1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
8 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997,
9 1998, 1999, 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
11 Permission is granted to copy, distribute and/or modify this document
12 under the terms of the GNU Free Documentation License, Version 1.2 or
13 any later version published by the Free Software Foundation; with the
14 Invariant Sections being ``GNU General Public License'' and ``Funding
15 Free Software'', the Front-Cover texts being (a) (see below), and with
16 the Back-Cover Texts being (b) (see below). A copy of the license is
17 included in the gfdl(7) man page.
19 (a) The FSF's Front-Cover Text is:
23 (b) The FSF's Back-Cover Text is:
25 You have freedom to copy and modify this GNU Manual, like GNU
26 software. Copies published by the Free Software Foundation raise
27 funds for GNU development.
29 @c Set file name and title for the man page.
31 @settitle GNU project C and C++ compiler
33 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
34 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
35 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
36 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
37 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
38 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
39 [@option{-o} @var{outfile}] @var{infile}@dots{}
41 Only the most useful options are listed here; see below for the
42 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
45 gpl(7), gfdl(7), fsf-funding(7),
46 cpp(1), gcov(1), g77(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
47 and the Info entries for @file{gcc}, @file{cpp}, @file{g77}, @file{as},
48 @file{ld}, @file{binutils} and @file{gdb}.
51 For instructions on reporting bugs, see
52 @w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug}
53 script to report bugs is recommended.
56 See the Info entry for @command{gcc}, or
57 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
58 for contributors to GCC@.
63 @chapter GCC Command Options
64 @cindex GCC command options
65 @cindex command options
66 @cindex options, GCC command
68 @c man begin DESCRIPTION
69 When you invoke GCC, it normally does preprocessing, compilation,
70 assembly and linking. The ``overall options'' allow you to stop this
71 process at an intermediate stage. For example, the @option{-c} option
72 says not to run the linker. Then the output consists of object files
73 output by the assembler.
75 Other options are passed on to one stage of processing. Some options
76 control the preprocessor and others the compiler itself. Yet other
77 options control the assembler and linker; most of these are not
78 documented here, since you rarely need to use any of them.
80 @cindex C compilation options
81 Most of the command line options that you can use with GCC are useful
82 for C programs; when an option is only useful with another language
83 (usually C++), the explanation says so explicitly. If the description
84 for a particular option does not mention a source language, you can use
85 that option with all supported languages.
87 @cindex C++ compilation options
88 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
89 options for compiling C++ programs.
91 @cindex grouping options
92 @cindex options, grouping
93 The @command{gcc} program accepts options and file names as operands. Many
94 options have multi-letter names; therefore multiple single-letter options
95 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
98 @cindex order of options
99 @cindex options, order
100 You can mix options and other arguments. For the most part, the order
101 you use doesn't matter. Order does matter when you use several options
102 of the same kind; for example, if you specify @option{-L} more than once,
103 the directories are searched in the order specified.
105 Many options have long names starting with @samp{-f} or with
106 @samp{-W}---for example, @option{-fforce-mem},
107 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
108 these have both positive and negative forms; the negative form of
109 @option{-ffoo} would be @option{-fno-foo}. This manual documents
110 only one of these two forms, whichever one is not the default.
114 @xref{Option Index}, for an index to GCC's options.
117 * Option Summary:: Brief list of all options, without explanations.
118 * Overall Options:: Controlling the kind of output:
119 an executable, object files, assembler files,
120 or preprocessed source.
121 * Invoking G++:: Compiling C++ programs.
122 * C Dialect Options:: Controlling the variant of C language compiled.
123 * C++ Dialect Options:: Variations on C++.
124 * Objective-C Dialect Options:: Variations on Objective-C.
125 * Language Independent Options:: Controlling how diagnostics should be
127 * Warning Options:: How picky should the compiler be?
128 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
129 * Optimize Options:: How much optimization?
130 * Preprocessor Options:: Controlling header files and macro definitions.
131 Also, getting dependency information for Make.
132 * Assembler Options:: Passing options to the assembler.
133 * Link Options:: Specifying libraries and so on.
134 * Directory Options:: Where to find header files and libraries.
135 Where to find the compiler executable files.
136 * Spec Files:: How to pass switches to sub-processes.
137 * Target Options:: Running a cross-compiler, or an old version of GCC.
138 * Submodel Options:: Specifying minor hardware or convention variations,
139 such as 68010 vs 68020.
140 * Code Gen Options:: Specifying conventions for function calls, data layout
142 * Environment Variables:: Env vars that affect GCC.
143 * Precompiled Headers:: Compiling a header once, and using it many times.
144 * Running Protoize:: Automatically adding or removing function prototypes.
150 @section Option Summary
152 Here is a summary of all the options, grouped by type. Explanations are
153 in the following sections.
156 @item Overall Options
157 @xref{Overall Options,,Options Controlling the Kind of Output}.
158 @gccoptlist{-c -S -E -o @var{file} -pipe -pass-exit-codes @gol
159 -x @var{language} -v -### --help --target-help --version}
161 @item C Language Options
162 @xref{C Dialect Options,,Options Controlling C Dialect}.
163 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
164 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
165 -fhosted -ffreestanding -fms-extensions @gol
166 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
167 -fallow-single-precision -fcond-mismatch @gol
168 -fsigned-bitfields -fsigned-char @gol
169 -funsigned-bitfields -funsigned-char}
171 @item C++ Language Options
172 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
173 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
174 -fconserve-space -fno-const-strings @gol
175 -fno-elide-constructors @gol
176 -fno-enforce-eh-specs @gol
177 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
178 -fno-implicit-templates @gol
179 -fno-implicit-inline-templates @gol
180 -fno-implement-inlines -fms-extensions @gol
181 -fno-nonansi-builtins -fno-operator-names @gol
182 -fno-optional-diags -fpermissive @gol
183 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
184 -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
185 -fno-default-inline -Wabi -Wctor-dtor-privacy @gol
186 -Wnon-virtual-dtor -Wreorder @gol
187 -Weffc++ -Wno-deprecated @gol
188 -Wno-non-template-friend -Wold-style-cast @gol
189 -Woverloaded-virtual -Wno-pmf-conversions @gol
190 -Wsign-promo -Wsynth}
192 @item Objective-C Language Options
193 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
195 -fconstant-string-class=@var{class-name} @gol
196 -fgnu-runtime -fnext-runtime @gol
197 -fno-nil-receivers @gol
198 -fobjc-exceptions @gol
199 -freplace-objc-classes @gol
202 -Wno-protocol -Wselector -Wundeclared-selector}
204 @item Language Independent Options
205 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
206 @gccoptlist{-fmessage-length=@var{n} @gol
207 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
209 @item Warning Options
210 @xref{Warning Options,,Options to Request or Suppress Warnings}.
211 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
212 -w -Wextra -Wall -Waggregate-return @gol
213 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
214 -Wconversion -Wno-deprecated-declarations @gol
215 -Wdisabled-optimization -Wno-div-by-zero -Wendif-labels @gol
216 -Werror -Werror-implicit-function-declaration @gol
217 -Wfloat-equal -Wformat -Wformat=2 @gol
218 -Wno-format-extra-args -Wformat-nonliteral @gol
219 -Wformat-security -Wformat-y2k @gol
220 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
221 -Wimport -Wno-import -Winit-self -Winline @gol
222 -Wno-invalid-offsetof -Winvalid-pch @gol
223 -Wlarger-than-@var{len} -Wlong-long @gol
224 -Wmain -Wmissing-braces @gol
225 -Wmissing-format-attribute -Wmissing-noreturn @gol
226 -Wno-multichar -Wnonnull -Wpacked -Wpadded @gol
227 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
228 -Wreturn-type -Wsequence-point -Wshadow @gol
229 -Wsign-compare -Wstrict-aliasing -Wstrict-aliasing=2 @gol
230 -Wswitch -Wswitch-default -Wswitch-enum @gol
231 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
232 -Wunknown-pragmas -Wunreachable-code @gol
233 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
234 -Wunused-value -Wunused-variable -Wwrite-strings @gol
237 @item C-only Warning Options
238 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
239 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
240 -Wstrict-prototypes -Wtraditional @gol
241 -Wdeclaration-after-statement}
243 @item Debugging Options
244 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
245 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
246 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
247 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
248 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
249 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
250 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
251 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
252 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs @gol
253 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
254 -ftest-coverage -ftime-report -fvar-tracking @gol
255 -g -g@var{level} -gcoff -gdwarf-2 @gol
256 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
257 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
258 -print-multi-directory -print-multi-lib @gol
259 -print-prog-name=@var{program} -print-search-dirs -Q @gol
262 @item Optimization Options
263 @xref{Optimize Options,,Options that Control Optimization}.
264 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
265 -falign-labels=@var{n} -falign-loops=@var{n} @gol
266 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
267 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
268 -fcaller-saves -fcprop-registers @gol
269 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
270 -fdelayed-branch -fdelete-null-pointer-checks @gol
271 -fexpensive-optimizations -ffast-math -ffloat-store @gol
272 -fforce-addr -fforce-mem -ffunction-sections @gol
273 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -floop-optimize @gol
274 -fcrossjumping -fif-conversion -fif-conversion2 @gol
275 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
276 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
277 -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
278 -fno-default-inline -fno-defer-pop @gol
279 -fno-function-cse -fno-guess-branch-probability @gol
280 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
281 -funsafe-math-optimizations -ffinite-math-only @gol
282 -fno-trapping-math -fno-zero-initialized-in-bss @gol
283 -fomit-frame-pointer -foptimize-register-move @gol
284 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
285 -fprofile-generate -fprofile-use @gol
286 -freduce-all-givs -fregmove -frename-registers @gol
287 -freorder-blocks -freorder-functions @gol
288 -frerun-cse-after-loop -frerun-loop-opt @gol
289 -frounding-math -fschedule-insns -fschedule-insns2 @gol
290 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
291 -fsched-spec-load-dangerous @gol
292 -fsched-stalled-insns=@var{n} -sched-stalled-insns-dep=@var{n} @gol
293 -fsched2-use-superblocks @gol
294 -fsched2-use-traces -fsignaling-nans @gol
295 -fsingle-precision-constant @gol
296 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
297 -funroll-all-loops -funroll-loops -fpeel-loops @gol
298 -funswitch-loops -fold-unroll-loops -fold-unroll-all-loops @gol
299 --param @var{name}=@var{value}
300 -O -O0 -O1 -O2 -O3 -Os}
302 @item Preprocessor Options
303 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
304 @gccoptlist{-A@var{question}=@var{answer} @gol
305 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
306 -C -dD -dI -dM -dN @gol
307 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
308 -idirafter @var{dir} @gol
309 -include @var{file} -imacros @var{file} @gol
310 -iprefix @var{file} -iwithprefix @var{dir} @gol
311 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
312 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
313 -P -fworking-directory -remap @gol
314 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
315 -Xpreprocessor @var{option}}
317 @item Assembler Option
318 @xref{Assembler Options,,Passing Options to the Assembler}.
319 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
322 @xref{Link Options,,Options for Linking}.
323 @gccoptlist{@var{object-file-name} -l@var{library} @gol
324 -nostartfiles -nodefaultlibs -nostdlib -pie @gol
325 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
326 -Wl,@var{option} -Xlinker @var{option} @gol
329 @item Directory Options
330 @xref{Directory Options,,Options for Directory Search}.
331 @gccoptlist{-B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
334 @c I wrote this xref this way to avoid overfull hbox. -- rms
335 @xref{Target Options}.
336 @gccoptlist{-V @var{version} -b @var{machine}}
338 @item Machine Dependent Options
339 @xref{Submodel Options,,Hardware Models and Configurations}.
341 @emph{M680x0 Options}
342 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
343 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
344 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
345 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
346 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
348 @emph{M68hc1x Options}
349 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
350 -mauto-incdec -minmax -mlong-calls -mshort @gol
351 -msoft-reg-count=@var{count}}
354 @gccoptlist{-mg -mgnu -munix}
357 @gccoptlist{-mcpu=@var{cpu-type} @gol
358 -mtune=@var{cpu-type} @gol
359 -mcmodel=@var{code-model} @gol
360 -m32 -m64 -mapp-regs -mno-app-regs @gol
361 -mfaster-structs -mno-faster-structs @gol
362 -mfpu -mno-fpu -mhard-float -msoft-float @gol
363 -mhard-quad-float -msoft-quad-float @gol
364 -mimpure-text -mno-impure-text -mlittle-endian @gol
365 -mstack-bias -mno-stack-bias @gol
366 -munaligned-doubles -mno-unaligned-doubles @gol
367 -mv8plus -mno-v8plus -mvis -mno-vis}
370 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
371 -mapcs-26 -mapcs-32 @gol
372 -mapcs-stack-check -mno-apcs-stack-check @gol
373 -mapcs-float -mno-apcs-float @gol
374 -mapcs-reentrant -mno-apcs-reentrant @gol
375 -msched-prolog -mno-sched-prolog @gol
376 -mlittle-endian -mbig-endian -mwords-little-endian @gol
377 -malignment-traps -mno-alignment-traps @gol
378 -mfloat-abi=@var{name} soft-float -mhard-float -mfpe @gol
379 -mthumb-interwork -mno-thumb-interwork @gol
380 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
381 -mstructure-size-boundary=@var{n} @gol
382 -mabort-on-noreturn @gol
383 -mlong-calls -mno-long-calls @gol
384 -msingle-pic-base -mno-single-pic-base @gol
385 -mpic-register=@var{reg} @gol
386 -mnop-fun-dllimport @gol
387 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
388 -mpoke-function-name @gol
390 -mtpcs-frame -mtpcs-leaf-frame @gol
391 -mcaller-super-interworking -mcallee-super-interworking}
393 @emph{MN10300 Options}
394 @gccoptlist{-mmult-bug -mno-mult-bug @gol
395 -mam33 -mno-am33 @gol
396 -mam33-2 -mno-am33-2 @gol
399 @emph{M32R/D Options}
400 @gccoptlist{-m32r2 -m32rx -m32r @gol
402 -malign-loops -mno-align-loops @gol
403 -missue-rate=@var{number} @gol
404 -mbranch-cost=@var{number} @gol
405 -mmodel=@var{code-size-model-type} @gol
406 -msdata=@var{sdata-type} @gol
407 -mno-flush-func -mflush-func=@var{name} @gol
408 -mno-flush-trap -mflush-trap=@var{number} @gol
411 @emph{RS/6000 and PowerPC Options}
412 @gccoptlist{-mcpu=@var{cpu-type} @gol
413 -mtune=@var{cpu-type} @gol
414 -mpower -mno-power -mpower2 -mno-power2 @gol
415 -mpowerpc -mpowerpc64 -mno-powerpc @gol
416 -maltivec -mno-altivec @gol
417 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
418 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
419 -mnew-mnemonics -mold-mnemonics @gol
420 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
421 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
422 -malign-power -malign-natural @gol
423 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
424 -mstring -mno-string -mupdate -mno-update @gol
425 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
426 -mstrict-align -mno-strict-align -mrelocatable @gol
427 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
428 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
429 -mdynamic-no-pic @gol
430 -mprioritize-restricted-insns=@var{priority} @gol
431 -msched-costly-dep=@var{dependence_type} @gol
432 -minsert-sched-nops=@var{scheme} @gol
433 -mcall-sysv -mcall-netbsd @gol
434 -maix-struct-return -msvr4-struct-return @gol
435 -mabi=altivec -mabi=no-altivec @gol
436 -mabi=spe -mabi=no-spe @gol
437 -misel=yes -misel=no @gol
438 -mspe=yes -mspe=no @gol
439 -mfloat-gprs=yes -mfloat-gprs=no @gol
440 -mprototype -mno-prototype @gol
441 -msim -mmvme -mads -myellowknife -memb -msdata @gol
442 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
444 @emph{Darwin Options}
445 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
446 -arch_only -bind_at_load -bundle -bundle_loader @gol
447 -client_name -compatibility_version -current_version @gol
448 -dependency-file -dylib_file -dylinker_install_name @gol
449 -dynamic -dynamiclib -exported_symbols_list @gol
450 -filelist -flat_namespace -force_cpusubtype_ALL @gol
451 -force_flat_namespace -headerpad_max_install_names @gol
452 -image_base -init -install_name -keep_private_externs @gol
453 -multi_module -multiply_defined -multiply_defined_unused @gol
454 -noall_load -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
455 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
456 -private_bundle -read_only_relocs -sectalign @gol
457 -sectobjectsymbols -whyload -seg1addr @gol
458 -sectcreate -sectobjectsymbols -sectorder @gol
459 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
460 -segprot -segs_read_only_addr -segs_read_write_addr @gol
461 -single_module -static -sub_library -sub_umbrella @gol
462 -twolevel_namespace -umbrella -undefined @gol
463 -unexported_symbols_list -weak_reference_mismatches @gol
467 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
468 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
469 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
470 -mxgot -mno-xgot -membedded-pic -mno-embedded-pic @gol
471 -mgp32 -mgp64 -mfp32 -mfp64 -mhard-float -msoft-float @gol
472 -msingle-float -mdouble-float -mint64 -mlong64 -mlong32 @gol
473 -G@var{num} -membedded-data -mno-embedded-data @gol
474 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
475 -msplit-addresses -mno-split-addresses @gol
476 -mexplicit-relocs -mno-explicit-relocs @gol
477 -mrnames -mno-rnames @gol
478 -mcheck-zero-division -mno-check-zero-division @gol
479 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
480 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
481 -mfix-r4000 -mno-fix-r4000 -mfix-sb1 -mno-fix-sb1 @gol
482 -mflush-func=@var{func} -mno-flush-func @gol
483 -mbranch-likely -mno-branch-likely}
485 @emph{i386 and x86-64 Options}
486 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
487 -mfpmath=@var{unit} @gol
488 -masm=@var{dialect} -mno-fancy-math-387 @gol
489 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
490 -mno-wide-multiply -mrtd -malign-double @gol
491 -mpreferred-stack-boundary=@var{num} @gol
492 -mmmx -msse -msse2 -msse3 -m3dnow @gol
493 -mthreads -mno-align-stringops -minline-all-stringops @gol
494 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
495 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
496 -mno-red-zone -mno-tls-direct-seg-refs @gol
497 -mcmodel=@var{code-model} @gol
501 @gccoptlist{-march=@var{architecture-type} @gol
502 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
503 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
504 -mjump-in-delay -mlinker-opt -mlong-calls @gol
505 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
506 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
507 -mno-jump-in-delay -mno-long-load-store @gol
508 -mno-portable-runtime -mno-soft-float @gol
509 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
510 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
511 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
512 -nolibdld -static -threads}
514 @emph{DEC Alpha Options}
515 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
516 -mieee -mieee-with-inexact -mieee-conformant @gol
517 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
518 -mtrap-precision=@var{mode} -mbuild-constants @gol
519 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
520 -mbwx -mmax -mfix -mcix @gol
521 -mfloat-vax -mfloat-ieee @gol
522 -mexplicit-relocs -msmall-data -mlarge-data @gol
523 -msmall-text -mlarge-text @gol
524 -mmemory-latency=@var{time}}
526 @emph{DEC Alpha/VMS Options}
527 @gccoptlist{-mvms-return-codes}
529 @emph{H8/300 Options}
530 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
533 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
534 -m4-nofpu -m4-single-only -m4-single -m4 @gol
535 -m5-64media -m5-64media-nofpu @gol
536 -m5-32media -m5-32media-nofpu @gol
537 -m5-compact -m5-compact-nofpu @gol
538 -mb -ml -mdalign -mrelax @gol
539 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
540 -mieee -misize -mpadstruct -mspace @gol
541 -mprefergot -musermode}
543 @emph{System V Options}
544 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
547 @gccoptlist{-EB -EL @gol
548 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
549 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
551 @emph{TMS320C3x/C4x Options}
552 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
553 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
554 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
555 -mparallel-insns -mparallel-mpy -mpreserve-float}
558 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
559 -mprolog-function -mno-prolog-function -mspace @gol
560 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
561 -mapp-regs -mno-app-regs @gol
562 -mdisable-callt -mno-disable-callt @gol
568 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
569 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
570 -mregparam -mnoregparam -msb -mnosb @gol
571 -mbitfield -mnobitfield -mhimem -mnohimem}
574 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
575 -mcall-prologues -mno-tablejump -mtiny-stack}
578 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
579 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
580 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
581 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
582 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
585 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
586 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
587 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
588 -mno-base-addresses -msingle-exit -mno-single-exit}
591 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
592 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
593 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
594 -minline-float-divide-max-throughput @gol
595 -minline-int-divide-min-latency @gol
596 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
597 -mfixed-range=@var{register-range}}
599 @emph{S/390 and zSeries Options}
600 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
601 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
602 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
603 -m64 -m31 -mdebug -mno-debug -mesa -mzarch -mfused-madd -mno-fused-madd}
606 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
607 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
608 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
609 -mstack-align -mdata-align -mconst-align @gol
610 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
611 -melf -maout -melinux -mlinux -sim -sim2 @gol
612 -mmul-bug-workaround -mno-mul-bug-workaround}
614 @emph{PDP-11 Options}
615 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
616 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
617 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
618 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
619 -mbranch-expensive -mbranch-cheap @gol
620 -msplit -mno-split -munix-asm -mdec-asm}
622 @emph{Xstormy16 Options}
625 @emph{Xtensa Options}
626 @gccoptlist{-mconst16 -mno-const16 @gol
627 -mfused-madd -mno-fused-madd @gol
628 -mtext-section-literals -mno-text-section-literals @gol
629 -mtarget-align -mno-target-align @gol
630 -mlongcalls -mno-longcalls}
633 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
634 -mhard-float -msoft-float @gol
635 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
636 -mdouble -mno-double @gol
637 -mmedia -mno-media -mmuladd -mno-muladd @gol
638 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic -mlinked-fp @gol
639 -mlibrary-pic -macc-4 -macc-8 @gol
640 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
641 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
642 -mvliw-branch -mno-vliw-branch @gol
643 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
644 -mno-nested-cond-exec -mtomcat-stats @gol
647 @item Code Generation Options
648 @xref{Code Gen Options,,Options for Code Generation Conventions}.
