
5032 @option{-fno-delete-null-pointer-checks} to disable this optimization
5033 for programs which depend on that behavior.
5035 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5037 @item -fexpensive-optimizations
5038 @opindex fexpensive-optimizations
5039 Perform a number of minor optimizations that are relatively expensive.
5041 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5043 @item -foptimize-register-move
5045 @opindex foptimize-register-move
5047 Attempt to reassign register numbers in move instructions and as
5048 operands of other simple instructions in order to maximize the amount of
5049 register tying. This is especially helpful on machines with two-operand
5052 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5055 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5057 @item -fdelayed-branch
5058 @opindex fdelayed-branch
5059 If supported for the target machine, attempt to reorder instructions
5060 to exploit instruction slots available after delayed branch
5063 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5065 @item -fschedule-insns
5066 @opindex fschedule-insns
5067 If supported for the target machine, attempt to reorder instructions to
5068 eliminate execution stalls due to required data being unavailable. This
5069 helps machines that have slow floating point or memory load instructions
5070 by allowing other instructions to be issued until the result of the load
5071 or floating point instruction is required.
5073 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5075 @item -fschedule-insns2
5076 @opindex fschedule-insns2
5077 Similar to @option{-fschedule-insns}, but requests an additional pass of
5078 instruction scheduling after register allocation has been done. This is
5079 especially useful on machines with a relatively small number of
5080 registers and where memory load instructions take more than one cycle.
5082 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5084 @item -fno-sched-interblock
5085 @opindex fno-sched-interblock
5086 Don't schedule instructions across basic blocks. This is normally
5087 enabled by default when scheduling before register allocation, i.e.@:
5088 with @option{-fschedule-insns} or at @option{-O2} or higher.
5090 @item -fno-sched-spec
5091 @opindex fno-sched-spec
5092 Don't allow speculative motion of non-load instructions. This is normally
5093 enabled by default when scheduling before register allocation, i.e.@:
5094 with @option{-fschedule-insns} or at @option{-O2} or higher.
5096 @item -fsched-spec-load
5097 @opindex fsched-spec-load
5098 Allow speculative motion of some load instructions. This only makes
5099 sense when scheduling before register allocation, i.e.@: with
5100 @option{-fschedule-insns} or at @option{-O2} or higher.
5102 @item -fsched-spec-load-dangerous
5103 @opindex fsched-spec-load-dangerous
5104 Allow speculative motion of more load instructions. This only makes
5105 sense when scheduling before register allocation, i.e.@: with
5106 @option{-fschedule-insns} or at @option{-O2} or higher.
5108 @item -fsched-stalled-insns=@var{n}
5109 @opindex fsched-stalled-insns
5110 Define how many insns (if any) can be moved prematurely from the queue
5111 of stalled insns into the ready list, during the second scheduling pass.
5113 @item -fsched-stalled-insns-dep=@var{n}
5114 @opindex fsched-stalled-insns-dep
5115 Define how many insn groups (cycles) will be examined for a dependency
5116 on a stalled insn that is candidate for premature removal from the queue
5117 of stalled insns. Has an effect only during the second scheduling pass,
5118 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
5120 @item -fsched2-use-superblocks
5121 @opindex fsched2-use-superblocks
5122 When scheduling after register allocation, do use superblock scheduling
5123 algorithm. Superblock scheduling allows motion across basic block boundaries
5124 resulting on faster schedules. This option is experimental, as not all machine
5125 descriptions used by GCC model the CPU closely enough to avoid unreliable
5126 results from the algorithm.
5128 This only makes sense when scheduling after register allocation, i.e.@: with
5129 @option{-fschedule-insns2} or at @option{-O2} or higher.
5131 @item -fsched2-use-traces
5132 @opindex fsched2-use-traces
5133 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5134 allocation and additionally perform code duplication in order to increase the
5135 size of superblocks using tracer pass. See @option{-ftracer} for details on
5138 This mode should produce faster but significantly longer programs. Also
5139 without @option{-fbranch-probabilities} the traces constructed may not
5140 match the reality and hurt the performance. This only makes
5141 sense when scheduling after register allocation, i.e.@: with
5142 @option{-fschedule-insns2} or at @option{-O2} or higher.
5146 Eliminates redundant extension instructions and move the non redundant
5147 ones to optimal placement using LCM.
5149 @item -freschedule-modulo-scheduled-loops
5150 @opindex fscheduling-in-modulo-scheduled-loops
5151 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
5152 we may want to prevent the later scheduling passes from changing its schedule, we use this
5153 option to control that.
5155 @item -fcaller-saves
5156 @opindex fcaller-saves
5157 Enable values to be allocated in registers that will be clobbered by
5158 function calls, by emitting extra instructions to save and restore the
5159 registers around such calls. Such allocation is done only when it
5160 seems to result in better code than would otherwise be produced.
5162 This option is always enabled by default on certain machines, usually
5163 those which have no call-preserved registers to use instead.
5165 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5168 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
5169 enabled by default at @option{-O2} and @option{-O3}.
5172 Perform Full Redundancy Elimination (FRE) on trees. The difference
5173 between FRE and PRE is that FRE only considers expressions
5174 that are computed on all paths leading to the redundant computation.
5175 This analysis faster than PRE, though it exposes fewer redundancies.
5176 This flag is enabled by default at @option{-O} and higher.
5178 @item -ftree-copy-prop
5179 Perform copy propagation on trees. This pass eliminates unnecessary
5180 copy operations. This flag is enabled by default at @option{-O} and
5183 @item -ftree-store-copy-prop
5184 Perform copy propagation of memory loads and stores. This pass
5185 eliminates unnecessary copy operations in memory references
5186 (structures, global variables, arrays, etc). This flag is enabled by
5187 default at @option{-O2} and higher.
5190 Perform structural alias analysis on trees. This flag
5191 is enabled by default at @option{-O} and higher.
5194 Perform interprocedural pointer analysis.
5197 Perform forward store motion on trees. This flag is
5198 enabled by default at @option{-O} and higher.
5201 Perform sparse conditional constant propagation (CCP) on trees. This
5202 pass only operates on local scalar variables and is enabled by default
5203 at @option{-O} and higher.
5205 @item -ftree-store-ccp
5206 Perform sparse conditional constant propagation (CCP) on trees. This
5207 pass operates on both local scalar variables and memory stores and
5208 loads (global variables, structures, arrays, etc). This flag is
5209 enabled by default at @option{-O2} and higher.
5212 Perform dead code elimination (DCE) on trees. This flag is enabled by
5213 default at @option{-O} and higher.
5215 @item -ftree-dominator-opts
5216 Perform a variety of simple scalar cleanups (constant/copy
5217 propagation, redundancy elimination, range propagation and expression
5218 simplification) based on a dominator tree traversal. This also
5219 performs jump threading (to reduce jumps to jumps). This flag is
5220 enabled by default at @option{-O} and higher.
5223 Perform loop header copying on trees. This is beneficial since it increases
5224 effectiveness of code motion optimizations. It also saves one jump. This flag
5225 is enabled by default at @option{-O} and higher. It is not enabled
5226 for @option{-Os}, since it usually increases code size.
5228 @item -ftree-loop-optimize
5229 Perform loop optimizations on trees. This flag is enabled by default
5230 at @option{-O} and higher.
5232 @item -ftree-loop-linear
5233 Perform linear loop transformations on tree. This flag can improve cache
5234 performance and allow further loop optimizations to take place.
5236 @item -ftree-loop-im
5237 Perform loop invariant motion on trees. This pass moves only invariants that
5238 would be hard to handle at RTL level (function calls, operations that expand to
5239 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5240 operands of conditions that are invariant out of the loop, so that we can use
5241 just trivial invariantness analysis in loop unswitching. The pass also includes
5244 @item -ftree-loop-ivcanon
5245 Create a canonical counter for number of iterations in the loop for that
5246 determining number of iterations requires complicated analysis. Later
5247 optimizations then may determine the number easily. Useful especially
5248 in connection with unrolling.
5251 Perform induction variable optimizations (strength reduction, induction
5252 variable merging and induction variable elimination) on trees.
5255 Perform scalar replacement of aggregates. This pass replaces structure
5256 references with scalars to prevent committing structures to memory too
5257 early. This flag is enabled by default at @option{-O} and higher.
5259 @item -ftree-copyrename
5260 Perform copy renaming on trees. This pass attempts to rename compiler
5261 temporaries to other variables at copy locations, usually resulting in
5262 variable names which more closely resemble the original variables. This flag
5263 is enabled by default at @option{-O} and higher.
5266 Perform temporary expression replacement during the SSA->normal phase. Single
5267 use/single def temporaries are replaced at their use location with their
5268 defining expression. This results in non-GIMPLE code, but gives the expanders
5269 much more complex trees to work on resulting in better RTL generation. This is
5270 enabled by default at @option{-O} and higher.
5273 Perform live range splitting during the SSA->normal phase. Distinct live
5274 ranges of a variable are split into unique variables, allowing for better
5275 optimization later. This is enabled by default at @option{-O} and higher.
5277 @item -ftree-vectorize
5278 Perform loop vectorization on trees.
5280 @item -ftree-vect-loop-version
5281 @opindex ftree-vect-loop-version
5282 Perform loop versioning when doing loop vectorization on trees. When a loop
5283 appears to be vectorizable except that data alignment or data dependence cannot
5284 be determined at compile time then vectorized and non-vectorized versions of
5285 the loop are generated along with runtime checks for alignment or dependence
5286 to control which version is executed. This option is enabled by default
5287 except at level @option{-Os} where it is disabled.
5290 Perform Value Range Propagation on trees. This is similar to the
5291 constant propagation pass, but instead of values, ranges of values are
5292 propagated. This allows the optimizers to remove unnecessary range
5293 checks like array bound checks and null pointer checks. This is
5294 enabled by default at @option{-O2} and higher. Null pointer check
5295 elimination is only done if @option{-fdelete-null-pointer-checks} is
5300 Perform tail duplication to enlarge superblock size. This transformation
5301 simplifies the control flow of the function allowing other optimizations to do
5304 @item -funroll-loops
5305 @opindex funroll-loops
5306 Unroll loops whose number of iterations can be determined at compile
5307 time or upon entry to the loop. @option{-funroll-loops} implies
5308 @option{-frerun-cse-after-loop}. This option makes code larger,
5309 and may or may not make it run faster.
5311 @item -funroll-all-loops
5312 @opindex funroll-all-loops
5313 Unroll all loops, even if their number of iterations is uncertain when
5314 the loop is entered. This usually makes programs run more slowly.
5315 @option{-funroll-all-loops} implies the same options as
5316 @option{-funroll-loops},
5318 @item -fsplit-ivs-in-unroller
5319 @opindex -fsplit-ivs-in-unroller
5320 Enables expressing of values of induction variables in later iterations
5321 of the unrolled loop using the value in the first iteration. This breaks
5322 long dependency chains, thus improving efficiency of the scheduling passes.
5324 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5325 same effect. However in cases the loop body is more complicated than
5326 a single basic block, this is not reliable. It also does not work at all
5327 on some of the architectures due to restrictions in the CSE pass.
5329 This optimization is enabled by default.
5331 @item -fvariable-expansion-in-unroller
5332 @opindex -fvariable-expansion-in-unroller
5333 With this option, the compiler will create multiple copies of some
5334 local variables when unrolling a loop which can result in superior code.
5336 @item -fprefetch-loop-arrays
5337 @opindex fprefetch-loop-arrays
5338 If supported by the target machine, generate instructions to prefetch
5339 memory to improve the performance of loops that access large arrays.
5341 This option may generate better or worse code; results are highly
5342 dependent on the structure of loops within the source code.
5344 Disabled at level @option{-Os}.
5347 @itemx -fno-peephole2
5348 @opindex fno-peephole
5349 @opindex fno-peephole2
5350 Disable any machine-specific peephole optimizations. The difference
5351 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5352 are implemented in the compiler; some targets use one, some use the
5353 other, a few use both.
5355 @option{-fpeephole} is enabled by default.
5356 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5358 @item -fno-guess-branch-probability
5359 @opindex fno-guess-branch-probability
5360 Do not guess branch probabilities using heuristics.
5362 GCC will use heuristics to guess branch probabilities if they are
5363 not provided by profiling feedback (@option{-fprofile-arcs}). These
5364 heuristics are based on the control flow graph. If some branch probabilities
5365 are specified by @samp{__builtin_expect}, then the heuristics will be
5366 used to guess branch probabilities for the rest of the control flow graph,
5367 taking the @samp{__builtin_expect} info into account. The interactions
5368 between the heuristics and @samp{__builtin_expect} can be complex, and in
5369 some cases, it may be useful to disable the heuristics so that the effects
5370 of @samp{__builtin_expect} are easier to understand.
5372 The default is @option{-fguess-branch-probability} at levels
5373 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5375 @item -freorder-blocks
5376 @opindex freorder-blocks
5377 Reorder basic blocks in the compiled function in order to reduce number of
5378 taken branches and improve code locality.
5380 Enabled at levels @option{-O2}, @option{-O3}.
5382 @item -freorder-blocks-and-partition
5383 @opindex freorder-blocks-and-partition
5384 In addition to reordering basic blocks in the compiled function, in order
5385 to reduce number of taken branches, partitions hot and cold basic blocks
5386 into separate sections of the assembly and .o files, to improve
5387 paging and cache locality performance.
5389 This optimization is automatically turned off in the presence of
5390 exception handling, for linkonce sections, for functions with a user-defined
5391 section attribute and on any architecture that does not support named
5394 @item -freorder-functions
5395 @opindex freorder-functions
5396 Reorder functions in the object file in order to
5397 improve code locality. This is implemented by using special
5398 subsections @code{.text.hot} for most frequently executed functions and
5399 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5400 the linker so object file format must support named sections and linker must
5401 place them in a reasonable way.
5403 Also profile feedback must be available in to make this option effective. See
5404 @option{-fprofile-arcs} for details.
5406 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5408 @item -fstrict-aliasing
5409 @opindex fstrict-aliasing
5410 Allows the compiler to assume the strictest aliasing rules applicable to
5411 the language being compiled. For C (and C++), this activates
5412 optimizations based on the type of expressions. In particular, an
5413 object of one type is assumed never to reside at the same address as an
5414 object of a different type, unless the types are almost the same. For
5415 example, an @code{unsigned int} can alias an @code{int}, but not a
5416 @code{void*} or a @code{double}. A character type may alias any other
5419 Pay special attention to code like this:
5432 The practice of reading from a different union member than the one most
5433 recently written to (called ``type-punning'') is common. Even with
5434 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5435 is accessed through the union type. So, the code above will work as
5436 expected. However, this code might not:
5447 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5449 @item -falign-functions
5450 @itemx -falign-functions=@var{n}
5451 @opindex falign-functions
5452 Align the start of functions to the next power-of-two greater than
5453 @var{n}, skipping up to @var{n} bytes. For instance,
5454 @option{-falign-functions=32} aligns functions to the next 32-byte
5455 boundary, but @option{-falign-functions=24} would align to the next
5456 32-byte boundary only if this can be done by skipping 23 bytes or less.
5458 @option{-fno-align-functions} and @option{-falign-functions=1} are
5459 equivalent and mean that functions will not be aligned.
5461 Some assemblers only support this flag when @var{n} is a power of two;
5462 in that case, it is rounded up.
5464 If @var{n} is not specified or is zero, use a machine-dependent default.
5466 Enabled at levels @option{-O2}, @option{-O3}.
5468 @item -falign-labels
5469 @itemx -falign-labels=@var{n}
5470 @opindex falign-labels
5471 Align all branch targets to a power-of-two boundary, skipping up to
5472 @var{n} bytes like @option{-falign-functions}. This option can easily
5473 make code slower, because it must insert dummy operations for when the
5474 branch target is reached in the usual flow of the code.
5476 @option{-fno-align-labels} and @option{-falign-labels=1} are
5477 equivalent and mean that labels will not be aligned.
5479 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
5480 are greater than this value, then their values are used instead.
5482 If @var{n} is not specified or is zero, use a machine-dependent default
5483 which is very likely to be @samp{1}, meaning no alignment.
5485 Enabled at levels @option{-O2}, @option{-O3}.
5488 @itemx -falign-loops=@var{n}
5489 @opindex falign-loops
5490 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
5491 like @option{-falign-functions}. The hope is that the loop will be
5492 executed many times, which will make up for any execution of the dummy
5495 @option{-fno-align-loops} and @option{-falign-loops=1} are
5496 equivalent and mean that loops will not be aligned.
5498 If @var{n} is not specified or is zero, use a machine-dependent default.
5500 Enabled at levels @option{-O2}, @option{-O3}.
5503 @itemx -falign-jumps=@var{n}
5504 @opindex falign-jumps
5505 Align branch targets to a power-of-two boundary, for branch targets
5506 where the targets can only be reached by jumping, skipping up to @var{n}
5507 bytes like @option{-falign-functions}. In this case, no dummy operations
5510 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
5511 equivalent and mean that loops will not be aligned.
5513 If @var{n} is not specified or is zero, use a machine-dependent default.
5515 Enabled at levels @option{-O2}, @option{-O3}.
5517 @item -funit-at-a-time
5518 @opindex funit-at-a-time
5519 Parse the whole compilation unit before starting to produce code.
5520 This allows some extra optimizations to take place but consumes
5521 more memory (in general). There are some compatibility issues
5522 with @emph{unit-at-a-time} mode:
5525 enabling @emph{unit-at-a-time} mode may change the order
5526 in which functions, variables, and top-level @code{asm} statements
5527 are emitted, and will likely break code relying on some particular
5528 ordering. The majority of such top-level @code{asm} statements,
5529 though, can be replaced by @code{section} attributes. The
5530 @option{fno-toplevel-reorder} option may be used to keep the ordering
5531 used in the input file, at the cost of some optimizations.
5534 @emph{unit-at-a-time} mode removes unreferenced static variables
5535 and functions. This may result in undefined references
5536 when an @code{asm} statement refers directly to variables or functions
5537 that are otherwise unused. In that case either the variable/function
5538 shall be listed as an operand of the @code{asm} statement operand or,
5539 in the case of top-level @code{asm} statements the attribute @code{used}
5540 shall be used on the declaration.
5543 Static functions now can use non-standard passing conventions that
5544 may break @code{asm} statements calling functions directly. Again,
5545 attribute @code{used} will prevent this behavior.
5548 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
5549 but this scheme may not be supported by future releases of GCC@.
5551 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5553 @item -fno-toplevel-reorder
5554 Do not reorder top-level functions, variables, and @code{asm}
5555 statements. Output them in the same order that they appear in the
5556 input file. When this option is used, unreferenced static variables
5557 will not be removed. This option is intended to support existing code
5558 which relies on a particular ordering. For new code, it is better to
5563 Constructs webs as commonly used for register allocation purposes and assign
5564 each web individual pseudo register. This allows the register allocation pass
5565 to operate on pseudos directly, but also strengthens several other optimization
5566 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
5567 however, make debugging impossible, since variables will no longer stay in a
5570 Enabled by default with @option{-funroll-loops}.
5572 @item -fwhole-program
5573 @opindex fwhole-program
5574 Assume that the current compilation unit represents whole program being
5575 compiled. All public functions and variables with the exception of @code{main}
5576 and those merged by attribute @code{externally_visible} become static functions
5577 and in a affect gets more aggressively optimized by interprocedural optimizers.
5578 While this option is equivalent to proper use of @code{static} keyword for
5579 programs consisting of single file, in combination with option
5580 @option{--combine} this flag can be used to compile most of smaller scale C
5581 programs since the functions and variables become local for the whole combined
5582 compilation unit, not for the single source file itself.
5585 @item -fno-cprop-registers
5586 @opindex fno-cprop-registers
5587 After register allocation and post-register allocation instruction splitting,
5588 we perform a copy-propagation pass to try to reduce scheduling dependencies
5589 and occasionally eliminate the copy.
5591 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5593 @item -fprofile-generate
5594 @opindex fprofile-generate
5596 Enable options usually used for instrumenting application to produce
5597 profile useful for later recompilation with profile feedback based
5598 optimization. You must use @option{-fprofile-generate} both when
5599 compiling and when linking your program.
5601 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
5604 @opindex fprofile-use
5605 Enable profile feedback directed optimizations, and optimizations
5606 generally profitable only with profile feedback available.
5608 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
5609 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
5613 The following options control compiler behavior regarding floating
5614 point arithmetic. These options trade off between speed and
5615 correctness. All must be specifically enabled.
5619 @opindex ffloat-store
5620 Do not store floating point variables in registers, and inhibit other
5621 options that might change whether a floating point value is taken from a
5624 @cindex floating point precision
5625 This option prevents undesirable excess precision on machines such as
5626 the 68000 where the floating registers (of the 68881) keep more
5627 precision than a @code{double} is supposed to have. Similarly for the
5628 x86 architecture. For most programs, the excess precision does only
5629 good, but a few programs rely on the precise definition of IEEE floating
5630 point. Use @option{-ffloat-store} for such programs, after modifying
5631 them to store all pertinent intermediate computations into variables.
5635 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5636 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5637 @option{-fno-rounding-math}, @option{-fno-signaling-nans},
5638 @option{-fno-signed-zeros} and @option{fcx-limited-range}.
5640 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5642 This option should never be turned on by any @option{-O} option since
5643 it can result in incorrect output for programs which depend on
5644 an exact implementation of IEEE or ISO rules/specifications for
5647 @item -fno-math-errno
5648 @opindex fno-math-errno
5649 Do not set ERRNO after calling math functions that are executed
5650 with a single instruction, e.g., sqrt. A program that relies on
5651 IEEE exceptions for math error handling may want to use this flag
5652 for speed while maintaining IEEE arithmetic compatibility.
5654 This option should never be turned on by any @option{-O} option since
5655 it can result in incorrect output for programs which depend on
5656 an exact implementation of IEEE or ISO rules/specifications for
5659 The default is @option{-fmath-errno}.
5661 On Darwin systems, the math library never sets @code{errno}. There is
5662 therefore no reason for the compiler to consider the possibility that
5663 it might, and @option{-fno-math-errno} is the default.
5665 @item -funsafe-math-optimizations
5666 @opindex funsafe-math-optimizations
5667 Allow optimizations for floating-point arithmetic that (a) assume
5668 that arguments and results are valid and (b) may violate IEEE or
5669 ANSI standards. When used at link-time, it may include libraries
5670 or startup files that change the default FPU control word or other
5671 similar optimizations.
5673 This option should never be turned on by any @option{-O} option since
5674 it can result in incorrect output for programs which depend on
5675 an exact implementation of IEEE or ISO rules/specifications for
5678 The default is @option{-fno-unsafe-math-optimizations}.
5680 @item -ffinite-math-only
5681 @opindex ffinite-math-only
5682 Allow optimizations for floating-point arithmetic that assume
5683 that arguments and results are not NaNs or +-Infs.
5685 This option should never be turned on by any @option{-O} option since
5686 it can result in incorrect output for programs which depend on
5687 an exact implementation of IEEE or ISO rules/specifications.
5689 The default is @option{-fno-finite-math-only}.
5691 @item -fno-signed-zeros
5692 @opindex fno-signed-zeros
5693 Allow optimizations for floating point arithmetic that ignore the
5694 signedness of zero. IEEE arithmetic specifies the behavior of
5695 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
5696 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
5697 This option implies that the sign of a zero result isn't significant.
5699 The default is @option{-fsigned-zeros}.
5701 @item -fno-trapping-math
5702 @opindex fno-trapping-math
5703 Compile code assuming that floating-point operations cannot generate
5704 user-visible traps. These traps include division by zero, overflow,
5705 underflow, inexact result and invalid operation. This option implies
5706 @option{-fno-signaling-nans}. Setting this option may allow faster
5707 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5709 This option should never be turned on by any @option{-O} option since
5710 it can result in incorrect output for programs which depend on
5711 an exact implementation of IEEE or ISO rules/specifications for
5714 The default is @option{-ftrapping-math}.
5716 @item -frounding-math
5717 @opindex frounding-math
5718 Disable transformations and optimizations that assume default floating
5719 point rounding behavior. This is round-to-zero for all floating point
5720 to integer conversions, and round-to-nearest for all other arithmetic
5721 truncations. This option should be specified for programs that change
5722 the FP rounding mode dynamically, or that may be executed with a
5723 non-default rounding mode. This option disables constant folding of
5724 floating point expressions at compile-time (which may be affected by
5725 rounding mode) and arithmetic transformations that are unsafe in the
5726 presence of sign-dependent rounding modes.
5728 The default is @option{-fno-rounding-math}.
5730 This option is experimental and does not currently guarantee to
5731 disable all GCC optimizations that are affected by rounding mode.
5732 Future versions of GCC may provide finer control of this setting
5733 using C99's @code{FENV_ACCESS} pragma. This command line option
5734 will be used to specify the default state for @code{FENV_ACCESS}.
5736 @item -frtl-abstract-sequences
5737 @opindex frtl-abstract-sequences
5738 It is a size optimization method. This option is to find identical
5739 sequences of code, which can be turned into pseudo-procedures and
5740 then replace all occurrences with calls to the newly created
5741 subroutine. It is kind of an opposite of @option{-finline-functions}.
5742 This optimization runs at RTL level.
5744 @item -fsignaling-nans
5745 @opindex fsignaling-nans
5746 Compile code assuming that IEEE signaling NaNs may generate user-visible
5747 traps during floating-point operations. Setting this option disables
5748 optimizations that may change the number of exceptions visible with
5749 signaling NaNs. This option implies @option{-ftrapping-math}.
5751 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5754 The default is @option{-fno-signaling-nans}.
5756 This option is experimental and does not currently guarantee to
5757 disable all GCC optimizations that affect signaling NaN behavior.
5759 @item -fsingle-precision-constant
5760 @opindex fsingle-precision-constant
5761 Treat floating point constant as single precision constant instead of
5762 implicitly converting it to double precision constant.
5764 @item -fcx-limited-range
5765 @itemx -fno-cx-limited-range
5766 @opindex fcx-limited-range
5767 @opindex fno-cx-limited-range
5768 When enabled, this option states that a range reduction step is not
5769 needed when performing complex division. The default is
5770 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
5772 This option controls the default setting of the ISO C99
5773 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
5778 The following options control optimizations that may improve
5779 performance, but are not enabled by any @option{-O} options. This
5780 section includes experimental options that may produce broken code.
5783 @item -fbranch-probabilities
5784 @opindex fbranch-probabilities
5785 After running a program compiled with @option{-fprofile-arcs}
5786 (@pxref{Debugging Options,, Options for Debugging Your Program or
5787 @command{gcc}}), you can compile it a second time using
5788 @option{-fbranch-probabilities}, to improve optimizations based on
5789 the number of times each branch was taken. When the program
5790 compiled with @option{-fprofile-arcs} exits it saves arc execution
5791 counts to a file called @file{@var{sourcename}.gcda} for each source
5792 file. The information in this data file is very dependent on the
5793 structure of the generated code, so you must use the same source code
5794 and the same optimization options for both compilations.
5796 With @option{-fbranch-probabilities}, GCC puts a
5797 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5798 These can be used to improve optimization. Currently, they are only
5799 used in one place: in @file{reorg.c}, instead of guessing which path a
5800 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5801 exactly determine which path is taken more often.
5803 @item -fprofile-values
5804 @opindex fprofile-values
5805 If combined with @option{-fprofile-arcs}, it adds code so that some
5806 data about values of expressions in the program is gathered.
5808 With @option{-fbranch-probabilities}, it reads back the data gathered
5809 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5810 notes to instructions for their later usage in optimizations.
5812 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5816 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5817 a code to gather information about values of expressions.
5819 With @option{-fbranch-probabilities}, it reads back the data gathered
5820 and actually performs the optimizations based on them.
5821 Currently the optimizations include specialization of division operation
5822 using the knowledge about the value of the denominator.
5824 @item -frename-registers
5825 @opindex frename-registers
5826 Attempt to avoid false dependencies in scheduled code by making use
5827 of registers left over after register allocation. This optimization
5828 will most benefit processors with lots of registers. Depending on the
5829 debug information format adopted by the target, however, it can
5830 make debugging impossible, since variables will no longer stay in
5831 a ``home register''.
5833 Enabled by default with @option{-funroll-loops}.
5837 Perform tail duplication to enlarge superblock size. This transformation
5838 simplifies the control flow of the function allowing other optimizations to do
5841 Enabled with @option{-fprofile-use}.
5843 @item -funroll-loops
5844 @opindex funroll-loops
5845 Unroll loops whose number of iterations can be determined at compile time or
5846 upon entry to the loop. @option{-funroll-loops} implies
5847 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
5848 It also turns on complete loop peeling (i.e.@: complete removal of loops with
5849 small constant number of iterations). This option makes code larger, and may
5850 or may not make it run faster.
5852 Enabled with @option{-fprofile-use}.
5854 @item -funroll-all-loops
5855 @opindex funroll-all-loops
5856 Unroll all loops, even if their number of iterations is uncertain when
5857 the loop is entered. This usually makes programs run more slowly.
5858 @option{-funroll-all-loops} implies the same options as
5859 @option{-funroll-loops}.
5862 @opindex fpeel-loops
5863 Peels the loops for that there is enough information that they do not
5864 roll much (from profile feedback). It also turns on complete loop peeling
5865 (i.e.@: complete removal of loops with small constant number of iterations).
5867 Enabled with @option{-fprofile-use}.
5869 @item -fmove-loop-invariants
5870 @opindex fmove-loop-invariants
5871 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
5872 at level @option{-O1}
5874 @item -funswitch-loops
5875 @opindex funswitch-loops
5876 Move branches with loop invariant conditions out of the loop, with duplicates
5877 of the loop on both branches (modified according to result of the condition).
5879 @item -ffunction-sections
5880 @itemx -fdata-sections
5881 @opindex ffunction-sections
5882 @opindex fdata-sections
5883 Place each function or data item into its own section in the output
5884 file if the target supports arbitrary sections. The name of the
5885 function or the name of the data item determines the section's name
5888 Use these options on systems where the linker can perform optimizations
5889 to improve locality of reference in the instruction space. Most systems
5890 using the ELF object format and SPARC processors running Solaris 2 have
5891 linkers with such optimizations. AIX may have these optimizations in
5894 Only use these options when there are significant benefits from doing
5895 so. When you specify these options, the assembler and linker will
5896 create larger object and executable files and will also be slower.
5897 You will not be able to use @code{gprof} on all systems if you
5898 specify this option and you may have problems with debugging if
5899 you specify both this option and @option{-g}.
5901 @item -fbranch-target-load-optimize
5902 @opindex fbranch-target-load-optimize
5903 Perform branch target register load optimization before prologue / epilogue
5905 The use of target registers can typically be exposed only during reload,
5906 thus hoisting loads out of loops and doing inter-block scheduling needs
5907 a separate optimization pass.
5909 @item -fbranch-target-load-optimize2
5910 @opindex fbranch-target-load-optimize2
5911 Perform branch target register load optimization after prologue / epilogue
5914 @item -fbtr-bb-exclusive
5915 @opindex fbtr-bb-exclusive
5916 When performing branch target register load optimization, don't reuse
5917 branch target registers in within any basic block.
5919 @item -fstack-protector
5920 Emit extra code to check for buffer overflows, such as stack smashing
5921 attacks. This is done by adding a guard variable to functions with
5922 vulnerable objects. This includes functions that call alloca, and
5923 functions with buffers larger than 8 bytes. The guards are initialized
5924 when a function is entered and then checked when the function exits.
5925 If a guard check fails, an error message is printed and the program exits.
5927 @item -fstack-protector-all
5928 Like @option{-fstack-protector} except that all functions are protected.
5930 @item -fsection-anchors
5931 @opindex fsection-anchors
5932 Try to reduce the number of symbolic address calculations by using
5933 shared ``anchor'' symbols to address nearby objects. This transformation
5934 can help to reduce the number of GOT entries and GOT accesses on some
5937 For example, the implementation of the following function @code{foo}:
5941 int foo (void) @{ return a + b + c; @}
5944 would usually calculate the addresses of all three variables, but if you
5945 compile it with @option{-fsection-anchors}, it will access the variables
5946 from a common anchor point instead. The effect is similar to the
5947 following pseudocode (which isn't valid C):
5952 register int *xr = &x;
5953 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5957 Not all targets support this option.
5959 @item --param @var{name}=@var{value}
5961 In some places, GCC uses various constants to control the amount of
5962 optimization that is done. For example, GCC will not inline functions
5963 that contain more that a certain number of instructions. You can
5964 control some of these constants on the command-line using the
5965 @option{--param} option.
5967 The names of specific parameters, and the meaning of the values, are
5968 tied to the internals of the compiler, and are subject to change
5969 without notice in future releases.
5971 In each case, the @var{value} is an integer. The allowable choices for
5972 @var{name} are given in the following table:
5975 @item salias-max-implicit-fields
5976 The maximum number of fields in a variable without direct
5977 structure accesses for which structure aliasing will consider trying
5978 to track each field. The default is 5
5980 @item salias-max-array-elements
5981 The maximum number of elements an array can have and its elements
5982 still be tracked individually by structure aliasing. The default is 4
5984 @item sra-max-structure-size
5985 The maximum structure size, in bytes, at which the scalar replacement
5986 of aggregates (SRA) optimization will perform block copies. The
5987 default value, 0, implies that GCC will select the most appropriate
5990 @item sra-field-structure-ratio
5991 The threshold ratio (as a percentage) between instantiated fields and
5992 the complete structure size. We say that if the ratio of the number
5993 of bytes in instantiated fields to the number of bytes in the complete
5994 structure exceeds this parameter, then block copies are not used. The
5997 @item max-crossjump-edges
5998 The maximum number of incoming edges to consider for crossjumping.
5999 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
6000 the number of edges incoming to each block. Increasing values mean
6001 more aggressive optimization, making the compile time increase with
6002 probably small improvement in executable size.
6004 @item min-crossjump-insns
6005 The minimum number of instructions which must be matched at the end
6006 of two blocks before crossjumping will be performed on them. This
6007 value is ignored in the case where all instructions in the block being
6008 crossjumped from are matched. The default value is 5.
6010 @item max-grow-copy-bb-insns
6011 The maximum code size expansion factor when copying basic blocks
6012 instead of jumping. The expansion is relative to a jump instruction.
6013 The default value is 8.
6015 @item max-goto-duplication-insns
6016 The maximum number of instructions to duplicate to a block that jumps
6017 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
6018 passes, GCC factors computed gotos early in the compilation process,
6019 and unfactors them as late as possible. Only computed jumps at the
6020 end of a basic blocks with no more than max-goto-duplication-insns are
6021 unfactored. The default value is 8.
6023 @item max-delay-slot-insn-search
6024 The maximum number of instructions to consider when looking for an
6025 instruction to fill a delay slot. If more than this arbitrary number of
6026 instructions is searched, the time savings from filling the delay slot
6027 will be minimal so stop searching. Increasing values mean more
6028 aggressive optimization, making the compile time increase with probably
6029 small improvement in executable run time.
6031 @item max-delay-slot-live-search
6032 When trying to fill delay slots, the maximum number of instructions to
6033 consider when searching for a block with valid live register
6034 information. Increasing this arbitrarily chosen value means more
6035 aggressive optimization, increasing the compile time. This parameter
6036 should be removed when the delay slot code is rewritten to maintain the
6039 @item max-gcse-memory
6040 The approximate maximum amount of memory that will be allocated in
6041 order to perform the global common subexpression elimination
6042 optimization. If more memory than specified is required, the
6043 optimization will not be done.
6045 @item max-gcse-passes
6046 The maximum number of passes of GCSE to run. The default is 1.
6048 @item max-pending-list-length
6049 The maximum number of pending dependencies scheduling will allow
6050 before flushing the current state and starting over. Large functions
6051 with few branches or calls can create excessively large lists which
6052 needlessly consume memory and resources.
6054 @item max-inline-insns-single
6055 Several parameters control the tree inliner used in gcc.
6056 This number sets the maximum number of instructions (counted in GCC's
6057 internal representation) in a single function that the tree inliner
6058 will consider for inlining. This only affects functions declared
6059 inline and methods implemented in a class declaration (C++).
6060 The default value is 450.
6062 @item max-inline-insns-auto
6063 When you use @option{-finline-functions} (included in @option{-O3}),
6064 a lot of functions that would otherwise not be considered for inlining
6065 by the compiler will be investigated. To those functions, a different
6066 (more restrictive) limit compared to functions declared inline can
6068 The default value is 90.
6070 @item large-function-insns
6071 The limit specifying really large functions. For functions larger than this
6072 limit after inlining inlining is constrained by
6073 @option{--param large-function-growth}. This parameter is useful primarily
6074 to avoid extreme compilation time caused by non-linear algorithms used by the
6076 This parameter is ignored when @option{-funit-at-a-time} is not used.
6077 The default value is 2700.
6079 @item large-function-growth
6080 Specifies maximal growth of large function caused by inlining in percents.
6081 This parameter is ignored when @option{-funit-at-a-time} is not used.
6082 The default value is 100 which limits large function growth to 2.0 times
6085 @item large-unit-insns
6086 The limit specifying large translation unit. Growth caused by inlining of
6087 units larger than this limit is limited by @option{--param inline-unit-growth}.