649 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
650 -ffixed-@var{reg} -fexceptions @gol
651 -fnon-call-exceptions -funwind-tables @gol
652 -fasynchronous-unwind-tables @gol
653 -finhibit-size-directive -finstrument-functions @gol
654 -fno-common -fno-ident @gol
655 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
656 -freg-struct-return -fshared-data -fshort-enums @gol
657 -fshort-double -fshort-wchar @gol
658 -fverbose-asm -fpack-struct -fstack-check @gol
659 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
660 -fargument-alias -fargument-noalias @gol
661 -fargument-noalias-global -fleading-underscore @gol
662 -ftls-model=@var{model} @gol
663 -ftrapv -fwrapv -fbounds-check}
667 * Overall Options:: Controlling the kind of output:
668 an executable, object files, assembler files,
669 or preprocessed source.
670 * C Dialect Options:: Controlling the variant of C language compiled.
671 * C++ Dialect Options:: Variations on C++.
672 * Objective-C Dialect Options:: Variations on Objective-C.
673 * Language Independent Options:: Controlling how diagnostics should be
675 * Warning Options:: How picky should the compiler be?
676 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
677 * Optimize Options:: How much optimization?
678 * Preprocessor Options:: Controlling header files and macro definitions.
679 Also, getting dependency information for Make.
680 * Assembler Options:: Passing options to the assembler.
681 * Link Options:: Specifying libraries and so on.
682 * Directory Options:: Where to find header files and libraries.
683 Where to find the compiler executable files.
684 * Spec Files:: How to pass switches to sub-processes.
685 * Target Options:: Running a cross-compiler, or an old version of GCC.
688 @node Overall Options
689 @section Options Controlling the Kind of Output
691 Compilation can involve up to four stages: preprocessing, compilation
692 proper, assembly and linking, always in that order. GCC is capable of
693 preprocessing and compiling several files either into several
694 assembler input files, or into one assembler input file; then each
695 assembler input file produces an object file, and linking combines all
696 the object files (those newly compiled, and those specified as input)
697 into an executable file.
699 @cindex file name suffix
700 For any given input file, the file name suffix determines what kind of
705 C source code which must be preprocessed.
708 C source code which should not be preprocessed.
711 C++ source code which should not be preprocessed.
714 Objective-C source code. Note that you must link with the library
715 @file{libobjc.a} to make an Objective-C program work.
718 Objective-C source code which should not be preprocessed.
721 C or C++ header file to be turned into a precompiled header.
725 @itemx @var{file}.cxx
726 @itemx @var{file}.cpp
727 @itemx @var{file}.CPP
728 @itemx @var{file}.c++
730 C++ source code which must be preprocessed. Note that in @samp{.cxx},
731 the last two letters must both be literally @samp{x}. Likewise,
732 @samp{.C} refers to a literal capital C@.
736 C++ header file to be turned into a precompiled header.
739 @itemx @var{file}.for
740 @itemx @var{file}.FOR
741 Fortran source code which should not be preprocessed.
744 @itemx @var{file}.fpp
745 @itemx @var{file}.FPP
746 Fortran source code which must be preprocessed (with the traditional
750 Fortran source code which must be preprocessed with a RATFOR
751 preprocessor (not included with GCC)@.
753 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
754 Using and Porting GNU Fortran}, for more details of the handling of
757 @c FIXME: Descriptions of Java file types.
764 Ada source code file which contains a library unit declaration (a
765 declaration of a package, subprogram, or generic, or a generic
766 instantiation), or a library unit renaming declaration (a package,
767 generic, or subprogram renaming declaration). Such files are also
770 @itemx @var{file}.adb
771 Ada source code file containing a library unit body (a subprogram or
772 package body). Such files are also called @dfn{bodies}.
774 @c GCC also knows about some suffixes for languages not yet included:
783 Assembler code which must be preprocessed.
786 An object file to be fed straight into linking.
787 Any file name with no recognized suffix is treated this way.
791 You can specify the input language explicitly with the @option{-x} option:
794 @item -x @var{language}
795 Specify explicitly the @var{language} for the following input files
796 (rather than letting the compiler choose a default based on the file
797 name suffix). This option applies to all following input files until
798 the next @option{-x} option. Possible values for @var{language} are:
800 c c-header cpp-output
801 c++ c++-header c++-cpp-output
802 objective-c objective-c-header objc-cpp-output
803 assembler assembler-with-cpp
805 f77 f77-cpp-input ratfor
811 Turn off any specification of a language, so that subsequent files are
812 handled according to their file name suffixes (as they are if @option{-x}
813 has not been used at all).
815 @item -pass-exit-codes
816 @opindex pass-exit-codes
817 Normally the @command{gcc} program will exit with the code of 1 if any
818 phase of the compiler returns a non-success return code. If you specify
819 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
820 numerically highest error produced by any phase that returned an error
824 If you only want some of the stages of compilation, you can use
825 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
826 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
827 @command{gcc} is to stop. Note that some combinations (for example,
828 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
833 Compile or assemble the source files, but do not link. The linking
834 stage simply is not done. The ultimate output is in the form of an
835 object file for each source file.
837 By default, the object file name for a source file is made by replacing
838 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
840 Unrecognized input files, not requiring compilation or assembly, are
845 Stop after the stage of compilation proper; do not assemble. The output
846 is in the form of an assembler code file for each non-assembler input
849 By default, the assembler file name for a source file is made by
850 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
852 Input files that don't require compilation are ignored.
856 Stop after the preprocessing stage; do not run the compiler proper. The
857 output is in the form of preprocessed source code, which is sent to the
860 Input files which don't require preprocessing are ignored.
862 @cindex output file option
865 Place output in file @var{file}. This applies regardless to whatever
866 sort of output is being produced, whether it be an executable file,
867 an object file, an assembler file or preprocessed C code.
869 If you specify @option{-o} when compiling more than one input file, or
870 you are producing an executable file as output, all the source files
871 on the command line will be compiled at once.
873 If @option{-o} is not specified, the default is to put an executable file
874 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
875 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
876 all preprocessed C source on standard output.
880 Print (on standard error output) the commands executed to run the stages
881 of compilation. Also print the version number of the compiler driver
882 program and of the preprocessor and the compiler proper.
886 Like @option{-v} except the commands are not executed and all command
887 arguments are quoted. This is useful for shell scripts to capture the
888 driver-generated command lines.
892 Use pipes rather than temporary files for communication between the
893 various stages of compilation. This fails to work on some systems where
894 the assembler is unable to read from a pipe; but the GNU assembler has
899 Print (on the standard output) a description of the command line options
900 understood by @command{gcc}. If the @option{-v} option is also specified
901 then @option{--help} will also be passed on to the various processes
902 invoked by @command{gcc}, so that they can display the command line options
903 they accept. If the @option{-Wextra} option is also specified then command
904 line options which have no documentation associated with them will also
909 Print (on the standard output) a description of target specific command
910 line options for each tool.
914 Display the version number and copyrights of the invoked GCC.
918 @section Compiling C++ Programs
920 @cindex suffixes for C++ source
921 @cindex C++ source file suffixes
922 C++ source files conventionally use one of the suffixes @samp{.C},
923 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
924 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
925 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
926 files with these names and compiles them as C++ programs even if you
927 call the compiler the same way as for compiling C programs (usually
928 with the name @command{gcc}).
932 However, C++ programs often require class libraries as well as a
933 compiler that understands the C++ language---and under some
934 circumstances, you might want to compile programs or header files from
935 standard input, or otherwise without a suffix that flags them as C++
936 programs. You might also like to precompile a C header file with a
937 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
938 program that calls GCC with the default language set to C++, and
939 automatically specifies linking against the C++ library. On many
940 systems, @command{g++} is also installed with the name @command{c++}.
942 @cindex invoking @command{g++}
943 When you compile C++ programs, you may specify many of the same
944 command-line options that you use for compiling programs in any
945 language; or command-line options meaningful for C and related
946 languages; or options that are meaningful only for C++ programs.
947 @xref{C Dialect Options,,Options Controlling C Dialect}, for
948 explanations of options for languages related to C@.
949 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
950 explanations of options that are meaningful only for C++ programs.
952 @node C Dialect Options
953 @section Options Controlling C Dialect
954 @cindex dialect options
955 @cindex language dialect options
956 @cindex options, dialect
958 The following options control the dialect of C (or languages derived
959 from C, such as C++ and Objective-C) that the compiler accepts:
966 In C mode, support all ISO C90 programs. In C++ mode,
967 remove GNU extensions that conflict with ISO C++.
969 This turns off certain features of GCC that are incompatible with ISO
970 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
971 such as the @code{asm} and @code{typeof} keywords, and
972 predefined macros such as @code{unix} and @code{vax} that identify the
973 type of system you are using. It also enables the undesirable and
974 rarely used ISO trigraph feature. For the C compiler,
975 it disables recognition of C++ style @samp{//} comments as well as
976 the @code{inline} keyword.
978 The alternate keywords @code{__asm__}, @code{__extension__},
979 @code{__inline__} and @code{__typeof__} continue to work despite
980 @option{-ansi}. You would not want to use them in an ISO C program, of
981 course, but it is useful to put them in header files that might be included
982 in compilations done with @option{-ansi}. Alternate predefined macros
983 such as @code{__unix__} and @code{__vax__} are also available, with or
984 without @option{-ansi}.
986 The @option{-ansi} option does not cause non-ISO programs to be
987 rejected gratuitously. For that, @option{-pedantic} is required in
988 addition to @option{-ansi}. @xref{Warning Options}.
990 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
991 option is used. Some header files may notice this macro and refrain
992 from declaring certain functions or defining certain macros that the
993 ISO standard doesn't call for; this is to avoid interfering with any
994 programs that might use these names for other things.
996 Functions which would normally be built in but do not have semantics
997 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
998 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
999 built-in functions provided by GCC}, for details of the functions
1004 Determine the language standard. This option is currently only
1005 supported when compiling C or C++. A value for this option must be
1006 provided; possible values are
1011 ISO C90 (same as @option{-ansi}).
1013 @item iso9899:199409
1014 ISO C90 as modified in amendment 1.
1020 ISO C99. Note that this standard is not yet fully supported; see
1021 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1022 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1025 Default, ISO C90 plus GNU extensions (including some C99 features).
1029 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1030 this will become the default. The name @samp{gnu9x} is deprecated.
1033 The 1998 ISO C++ standard plus amendments.
1036 The same as @option{-std=c++98} plus GNU extensions. This is the
1037 default for C++ code.
1040 Even when this option is not specified, you can still use some of the
1041 features of newer standards in so far as they do not conflict with
1042 previous C standards. For example, you may use @code{__restrict__} even
1043 when @option{-std=c99} is not specified.
1045 The @option{-std} options specifying some version of ISO C have the same
1046 effects as @option{-ansi}, except that features that were not in ISO C90
1047 but are in the specified version (for example, @samp{//} comments and
1048 the @code{inline} keyword in ISO C99) are not disabled.
1050 @xref{Standards,,Language Standards Supported by GCC}, for details of
1051 these standard versions.
1053 @item -aux-info @var{filename}
1055 Output to the given filename prototyped declarations for all functions
1056 declared and/or defined in a translation unit, including those in header
1057 files. This option is silently ignored in any language other than C@.
1059 Besides declarations, the file indicates, in comments, the origin of
1060 each declaration (source file and line), whether the declaration was
1061 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1062 @samp{O} for old, respectively, in the first character after the line
1063 number and the colon), and whether it came from a declaration or a
1064 definition (@samp{C} or @samp{F}, respectively, in the following
1065 character). In the case of function definitions, a K&R-style list of
1066 arguments followed by their declarations is also provided, inside
1067 comments, after the declaration.
1071 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1072 keyword, so that code can use these words as identifiers. You can use
1073 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1074 instead. @option{-ansi} implies @option{-fno-asm}.
1076 In C++, this switch only affects the @code{typeof} keyword, since
1077 @code{asm} and @code{inline} are standard keywords. You may want to
1078 use the @option{-fno-gnu-keywords} flag instead, which has the same
1079 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1080 switch only affects the @code{asm} and @code{typeof} keywords, since
1081 @code{inline} is a standard keyword in ISO C99.
1084 @itemx -fno-builtin-@var{function}
1085 @opindex fno-builtin
1086 @cindex built-in functions
1087 Don't recognize built-in functions that do not begin with
1088 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1089 functions provided by GCC}, for details of the functions affected,
1090 including those which are not built-in functions when @option{-ansi} or
1091 @option{-std} options for strict ISO C conformance are used because they
1092 do not have an ISO standard meaning.
1094 GCC normally generates special code to handle certain built-in functions
1095 more efficiently; for instance, calls to @code{alloca} may become single
1096 instructions that adjust the stack directly, and calls to @code{memcpy}
1097 may become inline copy loops. The resulting code is often both smaller
1098 and faster, but since the function calls no longer appear as such, you
1099 cannot set a breakpoint on those calls, nor can you change the behavior
1100 of the functions by linking with a different library.
1102 With the @option{-fno-builtin-@var{function}} option
1103 only the built-in function @var{function} is
1104 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1105 function is named this is not built-in in this version of GCC, this
1106 option is ignored. There is no corresponding
1107 @option{-fbuiltin-@var{function}} option; if you wish to enable
1108 built-in functions selectively when using @option{-fno-builtin} or
1109 @option{-ffreestanding}, you may define macros such as:
1112 #define abs(n) __builtin_abs ((n))
1113 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1118 @cindex hosted environment
1120 Assert that compilation takes place in a hosted environment. This implies
1121 @option{-fbuiltin}. A hosted environment is one in which the
1122 entire standard library is available, and in which @code{main} has a return
1123 type of @code{int}. Examples are nearly everything except a kernel.
1124 This is equivalent to @option{-fno-freestanding}.
1126 @item -ffreestanding
1127 @opindex ffreestanding
1128 @cindex hosted environment
1130 Assert that compilation takes place in a freestanding environment. This
1131 implies @option{-fno-builtin}. A freestanding environment
1132 is one in which the standard library may not exist, and program startup may
1133 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1134 This is equivalent to @option{-fno-hosted}.
1136 @xref{Standards,,Language Standards Supported by GCC}, for details of
1137 freestanding and hosted environments.
1139 @item -fms-extensions
1140 @opindex fms-extensions
1141 Accept some non-standard constructs used in Microsoft header files.
1145 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1146 options for strict ISO C conformance) implies @option{-trigraphs}.
1148 @item -no-integrated-cpp
1149 @opindex no-integrated-cpp
1150 Performs a compilation in two passes: preprocessing and compiling. This
1151 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1152 @option{-B} option. The user supplied compilation step can then add in
1153 an additional preprocessing step after normal preprocessing but before
1154 compiling. The default is to use the integrated cpp (internal cpp)
1156 The semantics of this option will change if "cc1", "cc1plus", and
1157 "cc1obj" are merged.
1159 @cindex traditional C language
1160 @cindex C language, traditional
1162 @itemx -traditional-cpp
1163 @opindex traditional-cpp
1164 @opindex traditional
1165 Formerly, these options caused GCC to attempt to emulate a pre-standard
1166 C compiler. They are now only supported with the @option{-E} switch.
1167 The preprocessor continues to support a pre-standard mode. See the GNU
1168 CPP manual for details.
1170 @item -fcond-mismatch
1171 @opindex fcond-mismatch
1172 Allow conditional expressions with mismatched types in the second and
1173 third arguments. The value of such an expression is void. This option
1174 is not supported for C++.
1176 @item -funsigned-char
1177 @opindex funsigned-char
1178 Let the type @code{char} be unsigned, like @code{unsigned char}.
1180 Each kind of machine has a default for what @code{char} should
1181 be. It is either like @code{unsigned char} by default or like
1182 @code{signed char} by default.
1184 Ideally, a portable program should always use @code{signed char} or
1185 @code{unsigned char} when it depends on the signedness of an object.
1186 But many programs have been written to use plain @code{char} and
1187 expect it to be signed, or expect it to be unsigned, depending on the
1188 machines they were written for. This option, and its inverse, let you
1189 make such a program work with the opposite default.
1191 The type @code{char} is always a distinct type from each of
1192 @code{signed char} or @code{unsigned char}, even though its behavior
1193 is always just like one of those two.
1196 @opindex fsigned-char
1197 Let the type @code{char} be signed, like @code{signed char}.
1199 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1200 the negative form of @option{-funsigned-char}. Likewise, the option
1201 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1203 @item -fsigned-bitfields
1204 @itemx -funsigned-bitfields
1205 @itemx -fno-signed-bitfields
1206 @itemx -fno-unsigned-bitfields
1207 @opindex fsigned-bitfields
1208 @opindex funsigned-bitfields
1209 @opindex fno-signed-bitfields
1210 @opindex fno-unsigned-bitfields
1211 These options control whether a bit-field is signed or unsigned, when the
1212 declaration does not use either @code{signed} or @code{unsigned}. By
1213 default, such a bit-field is signed, because this is consistent: the
1214 basic integer types such as @code{int} are signed types.
1217 @node C++ Dialect Options
1218 @section Options Controlling C++ Dialect
1220 @cindex compiler options, C++
1221 @cindex C++ options, command line
1222 @cindex options, C++
1223 This section describes the command-line options that are only meaningful
1224 for C++ programs; but you can also use most of the GNU compiler options
1225 regardless of what language your program is in. For example, you
1226 might compile a file @code{firstClass.C} like this:
1229 g++ -g -frepo -O -c firstClass.C
1233 In this example, only @option{-frepo} is an option meant
1234 only for C++ programs; you can use the other options with any
1235 language supported by GCC@.
1237 Here is a list of options that are @emph{only} for compiling C++ programs:
1241 @item -fabi-version=@var{n}
1242 @opindex fabi-version
1243 Use version @var{n} of the C++ ABI. Version 2 is the version of the
1244 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1245 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1246 the version that conforms most closely to the C++ ABI specification.