6088 For small units this might be too tight (consider unit consisting of function A
6089 that is inline and B that just calls A three time. If B is small relative to
6090 A, the growth of unit is 300\% and yet such inlining is very sane. For very
6091 large units consisting of small inlininable functions however the overall unit
6092 growth limit is needed to avoid exponential explosion of code size. Thus for
6093 smaller units, the size is increased to @option{--param large-unit-insns}
6094 before applying @option{--param inline-unit-growth}. The default is 10000
6096 @item inline-unit-growth
6097 Specifies maximal overall growth of the compilation unit caused by inlining.
6098 This parameter is ignored when @option{-funit-at-a-time} is not used.
6099 The default value is 50 which limits unit growth to 1.5 times the original
6102 @item large-stack-frame
6103 The limit specifying large stack frames. While inlining the algorithm is trying
6104 to not grow past this limit too much. Default value is 256 bytes.
6106 @item large-stack-frame-growth
6107 Specifies maximal growth of large stack frames caused by inlining in percents.
6108 The default value is 1000 which limits large stack frame growth to 11 times
6111 @item max-inline-insns-recursive
6112 @itemx max-inline-insns-recursive-auto
6113 Specifies maximum number of instructions out-of-line copy of self recursive inline
6114 function can grow into by performing recursive inlining.
6116 For functions declared inline @option{--param max-inline-insns-recursive} is
6117 taken into account. For function not declared inline, recursive inlining
6118 happens only when @option{-finline-functions} (included in @option{-O3}) is
6119 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
6120 default value is 450.
6122 @item max-inline-recursive-depth
6123 @itemx max-inline-recursive-depth-auto
6124 Specifies maximum recursion depth used by the recursive inlining.
6126 For functions declared inline @option{--param max-inline-recursive-depth} is
6127 taken into account. For function not declared inline, recursive inlining
6128 happens only when @option{-finline-functions} (included in @option{-O3}) is
6129 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6130 default value is 450.
6132 @item min-inline-recursive-probability
6133 Recursive inlining is profitable only for function having deep recursion
6134 in average and can hurt for function having little recursion depth by
6135 increasing the prologue size or complexity of function body to other
6138 When profile feedback is available (see @option{-fprofile-generate}) the actual
6139 recursion depth can be guessed from probability that function will recurse via
6140 given call expression. This parameter limits inlining only to call expression
6141 whose probability exceeds given threshold (in percents). The default value is
6144 @item inline-call-cost
6145 Specify cost of call instruction relative to simple arithmetics operations
6146 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6147 functions and at the same time increases size of leaf function that is believed to
6148 reduce function size by being inlined. In effect it increases amount of
6149 inlining for code having large abstraction penalty (many functions that just
6150 pass the arguments to other functions) and decrease inlining for code with low
6151 abstraction penalty. The default value is 16.
6153 @item min-vect-loop-bound
6154 The minimum number of iterations under which a loop will not get vectorized
6155 when @option{-ftree-vectorize} is used. The number of iterations after
6156 vectorization needs to be greater than the value specified by this option
6157 to allow vectorization. The default value is 0.
6159 @item max-unrolled-insns
6160 The maximum number of instructions that a loop should have if that loop
6161 is unrolled, and if the loop is unrolled, it determines how many times
6162 the loop code is unrolled.
6164 @item max-average-unrolled-insns
6165 The maximum number of instructions biased by probabilities of their execution
6166 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6167 it determines how many times the loop code is unrolled.
6169 @item max-unroll-times
6170 The maximum number of unrollings of a single loop.
6172 @item max-peeled-insns
6173 The maximum number of instructions that a loop should have if that loop
6174 is peeled, and if the loop is peeled, it determines how many times
6175 the loop code is peeled.
6177 @item max-peel-times
6178 The maximum number of peelings of a single loop.
6180 @item max-completely-peeled-insns
6181 The maximum number of insns of a completely peeled loop.
6183 @item max-completely-peel-times
6184 The maximum number of iterations of a loop to be suitable for complete peeling.
6186 @item max-unswitch-insns
6187 The maximum number of insns of an unswitched loop.
6189 @item max-unswitch-level
6190 The maximum number of branches unswitched in a single loop.
6193 The minimum cost of an expensive expression in the loop invariant motion.
6195 @item iv-consider-all-candidates-bound
6196 Bound on number of candidates for induction variables below that
6197 all candidates are considered for each use in induction variable
6198 optimizations. Only the most relevant candidates are considered
6199 if there are more candidates, to avoid quadratic time complexity.
6201 @item iv-max-considered-uses
6202 The induction variable optimizations give up on loops that contain more
6203 induction variable uses.
6205 @item iv-always-prune-cand-set-bound
6206 If number of candidates in the set is smaller than this value,
6207 we always try to remove unnecessary ivs from the set during its
6208 optimization when a new iv is added to the set.
6210 @item scev-max-expr-size
6211 Bound on size of expressions used in the scalar evolutions analyzer.
6212 Large expressions slow the analyzer.
6214 @item vect-max-version-checks
6215 The maximum number of runtime checks that can be performed when doing
6216 loop versioning in the vectorizer. See option ftree-vect-loop-version
6217 for more information.
6219 @item max-iterations-to-track
6221 The maximum number of iterations of a loop the brute force algorithm
6222 for analysis of # of iterations of the loop tries to evaluate.
6224 @item hot-bb-count-fraction
6225 Select fraction of the maximal count of repetitions of basic block in program
6226 given basic block needs to have to be considered hot.
6228 @item hot-bb-frequency-fraction
6229 Select fraction of the maximal frequency of executions of basic block in
6230 function given basic block needs to have to be considered hot
6232 @item max-predicted-iterations
6233 The maximum number of loop iterations we predict statically. This is useful
6234 in cases where function contain single loop with known bound and other loop
6235 with unknown. We predict the known number of iterations correctly, while
6236 the unknown number of iterations average to roughly 10. This means that the
6237 loop without bounds would appear artificially cold relative to the other one.
6239 @item tracer-dynamic-coverage
6240 @itemx tracer-dynamic-coverage-feedback
6242 This value is used to limit superblock formation once the given percentage of
6243 executed instructions is covered. This limits unnecessary code size
6246 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6247 feedback is available. The real profiles (as opposed to statically estimated
6248 ones) are much less balanced allowing the threshold to be larger value.
6250 @item tracer-max-code-growth
6251 Stop tail duplication once code growth has reached given percentage. This is
6252 rather hokey argument, as most of the duplicates will be eliminated later in
6253 cross jumping, so it may be set to much higher values than is the desired code
6256 @item tracer-min-branch-ratio
6258 Stop reverse growth when the reverse probability of best edge is less than this
6259 threshold (in percent).
6261 @item tracer-min-branch-ratio
6262 @itemx tracer-min-branch-ratio-feedback
6264 Stop forward growth if the best edge do have probability lower than this
6267 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6268 compilation for profile feedback and one for compilation without. The value
6269 for compilation with profile feedback needs to be more conservative (higher) in
6270 order to make tracer effective.
6272 @item max-cse-path-length
6274 Maximum number of basic blocks on path that cse considers. The default is 10.
6277 The maximum instructions CSE process before flushing. The default is 1000.
6279 @item max-aliased-vops
6281 Maximum number of virtual operands per statement allowed to represent
6282 aliases before triggering the alias grouping heuristic. Alias
6283 grouping reduces compile times and memory consumption needed for
6284 aliasing at the expense of precision loss in alias information.
6286 @item ggc-min-expand
6288 GCC uses a garbage collector to manage its own memory allocation. This
6289 parameter specifies the minimum percentage by which the garbage
6290 collector's heap should be allowed to expand between collections.
6291 Tuning this may improve compilation speed; it has no effect on code
6294 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6295 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6296 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6297 GCC is not able to calculate RAM on a particular platform, the lower
6298 bound of 30% is used. Setting this parameter and
6299 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6300 every opportunity. This is extremely slow, but can be useful for
6303 @item ggc-min-heapsize
6305 Minimum size of the garbage collector's heap before it begins bothering
6306 to collect garbage. The first collection occurs after the heap expands
6307 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6308 tuning this may improve compilation speed, and has no effect on code
6311 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6312 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6313 with a lower bound of 4096 (four megabytes) and an upper bound of
6314 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6315 particular platform, the lower bound is used. Setting this parameter
6316 very large effectively disables garbage collection. Setting this
6317 parameter and @option{ggc-min-expand} to zero causes a full collection
6318 to occur at every opportunity.
6320 @item max-reload-search-insns
6321 The maximum number of instruction reload should look backward for equivalent
6322 register. Increasing values mean more aggressive optimization, making the
6323 compile time increase with probably slightly better performance. The default
6326 @item max-cselib-memory-locations
6327 The maximum number of memory locations cselib should take into account.
6328 Increasing values mean more aggressive optimization, making the compile time
6329 increase with probably slightly better performance. The default value is 500.
6331 @item max-flow-memory-locations
6332 Similar as @option{max-cselib-memory-locations} but for dataflow liveness.
6333 The default value is 100.
6335 @item reorder-blocks-duplicate
6336 @itemx reorder-blocks-duplicate-feedback
6338 Used by basic block reordering pass to decide whether to use unconditional
6339 branch or duplicate the code on its destination. Code is duplicated when its
6340 estimated size is smaller than this value multiplied by the estimated size of
6341 unconditional jump in the hot spots of the program.
6343 The @option{reorder-block-duplicate-feedback} is used only when profile
6344 feedback is available and may be set to higher values than
6345 @option{reorder-block-duplicate} since information about the hot spots is more
6348 @item max-sched-ready-insns
6349 The maximum number of instructions ready to be issued the scheduler should
6350 consider at any given time during the first scheduling pass. Increasing
6351 values mean more thorough searches, making the compilation time increase
6352 with probably little benefit. The default value is 100.
6354 @item max-sched-region-blocks
6355 The maximum number of blocks in a region to be considered for
6356 interblock scheduling. The default value is 10.
6358 @item max-sched-region-insns
6359 The maximum number of insns in a region to be considered for
6360 interblock scheduling. The default value is 100.
6363 The minimum probability (in percents) of reaching a source block
6364 for interblock speculative scheduling. The default value is 40.
6366 @item max-sched-extend-regions-iters
6367 The maximum number of iterations through CFG to extend regions.
6368 0 - disable region extension,
6369 N - do at most N iterations.
6370 The default value is 0.
6372 @item max-sched-insn-conflict-delay
6373 The maximum conflict delay for an insn to be considered for speculative motion.
6374 The default value is 3.
6376 @item sched-spec-prob-cutoff
6377 The minimal probability of speculation success (in percents), so that
6378 speculative insn will be scheduled.
6379 The default value is 40.
6381 @item max-last-value-rtl
6383 The maximum size measured as number of RTLs that can be recorded in an expression
6384 in combiner for a pseudo register as last known value of that register. The default
6387 @item integer-share-limit
6388 Small integer constants can use a shared data structure, reducing the
6389 compiler's memory usage and increasing its speed. This sets the maximum
6390 value of a shared integer constant's. The default value is 256.
6392 @item min-virtual-mappings
6393 Specifies the minimum number of virtual mappings in the incremental
6394 SSA updater that should be registered to trigger the virtual mappings
6395 heuristic defined by virtual-mappings-ratio. The default value is
6398 @item virtual-mappings-ratio
6399 If the number of virtual mappings is virtual-mappings-ratio bigger
6400 than the number of virtual symbols to be updated, then the incremental
6401 SSA updater switches to a full update for those symbols. The default
6404 @item ssp-buffer-size
6405 The minimum size of buffers (i.e. arrays) that will receive stack smashing
6406 protection when @option{-fstack-protection} is used.
6408 @item max-jump-thread-duplication-stmts
6409 Maximum number of statements allowed in a block that needs to be
6410 duplicated when threading jumps.
6412 @item max-fields-for-field-sensitive
6413 Maximum number of fields in a structure we will treat in
6414 a field sensitive manner during pointer analysis.
6416 @item prefetch-latency
6417 Estimate on average number of instructions that are executed before
6418 prefetch finishes. The distance we prefetch ahead is proportional
6419 to this constant. Increasing this number may also lead to less
6420 streams being prefetched (see @option{simultaneous-prefetches}).
6422 @item simultaneous-prefetches
6423 Maximum number of prefetches that can run at the same time.
6425 @item l1-cache-line-size
6426 The size of cache line in L1 cache, in bytes.
6429 The number of cache lines in L1 cache.
6431 @item verify-canonical-types
6432 Whether the compiler should verify the ``canonical'' types used for
6433 type equality comparisons within the C++ and Objective-C++ front
6434 ends. Set to 1 (the default when GCC is configured with
6435 --enable-checking) to enable verification, 0 to disable verification
6436 (the default when GCC is configured with --disable-checking).
6441 @node Preprocessor Options
6442 @section Options Controlling the Preprocessor
6443 @cindex preprocessor options
6444 @cindex options, preprocessor
6446 These options control the C preprocessor, which is run on each C source
6447 file before actual compilation.
6449 If you use the @option{-E} option, nothing is done except preprocessing.
6450 Some of these options make sense only together with @option{-E} because
6451 they cause the preprocessor output to be unsuitable for actual
6456 You can use @option{-Wp,@var{option}} to bypass the compiler driver
6457 and pass @var{option} directly through to the preprocessor. If
6458 @var{option} contains commas, it is split into multiple options at the
6459 commas. However, many options are modified, translated or interpreted
6460 by the compiler driver before being passed to the preprocessor, and
6461 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
6462 interface is undocumented and subject to change, so whenever possible
6463 you should avoid using @option{-Wp} and let the driver handle the
6466 @item -Xpreprocessor @var{option}
6467 @opindex preprocessor
6468 Pass @var{option} as an option to the preprocessor. You can use this to
6469 supply system-specific preprocessor options which GCC does not know how to
6472 If you want to pass an option that takes an argument, you must use
6473 @option{-Xpreprocessor} twice, once for the option and once for the argument.
6476 @include cppopts.texi
6478 @node Assembler Options
6479 @section Passing Options to the Assembler
6481 @c prevent bad page break with this line
6482 You can pass options to the assembler.
6485 @item -Wa,@var{option}
6487 Pass @var{option} as an option to the assembler. If @var{option}
6488 contains commas, it is split into multiple options at the commas.
6490 @item -Xassembler @var{option}
6492 Pass @var{option} as an option to the assembler. You can use this to
6493 supply system-specific assembler options which GCC does not know how to
6496 If you want to pass an option that takes an argument, you must use
6497 @option{-Xassembler} twice, once for the option and once for the argument.
6502 @section Options for Linking
6503 @cindex link options
6504 @cindex options, linking
6506 These options come into play when the compiler links object files into
6507 an executable output file. They are meaningless if the compiler is
6508 not doing a link step.
6512 @item @var{object-file-name}
6513 A file name that does not end in a special recognized suffix is
6514 considered to name an object file or library. (Object files are
6515 distinguished from libraries by the linker according to the file
6516 contents.) If linking is done, these object files are used as input
6525 If any of these options is used, then the linker is not run, and
6526 object file names should not be used as arguments. @xref{Overall
6530 @item -l@var{library}
6531 @itemx -l @var{library}
6533 Search the library named @var{library} when linking. (The second
6534 alternative with the library as a separate argument is only for
6535 POSIX compliance and is not recommended.)
6537 It makes a difference where in the command you write this option; the
6538 linker searches and processes libraries and object files in the order they
6539 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
6540 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
6541 to functions in @samp{z}, those functions may not be loaded.
6543 The linker searches a standard list of directories for the library,
6544 which is actually a file named @file{lib@var{library}.a}. The linker
6545 then uses this file as if it had been specified precisely by name.
6547 The directories searched include several standard system directories
6548 plus any that you specify with @option{-L}.
6550 Normally the files found this way are library files---archive files
6551 whose members are object files. The linker handles an archive file by
6552 scanning through it for members which define symbols that have so far
6553 been referenced but not defined. But if the file that is found is an
6554 ordinary object file, it is linked in the usual fashion. The only
6555 difference between using an @option{-l} option and specifying a file name
6556 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
6557 and searches several directories.
6561 You need this special case of the @option{-l} option in order to
6562 link an Objective-C or Objective-C++ program.
6565 @opindex nostartfiles
6566 Do not use the standard system startup files when linking.
6567 The standard system libraries are used normally, unless @option{-nostdlib}
6568 or @option{-nodefaultlibs} is used.
6570 @item -nodefaultlibs
6571 @opindex nodefaultlibs
6572 Do not use the standard system libraries when linking.
6573 Only the libraries you specify will be passed to the linker.
6574 The standard startup files are used normally, unless @option{-nostartfiles}
6575 is used. The compiler may generate calls to @code{memcmp},
6576 @code{memset}, @code{memcpy} and @code{memmove}.
6577 These entries are usually resolved by entries in
6578 libc. These entry points should be supplied through some other
6579 mechanism when this option is specified.
6583 Do not use the standard system startup files or libraries when linking.
6584 No startup files and only the libraries you specify will be passed to
6585 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
6586 @code{memcpy} and @code{memmove}.
6587 These entries are usually resolved by entries in
6588 libc. These entry points should be supplied through some other
6589 mechanism when this option is specified.
6591 @cindex @option{-lgcc}, use with @option{-nostdlib}
6592 @cindex @option{-nostdlib} and unresolved references
6593 @cindex unresolved references and @option{-nostdlib}
6594 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
6595 @cindex @option{-nodefaultlibs} and unresolved references
6596 @cindex unresolved references and @option{-nodefaultlibs}
6597 One of the standard libraries bypassed by @option{-nostdlib} and
6598 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
6599 that GCC uses to overcome shortcomings of particular machines, or special
6600 needs for some languages.
6601 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
6602 Collection (GCC) Internals},
6603 for more discussion of @file{libgcc.a}.)
6604 In most cases, you need @file{libgcc.a} even when you want to avoid
6605 other standard libraries. In other words, when you specify @option{-nostdlib}
6606 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
6607 This ensures that you have no unresolved references to internal GCC
6608 library subroutines. (For example, @samp{__main}, used to ensure C++
6609 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
6610 GNU Compiler Collection (GCC) Internals}.)
6614 Produce a position independent executable on targets which support it.
6615 For predictable results, you must also specify the same set of options
6616 that were used to generate code (@option{-fpie}, @option{-fPIE},
6617 or model suboptions) when you specify this option.
6621 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
6622 that support it. This instructs the linker to add all symbols, not
6623 only used ones, to the dynamic symbol table. This option is needed
6624 for some uses of @code{dlopen} or to allow obtaining backtraces
6625 from within a program.
6629 Remove all symbol table and relocation information from the executable.
6633 On systems that support dynamic linking, this prevents linking with the shared
6634 libraries. On other systems, this option has no effect.
6638 Produce a shared object which can then be linked with other objects to
6639 form an executable. Not all systems support this option. For predictable
6640 results, you must also specify the same set of options that were used to
6641 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
6642 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
6643 needs to build supplementary stub code for constructors to work. On
6644 multi-libbed systems, @samp{gcc -shared} must select the correct support
6645 libraries to link against. Failing to supply the correct flags may lead
6646 to subtle defects. Supplying them in cases where they are not necessary
6649 @item -shared-libgcc
6650 @itemx -static-libgcc
6651 @opindex shared-libgcc
6652 @opindex static-libgcc
6653 On systems that provide @file{libgcc} as a shared library, these options
6654 force the use of either the shared or static version respectively.
6655 If no shared version of @file{libgcc} was built when the compiler was
6656 configured, these options have no effect.
6658 There are several situations in which an application should use the
6659 shared @file{libgcc} instead of the static version. The most common
6660 of these is when the application wishes to throw and catch exceptions
6661 across different shared libraries. In that case, each of the libraries
6662 as well as the application itself should use the shared @file{libgcc}.
6664 Therefore, the G++ and GCJ drivers automatically add
6665 @option{-shared-libgcc} whenever you build a shared library or a main
6666 executable, because C++ and Java programs typically use exceptions, so
6667 this is the right thing to do.
6669 If, instead, you use the GCC driver to create shared libraries, you may
6670 find that they will not always be linked with the shared @file{libgcc}.
6671 If GCC finds, at its configuration time, that you have a non-GNU linker
6672 or a GNU linker that does not support option @option{--eh-frame-hdr},
6673 it will link the shared version of @file{libgcc} into shared libraries
6674 by default. Otherwise, it will take advantage of the linker and optimize
6675 away the linking with the shared version of @file{libgcc}, linking with
6676 the static version of libgcc by default. This allows exceptions to
6677 propagate through such shared libraries, without incurring relocation
6678 costs at library load time.
6680 However, if a library or main executable is supposed to throw or catch
6681 exceptions, you must link it using the G++ or GCJ driver, as appropriate
6682 for the languages used in the program, or using the option
6683 @option{-shared-libgcc}, such that it is linked with the shared
6688 Bind references to global symbols when building a shared object. Warn
6689 about any unresolved references (unless overridden by the link editor
6690 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
6693 @item -Xlinker @var{option}
6695 Pass @var{option} as an option to the linker. You can use this to
6696 supply system-specific linker options which GCC does not know how to
6699 If you want to pass an option that takes an argument, you must use
6700 @option{-Xlinker} twice, once for the option and once for the argument.
6701 For example, to pass @option{-assert definitions}, you must write
6702 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
6703 @option{-Xlinker "-assert definitions"}, because this passes the entire
6704 string as a single argument, which is not what the linker expects.
6706 @item -Wl,@var{option}
6708 Pass @var{option} as an option to the linker. If @var{option} contains
6709 commas, it is split into multiple options at the commas.
6711 @item -u @var{symbol}
6713 Pretend the symbol @var{symbol} is undefined, to force linking of
6714 library modules to define it. You can use @option{-u} multiple times with
6715 different symbols to force loading of additional library modules.
6718 @node Directory Options
6719 @section Options for Directory Search
6720 @cindex directory options
6721 @cindex options, directory search
6724 These options specify directories to search for header files, for
6725 libraries and for parts of the compiler:
6730 Add the directory @var{dir} to the head of the list of directories to be
6731 searched for header files. This can be used to override a system header
6732 file, substituting your own version, since these directories are
6733 searched before the system header file directories. However, you should
6734 not use this option to add directories that contain vendor-supplied
6735 system header files (use @option{-isystem} for that). If you use more than
6736 one @option{-I} option, the directories are scanned in left-to-right
6737 order; the standard system directories come after.
6739 If a standard system include directory, or a directory specified with
6740 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
6741 option will be ignored. The directory will still be searched but as a
6742 system directory at its normal position in the system include chain.
6743 This is to ensure that GCC's procedure to fix buggy system headers and
6744 the ordering for the include_next directive are not inadvertently changed.
6745 If you really need to change the search order for system directories,
6746 use the @option{-nostdinc} and/or @option{-isystem} options.
6748 @item -iquote@var{dir}
6750 Add the directory @var{dir} to the head of the list of directories to
6751 be searched for header files only for the case of @samp{#include
6752 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
6753 otherwise just like @option{-I}.
6757 Add directory @var{dir} to the list of directories to be searched
6760 @item -B@var{prefix}
6762 This option specifies where to find the executables, libraries,
6763 include files, and data files of the compiler itself.
6765 The compiler driver program runs one or more of the subprograms
6766 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
6767 @var{prefix} as a prefix for each program it tries to run, both with and
6768 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
6770 For each subprogram to be run, the compiler driver first tries the
6771 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
6772 was not specified, the driver tries two standard prefixes, which are
6773 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
6774 those results in a file name that is found, the unmodified program
6775 name is searched for using the directories specified in your
6776 @env{PATH} environment variable.
6778 The compiler will check to see if the path provided by the @option{-B}
6779 refers to a directory, and if necessary it will add a directory
6780 separator character at the end of the path.
6782 @option{-B} prefixes that effectively specify directory names also apply
6783 to libraries in the linker, because the compiler translates these
6784 options into @option{-L} options for the linker. They also apply to
6785 includes files in the preprocessor, because the compiler translates these
6786 options into @option{-isystem} options for the preprocessor. In this case,
6787 the compiler appends @samp{include} to the prefix.
6789 The run-time support file @file{libgcc.a} can also be searched for using
6790 the @option{-B} prefix, if needed. If it is not found there, the two
6791 standard prefixes above are tried, and that is all. The file is left
6792 out of the link if it is not found by those means.
6794 Another way to specify a prefix much like the @option{-B} prefix is to use
6795 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
6798 As a special kludge, if the path provided by @option{-B} is
6799 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
6800 9, then it will be replaced by @file{[dir/]include}. This is to help
6801 with boot-strapping the compiler.
6803 @item -specs=@var{file}
6805 Process @var{file} after the compiler reads in the standard @file{specs}
6806 file, in order to override the defaults that the @file{gcc} driver
6807 program uses when determining what switches to pass to @file{cc1},
6808 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
6809 @option{-specs=@var{file}} can be specified on the command line, and they
6810 are processed in order, from left to right.
6812 @item --sysroot=@var{dir}
6814 Use @var{dir} as the logical root directory for headers and libraries.
6815 For example, if the compiler would normally search for headers in
6816 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
6817 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
6819 If you use both this option and the @option{-isysroot} option, then
6820 the @option{--sysroot} option will apply to libraries, but the
6821 @option{-isysroot} option will apply to header files.
6823 The GNU linker (beginning with version 2.16) has the necessary support
6824 for this option. If your linker does not support this option, the
6825 header file aspect of @option{--sysroot} will still work, but the
6826 library aspect will not.
6830 This option has been deprecated. Please use @option{-iquote} instead for
6831 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
6832 Any directories you specify with @option{-I} options before the @option{-I-}
6833 option are searched only for the case of @samp{#include "@var{file}"};
6834 they are not searched for @samp{#include <@var{file}>}.
6836 If additional directories are specified with @option{-I} options after
6837 the @option{-I-}, these directories are searched for all @samp{#include}
6838 directives. (Ordinarily @emph{all} @option{-I} directories are used
6841 In addition, the @option{-I-} option inhibits the use of the current
6842 directory (where the current input file came from) as the first search
6843 directory for @samp{#include "@var{file}"}. There is no way to
6844 override this effect of @option{-I-}. With @option{-I.} you can specify
6845 searching the directory which was current when the compiler was
6846 invoked. That is not exactly the same as what the preprocessor does
6847 by default, but it is often satisfactory.
6849 @option{-I-} does not inhibit the use of the standard system directories
6850 for header files. Thus, @option{-I-} and @option{-nostdinc} are
6857 @section Specifying subprocesses and the switches to pass to them
6860 @command{gcc} is a driver program. It performs its job by invoking a
6861 sequence of other programs to do the work of compiling, assembling and
6862 linking. GCC interprets its command-line parameters and uses these to
6863 deduce which programs it should invoke, and which command-line options
6864 it ought to place on their command lines. This behavior is controlled
6865 by @dfn{spec strings}. In most cases there is one spec string for each
6866 program that GCC can invoke, but a few programs have multiple spec
6867 strings to control their behavior. The spec strings built into GCC can
6868 be overridden by using the @option{-specs=} command-line switch to specify
6871 @dfn{Spec files} are plaintext files that are used to construct spec
6872 strings. They consist of a sequence of directives separated by blank
6873 lines. The type of directive is determined by the first non-whitespace
6874 character on the line and it can be one of the following:
6877 @item %@var{command}
6878 Issues a @var{command} to the spec file processor. The commands that can
6882 @item %include <@var{file}>
6884 Search for @var{file} and insert its text at the current point in the
6887 @item %include_noerr <@var{file}>
6888 @cindex %include_noerr
6889 Just like @samp{%include}, but do not generate an error message if the include
6890 file cannot be found.
6892 @item %rename @var{old_name} @var{new_name}
6894 Rename the spec string @var{old_name} to @var{new_name}.
6898 @item *[@var{spec_name}]:
6899 This tells the compiler to create, override or delete the named spec
6900 string. All lines after this directive up to the next directive or
6901 blank line are considered to be the text for the spec string. If this
6902 results in an empty string then the spec will be deleted. (Or, if the
6903 spec did not exist, then nothing will happened.) Otherwise, if the spec
6904 does not currently exist a new spec will be created. If the spec does
6905 exist then its contents will be overridden by the text of this
6906 directive, unless the first character of that text is the @samp{+}
6907 character, in which case the text will be appended to the spec.
6909 @item [@var{suffix}]:
6910 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6911 and up to the next directive or blank line are considered to make up the
6912 spec string for the indicated suffix. When the compiler encounters an
6913 input file with the named suffix, it will processes the spec string in
6914 order to work out how to compile that file. For example:
6921 This says that any input file whose name ends in @samp{.ZZ} should be
6922 passed to the program @samp{z-compile}, which should be invoked with the
6923 command-line switch @option{-input} and with the result of performing the
6924 @samp{%i} substitution. (See below.)
6926 As an alternative to providing a spec string, the text that follows a
6927 suffix directive can be one of the following:
6930 @item @@@var{language}
6931 This says that the suffix is an alias for a known @var{language}. This is
6932 similar to using the @option{-x} command-line switch to GCC to specify a
6933 language explicitly. For example:
6940 Says that .ZZ files are, in fact, C++ source files.
6943 This causes an error messages saying:
6946 @var{name} compiler not installed on this system.
6950 GCC already has an extensive list of suffixes built into it.
6951 This directive will add an entry to the end of the list of suffixes, but
6952 since the list is searched from the end backwards, it is effectively
6953 possible to override earlier entries using this technique.
6957 GCC has the following spec strings built into it. Spec files can
6958 override these strings or create their own. Note that individual
6959 targets can also add their own spec strings to this list.
6962 asm Options to pass to the assembler
6963 asm_final Options to pass to the assembler post-processor
6964 cpp Options to pass to the C preprocessor
6965 cc1 Options to pass to the C compiler
6966 cc1plus Options to pass to the C++ compiler
6967 endfile Object files to include at the end of the link
6968 link Options to pass to the linker
6969 lib Libraries to include on the command line to the linker
6970 libgcc Decides which GCC support library to pass to the linker
6971 linker Sets the name of the linker
6972 predefines Defines to be passed to the C preprocessor
6973 signed_char Defines to pass to CPP to say whether @code{char} is signed
6975 startfile Object files to include at the start of the link
6978 Here is a small example of a spec file:
6984 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6987 This example renames the spec called @samp{lib} to @samp{old_lib} and
6988 then overrides the previous definition of @samp{lib} with a new one.
6989 The new definition adds in some extra command-line options before
6990 including the text of the old definition.
6992 @dfn{Spec strings} are a list of command-line options to be passed to their
6993 corresponding program. In addition, the spec strings can contain
6994 @samp{%}-prefixed sequences to substitute variable text or to
6995 conditionally insert text into the command line. Using these constructs
6996 it is possible to generate quite complex command lines.
6998 Here is a table of all defined @samp{%}-sequences for spec
6999 strings. Note that spaces are not generated automatically around the
7000 results of expanding these sequences. Therefore you can concatenate them
7001 together or combine them with constant text in a single argument.
7005 Substitute one @samp{%} into the program name or argument.
7008 Substitute the name of the input file being processed.
7011 Substitute the basename of the input file being processed.
7012 This is the substring up to (and not including) the last period
7013 and not including the directory.
7016 This is the same as @samp{%b}, but include the file suffix (text after
7020 Marks the argument containing or following the @samp{%d} as a
7021 temporary file name, so that that file will be deleted if GCC exits
7022 successfully. Unlike @samp{%g}, this contributes no text to the
7025 @item %g@var{suffix}
7026 Substitute a file name that has suffix @var{suffix} and is chosen
7027 once per compilation, and mark the argument in the same way as
7028 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
7029 name is now chosen in a way that is hard to predict even when previously
7030 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
7031 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
7032 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
7033 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
7034 was simply substituted with a file name chosen once per compilation,
7035 without regard to any appended suffix (which was therefore treated
7036 just like ordinary text), making such attacks more likely to succeed.
7038 @item %u@var{suffix}
7039 Like @samp{%g}, but generates a new temporary file name even if
7040 @samp{%u@var{suffix}} was already seen.
7042 @item %U@var{suffix}
7043 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
7044 new one if there is no such last file name. In the absence of any
7045 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
7046 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
7047 would involve the generation of two distinct file names, one
7048 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
7049 simply substituted with a file name chosen for the previous @samp{%u},
7050 without regard to any appended suffix.
7052 @item %j@var{suffix}
7053 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
7054 writable, and if save-temps is off; otherwise, substitute the name
7055 of a temporary file, just like @samp{%u}. This temporary file is not
7056 meant for communication between processes, but rather as a junk
7059 @item %|@var{suffix}
7060 @itemx %m@var{suffix}
7061 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
7062 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
7063 all. These are the two most common ways to instruct a program that it
7064 should read from standard input or write to standard output. If you
7065 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
7066 construct: see for example @file{f/lang-specs.h}.
7068 @item %.@var{SUFFIX}
7069 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
7070 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
7071 terminated by the next space or %.
7074 Marks the argument containing or following the @samp{%w} as the
7075 designated output file of this compilation. This puts the argument
7076 into the sequence of arguments that @samp{%o} will substitute later.
7079 Substitutes the names of all the output files, with spaces
7080 automatically placed around them. You should write spaces
7081 around the @samp{%o} as well or the results are undefined.
7082 @samp{%o} is for use in the specs for running the linker.
7083 Input files whose names have no recognized suffix are not compiled
7084 at all, but they are included among the output files, so they will
7088 Substitutes the suffix for object files. Note that this is
7089 handled specially when it immediately follows @samp{%g, %u, or %U},
7090 because of the need for those to form complete file names. The
7091 handling is such that @samp{%O} is treated exactly as if it had already
7092 been substituted, except that @samp{%g, %u, and %U} do not currently
7093 support additional @var{suffix} characters following @samp{%O} as they would
7094 following, for example, @samp{.o}.
7097 Substitutes the standard macro predefinitions for the
7098 current target machine. Use this when running @code{cpp}.
7101 Like @samp{%p}, but puts @samp{__} before and after the name of each
7102 predefined macro, except for macros that start with @samp{__} or with
7103 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
7107 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
7108 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
7109 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
7110 and @option{-imultilib} as necessary.
7113 Current argument is the name of a library or startup file of some sort.
7114 Search for that file in a standard list of directories and substitute
7115 the full name found.
7118 Print @var{str} as an error message. @var{str} is terminated by a newline.
7119 Use this when inconsistent options are detected.
7122 Substitute the contents of spec string @var{name} at this point.
7125 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
7127 @item %x@{@var{option}@}
7128 Accumulate an option for @samp{%X}.
7131 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
7135 Output the accumulated assembler options specified by @option{-Wa}.
7138 Output the accumulated preprocessor options specified by @option{-Wp}.
7141 Process the @code{asm} spec. This is used to compute the
7142 switches to be passed to the assembler.
7145 Process the @code{asm_final} spec. This is a spec string for
7146 passing switches to an assembler post-processor, if such a program is
7150 Process the @code{link} spec. This is the spec for computing the
7151 command line passed to the linker. Typically it will make use of the
7152 @samp{%L %G %S %D and %E} sequences.
7155 Dump out a @option{-L} option for each directory that GCC believes might
7156 contain startup files. If the target supports multilibs then the
7157 current multilib directory will be prepended to each of these paths.
7160 Process the @code{lib} spec. This is a spec string for deciding which
7161 libraries should be included on the command line to the linker.
7164 Process the @code{libgcc} spec. This is a spec string for deciding
7165 which GCC support library should be included on the command line to the linker.
7168 Process the @code{startfile} spec. This is a spec for deciding which
7169 object files should be the first ones passed to the linker. Typically
7170 this might be a file named @file{crt0.o}.
7173 Process the @code{endfile} spec. This is a spec string that specifies
7174 the last object files that will be passed to the linker.
7177 Process the @code{cpp} spec. This is used to construct the arguments
7178 to be passed to the C preprocessor.
7181 Process the @code{cc1} spec. This is used to construct the options to be
7182 passed to the actual C compiler (@samp{cc1}).
7185 Process the @code{cc1plus} spec. This is used to construct the options to be
7186 passed to the actual C++ compiler (@samp{cc1plus}).
7189 Substitute the variable part of a matched option. See below.
7190 Note that each comma in the substituted string is replaced by
7194 Remove all occurrences of @code{-S} from the command line. Note---this
7195 command is position dependent. @samp{%} commands in the spec string
7196 before this one will see @code{-S}, @samp{%} commands in the spec string
7197 after this one will not.