1247 Therefore, the ABI obtained using version 0 will change as ABI bugs
1250 The default is version 2.
1252 @item -fno-access-control
1253 @opindex fno-access-control
1254 Turn off all access checking. This switch is mainly useful for working
1255 around bugs in the access control code.
1259 Check that the pointer returned by @code{operator new} is non-null
1260 before attempting to modify the storage allocated. This check is
1261 normally unnecessary because the C++ standard specifies that
1262 @code{operator new} will only return @code{0} if it is declared
1263 @samp{throw()}, in which case the compiler will always check the
1264 return value even without this option. In all other cases, when
1265 @code{operator new} has a non-empty exception specification, memory
1266 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1267 @samp{new (nothrow)}.
1269 @item -fconserve-space
1270 @opindex fconserve-space
1271 Put uninitialized or runtime-initialized global variables into the
1272 common segment, as C does. This saves space in the executable at the
1273 cost of not diagnosing duplicate definitions. If you compile with this
1274 flag and your program mysteriously crashes after @code{main()} has
1275 completed, you may have an object that is being destroyed twice because
1276 two definitions were merged.
1278 This option is no longer useful on most targets, now that support has
1279 been added for putting variables into BSS without making them common.
1281 @item -fno-const-strings
1282 @opindex fno-const-strings
1283 Give string constants type @code{char *} instead of type @code{const
1284 char *}. By default, G++ uses type @code{const char *} as required by
1285 the standard. Even if you use @option{-fno-const-strings}, you cannot
1286 actually modify the value of a string constant.
1288 This option might be removed in a future release of G++. For maximum
1289 portability, you should structure your code so that it works with
1290 string constants that have type @code{const char *}.
1292 @item -fno-elide-constructors
1293 @opindex fno-elide-constructors
1294 The C++ standard allows an implementation to omit creating a temporary
1295 which is only used to initialize another object of the same type.
1296 Specifying this option disables that optimization, and forces G++ to
1297 call the copy constructor in all cases.
1299 @item -fno-enforce-eh-specs
1300 @opindex fno-enforce-eh-specs
1301 Don't check for violation of exception specifications at runtime. This
1302 option violates the C++ standard, but may be useful for reducing code
1303 size in production builds, much like defining @samp{NDEBUG}. The compiler
1304 will still optimize based on the exception specifications.
1307 @itemx -fno-for-scope
1309 @opindex fno-for-scope
1310 If @option{-ffor-scope} is specified, the scope of variables declared in
1311 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1312 as specified by the C++ standard.
1313 If @option{-fno-for-scope} is specified, the scope of variables declared in
1314 a @i{for-init-statement} extends to the end of the enclosing scope,
1315 as was the case in old versions of G++, and other (traditional)
1316 implementations of C++.
1318 The default if neither flag is given to follow the standard,
1319 but to allow and give a warning for old-style code that would
1320 otherwise be invalid, or have different behavior.
1322 @item -fno-gnu-keywords
1323 @opindex fno-gnu-keywords
1324 Do not recognize @code{typeof} as a keyword, so that code can use this
1325 word as an identifier. You can use the keyword @code{__typeof__} instead.
1326 @option{-ansi} implies @option{-fno-gnu-keywords}.
1328 @item -fno-implicit-templates
1329 @opindex fno-implicit-templates
1330 Never emit code for non-inline templates which are instantiated
1331 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1332 @xref{Template Instantiation}, for more information.
1334 @item -fno-implicit-inline-templates
1335 @opindex fno-implicit-inline-templates
1336 Don't emit code for implicit instantiations of inline templates, either.
1337 The default is to handle inlines differently so that compiles with and
1338 without optimization will need the same set of explicit instantiations.
1340 @item -fno-implement-inlines
1341 @opindex fno-implement-inlines
1342 To save space, do not emit out-of-line copies of inline functions
1343 controlled by @samp{#pragma implementation}. This will cause linker
1344 errors if these functions are not inlined everywhere they are called.
1346 @item -fms-extensions
1347 @opindex fms-extensions
1348 Disable pedantic warnings about constructs used in MFC, such as implicit
1349 int and getting a pointer to member function via non-standard syntax.
1351 @item -fno-nonansi-builtins
1352 @opindex fno-nonansi-builtins
1353 Disable built-in declarations of functions that are not mandated by
1354 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1355 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1357 @item -fno-operator-names
1358 @opindex fno-operator-names
1359 Do not treat the operator name keywords @code{and}, @code{bitand},
1360 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1361 synonyms as keywords.
1363 @item -fno-optional-diags
1364 @opindex fno-optional-diags
1365 Disable diagnostics that the standard says a compiler does not need to
1366 issue. Currently, the only such diagnostic issued by G++ is the one for
1367 a name having multiple meanings within a class.
1370 @opindex fpermissive
1371 Downgrade some diagnostics about nonconformant code from errors to
1372 warnings. Thus, using @option{-fpermissive} will allow some
1373 nonconforming code to compile.
1377 Enable automatic template instantiation at link time. This option also
1378 implies @option{-fno-implicit-templates}. @xref{Template
1379 Instantiation}, for more information.
1383 Disable generation of information about every class with virtual
1384 functions for use by the C++ runtime type identification features
1385 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1386 of the language, you can save some space by using this flag. Note that
1387 exception handling uses the same information, but it will generate it as
1392 Emit statistics about front-end processing at the end of the compilation.
1393 This information is generally only useful to the G++ development team.
1395 @item -ftemplate-depth-@var{n}
1396 @opindex ftemplate-depth
1397 Set the maximum instantiation depth for template classes to @var{n}.
1398 A limit on the template instantiation depth is needed to detect
1399 endless recursions during template class instantiation. ANSI/ISO C++
1400 conforming programs must not rely on a maximum depth greater than 17.
1402 @item -fuse-cxa-atexit
1403 @opindex fuse-cxa-atexit
1404 Register destructors for objects with static storage duration with the
1405 @code{__cxa_atexit} function rather than the @code{atexit} function.
1406 This option is required for fully standards-compliant handling of static
1407 destructors, but will only work if your C library supports
1408 @code{__cxa_atexit}.
1412 Do not use weak symbol support, even if it is provided by the linker.
1413 By default, G++ will use weak symbols if they are available. This
1414 option exists only for testing, and should not be used by end-users;
1415 it will result in inferior code and has no benefits. This option may
1416 be removed in a future release of G++.
1420 Do not search for header files in the standard directories specific to
1421 C++, but do still search the other standard directories. (This option
1422 is used when building the C++ library.)
1425 In addition, these optimization, warning, and code generation options
1426 have meanings only for C++ programs:
1429 @item -fno-default-inline
1430 @opindex fno-default-inline
1431 Do not assume @samp{inline} for functions defined inside a class scope.
1432 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1433 functions will have linkage like inline functions; they just won't be
1436 @item -Wabi @r{(C++ only)}
1438 Warn when G++ generates code that is probably not compatible with the
1439 vendor-neutral C++ ABI. Although an effort has been made to warn about
1440 all such cases, there are probably some cases that are not warned about,
1441 even though G++ is generating incompatible code. There may also be
1442 cases where warnings are emitted even though the code that is generated
1445 You should rewrite your code to avoid these warnings if you are
1446 concerned about the fact that code generated by G++ may not be binary
1447 compatible with code generated by other compilers.
1449 The known incompatibilities at this point include:
1454 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1455 pack data into the same byte as a base class. For example:
1458 struct A @{ virtual void f(); int f1 : 1; @};
1459 struct B : public A @{ int f2 : 1; @};
1463 In this case, G++ will place @code{B::f2} into the same byte
1464 as@code{A::f1}; other compilers will not. You can avoid this problem
1465 by explicitly padding @code{A} so that its size is a multiple of the
1466 byte size on your platform; that will cause G++ and other compilers to
1467 layout @code{B} identically.
1470 Incorrect handling of tail-padding for virtual bases. G++ does not use
1471 tail padding when laying out virtual bases. For example:
1474 struct A @{ virtual void f(); char c1; @};
1475 struct B @{ B(); char c2; @};
1476 struct C : public A, public virtual B @{@};
1480 In this case, G++ will not place @code{B} into the tail-padding for
1481 @code{A}; other compilers will. You can avoid this problem by
1482 explicitly padding @code{A} so that its size is a multiple of its
1483 alignment (ignoring virtual base classes); that will cause G++ and other
1484 compilers to layout @code{C} identically.
1487 Incorrect handling of bit-fields with declared widths greater than that
1488 of their underlying types, when the bit-fields appear in a union. For
1492 union U @{ int i : 4096; @};
1496 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1497 union too small by the number of bits in an @code{int}.
1500 Empty classes can be placed at incorrect offsets. For example:
1510 struct C : public B, public A @{@};
1514 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1515 it should be placed at offset zero. G++ mistakenly believes that the
1516 @code{A} data member of @code{B} is already at offset zero.
1519 Names of template functions whose types involve @code{typename} or
1520 template template parameters can be mangled incorrectly.
1523 template <typename Q>
1524 void f(typename Q::X) @{@}
1526 template <template <typename> class Q>
1527 void f(typename Q<int>::X) @{@}
1531 Instantiations of these templates may be mangled incorrectly.
1535 @item -Wctor-dtor-privacy @r{(C++ only)}
1536 @opindex Wctor-dtor-privacy
1537 Warn when a class seems unusable because all the constructors or
1538 destructors in that class are private, and it has neither friends nor
1539 public static member functions.
1541 @item -Wnon-virtual-dtor @r{(C++ only)}
1542 @opindex Wnon-virtual-dtor
1543 Warn when a class appears to be polymorphic, thereby requiring a virtual
1544 destructor, yet it declares a non-virtual one.
1545 This warning is enabled by @option{-Wall}.
1547 @item -Wreorder @r{(C++ only)}
1549 @cindex reordering, warning
1550 @cindex warning for reordering of member initializers
1551 Warn when the order of member initializers given in the code does not
1552 match the order in which they must be executed. For instance:
1558 A(): j (0), i (1) @{ @}
1562 The compiler will rearrange the member initializers for @samp{i}
1563 and @samp{j} to match the declaration order of the members, emitting
1564 a warning to that effect. This warning is enabled by @option{-Wall}.
1567 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1570 @item -Weffc++ @r{(C++ only)}
1572 Warn about violations of the following style guidelines from Scott Meyers'
1573 @cite{Effective C++} book:
1577 Item 11: Define a copy constructor and an assignment operator for classes
1578 with dynamically allocated memory.
1581 Item 12: Prefer initialization to assignment in constructors.
1584 Item 14: Make destructors virtual in base classes.
1587 Item 15: Have @code{operator=} return a reference to @code{*this}.
1590 Item 23: Don't try to return a reference when you must return an object.
1594 Also warn about violations of the following style guidelines from
1595 Scott Meyers' @cite{More Effective C++} book:
1599 Item 6: Distinguish between prefix and postfix forms of increment and
1600 decrement operators.
1603 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1607 When selecting this option, be aware that the standard library
1608 headers do not obey all of these guidelines; use @samp{grep -v}
1609 to filter out those warnings.
1611 @item -Wno-deprecated @r{(C++ only)}
1612 @opindex Wno-deprecated
1613 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1615 @item -Wno-non-template-friend @r{(C++ only)}
1616 @opindex Wno-non-template-friend
1617 Disable warnings when non-templatized friend functions are declared
1618 within a template. Since the advent of explicit template specification
1619 support in G++, if the name of the friend is an unqualified-id (i.e.,
1620 @samp{friend foo(int)}), the C++ language specification demands that the
1621 friend declare or define an ordinary, nontemplate function. (Section
1622 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1623 could be interpreted as a particular specialization of a templatized
1624 function. Because this non-conforming behavior is no longer the default
1625 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1626 check existing code for potential trouble spots and is on by default.
1627 This new compiler behavior can be turned off with
1628 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1629 but disables the helpful warning.
1631 @item -Wold-style-cast @r{(C++ only)}
1632 @opindex Wold-style-cast
1633 Warn if an old-style (C-style) cast to a non-void type is used within
1634 a C++ program. The new-style casts (@samp{static_cast},
1635 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1636 unintended effects and much easier to search for.
1638 @item -Woverloaded-virtual @r{(C++ only)}
1639 @opindex Woverloaded-virtual
1640 @cindex overloaded virtual fn, warning
1641 @cindex warning for overloaded virtual fn
1642 Warn when a function declaration hides virtual functions from a
1643 base class. For example, in:
1650 struct B: public A @{
1655 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1663 will fail to compile.
1665 @item -Wno-pmf-conversions @r{(C++ only)}
1666 @opindex Wno-pmf-conversions
1667 Disable the diagnostic for converting a bound pointer to member function
1670 @item -Wsign-promo @r{(C++ only)}
1671 @opindex Wsign-promo
1672 Warn when overload resolution chooses a promotion from unsigned or
1673 enumeral type to a signed type, over a conversion to an unsigned type of
1674 the same size. Previous versions of G++ would try to preserve
1675 unsignedness, but the standard mandates the current behavior.
1677 @item -Wsynth @r{(C++ only)}
1679 @cindex warning for synthesized methods
1680 @cindex synthesized methods, warning
1681 Warn when G++'s synthesis behavior does not match that of cfront. For
1687 A& operator = (int);
1697 In this example, G++ will synthesize a default @samp{A& operator =
1698 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1701 @node Objective-C Dialect Options
1702 @section Options Controlling Objective-C Dialect
1704 @cindex compiler options, Objective-C
1705 @cindex Objective-C options, command line
1706 @cindex options, Objective-C
1707 (NOTE: This manual does not describe the Objective-C language itself. See
1708 @w{@uref{http://gcc.gnu.org/readings.html}} for references.)
1710 This section describes the command-line options that are only meaningful
1711 for Objective-C programs, but you can also use most of the GNU compiler
1712 options regardless of what language your program is in. For example,
1713 you might compile a file @code{some_class.m} like this:
1716 gcc -g -fgnu-runtime -O -c some_class.m
1720 In this example, @option{-fgnu-runtime} is an option meant only for
1721 Objective-C programs; you can use the other options with any language
1724 Here is a list of options that are @emph{only} for compiling Objective-C
1728 @item -fconstant-string-class=@var{class-name}
1729 @opindex fconstant-string-class
1730 Use @var{class-name} as the name of the class to instantiate for each
1731 literal string specified with the syntax @code{@@"@dots{}"}. The default
1732 class name is @code{NXConstantString} if the GNU runtime is being used, and
1733 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1734 @option{-fconstant-cfstrings} option, if also present, will override the
1735 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1736 to be laid out as constant CoreFoundation strings.
1739 @opindex fgnu-runtime
1740 Generate object code compatible with the standard GNU Objective-C
1741 runtime. This is the default for most types of systems.
1743 @item -fnext-runtime
1744 @opindex fnext-runtime
1745 Generate output compatible with the NeXT runtime. This is the default
1746 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1747 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1750 @item -fno-nil-receivers
1751 @opindex -fno-nil-receivers
1752 Assume that all Objective-C message dispatches (e.g.,
1753 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1754 is not @code{nil}. This allows for more efficient entry points in the runtime to be
1755 used. Currently, this option is only available in conjunction with
1756 the NeXT runtime on Mac OS X 10.3 and later.
1758 @item -fobjc-exceptions
1759 @opindex -fobjc-exceptions
1760 Enable syntactic support for structured exception handling in Objective-C,
1761 similar to what is offered by C++ and Java. Currently, this option is only
1762 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1770 @@catch (AnObjCClass *exc) @{
1777 @@catch (AnotherClass *exc) @{
1780 @@catch (id allOthers) @{
1790 The @code{@@throw} statement may appear anywhere in an Objective-C or
1791 Objective-C++ program; when used inside of a @code{@@catch} block, the
1792 @code{@@throw} may appear without an argument (as shown above), in which case
1793 the object caught by the @code{@@catch} will be rethrown.
1795 Note that only (pointers to) Objective-C objects may be thrown and
1796 caught using this scheme. When an object is thrown, it will be caught
1797 by the nearest @code{@@catch} clause capable of handling objects of that type,
1798 analogously to how @code{catch} blocks work in C++ and Java. A
1799 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1800 any and all Objective-C exceptions not caught by previous @code{@@catch}
1803 The @code{@@finally} clause, if present, will be executed upon exit from the
1804 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1805 regardless of whether any exceptions are thrown, caught or rethrown
1806 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1807 of the @code{finally} clause in Java.
1809 There are several caveats to using the new exception mechanism:
1813 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1814 idioms provided by the @code{NSException} class, the new
1815 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1816 systems, due to additional functionality needed in the (NeXT) Objective-C
1820 As mentioned above, the new exceptions do not support handling
1821 types other than Objective-C objects. Furthermore, when used from
1822 Objective-C++, the Objective-C exception model does not interoperate with C++
1823 exceptions at this time. This means you cannot @code{@@throw} an exception
1824 from Objective-C and @code{catch} it in C++, or vice versa
1825 (i.e., @code{throw @dots{} @@catch}).
1828 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1829 blocks for thread-safe execution:
1832 @@synchronized (ObjCClass *guard) @{
1837 Upon entering the @code{@@synchronized} block, a thread of execution shall
1838 first check whether a lock has been placed on the corresponding @code{guard}
1839 object by another thread. If it has, the current thread shall wait until
1840 the other thread relinquishes its lock. Once @code{guard} becomes available,
1841 the current thread will place its own lock on it, execute the code contained in
1842 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1843 making @code{guard} available to other threads).
1845 Unlike Java, Objective-C does not allow for entire methods to be marked
1846 @code{@@synchronized}. Note that throwing exceptions out of
1847 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1848 to be unlocked properly.
1850 @item -freplace-objc-classes
1851 @opindex -freplace-objc-classes
1852 Emit a special marker instructing @command{ld(1)} not to statically link in
1853 the resulting object file, and allow @command{dyld(1)} to load it in at
1854 run time instead. This is used in conjunction with the Fix-and-Continue
1855 debugging mode, where the object file in question may be recompiled and
1856 dynamically reloaded in the course of program execution, without the need
1857 to restart the program itself. Currently, Fix-and-Continue functionality
1858 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
1862 @opindex -fzero-link
1863 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1864 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1865 compile time) with static class references that get initialized at load time,
1866 which improves run-time performance. Specifying the @option{-fzero-link} flag
1867 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1868 to be retained. This is useful in Zero-Link debugging mode, since it allows
1869 for individual class implementations to be modified during program execution.
1873 Dump interface declarations for all classes seen in the source file to a
1874 file named @file{@var{sourcename}.decl}.
1877 @opindex Wno-protocol
1878 If a class is declared to implement a protocol, a warning is issued for
1879 every method in the protocol that is not implemented by the class. The
1880 default behavior is to issue a warning for every method not explicitly
1881 implemented in the class, even if a method implementation is inherited
1882 from the superclass. If you use the @code{-Wno-protocol} option, then
1883 methods inherited from the superclass are considered to be implemented,
1884 and no warning is issued for them.
1888 Warn if multiple methods of different types for the same selector are
1889 found during compilation. The check is performed on the list of methods
1890 in the final stage of compilation. Additionally, a check is performed
1891 for each selector appearing in a @code{@@selector(@dots{})}
1892 expression, and a corresponding method for that selector has been found
1893 during compilation. Because these checks scan the method table only at
1894 the end of compilation, these warnings are not produced if the final
1895 stage of compilation is not reached, for example because an error is
1896 found during compilation, or because the @code{-fsyntax-only} option is
1899 @item -Wundeclared-selector
1900 @opindex Wundeclared-selector
1901 Warn if a @code{@@selector(@dots{})} expression referring to an
1902 undeclared selector is found. A selector is considered undeclared if no
1903 method with that name has been declared before the
1904 @code{@@selector(@dots{})} expression, either explicitly in an
1905 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
1906 an @code{@@implementation} section. This option always performs its
1907 checks as soon as a @code{@@selector(@dots{})} expression is found,
1908 while @code{-Wselector} only performs its checks in the final stage of
1909 compilation. This also enforces the coding style convention
1910 that methods and selectors must be declared before being used.