7199 @item %:@var{function}(@var{args})
7200 Call the named function @var{function}, passing it @var{args}.
7201 @var{args} is first processed as a nested spec string, then split
7202 into an argument vector in the usual fashion. The function returns
7203 a string which is processed as if it had appeared literally as part
7204 of the current spec.
7206 The following built-in spec functions are provided:
7209 @item @code{if-exists}
7210 The @code{if-exists} spec function takes one argument, an absolute
7211 pathname to a file. If the file exists, @code{if-exists} returns the
7212 pathname. Here is a small example of its usage:
7216 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7219 @item @code{if-exists-else}
7220 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7221 spec function, except that it takes two arguments. The first argument is
7222 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7223 returns the pathname. If it does not exist, it returns the second argument.
7224 This way, @code{if-exists-else} can be used to select one file or another,
7225 based on the existence of the first. Here is a small example of its usage:
7229 crt0%O%s %:if-exists(crti%O%s) \
7230 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
7233 @item @code{replace-outfile}
7234 The @code{replace-outfile} spec function takes two arguments. It looks for the
7235 first argument in the outfiles array and replaces it with the second argument. Here
7236 is a small example of its usage:
7239 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7245 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7246 If that switch was not specified, this substitutes nothing. Note that
7247 the leading dash is omitted when specifying this option, and it is
7248 automatically inserted if the substitution is performed. Thus the spec
7249 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7250 and would output the command line option @option{-foo}.
7252 @item %W@{@code{S}@}
7253 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7256 @item %@{@code{S}*@}
7257 Substitutes all the switches specified to GCC whose names start
7258 with @code{-S}, but which also take an argument. This is used for
7259 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7260 GCC considers @option{-o foo} as being
7261 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7262 text, including the space. Thus two arguments would be generated.
7264 @item %@{@code{S}*&@code{T}*@}
7265 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7266 (the order of @code{S} and @code{T} in the spec is not significant).
7267 There can be any number of ampersand-separated variables; for each the
7268 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
7270 @item %@{@code{S}:@code{X}@}
7271 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
7273 @item %@{!@code{S}:@code{X}@}
7274 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
7276 @item %@{@code{S}*:@code{X}@}
7277 Substitutes @code{X} if one or more switches whose names start with
7278 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
7279 once, no matter how many such switches appeared. However, if @code{%*}
7280 appears somewhere in @code{X}, then @code{X} will be substituted once
7281 for each matching switch, with the @code{%*} replaced by the part of
7282 that switch that matched the @code{*}.
7284 @item %@{.@code{S}:@code{X}@}
7285 Substitutes @code{X}, if processing a file with suffix @code{S}.
7287 @item %@{!.@code{S}:@code{X}@}
7288 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
7290 @item %@{@code{S}|@code{P}:@code{X}@}
7291 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
7292 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
7293 although they have a stronger binding than the @samp{|}. If @code{%*}
7294 appears in @code{X}, all of the alternatives must be starred, and only
7295 the first matching alternative is substituted.
7297 For example, a spec string like this:
7300 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
7303 will output the following command-line options from the following input
7304 command-line options:
7309 -d fred.c -foo -baz -boggle
7310 -d jim.d -bar -baz -boggle
7313 @item %@{S:X; T:Y; :D@}
7315 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
7316 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
7317 be as many clauses as you need. This may be combined with @code{.},
7318 @code{!}, @code{|}, and @code{*} as needed.
7323 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
7324 construct may contain other nested @samp{%} constructs or spaces, or
7325 even newlines. They are processed as usual, as described above.
7326 Trailing white space in @code{X} is ignored. White space may also
7327 appear anywhere on the left side of the colon in these constructs,
7328 except between @code{.} or @code{*} and the corresponding word.
7330 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
7331 handled specifically in these constructs. If another value of
7332 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7333 @option{-W} switch is found later in the command line, the earlier
7334 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7335 just one letter, which passes all matching options.
7337 The character @samp{|} at the beginning of the predicate text is used to
7338 indicate that a command should be piped to the following command, but
7339 only if @option{-pipe} is specified.
7341 It is built into GCC which switches take arguments and which do not.
7342 (You might think it would be useful to generalize this to allow each
7343 compiler's spec to say which switches take arguments. But this cannot
7344 be done in a consistent fashion. GCC cannot even decide which input
7345 files have been specified without knowing which switches take arguments,
7346 and it must know which input files to compile in order to tell which
7349 GCC also knows implicitly that arguments starting in @option{-l} are to be
7350 treated as compiler output files, and passed to the linker in their
7351 proper position among the other output files.
7353 @c man begin OPTIONS
7355 @node Target Options
7356 @section Specifying Target Machine and Compiler Version
7357 @cindex target options
7358 @cindex cross compiling
7359 @cindex specifying machine version
7360 @cindex specifying compiler version and target machine
7361 @cindex compiler version, specifying
7362 @cindex target machine, specifying
7364 The usual way to run GCC is to run the executable called @file{gcc}, or
7365 @file{<machine>-gcc} when cross-compiling, or
7366 @file{<machine>-gcc-<version>} to run a version other than the one that
7367 was installed last. Sometimes this is inconvenient, so GCC provides
7368 options that will switch to another cross-compiler or version.
7371 @item -b @var{machine}
7373 The argument @var{machine} specifies the target machine for compilation.
7375 The value to use for @var{machine} is the same as was specified as the
7376 machine type when configuring GCC as a cross-compiler. For
7377 example, if a cross-compiler was configured with @samp{configure
7378 arm-elf}, meaning to compile for an arm processor with elf binaries,
7379 then you would specify @option{-b arm-elf} to run that cross compiler.
7380 Because there are other options beginning with @option{-b}, the
7381 configuration must contain a hyphen.
7383 @item -V @var{version}
7385 The argument @var{version} specifies which version of GCC to run.
7386 This is useful when multiple versions are installed. For example,
7387 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
7390 The @option{-V} and @option{-b} options work by running the
7391 @file{<machine>-gcc-<version>} executable, so there's no real reason to
7392 use them if you can just run that directly.
7394 @node Submodel Options
7395 @section Hardware Models and Configurations
7396 @cindex submodel options
7397 @cindex specifying hardware config
7398 @cindex hardware models and configurations, specifying
7399 @cindex machine dependent options
7401 Earlier we discussed the standard option @option{-b} which chooses among
7402 different installed compilers for completely different target
7403 machines, such as VAX vs.@: 68000 vs.@: 80386.
7405 In addition, each of these target machine types can have its own
7406 special options, starting with @samp{-m}, to choose among various
7407 hardware models or configurations---for example, 68010 vs 68020,
7408 floating coprocessor or none. A single installed version of the
7409 compiler can compile for any model or configuration, according to the
7412 Some configurations of the compiler also support additional special
7413 options, usually for compatibility with other compilers on the same
7416 @c This list is ordered alphanumerically by subsection name.
7417 @c It should be the same order and spelling as these options are listed
7418 @c in Machine Dependent Options
7424 * Blackfin Options::
7428 * DEC Alpha Options::
7429 * DEC Alpha/VMS Options::
7431 * GNU/Linux Options::
7434 * i386 and x86-64 Options::
7447 * RS/6000 and PowerPC Options::
7448 * S/390 and zSeries Options::
7453 * System V Options::
7454 * TMS320C3x/C4x Options::
7458 * Xstormy16 Options::
7464 @subsection ARC Options
7467 These options are defined for ARC implementations:
7472 Compile code for little endian mode. This is the default.
7476 Compile code for big endian mode.
7479 @opindex mmangle-cpu
7480 Prepend the name of the cpu to all public symbol names.
7481 In multiple-processor systems, there are many ARC variants with different
7482 instruction and register set characteristics. This flag prevents code
7483 compiled for one cpu to be linked with code compiled for another.
7484 No facility exists for handling variants that are ``almost identical''.
7485 This is an all or nothing option.
7487 @item -mcpu=@var{cpu}
7489 Compile code for ARC variant @var{cpu}.
7490 Which variants are supported depend on the configuration.
7491 All variants support @option{-mcpu=base}, this is the default.
7493 @item -mtext=@var{text-section}
7494 @itemx -mdata=@var{data-section}
7495 @itemx -mrodata=@var{readonly-data-section}
7499 Put functions, data, and readonly data in @var{text-section},
7500 @var{data-section}, and @var{readonly-data-section} respectively
7501 by default. This can be overridden with the @code{section} attribute.
7502 @xref{Variable Attributes}.
7507 @subsection ARM Options
7510 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
7514 @item -mabi=@var{name}
7516 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
7517 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
7520 @opindex mapcs-frame
7521 Generate a stack frame that is compliant with the ARM Procedure Call
7522 Standard for all functions, even if this is not strictly necessary for
7523 correct execution of the code. Specifying @option{-fomit-frame-pointer}
7524 with this option will cause the stack frames not to be generated for
7525 leaf functions. The default is @option{-mno-apcs-frame}.
7529 This is a synonym for @option{-mapcs-frame}.
7532 @c not currently implemented
7533 @item -mapcs-stack-check
7534 @opindex mapcs-stack-check
7535 Generate code to check the amount of stack space available upon entry to
7536 every function (that actually uses some stack space). If there is
7537 insufficient space available then either the function
7538 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
7539 called, depending upon the amount of stack space required. The run time
7540 system is required to provide these functions. The default is
7541 @option{-mno-apcs-stack-check}, since this produces smaller code.
7543 @c not currently implemented
7545 @opindex mapcs-float
7546 Pass floating point arguments using the float point registers. This is
7547 one of the variants of the APCS@. This option is recommended if the
7548 target hardware has a floating point unit or if a lot of floating point
7549 arithmetic is going to be performed by the code. The default is
7550 @option{-mno-apcs-float}, since integer only code is slightly increased in
7551 size if @option{-mapcs-float} is used.
7553 @c not currently implemented
7554 @item -mapcs-reentrant
7555 @opindex mapcs-reentrant
7556 Generate reentrant, position independent code. The default is
7557 @option{-mno-apcs-reentrant}.
7560 @item -mthumb-interwork
7561 @opindex mthumb-interwork
7562 Generate code which supports calling between the ARM and Thumb
7563 instruction sets. Without this option the two instruction sets cannot
7564 be reliably used inside one program. The default is
7565 @option{-mno-thumb-interwork}, since slightly larger code is generated
7566 when @option{-mthumb-interwork} is specified.
7568 @item -mno-sched-prolog
7569 @opindex mno-sched-prolog
7570 Prevent the reordering of instructions in the function prolog, or the
7571 merging of those instruction with the instructions in the function's
7572 body. This means that all functions will start with a recognizable set
7573 of instructions (or in fact one of a choice from a small set of
7574 different function prologues), and this information can be used to
7575 locate the start if functions inside an executable piece of code. The
7576 default is @option{-msched-prolog}.
7579 @opindex mhard-float
7580 Generate output containing floating point instructions. This is the
7584 @opindex msoft-float
7585 Generate output containing library calls for floating point.
7586 @strong{Warning:} the requisite libraries are not available for all ARM
7587 targets. Normally the facilities of the machine's usual C compiler are
7588 used, but this cannot be done directly in cross-compilation. You must make
7589 your own arrangements to provide suitable library functions for
7592 @option{-msoft-float} changes the calling convention in the output file;
7593 therefore, it is only useful if you compile @emph{all} of a program with
7594 this option. In particular, you need to compile @file{libgcc.a}, the
7595 library that comes with GCC, with @option{-msoft-float} in order for
7598 @item -mfloat-abi=@var{name}
7600 Specifies which ABI to use for floating point values. Permissible values
7601 are: @samp{soft}, @samp{softfp} and @samp{hard}.
7603 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
7604 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
7605 of floating point instructions, but still uses the soft-float calling
7608 @item -mlittle-endian
7609 @opindex mlittle-endian
7610 Generate code for a processor running in little-endian mode. This is
7611 the default for all standard configurations.
7614 @opindex mbig-endian
7615 Generate code for a processor running in big-endian mode; the default is
7616 to compile code for a little-endian processor.
7618 @item -mwords-little-endian
7619 @opindex mwords-little-endian
7620 This option only applies when generating code for big-endian processors.
7621 Generate code for a little-endian word order but a big-endian byte
7622 order. That is, a byte order of the form @samp{32107654}. Note: this
7623 option should only be used if you require compatibility with code for
7624 big-endian ARM processors generated by versions of the compiler prior to
7627 @item -mcpu=@var{name}
7629 This specifies the name of the target ARM processor. GCC uses this name
7630 to determine what kind of instructions it can emit when generating
7631 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
7632 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
7633 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
7634 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
7635 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
7636 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
7637 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
7638 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
7639 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
7640 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
7641 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
7642 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
7643 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
7644 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
7645 @samp{cortex-a8}, @samp{cortex-r4}, @samp{cortex-m3},
7646 @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
7648 @itemx -mtune=@var{name}
7650 This option is very similar to the @option{-mcpu=} option, except that
7651 instead of specifying the actual target processor type, and hence
7652 restricting which instructions can be used, it specifies that GCC should
7653 tune the performance of the code as if the target were of the type
7654 specified in this option, but still choosing the instructions that it
7655 will generate based on the cpu specified by a @option{-mcpu=} option.
7656 For some ARM implementations better performance can be obtained by using
7659 @item -march=@var{name}
7661 This specifies the name of the target ARM architecture. GCC uses this
7662 name to determine what kind of instructions it can emit when generating
7663 assembly code. This option can be used in conjunction with or instead
7664 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
7665 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
7666 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
7667 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv7}, @samp{armv7-a},
7668 @samp{armv7-r}, @samp{armv7-m}, @samp{iwmmxt}, @samp{ep9312}.
7670 @item -mfpu=@var{name}
7671 @itemx -mfpe=@var{number}
7672 @itemx -mfp=@var{number}
7676 This specifies what floating point hardware (or hardware emulation) is
7677 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
7678 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
7679 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
7680 with older versions of GCC@.
7682 If @option{-msoft-float} is specified this specifies the format of
7683 floating point values.
7685 @item -mstructure-size-boundary=@var{n}
7686 @opindex mstructure-size-boundary
7687 The size of all structures and unions will be rounded up to a multiple
7688 of the number of bits set by this option. Permissible values are 8, 32
7689 and 64. The default value varies for different toolchains. For the COFF
7690 targeted toolchain the default value is 8. A value of 64 is only allowed
7691 if the underlying ABI supports it.
7693 Specifying the larger number can produce faster, more efficient code, but
7694 can also increase the size of the program. Different values are potentially
7695 incompatible. Code compiled with one value cannot necessarily expect to
7696 work with code or libraries compiled with another value, if they exchange
7697 information using structures or unions.
7699 @item -mabort-on-noreturn
7700 @opindex mabort-on-noreturn
7701 Generate a call to the function @code{abort} at the end of a
7702 @code{noreturn} function. It will be executed if the function tries to
7706 @itemx -mno-long-calls
7707 @opindex mlong-calls
7708 @opindex mno-long-calls
7709 Tells the compiler to perform function calls by first loading the
7710 address of the function into a register and then performing a subroutine
7711 call on this register. This switch is needed if the target function
7712 will lie outside of the 64 megabyte addressing range of the offset based
7713 version of subroutine call instruction.
7715 Even if this switch is enabled, not all function calls will be turned
7716 into long calls. The heuristic is that static functions, functions
7717 which have the @samp{short-call} attribute, functions that are inside
7718 the scope of a @samp{#pragma no_long_calls} directive and functions whose
7719 definitions have already been compiled within the current compilation
7720 unit, will not be turned into long calls. The exception to this rule is
7721 that weak function definitions, functions with the @samp{long-call}
7722 attribute or the @samp{section} attribute, and functions that are within
7723 the scope of a @samp{#pragma long_calls} directive, will always be
7724 turned into long calls.
7726 This feature is not enabled by default. Specifying
7727 @option{-mno-long-calls} will restore the default behavior, as will
7728 placing the function calls within the scope of a @samp{#pragma
7729 long_calls_off} directive. Note these switches have no effect on how
7730 the compiler generates code to handle function calls via function
7733 @item -mnop-fun-dllimport
7734 @opindex mnop-fun-dllimport
7735 Disable support for the @code{dllimport} attribute.
7737 @item -msingle-pic-base
7738 @opindex msingle-pic-base
7739 Treat the register used for PIC addressing as read-only, rather than
7740 loading it in the prologue for each function. The run-time system is
7741 responsible for initializing this register with an appropriate value
7742 before execution begins.
7744 @item -mpic-register=@var{reg}
7745 @opindex mpic-register
7746 Specify the register to be used for PIC addressing. The default is R10
7747 unless stack-checking is enabled, when R9 is used.
7749 @item -mcirrus-fix-invalid-insns
7750 @opindex mcirrus-fix-invalid-insns
7751 @opindex mno-cirrus-fix-invalid-insns
7752 Insert NOPs into the instruction stream to in order to work around
7753 problems with invalid Maverick instruction combinations. This option
7754 is only valid if the @option{-mcpu=ep9312} option has been used to
7755 enable generation of instructions for the Cirrus Maverick floating
7756 point co-processor. This option is not enabled by default, since the
7757 problem is only present in older Maverick implementations. The default
7758 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
7761 @item -mpoke-function-name
7762 @opindex mpoke-function-name
7763 Write the name of each function into the text section, directly
7764 preceding the function prologue. The generated code is similar to this:
7768 .ascii "arm_poke_function_name", 0
7771 .word 0xff000000 + (t1 - t0)
7772 arm_poke_function_name
7774 stmfd sp!, @{fp, ip, lr, pc@}
7778 When performing a stack backtrace, code can inspect the value of
7779 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
7780 location @code{pc - 12} and the top 8 bits are set, then we know that
7781 there is a function name embedded immediately preceding this location
7782 and has length @code{((pc[-3]) & 0xff000000)}.
7786 Generate code for the Thumb instruction set. The default is to
7787 use the 32-bit ARM instruction set.
7788 This option automatically enables either 16-bit Thumb-1 or
7789 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
7790 and @option{-march=@var{name}} options.
7793 @opindex mtpcs-frame
7794 Generate a stack frame that is compliant with the Thumb Procedure Call
7795 Standard for all non-leaf functions. (A leaf function is one that does
7796 not call any other functions.) The default is @option{-mno-tpcs-frame}.
7798 @item -mtpcs-leaf-frame
7799 @opindex mtpcs-leaf-frame
7800 Generate a stack frame that is compliant with the Thumb Procedure Call
7801 Standard for all leaf functions. (A leaf function is one that does
7802 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
7804 @item -mcallee-super-interworking
7805 @opindex mcallee-super-interworking
7806 Gives all externally visible functions in the file being compiled an ARM
7807 instruction set header which switches to Thumb mode before executing the
7808 rest of the function. This allows these functions to be called from
7809 non-interworking code.
7811 @item -mcaller-super-interworking
7812 @opindex mcaller-super-interworking
7813 Allows calls via function pointers (including virtual functions) to
7814 execute correctly regardless of whether the target code has been
7815 compiled for interworking or not. There is a small overhead in the cost
7816 of executing a function pointer if this option is enabled.
7818 @item -mtp=@var{name}
7820 Specify the access model for the thread local storage pointer. The valid
7821 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
7822 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
7823 (supported in the arm6k architecture), and @option{auto}, which uses the
7824 best available method for the selected processor. The default setting is
7830 @subsection AVR Options
7833 These options are defined for AVR implementations:
7836 @item -mmcu=@var{mcu}
7838 Specify ATMEL AVR instruction set or MCU type.
7840 Instruction set avr1 is for the minimal AVR core, not supported by the C
7841 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7842 attiny11, attiny12, attiny15, attiny28).
7844 Instruction set avr2 (default) is for the classic AVR core with up to
7845 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7846 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7847 at90c8534, at90s8535).
7849 Instruction set avr3 is for the classic AVR core with up to 128K program
7850 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
7852 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7853 memory space (MCU types: atmega8, atmega83, atmega85).
7855 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7856 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
7857 atmega64, atmega128, at43usb355, at94k).
7861 Output instruction sizes to the asm file.
7863 @item -minit-stack=@var{N}
7864 @opindex minit-stack
7865 Specify the initial stack address, which may be a symbol or numeric value,
7866 @samp{__stack} is the default.
7868 @item -mno-interrupts
7869 @opindex mno-interrupts
7870 Generated code is not compatible with hardware interrupts.
7871 Code size will be smaller.
7873 @item -mcall-prologues
7874 @opindex mcall-prologues
7875 Functions prologues/epilogues expanded as call to appropriate
7876 subroutines. Code size will be smaller.
7878 @item -mno-tablejump
7879 @opindex mno-tablejump
7880 Do not generate tablejump insns which sometimes increase code size.
7883 @opindex mtiny-stack
7884 Change only the low 8 bits of the stack pointer.
7888 Assume int to be 8 bit integer. This affects the sizes of all types: A
7889 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7890 and long long will be 4 bytes. Please note that this option does not
7891 comply to the C standards, but it will provide you with smaller code
7895 @node Blackfin Options
7896 @subsection Blackfin Options
7897 @cindex Blackfin Options
7900 @item -momit-leaf-frame-pointer
7901 @opindex momit-leaf-frame-pointer
7902 Don't keep the frame pointer in a register for leaf functions. This
7903 avoids the instructions to save, set up and restore frame pointers and
7904 makes an extra register available in leaf functions. The option
7905 @option{-fomit-frame-pointer} removes the frame pointer for all functions
7906 which might make debugging harder.
7908 @item -mspecld-anomaly
7909 @opindex mspecld-anomaly
7910 When enabled, the compiler will ensure that the generated code does not
7911 contain speculative loads after jump instructions. This option is enabled
7914 @item -mno-specld-anomaly
7915 @opindex mno-specld-anomaly
7916 Don't generate extra code to prevent speculative loads from occurring.
7918 @item -mcsync-anomaly
7919 @opindex mcsync-anomaly
7920 When enabled, the compiler will ensure that the generated code does not
7921 contain CSYNC or SSYNC instructions too soon after conditional branches.
7922 This option is enabled by default.
7924 @item -mno-csync-anomaly
7925 @opindex mno-csync-anomaly
7926 Don't generate extra code to prevent CSYNC or SSYNC instructions from
7927 occurring too soon after a conditional branch.
7931 When enabled, the compiler is free to take advantage of the knowledge that
7932 the entire program fits into the low 64k of memory.
7935 @opindex mno-low-64k
7936 Assume that the program is arbitrarily large. This is the default.
7938 @item -mstack-check-l1
7939 @opindex mstack-check-l1
7940 Do stack checking using information placed into L1 scratchpad memory by the
7943 @item -mid-shared-library
7944 @opindex mid-shared-library
7945 Generate code that supports shared libraries via the library ID method.
7946 This allows for execute in place and shared libraries in an environment
7947 without virtual memory management. This option implies @option{-fPIC}.
7949 @item -mno-id-shared-library
7950 @opindex mno-id-shared-library
7951 Generate code that doesn't assume ID based shared libraries are being used.
7952 This is the default.
7954 @item -mleaf-id-shared-library
7955 @opindex mleaf-id-shared-library
7956 Generate code that supports shared libraries via the library ID method,
7957 but assumes that this library or executable won't link against any other
7958 ID shared libraries. That allows the compiler to use faster code for jumps
7961 @item -mno-leaf-id-shared-library
7962 @opindex mno-leaf-id-shared-library
7963 Do not assume that the code being compiled won't link against any ID shared
7964 libraries. Slower code will be generated for jump and call insns.
7966 @item -mshared-library-id=n
7967 @opindex mshared-library-id
7968 Specified the identification number of the ID based shared library being
7969 compiled. Specifying a value of 0 will generate more compact code, specifying
7970 other values will force the allocation of that number to the current
7971 library but is no more space or time efficient than omitting this option.
7975 Generate code that allows the data segment to be located in a different
7976 area of memory from the text segment. This allows for execute in place in
7977 an environment without virtual memory management by eliminating relocations
7978 against the text section.
7981 @opindex mno-sep-data
7982 Generate code that assumes that the data segment follows the text segment.
7983 This is the default.
7986 @itemx -mno-long-calls
7987 @opindex mlong-calls
7988 @opindex mno-long-calls
7989 Tells the compiler to perform function calls by first loading the
7990 address of the function into a register and then performing a subroutine
7991 call on this register. This switch is needed if the target function
7992 will lie outside of the 24 bit addressing range of the offset based
7993 version of subroutine call instruction.
7995 This feature is not enabled by default. Specifying
7996 @option{-mno-long-calls} will restore the default behavior. Note these
7997 switches have no effect on how the compiler generates code to handle
7998 function calls via function pointers.
8002 @subsection CRIS Options
8003 @cindex CRIS Options
8005 These options are defined specifically for the CRIS ports.
8008 @item -march=@var{architecture-type}
8009 @itemx -mcpu=@var{architecture-type}
8012 Generate code for the specified architecture. The choices for
8013 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
8014 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
8015 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
8018 @item -mtune=@var{architecture-type}
8020 Tune to @var{architecture-type} everything applicable about the generated
8021 code, except for the ABI and the set of available instructions. The
8022 choices for @var{architecture-type} are the same as for
8023 @option{-march=@var{architecture-type}}.
8025 @item -mmax-stack-frame=@var{n}
8026 @opindex mmax-stack-frame
8027 Warn when the stack frame of a function exceeds @var{n} bytes.
8029 @item -melinux-stacksize=@var{n}
8030 @opindex melinux-stacksize
8031 Only available with the @samp{cris-axis-aout} target. Arranges for
8032 indications in the program to the kernel loader that the stack of the
8033 program should be set to @var{n} bytes.
8039 The options @option{-metrax4} and @option{-metrax100} are synonyms for
8040 @option{-march=v3} and @option{-march=v8} respectively.
8042 @item -mmul-bug-workaround
8043 @itemx -mno-mul-bug-workaround
8044 @opindex mmul-bug-workaround
8045 @opindex mno-mul-bug-workaround
8046 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
8047 models where it applies. This option is active by default.
8051 Enable CRIS-specific verbose debug-related information in the assembly
8052 code. This option also has the effect to turn off the @samp{#NO_APP}
8053 formatted-code indicator to the assembler at the beginning of the
8058 Do not use condition-code results from previous instruction; always emit
8059 compare and test instructions before use of condition codes.
8061 @item -mno-side-effects
8062 @opindex mno-side-effects
8063 Do not emit instructions with side-effects in addressing modes other than
8067 @itemx -mno-stack-align
8069 @itemx -mno-data-align
8070 @itemx -mconst-align
8071 @itemx -mno-const-align
8072 @opindex mstack-align
8073 @opindex mno-stack-align
8074 @opindex mdata-align
8075 @opindex mno-data-align
8076 @opindex mconst-align
8077 @opindex mno-const-align
8078 These options (no-options) arranges (eliminate arrangements) for the
8079 stack-frame, individual data and constants to be aligned for the maximum
8080 single data access size for the chosen CPU model. The default is to
8081 arrange for 32-bit alignment. ABI details such as structure layout are
8082 not affected by these options.
8090 Similar to the stack- data- and const-align options above, these options
8091 arrange for stack-frame, writable data and constants to all be 32-bit,
8092 16-bit or 8-bit aligned. The default is 32-bit alignment.
8094 @item -mno-prologue-epilogue
8095 @itemx -mprologue-epilogue
8096 @opindex mno-prologue-epilogue
8097 @opindex mprologue-epilogue
8098 With @option{-mno-prologue-epilogue}, the normal function prologue and
8099 epilogue that sets up the stack-frame are omitted and no return
8100 instructions or return sequences are generated in the code. Use this
8101 option only together with visual inspection of the compiled code: no
8102 warnings or errors are generated when call-saved registers must be saved,
8103 or storage for local variable needs to be allocated.
8109 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
8110 instruction sequences that load addresses for functions from the PLT part
8111 of the GOT rather than (traditional on other architectures) calls to the
8112 PLT@. The default is @option{-mgotplt}.
8116 Legacy no-op option only recognized with the cris-axis-aout target.
8120 Legacy no-op option only recognized with the cris-axis-elf and
8121 cris-axis-linux-gnu targets.
8125 Only recognized with the cris-axis-aout target, where it selects a
8126 GNU/linux-like multilib, include files and instruction set for
8131 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
8135 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
8136 to link with input-output functions from a simulator library. Code,
8137 initialized data and zero-initialized data are allocated consecutively.
8141 Like @option{-sim}, but pass linker options to locate initialized data at
8142 0x40000000 and zero-initialized data at 0x80000000.
8146 @subsection CRX Options
8149 These options are defined specifically for the CRX ports.
8155 Enable the use of multiply-accumulate instructions. Disabled by default.
8159 Push instructions will be used to pass outgoing arguments when functions
8160 are called. Enabled by default.
8163 @node Darwin Options
8164 @subsection Darwin Options
8165 @cindex Darwin options
8167 These options are defined for all architectures running the Darwin operating
8170 FSF GCC on Darwin does not create ``fat'' object files; it will create
8171 an object file for the single architecture that it was built to
8172 target. Apple's GCC on Darwin does create ``fat'' files if multiple
8173 @option{-arch} options are used; it does so by running the compiler or
8174 linker multiple times and joining the results together with
8177 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
8178 @samp{i686}) is determined by the flags that specify the ISA
8179 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
8180 @option{-force_cpusubtype_ALL} option can be used to override this.
8182 The Darwin tools vary in their behavior when presented with an ISA
8183 mismatch. The assembler, @file{as}, will only permit instructions to
8184 be used that are valid for the subtype of the file it is generating,
8185 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
8186 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
8187 and print an error if asked to create a shared library with a less
8188 restrictive subtype than its input files (for instance, trying to put
8189 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
8190 for executables, @file{ld}, will quietly give the executable the most
8191 restrictive subtype of any of its input files.
8196 Add the framework directory @var{dir} to the head of the list of
8197 directories to be searched for header files. These directories are
8198 interleaved with those specified by @option{-I} options and are
8199 scanned in a left-to-right order.
8201 A framework directory is a directory with frameworks in it. A
8202 framework is a directory with a @samp{"Headers"} and/or
8203 @samp{"PrivateHeaders"} directory contained directly in it that ends
8204 in @samp{".framework"}. The name of a framework is the name of this
8205 directory excluding the @samp{".framework"}. Headers associated with
8206 the framework are found in one of those two directories, with
8207 @samp{"Headers"} being searched first. A subframework is a framework
8208 directory that is in a framework's @samp{"Frameworks"} directory.
8209 Includes of subframework headers can only appear in a header of a
8210 framework that contains the subframework, or in a sibling subframework
8211 header. Two subframeworks are siblings if they occur in the same
8212 framework. A subframework should not have the same name as a
8213 framework, a warning will be issued if this is violated. Currently a
8214 subframework cannot have subframeworks, in the future, the mechanism
8215 may be extended to support this. The standard frameworks can be found
8216 in @samp{"/System/Library/Frameworks"} and
8217 @samp{"/Library/Frameworks"}. An example include looks like
8218 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
8219 the name of the framework and header.h is found in the
8220 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
8224 Emit debugging information for symbols that are used. For STABS
8225 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
8226 This is by default ON@.
8230 Emit debugging information for all symbols and types.
8232 @item -mmacosx-version-min=@var{version}
8233 The earliest version of MacOS X that this executable will run on
8234 is @var{version}. Typical values of @var{version} include @code{10.1},
8235 @code{10.2}, and @code{10.3.9}.
8237 The default for this option is to make choices that seem to be most
8242 Enable kernel development mode. The @option{-mkernel} option sets
8243 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
8244 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
8245 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
8246 applicable. This mode also sets @option{-mno-altivec},
8247 @option{-msoft-float}, @option{-fno-builtin} and
8248 @option{-mlong-branch} for PowerPC targets.
8250 @item -mone-byte-bool
8251 @opindex -mone-byte-bool
8252 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
8253 By default @samp{sizeof(bool)} is @samp{4} when compiling for
8254 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
8255 option has no effect on x86.
8257 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
8258 to generate code that is not binary compatible with code generated
8259 without that switch. Using this switch may require recompiling all
8260 other modules in a program, including system libraries. Use this
8261 switch to conform to a non-default data model.
8263 @item -mfix-and-continue
8264 @itemx -ffix-and-continue
8265 @itemx -findirect-data
8266 @opindex mfix-and-continue
8267 @opindex ffix-and-continue
8268 @opindex findirect-data
8269 Generate code suitable for fast turn around development. Needed to
8270 enable gdb to dynamically load @code{.o} files into already running
8271 programs. @option{-findirect-data} and @option{-ffix-and-continue}
8272 are provided for backwards compatibility.
8276 Loads all members of static archive libraries.
8277 See man ld(1) for more information.
8279 @item -arch_errors_fatal
8280 @opindex arch_errors_fatal
8281 Cause the errors having to do with files that have the wrong architecture
8285 @opindex bind_at_load
8286 Causes the output file to be marked such that the dynamic linker will
8287 bind all undefined references when the file is loaded or launched.
8291 Produce a Mach-o bundle format file.
8292 See man ld(1) for more information.
8294 @item -bundle_loader @var{executable}
8295 @opindex bundle_loader
8296 This option specifies the @var{executable} that will be loading the build
8297 output file being linked. See man ld(1) for more information.
8300 @opindex -dynamiclib
8301 When passed this option, GCC will produce a dynamic library instead of
8302 an executable when linking, using the Darwin @file{libtool} command.
8304 @item -force_cpusubtype_ALL
8305 @opindex -force_cpusubtype_ALL
8306 This causes GCC's output file to have the @var{ALL} subtype, instead of
8307 one controlled by the @option{-mcpu} or @option{-march} option.
8309 @item -allowable_client @var{client_name}
8311 @itemx -compatibility_version
8312 @itemx -current_version
8314 @itemx -dependency-file
8316 @itemx -dylinker_install_name
8318 @itemx -exported_symbols_list
8320 @itemx -flat_namespace
8321 @itemx -force_flat_namespace
8322 @itemx -headerpad_max_install_names
8325 @itemx -install_name
8326 @itemx -keep_private_externs
8327 @itemx -multi_module
8328 @itemx -multiply_defined
8329 @itemx -multiply_defined_unused
8331 @itemx -no_dead_strip_inits_and_terms
8332 @itemx -nofixprebinding
8335 @itemx -noseglinkedit
8336 @itemx -pagezero_size
8338 @itemx -prebind_all_twolevel_modules
8339 @itemx -private_bundle
8340 @itemx -read_only_relocs
8342 @itemx -sectobjectsymbols
8346 @itemx -sectobjectsymbols
8349 @itemx -segs_read_only_addr
8350 @itemx -segs_read_write_addr
8351 @itemx -seg_addr_table
8352 @itemx -seg_addr_table_filename
8355 @itemx -segs_read_only_addr
8356 @itemx -segs_read_write_addr
8357 @itemx -single_module
8360 @itemx -sub_umbrella
8361 @itemx -twolevel_namespace
8364 @itemx -unexported_symbols_list
8365 @itemx -weak_reference_mismatches
8368 @opindex allowable_client
8369 @opindex client_name
8370 @opindex compatibility_version
8371 @opindex current_version
8373 @opindex dependency-file
8375 @opindex dylinker_install_name
8377 @opindex exported_symbols_list
8379 @opindex flat_namespace
8380 @opindex force_flat_namespace
8381 @opindex headerpad_max_install_names
8384 @opindex install_name
8385 @opindex keep_private_externs
8386 @opindex multi_module
8387 @opindex multiply_defined
8388 @opindex multiply_defined_unused
8390 @opindex no_dead_strip_inits_and_terms
8391 @opindex nofixprebinding
8392 @opindex nomultidefs
8394 @opindex noseglinkedit
8395 @opindex pagezero_size
8397 @opindex prebind_all_twolevel_modules
8398 @opindex private_bundle
8399 @opindex read_only_relocs
8401 @opindex sectobjectsymbols
8405 @opindex sectobjectsymbols
8408 @opindex segs_read_only_addr
8409 @opindex segs_read_write_addr
8410 @opindex seg_addr_table
8411 @opindex seg_addr_table_filename
8412 @opindex seglinkedit
8414 @opindex segs_read_only_addr
8415 @opindex segs_read_write_addr
8416 @opindex single_module
8418 @opindex sub_library
8419 @opindex sub_umbrella
8420 @opindex twolevel_namespace
8423 @opindex unexported_symbols_list
8424 @opindex weak_reference_mismatches
8425 @opindex whatsloaded
8427 These options are passed to the Darwin linker. The Darwin linker man page
8428 describes them in detail.
8431 @node DEC Alpha Options
8432 @subsection DEC Alpha Options
8434 These @samp{-m} options are defined for the DEC Alpha implementations:
8437 @item -mno-soft-float
8439 @opindex mno-soft-float
8440 @opindex msoft-float
8441 Use (do not use) the hardware floating-point instructions for
8442 floating-point operations. When @option{-msoft-float} is specified,
8443 functions in @file{libgcc.a} will be used to perform floating-point
8444 operations. Unless they are replaced by routines that emulate the
8445 floating-point operations, or compiled in such a way as to call such
8446 emulations routines, these routines will issue floating-point
8447 operations. If you are compiling for an Alpha without floating-point
8448 operations, you must ensure that the library is built so as not to call
8451 Note that Alpha implementations without floating-point operations are
8452 required to have floating-point registers.
8457 @opindex mno-fp-regs
8458 Generate code that uses (does not use) the floating-point register set.
8459 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8460 register set is not used, floating point operands are passed in integer
8461 registers as if they were integers and floating-point results are passed
8462 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8463 so any function with a floating-point argument or return value called by code
8464 compiled with @option{-mno-fp-regs} must also be compiled with that
8467 A typical use of this option is building a kernel that does not use,
8468 and hence need not save and restore, any floating-point registers.
8472 The Alpha architecture implements floating-point hardware optimized for
8473 maximum performance. It is mostly compliant with the IEEE floating
8474 point standard. However, for full compliance, software assistance is
8475 required. This option generates code fully IEEE compliant code
8476 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8477 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8478 defined during compilation. The resulting code is less efficient but is
8479 able to correctly support denormalized numbers and exceptional IEEE
8480 values such as not-a-number and plus/minus infinity. Other Alpha
8481 compilers call this option @option{-ieee_with_no_inexact}.
8483 @item -mieee-with-inexact
8484 @opindex mieee-with-inexact
8485 This is like @option{-mieee} except the generated code also maintains
8486 the IEEE @var{inexact-flag}. Turning on this option causes the
8487 generated code to implement fully-compliant IEEE math. In addition to
8488 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8489 macro. On some Alpha implementations the resulting code may execute
8490 significantly slower than the code generated by default. Since there is
8491 very little code that depends on the @var{inexact-flag}, you should
8492 normally not specify this option. Other Alpha compilers call this
8493 option @option{-ieee_with_inexact}.
8495 @item -mfp-trap-mode=@var{trap-mode}
8496 @opindex mfp-trap-mode
8497 This option controls what floating-point related traps are enabled.
8498 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8499 The trap mode can be set to one of four values:
8503 This is the default (normal) setting. The only traps that are enabled
8504 are the ones that cannot be disabled in software (e.g., division by zero
8508 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8512 Like @samp{u}, but the instructions are marked to be safe for software
8513 completion (see Alpha architecture manual for details).
8516 Like @samp{su}, but inexact traps are enabled as well.
8519 @item -mfp-rounding-mode=@var{rounding-mode}
8520 @opindex mfp-rounding-mode
8521 Selects the IEEE rounding mode. Other Alpha compilers call this option
8522 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8527 Normal IEEE rounding mode. Floating point numbers are rounded towards
8528 the nearest machine number or towards the even machine number in case
8532 Round towards minus infinity.
8535 Chopped rounding mode. Floating point numbers are rounded towards zero.
8538 Dynamic rounding mode. A field in the floating point control register
8539 (@var{fpcr}, see Alpha architecture reference manual) controls the
8540 rounding mode in effect. The C library initializes this register for
8541 rounding towards plus infinity. Thus, unless your program modifies the
8542 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8545 @item -mtrap-precision=@var{trap-precision}
8546 @opindex mtrap-precision
8547 In the Alpha architecture, floating point traps are imprecise. This
8548 means without software assistance it is impossible to recover from a
8549 floating trap and program execution normally needs to be terminated.
8550 GCC can generate code that can assist operating system trap handlers
8551 in determining the exact location that caused a floating point trap.
8552 Depending on the requirements of an application, different levels of
8553 precisions can be selected:
8557 Program precision. This option is the default and means a trap handler
8558 can only identify which program caused a floating point exception.
8561 Function precision. The trap handler can determine the function that
8562 caused a floating point exception.
8565 Instruction precision. The trap handler can determine the exact
8566 instruction that caused a floating point exception.
8569 Other Alpha compilers provide the equivalent options called
8570 @option{-scope_safe} and @option{-resumption_safe}.
8572 @item -mieee-conformant
8573 @opindex mieee-conformant
8574 This option marks the generated code as IEEE conformant. You must not
8575 use this option unless you also specify @option{-mtrap-precision=i} and either
8576 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8577 is to emit the line @samp{.eflag 48} in the function prologue of the
8578 generated assembly file. Under DEC Unix, this has the effect that
8579 IEEE-conformant math library routines will be linked in.
8581 @item -mbuild-constants
8582 @opindex mbuild-constants
8583 Normally GCC examines a 32- or 64-bit integer constant to
8584 see if it can construct it from smaller constants in two or three
8585 instructions. If it cannot, it will output the constant as a literal and
8586 generate code to load it from the data segment at runtime.
8588 Use this option to require GCC to construct @emph{all} integer constants
8589 using code, even if it takes more instructions (the maximum is six).
8591 You would typically use this option to build a shared library dynamic
8592 loader. Itself a shared library, it must relocate itself in memory
8593 before it can find the variables and constants in its own data segment.
8599 Select whether to generate code to be assembled by the vendor-supplied
8600 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8618 Indicate whether GCC should generate code to use the optional BWX,
8619 CIX, FIX and MAX instruction sets. The default is to use the instruction
8620 sets supported by the CPU type specified via @option{-mcpu=} option or that
8621 of the CPU on which GCC was built if none was specified.
8626 @opindex mfloat-ieee
8627 Generate code that uses (does not use) VAX F and G floating point
8628 arithmetic instead of IEEE single and double precision.
8630 @item -mexplicit-relocs
8631 @itemx -mno-explicit-relocs
8632 @opindex mexplicit-relocs
8633 @opindex mno-explicit-relocs
8634 Older Alpha assemblers provided no way to generate symbol relocations
8635 except via assembler macros. Use of these macros does not allow
8636 optimal instruction scheduling. GNU binutils as of version 2.12
8637 supports a new syntax that allows the compiler to explicitly mark
8638 which relocations should apply to which instructions. This option
8639 is mostly useful for debugging, as GCC detects the capabilities of
8640 the assembler when it is built and sets the default accordingly.
8644 @opindex msmall-data
8645 @opindex mlarge-data
8646 When @option{-mexplicit-relocs} is in effect, static data is
8647 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8648 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8649 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8650 16-bit relocations off of the @code{$gp} register. This limits the
8651 size of the small data area to 64KB, but allows the variables to be
8652 directly accessed via a single instruction.
8654 The default is @option{-mlarge-data}. With this option the data area
8655 is limited to just below 2GB@. Programs that require more than 2GB of
8656 data must use @code{malloc} or @code{mmap} to allocate the data in the
8657 heap instead of in the program's data segment.
8659 When generating code for shared libraries, @option{-fpic} implies
8660 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8664 @opindex msmall-text
8665 @opindex mlarge-text
8666 When @option{-msmall-text} is used, the compiler assumes that the
8667 code of the entire program (or shared library) fits in 4MB, and is
8668 thus reachable with a branch instruction. When @option{-msmall-data}
8669 is used, the compiler can assume that all local symbols share the
8670 same @code{$gp} value, and thus reduce the number of instructions
8671 required for a function call from 4 to 1.
8673 The default is @option{-mlarge-text}.
8675 @item -mcpu=@var{cpu_type}
8677 Set the instruction set and instruction scheduling parameters for
8678 machine type @var{cpu_type}. You can specify either the @samp{EV}
8679 style name or the corresponding chip number. GCC supports scheduling
8680 parameters for the EV4, EV5 and EV6 family of processors and will
8681 choose the default values for the instruction set from the processor
8682 you specify. If you do not specify a processor type, GCC will default
8683 to the processor on which the compiler was built.
8685 Supported values for @var{cpu_type} are
8691 Schedules as an EV4 and has no instruction set extensions.
8695 Schedules as an EV5 and has no instruction set extensions.
8699 Schedules as an EV5 and supports the BWX extension.
8704 Schedules as an EV5 and supports the BWX and MAX extensions.
8708 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8712 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8715 @item -mtune=@var{cpu_type}
8717 Set only the instruction scheduling parameters for machine type
8718 @var{cpu_type}. The instruction set is not changed.
8720 @item -mmemory-latency=@var{time}
8721 @opindex mmemory-latency
8722 Sets the latency the scheduler should assume for typical memory
8723 references as seen by the application. This number is highly
8724 dependent on the memory access patterns used by the application
8725 and the size of the external cache on the machine.
8727 Valid options for @var{time} are
8731 A decimal number representing clock cycles.
8737 The compiler contains estimates of the number of clock cycles for
8738 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8739 (also called Dcache, Scache, and Bcache), as well as to main memory.
8740 Note that L3 is only valid for EV5.
8745 @node DEC Alpha/VMS Options
8746 @subsection DEC Alpha/VMS Options
8748 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8751 @item -mvms-return-codes
8752 @opindex mvms-return-codes
8753 Return VMS condition codes from main. The default is to return POSIX
8754 style condition (e.g.@ error) codes.
8758 @subsection FRV Options
8765 Only use the first 32 general purpose registers.
8770 Use all 64 general purpose registers.
8775 Use only the first 32 floating point registers.
8780 Use all 64 floating point registers
8783 @opindex mhard-float
8785 Use hardware instructions for floating point operations.
8788 @opindex msoft-float
8790 Use library routines for floating point operations.
8795 Dynamically allocate condition code registers.
8800 Do not try to dynamically allocate condition code registers, only
8801 use @code{icc0} and @code{fcc0}.
8806 Change ABI to use double word insns.
8811 Do not use double word instructions.
8816 Use floating point double instructions.
8821 Do not use floating point double instructions.
8826 Use media instructions.
8831 Do not use media instructions.
8836 Use multiply and add/subtract instructions.
8841 Do not use multiply and add/subtract instructions.
8846 Select the FDPIC ABI, that uses function descriptors to represent
8847 pointers to functions. Without any PIC/PIE-related options, it
8848 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
8849 assumes GOT entries and small data are within a 12-bit range from the
8850 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
8851 are computed with 32 bits.
8854 @opindex minline-plt
8856 Enable inlining of PLT entries in function calls to functions that are
8857 not known to bind locally. It has no effect without @option{-mfdpic}.
8858 It's enabled by default if optimizing for speed and compiling for
8859 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
8860 optimization option such as @option{-O3} or above is present in the
8866 Assume a large TLS segment when generating thread-local code.
8871 Do not assume a large TLS segment when generating thread-local code.
8876 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
8877 that is known to be in read-only sections. It's enabled by default,
8878 except for @option{-fpic} or @option{-fpie}: even though it may help
8879 make the global offset table smaller, it trades 1 instruction for 4.
8880 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
8881 one of which may be shared by multiple symbols, and it avoids the need
8882 for a GOT entry for the referenced symbol, so it's more likely to be a
8883 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
8885 @item -multilib-library-pic
8886 @opindex multilib-library-pic
8888 Link with the (library, not FD) pic libraries. It's implied by
8889 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
8890 @option{-fpic} without @option{-mfdpic}. You should never have to use
8896 Follow the EABI requirement of always creating a frame pointer whenever
8897 a stack frame is allocated. This option is enabled by default and can
8898 be disabled with @option{-mno-linked-fp}.
8901 @opindex mlong-calls
8903 Use indirect addressing to call functions outside the current
8904 compilation unit. This allows the functions to be placed anywhere
8905 within the 32-bit address space.
8907 @item -malign-labels
8908 @opindex malign-labels
8910 Try to align labels to an 8-byte boundary by inserting nops into the
8911 previous packet. This option only has an effect when VLIW packing
8912 is enabled. It doesn't create new packets; it merely adds nops to
8916 @opindex mlibrary-pic
8918 Generate position-independent EABI code.
8923 Use only the first four media accumulator registers.
8928 Use all eight media accumulator registers.
8933 Pack VLIW instructions.
8938 Do not pack VLIW instructions.
8943 Do not mark ABI switches in e_flags.
8948 Enable the use of conditional-move instructions (default).
8950 This switch is mainly for debugging the compiler and will likely be removed
8951 in a future version.
8953 @item -mno-cond-move
8954 @opindex mno-cond-move
8956 Disable the use of conditional-move instructions.
8958 This switch is mainly for debugging the compiler and will likely be removed
8959 in a future version.
8964 Enable the use of conditional set instructions (default).
8966 This switch is mainly for debugging the compiler and will likely be removed
8967 in a future version.
8972 Disable the use of conditional set instructions.
8974 This switch is mainly for debugging the compiler and will likely be removed
8975 in a future version.
8980 Enable the use of conditional execution (default).
8982 This switch is mainly for debugging the compiler and will likely be removed
8983 in a future version.
8985 @item -mno-cond-exec
8986 @opindex mno-cond-exec
8988 Disable the use of conditional execution.
8990 This switch is mainly for debugging the compiler and will likely be removed
8991 in a future version.
8994 @opindex mvliw-branch
8996 Run a pass to pack branches into VLIW instructions (default).
8998 This switch is mainly for debugging the compiler and will likely be removed
8999 in a future version.
9001 @item -mno-vliw-branch
9002 @opindex mno-vliw-branch
9004 Do not run a pass to pack branches into VLIW instructions.
9006 This switch is mainly for debugging the compiler and will likely be removed
9007 in a future version.
9009 @item -mmulti-cond-exec
9010 @opindex mmulti-cond-exec
9012 Enable optimization of @code{&&} and @code{||} in conditional execution
9015 This switch is mainly for debugging the compiler and will likely be removed
9016 in a future version.
9018 @item -mno-multi-cond-exec
9019 @opindex mno-multi-cond-exec
9021 Disable optimization of @code{&&} and @code{||} in conditional execution.
9023 This switch is mainly for debugging the compiler and will likely be removed
9024 in a future version.
9026 @item -mnested-cond-exec
9027 @opindex mnested-cond-exec
9029 Enable nested conditional execution optimizations (default).
9031 This switch is mainly for debugging the compiler and will likely be removed
9032 in a future version.
9034 @item -mno-nested-cond-exec
9035 @opindex mno-nested-cond-exec
9037 Disable nested conditional execution optimizations.
9039 This switch is mainly for debugging the compiler and will likely be removed
9040 in a future version.
9042 @item -moptimize-membar
9043 @opindex moptimize-membar
9045 This switch removes redundant @code{membar} instructions from the
9046 compiler generated code. It is enabled by default.
9048 @item -mno-optimize-membar
9049 @opindex mno-optimize-membar
9051 This switch disables the automatic removal of redundant @code{membar}
9052 instructions from the generated code.
9054 @item -mtomcat-stats
9055 @opindex mtomcat-stats
9057 Cause gas to print out tomcat statistics.
9059 @item -mcpu=@var{cpu}
9062 Select the processor type for which to generate code. Possible values are
9063 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
9064 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
9068 @node GNU/Linux Options
9069 @subsection GNU/Linux Options
9071 These @samp{-m} options are defined for GNU/Linux targets:
9076 Use the GNU C library instead of uClibc. This is the default except
9077 on @samp{*-*-linux-*uclibc*} targets.
9081 Use uClibc instead of the GNU C library. This is the default on
9082 @samp{*-*-linux-*uclibc*} targets.
9085 @node H8/300 Options
9086 @subsection H8/300 Options
9088 These @samp{-m} options are defined for the H8/300 implementations:
9093 Shorten some address references at link time, when possible; uses the
9094 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9095 ld, Using ld}, for a fuller description.
9099 Generate code for the H8/300H@.
9103 Generate code for the H8S@.
9107 Generate code for the H8S and H8/300H in the normal mode. This switch
9108 must be used either with @option{-mh} or @option{-ms}.
9112 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9116 Make @code{int} data 32 bits by default.
9120 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9121 The default for the H8/300H and H8S is to align longs and floats on 4
9123 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9124 This option has no effect on the H8/300.
9128 @subsection HPPA Options
9129 @cindex HPPA Options
9131 These @samp{-m} options are defined for the HPPA family of computers:
9134 @item -march=@var{architecture-type}
9136 Generate code for the specified architecture. The choices for
9137 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
9138 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
9139 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
9140 architecture option for your machine. Code compiled for lower numbered
9141 architectures will run on higher numbered architectures, but not the
9145 @itemx -mpa-risc-1-1
9146 @itemx -mpa-risc-2-0
9147 @opindex mpa-risc-1-0
9148 @opindex mpa-risc-1-1
9149 @opindex mpa-risc-2-0
9150 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
9153 @opindex mbig-switch
9154 Generate code suitable for big switch tables. Use this option only if
9155 the assembler/linker complain about out of range branches within a switch
9158 @item -mjump-in-delay
9159 @opindex mjump-in-delay
9160 Fill delay slots of function calls with unconditional jump instructions
9161 by modifying the return pointer for the function call to be the target
9162 of the conditional jump.
9164 @item -mdisable-fpregs
9165 @opindex mdisable-fpregs
9166 Prevent floating point registers from being used in any manner. This is
9167 necessary for compiling kernels which perform lazy context switching of
9168 floating point registers. If you use this option and attempt to perform
9169 floating point operations, the compiler will abort.
9171 @item -mdisable-indexing
9172 @opindex mdisable-indexing
9173 Prevent the compiler from using indexing address modes. This avoids some
9174 rather obscure problems when compiling MIG generated code under MACH@.
9176 @item -mno-space-regs
9177 @opindex mno-space-regs
9178 Generate code that assumes the target has no space registers. This allows
9179 GCC to generate faster indirect calls and use unscaled index address modes.
9181 Such code is suitable for level 0 PA systems and kernels.
9183 @item -mfast-indirect-calls
9184 @opindex mfast-indirect-calls
9185 Generate code that assumes calls never cross space boundaries. This
9186 allows GCC to emit code which performs faster indirect calls.
9188 This option will not work in the presence of shared libraries or nested
9191 @item -mfixed-range=@var{register-range}
9192 @opindex mfixed-range
9193 Generate code treating the given register range as fixed registers.
9194 A fixed register is one that the register allocator can not use. This is
9195 useful when compiling kernel code. A register range is specified as
9196 two registers separated by a dash. Multiple register ranges can be
9197 specified separated by a comma.
9199 @item -mlong-load-store
9200 @opindex mlong-load-store
9201 Generate 3-instruction load and store sequences as sometimes required by
9202 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
9205 @item -mportable-runtime
9206 @opindex mportable-runtime
9207 Use the portable calling conventions proposed by HP for ELF systems.
9211 Enable the use of assembler directives only GAS understands.
9213 @item -mschedule=@var{cpu-type}
9215 Schedule code according to the constraints for the machine type
9216 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
9217 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
9218 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
9219 proper scheduling option for your machine. The default scheduling is
9223 @opindex mlinker-opt
9224 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
9225 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
9226 linkers in which they give bogus error messages when linking some programs.
9229 @opindex msoft-float
9230 Generate output containing library calls for floating point.
9231 @strong{Warning:} the requisite libraries are not available for all HPPA
9232 targets. Normally the facilities of the machine's usual C compiler are
9233 used, but this cannot be done directly in cross-compilation. You must make
9234 your own arrangements to provide suitable library functions for
9235 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
9236 does provide software floating point support.
9238 @option{-msoft-float} changes the calling convention in the output file;
9239 therefore, it is only useful if you compile @emph{all} of a program with
9240 this option. In particular, you need to compile @file{libgcc.a}, the
9241 library that comes with GCC, with @option{-msoft-float} in order for
9246 Generate the predefine, @code{_SIO}, for server IO@. The default is
9247 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
9248 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
9249 options are available under HP-UX and HI-UX@.
9253 Use GNU ld specific options. This passes @option{-shared} to ld when
9254 building a shared library. It is the default when GCC is configured,
9255 explicitly or implicitly, with the GNU linker. This option does not
9256 have any affect on which ld is called, it only changes what parameters
9257 are passed to that ld. The ld that is called is determined by the
9258 @option{--with-ld} configure option, GCC's program search path, and
9259 finally by the user's @env{PATH}. The linker used by GCC can be printed
9260 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
9261 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9265 Use HP ld specific options. This passes @option{-b} to ld when building
9266 a shared library and passes @option{+Accept TypeMismatch} to ld on all
9267 links. It is the default when GCC is configured, explicitly or
9268 implicitly, with the HP linker. This option does not have any affect on
9269 which ld is called, it only changes what parameters are passed to that
9270 ld. The ld that is called is determined by the @option{--with-ld}
9271 configure option, GCC's program search path, and finally by the user's
9272 @env{PATH}. The linker used by GCC can be printed using @samp{which
9273 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
9274 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9277 @opindex mno-long-calls
9278 Generate code that uses long call sequences. This ensures that a call
9279 is always able to reach linker generated stubs. The default is to generate
9280 long calls only when the distance from the call site to the beginning
9281 of the function or translation unit, as the case may be, exceeds a
9282 predefined limit set by the branch type being used. The limits for
9283 normal calls are 7,600,000 and 240,000 bytes, respectively for the
9284 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
9287 Distances are measured from the beginning of functions when using the
9288 @option{-ffunction-sections} option, or when using the @option{-mgas}
9289 and @option{-mno-portable-runtime} options together under HP-UX with
9292 It is normally not desirable to use this option as it will degrade
9293 performance. However, it may be useful in large applications,
9294 particularly when partial linking is used to build the application.
9296 The types of long calls used depends on the capabilities of the
9297 assembler and linker, and the type of code being generated. The
9298 impact on systems that support long absolute calls, and long pic
9299 symbol-difference or pc-relative calls should be relatively small.
9300 However, an indirect call is used on 32-bit ELF systems in pic code
9301 and it is quite long.
9303 @item -munix=@var{unix-std}
9305 Generate compiler predefines and select a startfile for the specified
9306 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
9307 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
9308 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
9309 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
9310 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
9313 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
9314 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
9315 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
9316 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
9317 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
9318 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
9320 It is @emph{important} to note that this option changes the interfaces
9321 for various library routines. It also affects the operational behavior
9322 of the C library. Thus, @emph{extreme} care is needed in using this
9325 Library code that is intended to operate with more than one UNIX
9326 standard must test, set and restore the variable @var{__xpg4_extended_mask}
9327 as appropriate. Most GNU software doesn't provide this capability.
9331 Suppress the generation of link options to search libdld.sl when the
9332 @option{-static} option is specified on HP-UX 10 and later.
9336 The HP-UX implementation of setlocale in libc has a dependency on
9337 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9338 when the @option{-static} option is specified, special link options
9339 are needed to resolve this dependency.
9341 On HP-UX 10 and later, the GCC driver adds the necessary options to
9342 link with libdld.sl when the @option{-static} option is specified.
9343 This causes the resulting binary to be dynamic. On the 64-bit port,
9344 the linkers generate dynamic binaries by default in any case. The
9345 @option{-nolibdld} option can be used to prevent the GCC driver from
9346 adding these link options.
9350 Add support for multithreading with the @dfn{dce thread} library
9351 under HP-UX@. This option sets flags for both the preprocessor and
9355 @node i386 and x86-64 Options
9356 @subsection Intel 386 and AMD x86-64 Options
9357 @cindex i386 Options
9358 @cindex x86-64 Options
9359 @cindex Intel 386 Options
9360 @cindex AMD x86-64 Options
9362 These @samp{-m} options are defined for the i386 and x86-64 family of
9366 @item -mtune=@var{cpu-type}
9368 Tune to @var{cpu-type} everything applicable about the generated code, except
9369 for the ABI and the set of available instructions. The choices for
9373 Produce code optimized for the most common IA32/AMD64/EM64T processors.
9374 If you know the CPU on which your code will run, then you should use
9375 the corresponding @option{-mtune} option instead of
9376 @option{-mtune=generic}. But, if you do not know exactly what CPU users
9377 of your application will have, then you should use this option.
9379 As new processors are deployed in the marketplace, the behavior of this
9380 option will change. Therefore, if you upgrade to a newer version of
9381 GCC, the code generated option will change to reflect the processors
9382 that were most common when that version of GCC was released.
9384 There is no @option{-march=generic} option because @option{-march}
9385 indicates the instruction set the compiler can use, and there is no
9386 generic instruction set applicable to all processors. In contrast,
9387 @option{-mtune} indicates the processor (or, in this case, collection of
9388 processors) for which the code is optimized.
9390 This selects the CPU to tune for at compilation time by determining
9391 the processor type of the compiling machine. Using @option{-mtune=native}
9392 will produce code optimized for the local machine under the constraints
9393 of the selected instruction set. Using @option{-march=native} will
9394 enable all instruction subsets supported by the local machine (hence
9395 the result might not run on different machines).
9397 Original Intel's i386 CPU@.
9399 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
9401 Intel Pentium CPU with no MMX support.
9403 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
9405 Intel PentiumPro CPU@.
9407 Same as @code{generic}, but when used as @code{march} option, PentiumPro
9408 instruction set will be used, so the code will run on all i686 family chips.
9410 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
9411 @item pentium3, pentium3m
9412 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
9415 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
9416 support. Used by Centrino notebooks.
9417 @item pentium4, pentium4m
9418 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
9420 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
9423 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
9424 SSE2 and SSE3 instruction set support.
9426 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
9427 instruction set support.
9429 AMD K6 CPU with MMX instruction set support.
9431 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
9432 @item athlon, athlon-tbird
9433 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
9435 @item athlon-4, athlon-xp, athlon-mp
9436 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
9437 instruction set support.
9438 @item k8, opteron, athlon64, athlon-fx
9439 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
9440 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
9442 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
9445 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
9446 instruction set support.
9448 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
9449 implemented for this chip.)
9451 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
9452 implemented for this chip.)
9454 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
9457 While picking a specific @var{cpu-type} will schedule things appropriately
9458 for that particular chip, the compiler will not generate any code that
9459 does not run on the i386 without the @option{-march=@var{cpu-type}} option
9462 @item -march=@var{cpu-type}
9464 Generate instructions for the machine type @var{cpu-type}. The choices
9465 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
9466 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
9468 @item -mcpu=@var{cpu-type}
9470 A deprecated synonym for @option{-mtune}.
9479 @opindex mpentiumpro
9480 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
9481 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
9482 These synonyms are deprecated.
9484 @item -mfpmath=@var{unit}
9486 Generate floating point arithmetics for selected unit @var{unit}. The choices
9491 Use the standard 387 floating point coprocessor present majority of chips and
9492 emulated otherwise. Code compiled with this option will run almost everywhere.
9493 The temporary results are computed in 80bit precision instead of precision
9494 specified by the type resulting in slightly different results compared to most
9495 of other chips. See @option{-ffloat-store} for more detailed description.
9497 This is the default choice for i386 compiler.
9500 Use scalar floating point instructions present in the SSE instruction set.
9501 This instruction set is supported by Pentium3 and newer chips, in the AMD line
9502 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
9503 instruction set supports only single precision arithmetics, thus the double and
9504 extended precision arithmetics is still done using 387. Later version, present
9505 only in Pentium4 and the future AMD x86-64 chips supports double precision
9508 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
9509 or @option{-msse2} switches to enable SSE extensions and make this option
9510 effective. For the x86-64 compiler, these extensions are enabled by default.
9512 The resulting code should be considerably faster in the majority of cases and avoid
9513 the numerical instability problems of 387 code, but may break some existing
9514 code that expects temporaries to be 80bit.
9516 This is the default choice for the x86-64 compiler.
9519 Attempt to utilize both instruction sets at once. This effectively double the
9520 amount of available registers and on chips with separate execution units for
9521 387 and SSE the execution resources too. Use this option with care, as it is
9522 still experimental, because the GCC register allocator does not model separate
9523 functional units well resulting in instable performance.
9526 @item -masm=@var{dialect}
9527 @opindex masm=@var{dialect}
9528 Output asm instructions using selected @var{dialect}. Supported
9529 choices are @samp{intel} or @samp{att} (the default one). Darwin does
9530 not support @samp{intel}.
9535 @opindex mno-ieee-fp
9536 Control whether or not the compiler uses IEEE floating point
9537 comparisons. These handle correctly the case where the result of a
9538 comparison is unordered.
9541 @opindex msoft-float
9542 Generate output containing library calls for floating point.
9543 @strong{Warning:} the requisite libraries are not part of GCC@.
9544 Normally the facilities of the machine's usual C compiler are used, but
9545 this can't be done directly in cross-compilation. You must make your
9546 own arrangements to provide suitable library functions for
9549 On machines where a function returns floating point results in the 80387
9550 register stack, some floating point opcodes may be emitted even if
9551 @option{-msoft-float} is used.
9553 @item -mno-fp-ret-in-387
9554 @opindex mno-fp-ret-in-387
9555 Do not use the FPU registers for return values of functions.
9557 The usual calling convention has functions return values of types
9558 @code{float} and @code{double} in an FPU register, even if there
9559 is no FPU@. The idea is that the operating system should emulate
9562 The option @option{-mno-fp-ret-in-387} causes such values to be returned
9563 in ordinary CPU registers instead.
9565 @item -mno-fancy-math-387
9566 @opindex mno-fancy-math-387
9567 Some 387 emulators do not support the @code{sin}, @code{cos} and
9568 @code{sqrt} instructions for the 387. Specify this option to avoid
9569 generating those instructions. This option is the default on FreeBSD,
9570 OpenBSD and NetBSD@. This option is overridden when @option{-march}
9571 indicates that the target cpu will always have an FPU and so the
9572 instruction will not need emulation. As of revision 2.6.1, these
9573 instructions are not generated unless you also use the
9574 @option{-funsafe-math-optimizations} switch.
9576 @item -malign-double
9577 @itemx -mno-align-double
9578 @opindex malign-double
9579 @opindex mno-align-double
9580 Control whether GCC aligns @code{double}, @code{long double}, and
9581 @code{long long} variables on a two word boundary or a one word
9582 boundary. Aligning @code{double} variables on a two word boundary will
9583 produce code that runs somewhat faster on a @samp{Pentium} at the
9584 expense of more memory.
9586 On x86-64, @option{-malign-double} is enabled by default.
9588 @strong{Warning:} if you use the @option{-malign-double} switch,
9589 structures containing the above types will be aligned differently than
9590 the published application binary interface specifications for the 386
9591 and will not be binary compatible with structures in code compiled
9592 without that switch.
9594 @item -m96bit-long-double
9595 @itemx -m128bit-long-double
9596 @opindex m96bit-long-double
9597 @opindex m128bit-long-double
9598 These switches control the size of @code{long double} type. The i386
9599 application binary interface specifies the size to be 96 bits,
9600 so @option{-m96bit-long-double} is the default in 32 bit mode.
9602 Modern architectures (Pentium and newer) would prefer @code{long double}
9603 to be aligned to an 8 or 16 byte boundary. In arrays or structures
9604 conforming to the ABI, this would not be possible. So specifying a
9605 @option{-m128bit-long-double} will align @code{long double}
9606 to a 16 byte boundary by padding the @code{long double} with an additional
9609 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
9610 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
9612 Notice that neither of these options enable any extra precision over the x87
9613 standard of 80 bits for a @code{long double}.
9615 @strong{Warning:} if you override the default value for your target ABI, the
9616 structures and arrays containing @code{long double} variables will change
9617 their size as well as function calling convention for function taking
9618 @code{long double} will be modified. Hence they will not be binary
9619 compatible with arrays or structures in code compiled without that switch.
9621 @item -mmlarge-data-threshold=@var{number}
9622 @opindex mlarge-data-threshold=@var{number}
9623 When @option{-mcmodel=medium} is specified, the data greater than
9624 @var{threshold} are placed in large data section. This value must be the
9625 same across all object linked into the binary and defaults to 65535.
9628 @itemx -mno-svr3-shlib
9629 @opindex msvr3-shlib
9630 @opindex mno-svr3-shlib
9631 Control whether GCC places uninitialized local variables into the
9632 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
9633 into @code{bss}. These options are meaningful only on System V Release 3.
9637 Use a different function-calling convention, in which functions that
9638 take a fixed number of arguments return with the @code{ret} @var{num}
9639 instruction, which pops their arguments while returning. This saves one
9640 instruction in the caller since there is no need to pop the arguments
9643 You can specify that an individual function is called with this calling
9644 sequence with the function attribute @samp{stdcall}. You can also
9645 override the @option{-mrtd} option by using the function attribute
9646 @samp{cdecl}. @xref{Function Attributes}.
9648 @strong{Warning:} this calling convention is incompatible with the one
9649 normally used on Unix, so you cannot use it if you need to call
9650 libraries compiled with the Unix compiler.
9652 Also, you must provide function prototypes for all functions that
9653 take variable numbers of arguments (including @code{printf});
9654 otherwise incorrect code will be generated for calls to those
9657 In addition, seriously incorrect code will result if you call a
9658 function with too many arguments. (Normally, extra arguments are
9659 harmlessly ignored.)
9661 @item -mregparm=@var{num}
9663 Control how many registers are used to pass integer arguments. By
9664 default, no registers are used to pass arguments, and at most 3
9665 registers can be used. You can control this behavior for a specific
9666 function by using the function attribute @samp{regparm}.
9667 @xref{Function Attributes}.
9669 @strong{Warning:} if you use this switch, and
9670 @var{num} is nonzero, then you must build all modules with the same
9671 value, including any libraries. This includes the system libraries and
9675 @opindex msseregparm
9676 Use SSE register passing conventions for float and double arguments
9677 and return values. You can control this behavior for a specific
9678 function by using the function attribute @samp{sseregparm}.
9679 @xref{Function Attributes}.
9681 @strong{Warning:} if you use this switch then you must build all
9682 modules with the same value, including any libraries. This includes
9683 the system libraries and startup modules.
9685 @item -mstackrealign
9686 @opindex mstackrealign
9687 Realign the stack at entry. On the Intel x86, the
9688 @option{-mstackrealign} option will generate an alternate prologue and
9689 epilogue that realigns the runtime stack. This supports mixing legacy
9690 codes that keep a 4-byte aligned stack with modern codes that keep a
9691 16-byte stack for SSE compatibility. The alternate prologue and
9692 epilogue are slower and bigger than the regular ones, and the
9693 alternate prologue requires an extra scratch register; this lowers the
9694 number of registers available if used in conjunction with the
9695 @code{regparm} attribute. The @option{-mstackrealign} option is
9696 incompatible with the nested function prologue; this is considered a
9697 hard error. See also the attribute @code{force_align_arg_pointer},
9698 applicable to individual functions.
9700 @item -mpreferred-stack-boundary=@var{num}
9701 @opindex mpreferred-stack-boundary
9702 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
9703 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
9704 the default is 4 (16 bytes or 128 bits).
9706 On Pentium and PentiumPro, @code{double} and @code{long double} values
9707 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
9708 suffer significant run time performance penalties. On Pentium III, the
9709 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
9710 properly if it is not 16 byte aligned.
9712 To ensure proper alignment of this values on the stack, the stack boundary
9713 must be as aligned as that required by any value stored on the stack.
9714 Further, every function must be generated such that it keeps the stack
9715 aligned. Thus calling a function compiled with a higher preferred
9716 stack boundary from a function compiled with a lower preferred stack
9717 boundary will most likely misalign the stack. It is recommended that
9718 libraries that use callbacks always use the default setting.
9720 This extra alignment does consume extra stack space, and generally
9721 increases code size. Code that is sensitive to stack space usage, such
9722 as embedded systems and operating system kernels, may want to reduce the
9723 preferred alignment to @option{-mpreferred-stack-boundary=2}.
9743 These switches enable or disable the use of instructions in the MMX,
9744 SSE, SSE2, SSE3, SSSE3 or 3DNow! extended instruction sets.
9745 These extensions are also available as built-in functions: see
9746 @ref{X86 Built-in Functions}, for details of the functions enabled and
9747 disabled by these switches.
9749 To have SSE/SSE2 instructions generated automatically from floating-point
9750 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
9752 These options will enable GCC to use these extended instructions in
9753 generated code, even without @option{-mfpmath=sse}. Applications which
9754 perform runtime CPU detection must compile separate files for each
9755 supported architecture, using the appropriate flags. In particular,
9756 the file containing the CPU detection code should be compiled without
9760 @itemx -mno-push-args
9762 @opindex mno-push-args
9763 Use PUSH operations to store outgoing parameters. This method is shorter
9764 and usually equally fast as method using SUB/MOV operations and is enabled
9765 by default. In some cases disabling it may improve performance because of
9766 improved scheduling and reduced dependencies.
9768 @item -maccumulate-outgoing-args
9769 @opindex maccumulate-outgoing-args
9770 If enabled, the maximum amount of space required for outgoing arguments will be
9771 computed in the function prologue. This is faster on most modern CPUs
9772 because of reduced dependencies, improved scheduling and reduced stack usage
9773 when preferred stack boundary is not equal to 2. The drawback is a notable
9774 increase in code size. This switch implies @option{-mno-push-args}.