1912 @item -print-objc-runtime-info
1913 @opindex -print-objc-runtime-info
1914 Generate C header describing the largest structure that is passed by
1919 @node Language Independent Options
1920 @section Options to Control Diagnostic Messages Formatting
1921 @cindex options to control diagnostics formatting
1922 @cindex diagnostic messages
1923 @cindex message formatting
1925 Traditionally, diagnostic messages have been formatted irrespective of
1926 the output device's aspect (e.g.@: its width, @dots{}). The options described
1927 below can be used to control the diagnostic messages formatting
1928 algorithm, e.g.@: how many characters per line, how often source location
1929 information should be reported. Right now, only the C++ front end can
1930 honor these options. However it is expected, in the near future, that
1931 the remaining front ends would be able to digest them correctly.
1934 @item -fmessage-length=@var{n}
1935 @opindex fmessage-length
1936 Try to format error messages so that they fit on lines of about @var{n}
1937 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1938 the front ends supported by GCC@. If @var{n} is zero, then no
1939 line-wrapping will be done; each error message will appear on a single
1942 @opindex fdiagnostics-show-location
1943 @item -fdiagnostics-show-location=once
1944 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1945 reporter to emit @emph{once} source location information; that is, in
1946 case the message is too long to fit on a single physical line and has to
1947 be wrapped, the source location won't be emitted (as prefix) again,
1948 over and over, in subsequent continuation lines. This is the default
1951 @item -fdiagnostics-show-location=every-line
1952 Only meaningful in line-wrapping mode. Instructs the diagnostic
1953 messages reporter to emit the same source location information (as
1954 prefix) for physical lines that result from the process of breaking
1955 a message which is too long to fit on a single line.
1959 @node Warning Options
1960 @section Options to Request or Suppress Warnings
1961 @cindex options to control warnings
1962 @cindex warning messages
1963 @cindex messages, warning
1964 @cindex suppressing warnings
1966 Warnings are diagnostic messages that report constructions which
1967 are not inherently erroneous but which are risky or suggest there
1968 may have been an error.
1970 You can request many specific warnings with options beginning @samp{-W},
1971 for example @option{-Wimplicit} to request warnings on implicit
1972 declarations. Each of these specific warning options also has a
1973 negative form beginning @samp{-Wno-} to turn off warnings;
1974 for example, @option{-Wno-implicit}. This manual lists only one of the
1975 two forms, whichever is not the default.
1977 The following options control the amount and kinds of warnings produced
1978 by GCC; for further, language-specific options also refer to
1979 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
1982 @cindex syntax checking
1984 @opindex fsyntax-only
1985 Check the code for syntax errors, but don't do anything beyond that.
1989 Issue all the warnings demanded by strict ISO C and ISO C++;
1990 reject all programs that use forbidden extensions, and some other
1991 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1992 version of the ISO C standard specified by any @option{-std} option used.
1994 Valid ISO C and ISO C++ programs should compile properly with or without
1995 this option (though a rare few will require @option{-ansi} or a
1996 @option{-std} option specifying the required version of ISO C)@. However,
1997 without this option, certain GNU extensions and traditional C and C++
1998 features are supported as well. With this option, they are rejected.
2000 @option{-pedantic} does not cause warning messages for use of the
2001 alternate keywords whose names begin and end with @samp{__}. Pedantic
2002 warnings are also disabled in the expression that follows
2003 @code{__extension__}. However, only system header files should use
2004 these escape routes; application programs should avoid them.
2005 @xref{Alternate Keywords}.
2007 Some users try to use @option{-pedantic} to check programs for strict ISO
2008 C conformance. They soon find that it does not do quite what they want:
2009 it finds some non-ISO practices, but not all---only those for which
2010 ISO C @emph{requires} a diagnostic, and some others for which
2011 diagnostics have been added.
2013 A feature to report any failure to conform to ISO C might be useful in
2014 some instances, but would require considerable additional work and would
2015 be quite different from @option{-pedantic}. We don't have plans to
2016 support such a feature in the near future.
2018 Where the standard specified with @option{-std} represents a GNU
2019 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2020 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2021 extended dialect is based. Warnings from @option{-pedantic} are given
2022 where they are required by the base standard. (It would not make sense
2023 for such warnings to be given only for features not in the specified GNU
2024 C dialect, since by definition the GNU dialects of C include all
2025 features the compiler supports with the given option, and there would be
2026 nothing to warn about.)
2028 @item -pedantic-errors
2029 @opindex pedantic-errors
2030 Like @option{-pedantic}, except that errors are produced rather than
2035 Inhibit all warning messages.
2039 Inhibit warning messages about the use of @samp{#import}.
2041 @item -Wchar-subscripts
2042 @opindex Wchar-subscripts
2043 Warn if an array subscript has type @code{char}. This is a common cause
2044 of error, as programmers often forget that this type is signed on some
2049 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2050 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2054 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2055 the arguments supplied have types appropriate to the format string
2056 specified, and that the conversions specified in the format string make
2057 sense. This includes standard functions, and others specified by format
2058 attributes (@pxref{Function Attributes}), in the @code{printf},
2059 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2060 not in the C standard) families.
2062 The formats are checked against the format features supported by GNU
2063 libc version 2.2. These include all ISO C90 and C99 features, as well
2064 as features from the Single Unix Specification and some BSD and GNU
2065 extensions. Other library implementations may not support all these
2066 features; GCC does not support warning about features that go beyond a
2067 particular library's limitations. However, if @option{-pedantic} is used
2068 with @option{-Wformat}, warnings will be given about format features not
2069 in the selected standard version (but not for @code{strfmon} formats,
2070 since those are not in any version of the C standard). @xref{C Dialect
2071 Options,,Options Controlling C Dialect}.
2073 Since @option{-Wformat} also checks for null format arguments for
2074 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2076 @option{-Wformat} is included in @option{-Wall}. For more control over some
2077 aspects of format checking, the options @option{-Wformat-y2k},
2078 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2079 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2080 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2083 @opindex Wformat-y2k
2084 If @option{-Wformat} is specified, also warn about @code{strftime}
2085 formats which may yield only a two-digit year.
2087 @item -Wno-format-extra-args
2088 @opindex Wno-format-extra-args
2089 If @option{-Wformat} is specified, do not warn about excess arguments to a
2090 @code{printf} or @code{scanf} format function. The C standard specifies
2091 that such arguments are ignored.
2093 Where the unused arguments lie between used arguments that are
2094 specified with @samp{$} operand number specifications, normally
2095 warnings are still given, since the implementation could not know what
2096 type to pass to @code{va_arg} to skip the unused arguments. However,
2097 in the case of @code{scanf} formats, this option will suppress the
2098 warning if the unused arguments are all pointers, since the Single
2099 Unix Specification says that such unused arguments are allowed.
2101 @item -Wno-format-zero-length
2102 @opindex Wno-format-zero-length
2103 If @option{-Wformat} is specified, do not warn about zero-length formats.
2104 The C standard specifies that zero-length formats are allowed.
2106 @item -Wformat-nonliteral
2107 @opindex Wformat-nonliteral
2108 If @option{-Wformat} is specified, also warn if the format string is not a
2109 string literal and so cannot be checked, unless the format function
2110 takes its format arguments as a @code{va_list}.
2112 @item -Wformat-security
2113 @opindex Wformat-security
2114 If @option{-Wformat} is specified, also warn about uses of format
2115 functions that represent possible security problems. At present, this
2116 warns about calls to @code{printf} and @code{scanf} functions where the
2117 format string is not a string literal and there are no format arguments,
2118 as in @code{printf (foo);}. This may be a security hole if the format
2119 string came from untrusted input and contains @samp{%n}. (This is
2120 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2121 in future warnings may be added to @option{-Wformat-security} that are not
2122 included in @option{-Wformat-nonliteral}.)
2126 Enable @option{-Wformat} plus format checks not included in
2127 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2128 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2132 Warn about passing a null pointer for arguments marked as
2133 requiring a non-null value by the @code{nonnull} function attribute.
2135 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2136 can be disabled with the @option{-Wno-nonnull} option.
2138 @item -Winit-self @r{(C, C++, and Objective-C only)}
2140 Warn about uninitialized variables which are initialized with themselves.
2141 Note this option can only be used with the @option{-Wuninitialized} option,
2142 which in turn only works with @option{-O1} and above.
2144 For example, GCC will warn about @code{i} being uninitialized in the
2145 following snippet only when @option{-Winit-self} has been specified:
2156 @item -Wimplicit-int
2157 @opindex Wimplicit-int
2158 Warn when a declaration does not specify a type.
2160 @item -Wimplicit-function-declaration
2161 @itemx -Werror-implicit-function-declaration
2162 @opindex Wimplicit-function-declaration
2163 @opindex Werror-implicit-function-declaration
2164 Give a warning (or error) whenever a function is used before being
2169 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2173 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2174 function with external linkage, returning int, taking either zero
2175 arguments, two, or three arguments of appropriate types.
2177 @item -Wmissing-braces
2178 @opindex Wmissing-braces
2179 Warn if an aggregate or union initializer is not fully bracketed. In
2180 the following example, the initializer for @samp{a} is not fully
2181 bracketed, but that for @samp{b} is fully bracketed.
2184 int a[2][2] = @{ 0, 1, 2, 3 @};
2185 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2189 @opindex Wparentheses
2190 Warn if parentheses are omitted in certain contexts, such
2191 as when there is an assignment in a context where a truth value
2192 is expected, or when operators are nested whose precedence people
2193 often get confused about.
2195 Also warn about constructions where there may be confusion to which
2196 @code{if} statement an @code{else} branch belongs. Here is an example of
2211 In C, every @code{else} branch belongs to the innermost possible @code{if}
2212 statement, which in this example is @code{if (b)}. This is often not
2213 what the programmer expected, as illustrated in the above example by
2214 indentation the programmer chose. When there is the potential for this
2215 confusion, GCC will issue a warning when this flag is specified.
2216 To eliminate the warning, add explicit braces around the innermost
2217 @code{if} statement so there is no way the @code{else} could belong to
2218 the enclosing @code{if}. The resulting code would look like this:
2234 @item -Wsequence-point
2235 @opindex Wsequence-point
2236 Warn about code that may have undefined semantics because of violations
2237 of sequence point rules in the C standard.
2239 The C standard defines the order in which expressions in a C program are
2240 evaluated in terms of @dfn{sequence points}, which represent a partial
2241 ordering between the execution of parts of the program: those executed
2242 before the sequence point, and those executed after it. These occur
2243 after the evaluation of a full expression (one which is not part of a
2244 larger expression), after the evaluation of the first operand of a
2245 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2246 function is called (but after the evaluation of its arguments and the
2247 expression denoting the called function), and in certain other places.
2248 Other than as expressed by the sequence point rules, the order of
2249 evaluation of subexpressions of an expression is not specified. All
2250 these rules describe only a partial order rather than a total order,
2251 since, for example, if two functions are called within one expression
2252 with no sequence point between them, the order in which the functions
2253 are called is not specified. However, the standards committee have
2254 ruled that function calls do not overlap.
2256 It is not specified when between sequence points modifications to the
2257 values of objects take effect. Programs whose behavior depends on this
2258 have undefined behavior; the C standard specifies that ``Between the
2259 previous and next sequence point an object shall have its stored value
2260 modified at most once by the evaluation of an expression. Furthermore,
2261 the prior value shall be read only to determine the value to be
2262 stored.''. If a program breaks these rules, the results on any
2263 particular implementation are entirely unpredictable.
2265 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2266 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2267 diagnosed by this option, and it may give an occasional false positive
2268 result, but in general it has been found fairly effective at detecting
2269 this sort of problem in programs.
2271 The present implementation of this option only works for C programs. A
2272 future implementation may also work for C++ programs.
2274 The C standard is worded confusingly, therefore there is some debate
2275 over the precise meaning of the sequence point rules in subtle cases.
2276 Links to discussions of the problem, including proposed formal
2277 definitions, may be found on our readings page, at
2278 @w{@uref{http://gcc.gnu.org/readings.html}}.
2281 @opindex Wreturn-type
2282 Warn whenever a function is defined with a return-type that defaults to
2283 @code{int}. Also warn about any @code{return} statement with no
2284 return-value in a function whose return-type is not @code{void}.
2286 For C++, a function without return type always produces a diagnostic
2287 message, even when @option{-Wno-return-type} is specified. The only
2288 exceptions are @samp{main} and functions defined in system headers.
2292 Warn whenever a @code{switch} statement has an index of enumeral type
2293 and lacks a @code{case} for one or more of the named codes of that
2294 enumeration. (The presence of a @code{default} label prevents this
2295 warning.) @code{case} labels outside the enumeration range also
2296 provoke warnings when this option is used.
2298 @item -Wswitch-default
2299 @opindex Wswitch-switch
2300 Warn whenever a @code{switch} statement does not have a @code{default}
2304 @opindex Wswitch-enum
2305 Warn whenever a @code{switch} statement has an index of enumeral type
2306 and lacks a @code{case} for one or more of the named codes of that
2307 enumeration. @code{case} labels outside the enumeration range also
2308 provoke warnings when this option is used.
2312 Warn if any trigraphs are encountered that might change the meaning of
2313 the program (trigraphs within comments are not warned about).
2315 @item -Wunused-function
2316 @opindex Wunused-function
2317 Warn whenever a static function is declared but not defined or a
2318 non\-inline static function is unused.
2320 @item -Wunused-label
2321 @opindex Wunused-label
2322 Warn whenever a label is declared but not used.
2324 To suppress this warning use the @samp{unused} attribute
2325 (@pxref{Variable Attributes}).
2327 @item -Wunused-parameter
2328 @opindex Wunused-parameter
2329 Warn whenever a function parameter is unused aside from its declaration.
2331 To suppress this warning use the @samp{unused} attribute
2332 (@pxref{Variable Attributes}).
2334 @item -Wunused-variable
2335 @opindex Wunused-variable
2336 Warn whenever a local variable or non-constant static variable is unused
2337 aside from its declaration
2339 To suppress this warning use the @samp{unused} attribute
2340 (@pxref{Variable Attributes}).
2342 @item -Wunused-value
2343 @opindex Wunused-value
2344 Warn whenever a statement computes a result that is explicitly not used.
2346 To suppress this warning cast the expression to @samp{void}.
2350 All the above @option{-Wunused} options combined.
2352 In order to get a warning about an unused function parameter, you must
2353 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2354 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2356 @item -Wuninitialized
2357 @opindex Wuninitialized
2358 Warn if an automatic variable is used without first being initialized or
2359 if a variable may be clobbered by a @code{setjmp} call.
2361 These warnings are possible only in optimizing compilation,
2362 because they require data flow information that is computed only
2363 when optimizing. If you don't specify @option{-O}, you simply won't
2366 If you want to warn about code which uses the uninitialized value of the
2367 variable in its own initializer, use the @option{-Winit-self} option.
2369 These warnings occur only for variables that are candidates for
2370 register allocation. Therefore, they do not occur for a variable that
2371 is declared @code{volatile}, or whose address is taken, or whose size
2372 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2373 structures, unions or arrays, even when they are in registers.
2375 Note that there may be no warning about a variable that is used only
2376 to compute a value that itself is never used, because such
2377 computations may be deleted by data flow analysis before the warnings
2380 These warnings are made optional because GCC is not smart
2381 enough to see all the reasons why the code might be correct
2382 despite appearing to have an error. Here is one example of how
2403 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2404 always initialized, but GCC doesn't know this. Here is
2405 another common case:
2410 if (change_y) save_y = y, y = new_y;
2412 if (change_y) y = save_y;
2417 This has no bug because @code{save_y} is used only if it is set.
2419 @cindex @code{longjmp} warnings
2420 This option also warns when a non-volatile automatic variable might be
2421 changed by a call to @code{longjmp}. These warnings as well are possible
2422 only in optimizing compilation.
2424 The compiler sees only the calls to @code{setjmp}. It cannot know
2425 where @code{longjmp} will be called; in fact, a signal handler could
2426 call it at any point in the code. As a result, you may get a warning
2427 even when there is in fact no problem because @code{longjmp} cannot
2428 in fact be called at the place which would cause a problem.
2430 Some spurious warnings can be avoided if you declare all the functions
2431 you use that never return as @code{noreturn}. @xref{Function
2434 @item -Wunknown-pragmas
2435 @opindex Wunknown-pragmas
2436 @cindex warning for unknown pragmas
2437 @cindex unknown pragmas, warning
2438 @cindex pragmas, warning of unknown
2439 Warn when a #pragma directive is encountered which is not understood by
2440 GCC@. If this command line option is used, warnings will even be issued
2441 for unknown pragmas in system header files. This is not the case if
2442 the warnings were only enabled by the @option{-Wall} command line option.
2444 @item -Wstrict-aliasing
2445 @opindex Wstrict-aliasing
2446 This option is only active when @option{-fstrict-aliasing} is active.
2447 It warns about code which might break the strict aliasing rules that the
2448 compiler is using for optimization. The warning does not catch all
2449 cases, but does attempt to catch the more common pitfalls. It is
2450 included in @option{-Wall}.
2452 @item -Wstrict-aliasing=2
2453 @opindex Wstrict-aliasing=2
2454 This option is only active when @option{-fstrict-aliasing} is active.
2455 It warns about all code which might break the strict aliasing rules that the
2456 compiler is using for optimization. This warning catches all cases, but
2457 it will also give a warning for some ambiguous cases that are safe.
2461 All of the above @samp{-W} options combined. This enables all the
2462 warnings about constructions that some users consider questionable, and
2463 that are easy to avoid (or modify to prevent the warning), even in
2464 conjunction with macros. This also enables some language-specific
2465 warnings described in @ref{C++ Dialect Options} and
2466 @ref{Objective-C Dialect Options}.
2469 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2470 Some of them warn about constructions that users generally do not
2471 consider questionable, but which occasionally you might wish to check
2472 for; others warn about constructions that are necessary or hard to avoid
2473 in some cases, and there is no simple way to modify the code to suppress
2480 (This option used to be called @option{-W}. The older name is still
2481 supported, but the newer name is more descriptive.) Print extra warning
2482 messages for these events:
2486 A function can return either with or without a value. (Falling
2487 off the end of the function body is considered returning without
2488 a value.) For example, this function would evoke such a
2502 An expression-statement or the left-hand side of a comma expression
2503 contains no side effects.
2504 To suppress the warning, cast the unused expression to void.
2505 For example, an expression such as @samp{x[i,j]} will cause a warning,
2506 but @samp{x[(void)i,j]} will not.
2509 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2512 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2513 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2514 that of ordinary mathematical notation.
2517 Storage-class specifiers like @code{static} are not the first things in
2518 a declaration. According to the C Standard, this usage is obsolescent.
2521 The return type of a function has a type qualifier such as @code{const}.
2522 Such a type qualifier has no effect, since the value returned by a
2523 function is not an lvalue. (But don't warn about the GNU extension of
2524 @code{volatile void} return types. That extension will be warned about
2525 if @option{-pedantic} is specified.)
2528 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2532 A comparison between signed and unsigned values could produce an
2533 incorrect result when the signed value is converted to unsigned.
2534 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2537 An aggregate has an initializer which does not initialize all members.
2538 For example, the following code would cause such a warning, because
2539 @code{x.h} would be implicitly initialized to zero:
2542 struct s @{ int f, g, h; @};
2543 struct s x = @{ 3, 4 @};
2547 A function parameter is declared without a type specifier in K&R-style
2555 An empty body occurs in an @samp{if} or @samp{else} statement.
2558 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2559 @samp{>}, or @samp{>=}.
2562 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2565 Any of several floating-point events that often indicate errors, such as
2566 overflow, underflow, loss of precision, etc.
2568 @item @r{(C++ only)}
2569 An enumerator and a non-enumerator both appear in a conditional expression.
2571 @item @r{(C++ only)}
2572 A non-static reference or non-static @samp{const} member appears in a
2573 class without constructors.
2575 @item @r{(C++ only)}
2576 Ambiguous virtual bases.
2578 @item @r{(C++ only)}
2579 Subscripting an array which has been declared @samp{register}.
2581 @item @r{(C++ only)}
2582 Taking the address of a variable which has been declared @samp{register}.
2584 @item @r{(C++ only)}
2585 A base class is not initialized in a derived class' copy constructor.
2588 @item -Wno-div-by-zero
2589 @opindex Wno-div-by-zero
2590 @opindex Wdiv-by-zero
2591 Do not warn about compile-time integer division by zero. Floating point
2592 division by zero is not warned about, as it can be a legitimate way of
2593 obtaining infinities and NaNs.
2595 @item -Wsystem-headers
2596 @opindex Wsystem-headers
2597 @cindex warnings from system headers
2598 @cindex system headers, warnings from
2599 Print warning messages for constructs found in system header files.
2600 Warnings from system headers are normally suppressed, on the assumption
2601 that they usually do not indicate real problems and would only make the
2602 compiler output harder to read. Using this command line option tells
2603 GCC to emit warnings from system headers as if they occurred in user
2604 code. However, note that using @option{-Wall} in conjunction with this
2605 option will @emph{not} warn about unknown pragmas in system
2606 headers---for that, @option{-Wunknown-pragmas} must also be used.
2609 @opindex Wfloat-equal
2610 Warn if floating point values are used in equality comparisons.
2612 The idea behind this is that sometimes it is convenient (for the
2613 programmer) to consider floating-point values as approximations to
2614 infinitely precise real numbers. If you are doing this, then you need
2615 to compute (by analyzing the code, or in some other way) the maximum or
2616 likely maximum error that the computation introduces, and allow for it
2617 when performing comparisons (and when producing output, but that's a
2618 different problem). In particular, instead of testing for equality, you
2619 would check to see whether the two values have ranges that overlap; and
2620 this is done with the relational operators, so equality comparisons are
2623 @item -Wtraditional @r{(C only)}
2624 @opindex Wtraditional
2625 Warn about certain constructs that behave differently in traditional and
2626 ISO C@. Also warn about ISO C constructs that have no traditional C
2627 equivalent, and/or problematic constructs which should be avoided.
2631 Macro parameters that appear within string literals in the macro body.
2632 In traditional C macro replacement takes place within string literals,
2633 but does not in ISO C@.
2636 In traditional C, some preprocessor directives did not exist.
2637 Traditional preprocessors would only consider a line to be a directive
2638 if the @samp{#} appeared in column 1 on the line. Therefore
2639 @option{-Wtraditional} warns about directives that traditional C
2640 understands but would ignore because the @samp{#} does not appear as the
2641 first character on the line. It also suggests you hide directives like
2642 @samp{#pragma} not understood by traditional C by indenting them. Some
2643 traditional implementations would not recognize @samp{#elif}, so it
2644 suggests avoiding it altogether.
2647 A function-like macro that appears without arguments.
2650 The unary plus operator.
2653 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2654 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2655 constants.) Note, these suffixes appear in macros defined in the system
2656 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2657 Use of these macros in user code might normally lead to spurious
2658 warnings, however GCC's integrated preprocessor has enough context to
2659 avoid warning in these cases.
2662 A function declared external in one block and then used after the end of
2666 A @code{switch} statement has an operand of type @code{long}.
2669 A non-@code{static} function declaration follows a @code{static} one.
2670 This construct is not accepted by some traditional C compilers.
2673 The ISO type of an integer constant has a different width or
2674 signedness from its traditional type. This warning is only issued if
2675 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2676 typically represent bit patterns, are not warned about.
2679 Usage of ISO string concatenation is detected.
2682 Initialization of automatic aggregates.
2685 Identifier conflicts with labels. Traditional C lacks a separate
2686 namespace for labels.
2689 Initialization of unions. If the initializer is zero, the warning is
2690 omitted. This is done under the assumption that the zero initializer in
2691 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2692 initializer warnings and relies on default initialization to zero in the
2696 Conversions by prototypes between fixed/floating point values and vice
2697 versa. The absence of these prototypes when compiling with traditional
2698 C would cause serious problems. This is a subset of the possible
2699 conversion warnings, for the full set use @option{-Wconversion}.
2702 Use of ISO C style function definitions. This warning intentionally is
2703 @emph{not} issued for prototype declarations or variadic functions
2704 because these ISO C features will appear in your code when using
2705 libiberty's traditional C compatibility macros, @code{PARAMS} and
2706 @code{VPARAMS}. This warning is also bypassed for nested functions
2707 because that feature is already a GCC extension and thus not relevant to
2708 traditional C compatibility.
2711 @item -Wdeclaration-after-statement @r{(C only)}
2712 @opindex Wdeclaration-after-statement
2713 Warn when a declaration is found after a statement in a block. This
2714 construct, known from C++, was introduced with ISO C99 and is by default
2715 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2716 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2720 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2722 @item -Wendif-labels
2723 @opindex Wendif-labels
2724 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2728 Warn whenever a local variable shadows another local variable, parameter or
2729 global variable or whenever a built-in function is shadowed.
2731 @item -Wlarger-than-@var{len}
2732 @opindex Wlarger-than
2733 Warn whenever an object of larger than @var{len} bytes is defined.
2735 @item -Wpointer-arith
2736 @opindex Wpointer-arith
2737 Warn about anything that depends on the ``size of'' a function type or
2738 of @code{void}. GNU C assigns these types a size of 1, for
2739 convenience in calculations with @code{void *} pointers and pointers
2742 @item -Wbad-function-cast @r{(C only)}
2743 @opindex Wbad-function-cast
2744 Warn whenever a function call is cast to a non-matching type.
2745 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2749 Warn whenever a pointer is cast so as to remove a type qualifier from
2750 the target type. For example, warn if a @code{const char *} is cast
2751 to an ordinary @code{char *}.
2754 @opindex Wcast-align
2755 Warn whenever a pointer is cast such that the required alignment of the
2756 target is increased. For example, warn if a @code{char *} is cast to
2757 an @code{int *} on machines where integers can only be accessed at
2758 two- or four-byte boundaries.
2760 @item -Wwrite-strings
2761 @opindex Wwrite-strings
2762 When compiling C, give string constants the type @code{const
2763 char[@var{length}]} so that
2764 copying the address of one into a non-@code{const} @code{char *}
2765 pointer will get a warning; when compiling C++, warn about the
2766 deprecated conversion from string constants to @code{char *}.
2767 These warnings will help you find at
2768 compile time code that can try to write into a string constant, but
2769 only if you have been very careful about using @code{const} in
2770 declarations and prototypes. Otherwise, it will just be a nuisance;
2771 this is why we did not make @option{-Wall} request these warnings.
2774 @opindex Wconversion
2775 Warn if a prototype causes a type conversion that is different from what
2776 would happen to the same argument in the absence of a prototype. This
2777 includes conversions of fixed point to floating and vice versa, and
2778 conversions changing the width or signedness of a fixed point argument
2779 except when the same as the default promotion.
2781 Also, warn if a negative integer constant expression is implicitly
2782 converted to an unsigned type. For example, warn about the assignment
2783 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2784 casts like @code{(unsigned) -1}.
2786 @item -Wsign-compare
2787 @opindex Wsign-compare
2788 @cindex warning for comparison of signed and unsigned values
2789 @cindex comparison of signed and unsigned values, warning
2790 @cindex signed and unsigned values, comparison warning
2791 Warn when a comparison between signed and unsigned values could produce
2792 an incorrect result when the signed value is converted to unsigned.
2793 This warning is also enabled by @option{-Wextra}; to get the other warnings
2794 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2796 @item -Waggregate-return
2797 @opindex Waggregate-return
2798 Warn if any functions that return structures or unions are defined or
2799 called. (In languages where you can return an array, this also elicits
2802 @item -Wstrict-prototypes @r{(C only)}
2803 @opindex Wstrict-prototypes
2804 Warn if a function is declared or defined without specifying the
2805 argument types. (An old-style function definition is permitted without
2806 a warning if preceded by a declaration which specifies the argument
2809 @item -Wold-style-definition @r{(C only)}
2810 @opindex Wold-style-definition
2811 Warn if an old-style function definition is used. A warning is given
2812 even if there is a previous prototype.
2814 @item -Wmissing-prototypes @r{(C only)}
2815 @opindex Wmissing-prototypes
2816 Warn if a global function is defined without a previous prototype
2817 declaration. This warning is issued even if the definition itself
2818 provides a prototype. The aim is to detect global functions that fail
2819 to be declared in header files.
2821 @item -Wmissing-declarations @r{(C only)}
2822 @opindex Wmissing-declarations
2823 Warn if a global function is defined without a previous declaration.
2824 Do so even if the definition itself provides a prototype.
2825 Use this option to detect global functions that are not declared in
2828 @item -Wmissing-noreturn
2829 @opindex Wmissing-noreturn
2830 Warn about functions which might be candidates for attribute @code{noreturn}.
2831 Note these are only possible candidates, not absolute ones. Care should
2832 be taken to manually verify functions actually do not ever return before
2833 adding the @code{noreturn} attribute, otherwise subtle code generation
2834 bugs could be introduced. You will not get a warning for @code{main} in
2835 hosted C environments.
2837 @item -Wmissing-format-attribute
2838 @opindex Wmissing-format-attribute
2840 If @option{-Wformat} is enabled, also warn about functions which might be
2841 candidates for @code{format} attributes. Note these are only possible
2842 candidates, not absolute ones. GCC will guess that @code{format}
2843 attributes might be appropriate for any function that calls a function
2844 like @code{vprintf} or @code{vscanf}, but this might not always be the
2845 case, and some functions for which @code{format} attributes are
2846 appropriate may not be detected. This option has no effect unless
2847 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2849 @item -Wno-multichar
2850 @opindex Wno-multichar
2852 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2853 Usually they indicate a typo in the user's code, as they have
2854 implementation-defined values, and should not be used in portable code.
2856 @item -Wno-deprecated-declarations
2857 @opindex Wno-deprecated-declarations
2858 Do not warn about uses of functions, variables, and types marked as
2859 deprecated by using the @code{deprecated} attribute.
2860 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2861 @pxref{Type Attributes}.)
2865 Warn if a structure is given the packed attribute, but the packed
2866 attribute has no effect on the layout or size of the structure.
2867 Such structures may be mis-aligned for little benefit. For
2868 instance, in this code, the variable @code{f.x} in @code{struct bar}
2869 will be misaligned even though @code{struct bar} does not itself
2870 have the packed attribute:
2877 @} __attribute__((packed));
2887 Warn if padding is included in a structure, either to align an element
2888 of the structure or to align the whole structure. Sometimes when this
2889 happens it is possible to rearrange the fields of the structure to
2890 reduce the padding and so make the structure smaller.
2892 @item -Wredundant-decls
2893 @opindex Wredundant-decls
2894 Warn if anything is declared more than once in the same scope, even in
2895 cases where multiple declaration is valid and changes nothing.
2897 @item -Wnested-externs @r{(C only)}
2898 @opindex Wnested-externs
2899 Warn if an @code{extern} declaration is encountered within a function.
2901 @item -Wunreachable-code
2902 @opindex Wunreachable-code
2903 Warn if the compiler detects that code will never be executed.
2905 This option is intended to warn when the compiler detects that at
2906 least a whole line of source code will never be executed, because
2907 some condition is never satisfied or because it is after a
2908 procedure that never returns.
2910 It is possible for this option to produce a warning even though there
2911 are circumstances under which part of the affected line can be executed,
2912 so care should be taken when removing apparently-unreachable code.
2914 For instance, when a function is inlined, a warning may mean that the
2915 line is unreachable in only one inlined copy of the function.
2917 This option is not made part of @option{-Wall} because in a debugging
2918 version of a program there is often substantial code which checks
2919 correct functioning of the program and is, hopefully, unreachable
2920 because the program does work. Another common use of unreachable
2921 code is to provide behavior which is selectable at compile-time.
2925 Warn if a function can not be inlined and it was declared as inline.
2926 Even with this option, the compiler will not warn about failures to
2927 inline functions declared in system headers.
2929 The compiler uses a variety of heuristics to determine whether or not
2930 to inline a function. For example, the compiler takes into account
2931 the size of the function being inlined and the the amount of inlining
2932 that has already been done in the current function. Therefore,
2933 seemingly insignificant changes in the source program can cause the
2934 warnings produced by @option{-Winline} to appear or disappear.
2936 @item -Wno-invalid-offsetof @r{(C++ only)}
2937 @opindex Wno-invalid-offsetof
2938 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
2939 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
2940 to a non-POD type is undefined. In existing C++ implementations,
2941 however, @samp{offsetof} typically gives meaningful results even when
2942 applied to certain kinds of non-POD types. (Such as a simple
2943 @samp{struct} that fails to be a POD type only by virtue of having a
2944 constructor.) This flag is for users who are aware that they are
2945 writing nonportable code and who have deliberately chosen to ignore the
2948 The restrictions on @samp{offsetof} may be relaxed in a future version
2949 of the C++ standard.
2952 @opindex Winvalid-pch
2953 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
2954 the search path but can't be used.
2958 @opindex Wno-long-long
2959 Warn if @samp{long long} type is used. This is default. To inhibit
2960 the warning messages, use @option{-Wno-long-long}. Flags
2961 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2962 only when @option{-pedantic} flag is used.
2964 @item -Wvariadic-macros
2965 @opindex Wvariadic-macros
2966 @opindex Wno-variadic-macros
2967 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
2968 alternate syntax when in pedantic ISO C99 mode. This is default.
2969 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
2971 @item -Wdisabled-optimization
2972 @opindex Wdisabled-optimization
2973 Warn if a requested optimization pass is disabled. This warning does
2974 not generally indicate that there is anything wrong with your code; it
2975 merely indicates that GCC's optimizers were unable to handle the code
2976 effectively. Often, the problem is that your code is too big or too
2977 complex; GCC will refuse to optimize programs when the optimization
2978 itself is likely to take inordinate amounts of time.
2982 Make all warnings into errors.
2985 @node Debugging Options
2986 @section Options for Debugging Your Program or GCC
2987 @cindex options, debugging
2988 @cindex debugging information options
2990 GCC has various special options that are used for debugging
2991 either your program or GCC:
2996 Produce debugging information in the operating system's native format
2997 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
3000 On most systems that use stabs format, @option{-g} enables use of extra
3001 debugging information that only GDB can use; this extra information
3002 makes debugging work better in GDB but will probably make other debuggers
3004 refuse to read the program. If you want to control for certain whether
3005 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3006 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3008 Unlike most other C compilers, GCC allows you to use @option{-g} with
3009 @option{-O}. The shortcuts taken by optimized code may occasionally
3010 produce surprising results: some variables you declared may not exist
3011 at all; flow of control may briefly move where you did not expect it;
3012 some statements may not be executed because they compute constant
3013 results or their values were already at hand; some statements may
3014 execute in different places because they were moved out of loops.
3016 Nevertheless it proves possible to debug optimized output. This makes
3017 it reasonable to use the optimizer for programs that might have bugs.
3019 The following options are useful when GCC is generated with the
3020 capability for more than one debugging format.
3024 Produce debugging information for use by GDB@. This means to use the
3025 most expressive format available (DWARF 2, stabs, or the native format
3026 if neither of those are supported), including GDB extensions if at all
3031 Produce debugging information in stabs format (if that is supported),
3032 without GDB extensions. This is the format used by DBX on most BSD
3033 systems. On MIPS, Alpha and System V Release 4 systems this option
3034 produces stabs debugging output which is not understood by DBX or SDB@.
3035 On System V Release 4 systems this option requires the GNU assembler.
3037 @item -feliminate-unused-debug-symbols
3038 @opindex feliminate-unused-debug-symbols
3039 Produce debugging information in stabs format (if that is supported),
3040 for only symbols that are actually used.
3044 Produce debugging information in stabs format (if that is supported),
3045 using GNU extensions understood only by the GNU debugger (GDB)@. The
3046 use of these extensions is likely to make other debuggers crash or
3047 refuse to read the program.
3051 Produce debugging information in COFF format (if that is supported).
3052 This is the format used by SDB on most System V systems prior to
3057 Produce debugging information in XCOFF format (if that is supported).
3058 This is the format used by the DBX debugger on IBM RS/6000 systems.
3062 Produce debugging information in XCOFF format (if that is supported),
3063 using GNU extensions understood only by the GNU debugger (GDB)@. The
3064 use of these extensions is likely to make other debuggers crash or
3065 refuse to read the program, and may cause assemblers other than the GNU
3066 assembler (GAS) to fail with an error.
3070 Produce debugging information in DWARF version 2 format (if that is
3071 supported). This is the format used by DBX on IRIX 6.
3075 Produce debugging information in VMS debug format (if that is
3076 supported). This is the format used by DEBUG on VMS systems.
3079 @itemx -ggdb@var{level}
3080 @itemx -gstabs@var{level}
3081 @itemx -gcoff@var{level}
3082 @itemx -gxcoff@var{level}
3083 @itemx -gvms@var{level}
3084 Request debugging information and also use @var{level} to specify how
3085 much information. The default level is 2.
3087 Level 1 produces minimal information, enough for making backtraces in
3088 parts of the program that you don't plan to debug. This includes
3089 descriptions of functions and external variables, but no information
3090 about local variables and no line numbers.
3092 Level 3 includes extra information, such as all the macro definitions
3093 present in the program. Some debuggers support macro expansion when
3094 you use @option{-g3}.
3096 Note that in order to avoid confusion between DWARF1 debug level 2,
3097 and DWARF2 @option{-gdwarf-2} does not accept a concatenated debug
3098 level. Instead use an additional @option{-g@var{level}} option to
3099 change the debug level for DWARF2.
3101 @item -feliminate-dwarf2-dups
3102 @opindex feliminate-dwarf2-dups
3103 Compress DWARF2 debugging information by eliminating duplicated
3104 information about each symbol. This option only makes sense when
3105 generating DWARF2 debugging information with @option{-gdwarf-2}.
3107 @cindex @command{prof}
3110 Generate extra code to write profile information suitable for the
3111 analysis program @command{prof}. You must use this option when compiling
3112 the source files you want data about, and you must also use it when
3115 @cindex @command{gprof}
3118 Generate extra code to write profile information suitable for the
3119 analysis program @command{gprof}. You must use this option when compiling
3120 the source files you want data about, and you must also use it when
3125 Makes the compiler print out each function name as it is compiled, and
3126 print some statistics about each pass when it finishes.
3129 @opindex ftime-report
3130 Makes the compiler print some statistics about the time consumed by each
3131 pass when it finishes.
3134 @opindex fmem-report
3135 Makes the compiler print some statistics about permanent memory
3136 allocation when it finishes.
3138 @item -fprofile-arcs
3139 @opindex fprofile-arcs
3140 Add code so that program flow @dfn{arcs} are instrumented. During
3141 execution the program records how many times each branch and call is
3142 executed and how many times it is taken or returns. When the compiled
3143 program exits it saves this data to a file called
3144 @file{@var{auxname}.gcda} for each source file. The data may be used for
3145 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3146 test coverage analysis (@option{-ftest-coverage}). Each object file's
3147 @var{auxname} is generated from the name of the output file, if
3148 explicitly specified and it is not the final executable, otherwise it is
3149 the basename of the source file. In both cases any suffix is removed
3150 (e.g. @file{foo.gcda} for input file @file{dir/foo.c}, or
3151 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3156 Compile the source files with @option{-fprofile-arcs} plus optimization
3157 and code generation options. For test coverage analysis, use the
3158 additional @option{-ftest-coverage} option. You do not need to profile
3159 every source file in a program.
3162 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3163 (the latter implies the former).
3166 Run the program on a representative workload to generate the arc profile
3167 information. This may be repeated any number of times. You can run
3168 concurrent instances of your program, and provided that the file system
3169 supports locking, the data files will be correctly updated. Also
3170 @code{fork} calls are detected and correctly handled (double counting
3174 For profile-directed optimizations, compile the source files again with
3175 the same optimization and code generation options plus
3176 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3177 Control Optimization}).