9778 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
9779 on thread-safe exception handling must compile and link all code with the
9780 @option{-mthreads} option. When compiling, @option{-mthreads} defines
9781 @option{-D_MT}; when linking, it links in a special thread helper library
9782 @option{-lmingwthrd} which cleans up per thread exception handling data.
9784 @item -mno-align-stringops
9785 @opindex mno-align-stringops
9786 Do not align destination of inlined string operations. This switch reduces
9787 code size and improves performance in case the destination is already aligned,
9788 but GCC doesn't know about it.
9790 @item -minline-all-stringops
9791 @opindex minline-all-stringops
9792 By default GCC inlines string operations only when destination is known to be
9793 aligned at least to 4 byte boundary. This enables more inlining, increase code
9794 size, but may improve performance of code that depends on fast memcpy, strlen
9795 and memset for short lengths.
9797 @item -minline-stringops-dynamically
9798 @opindex minline-stringops-dynamically
9799 For string operation of unknown size, inline runtime checks so for small
9800 blocks inline code is used, while for large blocks library call is used.
9802 @item -mstringop-strategy=@var{alg}
9803 @opindex mstringop-strategy=@var{alg}
9804 Overwrite internal decision heuristic about particular algorithm to inline
9805 string operation with. The allowed values are @code{rep_byte},
9806 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
9807 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
9808 expanding inline loop, @code{libcall} for always expanding library call.
9810 @item -momit-leaf-frame-pointer
9811 @opindex momit-leaf-frame-pointer
9812 Don't keep the frame pointer in a register for leaf functions. This
9813 avoids the instructions to save, set up and restore frame pointers and
9814 makes an extra register available in leaf functions. The option
9815 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9816 which might make debugging harder.
9818 @item -mtls-direct-seg-refs
9819 @itemx -mno-tls-direct-seg-refs
9820 @opindex mtls-direct-seg-refs
9821 Controls whether TLS variables may be accessed with offsets from the
9822 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
9823 or whether the thread base pointer must be added. Whether or not this
9824 is legal depends on the operating system, and whether it maps the
9825 segment to cover the entire TLS area.
9827 For systems that use GNU libc, the default is on.
9830 These @samp{-m} switches are supported in addition to the above
9831 on AMD x86-64 processors in 64-bit environments.
9838 Generate code for a 32-bit or 64-bit environment.
9839 The 32-bit environment sets int, long and pointer to 32 bits and
9840 generates code that runs on any i386 system.
9841 The 64-bit environment sets int to 32 bits and long and pointer
9842 to 64 bits and generates code for AMD's x86-64 architecture.
9845 @opindex no-red-zone
9846 Do not use a so called red zone for x86-64 code. The red zone is mandated
9847 by the x86-64 ABI, it is a 128-byte area beyond the location of the
9848 stack pointer that will not be modified by signal or interrupt handlers
9849 and therefore can be used for temporary data without adjusting the stack
9850 pointer. The flag @option{-mno-red-zone} disables this red zone.
9852 @item -mcmodel=small
9853 @opindex mcmodel=small
9854 Generate code for the small code model: the program and its symbols must
9855 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
9856 Programs can be statically or dynamically linked. This is the default
9859 @item -mcmodel=kernel
9860 @opindex mcmodel=kernel
9861 Generate code for the kernel code model. The kernel runs in the
9862 negative 2 GB of the address space.
9863 This model has to be used for Linux kernel code.
9865 @item -mcmodel=medium
9866 @opindex mcmodel=medium
9867 Generate code for the medium model: The program is linked in the lower 2
9868 GB of the address space but symbols can be located anywhere in the
9869 address space. Programs can be statically or dynamically linked, but
9870 building of shared libraries are not supported with the medium model.
9872 @item -mcmodel=large
9873 @opindex mcmodel=large
9874 Generate code for the large model: This model makes no assumptions
9875 about addresses and sizes of sections. Currently GCC does not implement
9880 @subsection IA-64 Options
9881 @cindex IA-64 Options
9883 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9887 @opindex mbig-endian
9888 Generate code for a big endian target. This is the default for HP-UX@.
9890 @item -mlittle-endian
9891 @opindex mlittle-endian
9892 Generate code for a little endian target. This is the default for AIX5
9899 Generate (or don't) code for the GNU assembler. This is the default.
9900 @c Also, this is the default if the configure option @option{--with-gnu-as}
9907 Generate (or don't) code for the GNU linker. This is the default.
9908 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9913 Generate code that does not use a global pointer register. The result
9914 is not position independent code, and violates the IA-64 ABI@.
9916 @item -mvolatile-asm-stop
9917 @itemx -mno-volatile-asm-stop
9918 @opindex mvolatile-asm-stop
9919 @opindex mno-volatile-asm-stop
9920 Generate (or don't) a stop bit immediately before and after volatile asm
9923 @item -mregister-names
9924 @itemx -mno-register-names
9925 @opindex mregister-names
9926 @opindex mno-register-names
9927 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9928 the stacked registers. This may make assembler output more readable.
9934 Disable (or enable) optimizations that use the small data section. This may
9935 be useful for working around optimizer bugs.
9938 @opindex mconstant-gp
9939 Generate code that uses a single constant global pointer value. This is
9940 useful when compiling kernel code.
9944 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9945 This is useful when compiling firmware code.
9947 @item -minline-float-divide-min-latency
9948 @opindex minline-float-divide-min-latency
9949 Generate code for inline divides of floating point values
9950 using the minimum latency algorithm.
9952 @item -minline-float-divide-max-throughput
9953 @opindex minline-float-divide-max-throughput
9954 Generate code for inline divides of floating point values
9955 using the maximum throughput algorithm.
9957 @item -minline-int-divide-min-latency
9958 @opindex minline-int-divide-min-latency
9959 Generate code for inline divides of integer values
9960 using the minimum latency algorithm.
9962 @item -minline-int-divide-max-throughput
9963 @opindex minline-int-divide-max-throughput
9964 Generate code for inline divides of integer values
9965 using the maximum throughput algorithm.
9967 @item -minline-sqrt-min-latency
9968 @opindex minline-sqrt-min-latency
9969 Generate code for inline square roots
9970 using the minimum latency algorithm.
9972 @item -minline-sqrt-max-throughput
9973 @opindex minline-sqrt-max-throughput
9974 Generate code for inline square roots
9975 using the maximum throughput algorithm.
9977 @item -mno-dwarf2-asm
9979 @opindex mno-dwarf2-asm
9980 @opindex mdwarf2-asm
9981 Don't (or do) generate assembler code for the DWARF2 line number debugging
9982 info. This may be useful when not using the GNU assembler.
9984 @item -mearly-stop-bits
9985 @itemx -mno-early-stop-bits
9986 @opindex mearly-stop-bits
9987 @opindex mno-early-stop-bits
9988 Allow stop bits to be placed earlier than immediately preceding the
9989 instruction that triggered the stop bit. This can improve instruction
9990 scheduling, but does not always do so.
9992 @item -mfixed-range=@var{register-range}
9993 @opindex mfixed-range
9994 Generate code treating the given register range as fixed registers.
9995 A fixed register is one that the register allocator can not use. This is
9996 useful when compiling kernel code. A register range is specified as
9997 two registers separated by a dash. Multiple register ranges can be
9998 specified separated by a comma.
10000 @item -mtls-size=@var{tls-size}
10002 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
10005 @item -mtune=@var{cpu-type}
10007 Tune the instruction scheduling for a particular CPU, Valid values are
10008 itanium, itanium1, merced, itanium2, and mckinley.
10014 Add support for multithreading using the POSIX threads library. This
10015 option sets flags for both the preprocessor and linker. It does
10016 not affect the thread safety of object code produced by the compiler or
10017 that of libraries supplied with it. These are HP-UX specific flags.
10023 Generate code for a 32-bit or 64-bit environment.
10024 The 32-bit environment sets int, long and pointer to 32 bits.
10025 The 64-bit environment sets int to 32 bits and long and pointer
10026 to 64 bits. These are HP-UX specific flags.
10028 @item -mno-sched-br-data-spec
10029 @itemx -msched-br-data-spec
10030 @opindex -mno-sched-br-data-spec
10031 @opindex -msched-br-data-spec
10032 (Dis/En)able data speculative scheduling before reload.
10033 This will result in generation of the ld.a instructions and
10034 the corresponding check instructions (ld.c / chk.a).
10035 The default is 'disable'.
10037 @item -msched-ar-data-spec
10038 @itemx -mno-sched-ar-data-spec
10039 @opindex -msched-ar-data-spec
10040 @opindex -mno-sched-ar-data-spec
10041 (En/Dis)able data speculative scheduling after reload.
10042 This will result in generation of the ld.a instructions and
10043 the corresponding check instructions (ld.c / chk.a).
10044 The default is 'enable'.
10046 @item -mno-sched-control-spec
10047 @itemx -msched-control-spec
10048 @opindex -mno-sched-control-spec
10049 @opindex -msched-control-spec
10050 (Dis/En)able control speculative scheduling. This feature is
10051 available only during region scheduling (i.e. before reload).
10052 This will result in generation of the ld.s instructions and
10053 the corresponding check instructions chk.s .
10054 The default is 'disable'.
10056 @item -msched-br-in-data-spec
10057 @itemx -mno-sched-br-in-data-spec
10058 @opindex -msched-br-in-data-spec
10059 @opindex -mno-sched-br-in-data-spec
10060 (En/Dis)able speculative scheduling of the instructions that
10061 are dependent on the data speculative loads before reload.
10062 This is effective only with @option{-msched-br-data-spec} enabled.
10063 The default is 'enable'.
10065 @item -msched-ar-in-data-spec
10066 @itemx -mno-sched-ar-in-data-spec
10067 @opindex -msched-ar-in-data-spec
10068 @opindex -mno-sched-ar-in-data-spec
10069 (En/Dis)able speculative scheduling of the instructions that
10070 are dependent on the data speculative loads after reload.
10071 This is effective only with @option{-msched-ar-data-spec} enabled.
10072 The default is 'enable'.
10074 @item -msched-in-control-spec
10075 @itemx -mno-sched-in-control-spec
10076 @opindex -msched-in-control-spec
10077 @opindex -mno-sched-in-control-spec
10078 (En/Dis)able speculative scheduling of the instructions that
10079 are dependent on the control speculative loads.
10080 This is effective only with @option{-msched-control-spec} enabled.
10081 The default is 'enable'.
10084 @itemx -mno-sched-ldc
10085 @opindex -msched-ldc
10086 @opindex -mno-sched-ldc
10087 (En/Dis)able use of simple data speculation checks ld.c .
10088 If disabled, only chk.a instructions will be emitted to check
10089 data speculative loads.
10090 The default is 'enable'.
10092 @item -mno-sched-control-ldc
10093 @itemx -msched-control-ldc
10094 @opindex -mno-sched-control-ldc
10095 @opindex -msched-control-ldc
10096 (Dis/En)able use of ld.c instructions to check control speculative loads.
10097 If enabled, in case of control speculative load with no speculatively
10098 scheduled dependent instructions this load will be emitted as ld.sa and
10099 ld.c will be used to check it.
10100 The default is 'disable'.
10102 @item -mno-sched-spec-verbose
10103 @itemx -msched-spec-verbose
10104 @opindex -mno-sched-spec-verbose
10105 @opindex -msched-spec-verbose
10106 (Dis/En)able printing of the information about speculative motions.
10108 @item -mno-sched-prefer-non-data-spec-insns
10109 @itemx -msched-prefer-non-data-spec-insns
10110 @opindex -mno-sched-prefer-non-data-spec-insns
10111 @opindex -msched-prefer-non-data-spec-insns
10112 If enabled, data speculative instructions will be chosen for schedule
10113 only if there are no other choices at the moment. This will make
10114 the use of the data speculation much more conservative.
10115 The default is 'disable'.
10117 @item -mno-sched-prefer-non-control-spec-insns
10118 @itemx -msched-prefer-non-control-spec-insns
10119 @opindex -mno-sched-prefer-non-control-spec-insns
10120 @opindex -msched-prefer-non-control-spec-insns
10121 If enabled, control speculative instructions will be chosen for schedule
10122 only if there are no other choices at the moment. This will make
10123 the use of the control speculation much more conservative.
10124 The default is 'disable'.
10126 @item -mno-sched-count-spec-in-critical-path
10127 @itemx -msched-count-spec-in-critical-path
10128 @opindex -mno-sched-count-spec-in-critical-path
10129 @opindex -msched-count-spec-in-critical-path
10130 If enabled, speculative dependencies will be considered during
10131 computation of the instructions priorities. This will make the use of the
10132 speculation a bit more conservative.
10133 The default is 'disable'.
10138 @subsection M32C Options
10139 @cindex M32C options
10142 @item -mcpu=@var{name}
10144 Select the CPU for which code is generated. @var{name} may be one of
10145 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
10146 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
10147 the M32C/80 series.
10151 Specifies that the program will be run on the simulator. This causes
10152 an alternate runtime library to be linked in which supports, for
10153 example, file I/O. You must not use this option when generating
10154 programs that will run on real hardware; you must provide your own
10155 runtime library for whatever I/O functions are needed.
10157 @item -memregs=@var{number}
10159 Specifies the number of memory-based pseudo-registers GCC will use
10160 during code generation. These pseudo-registers will be used like real
10161 registers, so there is a tradeoff between GCC's ability to fit the
10162 code into available registers, and the performance penalty of using
10163 memory instead of registers. Note that all modules in a program must
10164 be compiled with the same value for this option. Because of that, you
10165 must not use this option with the default runtime libraries gcc
10170 @node M32R/D Options
10171 @subsection M32R/D Options
10172 @cindex M32R/D options
10174 These @option{-m} options are defined for Renesas M32R/D architectures:
10179 Generate code for the M32R/2@.
10183 Generate code for the M32R/X@.
10187 Generate code for the M32R@. This is the default.
10189 @item -mmodel=small
10190 @opindex mmodel=small
10191 Assume all objects live in the lower 16MB of memory (so that their addresses
10192 can be loaded with the @code{ld24} instruction), and assume all subroutines
10193 are reachable with the @code{bl} instruction.
10194 This is the default.
10196 The addressability of a particular object can be set with the
10197 @code{model} attribute.
10199 @item -mmodel=medium
10200 @opindex mmodel=medium
10201 Assume objects may be anywhere in the 32-bit address space (the compiler
10202 will generate @code{seth/add3} instructions to load their addresses), and
10203 assume all subroutines are reachable with the @code{bl} instruction.
10205 @item -mmodel=large
10206 @opindex mmodel=large
10207 Assume objects may be anywhere in the 32-bit address space (the compiler
10208 will generate @code{seth/add3} instructions to load their addresses), and
10209 assume subroutines may not be reachable with the @code{bl} instruction
10210 (the compiler will generate the much slower @code{seth/add3/jl}
10211 instruction sequence).
10214 @opindex msdata=none
10215 Disable use of the small data area. Variables will be put into
10216 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
10217 @code{section} attribute has been specified).
10218 This is the default.
10220 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
10221 Objects may be explicitly put in the small data area with the
10222 @code{section} attribute using one of these sections.
10224 @item -msdata=sdata
10225 @opindex msdata=sdata
10226 Put small global and static data in the small data area, but do not
10227 generate special code to reference them.
10230 @opindex msdata=use
10231 Put small global and static data in the small data area, and generate
10232 special instructions to reference them.
10236 @cindex smaller data references
10237 Put global and static objects less than or equal to @var{num} bytes
10238 into the small data or bss sections instead of the normal data or bss
10239 sections. The default value of @var{num} is 8.
10240 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
10241 for this option to have any effect.
10243 All modules should be compiled with the same @option{-G @var{num}} value.
10244 Compiling with different values of @var{num} may or may not work; if it
10245 doesn't the linker will give an error message---incorrect code will not be
10250 Makes the M32R specific code in the compiler display some statistics
10251 that might help in debugging programs.
10253 @item -malign-loops
10254 @opindex malign-loops
10255 Align all loops to a 32-byte boundary.
10257 @item -mno-align-loops
10258 @opindex mno-align-loops
10259 Do not enforce a 32-byte alignment for loops. This is the default.
10261 @item -missue-rate=@var{number}
10262 @opindex missue-rate=@var{number}
10263 Issue @var{number} instructions per cycle. @var{number} can only be 1
10266 @item -mbranch-cost=@var{number}
10267 @opindex mbranch-cost=@var{number}
10268 @var{number} can only be 1 or 2. If it is 1 then branches will be
10269 preferred over conditional code, if it is 2, then the opposite will
10272 @item -mflush-trap=@var{number}
10273 @opindex mflush-trap=@var{number}
10274 Specifies the trap number to use to flush the cache. The default is
10275 12. Valid numbers are between 0 and 15 inclusive.
10277 @item -mno-flush-trap
10278 @opindex mno-flush-trap
10279 Specifies that the cache cannot be flushed by using a trap.
10281 @item -mflush-func=@var{name}
10282 @opindex mflush-func=@var{name}
10283 Specifies the name of the operating system function to call to flush
10284 the cache. The default is @emph{_flush_cache}, but a function call
10285 will only be used if a trap is not available.
10287 @item -mno-flush-func
10288 @opindex mno-flush-func
10289 Indicates that there is no OS function for flushing the cache.
10293 @node M680x0 Options
10294 @subsection M680x0 Options
10295 @cindex M680x0 options
10297 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
10298 The default settings depend on which architecture was selected when
10299 the compiler was configured; the defaults for the most common choices
10303 @item -march=@var{arch}
10305 Generate code for a specific M680x0 or ColdFire instruction set
10306 architecture. Permissible values of @var{arch} for M680x0
10307 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
10308 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
10309 architectures are selected according to Freescale's ISA classification
10310 and the permissible values are: @samp{isaa}, @samp{isaaplus},
10311 @samp{isab} and @samp{isac}.
10313 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
10314 code for a ColdFire target. The @var{arch} in this macro is one of the
10315 @option{-march} arguments given above.
10317 When used together, @option{-march} and @option{-mtune} select code
10318 that runs on a family of similar processors but that is optimized
10319 for a particular microarchitecture.
10321 @item -mcpu=@var{cpu}
10323 Generate code for a specific M680x0 or ColdFire processor.
10324 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
10325 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
10326 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
10327 below, which also classifies the CPUs into families:
10329 @multitable @columnfractions 0.20 0.80
10330 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
10331 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
10332 @item @samp{5206e} @tab @samp{5206e}
10333 @item @samp{5208} @tab @samp{5207} @samp{5208}
10334 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
10335 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
10336 @item @samp{5216} @tab @samp{5214} @samp{5216}
10337 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
10338 @item @samp{5225} @tab @samp{5224} @samp{5225}
10339 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
10340 @item @samp{5249} @tab @samp{5249}
10341 @item @samp{5250} @tab @samp{5250}
10342 @item @samp{5271} @tab @samp{5270} @samp{5271}
10343 @item @samp{5272} @tab @samp{5272}
10344 @item @samp{5275} @tab @samp{5274} @samp{5275}
10345 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
10346 @item @samp{5307} @tab @samp{5307}
10347 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
10348 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
10349 @item @samp{5407} @tab @samp{5407}
10350 @item @samp{5475} @tab @samp{5470} @samp{5471} @samp{5472} @samp{5473} @samp{5474} @samp{5475} @samp{547x} @samp{5480} @samp{5481} @samp{5482} @samp{5483} @samp{5484} @samp{5485}
10353 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
10354 @var{arch} is compatible with @var{cpu}. Other combinations of
10355 @option{-mcpu} and @option{-march} are rejected.
10357 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
10358 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
10359 where the value of @var{family} is given by the table above.
10361 @item -mtune=@var{tune}
10363 Tune the code for a particular microarchitecture, within the
10364 constraints set by @option{-march} and @option{-mcpu}.
10365 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
10366 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
10367 and @samp{cpu32}. The ColdFire microarchitectures
10368 are: @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
10370 You can also use @option{-mtune=68020-40} for code that needs
10371 to run relatively well on 68020, 68030 and 68040 targets.
10372 @option{-mtune=68020-60} is similar but includes 68060 targets
10373 as well. These two options select the same tuning decisions as
10374 @option{-m68020-40} and @option{-m68020-60} respectively.
10376 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
10377 when tuning for 680x0 architecture @var{arch}. It also defines
10378 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
10379 option is used. If gcc is tuning for a range of architectures,
10380 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
10381 it defines the macros for every architecture in the range.
10383 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
10384 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
10385 of the arguments given above.
10391 Generate output for a 68000. This is the default
10392 when the compiler is configured for 68000-based systems.
10393 It is equivalent to @option{-march=68000}.
10395 Use this option for microcontrollers with a 68000 or EC000 core,
10396 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
10400 Generate output for a 68010. This is the default
10401 when the compiler is configured for 68010-based systems.
10402 It is equivalent to @option{-march=68010}.
10408 Generate output for a 68020. This is the default
10409 when the compiler is configured for 68020-based systems.
10410 It is equivalent to @option{-march=68020}.
10414 Generate output for a 68030. This is the default when the compiler is
10415 configured for 68030-based systems. It is equivalent to
10416 @option{-march=68030}.
10420 Generate output for a 68040. This is the default when the compiler is
10421 configured for 68040-based systems. It is equivalent to
10422 @option{-march=68040}.
10424 This option inhibits the use of 68881/68882 instructions that have to be
10425 emulated by software on the 68040. Use this option if your 68040 does not
10426 have code to emulate those instructions.
10430 Generate output for a 68060. This is the default when the compiler is
10431 configured for 68060-based systems. It is equivalent to
10432 @option{-march=68060}.
10434 This option inhibits the use of 68020 and 68881/68882 instructions that
10435 have to be emulated by software on the 68060. Use this option if your 68060
10436 does not have code to emulate those instructions.
10440 Generate output for a CPU32. This is the default
10441 when the compiler is configured for CPU32-based systems.
10442 It is equivalent to @option{-march=cpu32}.
10444 Use this option for microcontrollers with a
10445 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
10446 68336, 68340, 68341, 68349 and 68360.
10450 Generate output for a 520X ColdFire CPU. This is the default
10451 when the compiler is configured for 520X-based systems.
10452 It is equivalent to @option{-mcpu=5206}, and is now deprecated
10453 in favor of that option.
10455 Use this option for microcontroller with a 5200 core, including
10456 the MCF5202, MCF5203, MCF5204 and MCF5206.
10460 Generate output for a 5206e ColdFire CPU. The option is now
10461 deprecated in favor of the equivalent @option{-mcpu=5206e}.
10465 Generate output for a member of the ColdFire 528X family.
10466 The option is now deprecated in favor of the equivalent
10467 @option{-mcpu=528x}.
10471 Generate output for a ColdFire 5307 CPU. The option is now deprecated
10472 in favor of the equivalent @option{-mcpu=5307}.
10476 Generate output for a ColdFire 5407 CPU. The option is now deprecated
10477 in favor of the equivalent @option{-mcpu=5407}.
10481 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
10482 This includes use of hardware floating point instructions.
10483 The option is equivalent to @option{-mcpu=547x}, and is now
10484 deprecated in favor of that option.
10488 Generate output for a 68040, without using any of the new instructions.
10489 This results in code which can run relatively efficiently on either a
10490 68020/68881 or a 68030 or a 68040. The generated code does use the
10491 68881 instructions that are emulated on the 68040.
10493 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
10497 Generate output for a 68060, without using any of the new instructions.
10498 This results in code which can run relatively efficiently on either a
10499 68020/68881 or a 68030 or a 68040. The generated code does use the
10500 68881 instructions that are emulated on the 68060.
10502 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
10506 @opindex mhard-float
10508 Generate floating-point instructions. This is the default for 68020
10509 and above, and for ColdFire devices that have an FPU. It defines the
10510 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
10511 on ColdFire targets.
10514 @opindex msoft-float
10515 Do not generate floating-point instructions; use library calls instead.
10516 This is the default for 68000, 68010, and 68832 targets. It is also
10517 the default for ColdFire devices that have no FPU.
10523 Generate (do not generate) ColdFire hardware divide and remainder
10524 instructions. If @option{-march} is used without @option{-mcpu},
10525 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
10526 architectures. Otherwise, the default is taken from the target CPU
10527 (either the default CPU, or the one specified by @option{-mcpu}). For
10528 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
10529 @option{-mcpu=5206e}.
10531 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
10535 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10536 Additionally, parameters passed on the stack are also aligned to a
10537 16-bit boundary even on targets whose API mandates promotion to 32-bit.
10540 @opindex -mno-short
10541 Do not consider type @code{int} to be 16 bits wide. This is the default.
10544 @itemx -mno-bitfield
10545 @opindex mnobitfield
10546 @opindex mno-bitfield
10547 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
10548 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
10552 Do use the bit-field instructions. The @option{-m68020} option implies
10553 @option{-mbitfield}. This is the default if you use a configuration
10554 designed for a 68020.
10558 Use a different function-calling convention, in which functions
10559 that take a fixed number of arguments return with the @code{rtd}
10560 instruction, which pops their arguments while returning. This
10561 saves one instruction in the caller since there is no need to pop
10562 the arguments there.
10564 This calling convention is incompatible with the one normally
10565 used on Unix, so you cannot use it if you need to call libraries
10566 compiled with the Unix compiler.
10568 Also, you must provide function prototypes for all functions that
10569 take variable numbers of arguments (including @code{printf});
10570 otherwise incorrect code will be generated for calls to those
10573 In addition, seriously incorrect code will result if you call a
10574 function with too many arguments. (Normally, extra arguments are
10575 harmlessly ignored.)
10577 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
10578 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
10582 Do not use the calling conventions selected by @option{-mrtd}.
10583 This is the default.
10586 @itemx -mno-align-int
10587 @opindex malign-int
10588 @opindex mno-align-int
10589 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
10590 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
10591 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
10592 Aligning variables on 32-bit boundaries produces code that runs somewhat
10593 faster on processors with 32-bit busses at the expense of more memory.
10595 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
10596 align structures containing the above types differently than
10597 most published application binary interface specifications for the m68k.
10601 Use the pc-relative addressing mode of the 68000 directly, instead of
10602 using a global offset table. At present, this option implies @option{-fpic},
10603 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
10604 not presently supported with @option{-mpcrel}, though this could be supported for
10605 68020 and higher processors.
10607 @item -mno-strict-align
10608 @itemx -mstrict-align
10609 @opindex mno-strict-align
10610 @opindex mstrict-align
10611 Do not (do) assume that unaligned memory references will be handled by
10615 Generate code that allows the data segment to be located in a different
10616 area of memory from the text segment. This allows for execute in place in
10617 an environment without virtual memory management. This option implies
10620 @item -mno-sep-data
10621 Generate code that assumes that the data segment follows the text segment.
10622 This is the default.
10624 @item -mid-shared-library
10625 Generate code that supports shared libraries via the library ID method.
10626 This allows for execute in place and shared libraries in an environment
10627 without virtual memory management. This option implies @option{-fPIC}.
10629 @item -mno-id-shared-library
10630 Generate code that doesn't assume ID based shared libraries are being used.
10631 This is the default.
10633 @item -mshared-library-id=n
10634 Specified the identification number of the ID based shared library being
10635 compiled. Specifying a value of 0 will generate more compact code, specifying
10636 other values will force the allocation of that number to the current
10637 library but is no more space or time efficient than omitting this option.
10641 @node M68hc1x Options
10642 @subsection M68hc1x Options
10643 @cindex M68hc1x options
10645 These are the @samp{-m} options defined for the 68hc11 and 68hc12
10646 microcontrollers. The default values for these options depends on
10647 which style of microcontroller was selected when the compiler was configured;
10648 the defaults for the most common choices are given below.
10655 Generate output for a 68HC11. This is the default
10656 when the compiler is configured for 68HC11-based systems.
10662 Generate output for a 68HC12. This is the default
10663 when the compiler is configured for 68HC12-based systems.
10669 Generate output for a 68HCS12.
10671 @item -mauto-incdec
10672 @opindex mauto-incdec
10673 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
10680 Enable the use of 68HC12 min and max instructions.
10683 @itemx -mno-long-calls
10684 @opindex mlong-calls
10685 @opindex mno-long-calls
10686 Treat all calls as being far away (near). If calls are assumed to be
10687 far away, the compiler will use the @code{call} instruction to
10688 call a function and the @code{rtc} instruction for returning.
10692 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10694 @item -msoft-reg-count=@var{count}
10695 @opindex msoft-reg-count
10696 Specify the number of pseudo-soft registers which are used for the
10697 code generation. The maximum number is 32. Using more pseudo-soft
10698 register may or may not result in better code depending on the program.
10699 The default is 4 for 68HC11 and 2 for 68HC12.
10703 @node MCore Options
10704 @subsection MCore Options
10705 @cindex MCore options
10707 These are the @samp{-m} options defined for the Motorola M*Core
10713 @itemx -mno-hardlit
10715 @opindex mno-hardlit
10716 Inline constants into the code stream if it can be done in two
10717 instructions or less.
10723 Use the divide instruction. (Enabled by default).
10725 @item -mrelax-immediate
10726 @itemx -mno-relax-immediate
10727 @opindex mrelax-immediate
10728 @opindex mno-relax-immediate
10729 Allow arbitrary sized immediates in bit operations.
10731 @item -mwide-bitfields
10732 @itemx -mno-wide-bitfields
10733 @opindex mwide-bitfields
10734 @opindex mno-wide-bitfields
10735 Always treat bit-fields as int-sized.
10737 @item -m4byte-functions
10738 @itemx -mno-4byte-functions
10739 @opindex m4byte-functions
10740 @opindex mno-4byte-functions
10741 Force all functions to be aligned to a four byte boundary.
10743 @item -mcallgraph-data
10744 @itemx -mno-callgraph-data
10745 @opindex mcallgraph-data
10746 @opindex mno-callgraph-data
10747 Emit callgraph information.
10750 @itemx -mno-slow-bytes
10751 @opindex mslow-bytes
10752 @opindex mno-slow-bytes
10753 Prefer word access when reading byte quantities.
10755 @item -mlittle-endian
10756 @itemx -mbig-endian
10757 @opindex mlittle-endian
10758 @opindex mbig-endian
10759 Generate code for a little endian target.
10765 Generate code for the 210 processor.
10769 @subsection MIPS Options
10770 @cindex MIPS options
10776 Generate big-endian code.
10780 Generate little-endian code. This is the default for @samp{mips*el-*-*}
10783 @item -march=@var{arch}
10785 Generate code that will run on @var{arch}, which can be the name of a
10786 generic MIPS ISA, or the name of a particular processor.
10788 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
10789 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
10790 The processor names are:
10791 @samp{4kc}, @samp{4km}, @samp{4kp},
10792 @samp{4kec}, @samp{4kem}, @samp{4kep},
10793 @samp{5kc}, @samp{5kf},
10795 @samp{24kc}, @samp{24kf}, @samp{24kx},
10796 @samp{24kec}, @samp{24kef}, @samp{24kex},
10797 @samp{34kc}, @samp{34kf}, @samp{34kx},
10800 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
10801 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
10802 @samp{rm7000}, @samp{rm9000},
10805 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
10806 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
10807 The special value @samp{from-abi} selects the
10808 most compatible architecture for the selected ABI (that is,
10809 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
10811 In processor names, a final @samp{000} can be abbreviated as @samp{k}
10812 (for example, @samp{-march=r2k}). Prefixes are optional, and
10813 @samp{vr} may be written @samp{r}.
10815 GCC defines two macros based on the value of this option. The first
10816 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
10817 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
10818 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
10819 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
10820 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
10822 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
10823 above. In other words, it will have the full prefix and will not
10824 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
10825 the macro names the resolved architecture (either @samp{"mips1"} or
10826 @samp{"mips3"}). It names the default architecture when no
10827 @option{-march} option is given.
10829 @item -mtune=@var{arch}
10831 Optimize for @var{arch}. Among other things, this option controls
10832 the way instructions are scheduled, and the perceived cost of arithmetic
10833 operations. The list of @var{arch} values is the same as for
10836 When this option is not used, GCC will optimize for the processor
10837 specified by @option{-march}. By using @option{-march} and
10838 @option{-mtune} together, it is possible to generate code that will
10839 run on a family of processors, but optimize the code for one
10840 particular member of that family.
10842 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
10843 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
10844 @samp{-march} ones described above.
10848 Equivalent to @samp{-march=mips1}.
10852 Equivalent to @samp{-march=mips2}.
10856 Equivalent to @samp{-march=mips3}.
10860 Equivalent to @samp{-march=mips4}.
10864 Equivalent to @samp{-march=mips32}.
10868 Equivalent to @samp{-march=mips32r2}.
10872 Equivalent to @samp{-march=mips64}.
10877 @opindex mno-mips16
10878 Generate (do not generate) MIPS16 code. If GCC is targetting a
10879 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
10891 Generate code for the given ABI@.
10893 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
10894 generates 64-bit code when you select a 64-bit architecture, but you
10895 can use @option{-mgp32} to get 32-bit code instead.
10897 For information about the O64 ABI, see
10898 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
10900 GCC supports a variant of the o32 ABI in which floating-point registers
10901 are 64 rather than 32 bits wide. You can select this combination with
10902 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
10903 and @samp{mfhc1} instructions and is therefore only supported for
10904 MIPS32R2 processors.
10906 The register assignments for arguments and return values remain the
10907 same, but each scalar value is passed in a single 64-bit register
10908 rather than a pair of 32-bit registers. For example, scalar
10909 floating-point values are returned in @samp{$f0} only, not a
10910 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
10911 remains the same, but all 64 bits are saved.
10914 @itemx -mno-abicalls
10916 @opindex mno-abicalls
10917 Generate (do not generate) code that is suitable for SVR4-style
10918 dynamic objects. @option{-mabicalls} is the default for SVR4-based
10923 Generate (do not generate) code that is fully position-independent,
10924 and that can therefore be linked into shared libraries. This option
10925 only affects @option{-mabicalls}.
10927 All @option{-mabicalls} code has traditionally been position-independent,
10928 regardless of options like @option{-fPIC} and @option{-fpic}. However,
10929 as an extension, the GNU toolchain allows executables to use absolute
10930 accesses for locally-binding symbols. It can also use shorter GP
10931 initialization sequences and generate direct calls to locally-defined
10932 functions. This mode is selected by @option{-mno-shared}.
10934 @option{-mno-shared} depends on binutils 2.16 or higher and generates
10935 objects that can only be linked by the GNU linker. However, the option
10936 does not affect the ABI of the final executable; it only affects the ABI
10937 of relocatable objects. Using @option{-mno-shared} will generally make
10938 executables both smaller and quicker.
10940 @option{-mshared} is the default.
10946 Lift (do not lift) the usual restrictions on the size of the global
10949 GCC normally uses a single instruction to load values from the GOT@.
10950 While this is relatively efficient, it will only work if the GOT
10951 is smaller than about 64k. Anything larger will cause the linker
10952 to report an error such as:
10954 @cindex relocation truncated to fit (MIPS)
10956 relocation truncated to fit: R_MIPS_GOT16 foobar
10959 If this happens, you should recompile your code with @option{-mxgot}.
10960 It should then work with very large GOTs, although it will also be
10961 less efficient, since it will take three instructions to fetch the
10962 value of a global symbol.
10964 Note that some linkers can create multiple GOTs. If you have such a
10965 linker, you should only need to use @option{-mxgot} when a single object
10966 file accesses more than 64k's worth of GOT entries. Very few do.
10968 These options have no effect unless GCC is generating position
10973 Assume that general-purpose registers are 32 bits wide.
10977 Assume that general-purpose registers are 64 bits wide.
10981 Assume that floating-point registers are 32 bits wide.
10985 Assume that floating-point registers are 64 bits wide.
10988 @opindex mhard-float
10989 Use floating-point coprocessor instructions.
10992 @opindex msoft-float
10993 Do not use floating-point coprocessor instructions. Implement
10994 floating-point calculations using library calls instead.
10996 @item -msingle-float
10997 @opindex msingle-float
10998 Assume that the floating-point coprocessor only supports single-precision
11001 @itemx -mdouble-float
11002 @opindex mdouble-float
11003 Assume that the floating-point coprocessor supports double-precision
11004 operations. This is the default.
11010 Use (do not use) the MIPS DSP ASE. @xref{MIPS DSP Built-in Functions}.
11012 @itemx -mpaired-single
11013 @itemx -mno-paired-single
11014 @opindex mpaired-single
11015 @opindex mno-paired-single
11016 Use (do not use) paired-single floating-point instructions.
11017 @xref{MIPS Paired-Single Support}. This option can only be used
11018 when generating 64-bit code and requires hardware floating-point
11019 support to be enabled.
11024 @opindex mno-mips3d
11025 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
11026 The option @option{-mips3d} implies @option{-mpaired-single}.
11030 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
11031 an explanation of the default and the way that the pointer size is
11036 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
11038 The default size of @code{int}s, @code{long}s and pointers depends on
11039 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
11040 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
11041 32-bit @code{long}s. Pointers are the same size as @code{long}s,
11042 or the same size as integer registers, whichever is smaller.