3180 For test coverage analysis, use @command{gcov} to produce human readable
3181 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3182 @command{gcov} documentation for further information.
3186 With @option{-fprofile-arcs}, for each function of your program GCC
3187 creates a program flow graph, then finds a spanning tree for the graph.
3188 Only arcs that are not on the spanning tree have to be instrumented: the
3189 compiler adds code to count the number of times that these arcs are
3190 executed. When an arc is the only exit or only entrance to a block, the
3191 instrumentation code can be added to the block; otherwise, a new basic
3192 block must be created to hold the instrumentation code.
3195 @item -ftest-coverage
3196 @opindex ftest-coverage
3197 Produce a notes file that the @command{gcov} code-coverage utility
3198 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3199 show program coverage. Each source file's note file is called
3200 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3201 above for a description of @var{auxname} and instructions on how to
3202 generate test coverage data. Coverage data will match the source files
3203 more closely, if you do not optimize.
3205 @item -d@var{letters}
3207 Says to make debugging dumps during compilation at times specified by
3208 @var{letters}. This is used for debugging the compiler. The file names
3209 for most of the dumps are made by appending a pass number and a word to
3210 the @var{dumpname}. @var{dumpname} is generated from the name of the
3211 output file, if explicitly specified and it is not an executable,
3212 otherwise it is the basename of the source file. In both cases any
3213 suffix is removed (e.g. @file{foo.01.rtl} or @file{foo.02.sibling}).
3214 Here are the possible letters for use in @var{letters}, and their
3220 Annotate the assembler output with miscellaneous debugging information.
3223 Dump after computing branch probabilities, to @file{@var{file}.12.bp}.
3226 Dump after block reordering, to @file{@var{file}.31.bbro}.
3229 Dump after instruction combination, to the file @file{@var{file}.20.combine}.
3232 Dump after the first if conversion, to the file @file{@var{file}.14.ce1}.
3233 Also dump after the second if conversion, to the file @file{@var{file}.21.ce2}.
3236 Dump after branch target load optimization, to to @file{@var{file}.32.btl}.
3237 Also dump after delayed branch scheduling, to @file{@var{file}.36.dbr}.
3240 Dump all macro definitions, at the end of preprocessing, in addition to
3244 Dump after the third if conversion, to @file{@var{file}.30.ce3}.
3247 Dump after control and data flow analysis, to @file{@var{file}.11.cfg}.
3248 Also dump after life analysis, to @file{@var{file}.19.life}.
3251 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.07.addressof}.
3254 Dump after global register allocation, to @file{@var{file}.25.greg}.
3257 Dump after GCSE, to @file{@var{file}.08.gcse}.
3258 Also dump after jump bypassing and control flow optimizations, to
3259 @file{@var{file}.10.bypass}.
3262 Dump after finalization of EH handling code, to @file{@var{file}.03.eh}.
3265 Dump after sibling call optimizations, to @file{@var{file}.02.sibling}.
3268 Dump after the first jump optimization, to @file{@var{file}.04.jump}.
3271 Dump after conversion from registers to stack, to @file{@var{file}.34.stack}.
3274 Dump after local register allocation, to @file{@var{file}.24.lreg}.
3277 Dump after loop optimization passes, to @file{@var{file}.09.loop} and
3278 @file{@var{file}.16.loop2}.
3281 Dump after performing the machine dependent reorganization pass, to
3282 @file{@var{file}.35.mach}.
3285 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3288 Dump after the register move pass, to @file{@var{file}.22.regmove}.
3291 Dump after post-reload optimizations, to @file{@var{file}.26.postreload}.
3294 Dump after RTL generation, to @file{@var{file}.01.rtl}.
3297 Dump after the second scheduling pass, to @file{@var{file}.33.sched2}.
3300 Dump after CSE (including the jump optimization that sometimes follows
3301 CSE), to @file{@var{file}.06.cse}.
3304 Dump after the first scheduling pass, to @file{@var{file}.23.sched}.
3307 Dump after the second CSE pass (including the jump optimization that
3308 sometimes follows CSE), to @file{@var{file}.18.cse2}.
3311 Dump after running tracer, to @file{@var{file}.15.tracer}.
3314 Dump after null pointer elimination pass to @file{@var{file}.05.null}.
3317 Dump callgraph and unit-at-a-time optimization @file{@var{file}.00.unit}.
3320 Dump after the value profile transformations, to @file{@var{file}.13.vpt}.
3321 Also dump after variable tracking, to @file{@var{file}.35.vartrack}.
3324 Dump after the second flow pass, to @file{@var{file}.27.flow2}.
3327 Dump after the peephole pass, to @file{@var{file}.28.peephole2}.
3330 Dump after constructing the web, to @file{@var{file}.17.web}.
3333 Produce all the dumps listed above.
3336 Produce a core dump whenever an error occurs.
3339 Print statistics on memory usage, at the end of the run, to
3343 Annotate the assembler output with a comment indicating which
3344 pattern and alternative was used. The length of each instruction is
3348 Dump the RTL in the assembler output as a comment before each instruction.
3349 Also turns on @option{-dp} annotation.
3352 For each of the other indicated dump files (except for
3353 @file{@var{file}.01.rtl}), dump a representation of the control flow graph
3354 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3357 Just generate RTL for a function instead of compiling it. Usually used
3361 Dump debugging information during parsing, to standard error.
3364 @item -fdump-unnumbered
3365 @opindex fdump-unnumbered
3366 When doing debugging dumps (see @option{-d} option above), suppress instruction
3367 numbers and line number note output. This makes it more feasible to
3368 use diff on debugging dumps for compiler invocations with different
3369 options, in particular with and without @option{-g}.
3371 @item -fdump-translation-unit @r{(C and C++ only)}
3372 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3373 @opindex fdump-translation-unit
3374 Dump a representation of the tree structure for the entire translation
3375 unit to a file. The file name is made by appending @file{.tu} to the
3376 source file name. If the @samp{-@var{options}} form is used, @var{options}
3377 controls the details of the dump as described for the
3378 @option{-fdump-tree} options.
3380 @item -fdump-class-hierarchy @r{(C++ only)}
3381 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3382 @opindex fdump-class-hierarchy
3383 Dump a representation of each class's hierarchy and virtual function
3384 table layout to a file. The file name is made by appending @file{.class}
3385 to the source file name. If the @samp{-@var{options}} form is used,
3386 @var{options} controls the details of the dump as described for the
3387 @option{-fdump-tree} options.
3389 @item -fdump-tree-@var{switch} @r{(C++ only)}
3390 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3392 Control the dumping at various stages of processing the intermediate
3393 language tree to a file. The file name is generated by appending a switch
3394 specific suffix to the source file name. If the @samp{-@var{options}}
3395 form is used, @var{options} is a list of @samp{-} separated options that
3396 control the details of the dump. Not all options are applicable to all
3397 dumps, those which are not meaningful will be ignored. The following
3398 options are available
3402 Print the address of each node. Usually this is not meaningful as it
3403 changes according to the environment and source file. Its primary use
3404 is for tying up a dump file with a debug environment.
3406 Inhibit dumping of members of a scope or body of a function merely
3407 because that scope has been reached. Only dump such items when they
3408 are directly reachable by some other path.
3410 Turn on all options.
3413 The following tree dumps are possible:
3416 Dump before any tree based optimization, to @file{@var{file}.original}.
3418 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3420 Dump after function inlining, to @file{@var{file}.inlined}.
3423 @item -frandom-seed=@var{string}
3424 @opindex frandom-string
3425 This option provides a seed that GCC uses when it would otherwise use
3426 random numbers. It is used to generate certain symbol names
3427 that have to be different in every compiled file. It is also used to
3428 place unique stamps in coverage data files and the object files that
3429 produce them. You can use the @option{-frandom-seed} option to produce
3430 reproducibly identical object files.
3432 The @var{string} should be different for every file you compile.
3434 @item -fsched-verbose=@var{n}
3435 @opindex fsched-verbose
3436 On targets that use instruction scheduling, this option controls the
3437 amount of debugging output the scheduler prints. This information is
3438 written to standard error, unless @option{-dS} or @option{-dR} is
3439 specified, in which case it is output to the usual dump
3440 listing file, @file{.sched} or @file{.sched2} respectively. However
3441 for @var{n} greater than nine, the output is always printed to standard
3444 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3445 same information as @option{-dRS}. For @var{n} greater than one, it
3446 also output basic block probabilities, detailed ready list information
3447 and unit/insn info. For @var{n} greater than two, it includes RTL
3448 at abort point, control-flow and regions info. And for @var{n} over
3449 four, @option{-fsched-verbose} also includes dependence info.
3453 Store the usual ``temporary'' intermediate files permanently; place them
3454 in the current directory and name them based on the source file. Thus,
3455 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3456 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3457 preprocessed @file{foo.i} output file even though the compiler now
3458 normally uses an integrated preprocessor.
3462 Report the CPU time taken by each subprocess in the compilation
3463 sequence. For C source files, this is the compiler proper and assembler
3464 (plus the linker if linking is done). The output looks like this:
3471 The first number on each line is the ``user time,'' that is time spent
3472 executing the program itself. The second number is ``system time,''
3473 time spent executing operating system routines on behalf of the program.
3474 Both numbers are in seconds.
3476 @item -fvar-tracking
3477 @opindex fvar-tracking
3478 Run variable tracking pass. It computes where variables are stored at each
3479 position in code. Better debugging information is then generated
3480 (if the debugging information format supports this information).
3482 It is enabled by default when compiling with optimization (@option{-Os},
3483 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
3484 the debug info format supports it.
3486 @item -print-file-name=@var{library}
3487 @opindex print-file-name
3488 Print the full absolute name of the library file @var{library} that
3489 would be used when linking---and don't do anything else. With this
3490 option, GCC does not compile or link anything; it just prints the
3493 @item -print-multi-directory
3494 @opindex print-multi-directory
3495 Print the directory name corresponding to the multilib selected by any
3496 other switches present in the command line. This directory is supposed
3497 to exist in @env{GCC_EXEC_PREFIX}.
3499 @item -print-multi-lib
3500 @opindex print-multi-lib
3501 Print the mapping from multilib directory names to compiler switches
3502 that enable them. The directory name is separated from the switches by
3503 @samp{;}, and each switch starts with an @samp{@@} instead of the
3504 @samp{-}, without spaces between multiple switches. This is supposed to
3505 ease shell-processing.
3507 @item -print-prog-name=@var{program}
3508 @opindex print-prog-name
3509 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3511 @item -print-libgcc-file-name
3512 @opindex print-libgcc-file-name
3513 Same as @option{-print-file-name=libgcc.a}.
3515 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3516 but you do want to link with @file{libgcc.a}. You can do
3519 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3522 @item -print-search-dirs
3523 @opindex print-search-dirs
3524 Print the name of the configured installation directory and a list of
3525 program and library directories @command{gcc} will search---and don't do anything else.
3527 This is useful when @command{gcc} prints the error message
3528 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3529 To resolve this you either need to put @file{cpp0} and the other compiler
3530 components where @command{gcc} expects to find them, or you can set the environment
3531 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3532 Don't forget the trailing '/'.
3533 @xref{Environment Variables}.
3536 @opindex dumpmachine
3537 Print the compiler's target machine (for example,
3538 @samp{i686-pc-linux-gnu})---and don't do anything else.
3541 @opindex dumpversion
3542 Print the compiler version (for example, @samp{3.0})---and don't do
3547 Print the compiler's built-in specs---and don't do anything else. (This
3548 is used when GCC itself is being built.) @xref{Spec Files}.
3550 @item -feliminate-unused-debug-types
3551 @opindex feliminate-unused-debug-types
3552 Normally, when producing DWARF2 output, GCC will emit debugging
3553 information for all types declared in a compilation
3554 unit, regardless of whether or not they are actually used
3555 in that compilation unit. Sometimes this is useful, such as
3556 if, in the debugger, you want to cast a value to a type that is
3557 not actually used in your program (but is declared). More often,
3558 however, this results in a significant amount of wasted space.
3559 With this option, GCC will avoid producing debug symbol output
3560 for types that are nowhere used in the source file being compiled.
3563 @node Optimize Options
3564 @section Options That Control Optimization
3565 @cindex optimize options
3566 @cindex options, optimization
3568 These options control various sorts of optimizations.
3570 Without any optimization option, the compiler's goal is to reduce the
3571 cost of compilation and to make debugging produce the expected
3572 results. Statements are independent: if you stop the program with a
3573 breakpoint between statements, you can then assign a new value to any
3574 variable or change the program counter to any other statement in the
3575 function and get exactly the results you would expect from the source
3578 Turning on optimization flags makes the compiler attempt to improve
3579 the performance and/or code size at the expense of compilation time
3580 and possibly the ability to debug the program.
3582 The compiler performs optimization based on the knowledge it has of
3583 the program. Using the @option{-funit-at-a-time} flag will allow the
3584 compiler to consider information gained from later functions in the
3585 file when compiling a function. Compiling multiple files at once to a
3586 single output file (and using @option{-funit-at-a-time}) will allow
3587 the compiler to use information gained from all of the files when
3588 compiling each of them.
3590 Not all optimizations are controlled directly by a flag. Only
3591 optimizations that have a flag are listed.
3598 Optimize. Optimizing compilation takes somewhat more time, and a lot
3599 more memory for a large function.
3601 With @option{-O}, the compiler tries to reduce code size and execution
3602 time, without performing any optimizations that take a great deal of
3605 @option{-O} turns on the following optimization flags:
3606 @gccoptlist{-fdefer-pop @gol
3607 -fmerge-constants @gol
3609 -floop-optimize @gol
3610 -fif-conversion @gol
3611 -fif-conversion2 @gol
3612 -fdelayed-branch @gol
3613 -fguess-branch-probability @gol
3616 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3617 where doing so does not interfere with debugging.
3621 Optimize even more. GCC performs nearly all supported optimizations
3622 that do not involve a space-speed tradeoff. The compiler does not
3623 perform loop unrolling or function inlining when you specify @option{-O2}.
3624 As compared to @option{-O}, this option increases both compilation time
3625 and the performance of the generated code.
3627 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3628 also turns on the following optimization flags:
3629 @gccoptlist{-fforce-mem @gol
3630 -foptimize-sibling-calls @gol
3631 -fstrength-reduce @gol
3632 -fcse-follow-jumps -fcse-skip-blocks @gol
3633 -frerun-cse-after-loop -frerun-loop-opt @gol
3634 -fgcse -fgcse-lm -fgcse-sm -fgcse-las @gol
3635 -fdelete-null-pointer-checks @gol
3636 -fexpensive-optimizations @gol
3638 -fschedule-insns -fschedule-insns2 @gol
3639 -fsched-interblock -fsched-spec @gol
3642 -freorder-blocks -freorder-functions @gol
3643 -fstrict-aliasing @gol
3644 -funit-at-a-time @gol
3645 -falign-functions -falign-jumps @gol
3646 -falign-loops -falign-labels @gol
3649 Please note the warning under @option{-fgcse} about
3650 invoking @option{-O2} on programs that use computed gotos.
3654 Optimize yet more. @option{-O3} turns on all optimizations specified by
3655 @option{-O2} and also turns on the @option{-finline-functions},
3656 @option{-fweb} and @option{-frename-registers} options.
3660 Do not optimize. This is the default.
3664 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3665 do not typically increase code size. It also performs further
3666 optimizations designed to reduce code size.
3668 @option{-Os} disables the following optimization flags:
3669 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3670 -falign-labels -freorder-blocks -fprefetch-loop-arrays}
3672 If you use multiple @option{-O} options, with or without level numbers,
3673 the last such option is the one that is effective.
3676 Options of the form @option{-f@var{flag}} specify machine-independent
3677 flags. Most flags have both positive and negative forms; the negative
3678 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3679 below, only one of the forms is listed---the one you typically will
3680 use. You can figure out the other form by either removing @samp{no-}
3683 The following options control specific optimizations. They are either
3684 activated by @option{-O} options or are related to ones that are. You
3685 can use the following flags in the rare cases when ``fine-tuning'' of
3686 optimizations to be performed is desired.
3689 @item -fno-default-inline
3690 @opindex fno-default-inline
3691 Do not make member functions inline by default merely because they are
3692 defined inside the class scope (C++ only). Otherwise, when you specify
3693 @w{@option{-O}}, member functions defined inside class scope are compiled
3694 inline by default; i.e., you don't need to add @samp{inline} in front of
3695 the member function name.
3697 @item -fno-defer-pop
3698 @opindex fno-defer-pop
3699 Always pop the arguments to each function call as soon as that function
3700 returns. For machines which must pop arguments after a function call,
3701 the compiler normally lets arguments accumulate on the stack for several
3702 function calls and pops them all at once.
3704 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3708 Force memory operands to be copied into registers before doing
3709 arithmetic on them. This produces better code by making all memory
3710 references potential common subexpressions. When they are not common
3711 subexpressions, instruction combination should eliminate the separate
3714 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3717 @opindex fforce-addr
3718 Force memory address constants to be copied into registers before
3719 doing arithmetic on them. This may produce better code just as
3720 @option{-fforce-mem} may.
3722 @item -fomit-frame-pointer
3723 @opindex fomit-frame-pointer
3724 Don't keep the frame pointer in a register for functions that
3725 don't need one. This avoids the instructions to save, set up and
3726 restore frame pointers; it also makes an extra register available
3727 in many functions. @strong{It also makes debugging impossible on
3730 On some machines, such as the VAX, this flag has no effect, because
3731 the standard calling sequence automatically handles the frame pointer
3732 and nothing is saved by pretending it doesn't exist. The
3733 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3734 whether a target machine supports this flag. @xref{Registers,,Register
3735 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3737 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3739 @item -foptimize-sibling-calls
3740 @opindex foptimize-sibling-calls
3741 Optimize sibling and tail recursive calls.
3743 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3747 Don't pay attention to the @code{inline} keyword. Normally this option
3748 is used to keep the compiler from expanding any functions inline.
3749 Note that if you are not optimizing, no functions can be expanded inline.
3751 @item -finline-functions
3752 @opindex finline-functions
3753 Integrate all simple functions into their callers. The compiler
3754 heuristically decides which functions are simple enough to be worth
3755 integrating in this way.
3757 If all calls to a given function are integrated, and the function is
3758 declared @code{static}, then the function is normally not output as
3759 assembler code in its own right.
3761 Enabled at level @option{-O3}.
3763 @item -finline-limit=@var{n}
3764 @opindex finline-limit
3765 By default, GCC limits the size of functions that can be inlined. This flag
3766 allows the control of this limit for functions that are explicitly marked as
3767 inline (i.e., marked with the inline keyword or defined within the class
3768 definition in c++). @var{n} is the size of functions that can be inlined in
3769 number of pseudo instructions (not counting parameter handling). The default
3770 value of @var{n} is 600.
3771 Increasing this value can result in more inlined code at
3772 the cost of compilation time and memory consumption. Decreasing usually makes
3773 the compilation faster and less code will be inlined (which presumably
3774 means slower programs). This option is particularly useful for programs that
3775 use inlining heavily such as those based on recursive templates with C++.
3777 Inlining is actually controlled by a number of parameters, which may be
3778 specified individually by using @option{--param @var{name}=@var{value}}.
3779 The @option{-finline-limit=@var{n}} option sets some of these parameters
3783 @item max-inline-insns-single
3784 is set to @var{n}/2.
3785 @item max-inline-insns-auto
3786 is set to @var{n}/2.
3787 @item min-inline-insns
3788 is set to 130 or @var{n}/4, whichever is smaller.
3789 @item max-inline-insns-rtl
3793 See below for a documentation of the individual
3794 parameters controlling inlining.
3796 @emph{Note:} pseudo instruction represents, in this particular context, an
3797 abstract measurement of function's size. In no way, it represents a count
3798 of assembly instructions and as such its exact meaning might change from one
3799 release to an another.