11048 Assume (do not assume) that all symbols have 32-bit values, regardless
11049 of the selected ABI@. This option is useful in combination with
11050 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
11051 to generate shorter and faster references to symbolic addresses.
11055 @cindex smaller data references (MIPS)
11056 @cindex gp-relative references (MIPS)
11057 Put global and static items less than or equal to @var{num} bytes into
11058 the small data or bss section instead of the normal data or bss section.
11059 This allows the data to be accessed using a single instruction.
11061 All modules should be compiled with the same @option{-G @var{num}}
11064 @item -membedded-data
11065 @itemx -mno-embedded-data
11066 @opindex membedded-data
11067 @opindex mno-embedded-data
11068 Allocate variables to the read-only data section first if possible, then
11069 next in the small data section if possible, otherwise in data. This gives
11070 slightly slower code than the default, but reduces the amount of RAM required
11071 when executing, and thus may be preferred for some embedded systems.
11073 @item -muninit-const-in-rodata
11074 @itemx -mno-uninit-const-in-rodata
11075 @opindex muninit-const-in-rodata
11076 @opindex mno-uninit-const-in-rodata
11077 Put uninitialized @code{const} variables in the read-only data section.
11078 This option is only meaningful in conjunction with @option{-membedded-data}.
11080 @item -msplit-addresses
11081 @itemx -mno-split-addresses
11082 @opindex msplit-addresses
11083 @opindex mno-split-addresses
11084 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
11085 relocation operators. This option has been superseded by
11086 @option{-mexplicit-relocs} but is retained for backwards compatibility.
11088 @item -mexplicit-relocs
11089 @itemx -mno-explicit-relocs
11090 @opindex mexplicit-relocs
11091 @opindex mno-explicit-relocs
11092 Use (do not use) assembler relocation operators when dealing with symbolic
11093 addresses. The alternative, selected by @option{-mno-explicit-relocs},
11094 is to use assembler macros instead.
11096 @option{-mexplicit-relocs} is the default if GCC was configured
11097 to use an assembler that supports relocation operators.
11099 @item -mcheck-zero-division
11100 @itemx -mno-check-zero-division
11101 @opindex mcheck-zero-division
11102 @opindex mno-check-zero-division
11103 Trap (do not trap) on integer division by zero.
11105 The default is @option{-mcheck-zero-division}.
11107 @item -mdivide-traps
11108 @itemx -mdivide-breaks
11109 @opindex mdivide-traps
11110 @opindex mdivide-breaks
11111 MIPS systems check for division by zero by generating either a
11112 conditional trap or a break instruction. Using traps results in
11113 smaller code, but is only supported on MIPS II and later. Also, some
11114 versions of the Linux kernel have a bug that prevents trap from
11115 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
11116 allow conditional traps on architectures that support them and
11117 @option{-mdivide-breaks} to force the use of breaks.
11119 The default is usually @option{-mdivide-traps}, but this can be
11120 overridden at configure time using @option{--with-divide=breaks}.
11121 Divide-by-zero checks can be completely disabled using
11122 @option{-mno-check-zero-division}.
11127 @opindex mno-memcpy
11128 Force (do not force) the use of @code{memcpy()} for non-trivial block
11129 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
11130 most constant-sized copies.
11133 @itemx -mno-long-calls
11134 @opindex mlong-calls
11135 @opindex mno-long-calls
11136 Disable (do not disable) use of the @code{jal} instruction. Calling
11137 functions using @code{jal} is more efficient but requires the caller
11138 and callee to be in the same 256 megabyte segment.
11140 This option has no effect on abicalls code. The default is
11141 @option{-mno-long-calls}.
11147 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
11148 instructions, as provided by the R4650 ISA@.
11151 @itemx -mno-fused-madd
11152 @opindex mfused-madd
11153 @opindex mno-fused-madd
11154 Enable (disable) use of the floating point multiply-accumulate
11155 instructions, when they are available. The default is
11156 @option{-mfused-madd}.
11158 When multiply-accumulate instructions are used, the intermediate
11159 product is calculated to infinite precision and is not subject to
11160 the FCSR Flush to Zero bit. This may be undesirable in some
11165 Tell the MIPS assembler to not run its preprocessor over user
11166 assembler files (with a @samp{.s} suffix) when assembling them.
11169 @itemx -mno-fix-r4000
11170 @opindex mfix-r4000
11171 @opindex mno-fix-r4000
11172 Work around certain R4000 CPU errata:
11175 A double-word or a variable shift may give an incorrect result if executed
11176 immediately after starting an integer division.
11178 A double-word or a variable shift may give an incorrect result if executed
11179 while an integer multiplication is in progress.
11181 An integer division may give an incorrect result if started in a delay slot
11182 of a taken branch or a jump.
11186 @itemx -mno-fix-r4400
11187 @opindex mfix-r4400
11188 @opindex mno-fix-r4400
11189 Work around certain R4400 CPU errata:
11192 A double-word or a variable shift may give an incorrect result if executed
11193 immediately after starting an integer division.
11197 @itemx -mno-fix-vr4120
11198 @opindex mfix-vr4120
11199 Work around certain VR4120 errata:
11202 @code{dmultu} does not always produce the correct result.
11204 @code{div} and @code{ddiv} do not always produce the correct result if one
11205 of the operands is negative.
11207 The workarounds for the division errata rely on special functions in
11208 @file{libgcc.a}. At present, these functions are only provided by
11209 the @code{mips64vr*-elf} configurations.
11211 Other VR4120 errata require a nop to be inserted between certain pairs of
11212 instructions. These errata are handled by the assembler, not by GCC itself.
11215 @opindex mfix-vr4130
11216 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
11217 workarounds are implemented by the assembler rather than by GCC,
11218 although GCC will avoid using @code{mflo} and @code{mfhi} if the
11219 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
11220 instructions are available instead.
11223 @itemx -mno-fix-sb1
11225 Work around certain SB-1 CPU core errata.
11226 (This flag currently works around the SB-1 revision 2
11227 ``F1'' and ``F2'' floating point errata.)
11229 @item -mflush-func=@var{func}
11230 @itemx -mno-flush-func
11231 @opindex mflush-func
11232 Specifies the function to call to flush the I and D caches, or to not
11233 call any such function. If called, the function must take the same
11234 arguments as the common @code{_flush_func()}, that is, the address of the
11235 memory range for which the cache is being flushed, the size of the
11236 memory range, and the number 3 (to flush both caches). The default
11237 depends on the target GCC was configured for, but commonly is either
11238 @samp{_flush_func} or @samp{__cpu_flush}.
11240 @item -mbranch-likely
11241 @itemx -mno-branch-likely
11242 @opindex mbranch-likely
11243 @opindex mno-branch-likely
11244 Enable or disable use of Branch Likely instructions, regardless of the
11245 default for the selected architecture. By default, Branch Likely
11246 instructions may be generated if they are supported by the selected
11247 architecture. An exception is for the MIPS32 and MIPS64 architectures
11248 and processors which implement those architectures; for those, Branch
11249 Likely instructions will not be generated by default because the MIPS32
11250 and MIPS64 architectures specifically deprecate their use.
11252 @item -mfp-exceptions
11253 @itemx -mno-fp-exceptions
11254 @opindex mfp-exceptions
11255 Specifies whether FP exceptions are enabled. This affects how we schedule
11256 FP instructions for some processors. The default is that FP exceptions are
11259 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
11260 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
11263 @item -mvr4130-align
11264 @itemx -mno-vr4130-align
11265 @opindex mvr4130-align
11266 The VR4130 pipeline is two-way superscalar, but can only issue two
11267 instructions together if the first one is 8-byte aligned. When this
11268 option is enabled, GCC will align pairs of instructions that it
11269 thinks should execute in parallel.
11271 This option only has an effect when optimizing for the VR4130.
11272 It normally makes code faster, but at the expense of making it bigger.
11273 It is enabled by default at optimization level @option{-O3}.
11277 @subsection MMIX Options
11278 @cindex MMIX Options
11280 These options are defined for the MMIX:
11284 @itemx -mno-libfuncs
11286 @opindex mno-libfuncs
11287 Specify that intrinsic library functions are being compiled, passing all
11288 values in registers, no matter the size.
11291 @itemx -mno-epsilon
11293 @opindex mno-epsilon
11294 Generate floating-point comparison instructions that compare with respect
11295 to the @code{rE} epsilon register.
11297 @item -mabi=mmixware
11299 @opindex mabi-mmixware
11301 Generate code that passes function parameters and return values that (in
11302 the called function) are seen as registers @code{$0} and up, as opposed to
11303 the GNU ABI which uses global registers @code{$231} and up.
11305 @item -mzero-extend
11306 @itemx -mno-zero-extend
11307 @opindex mzero-extend
11308 @opindex mno-zero-extend
11309 When reading data from memory in sizes shorter than 64 bits, use (do not
11310 use) zero-extending load instructions by default, rather than
11311 sign-extending ones.
11314 @itemx -mno-knuthdiv
11316 @opindex mno-knuthdiv
11317 Make the result of a division yielding a remainder have the same sign as
11318 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
11319 remainder follows the sign of the dividend. Both methods are
11320 arithmetically valid, the latter being almost exclusively used.
11322 @item -mtoplevel-symbols
11323 @itemx -mno-toplevel-symbols
11324 @opindex mtoplevel-symbols
11325 @opindex mno-toplevel-symbols
11326 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
11327 code can be used with the @code{PREFIX} assembly directive.
11331 Generate an executable in the ELF format, rather than the default
11332 @samp{mmo} format used by the @command{mmix} simulator.
11334 @item -mbranch-predict
11335 @itemx -mno-branch-predict
11336 @opindex mbranch-predict
11337 @opindex mno-branch-predict
11338 Use (do not use) the probable-branch instructions, when static branch
11339 prediction indicates a probable branch.
11341 @item -mbase-addresses
11342 @itemx -mno-base-addresses
11343 @opindex mbase-addresses
11344 @opindex mno-base-addresses
11345 Generate (do not generate) code that uses @emph{base addresses}. Using a
11346 base address automatically generates a request (handled by the assembler
11347 and the linker) for a constant to be set up in a global register. The
11348 register is used for one or more base address requests within the range 0
11349 to 255 from the value held in the register. The generally leads to short
11350 and fast code, but the number of different data items that can be
11351 addressed is limited. This means that a program that uses lots of static
11352 data may require @option{-mno-base-addresses}.
11354 @item -msingle-exit
11355 @itemx -mno-single-exit
11356 @opindex msingle-exit
11357 @opindex mno-single-exit
11358 Force (do not force) generated code to have a single exit point in each
11362 @node MN10300 Options
11363 @subsection MN10300 Options
11364 @cindex MN10300 options
11366 These @option{-m} options are defined for Matsushita MN10300 architectures:
11371 Generate code to avoid bugs in the multiply instructions for the MN10300
11372 processors. This is the default.
11374 @item -mno-mult-bug
11375 @opindex mno-mult-bug
11376 Do not generate code to avoid bugs in the multiply instructions for the
11377 MN10300 processors.
11381 Generate code which uses features specific to the AM33 processor.
11385 Do not generate code which uses features specific to the AM33 processor. This
11388 @item -mreturn-pointer-on-d0
11389 @opindex mreturn-pointer-on-d0
11390 When generating a function which returns a pointer, return the pointer
11391 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
11392 only in a0, and attempts to call such functions without a prototype
11393 would result in errors. Note that this option is on by default; use
11394 @option{-mno-return-pointer-on-d0} to disable it.
11398 Do not link in the C run-time initialization object file.
11402 Indicate to the linker that it should perform a relaxation optimization pass
11403 to shorten branches, calls and absolute memory addresses. This option only
11404 has an effect when used on the command line for the final link step.
11406 This option makes symbolic debugging impossible.
11410 @subsection MT Options
11413 These @option{-m} options are defined for Morpho MT architectures:
11417 @item -march=@var{cpu-type}
11419 Generate code that will run on @var{cpu-type}, which is the name of a system
11420 representing a certain processor type. Possible values for
11421 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
11422 @samp{ms1-16-003} and @samp{ms2}.
11424 When this option is not used, the default is @option{-march=ms1-16-002}.
11428 Use byte loads and stores when generating code.
11432 Do not use byte loads and stores when generating code.
11436 Use simulator runtime
11440 Do not link in the C run-time initialization object file
11441 @file{crti.o}. Other run-time initialization and termination files
11442 such as @file{startup.o} and @file{exit.o} are still included on the
11443 linker command line.
11447 @node PDP-11 Options
11448 @subsection PDP-11 Options
11449 @cindex PDP-11 Options
11451 These options are defined for the PDP-11:
11456 Use hardware FPP floating point. This is the default. (FIS floating
11457 point on the PDP-11/40 is not supported.)
11460 @opindex msoft-float
11461 Do not use hardware floating point.
11465 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
11469 Return floating-point results in memory. This is the default.
11473 Generate code for a PDP-11/40.
11477 Generate code for a PDP-11/45. This is the default.
11481 Generate code for a PDP-11/10.
11483 @item -mbcopy-builtin
11484 @opindex bcopy-builtin
11485 Use inline @code{movmemhi} patterns for copying memory. This is the
11490 Do not use inline @code{movmemhi} patterns for copying memory.
11496 Use 16-bit @code{int}. This is the default.
11502 Use 32-bit @code{int}.
11505 @itemx -mno-float32
11507 @opindex mno-float32
11508 Use 64-bit @code{float}. This is the default.
11511 @itemx -mno-float64
11513 @opindex mno-float64
11514 Use 32-bit @code{float}.
11518 Use @code{abshi2} pattern. This is the default.
11522 Do not use @code{abshi2} pattern.
11524 @item -mbranch-expensive
11525 @opindex mbranch-expensive
11526 Pretend that branches are expensive. This is for experimenting with
11527 code generation only.
11529 @item -mbranch-cheap
11530 @opindex mbranch-cheap
11531 Do not pretend that branches are expensive. This is the default.
11535 Generate code for a system with split I&D@.
11539 Generate code for a system without split I&D@. This is the default.
11543 Use Unix assembler syntax. This is the default when configured for
11544 @samp{pdp11-*-bsd}.
11548 Use DEC assembler syntax. This is the default when configured for any
11549 PDP-11 target other than @samp{pdp11-*-bsd}.
11552 @node PowerPC Options
11553 @subsection PowerPC Options
11554 @cindex PowerPC options
11556 These are listed under @xref{RS/6000 and PowerPC Options}.
11558 @node RS/6000 and PowerPC Options
11559 @subsection IBM RS/6000 and PowerPC Options
11560 @cindex RS/6000 and PowerPC Options
11561 @cindex IBM RS/6000 and PowerPC Options
11563 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
11570 @itemx -mno-powerpc
11571 @itemx -mpowerpc-gpopt
11572 @itemx -mno-powerpc-gpopt
11573 @itemx -mpowerpc-gfxopt
11574 @itemx -mno-powerpc-gfxopt
11576 @itemx -mno-powerpc64
11580 @itemx -mno-popcntb
11588 @opindex mno-power2
11590 @opindex mno-powerpc
11591 @opindex mpowerpc-gpopt
11592 @opindex mno-powerpc-gpopt
11593 @opindex mpowerpc-gfxopt
11594 @opindex mno-powerpc-gfxopt
11595 @opindex mpowerpc64
11596 @opindex mno-powerpc64
11600 @opindex mno-popcntb
11604 @opindex mno-mfpgpr
11605 GCC supports two related instruction set architectures for the
11606 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
11607 instructions supported by the @samp{rios} chip set used in the original
11608 RS/6000 systems and the @dfn{PowerPC} instruction set is the
11609 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
11610 the IBM 4xx, 6xx, and follow-on microprocessors.
11612 Neither architecture is a subset of the other. However there is a
11613 large common subset of instructions supported by both. An MQ
11614 register is included in processors supporting the POWER architecture.
11616 You use these options to specify which instructions are available on the
11617 processor you are using. The default value of these options is
11618 determined when configuring GCC@. Specifying the
11619 @option{-mcpu=@var{cpu_type}} overrides the specification of these
11620 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
11621 rather than the options listed above.
11623 The @option{-mpower} option allows GCC to generate instructions that
11624 are found only in the POWER architecture and to use the MQ register.
11625 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
11626 to generate instructions that are present in the POWER2 architecture but
11627 not the original POWER architecture.
11629 The @option{-mpowerpc} option allows GCC to generate instructions that
11630 are found only in the 32-bit subset of the PowerPC architecture.
11631 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
11632 GCC to use the optional PowerPC architecture instructions in the
11633 General Purpose group, including floating-point square root. Specifying
11634 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
11635 use the optional PowerPC architecture instructions in the Graphics
11636 group, including floating-point select.
11638 The @option{-mmfcrf} option allows GCC to generate the move from
11639 condition register field instruction implemented on the POWER4
11640 processor and other processors that support the PowerPC V2.01
11642 The @option{-mpopcntb} option allows GCC to generate the popcount and
11643 double precision FP reciprocal estimate instruction implemented on the
11644 POWER5 processor and other processors that support the PowerPC V2.02
11646 The @option{-mfprnd} option allows GCC to generate the FP round to
11647 integer instructions implemented on the POWER5+ processor and other
11648 processors that support the PowerPC V2.03 architecture.
11649 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
11650 general purpose register instructions implemented on the POWER6X
11651 processor and other processors that support the extended PowerPC V2.05
11654 The @option{-mpowerpc64} option allows GCC to generate the additional
11655 64-bit instructions that are found in the full PowerPC64 architecture
11656 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
11657 @option{-mno-powerpc64}.
11659 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
11660 will use only the instructions in the common subset of both
11661 architectures plus some special AIX common-mode calls, and will not use
11662 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
11663 permits GCC to use any instruction from either architecture and to
11664 allow use of the MQ register; specify this for the Motorola MPC601.
11666 @item -mnew-mnemonics
11667 @itemx -mold-mnemonics
11668 @opindex mnew-mnemonics
11669 @opindex mold-mnemonics
11670 Select which mnemonics to use in the generated assembler code. With
11671 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
11672 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
11673 assembler mnemonics defined for the POWER architecture. Instructions
11674 defined in only one architecture have only one mnemonic; GCC uses that
11675 mnemonic irrespective of which of these options is specified.
11677 GCC defaults to the mnemonics appropriate for the architecture in
11678 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
11679 value of these option. Unless you are building a cross-compiler, you
11680 should normally not specify either @option{-mnew-mnemonics} or
11681 @option{-mold-mnemonics}, but should instead accept the default.
11683 @item -mcpu=@var{cpu_type}
11685 Set architecture type, register usage, choice of mnemonics, and
11686 instruction scheduling parameters for machine type @var{cpu_type}.
11687 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
11688 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
11689 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
11690 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
11691 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
11692 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
11693 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
11694 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6},
11695 @samp{power6x}, @samp{common}, @samp{powerpc}, @samp{powerpc64},
11696 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
11698 @option{-mcpu=common} selects a completely generic processor. Code
11699 generated under this option will run on any POWER or PowerPC processor.
11700 GCC will use only the instructions in the common subset of both
11701 architectures, and will not use the MQ register. GCC assumes a generic
11702 processor model for scheduling purposes.
11704 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
11705 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
11706 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
11707 types, with an appropriate, generic processor model assumed for
11708 scheduling purposes.
11710 The other options specify a specific processor. Code generated under
11711 those options will run best on that processor, and may not run at all on
11714 The @option{-mcpu} options automatically enable or disable the
11717 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
11718 -mnew-mnemonics -mpopcntb -mpower -mpower2 -mpowerpc64 @gol
11719 -mpowerpc-gpopt -mpowerpc-gfxopt -mstring -mmulhw -mdlmzb -mmfpgpr}
11721 The particular options set for any particular CPU will vary between
11722 compiler versions, depending on what setting seems to produce optimal
11723 code for that CPU; it doesn't necessarily reflect the actual hardware's
11724 capabilities. If you wish to set an individual option to a particular
11725 value, you may specify it after the @option{-mcpu} option, like
11726 @samp{-mcpu=970 -mno-altivec}.
11728 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
11729 not enabled or disabled by the @option{-mcpu} option at present because
11730 AIX does not have full support for these options. You may still
11731 enable or disable them individually if you're sure it'll work in your
11734 @item -mtune=@var{cpu_type}
11736 Set the instruction scheduling parameters for machine type
11737 @var{cpu_type}, but do not set the architecture type, register usage, or
11738 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
11739 values for @var{cpu_type} are used for @option{-mtune} as for
11740 @option{-mcpu}. If both are specified, the code generated will use the
11741 architecture, registers, and mnemonics set by @option{-mcpu}, but the
11742 scheduling parameters set by @option{-mtune}.
11748 Generate code to compute division as reciprocal estimate and iterative
11749 refinement, creating opportunities for increased throughput. This
11750 feature requires: optional PowerPC Graphics instruction set for single
11751 precision and FRE instruction for double precision, assuming divides
11752 cannot generate user-visible traps, and the domain values not include
11753 Infinities, denormals or zero denominator.
11756 @itemx -mno-altivec
11758 @opindex mno-altivec
11759 Generate code that uses (does not use) AltiVec instructions, and also
11760 enable the use of built-in functions that allow more direct access to
11761 the AltiVec instruction set. You may also need to set
11762 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
11768 @opindex mno-vrsave
11769 Generate VRSAVE instructions when generating AltiVec code.
11772 @opindex msecure-plt
11773 Generate code that allows ld and ld.so to build executables and shared
11774 libraries with non-exec .plt and .got sections. This is a PowerPC
11775 32-bit SYSV ABI option.
11779 Generate code that uses a BSS .plt section that ld.so fills in, and
11780 requires .plt and .got sections that are both writable and executable.
11781 This is a PowerPC 32-bit SYSV ABI option.
11787 This switch enables or disables the generation of ISEL instructions.
11789 @item -misel=@var{yes/no}
11790 This switch has been deprecated. Use @option{-misel} and
11791 @option{-mno-isel} instead.
11797 This switch enables or disables the generation of SPE simd
11800 @item -mspe=@var{yes/no}
11801 This option has been deprecated. Use @option{-mspe} and
11802 @option{-mno-spe} instead.
11804 @item -mfloat-gprs=@var{yes/single/double/no}
11805 @itemx -mfloat-gprs
11806 @opindex mfloat-gprs
11807 This switch enables or disables the generation of floating point
11808 operations on the general purpose registers for architectures that
11811 The argument @var{yes} or @var{single} enables the use of
11812 single-precision floating point operations.
11814 The argument @var{double} enables the use of single and
11815 double-precision floating point operations.
11817 The argument @var{no} disables floating point operations on the
11818 general purpose registers.
11820 This option is currently only available on the MPC854x.
11826 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
11827 targets (including GNU/Linux). The 32-bit environment sets int, long
11828 and pointer to 32 bits and generates code that runs on any PowerPC
11829 variant. The 64-bit environment sets int to 32 bits and long and
11830 pointer to 64 bits, and generates code for PowerPC64, as for
11831 @option{-mpowerpc64}.
11834 @itemx -mno-fp-in-toc
11835 @itemx -mno-sum-in-toc
11836 @itemx -mminimal-toc
11838 @opindex mno-fp-in-toc
11839 @opindex mno-sum-in-toc
11840 @opindex mminimal-toc
11841 Modify generation of the TOC (Table Of Contents), which is created for
11842 every executable file. The @option{-mfull-toc} option is selected by
11843 default. In that case, GCC will allocate at least one TOC entry for
11844 each unique non-automatic variable reference in your program. GCC
11845 will also place floating-point constants in the TOC@. However, only
11846 16,384 entries are available in the TOC@.
11848 If you receive a linker error message that saying you have overflowed
11849 the available TOC space, you can reduce the amount of TOC space used
11850 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
11851 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
11852 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
11853 generate code to calculate the sum of an address and a constant at
11854 run-time instead of putting that sum into the TOC@. You may specify one
11855 or both of these options. Each causes GCC to produce very slightly
11856 slower and larger code at the expense of conserving TOC space.
11858 If you still run out of space in the TOC even when you specify both of
11859 these options, specify @option{-mminimal-toc} instead. This option causes
11860 GCC to make only one TOC entry for every file. When you specify this
11861 option, GCC will produce code that is slower and larger but which
11862 uses extremely little TOC space. You may wish to use this option
11863 only on files that contain less frequently executed code.
11869 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
11870 @code{long} type, and the infrastructure needed to support them.
11871 Specifying @option{-maix64} implies @option{-mpowerpc64} and
11872 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
11873 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
11876 @itemx -mno-xl-compat
11877 @opindex mxl-compat
11878 @opindex mno-xl-compat
11879 Produce code that conforms more closely to IBM XL compiler semantics
11880 when using AIX-compatible ABI. Pass floating-point arguments to
11881 prototyped functions beyond the register save area (RSA) on the stack
11882 in addition to argument FPRs. Do not assume that most significant
11883 double in 128-bit long double value is properly rounded when comparing
11884 values and converting to double. Use XL symbol names for long double
11887 The AIX calling convention was extended but not initially documented to
11888 handle an obscure K&R C case of calling a function that takes the
11889 address of its arguments with fewer arguments than declared. IBM XL
11890 compilers access floating point arguments which do not fit in the
11891 RSA from the stack when a subroutine is compiled without
11892 optimization. Because always storing floating-point arguments on the
11893 stack is inefficient and rarely needed, this option is not enabled by
11894 default and only is necessary when calling subroutines compiled by IBM
11895 XL compilers without optimization.
11899 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
11900 application written to use message passing with special startup code to
11901 enable the application to run. The system must have PE installed in the
11902 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
11903 must be overridden with the @option{-specs=} option to specify the
11904 appropriate directory location. The Parallel Environment does not
11905 support threads, so the @option{-mpe} option and the @option{-pthread}
11906 option are incompatible.
11908 @item -malign-natural
11909 @itemx -malign-power
11910 @opindex malign-natural
11911 @opindex malign-power
11912 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
11913 @option{-malign-natural} overrides the ABI-defined alignment of larger
11914 types, such as floating-point doubles, on their natural size-based boundary.
11915 The option @option{-malign-power} instructs GCC to follow the ABI-specified
11916 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
11918 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
11922 @itemx -mhard-float
11923 @opindex msoft-float
11924 @opindex mhard-float
11925 Generate code that does not use (uses) the floating-point register set.
11926 Software floating point emulation is provided if you use the
11927 @option{-msoft-float} option, and pass the option to GCC when linking.
11930 @itemx -mno-multiple
11932 @opindex mno-multiple
11933 Generate code that uses (does not use) the load multiple word
11934 instructions and the store multiple word instructions. These
11935 instructions are generated by default on POWER systems, and not
11936 generated on PowerPC systems. Do not use @option{-mmultiple} on little
11937 endian PowerPC systems, since those instructions do not work when the
11938 processor is in little endian mode. The exceptions are PPC740 and
11939 PPC750 which permit the instructions usage in little endian mode.
11944 @opindex mno-string
11945 Generate code that uses (does not use) the load string instructions
11946 and the store string word instructions to save multiple registers and
11947 do small block moves. These instructions are generated by default on
11948 POWER systems, and not generated on PowerPC systems. Do not use
11949 @option{-mstring} on little endian PowerPC systems, since those
11950 instructions do not work when the processor is in little endian mode.
11951 The exceptions are PPC740 and PPC750 which permit the instructions
11952 usage in little endian mode.
11957 @opindex mno-update
11958 Generate code that uses (does not use) the load or store instructions
11959 that update the base register to the address of the calculated memory
11960 location. These instructions are generated by default. If you use
11961 @option{-mno-update}, there is a small window between the time that the
11962 stack pointer is updated and the address of the previous frame is
11963 stored, which means code that walks the stack frame across interrupts or
11964 signals may get corrupted data.
11967 @itemx -mno-fused-madd
11968 @opindex mfused-madd
11969 @opindex mno-fused-madd
11970 Generate code that uses (does not use) the floating point multiply and
11971 accumulate instructions. These instructions are generated by default if
11972 hardware floating is used.
11978 Generate code that uses (does not use) the half-word multiply and
11979 multiply-accumulate instructions on the IBM 405 and 440 processors.
11980 These instructions are generated by default when targetting those
11987 Generate code that uses (does not use) the string-search @samp{dlmzb}
11988 instruction on the IBM 405 and 440 processors. This instruction is
11989 generated by default when targetting those processors.
11991 @item -mno-bit-align
11993 @opindex mno-bit-align
11994 @opindex mbit-align
11995 On System V.4 and embedded PowerPC systems do not (do) force structures
11996 and unions that contain bit-fields to be aligned to the base type of the
11999 For example, by default a structure containing nothing but 8
12000 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
12001 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
12002 the structure would be aligned to a 1 byte boundary and be one byte in
12005 @item -mno-strict-align
12006 @itemx -mstrict-align
12007 @opindex mno-strict-align
12008 @opindex mstrict-align
12009 On System V.4 and embedded PowerPC systems do not (do) assume that
12010 unaligned memory references will be handled by the system.
12012 @item -mrelocatable
12013 @itemx -mno-relocatable
12014 @opindex mrelocatable
12015 @opindex mno-relocatable
12016 On embedded PowerPC systems generate code that allows (does not allow)
12017 the program to be relocated to a different address at runtime. If you
12018 use @option{-mrelocatable} on any module, all objects linked together must
12019 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
12021 @item -mrelocatable-lib
12022 @itemx -mno-relocatable-lib
12023 @opindex mrelocatable-lib
12024 @opindex mno-relocatable-lib
12025 On embedded PowerPC systems generate code that allows (does not allow)
12026 the program to be relocated to a different address at runtime. Modules
12027 compiled with @option{-mrelocatable-lib} can be linked with either modules
12028 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
12029 with modules compiled with the @option{-mrelocatable} options.
12035 On System V.4 and embedded PowerPC systems do not (do) assume that
12036 register 2 contains a pointer to a global area pointing to the addresses
12037 used in the program.
12040 @itemx -mlittle-endian
12042 @opindex mlittle-endian
12043 On System V.4 and embedded PowerPC systems compile code for the
12044 processor in little endian mode. The @option{-mlittle-endian} option is
12045 the same as @option{-mlittle}.
12048 @itemx -mbig-endian
12050 @opindex mbig-endian
12051 On System V.4 and embedded PowerPC systems compile code for the
12052 processor in big endian mode. The @option{-mbig-endian} option is
12053 the same as @option{-mbig}.
12055 @item -mdynamic-no-pic
12056 @opindex mdynamic-no-pic
12057 On Darwin and Mac OS X systems, compile code so that it is not
12058 relocatable, but that its external references are relocatable. The
12059 resulting code is suitable for applications, but not shared
12062 @item -mprioritize-restricted-insns=@var{priority}
12063 @opindex mprioritize-restricted-insns
12064 This option controls the priority that is assigned to
12065 dispatch-slot restricted instructions during the second scheduling
12066 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
12067 @var{no/highest/second-highest} priority to dispatch slot restricted
12070 @item -msched-costly-dep=@var{dependence_type}
12071 @opindex msched-costly-dep
12072 This option controls which dependences are considered costly
12073 by the target during instruction scheduling. The argument
12074 @var{dependence_type} takes one of the following values:
12075 @var{no}: no dependence is costly,
12076 @var{all}: all dependences are costly,
12077 @var{true_store_to_load}: a true dependence from store to load is costly,
12078 @var{store_to_load}: any dependence from store to load is costly,
12079 @var{number}: any dependence which latency >= @var{number} is costly.
12081 @item -minsert-sched-nops=@var{scheme}
12082 @opindex minsert-sched-nops
12083 This option controls which nop insertion scheme will be used during
12084 the second scheduling pass. The argument @var{scheme} takes one of the
12086 @var{no}: Don't insert nops.
12087 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
12088 according to the scheduler's grouping.
12089 @var{regroup_exact}: Insert nops to force costly dependent insns into
12090 separate groups. Insert exactly as many nops as needed to force an insn
12091 to a new group, according to the estimated processor grouping.
12092 @var{number}: Insert nops to force costly dependent insns into
12093 separate groups. Insert @var{number} nops to force an insn to a new group.
12096 @opindex mcall-sysv
12097 On System V.4 and embedded PowerPC systems compile code using calling
12098 conventions that adheres to the March 1995 draft of the System V
12099 Application Binary Interface, PowerPC processor supplement. This is the
12100 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
12102 @item -mcall-sysv-eabi
12103 @opindex mcall-sysv-eabi
12104 Specify both @option{-mcall-sysv} and @option{-meabi} options.
12106 @item -mcall-sysv-noeabi
12107 @opindex mcall-sysv-noeabi
12108 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
12110 @item -mcall-solaris
12111 @opindex mcall-solaris
12112 On System V.4 and embedded PowerPC systems compile code for the Solaris
12116 @opindex mcall-linux
12117 On System V.4 and embedded PowerPC systems compile code for the
12118 Linux-based GNU system.
12122 On System V.4 and embedded PowerPC systems compile code for the
12123 Hurd-based GNU system.
12125 @item -mcall-netbsd
12126 @opindex mcall-netbsd
12127 On System V.4 and embedded PowerPC systems compile code for the
12128 NetBSD operating system.
12130 @item -maix-struct-return
12131 @opindex maix-struct-return
12132 Return all structures in memory (as specified by the AIX ABI)@.
12134 @item -msvr4-struct-return
12135 @opindex msvr4-struct-return
12136 Return structures smaller than 8 bytes in registers (as specified by the
12139 @item -mabi=@var{abi-type}
12141 Extend the current ABI with a particular extension, or remove such extension.
12142 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
12143 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
12147 Extend the current ABI with SPE ABI extensions. This does not change
12148 the default ABI, instead it adds the SPE ABI extensions to the current
12152 @opindex mabi=no-spe
12153 Disable Booke SPE ABI extensions for the current ABI@.
12155 @item -mabi=ibmlongdouble
12156 @opindex mabi=ibmlongdouble
12157 Change the current ABI to use IBM extended precision long double.
12158 This is a PowerPC 32-bit SYSV ABI option.
12160 @item -mabi=ieeelongdouble
12161 @opindex mabi=ieeelongdouble
12162 Change the current ABI to use IEEE extended precision long double.
12163 This is a PowerPC 32-bit Linux ABI option.
12166 @itemx -mno-prototype
12167 @opindex mprototype
12168 @opindex mno-prototype
12169 On System V.4 and embedded PowerPC systems assume that all calls to
12170 variable argument functions are properly prototyped. Otherwise, the
12171 compiler must insert an instruction before every non prototyped call to
12172 set or clear bit 6 of the condition code register (@var{CR}) to
12173 indicate whether floating point values were passed in the floating point
12174 registers in case the function takes a variable arguments. With
12175 @option{-mprototype}, only calls to prototyped variable argument functions
12176 will set or clear the bit.
12180 On embedded PowerPC systems, assume that the startup module is called
12181 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
12182 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
12187 On embedded PowerPC systems, assume that the startup module is called
12188 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
12193 On embedded PowerPC systems, assume that the startup module is called
12194 @file{crt0.o} and the standard C libraries are @file{libads.a} and
12197 @item -myellowknife
12198 @opindex myellowknife
12199 On embedded PowerPC systems, assume that the startup module is called
12200 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
12205 On System V.4 and embedded PowerPC systems, specify that you are
12206 compiling for a VxWorks system.
12210 Specify that you are compiling for the WindISS simulation environment.
12214 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
12215 header to indicate that @samp{eabi} extended relocations are used.
12221 On System V.4 and embedded PowerPC systems do (do not) adhere to the
12222 Embedded Applications Binary Interface (eabi) which is a set of
12223 modifications to the System V.4 specifications. Selecting @option{-meabi}
12224 means that the stack is aligned to an 8 byte boundary, a function
12225 @code{__eabi} is called to from @code{main} to set up the eabi
12226 environment, and the @option{-msdata} option can use both @code{r2} and
12227 @code{r13} to point to two separate small data areas. Selecting
12228 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
12229 do not call an initialization function from @code{main}, and the
12230 @option{-msdata} option will only use @code{r13} to point to a single
12231 small data area. The @option{-meabi} option is on by default if you
12232 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
12235 @opindex msdata=eabi
12236 On System V.4 and embedded PowerPC systems, put small initialized
12237 @code{const} global and static data in the @samp{.sdata2} section, which
12238 is pointed to by register @code{r2}. Put small initialized
12239 non-@code{const} global and static data in the @samp{.sdata} section,
12240 which is pointed to by register @code{r13}. Put small uninitialized
12241 global and static data in the @samp{.sbss} section, which is adjacent to
12242 the @samp{.sdata} section. The @option{-msdata=eabi} option is
12243 incompatible with the @option{-mrelocatable} option. The
12244 @option{-msdata=eabi} option also sets the @option{-memb} option.