3801 @item -fkeep-inline-functions
3802 @opindex fkeep-inline-functions
3803 Even if all calls to a given function are integrated, and the function
3804 is declared @code{static}, nevertheless output a separate run-time
3805 callable version of the function. This switch does not affect
3806 @code{extern inline} functions.
3808 @item -fkeep-static-consts
3809 @opindex fkeep-static-consts
3810 Emit variables declared @code{static const} when optimization isn't turned
3811 on, even if the variables aren't referenced.
3813 GCC enables this option by default. If you want to force the compiler to
3814 check if the variable was referenced, regardless of whether or not
3815 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3817 @item -fmerge-constants
3818 Attempt to merge identical constants (string constants and floating point
3819 constants) across compilation units.
3821 This option is the default for optimized compilation if the assembler and
3822 linker support it. Use @option{-fno-merge-constants} to inhibit this
3825 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3827 @item -fmerge-all-constants
3828 Attempt to merge identical constants and identical variables.
3830 This option implies @option{-fmerge-constants}. In addition to
3831 @option{-fmerge-constants} this considers e.g. even constant initialized
3832 arrays or initialized constant variables with integral or floating point
3833 types. Languages like C or C++ require each non-automatic variable to
3834 have distinct location, so using this option will result in non-conforming
3839 Use a graph coloring register allocator. Currently this option is meant
3840 only for testing. Users should not specify this option, since it is not
3841 yet ready for production use.
3843 @item -fno-branch-count-reg
3844 @opindex fno-branch-count-reg
3845 Do not use ``decrement and branch'' instructions on a count register,
3846 but instead generate a sequence of instructions that decrement a
3847 register, compare it against zero, then branch based upon the result.
3848 This option is only meaningful on architectures that support such
3849 instructions, which include x86, PowerPC, IA-64 and S/390.
3851 The default is @option{-fbranch-count-reg}, enabled when
3852 @option{-fstrength-reduce} is enabled.
3854 @item -fno-function-cse
3855 @opindex fno-function-cse
3856 Do not put function addresses in registers; make each instruction that
3857 calls a constant function contain the function's address explicitly.
3859 This option results in less efficient code, but some strange hacks
3860 that alter the assembler output may be confused by the optimizations
3861 performed when this option is not used.
3863 The default is @option{-ffunction-cse}
3865 @item -fno-zero-initialized-in-bss
3866 @opindex fno-zero-initialized-in-bss
3867 If the target supports a BSS section, GCC by default puts variables that
3868 are initialized to zero into BSS@. This can save space in the resulting
3871 This option turns off this behavior because some programs explicitly
3872 rely on variables going to the data section. E.g., so that the
3873 resulting executable can find the beginning of that section and/or make
3874 assumptions based on that.
3876 The default is @option{-fzero-initialized-in-bss}.
3878 @item -fstrength-reduce
3879 @opindex fstrength-reduce
3880 Perform the optimizations of loop strength reduction and
3881 elimination of iteration variables.
3883 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3885 @item -fthread-jumps
3886 @opindex fthread-jumps
3887 Perform optimizations where we check to see if a jump branches to a
3888 location where another comparison subsumed by the first is found. If
3889 so, the first branch is redirected to either the destination of the
3890 second branch or a point immediately following it, depending on whether
3891 the condition is known to be true or false.
3893 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3895 @item -fcse-follow-jumps
3896 @opindex fcse-follow-jumps
3897 In common subexpression elimination, scan through jump instructions
3898 when the target of the jump is not reached by any other path. For
3899 example, when CSE encounters an @code{if} statement with an
3900 @code{else} clause, CSE will follow the jump when the condition
3903 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3905 @item -fcse-skip-blocks
3906 @opindex fcse-skip-blocks
3907 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3908 follow jumps which conditionally skip over blocks. When CSE
3909 encounters a simple @code{if} statement with no else clause,
3910 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3911 body of the @code{if}.
3913 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3915 @item -frerun-cse-after-loop
3916 @opindex frerun-cse-after-loop
3917 Re-run common subexpression elimination after loop optimizations has been
3920 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3922 @item -frerun-loop-opt
3923 @opindex frerun-loop-opt
3924 Run the loop optimizer twice.
3926 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3930 Perform a global common subexpression elimination pass.
3931 This pass also performs global constant and copy propagation.
3933 @emph{Note:} When compiling a program using computed gotos, a GCC
3934 extension, you may get better runtime performance if you disable
3935 the global common subexpression elimination pass by adding
3936 @option{-fno-gcse} to the command line.
3938 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3942 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3943 attempt to move loads which are only killed by stores into themselves. This
3944 allows a loop containing a load/store sequence to be changed to a load outside
3945 the loop, and a copy/store within the loop.
3947 Enabled by default when gcse is enabled.
3951 When @option{-fgcse-sm} is enabled, a store motion pass is run after
3952 global common subexpression elimination. This pass will attempt to move
3953 stores out of loops. When used in conjunction with @option{-fgcse-lm},
3954 loops containing a load/store sequence can be changed to a load before
3955 the loop and a store after the loop.
3957 Enabled by default when gcse is enabled.
3961 When @option{-fgcse-las} is enabled, the global common subexpression
3962 elimination pass eliminates redundant loads that come after stores to the
3963 same memory location (both partial and full redundancies).
3965 Enabled by default when gcse is enabled.
3967 @item -floop-optimize
3968 @opindex floop-optimize
3969 Perform loop optimizations: move constant expressions out of loops, simplify
3970 exit test conditions and optionally do strength-reduction and loop unrolling as
3973 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3975 @item -fcrossjumping
3976 @opindex crossjumping
3977 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
3978 resulting code may or may not perform better than without cross-jumping.
3980 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3982 @item -fif-conversion
3983 @opindex if-conversion
3984 Attempt to transform conditional jumps into branch-less equivalents. This
3985 include use of conditional moves, min, max, set flags and abs instructions, and
3986 some tricks doable by standard arithmetics. The use of conditional execution
3987 on chips where it is available is controlled by @code{if-conversion2}.
3989 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3991 @item -fif-conversion2
3992 @opindex if-conversion2
3993 Use conditional execution (where available) to transform conditional jumps into
3994 branch-less equivalents.
3996 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3998 @item -fdelete-null-pointer-checks
3999 @opindex fdelete-null-pointer-checks
4000 Use global dataflow analysis to identify and eliminate useless checks
4001 for null pointers. The compiler assumes that dereferencing a null
4002 pointer would have halted the program. If a pointer is checked after
4003 it has already been dereferenced, it cannot be null.
4005 In some environments, this assumption is not true, and programs can
4006 safely dereference null pointers. Use
4007 @option{-fno-delete-null-pointer-checks} to disable this optimization
4008 for programs which depend on that behavior.
4010 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4012 @item -fexpensive-optimizations
4013 @opindex fexpensive-optimizations
4014 Perform a number of minor optimizations that are relatively expensive.
4016 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4018 @item -foptimize-register-move
4020 @opindex foptimize-register-move
4022 Attempt to reassign register numbers in move instructions and as
4023 operands of other simple instructions in order to maximize the amount of
4024 register tying. This is especially helpful on machines with two-operand
4027 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4030 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4032 @item -fdelayed-branch
4033 @opindex fdelayed-branch
4034 If supported for the target machine, attempt to reorder instructions
4035 to exploit instruction slots available after delayed branch
4038 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4040 @item -fschedule-insns
4041 @opindex fschedule-insns
4042 If supported for the target machine, attempt to reorder instructions to
4043 eliminate execution stalls due to required data being unavailable. This
4044 helps machines that have slow floating point or memory load instructions
4045 by allowing other instructions to be issued until the result of the load
4046 or floating point instruction is required.
4048 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4050 @item -fschedule-insns2
4051 @opindex fschedule-insns2
4052 Similar to @option{-fschedule-insns}, but requests an additional pass of
4053 instruction scheduling after register allocation has been done. This is
4054 especially useful on machines with a relatively small number of
4055 registers and where memory load instructions take more than one cycle.
4057 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4059 @item -fno-sched-interblock
4060 @opindex fno-sched-interblock
4061 Don't schedule instructions across basic blocks. This is normally
4062 enabled by default when scheduling before register allocation, i.e.@:
4063 with @option{-fschedule-insns} or at @option{-O2} or higher.
4065 @item -fno-sched-spec
4066 @opindex fno-sched-spec
4067 Don't allow speculative motion of non-load instructions. This is normally
4068 enabled by default when scheduling before register allocation, i.e.@:
4069 with @option{-fschedule-insns} or at @option{-O2} or higher.
4071 @item -fsched-spec-load
4072 @opindex fsched-spec-load
4073 Allow speculative motion of some load instructions. This only makes
4074 sense when scheduling before register allocation, i.e.@: with
4075 @option{-fschedule-insns} or at @option{-O2} or higher.
4077 @item -fsched-spec-load-dangerous
4078 @opindex fsched-spec-load-dangerous
4079 Allow speculative motion of more load instructions. This only makes
4080 sense when scheduling before register allocation, i.e.@: with
4081 @option{-fschedule-insns} or at @option{-O2} or higher.
4083 @item -fsched-stalled-insns=@var{n}
4084 @opindex fsched-stalled-insns
4085 Define how many insns (if any) can be moved prematurely from the queue
4086 of stalled insns into the ready list, during the second scheduling pass.
4088 @item -fsched-stalled-insns-dep=@var{n}
4089 @opindex fsched-stalled-insns-dep
4090 Define how many insn groups (cycles) will be examined for a dependency
4091 on a stalled insn that is candidate for premature removal from the queue
4092 of stalled insns. Has an effect only during the second scheduling pass,
4093 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4095 @item -fsched2-use-superblocks
4096 @opindex fsched2-use-superblocks
4097 When scheduling after register allocation, do use superblock scheduling
4098 algorithm. Superblock scheduling allows motion across basic block boundaries
4099 resulting on faster schedules. This option is experimental, as not all machine
4100 descriptions used by GCC model the CPU closely enough to avoid unreliable
4101 results from the algorithm.
4103 This only makes sense when scheduling after register allocation, i.e.@: with
4104 @option{-fschedule-insns2} or at @option{-O2} or higher.
4106 @item -fsched2-use-traces
4107 @opindex fsched2-use-traces
4108 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4109 allocation and additionally perform code duplication in order to increase the
4110 size of superblocks using tracer pass. See @option{-ftracer} for details on
4113 This mode should produce faster but significantly longer programs. Also
4114 without @code{-fbranch-probabilities} the traces constructed may not match the
4115 reality and hurt the performance. This only makes
4116 sense when scheduling after register allocation, i.e.@: with
4117 @option{-fschedule-insns2} or at @option{-O2} or higher.
4119 @item -fcaller-saves
4120 @opindex fcaller-saves
4121 Enable values to be allocated in registers that will be clobbered by
4122 function calls, by emitting extra instructions to save and restore the
4123 registers around such calls. Such allocation is done only when it
4124 seems to result in better code than would otherwise be produced.
4126 This option is always enabled by default on certain machines, usually
4127 those which have no call-preserved registers to use instead.
4129 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4131 @item -fmove-all-movables
4132 @opindex fmove-all-movables
4133 Forces all invariant computations in loops to be moved
4136 @item -freduce-all-givs
4137 @opindex freduce-all-givs
4138 Forces all general-induction variables in loops to be
4141 @emph{Note:} When compiling programs written in Fortran,
4142 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4143 by default when you use the optimizer.
4145 These options may generate better or worse code; results are highly
4146 dependent on the structure of loops within the source code.
4148 These two options are intended to be removed someday, once
4149 they have helped determine the efficacy of various
4150 approaches to improving loop optimizations.
4152 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
4153 know how use of these options affects
4154 the performance of your production code.
4155 We're very interested in code that runs @emph{slower}
4156 when these options are @emph{enabled}.
4159 @itemx -fno-peephole2
4160 @opindex fno-peephole
4161 @opindex fno-peephole2
4162 Disable any machine-specific peephole optimizations. The difference
4163 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4164 are implemented in the compiler; some targets use one, some use the
4165 other, a few use both.
4167 @option{-fpeephole} is enabled by default.
4168 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4170 @item -fno-guess-branch-probability
4171 @opindex fno-guess-branch-probability
4172 Do not guess branch probabilities using a randomized model.
4174 Sometimes GCC will opt to use a randomized model to guess branch
4175 probabilities, when none are available from either profiling feedback
4176 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4177 different runs of the compiler on the same program may produce different
4180 In a hard real-time system, people don't want different runs of the
4181 compiler to produce code that has different behavior; minimizing
4182 non-determinism is of paramount import. This switch allows users to
4183 reduce non-determinism, possibly at the expense of inferior
4186 The default is @option{-fguess-branch-probability} at levels
4187 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4189 @item -freorder-blocks
4190 @opindex freorder-blocks
4191 Reorder basic blocks in the compiled function in order to reduce number of
4192 taken branches and improve code locality.
4194 Enabled at levels @option{-O2}, @option{-O3}.
4196 @item -freorder-functions
4197 @opindex freorder-functions
4198 Reorder basic blocks in the compiled function in order to reduce number of
4199 taken branches and improve code locality. This is implemented by using special
4200 subsections @code{text.hot} for most frequently executed functions and
4201 @code{text.unlikely} for unlikely executed functions. Reordering is done by
4202 the linker so object file format must support named sections and linker must
4203 place them in a reasonable way.
4205 Also profile feedback must be available in to make this option effective. See
4206 @option{-fprofile-arcs} for details.
4208 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4210 @item -fstrict-aliasing
4211 @opindex fstrict-aliasing
4212 Allows the compiler to assume the strictest aliasing rules applicable to
4213 the language being compiled. For C (and C++), this activates
4214 optimizations based on the type of expressions. In particular, an
4215 object of one type is assumed never to reside at the same address as an
4216 object of a different type, unless the types are almost the same. For
4217 example, an @code{unsigned int} can alias an @code{int}, but not a
4218 @code{void*} or a @code{double}. A character type may alias any other
4221 Pay special attention to code like this:
4234 The practice of reading from a different union member than the one most
4235 recently written to (called ``type-punning'') is common. Even with
4236 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4237 is accessed through the union type. So, the code above will work as
4238 expected. However, this code might not:
4249 Every language that wishes to perform language-specific alias analysis
4250 should define a function that computes, given an @code{tree}
4251 node, an alias set for the node. Nodes in different alias sets are not
4252 allowed to alias. For an example, see the C front-end function
4253 @code{c_get_alias_set}.
4255 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4257 @item -falign-functions
4258 @itemx -falign-functions=@var{n}
4259 @opindex falign-functions
4260 Align the start of functions to the next power-of-two greater than
4261 @var{n}, skipping up to @var{n} bytes. For instance,
4262 @option{-falign-functions=32} aligns functions to the next 32-byte
4263 boundary, but @option{-falign-functions=24} would align to the next
4264 32-byte boundary only if this can be done by skipping 23 bytes or less.
4266 @option{-fno-align-functions} and @option{-falign-functions=1} are
4267 equivalent and mean that functions will not be aligned.
4269 Some assemblers only support this flag when @var{n} is a power of two;
4270 in that case, it is rounded up.
4272 If @var{n} is not specified or is zero, use a machine-dependent default.
4274 Enabled at levels @option{-O2}, @option{-O3}.
4276 @item -falign-labels
4277 @itemx -falign-labels=@var{n}
4278 @opindex falign-labels
4279 Align all branch targets to a power-of-two boundary, skipping up to
4280 @var{n} bytes like @option{-falign-functions}. This option can easily
4281 make code slower, because it must insert dummy operations for when the
4282 branch target is reached in the usual flow of the code.
4284 @option{-fno-align-labels} and @option{-falign-labels=1} are
4285 equivalent and mean that labels will not be aligned.
4287 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4288 are greater than this value, then their values are used instead.
4290 If @var{n} is not specified or is zero, use a machine-dependent default
4291 which is very likely to be @samp{1}, meaning no alignment.
4293 Enabled at levels @option{-O2}, @option{-O3}.
4296 @itemx -falign-loops=@var{n}
4297 @opindex falign-loops
4298 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4299 like @option{-falign-functions}. The hope is that the loop will be
4300 executed many times, which will make up for any execution of the dummy
4303 @option{-fno-align-loops} and @option{-falign-loops=1} are
4304 equivalent and mean that loops will not be aligned.
4306 If @var{n} is not specified or is zero, use a machine-dependent default.
4308 Enabled at levels @option{-O2}, @option{-O3}.
4311 @itemx -falign-jumps=@var{n}
4312 @opindex falign-jumps
4313 Align branch targets to a power-of-two boundary, for branch targets
4314 where the targets can only be reached by jumping, skipping up to @var{n}
4315 bytes like @option{-falign-functions}. In this case, no dummy operations
4318 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4319 equivalent and mean that loops will not be aligned.
4321 If @var{n} is not specified or is zero, use a machine-dependent default.
4323 Enabled at levels @option{-O2}, @option{-O3}.
4325 @item -frename-registers
4326 @opindex frename-registers
4327 Attempt to avoid false dependencies in scheduled code by making use
4328 of registers left over after register allocation. This optimization
4329 will most benefit processors with lots of registers. It can, however,
4330 make debugging impossible, since variables will no longer stay in
4331 a ``home register''.
4335 Constructs webs as commonly used for register allocation purposes and assign
4336 each web individual pseudo register. This allows our register allocation pass
4337 to operate on pseudos directly, but also strengthens several other optimization
4338 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
4339 however, make debugging impossible, since variables will no longer stay in a
4342 Enabled at levels @option{-O3}.
4344 @item -fno-cprop-registers
4345 @opindex fno-cprop-registers
4346 After register allocation and post-register allocation instruction splitting,
4347 we perform a copy-propagation pass to try to reduce scheduling dependencies
4348 and occasionally eliminate the copy.
4350 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4352 @item -fprofile-generate
4353 @opindex fprofile-generate
4354 Enable options usually used for instrumenting application to produce profile usefull
4355 for later recompilation profile feedback based optimization.
4357 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}
4360 @opindex fprofile-use
4361 Enable profile feedback directed optimizations, and optimizations
4362 generally profitable only with profile feedback available.
4364 The following options are enabled: @code{-fbranch-probabilities},
4365 @code{-fvpt}, @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}.
4369 The following options control compiler behavior regarding floating
4370 point arithmetic. These options trade off between speed and
4371 correctness. All must be specifically enabled.
4375 @opindex ffloat-store
4376 Do not store floating point variables in registers, and inhibit other
4377 options that might change whether a floating point value is taken from a
4380 @cindex floating point precision
4381 This option prevents undesirable excess precision on machines such as
4382 the 68000 where the floating registers (of the 68881) keep more
4383 precision than a @code{double} is supposed to have. Similarly for the
4384 x86 architecture. For most programs, the excess precision does only
4385 good, but a few programs rely on the precise definition of IEEE floating
4386 point. Use @option{-ffloat-store} for such programs, after modifying
4387 them to store all pertinent intermediate computations into variables.
4391 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4392 @option{-fno-trapping-math}, @option{-ffinite-math-only},
4393 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
4395 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4397 This option should never be turned on by any @option{-O} option since
4398 it can result in incorrect output for programs which depend on
4399 an exact implementation of IEEE or ISO rules/specifications for
4402 @item -fno-math-errno
4403 @opindex fno-math-errno
4404 Do not set ERRNO after calling math functions that are executed
4405 with a single instruction, e.g., sqrt. A program that relies on
4406 IEEE exceptions for math error handling may want to use this flag
4407 for speed while maintaining IEEE arithmetic compatibility.
4409 This option should never be turned on by any @option{-O} option since
4410 it can result in incorrect output for programs which depend on
4411 an exact implementation of IEEE or ISO rules/specifications for
4414 The default is @option{-fmath-errno}.
4416 @item -funsafe-math-optimizations
4417 @opindex funsafe-math-optimizations
4418 Allow optimizations for floating-point arithmetic that (a) assume
4419 that arguments and results are valid and (b) may violate IEEE or
4420 ANSI standards. When used at link-time, it may include libraries
4421 or startup files that change the default FPU control word or other
4422 similar optimizations.