12247 @opindex msdata=sysv
12248 On System V.4 and embedded PowerPC systems, put small global and static
12249 data in the @samp{.sdata} section, which is pointed to by register
12250 @code{r13}. Put small uninitialized global and static data in the
12251 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
12252 The @option{-msdata=sysv} option is incompatible with the
12253 @option{-mrelocatable} option.
12255 @item -msdata=default
12257 @opindex msdata=default
12259 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
12260 compile code the same as @option{-msdata=eabi}, otherwise compile code the
12261 same as @option{-msdata=sysv}.
12264 @opindex msdata-data
12265 On System V.4 and embedded PowerPC systems, put small global
12266 data in the @samp{.sdata} section. Put small uninitialized global
12267 data in the @samp{.sbss} section. Do not use register @code{r13}
12268 to address small data however. This is the default behavior unless
12269 other @option{-msdata} options are used.
12273 @opindex msdata=none
12275 On embedded PowerPC systems, put all initialized global and static data
12276 in the @samp{.data} section, and all uninitialized data in the
12277 @samp{.bss} section.
12281 @cindex smaller data references (PowerPC)
12282 @cindex .sdata/.sdata2 references (PowerPC)
12283 On embedded PowerPC systems, put global and static items less than or
12284 equal to @var{num} bytes into the small data or bss sections instead of
12285 the normal data or bss section. By default, @var{num} is 8. The
12286 @option{-G @var{num}} switch is also passed to the linker.
12287 All modules should be compiled with the same @option{-G @var{num}} value.
12290 @itemx -mno-regnames
12292 @opindex mno-regnames
12293 On System V.4 and embedded PowerPC systems do (do not) emit register
12294 names in the assembly language output using symbolic forms.
12297 @itemx -mno-longcall
12299 @opindex mno-longcall
12300 By default assume that all calls are far away so that a longer more
12301 expensive calling sequence is required. This is required for calls
12302 further than 32 megabytes (33,554,432 bytes) from the current location.
12303 A short call will be generated if the compiler knows
12304 the call cannot be that far away. This setting can be overridden by
12305 the @code{shortcall} function attribute, or by @code{#pragma
12308 Some linkers are capable of detecting out-of-range calls and generating
12309 glue code on the fly. On these systems, long calls are unnecessary and
12310 generate slower code. As of this writing, the AIX linker can do this,
12311 as can the GNU linker for PowerPC/64. It is planned to add this feature
12312 to the GNU linker for 32-bit PowerPC systems as well.
12314 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
12315 callee, L42'', plus a ``branch island'' (glue code). The two target
12316 addresses represent the callee and the ``branch island''. The
12317 Darwin/PPC linker will prefer the first address and generate a ``bl
12318 callee'' if the PPC ``bl'' instruction will reach the callee directly;
12319 otherwise, the linker will generate ``bl L42'' to call the ``branch
12320 island''. The ``branch island'' is appended to the body of the
12321 calling function; it computes the full 32-bit address of the callee
12324 On Mach-O (Darwin) systems, this option directs the compiler emit to
12325 the glue for every direct call, and the Darwin linker decides whether
12326 to use or discard it.
12328 In the future, we may cause GCC to ignore all longcall specifications
12329 when the linker is known to generate glue.
12333 Adds support for multithreading with the @dfn{pthreads} library.
12334 This option sets flags for both the preprocessor and linker.
12338 @node S/390 and zSeries Options
12339 @subsection S/390 and zSeries Options
12340 @cindex S/390 and zSeries Options
12342 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
12346 @itemx -msoft-float
12347 @opindex mhard-float
12348 @opindex msoft-float
12349 Use (do not use) the hardware floating-point instructions and registers
12350 for floating-point operations. When @option{-msoft-float} is specified,
12351 functions in @file{libgcc.a} will be used to perform floating-point
12352 operations. When @option{-mhard-float} is specified, the compiler
12353 generates IEEE floating-point instructions. This is the default.
12355 @item -mlong-double-64
12356 @itemx -mlong-double-128
12357 @opindex mlong-double-64
12358 @opindex mlong-double-128
12359 These switches control the size of @code{long double} type. A size
12360 of 64bit makes the @code{long double} type equivalent to the @code{double}
12361 type. This is the default.
12364 @itemx -mno-backchain
12365 @opindex mbackchain
12366 @opindex mno-backchain
12367 Store (do not store) the address of the caller's frame as backchain pointer
12368 into the callee's stack frame.
12369 A backchain may be needed to allow debugging using tools that do not understand
12370 DWARF-2 call frame information.
12371 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
12372 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
12373 the backchain is placed into the topmost word of the 96/160 byte register
12376 In general, code compiled with @option{-mbackchain} is call-compatible with
12377 code compiled with @option{-mmo-backchain}; however, use of the backchain
12378 for debugging purposes usually requires that the whole binary is built with
12379 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
12380 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
12381 to build a linux kernel use @option{-msoft-float}.
12383 The default is to not maintain the backchain.
12385 @item -mpacked-stack
12386 @item -mno-packed-stack
12387 @opindex mpacked-stack
12388 @opindex mno-packed-stack
12389 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
12390 specified, the compiler uses the all fields of the 96/160 byte register save
12391 area only for their default purpose; unused fields still take up stack space.
12392 When @option{-mpacked-stack} is specified, register save slots are densely
12393 packed at the top of the register save area; unused space is reused for other
12394 purposes, allowing for more efficient use of the available stack space.
12395 However, when @option{-mbackchain} is also in effect, the topmost word of
12396 the save area is always used to store the backchain, and the return address
12397 register is always saved two words below the backchain.
12399 As long as the stack frame backchain is not used, code generated with
12400 @option{-mpacked-stack} is call-compatible with code generated with
12401 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
12402 S/390 or zSeries generated code that uses the stack frame backchain at run
12403 time, not just for debugging purposes. Such code is not call-compatible
12404 with code compiled with @option{-mpacked-stack}. Also, note that the
12405 combination of @option{-mbackchain},
12406 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
12407 to build a linux kernel use @option{-msoft-float}.
12409 The default is to not use the packed stack layout.
12412 @itemx -mno-small-exec
12413 @opindex msmall-exec
12414 @opindex mno-small-exec
12415 Generate (or do not generate) code using the @code{bras} instruction
12416 to do subroutine calls.
12417 This only works reliably if the total executable size does not
12418 exceed 64k. The default is to use the @code{basr} instruction instead,
12419 which does not have this limitation.
12425 When @option{-m31} is specified, generate code compliant to the
12426 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
12427 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
12428 particular to generate 64-bit instructions. For the @samp{s390}
12429 targets, the default is @option{-m31}, while the @samp{s390x}
12430 targets default to @option{-m64}.
12436 When @option{-mzarch} is specified, generate code using the
12437 instructions available on z/Architecture.
12438 When @option{-mesa} is specified, generate code using the
12439 instructions available on ESA/390. Note that @option{-mesa} is
12440 not possible with @option{-m64}.
12441 When generating code compliant to the GNU/Linux for S/390 ABI,
12442 the default is @option{-mesa}. When generating code compliant
12443 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
12449 Generate (or do not generate) code using the @code{mvcle} instruction
12450 to perform block moves. When @option{-mno-mvcle} is specified,
12451 use a @code{mvc} loop instead. This is the default unless optimizing for
12458 Print (or do not print) additional debug information when compiling.
12459 The default is to not print debug information.
12461 @item -march=@var{cpu-type}
12463 Generate code that will run on @var{cpu-type}, which is the name of a system
12464 representing a certain processor type. Possible values for
12465 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
12466 When generating code using the instructions available on z/Architecture,
12467 the default is @option{-march=z900}. Otherwise, the default is
12468 @option{-march=g5}.
12470 @item -mtune=@var{cpu-type}
12472 Tune to @var{cpu-type} everything applicable about the generated code,
12473 except for the ABI and the set of available instructions.
12474 The list of @var{cpu-type} values is the same as for @option{-march}.
12475 The default is the value used for @option{-march}.
12478 @itemx -mno-tpf-trace
12479 @opindex mtpf-trace
12480 @opindex mno-tpf-trace
12481 Generate code that adds (does not add) in TPF OS specific branches to trace
12482 routines in the operating system. This option is off by default, even
12483 when compiling for the TPF OS@.
12486 @itemx -mno-fused-madd
12487 @opindex mfused-madd
12488 @opindex mno-fused-madd
12489 Generate code that uses (does not use) the floating point multiply and
12490 accumulate instructions. These instructions are generated by default if
12491 hardware floating point is used.
12493 @item -mwarn-framesize=@var{framesize}
12494 @opindex mwarn-framesize
12495 Emit a warning if the current function exceeds the given frame size. Because
12496 this is a compile time check it doesn't need to be a real problem when the program
12497 runs. It is intended to identify functions which most probably cause
12498 a stack overflow. It is useful to be used in an environment with limited stack
12499 size e.g.@: the linux kernel.
12501 @item -mwarn-dynamicstack
12502 @opindex mwarn-dynamicstack
12503 Emit a warning if the function calls alloca or uses dynamically
12504 sized arrays. This is generally a bad idea with a limited stack size.
12506 @item -mstack-guard=@var{stack-guard}
12507 @item -mstack-size=@var{stack-size}
12508 @opindex mstack-guard
12509 @opindex mstack-size
12510 These arguments always have to be used in conjunction. If they are present the s390
12511 back end emits additional instructions in the function prologue which trigger a trap
12512 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
12513 (remember that the stack on s390 grows downward). These options are intended to
12514 be used to help debugging stack overflow problems. The additionally emitted code
12515 causes only little overhead and hence can also be used in production like systems
12516 without greater performance degradation. The given values have to be exact
12517 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard} without
12519 In order to be efficient the extra code makes the assumption that the stack starts
12520 at an address aligned to the value given by @var{stack-size}.
12523 @node Score Options
12524 @subsection Score Options
12525 @cindex Score Options
12527 These options are defined for Score implementations:
12532 Compile code for little endian mode.
12536 Compile code for big endian mode. This is the default.
12540 Enable the use of multiply-accumulate instructions. Disabled by default.
12544 Specify the SCORE5U of the target architecture.
12548 Specify the SCORE7 of the target architecture. This is the default.
12552 @subsection SH Options
12554 These @samp{-m} options are defined for the SH implementations:
12559 Generate code for the SH1.
12563 Generate code for the SH2.
12566 Generate code for the SH2e.
12570 Generate code for the SH3.
12574 Generate code for the SH3e.
12578 Generate code for the SH4 without a floating-point unit.
12580 @item -m4-single-only
12581 @opindex m4-single-only
12582 Generate code for the SH4 with a floating-point unit that only
12583 supports single-precision arithmetic.
12587 Generate code for the SH4 assuming the floating-point unit is in
12588 single-precision mode by default.
12592 Generate code for the SH4.
12596 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
12597 floating-point unit is not used.
12599 @item -m4a-single-only
12600 @opindex m4a-single-only
12601 Generate code for the SH4a, in such a way that no double-precision
12602 floating point operations are used.
12605 @opindex m4a-single
12606 Generate code for the SH4a assuming the floating-point unit is in
12607 single-precision mode by default.
12611 Generate code for the SH4a.
12615 Same as @option{-m4a-nofpu}, except that it implicitly passes
12616 @option{-dsp} to the assembler. GCC doesn't generate any DSP
12617 instructions at the moment.
12621 Compile code for the processor in big endian mode.
12625 Compile code for the processor in little endian mode.
12629 Align doubles at 64-bit boundaries. Note that this changes the calling
12630 conventions, and thus some functions from the standard C library will
12631 not work unless you recompile it first with @option{-mdalign}.
12635 Shorten some address references at link time, when possible; uses the
12636 linker option @option{-relax}.
12640 Use 32-bit offsets in @code{switch} tables. The default is to use
12645 Enable the use of the instruction @code{fmovd}.
12649 Comply with the calling conventions defined by Renesas.
12653 Comply with the calling conventions defined by Renesas.
12657 Comply with the calling conventions defined for GCC before the Renesas
12658 conventions were available. This option is the default for all
12659 targets of the SH toolchain except for @samp{sh-symbianelf}.
12662 @opindex mnomacsave
12663 Mark the @code{MAC} register as call-clobbered, even if
12664 @option{-mhitachi} is given.
12668 Increase IEEE-compliance of floating-point code.
12669 At the moment, this is equivalent to @option{-fno-finite-math-only}.
12670 When generating 16 bit SH opcodes, getting IEEE-conforming results for
12671 comparisons of NANs / infinities incurs extra overhead in every
12672 floating point comparison, therefore the default is set to
12673 @option{-ffinite-math-only}.
12675 @item -minline-ic_invalidate
12676 @opindex minline-ic_invalidate
12677 Inline code to invalidate instruction cache entries after setting up
12678 nested function trampolines.
12679 This option has no effect if -musermode is in effect and the selected
12680 code generation option (e.g. -m4) does not allow the use of the icbi
12682 If the selected code generation option does not allow the use of the icbi
12683 instruction, and -musermode is not in effect, the inlined code will
12684 manipulate the instruction cache address array directly with an associative
12685 write. This not only requires privileged mode, but it will also
12686 fail if the cache line had been mapped via the TLB and has become unmapped.
12690 Dump instruction size and location in the assembly code.
12693 @opindex mpadstruct
12694 This option is deprecated. It pads structures to multiple of 4 bytes,
12695 which is incompatible with the SH ABI@.
12699 Optimize for space instead of speed. Implied by @option{-Os}.
12702 @opindex mprefergot
12703 When generating position-independent code, emit function calls using
12704 the Global Offset Table instead of the Procedure Linkage Table.
12708 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
12709 if the inlined code would not work in user mode.
12710 This is the default when the target is @code{sh-*-linux*}.
12712 @item -multcost=@var{number}
12713 @opindex multcost=@var{number}
12714 Set the cost to assume for a multiply insn.
12716 @item -mdiv=@var{strategy}
12717 @opindex mdiv=@var{strategy}
12718 Set the division strategy to use for SHmedia code. @var{strategy} must be
12719 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
12720 inv:call2, inv:fp .
12721 "fp" performs the operation in floating point. This has a very high latency,
12722 but needs only a few instructions, so it might be a good choice if
12723 your code has enough easily exploitable ILP to allow the compiler to
12724 schedule the floating point instructions together with other instructions.
12725 Division by zero causes a floating point exception.
12726 "inv" uses integer operations to calculate the inverse of the divisor,
12727 and then multiplies the dividend with the inverse. This strategy allows
12728 cse and hoisting of the inverse calculation. Division by zero calculates
12729 an unspecified result, but does not trap.
12730 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
12731 have been found, or if the entire operation has been hoisted to the same
12732 place, the last stages of the inverse calculation are intertwined with the
12733 final multiply to reduce the overall latency, at the expense of using a few
12734 more instructions, and thus offering fewer scheduling opportunities with
12736 "call" calls a library function that usually implements the inv:minlat
12738 This gives high code density for m5-*media-nofpu compilations.
12739 "call2" uses a different entry point of the same library function, where it
12740 assumes that a pointer to a lookup table has already been set up, which
12741 exposes the pointer load to cse / code hoisting optimizations.
12742 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
12743 code generation, but if the code stays unoptimized, revert to the "call",
12744 "call2", or "fp" strategies, respectively. Note that the
12745 potentially-trapping side effect of division by zero is carried by a
12746 separate instruction, so it is possible that all the integer instructions
12747 are hoisted out, but the marker for the side effect stays where it is.
12748 A recombination to fp operations or a call is not possible in that case.
12749 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
12750 that the inverse calculation was nor separated from the multiply, they speed
12751 up division where the dividend fits into 20 bits (plus sign where applicable),
12752 by inserting a test to skip a number of operations in this case; this test
12753 slows down the case of larger dividends. inv20u assumes the case of a such
12754 a small dividend to be unlikely, and inv20l assumes it to be likely.
12756 @item -mdivsi3_libfunc=@var{name}
12757 @opindex mdivsi3_libfunc=@var{name}
12758 Set the name of the library function used for 32 bit signed division to
12759 @var{name}. This only affect the name used in the call and inv:call
12760 division strategies, and the compiler will still expect the same
12761 sets of input/output/clobbered registers as if this option was not present.
12763 @item -madjust-unroll
12764 @opindex madjust-unroll
12765 Throttle unrolling to avoid thrashing target registers.
12766 This option only has an effect if the gcc code base supports the
12767 TARGET_ADJUST_UNROLL_MAX target hook.
12769 @item -mindexed-addressing
12770 @opindex mindexed-addressing
12771 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
12772 This is only safe if the hardware and/or OS implement 32 bit wrap-around
12773 semantics for the indexed addressing mode. The architecture allows the
12774 implementation of processors with 64 bit MMU, which the OS could use to
12775 get 32 bit addressing, but since no current hardware implementation supports
12776 this or any other way to make the indexed addressing mode safe to use in
12777 the 32 bit ABI, the default is -mno-indexed-addressing.
12779 @item -mgettrcost=@var{number}
12780 @opindex mgettrcost=@var{number}
12781 Set the cost assumed for the gettr instruction to @var{number}.
12782 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
12786 Assume pt* instructions won't trap. This will generally generate better
12787 scheduled code, but is unsafe on current hardware. The current architecture
12788 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
12789 This has the unintentional effect of making it unsafe to schedule ptabs /
12790 ptrel before a branch, or hoist it out of a loop. For example,
12791 __do_global_ctors, a part of libgcc that runs constructors at program
12792 startup, calls functions in a list which is delimited by @minus{}1. With the
12793 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
12794 That means that all the constructors will be run a bit quicker, but when
12795 the loop comes to the end of the list, the program crashes because ptabs
12796 loads @minus{}1 into a target register. Since this option is unsafe for any
12797 hardware implementing the current architecture specification, the default
12798 is -mno-pt-fixed. Unless the user specifies a specific cost with
12799 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
12800 this deters register allocation using target registers for storing
12803 @item -minvalid-symbols
12804 @opindex minvalid-symbols
12805 Assume symbols might be invalid. Ordinary function symbols generated by
12806 the compiler will always be valid to load with movi/shori/ptabs or
12807 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
12808 to generate symbols that will cause ptabs / ptrel to trap.
12809 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
12810 It will then prevent cross-basic-block cse, hoisting and most scheduling
12811 of symbol loads. The default is @option{-mno-invalid-symbols}.
12814 @node SPARC Options
12815 @subsection SPARC Options
12816 @cindex SPARC options
12818 These @samp{-m} options are supported on the SPARC:
12821 @item -mno-app-regs
12823 @opindex mno-app-regs
12825 Specify @option{-mapp-regs} to generate output using the global registers
12826 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
12829 To be fully SVR4 ABI compliant at the cost of some performance loss,
12830 specify @option{-mno-app-regs}. You should compile libraries and system
12831 software with this option.
12834 @itemx -mhard-float
12836 @opindex mhard-float
12837 Generate output containing floating point instructions. This is the
12841 @itemx -msoft-float
12843 @opindex msoft-float
12844 Generate output containing library calls for floating point.
12845 @strong{Warning:} the requisite libraries are not available for all SPARC
12846 targets. Normally the facilities of the machine's usual C compiler are
12847 used, but this cannot be done directly in cross-compilation. You must make
12848 your own arrangements to provide suitable library functions for
12849 cross-compilation. The embedded targets @samp{sparc-*-aout} and
12850 @samp{sparclite-*-*} do provide software floating point support.
12852 @option{-msoft-float} changes the calling convention in the output file;
12853 therefore, it is only useful if you compile @emph{all} of a program with
12854 this option. In particular, you need to compile @file{libgcc.a}, the
12855 library that comes with GCC, with @option{-msoft-float} in order for
12858 @item -mhard-quad-float
12859 @opindex mhard-quad-float
12860 Generate output containing quad-word (long double) floating point
12863 @item -msoft-quad-float
12864 @opindex msoft-quad-float
12865 Generate output containing library calls for quad-word (long double)
12866 floating point instructions. The functions called are those specified
12867 in the SPARC ABI@. This is the default.
12869 As of this writing, there are no SPARC implementations that have hardware
12870 support for the quad-word floating point instructions. They all invoke
12871 a trap handler for one of these instructions, and then the trap handler
12872 emulates the effect of the instruction. Because of the trap handler overhead,
12873 this is much slower than calling the ABI library routines. Thus the
12874 @option{-msoft-quad-float} option is the default.
12876 @item -mno-unaligned-doubles
12877 @itemx -munaligned-doubles
12878 @opindex mno-unaligned-doubles
12879 @opindex munaligned-doubles
12880 Assume that doubles have 8 byte alignment. This is the default.
12882 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
12883 alignment only if they are contained in another type, or if they have an
12884 absolute address. Otherwise, it assumes they have 4 byte alignment.
12885 Specifying this option avoids some rare compatibility problems with code
12886 generated by other compilers. It is not the default because it results
12887 in a performance loss, especially for floating point code.
12889 @item -mno-faster-structs
12890 @itemx -mfaster-structs
12891 @opindex mno-faster-structs
12892 @opindex mfaster-structs
12893 With @option{-mfaster-structs}, the compiler assumes that structures
12894 should have 8 byte alignment. This enables the use of pairs of
12895 @code{ldd} and @code{std} instructions for copies in structure
12896 assignment, in place of twice as many @code{ld} and @code{st} pairs.
12897 However, the use of this changed alignment directly violates the SPARC
12898 ABI@. Thus, it's intended only for use on targets where the developer
12899 acknowledges that their resulting code will not be directly in line with
12900 the rules of the ABI@.
12902 @item -mimpure-text
12903 @opindex mimpure-text
12904 @option{-mimpure-text}, used in addition to @option{-shared}, tells
12905 the compiler to not pass @option{-z text} to the linker when linking a
12906 shared object. Using this option, you can link position-dependent
12907 code into a shared object.
12909 @option{-mimpure-text} suppresses the ``relocations remain against
12910 allocatable but non-writable sections'' linker error message.
12911 However, the necessary relocations will trigger copy-on-write, and the
12912 shared object is not actually shared across processes. Instead of
12913 using @option{-mimpure-text}, you should compile all source code with
12914 @option{-fpic} or @option{-fPIC}.
12916 This option is only available on SunOS and Solaris.
12918 @item -mcpu=@var{cpu_type}
12920 Set the instruction set, register set, and instruction scheduling parameters
12921 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
12922 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
12923 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
12924 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
12925 @samp{ultrasparc3}, and @samp{niagara}.
12927 Default instruction scheduling parameters are used for values that select
12928 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
12929 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
12931 Here is a list of each supported architecture and their supported
12936 v8: supersparc, hypersparc
12937 sparclite: f930, f934, sparclite86x
12939 v9: ultrasparc, ultrasparc3, niagara
12942 By default (unless configured otherwise), GCC generates code for the V7
12943 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
12944 additionally optimizes it for the Cypress CY7C602 chip, as used in the
12945 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
12946 SPARCStation 1, 2, IPX etc.
12948 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
12949 architecture. The only difference from V7 code is that the compiler emits
12950 the integer multiply and integer divide instructions which exist in SPARC-V8
12951 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
12952 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
12955 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
12956 the SPARC architecture. This adds the integer multiply, integer divide step
12957 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
12958 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
12959 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
12960 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
12961 MB86934 chip, which is the more recent SPARClite with FPU@.
12963 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
12964 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
12965 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
12966 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
12967 optimizes it for the TEMIC SPARClet chip.
12969 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
12970 architecture. This adds 64-bit integer and floating-point move instructions,
12971 3 additional floating-point condition code registers and conditional move
12972 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
12973 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
12974 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
12975 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
12976 @option{-mcpu=niagara}, the compiler additionally optimizes it for
12977 Sun UltraSPARC T1 chips.
12979 @item -mtune=@var{cpu_type}
12981 Set the instruction scheduling parameters for machine type
12982 @var{cpu_type}, but do not set the instruction set or register set that the
12983 option @option{-mcpu=@var{cpu_type}} would.
12985 The same values for @option{-mcpu=@var{cpu_type}} can be used for
12986 @option{-mtune=@var{cpu_type}}, but the only useful values are those
12987 that select a particular cpu implementation. Those are @samp{cypress},
12988 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
12989 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
12990 @samp{ultrasparc3}, and @samp{niagara}.
12995 @opindex mno-v8plus
12996 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
12997 difference from the V8 ABI is that the global and out registers are
12998 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
12999 mode for all SPARC-V9 processors.
13005 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
13006 Visual Instruction Set extensions. The default is @option{-mno-vis}.
13009 These @samp{-m} options are supported in addition to the above
13010 on SPARC-V9 processors in 64-bit environments:
13013 @item -mlittle-endian
13014 @opindex mlittle-endian
13015 Generate code for a processor running in little-endian mode. It is only
13016 available for a few configurations and most notably not on Solaris and Linux.
13022 Generate code for a 32-bit or 64-bit environment.
13023 The 32-bit environment sets int, long and pointer to 32 bits.
13024 The 64-bit environment sets int to 32 bits and long and pointer
13027 @item -mcmodel=medlow
13028 @opindex mcmodel=medlow
13029 Generate code for the Medium/Low code model: 64-bit addresses, programs
13030 must be linked in the low 32 bits of memory. Programs can be statically
13031 or dynamically linked.
13033 @item -mcmodel=medmid
13034 @opindex mcmodel=medmid
13035 Generate code for the Medium/Middle code model: 64-bit addresses, programs
13036 must be linked in the low 44 bits of memory, the text and data segments must
13037 be less than 2GB in size and the data segment must be located within 2GB of
13040 @item -mcmodel=medany
13041 @opindex mcmodel=medany
13042 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
13043 may be linked anywhere in memory, the text and data segments must be less
13044 than 2GB in size and the data segment must be located within 2GB of the
13047 @item -mcmodel=embmedany
13048 @opindex mcmodel=embmedany
13049 Generate code for the Medium/Anywhere code model for embedded systems:
13050 64-bit addresses, the text and data segments must be less than 2GB in
13051 size, both starting anywhere in memory (determined at link time). The
13052 global register %g4 points to the base of the data segment. Programs
13053 are statically linked and PIC is not supported.
13056 @itemx -mno-stack-bias
13057 @opindex mstack-bias
13058 @opindex mno-stack-bias
13059 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
13060 frame pointer if present, are offset by @minus{}2047 which must be added back
13061 when making stack frame references. This is the default in 64-bit mode.
13062 Otherwise, assume no such offset is present.
13065 These switches are supported in addition to the above on Solaris:
13070 Add support for multithreading using the Solaris threads library. This
13071 option sets flags for both the preprocessor and linker. This option does
13072 not affect the thread safety of object code produced by the compiler or
13073 that of libraries supplied with it.
13077 Add support for multithreading using the POSIX threads library. This
13078 option sets flags for both the preprocessor and linker. This option does
13079 not affect the thread safety of object code produced by the compiler or
13080 that of libraries supplied with it.
13084 This is a synonym for @option{-pthreads}.
13088 @subsection SPU Options
13089 @cindex SPU options
13091 These @samp{-m} options are supported on the SPU:
13095 @itemx -merror-reloc
13096 @opindex mwarn-reloc
13097 @opindex merror-reloc
13099 The loader for SPU does not handle dynamic relocations. By default, GCC
13100 will give an error when it generates code that requires a dynamic
13101 relocation. @option{-mno-error-reloc} disables the error,
13102 @option{-mwarn-reloc} will generate a warning instead.
13105 @itemx -munsafe-dma
13107 @opindex munsafe-dma
13109 Instructions which initiate or test completion of DMA must not be
13110 reordered with respect to loads and stores of the memory which is being
13111 accessed. Users typically address this problem using the volatile
13112 keyword, but that can lead to inefficient code in places where the
13113 memory is known to not change. Rather than mark the memory as volatile
13114 we treat the DMA instructions as potentially effecting all memory. With
13115 @option{-munsafe-dma} users must use the volatile keyword to protect
13118 @item -mbranch-hints
13119 @opindex mbranch-hints
13121 By default, GCC will generate a branch hint instruction to avoid
13122 pipeline stalls for always taken or probably taken branches. A hint
13123 will not be generated closer than 8 instructions away from its branch.
13124 There is little reason to disable them, except for debugging purposes,
13125 or to make an object a little bit smaller.
13129 @opindex msmall-mem
13130 @opindex mlarge-mem
13132 By default, GCC generates code assuming that addresses are never larger
13133 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
13134 a full 32 bit address.
13139 By default, GCC links against startup code that assumes the SPU-style
13140 main function interface (which has an unconventional parameter list).
13141 With @option{-mstdmain}, GCC will link your program against startup
13142 code that assumes a C99-style interface to @code{main}, including a
13143 local copy of @code{argv} strings.
13145 @item -mfixed-range=@var{register-range}
13146 @opindex mfixed-range
13147 Generate code treating the given register range as fixed registers.
13148 A fixed register is one that the register allocator can not use. This is
13149 useful when compiling kernel code. A register range is specified as
13150 two registers separated by a dash. Multiple register ranges can be
13151 specified separated by a comma.
13155 @node System V Options
13156 @subsection Options for System V
13158 These additional options are available on System V Release 4 for
13159 compatibility with other compilers on those systems:
13164 Create a shared object.
13165 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
13169 Identify the versions of each tool used by the compiler, in a
13170 @code{.ident} assembler directive in the output.
13174 Refrain from adding @code{.ident} directives to the output file (this is
13177 @item -YP,@var{dirs}
13179 Search the directories @var{dirs}, and no others, for libraries
13180 specified with @option{-l}.
13182 @item -Ym,@var{dir}
13184 Look in the directory @var{dir} to find the M4 preprocessor.
13185 The assembler uses this option.
13186 @c This is supposed to go with a -Yd for predefined M4 macro files, but
13187 @c the generic assembler that comes with Solaris takes just -Ym.
13190 @node TMS320C3x/C4x Options
13191 @subsection TMS320C3x/C4x Options
13192 @cindex TMS320C3x/C4x Options
13194 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
13198 @item -mcpu=@var{cpu_type}
13200 Set the instruction set, register set, and instruction scheduling
13201 parameters for machine type @var{cpu_type}. Supported values for
13202 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
13203 @samp{c44}. The default is @samp{c40} to generate code for the
13208 @itemx -msmall-memory
13210 @opindex mbig-memory
13212 @opindex msmall-memory
13214 Generates code for the big or small memory model. The small memory
13215 model assumed that all data fits into one 64K word page. At run-time
13216 the data page (DP) register must be set to point to the 64K page
13217 containing the .bss and .data program sections. The big memory model is
13218 the default and requires reloading of the DP register for every direct
13225 Allow (disallow) allocation of general integer operands into the block
13226 count register BK@.
13232 Enable (disable) generation of code using decrement and branch,
13233 DBcond(D), instructions. This is enabled by default for the C4x. To be
13234 on the safe side, this is disabled for the C3x, since the maximum
13235 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
13236 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
13237 that it can utilize the decrement and branch instruction, but will give
13238 up if there is more than one memory reference in the loop. Thus a loop
13239 where the loop counter is decremented can generate slightly more
13240 efficient code, in cases where the RPTB instruction cannot be utilized.
13242 @item -mdp-isr-reload
13244 @opindex mdp-isr-reload
13246 Force the DP register to be saved on entry to an interrupt service
13247 routine (ISR), reloaded to point to the data section, and restored on
13248 exit from the ISR@. This should not be required unless someone has
13249 violated the small memory model by modifying the DP register, say within
13256 For the C3x use the 24-bit MPYI instruction for integer multiplies
13257 instead of a library call to guarantee 32-bit results. Note that if one
13258 of the operands is a constant, then the multiplication will be performed
13259 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
13260 then squaring operations are performed inline instead of a library call.
13263 @itemx -mno-fast-fix
13265 @opindex mno-fast-fix
13266 The C3x/C4x FIX instruction to convert a floating point value to an
13267 integer value chooses the nearest integer less than or equal to the
13268 floating point value rather than to the nearest integer. Thus if the
13269 floating point number is negative, the result will be incorrectly
13270 truncated an additional code is necessary to detect and correct this
13271 case. This option can be used to disable generation of the additional
13272 code required to correct the result.
13278 Enable (disable) generation of repeat block sequences using the RPTB
13279 instruction for zero overhead looping. The RPTB construct is only used
13280 for innermost loops that do not call functions or jump across the loop
13281 boundaries. There is no advantage having nested RPTB loops due to the
13282 overhead required to save and restore the RC, RS, and RE registers.
13283 This is enabled by default with @option{-O2}.
13285 @item -mrpts=@var{count}
13289 Enable (disable) the use of the single instruction repeat instruction
13290 RPTS@. If a repeat block contains a single instruction, and the loop
13291 count can be guaranteed to be less than the value @var{count}, GCC will
13292 emit a RPTS instruction instead of a RPTB@. If no value is specified,
13293 then a RPTS will be emitted even if the loop count cannot be determined
13294 at compile time. Note that the repeated instruction following RPTS does
13295 not have to be reloaded from memory each iteration, thus freeing up the
13296 CPU buses for operands. However, since interrupts are blocked by this
13297 instruction, it is disabled by default.
13299 @item -mloop-unsigned
13300 @itemx -mno-loop-unsigned
13301 @opindex mloop-unsigned
13302 @opindex mno-loop-unsigned
13303 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
13304 is @math{2^{31} + 1} since these instructions test if the iteration count is
13305 negative to terminate the loop. If the iteration count is unsigned
13306 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
13307 exceeded. This switch allows an unsigned iteration count.
13311 Try to emit an assembler syntax that the TI assembler (asm30) is happy
13312 with. This also enforces compatibility with the API employed by the TI
13313 C3x C compiler. For example, long doubles are passed as structures
13314 rather than in floating point registers.
13320 Generate code that uses registers (stack) for passing arguments to functions.
13321 By default, arguments are passed in registers where possible rather
13322 than by pushing arguments on to the stack.
13324 @item -mparallel-insns
13325 @itemx -mno-parallel-insns
13326 @opindex mparallel-insns
13327 @opindex mno-parallel-insns
13328 Allow the generation of parallel instructions. This is enabled by
13329 default with @option{-O2}.
13331 @item -mparallel-mpy
13332 @itemx -mno-parallel-mpy
13333 @opindex mparallel-mpy
13334 @opindex mno-parallel-mpy
13335 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
13336 provided @option{-mparallel-insns} is also specified. These instructions have
13337 tight register constraints which can pessimize the code generation
13338 of large functions.
13343 @subsection V850 Options
13344 @cindex V850 Options
13346 These @samp{-m} options are defined for V850 implementations:
13350 @itemx -mno-long-calls
13351 @opindex mlong-calls
13352 @opindex mno-long-calls
13353 Treat all calls as being far away (near). If calls are assumed to be
13354 far away, the compiler will always load the functions address up into a
13355 register, and call indirect through the pointer.
13361 Do not optimize (do optimize) basic blocks that use the same index
13362 pointer 4 or more times to copy pointer into the @code{ep} register, and
13363 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
13364 option is on by default if you optimize.
13366 @item -mno-prolog-function
13367 @itemx -mprolog-function
13368 @opindex mno-prolog-function
13369 @opindex mprolog-function
13370 Do not use (do use) external functions to save and restore registers
13371 at the prologue and epilogue of a function. The external functions
13372 are slower, but use less code space if more than one function saves
13373 the same number of registers. The @option{-mprolog-function} option
13374 is on by default if you optimize.
13378 Try to make the code as small as possible. At present, this just turns
13379 on the @option{-mep} and @option{-mprolog-function} options.
13381 @item -mtda=@var{n}
13383 Put static or global variables whose size is @var{n} bytes or less into
13384 the tiny data area that register @code{ep} points to. The tiny data
13385 area can hold up to 256 bytes in total (128 bytes for byte references).
13387 @item -msda=@var{n}
13389 Put static or global variables whose size is @var{n} bytes or less into
13390 the small data area that register @code{gp} points to. The small data
13391 area can hold up to 64 kilobytes.
13393 @item -mzda=@var{n}
13395 Put static or global variables whose size is @var{n} bytes or less into
13396 the first 32 kilobytes of memory.
13400 Specify that the target processor is the V850.
13403 @opindex mbig-switch
13404 Generate code suitable for big switch tables. Use this option only if
13405 the assembler/linker complain about out of range branches within a switch
13410 This option will cause r2 and r5 to be used in the code generated by
13411 the compiler. This setting is the default.
13413 @item -mno-app-regs
13414 @opindex mno-app-regs
13415 This option will cause r2 and r5 to be treated as fixed registers.
13419 Specify that the target processor is the V850E1. The preprocessor
13420 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
13421 this option is used.
13425 Specify that the target processor is the V850E@. The preprocessor
13426 constant @samp{__v850e__} will be defined if this option is used.
13428 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
13429 are defined then a default target processor will be chosen and the
13430 relevant @samp{__v850*__} preprocessor constant will be defined.
13432 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
13433 defined, regardless of which processor variant is the target.
13435 @item -mdisable-callt
13436 @opindex mdisable-callt
13437 This option will suppress generation of the CALLT instruction for the
13438 v850e and v850e1 flavors of the v850 architecture. The default is
13439 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
13444 @subsection VAX Options
13445 @cindex VAX options
13447 These @samp{-m} options are defined for the VAX:
13452 Do not output certain jump instructions (@code{aobleq} and so on)
13453 that the Unix assembler for the VAX cannot handle across long
13458 Do output those jump instructions, on the assumption that you
13459 will assemble with the GNU assembler.
13463 Output code for g-format floating point numbers instead of d-format.
13466 @node x86-64 Options
13467 @subsection x86-64 Options
13468 @cindex x86-64 options
13470 These are listed under @xref{i386 and x86-64 Options}.
13472 @node Xstormy16 Options
13473 @subsection Xstormy16 Options
13474 @cindex Xstormy16 Options
13476 These options are defined for Xstormy16:
13481 Choose startup files and linker script suitable for the simulator.
13484 @node Xtensa Options
13485 @subsection Xtensa Options
13486 @cindex Xtensa Options
13488 These options are supported for Xtensa targets:
13492 @itemx -mno-const16
13494 @opindex mno-const16
13495 Enable or disable use of @code{CONST16} instructions for loading
13496 constant values. The @code{CONST16} instruction is currently not a
13497 standard option from Tensilica. When enabled, @code{CONST16}
13498 instructions are always used in place of the standard @code{L32R}
13499 instructions. The use of @code{CONST16} is enabled by default only if
13500 the @code{L32R} instruction is not available.
13503 @itemx -mno-fused-madd
13504 @opindex mfused-madd
13505 @opindex mno-fused-madd
13506 Enable or disable use of fused multiply/add and multiply/subtract
13507 instructions in the floating-point option. This has no effect if the
13508 floating-point option is not also enabled. Disabling fused multiply/add
13509 and multiply/subtract instructions forces the compiler to use separate
13510 instructions for the multiply and add/subtract operations. This may be
13511 desirable in some cases where strict IEEE 754-compliant results are
13512 required: the fused multiply add/subtract instructions do not round the
13513 intermediate result, thereby producing results with @emph{more} bits of
13514 precision than specified by the IEEE standard. Disabling fused multiply
13515 add/subtract instructions also ensures that the program output is not
13516 sensitive to the compiler's ability to combine multiply and add/subtract
13519 @item -mtext-section-literals
13520 @itemx -mno-text-section-literals
13521 @opindex mtext-section-literals
13522 @opindex mno-text-section-literals
13523 Control the treatment of literal pools. The default is
13524 @option{-mno-text-section-literals}, which places literals in a separate
13525 section in the output file. This allows the literal pool to be placed
13526 in a data RAM/ROM, and it also allows the linker to combine literal
13527 pools from separate object files to remove redundant literals and
13528 improve code size. With @option{-mtext-section-literals}, the literals
13529 are interspersed in the text section in order to keep them as close as
13530 possible to their references. This may be necessary for large assembly
13533 @item -mtarget-align
13534 @itemx -mno-target-align
13535 @opindex mtarget-align
13536 @opindex mno-target-align
13537 When this option is enabled, GCC instructs the assembler to
13538 automatically align instructions to reduce branch penalties at the
13539 expense of some code density. The assembler attempts to widen density
13540 instructions to align branch targets and the instructions following call
13541 instructions. If there are not enough preceding safe density
13542 instructions to align a target, no widening will be performed. The
13543 default is @option{-mtarget-align}. These options do not affect the
13544 treatment of auto-aligned instructions like @code{LOOP}, which the
13545 assembler will always align, either by widening density instructions or
13546 by inserting no-op instructions.
13549 @itemx -mno-longcalls
13550 @opindex mlongcalls
13551 @opindex mno-longcalls
13552 When this option is enabled, GCC instructs the assembler to translate
13553 direct calls to indirect calls unless it can determine that the target
13554 of a direct call is in the range allowed by the call instruction. This
13555 translation typically occurs for calls to functions in other source
13556 files. Specifically, the assembler translates a direct @code{CALL}
13557 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
13558 The default is @option{-mno-longcalls}. This option should be used in
13559 programs where the call target can potentially be out of range. This
13560 option is implemented in the assembler, not the compiler, so the
13561 assembly code generated by GCC will still show direct call
13562 instructions---look at the disassembled object code to see the actual
13563 instructions. Note that the assembler will use an indirect call for
13564 every cross-file call, not just those that really will be out of range.
13567 @node zSeries Options
13568 @subsection zSeries Options
13569 @cindex zSeries options
13571 These are listed under @xref{S/390 and zSeries Options}.
13573 @node Code Gen Options
13574 @section Options for Code Generation Conventions
13575 @cindex code generation conventions
13576 @cindex options, code generation
13577 @cindex run-time options
13579 These machine-independent options control the interface conventions
13580 used in code generation.
13582 Most of them have both positive and negative forms; the negative form
13583 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
13584 one of the forms is listed---the one which is not the default. You
13585 can figure out the other form by either removing @samp{no-} or adding
13589 @item -fbounds-check
13590 @opindex fbounds-check
13591 For front-ends that support it, generate additional code to check that
13592 indices used to access arrays are within the declared range. This is
13593 currently only supported by the Java and Fortran front-ends, where
13594 this option defaults to true and false respectively.
13598 This option generates traps for signed overflow on addition, subtraction,
13599 multiplication operations.
13603 This option instructs the compiler to assume that signed arithmetic
13604 overflow of addition, subtraction and multiplication wraps around
13605 using twos-complement representation. This flag enables some optimizations
13606 and disables others. This option is enabled by default for the Java
13607 front-end, as required by the Java language specification.
13610 @opindex fexceptions
13611 Enable exception handling. Generates extra code needed to propagate
13612 exceptions. For some targets, this implies GCC will generate frame
13613 unwind information for all functions, which can produce significant data
13614 size overhead, although it does not affect execution. If you do not
13615 specify this option, GCC will enable it by default for languages like
13616 C++ which normally require exception handling, and disable it for
13617 languages like C that do not normally require it. However, you may need
13618 to enable this option when compiling C code that needs to interoperate
13619 properly with exception handlers written in C++. You may also wish to
13620 disable this option if you are compiling older C++ programs that don't
13621 use exception handling.
13623 @item -fnon-call-exceptions
13624 @opindex fnon-call-exceptions
13625 Generate code that allows trapping instructions to throw exceptions.
13626 Note that this requires platform-specific runtime support that does
13627 not exist everywhere. Moreover, it only allows @emph{trapping}
13628 instructions to throw exceptions, i.e.@: memory references or floating
13629 point instructions. It does not allow exceptions to be thrown from
13630 arbitrary signal handlers such as @code{SIGALRM}.
13632 @item -funwind-tables
13633 @opindex funwind-tables
13634 Similar to @option{-fexceptions}, except that it will just generate any needed
13635 static data, but will not affect the generated code in any other way.
13636 You will normally not enable this option; instead, a language processor
13637 that needs this handling would enable it on your behalf.
13639 @item -fasynchronous-unwind-tables
13640 @opindex fasynchronous-unwind-tables
13641 Generate unwind table in dwarf2 format, if supported by target machine. The
13642 table is exact at each instruction boundary, so it can be used for stack
13643 unwinding from asynchronous events (such as debugger or garbage collector).
13645 @item -fpcc-struct-return
13646 @opindex fpcc-struct-return
13647 Return ``short'' @code{struct} and @code{union} values in memory like
13648 longer ones, rather than in registers. This convention is less
13649 efficient, but it has the advantage of allowing intercallability between
13650 GCC-compiled files and files compiled with other compilers, particularly
13651 the Portable C Compiler (pcc).
13653 The precise convention for returning structures in memory depends
13654 on the target configuration macros.
13656 Short structures and unions are those whose size and alignment match
13657 that of some integer type.
13659 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
13660 switch is not binary compatible with code compiled with the
13661 @option{-freg-struct-return} switch.
13662 Use it to conform to a non-default application binary interface.
13664 @item -freg-struct-return
13665 @opindex freg-struct-return
13666 Return @code{struct} and @code{union} values in registers when possible.
13667 This is more efficient for small structures than
13668 @option{-fpcc-struct-return}.
13670 If you specify neither @option{-fpcc-struct-return} nor
13671 @option{-freg-struct-return}, GCC defaults to whichever convention is
13672 standard for the target. If there is no standard convention, GCC
13673 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
13674 the principal compiler. In those cases, we can choose the standard, and
13675 we chose the more efficient register return alternative.
13677 @strong{Warning:} code compiled with the @option{-freg-struct-return}
13678 switch is not binary compatible with code compiled with the
13679 @option{-fpcc-struct-return} switch.
13680 Use it to conform to a non-default application binary interface.
13682 @item -fshort-enums
13683 @opindex fshort-enums
13684 Allocate to an @code{enum} type only as many bytes as it needs for the
13685 declared range of possible values. Specifically, the @code{enum} type
13686 will be equivalent to the smallest integer type which has enough room.
13688 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
13689 code that is not binary compatible with code generated without that switch.
13690 Use it to conform to a non-default application binary interface.
13692 @item -fshort-double
13693 @opindex fshort-double
13694 Use the same size for @code{double} as for @code{float}.
13696 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
13697 code that is not binary compatible with code generated without that switch.
13698 Use it to conform to a non-default application binary interface.
13700 @item -fshort-wchar
13701 @opindex fshort-wchar
13702 Override the underlying type for @samp{wchar_t} to be @samp{short
13703 unsigned int} instead of the default for the target. This option is
13704 useful for building programs to run under WINE@.
13706 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
13707 code that is not binary compatible with code generated without that switch.
13708 Use it to conform to a non-default application binary interface.
13711 @opindex fno-common
13712 In C, allocate even uninitialized global variables in the data section of the
13713 object file, rather than generating them as common blocks. This has the
13714 effect that if the same variable is declared (without @code{extern}) in
13715 two different compilations, you will get an error when you link them.
13716 The only reason this might be useful is if you wish to verify that the
13717 program will work on other systems which always work this way.
13721 Ignore the @samp{#ident} directive.
13723 @item -finhibit-size-directive
13724 @opindex finhibit-size-directive
13725 Don't output a @code{.size} assembler directive, or anything else that
13726 would cause trouble if the function is split in the middle, and the
13727 two halves are placed at locations far apart in memory. This option is
13728 used when compiling @file{crtstuff.c}; you should not need to use it
13731 @item -fverbose-asm
13732 @opindex fverbose-asm
13733 Put extra commentary information in the generated assembly code to
13734 make it more readable. This option is generally only of use to those
13735 who actually need to read the generated assembly code (perhaps while
13736 debugging the compiler itself).
13738 @option{-fno-verbose-asm}, the default, causes the
13739 extra information to be omitted and is useful when comparing two assembler
13742 @item -frecord-gcc-switches
13743 @opindex frecord-gcc-switches
13744 This switch causes the command line that was used to invoke the
13745 compiler to be recorded into the object file that is being created.
13746 This switch is only implemented on some targets and the exact format
13747 of the recording is target and binary file format dependent, but it
13748 usually takes the form of a section containing ASCII text. This
13749 switch is related to the @option{-fverbose-asm} switch, but that
13750 switch only records information in the assembler output file as
13751 comments, so it never reaches the object file.
13755 @cindex global offset table
13757 Generate position-independent code (PIC) suitable for use in a shared
13758 library, if supported for the target machine. Such code accesses all
13759 constant addresses through a global offset table (GOT)@. The dynamic
13760 loader resolves the GOT entries when the program starts (the dynamic
13761 loader is not part of GCC; it is part of the operating system). If
13762 the GOT size for the linked executable exceeds a machine-specific
13763 maximum size, you get an error message from the linker indicating that
13764 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
13765 instead. (These maximums are 8k on the SPARC and 32k
13766 on the m68k and RS/6000. The 386 has no such limit.)
13768 Position-independent code requires special support, and therefore works
13769 only on certain machines. For the 386, GCC supports PIC for System V
13770 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
13771 position-independent.
13773 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13778 If supported for the target machine, emit position-independent code,
13779 suitable for dynamic linking and avoiding any limit on the size of the
13780 global offset table. This option makes a difference on the m68k,
13781 PowerPC and SPARC@.
13783 Position-independent code requires special support, and therefore works
13784 only on certain machines.
13786 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13793 These options are similar to @option{-fpic} and @option{-fPIC}, but
13794 generated position independent code can be only linked into executables.
13795 Usually these options are used when @option{-pie} GCC option will be
13796 used during linking.
13798 @item -fno-jump-tables
13799 @opindex fno-jump-tables
13800 Do not use jump tables for switch statements even where it would be
13801 more efficient than other code generation strategies. This option is
13802 of use in conjunction with @option{-fpic} or @option{-fPIC} for
13803 building code which forms part of a dynamic linker and cannot
13804 reference the address of a jump table. On some targets, jump tables
13805 do not require a GOT and this option is not needed.
13807 @item -ffixed-@var{reg}
13809 Treat the register named @var{reg} as a fixed register; generated code
13810 should never refer to it (except perhaps as a stack pointer, frame
13811 pointer or in some other fixed role).
13813 @var{reg} must be the name of a register. The register names accepted
13814 are machine-specific and are defined in the @code{REGISTER_NAMES}
13815 macro in the machine description macro file.
13817 This flag does not have a negative form, because it specifies a
13820 @item -fcall-used-@var{reg}
13821 @opindex fcall-used
13822 Treat the register named @var{reg} as an allocable register that is
13823 clobbered by function calls. It may be allocated for temporaries or
13824 variables that do not live across a call. Functions compiled this way
13825 will not save and restore the register @var{reg}.
13827 It is an error to used this flag with the frame pointer or stack pointer.
13828 Use of this flag for other registers that have fixed pervasive roles in
13829 the machine's execution model will produce disastrous results.
13831 This flag does not have a negative form, because it specifies a
13834 @item -fcall-saved-@var{reg}
13835 @opindex fcall-saved
13836 Treat the register named @var{reg} as an allocable register saved by
13837 functions. It may be allocated even for temporaries or variables that
13838 live across a call. Functions compiled this way will save and restore
13839 the register @var{reg} if they use it.
13841 It is an error to used this flag with the frame pointer or stack pointer.
13842 Use of this flag for other registers that have fixed pervasive roles in
13843 the machine's execution model will produce disastrous results.
13845 A different sort of disaster will result from the use of this flag for
13846 a register in which function values may be returned.
13848 This flag does not have a negative form, because it specifies a
13851 @item -fpack-struct[=@var{n}]
13852 @opindex fpack-struct
13853 Without a value specified, pack all structure members together without
13854 holes. When a value is specified (which must be a small power of two), pack
13855 structure members according to this value, representing the maximum
13856 alignment (that is, objects with default alignment requirements larger than
13857 this will be output potentially unaligned at the next fitting location.
13859 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
13860 code that is not binary compatible with code generated without that switch.
13861 Additionally, it makes the code suboptimal.
13862 Use it to conform to a non-default application binary interface.
13864 @item -finstrument-functions
13865 @opindex finstrument-functions
13866 Generate instrumentation calls for entry and exit to functions. Just
13867 after function entry and just before function exit, the following
13868 profiling functions will be called with the address of the current
13869 function and its call site. (On some platforms,
13870 @code{__builtin_return_address} does not work beyond the current
13871 function, so the call site information may not be available to the
13872 profiling functions otherwise.)
13875 void __cyg_profile_func_enter (void *this_fn,
13877 void __cyg_profile_func_exit (void *this_fn,
13881 The first argument is the address of the start of the current function,
13882 which may be looked up exactly in the symbol table.
13884 This instrumentation is also done for functions expanded inline in other
13885 functions. The profiling calls will indicate where, conceptually, the
13886 inline function is entered and exited. This means that addressable
13887 versions of such functions must be available. If all your uses of a
13888 function are expanded inline, this may mean an additional expansion of
13889 code size. If you use @samp{extern inline} in your C code, an
13890 addressable version of such functions must be provided. (This is
13891 normally the case anyways, but if you get lucky and the optimizer always
13892 expands the functions inline, you might have gotten away without
13893 providing static copies.)
13895 A function may be given the attribute @code{no_instrument_function}, in
13896 which case this instrumentation will not be done. This can be used, for
13897 example, for the profiling functions listed above, high-priority
13898 interrupt routines, and any functions from which the profiling functions
13899 cannot safely be called (perhaps signal handlers, if the profiling
13900 routines generate output or allocate memory).
13902 @item -fstack-check
13903 @opindex fstack-check
13904 Generate code to verify that you do not go beyond the boundary of the
13905 stack. You should specify this flag if you are running in an
13906 environment with multiple threads, but only rarely need to specify it in
13907 a single-threaded environment since stack overflow is automatically
13908 detected on nearly all systems if there is only one stack.
13910 Note that this switch does not actually cause checking to be done; the
13911 operating system must do that. The switch causes generation of code
13912 to ensure that the operating system sees the stack being extended.
13914 @item -fstack-limit-register=@var{reg}
13915 @itemx -fstack-limit-symbol=@var{sym}
13916 @itemx -fno-stack-limit
13917 @opindex fstack-limit-register
13918 @opindex fstack-limit-symbol
13919 @opindex fno-stack-limit
13920 Generate code to ensure that the stack does not grow beyond a certain value,
13921 either the value of a register or the address of a symbol. If the stack
13922 would grow beyond the value, a signal is raised. For most targets,
13923 the signal is raised before the stack overruns the boundary, so
13924 it is possible to catch the signal without taking special precautions.
13926 For instance, if the stack starts at absolute address @samp{0x80000000}
13927 and grows downwards, you can use the flags
13928 @option{-fstack-limit-symbol=__stack_limit} and
13929 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
13930 of 128KB@. Note that this may only work with the GNU linker.
13932 @cindex aliasing of parameters
13933 @cindex parameters, aliased
13934 @item -fargument-alias
13935 @itemx -fargument-noalias
13936 @itemx -fargument-noalias-global
13937 @itemx -fargument-noalias-anything
13938 @opindex fargument-alias
13939 @opindex fargument-noalias
13940 @opindex fargument-noalias-global
13941 @opindex fargument-noalias-anything
13942 Specify the possible relationships among parameters and between
13943 parameters and global data.
13945 @option{-fargument-alias} specifies that arguments (parameters) may
13946 alias each other and may alias global storage.@*
13947 @option{-fargument-noalias} specifies that arguments do not alias
13948 each other, but may alias global storage.@*
13949 @option{-fargument-noalias-global} specifies that arguments do not
13950 alias each other and do not alias global storage.
13951 @option{-fargument-noalias-anything} specifies that arguments do not
13952 alias any other storage.
13954 Each language will automatically use whatever option is required by
13955 the language standard. You should not need to use these options yourself.
13957 @item -fleading-underscore
13958 @opindex fleading-underscore
13959 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
13960 change the way C symbols are represented in the object file. One use
13961 is to help link with legacy assembly code.
13963 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
13964 generate code that is not binary compatible with code generated without that
13965 switch. Use it to conform to a non-default application binary interface.
13966 Not all targets provide complete support for this switch.
13968 @item -ftls-model=@var{model}
13969 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
13970 The @var{model} argument should be one of @code{global-dynamic},
13971 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
13973 The default without @option{-fpic} is @code{initial-exec}; with
13974 @option{-fpic} the default is @code{global-dynamic}.
13976 @item -fvisibility=@var{default|internal|hidden|protected}
13977 @opindex fvisibility
13978 Set the default ELF image symbol visibility to the specified option---all
13979 symbols will be marked with this unless overridden within the code.
13980 Using this feature can very substantially improve linking and
13981 load times of shared object libraries, produce more optimized
13982 code, provide near-perfect API export and prevent symbol clashes.
13983 It is @strong{strongly} recommended that you use this in any shared objects
13986 Despite the nomenclature, @code{default} always means public ie;
13987 available to be linked against from outside the shared object.
13988 @code{protected} and @code{internal} are pretty useless in real-world
13989 usage so the only other commonly used option will be @code{hidden}.
13990 The default if @option{-fvisibility} isn't specified is
13991 @code{default}, i.e., make every
13992 symbol public---this causes the same behavior as previous versions of
13995 A good explanation of the benefits offered by ensuring ELF
13996 symbols have the correct visibility is given by ``How To Write
13997 Shared Libraries'' by Ulrich Drepper (which can be found at
13998 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
13999 solution made possible by this option to marking things hidden when
14000 the default is public is to make the default hidden and mark things
14001 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
14002 and @code{__attribute__ ((visibility("default")))} instead of
14003 @code{__declspec(dllexport)} you get almost identical semantics with
14004 identical syntax. This is a great boon to those working with
14005 cross-platform projects.
14007 For those adding visibility support to existing code, you may find
14008 @samp{#pragma GCC visibility} of use. This works by you enclosing
14009 the declarations you wish to set visibility for with (for example)
14010 @samp{#pragma GCC visibility push(hidden)} and
14011 @samp{#pragma GCC visibility pop}.
14012 Bear in mind that symbol visibility should be viewed @strong{as
14013 part of the API interface contract} and thus all new code should
14014 always specify visibility when it is not the default ie; declarations
14015 only for use within the local DSO should @strong{always} be marked explicitly
14016 as hidden as so to avoid PLT indirection overheads---making this
14017 abundantly clear also aids readability and self-documentation of the code.
14018 Note that due to ISO C++ specification requirements, operator new and
14019 operator delete must always be of default visibility.
14021 Be aware that headers from outside your project, in particular system
14022 headers and headers from any other library you use, may not be
14023 expecting to be compiled with visibility other than the default. You
14024 may need to explicitly say @samp{#pragma GCC visibility push(default)}
14025 before including any such headers.
14027 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
14028 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
14029 no modifications. However, this means that calls to @samp{extern}
14030 functions with no explicit visibility will use the PLT, so it is more
14031 effective to use @samp{__attribute ((visibility))} and/or
14032 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
14033 declarations should be treated as hidden.
14035 Note that @samp{-fvisibility} does affect C++ vague linkage
14036 entities. This means that, for instance, an exception class that will
14037 be thrown between DSOs must be explicitly marked with default
14038 visibility so that the @samp{type_info} nodes will be unified between
14041 An overview of these techniques, their benefits and how to use them
14042 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
14048 @node Environment Variables
14049 @section Environment Variables Affecting GCC
14050 @cindex environment variables
14052 @c man begin ENVIRONMENT
14053 This section describes several environment variables that affect how GCC
14054 operates. Some of them work by specifying directories or prefixes to use
14055 when searching for various kinds of files. Some are used to specify other
14056 aspects of the compilation environment.
14058 Note that you can also specify places to search using options such as
14059 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
14060 take precedence over places specified using environment variables, which
14061 in turn take precedence over those specified by the configuration of GCC@.
14062 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
14063 GNU Compiler Collection (GCC) Internals}.
14068 @c @itemx LC_COLLATE
14070 @c @itemx LC_MONETARY
14071 @c @itemx LC_NUMERIC
14076 @c @findex LC_COLLATE
14077 @findex LC_MESSAGES
14078 @c @findex LC_MONETARY
14079 @c @findex LC_NUMERIC
14083 These environment variables control the way that GCC uses
14084 localization information that allow GCC to work with different
14085 national conventions. GCC inspects the locale categories
14086 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
14087 so. These locale categories can be set to any value supported by your
14088 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
14089 Kingdom encoded in UTF-8.
14091 The @env{LC_CTYPE} environment variable specifies character
14092 classification. GCC uses it to determine the character boundaries in
14093 a string; this is needed for some multibyte encodings that contain quote
14094 and escape characters that would otherwise be interpreted as a string
14097 The @env{LC_MESSAGES} environment variable specifies the language to
14098 use in diagnostic messages.
14100 If the @env{LC_ALL} environment variable is set, it overrides the value
14101 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
14102 and @env{LC_MESSAGES} default to the value of the @env{LANG}
14103 environment variable. If none of these variables are set, GCC
14104 defaults to traditional C English behavior.
14108 If @env{TMPDIR} is set, it specifies the directory to use for temporary
14109 files. GCC uses temporary files to hold the output of one stage of
14110 compilation which is to be used as input to the next stage: for example,
14111 the output of the preprocessor, which is the input to the compiler
14114 @item GCC_EXEC_PREFIX
14115 @findex GCC_EXEC_PREFIX
14116 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
14117 names of the subprograms executed by the compiler. No slash is added
14118 when this prefix is combined with the name of a subprogram, but you can
14119 specify a prefix that ends with a slash if you wish.
14121 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
14122 an appropriate prefix to use based on the pathname it was invoked with.
14124 If GCC cannot find the subprogram using the specified prefix, it
14125 tries looking in the usual places for the subprogram.
14127 The default value of @env{GCC_EXEC_PREFIX} is
14128 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
14129 the installed compiler. In many cases @var{prefix} is the value
14130 of @code{prefix} when you ran the @file{configure} script.
14132 Other prefixes specified with @option{-B} take precedence over this prefix.
14134 This prefix is also used for finding files such as @file{crt0.o} that are
14137 In addition, the prefix is used in an unusual way in finding the
14138 directories to search for header files. For each of the standard
14139 directories whose name normally begins with @samp{/usr/local/lib/gcc}
14140 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
14141 replacing that beginning with the specified prefix to produce an
14142 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
14143 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
14144 These alternate directories are searched first; the standard directories
14145 come next. If a standard directory begins with the configured
14146 @var{prefix} then the value of @var{prefix} is replaced by
14147 @env{GCC_EXEC_PREFIX} when looking for header files.
14149 @item COMPILER_PATH
14150 @findex COMPILER_PATH
14151 The value of @env{COMPILER_PATH} is a colon-separated list of
14152 directories, much like @env{PATH}. GCC tries the directories thus
14153 specified when searching for subprograms, if it can't find the
14154 subprograms using @env{GCC_EXEC_PREFIX}.
14157 @findex LIBRARY_PATH
14158 The value of @env{LIBRARY_PATH} is a colon-separated list of
14159 directories, much like @env{PATH}. When configured as a native compiler,
14160 GCC tries the directories thus specified when searching for special
14161 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
14162 using GCC also uses these directories when searching for ordinary
14163 libraries for the @option{-l} option (but directories specified with
14164 @option{-L} come first).
14168 @cindex locale definition
14169 This variable is used to pass locale information to the compiler. One way in
14170 which this information is used is to determine the character set to be used
14171 when character literals, string literals and comments are parsed in C and C++.
14172 When the compiler is configured to allow multibyte characters,
14173 the following values for @env{LANG} are recognized:
14177 Recognize JIS characters.
14179 Recognize SJIS characters.
14181 Recognize EUCJP characters.
14184 If @env{LANG} is not defined, or if it has some other value, then the
14185 compiler will use mblen and mbtowc as defined by the default locale to
14186 recognize and translate multibyte characters.
14190 Some additional environments variables affect the behavior of the
14193 @include cppenv.texi
14197 @node Precompiled Headers
14198 @section Using Precompiled Headers
14199 @cindex precompiled headers
14200 @cindex speed of compilation
14202 Often large projects have many header files that are included in every
14203 source file. The time the compiler takes to process these header files
14204 over and over again can account for nearly all of the time required to
14205 build the project. To make builds faster, GCC allows users to
14206 `precompile' a header file; then, if builds can use the precompiled
14207 header file they will be much faster.
14209 To create a precompiled header file, simply compile it as you would any
14210 other file, if necessary using the @option{-x} option to make the driver
14211 treat it as a C or C++ header file. You will probably want to use a
14212 tool like @command{make} to keep the precompiled header up-to-date when
14213 the headers it contains change.
14215 A precompiled header file will be searched for when @code{#include} is
14216 seen in the compilation. As it searches for the included file
14217 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
14218 compiler looks for a precompiled header in each directory just before it
14219 looks for the include file in that directory. The name searched for is
14220 the name specified in the @code{#include} with @samp{.gch} appended. If
14221 the precompiled header file can't be used, it is ignored.
14223 For instance, if you have @code{#include "all.h"}, and you have
14224 @file{all.h.gch} in the same directory as @file{all.h}, then the
14225 precompiled header file will be used if possible, and the original
14226 header will be used otherwise.
14228 Alternatively, you might decide to put the precompiled header file in a
14229 directory and use @option{-I} to ensure that directory is searched
14230 before (or instead of) the directory containing the original header.
14231 Then, if you want to check that the precompiled header file is always
14232 used, you can put a file of the same name as the original header in this
14233 directory containing an @code{#error} command.
14235 This also works with @option{-include}. So yet another way to use
14236 precompiled headers, good for projects not designed with precompiled
14237 header files in mind, is to simply take most of the header files used by
14238 a project, include them from another header file, precompile that header
14239 file, and @option{-include} the precompiled header. If the header files
14240 have guards against multiple inclusion, they will be skipped because
14241 they've already been included (in the precompiled header).
14243 If you need to precompile the same header file for different
14244 languages, targets, or compiler options, you can instead make a
14245 @emph{directory} named like @file{all.h.gch}, and put each precompiled
14246 header in the directory, perhaps using @option{-o}. It doesn't matter
14247 what you call the files in the directory, every precompiled header in
14248 the directory will be considered. The first precompiled header
14249 encountered in the directory that is valid for this compilation will
14250 be used; they're searched in no particular order.
14252 There are many other possibilities, limited only by your imagination,
14253 good sense, and the constraints of your build system.
14255 A precompiled header file can be used only when these conditions apply:
14259 Only one precompiled header can be used in a particular compilation.
14262 A precompiled header can't be used once the first C token is seen. You
14263 can have preprocessor directives before a precompiled header; you can
14264 even include a precompiled header from inside another header, so long as
14265 there are no C tokens before the @code{#include}.
14268 The precompiled header file must be produced for the same language as
14269 the current compilation. You can't use a C precompiled header for a C++
14273 The precompiled header file must have been produced by the same compiler
14274 binary as the current compilation is using.
14277 Any macros defined before the precompiled header is included must
14278 either be defined in the same way as when the precompiled header was
14279 generated, or must not affect the precompiled header, which usually
14280 means that they don't appear in the precompiled header at all.
14282 The @option{-D} option is one way to define a macro before a
14283 precompiled header is included; using a @code{#define} can also do it.
14284 There are also some options that define macros implicitly, like
14285 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
14288 @item If debugging information is output when using the precompiled
14289 header, using @option{-g} or similar, the same kind of debugging information
14290 must have been output when building the precompiled header. However,
14291 a precompiled header built using @option{-g} can be used in a compilation
14292 when no debugging information is being output.
14294 @item The same @option{-m} options must generally be used when building
14295 and using the precompiled header. @xref{Submodel Options},
14296 for any cases where this rule is relaxed.
14298 @item Each of the following options must be the same when building and using
14299 the precompiled header:
14301 @gccoptlist{-fexceptions -funit-at-a-time}
14304 Some other command-line options starting with @option{-f},
14305 @option{-p}, or @option{-O} must be defined in the same way as when
14306 the precompiled header was generated. At present, it's not clear
14307 which options are safe to change and which are not; the safest choice
14308 is to use exactly the same options when generating and using the
14309 precompiled header. The following are known to be safe:
14311 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
14312 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
14313 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
14318 For all of these except the last, the compiler will automatically
14319 ignore the precompiled header if the conditions aren't met. If you
14320 find an option combination that doesn't work and doesn't cause the
14321 precompiled header to be ignored, please consider filing a bug report,
14324 If you do use differing options when generating and using the
14325 precompiled header, the actual behavior will be a mixture of the
14326 behavior for the options. For instance, if you use @option{-g} to
14327 generate the precompiled header but not when using it, you may or may
14328 not get debugging information for routines in the precompiled header.
14330 @node Running Protoize
14331 @section Running Protoize
14333 The program @code{protoize} is an optional part of GCC@. You can use
14334 it to add prototypes to a program, thus converting the program to ISO
14335 C in one respect. The companion program @code{unprotoize} does the
14336 reverse: it removes argument types from any prototypes that are found.
14338 When you run these programs, you must specify a set of source files as
14339 command line arguments. The conversion programs start out by compiling
14340 these files to see what functions they define. The information gathered
14341 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
14343 After scanning comes actual conversion. The specified files are all
14344 eligible to be converted; any files they include (whether sources or
14345 just headers) are eligible as well.
14347 But not all the eligible files are converted. By default,
14348 @code{protoize} and @code{unprotoize} convert only source and header
14349 files in the current directory. You can specify additional directories
14350 whose files should be converted with the @option{-d @var{directory}}
14351 option. You can also specify particular files to exclude with the
14352 @option{-x @var{file}} option. A file is converted if it is eligible, its
14353 directory name matches one of the specified directory names, and its
14354 name within the directory has not been excluded.
14356 Basic conversion with @code{protoize} consists of rewriting most
14357 function definitions and function declarations to specify the types of
14358 the arguments. The only ones not rewritten are those for varargs
14361 @code{protoize} optionally inserts prototype declarations at the
14362 beginning of the source file, to make them available for any calls that
14363 precede the function's definition. Or it can insert prototype
14364 declarations with block scope in the blocks where undeclared functions
14367 Basic conversion with @code{unprotoize} consists of rewriting most
14368 function declarations to remove any argument types, and rewriting
14369 function definitions to the old-style pre-ISO form.
14371 Both conversion programs print a warning for any function declaration or
14372 definition that they can't convert. You can suppress these warnings
14375 The output from @code{protoize} or @code{unprotoize} replaces the
14376 original source file. The original file is renamed to a name ending
14377 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
14378 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
14379 for DOS) file already exists, then the source file is simply discarded.
14381 @code{protoize} and @code{unprotoize} both depend on GCC itself to
14382 scan the program and collect information about the functions it uses.
14383 So neither of these programs will work until GCC is installed.
14385 Here is a table of the options you can use with @code{protoize} and
14386 @code{unprotoize}. Each option works with both programs unless
14390 @item -B @var{directory}
14391 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
14392 usual directory (normally @file{/usr/local/lib}). This file contains
14393 prototype information about standard system functions. This option
14394 applies only to @code{protoize}.
14396 @item -c @var{compilation-options}
14397 Use @var{compilation-options} as the options when running @command{gcc} to
14398 produce the @samp{.X} files. The special option @option{-aux-info} is
14399 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
14401 Note that the compilation options must be given as a single argument to
14402 @code{protoize} or @code{unprotoize}. If you want to specify several
14403 @command{gcc} options, you must quote the entire set of compilation options
14404 to make them a single word in the shell.
14406 There are certain @command{gcc} arguments that you cannot use, because they
14407 would produce the wrong kind of output. These include @option{-g},
14408 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
14409 the @var{compilation-options}, they are ignored.
14412 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
14413 systems) instead of @samp{.c}. This is convenient if you are converting
14414 a C program to C++. This option applies only to @code{protoize}.
14417 Add explicit global declarations. This means inserting explicit
14418 declarations at the beginning of each source file for each function
14419 that is called in the file and was not declared. These declarations
14420 precede the first function definition that contains a call to an
14421 undeclared function. This option applies only to @code{protoize}.
14423 @item -i @var{string}
14424 Indent old-style parameter declarations with the string @var{string}.
14425 This option applies only to @code{protoize}.
14427 @code{unprotoize} converts prototyped function definitions to old-style
14428 function definitions, where the arguments are declared between the
14429 argument list and the initial @samp{@{}. By default, @code{unprotoize}
14430 uses five spaces as the indentation. If you want to indent with just
14431 one space instead, use @option{-i " "}.
14434 Keep the @samp{.X} files. Normally, they are deleted after conversion
14438 Add explicit local declarations. @code{protoize} with @option{-l} inserts
14439 a prototype declaration for each function in each block which calls the
14440 function without any declaration. This option applies only to
14444 Make no real changes. This mode just prints information about the conversions
14445 that would have been done without @option{-n}.
14448 Make no @samp{.save} files. The original files are simply deleted.
14449 Use this option with caution.
14451 @item -p @var{program}
14452 Use the program @var{program} as the compiler. Normally, the name
14453 @file{gcc} is used.
14456 Work quietly. Most warnings are suppressed.
14459 Print the version number, just like @option{-v} for @command{gcc}.
14462 If you need special compiler options to compile one of your program's
14463 source files, then you should generate that file's @samp{.X} file
14464 specially, by running @command{gcc} on that source file with the
14465 appropriate options and the option @option{-aux-info}. Then run
14466 @code{protoize} on the entire set of files. @code{protoize} will use
14467 the existing @samp{.X} file because it is newer than the source file.
14471 gcc -Dfoo=bar file1.c -aux-info file1.X
14476 You need to include the special files along with the rest in the
14477 @code{protoize} command, even though their @samp{.X} files already
14478 exist, because otherwise they won't get converted.
14480 @xref{Protoize Caveats}, for more information on how to use
14481 @code{protoize} successfully.