4424 This option should never be turned on by any @option{-O} option since
4425 it can result in incorrect output for programs which depend on
4426 an exact implementation of IEEE or ISO rules/specifications for
4429 The default is @option{-fno-unsafe-math-optimizations}.
4431 @item -ffinite-math-only
4432 @opindex ffinite-math-only
4433 Allow optimizations for floating-point arithmetic that assume
4434 that arguments and results are not NaNs or +-Infs.
4436 This option should never be turned on by any @option{-O} option since
4437 it can result in incorrect output for programs which depend on
4438 an exact implementation of IEEE or ISO rules/specifications.
4440 The default is @option{-fno-finite-math-only}.
4442 @item -fno-trapping-math
4443 @opindex fno-trapping-math
4444 Compile code assuming that floating-point operations cannot generate
4445 user-visible traps. These traps include division by zero, overflow,
4446 underflow, inexact result and invalid operation. This option implies
4447 @option{-fno-signaling-nans}. Setting this option may allow faster
4448 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4450 This option should never be turned on by any @option{-O} option since
4451 it can result in incorrect output for programs which depend on
4452 an exact implementation of IEEE or ISO rules/specifications for
4455 The default is @option{-ftrapping-math}.
4457 @item -frounding-math
4458 @opindex frounding-math
4459 Disable transformations and optimizations that assume default floating
4460 point rounding behavior. This is round-to-zero for all floating point
4461 to integer conversions, and round-to-nearest for all other arithmetic
4462 truncations. This option should be specified for programs that change
4463 the FP rounding mode dynamically, or that may be executed with a
4464 non-default rounding mode. This option disables constant folding of
4465 floating point expressions at compile-time (which may be affected by
4466 rounding mode) and arithmetic transformations that are unsafe in the
4467 presence of sign-dependent rounding modes.
4469 The default is @option{-fno-rounding-math}.
4471 This option is experimental and does not currently guarantee to
4472 disable all GCC optimizations that are affected by rounding mode.
4473 Future versions of GCC may provide finer control of this setting
4474 using C99's @code{FENV_ACCESS} pragma. This command line option
4475 will be used to specify the default state for @code{FENV_ACCESS}.
4477 @item -fsignaling-nans
4478 @opindex fsignaling-nans
4479 Compile code assuming that IEEE signaling NaNs may generate user-visible
4480 traps during floating-point operations. Setting this option disables
4481 optimizations that may change the number of exceptions visible with
4482 signaling NaNs. This option implies @option{-ftrapping-math}.
4484 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4487 The default is @option{-fno-signaling-nans}.
4489 This option is experimental and does not currently guarantee to
4490 disable all GCC optimizations that affect signaling NaN behavior.
4492 @item -fsingle-precision-constant
4493 @opindex fsingle-precision-constant
4494 Treat floating point constant as single precision constant instead of
4495 implicitly converting it to double precision constant.
4500 The following options control optimizations that may improve
4501 performance, but are not enabled by any @option{-O} options. This
4502 section includes experimental options that may produce broken code.
4505 @item -fbranch-probabilities
4506 @opindex fbranch-probabilities
4507 After running a program compiled with @option{-fprofile-arcs}
4508 (@pxref{Debugging Options,, Options for Debugging Your Program or
4509 @command{gcc}}), you can compile it a second time using
4510 @option{-fbranch-probabilities}, to improve optimizations based on
4511 the number of times each branch was taken. When the program
4512 compiled with @option{-fprofile-arcs} exits it saves arc execution
4513 counts to a file called @file{@var{sourcename}.gcda} for each source
4514 file The information in this data file is very dependent on the
4515 structure of the generated code, so you must use the same source code
4516 and the same optimization options for both compilations.
4518 With @option{-fbranch-probabilities}, GCC puts a
4519 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4520 These can be used to improve optimization. Currently, they are only
4521 used in one place: in @file{reorg.c}, instead of guessing which path a
4522 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4523 exactly determine which path is taken more often.
4525 @item -fprofile-values
4526 @opindex fprofile-values
4527 If combined with @option{-fprofile-arcs}, it adds code so that some
4528 data about values of expressions in the program is gathered.
4530 With @option{-fbranch-probabilities}, it reads back the data gathered
4531 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
4532 notes to instructions for their later usage in optimizations.
4536 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
4537 a code to gather information about values of expressions.
4539 With @option{-fbranch-probabilities}, it reads back the data gathered
4540 and actually performs the optimizations based on them.
4541 Currently the optimizations include specialization of division operation
4542 using the knowledge about the value of the denominator.
4546 Use a graph coloring register allocator. Currently this option is meant
4547 for testing, so we are interested to hear about miscompilations with
4552 Perform tail duplication to enlarge superblock size. This transformation
4553 simplifies the control flow of the function allowing other optimizations to do
4556 @item -funit-at-a-time
4557 @opindex funit-at-a-time
4558 Parse the whole compilation unit before starting to produce code.
4559 This allows some extra optimizations to take place but consumes more
4562 @item -funroll-loops
4563 @opindex funroll-loops
4564 Unroll loops whose number of iterations can be determined at compile time or
4565 upon entry to the loop. @option{-funroll-loops} implies
4566 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4567 (i.e. complete removal of loops with small constant number of iterations).
4568 This option makes code larger, and may or may not make it run faster.
4570 @item -funroll-all-loops
4571 @opindex funroll-all-loops
4572 Unroll all loops, even if their number of iterations is uncertain when
4573 the loop is entered. This usually makes programs run more slowly.
4574 @option{-funroll-all-loops} implies the same options as
4575 @option{-funroll-loops}.
4578 @opindex fpeel-loops
4579 Peels the loops for that there is enough information that they do not
4580 roll much (from profile feedback). It also turns on complete loop peeling
4581 (i.e. complete removal of loops with small constant number of iterations).
4583 @item -funswitch-loops
4584 @opindex funswitch-loops
4585 Move branches with loop invariant conditions out of the loop, with duplicates
4586 of the loop on both branches (modified according to result of the condition).
4588 @item -fold-unroll-loops
4589 @opindex fold-unroll-loops
4590 Unroll loops whose number of iterations can be determined at compile
4591 time or upon entry to the loop, using the old loop unroller whose loop
4592 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4593 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4594 option makes code larger, and may or may not make it run faster.
4596 @item -fold-unroll-all-loops
4597 @opindex fold-unroll-all-loops
4598 Unroll all loops, even if their number of iterations is uncertain when
4599 the loop is entered. This is done using the old loop unroller whose loop
4600 recognition is based on notes from frontend. This usually makes programs run more slowly.
4601 @option{-fold-unroll-all-loops} implies the same options as
4602 @option{-fold-unroll-loops}.
4604 @item -funswitch-loops
4605 @opindex funswitch-loops
4606 Move branches with loop invariant conditions out of the loop, with duplicates
4607 of the loop on both branches (modified according to result of the condition).
4609 @item -funswitch-loops
4610 @opindex funswitch-loops
4611 Move branches with loop invariant conditions out of the loop, with duplicates
4612 of the loop on both branches (modified according to result of the condition).
4614 @item -fprefetch-loop-arrays
4615 @opindex fprefetch-loop-arrays
4616 If supported by the target machine, generate instructions to prefetch
4617 memory to improve the performance of loops that access large arrays.
4619 Disabled at level @option{-Os}.
4621 @item -ffunction-sections
4622 @itemx -fdata-sections
4623 @opindex ffunction-sections
4624 @opindex fdata-sections
4625 Place each function or data item into its own section in the output
4626 file if the target supports arbitrary sections. The name of the
4627 function or the name of the data item determines the section's name
4630 Use these options on systems where the linker can perform optimizations
4631 to improve locality of reference in the instruction space. Most systems
4632 using the ELF object format and SPARC processors running Solaris 2 have
4633 linkers with such optimizations. AIX may have these optimizations in
4636 Only use these options when there are significant benefits from doing
4637 so. When you specify these options, the assembler and linker will
4638 create larger object and executable files and will also be slower.
4639 You will not be able to use @code{gprof} on all systems if you
4640 specify this option and you may have problems with debugging if
4641 you specify both this option and @option{-g}.
4643 @item -fbranch-target-load-optimize
4644 @opindex fbranch-target-load-optimize
4645 Perform branch target register load optimization before prologue / epilogue
4647 The use of target registers can typically be exposed only during reload,
4648 thus hoisting loads out of loops and doing inter-block scheduling needs
4649 a separate optimization pass.
4651 @item -fbranch-target-load-optimize2
4652 @opindex fbranch-target-load-optimize2
4653 Perform branch target register load optimization after prologue / epilogue
4656 @item -fbtr-bb-exclusive
4657 @opindex fbtr-bb-exclusive
4658 WHen performing branch target register load optimization, don't reuse
4659 branch target registers in within any basic block.
4661 @item --param @var{name}=@var{value}
4663 In some places, GCC uses various constants to control the amount of
4664 optimization that is done. For example, GCC will not inline functions
4665 that contain more that a certain number of instructions. You can
4666 control some of these constants on the command-line using the
4667 @option{--param} option.
4669 The names of specific parameters, and the meaning of the values, are
4670 tied to the internals of the compiler, and are subject to change
4671 without notice in future releases.
4673 In each case, the @var{value} is an integer. The allowable choices for
4674 @var{name} are given in the following table:
4677 @item max-crossjump-edges
4678 The maximum number of incoming edges to consider for crossjumping.
4679 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
4680 the number of edges incoming to each block. Increasing values mean
4681 more aggressive optimization, making the compile time increase with
4682 probably small improvement in executable size.
4684 @item max-delay-slot-insn-search
4685 The maximum number of instructions to consider when looking for an
4686 instruction to fill a delay slot. If more than this arbitrary number of
4687 instructions is searched, the time savings from filling the delay slot
4688 will be minimal so stop searching. Increasing values mean more
4689 aggressive optimization, making the compile time increase with probably
4690 small improvement in executable run time.
4692 @item max-delay-slot-live-search
4693 When trying to fill delay slots, the maximum number of instructions to
4694 consider when searching for a block with valid live register
4695 information. Increasing this arbitrarily chosen value means more
4696 aggressive optimization, increasing the compile time. This parameter
4697 should be removed when the delay slot code is rewritten to maintain the
4700 @item max-gcse-memory
4701 The approximate maximum amount of memory that will be allocated in
4702 order to perform the global common subexpression elimination
4703 optimization. If more memory than specified is required, the
4704 optimization will not be done.
4706 @item max-gcse-passes
4707 The maximum number of passes of GCSE to run.
4709 @item max-pending-list-length
4710 The maximum number of pending dependencies scheduling will allow
4711 before flushing the current state and starting over. Large functions
4712 with few branches or calls can create excessively large lists which
4713 needlessly consume memory and resources.
4715 @item max-inline-insns-single
4716 Several parameters control the tree inliner used in gcc.
4717 This number sets the maximum number of instructions (counted in GCC's
4718 internal representation) in a single function that the tree inliner
4719 will consider for inlining. This only affects functions declared
4720 inline and methods implemented in a class declaration (C++).
4721 The default value is 500.
4723 @item max-inline-insns-auto
4724 When you use @option{-finline-functions} (included in @option{-O3}),
4725 a lot of functions that would otherwise not be considered for inlining
4726 by the compiler will be investigated. To those functions, a different
4727 (more restrictive) limit compared to functions declared inline can
4729 The default value is 120.
4731 @item large-function-insns
4732 The limit specifying really large functions. For functions greater than this
4733 limit inlining is constrained by @option{--param large-function-growth}.
4734 This parameter is useful primarily to avoid extreme compilation time caused by non-linear
4735 algorithms used by the backend.
4736 This parameter is ignored when @option{-funit-at-a-time} is not used.
4737 The default value is 3000.
4739 @item large-function-growth
4740 Specifies maximal growth of large function caused by inlining in percents.
4741 This parameter is ignored when @option{-funit-at-a-time} is not used.
4742 The default value is 200.
4744 @item inline-unit-growth
4745 Specifies maximal overall growth of the compilation unit caused by inlining.
4746 This parameter is ignored when @option{-funit-at-a-time} is not used.
4747 The default value is 150.
4749 @item max-inline-insns-rtl
4750 For languages that use the RTL inliner (this happens at a later stage
4751 than tree inlining), you can set the maximum allowable size (counted
4752 in RTL instructions) for the RTL inliner with this parameter.
4753 The default value is 600.
4755 @item max-unrolled-insns
4756 The maximum number of instructions that a loop should have if that loop
4757 is unrolled, and if the loop is unrolled, it determines how many times
4758 the loop code is unrolled.
4760 @item max-average-unrolled-insns
4761 The maximum number of instructions biased by probabilities of their execution
4762 that a loop should have if that loop is unrolled, and if the loop is unrolled,
4763 it determines how many times the loop code is unrolled.
4765 @item max-unroll-times
4766 The maximum number of unrollings of a single loop.
4768 @item max-peeled-insns
4769 The maximum number of instructions that a loop should have if that loop
4770 is peeled, and if the loop is peeled, it determines how many times
4771 the loop code is peeled.
4773 @item max-peel-times
4774 The maximum number of peelings of a single loop.
4776 @item max-completely-peeled-insns
4777 The maximum number of insns of a completely peeled loop.
4779 @item max-completely-peel-times
4780 The maximum number of iterations of a loop to be suitable for complete peeling.
4782 @item max-unswitch-insns
4783 The maximum number of insns of an unswitched loop.
4785 @item max-unswitch-level
4786 The maximum number of branches unswitched in a single loop.
4788 @item hot-bb-count-fraction
4789 Select fraction of the maximal count of repetitions of basic block in program
4790 given basic block needs to have to be considered hot.
4792 @item hot-bb-frequency-fraction
4793 Select fraction of the maximal frequency of executions of basic block in
4794 function given basic block needs to have to be considered hot
4796 @item tracer-dynamic-coverage
4797 @itemx tracer-dynamic-coverage-feedback
4799 This value is used to limit superblock formation once the given percentage of
4800 executed instructions is covered. This limits unnecessary code size
4803 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4804 feedback is available. The real profiles (as opposed to statically estimated
4805 ones) are much less balanced allowing the threshold to be larger value.
4807 @item tracer-max-code-growth
4808 Stop tail duplication once code growth has reached given percentage. This is
4809 rather hokey argument, as most of the duplicates will be eliminated later in
4810 cross jumping, so it may be set to much higher values than is the desired code
4813 @item tracer-min-branch-ratio
4815 Stop reverse growth when the reverse probability of best edge is less than this
4816 threshold (in percent).
4818 @item tracer-min-branch-ratio
4819 @itemx tracer-min-branch-ratio-feedback
4821 Stop forward growth if the best edge do have probability lower than this
4824 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
4825 compilation for profile feedback and one for compilation without. The value
4826 for compilation with profile feedback needs to be more conservative (higher) in
4827 order to make tracer effective.
4829 @item max-cse-path-length
4831 Maximum number of basic blocks on path that cse considers.
4833 @item ggc-min-expand
4835 GCC uses a garbage collector to manage its own memory allocation. This
4836 parameter specifies the minimum percentage by which the garbage
4837 collector's heap should be allowed to expand between collections.
4838 Tuning this may improve compilation speed; it has no effect on code
4841 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
4842 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
4843 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
4844 GCC is not able to calculate RAM on a particular platform, the lower
4845 bound of 30% is used. Setting this parameter and
4846 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4847 every opportunity. This is extremely slow, but can be useful for
4850 @item ggc-min-heapsize
4852 Minimum size of the garbage collector's heap before it begins bothering
4853 to collect garbage. The first collection occurs after the heap expands
4854 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4855 tuning this may improve compilation speed, and has no effect on code
4858 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
4859 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
4860 available, the notion of "RAM" is the smallest of actual RAM,
4861 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
4862 RAM on a particular platform, the lower bound is used. Setting this
4863 parameter very large effectively disables garbage collection. Setting
4864 this parameter and @option{ggc-min-expand} to zero causes a full
4865 collection to occur at every opportunity.
4867 @item max-reload-search-insns
4868 The maximum number of instruction reload should look backward for equivalent
4869 register. Increasing values mean more aggressive optimization, making the
4870 compile time increase with probably slightly better performance. The default
4873 @item max-cselib-memory-location
4874 The maximum number of memory locations cselib should take into acount.
4875 Increasing values mean more aggressive optimization, making the compile time
4876 increase with probably slightly better performance. The default value is 500.
4878 @item reorder-blocks-duplicate
4879 @itemx reorder-blocks-duplicate-feedback
4881 Used by basic block reordering pass to decide whether to use unconditional
4882 branch or duplicate the code on its destination. Code is duplicated when its
4883 estimated size is smaller than this value multiplied by the estimated size of
4884 unconditional jump in the hot spots of the program.
4886 The @option{reorder-block-duplicate-feedback} is used only when profile
4887 feedback is available and may be set to higher values than
4888 @option{reorder-block-duplicate} since information about the hot spots is more
4893 @node Preprocessor Options
4894 @section Options Controlling the Preprocessor
4895 @cindex preprocessor options
4896 @cindex options, preprocessor
4898 These options control the C preprocessor, which is run on each C source
4899 file before actual compilation.
4901 If you use the @option{-E} option, nothing is done except preprocessing.
4902 Some of these options make sense only together with @option{-E} because
4903 they cause the preprocessor output to be unsuitable for actual
4908 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4909 and pass @var{option} directly through to the preprocessor. If
4910 @var{option} contains commas, it is split into multiple options at the
4911 commas. However, many options are modified, translated or interpreted
4912 by the compiler driver before being passed to the preprocessor, and
4913 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4914 interface is undocumented and subject to change, so whenever possible
4915 you should avoid using @option{-Wp} and let the driver handle the
4918 @item -Xpreprocessor @var{option}
4919 @opindex preprocessor
4920 Pass @var{option} as an option to the preprocessor. You can use this to
4921 supply system-specific preprocessor options which GCC does not know how to
4924 If you want to pass an option that takes an argument, you must use
4925 @option{-Xpreprocessor} twice, once for the option and once for the argument.
4928 @include cppopts.texi
4930 @node Assembler Options
4931 @section Passing Options to the Assembler
4933 @c prevent bad page break with this line
4934 You can pass options to the assembler.
4937 @item -Wa,@var{option}
4939 Pass @var{option} as an option to the assembler. If @var{option}
4940 contains commas, it is split into multiple options at the commas.
4942 @item -Xassembler @var{option}
4944 Pass @var{option} as an option to the assembler. You can use this to
4945 supply system-specific assembler options which GCC does not know how to
4948 If you want to pass an option that takes an argument, you must use
4949 @option{-Xassembler} twice, once for the option and once for the argument.
4954 @section Options for Linking
4955 @cindex link options
4956 @cindex options, linking
4958 These options come into play when the compiler links object files into
4959 an executable output file. They are meaningless if the compiler is
4960 not doing a link step.
4964 @item @var{object-file-name}
4965 A file name that does not end in a special recognized suffix is
4966 considered to name an object file or library. (Object files are
4967 distinguished from libraries by the linker according to the file
4968 contents.) If linking is done, these object files are used as input
4977 If any of these options is used, then the linker is not run, and
4978 object file names should not be used as arguments. @xref{Overall
4982 @item -l@var{library}
4983 @itemx -l @var{library}
4985 Search the library named @var{library} when linking. (The second
4986 alternative with the library as a separate argument is only for
4987 POSIX compliance and is not recommended.)
4989 It makes a difference where in the command you write this option; the
4990 linker searches and processes libraries and object files in the order they
4991 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4992 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4993 to functions in @samp{z}, those functions may not be loaded.
4995 The linker searches a standard list of directories for the library,
4996 which is actually a file named @file{lib@var{library}.a}. The linker
4997 then uses this file as if it had been specified precisely by name.
4999 The directories searched include several standard system directories
5000 plus any that you specify with @option{-L}.
5002 Normally the files found this way are library files---archive files
5003 whose members are object files. The linker handles an archive file by
5004 scanning through it for members which define symbols that have so far
5005 been referenced but not defined. But if the file that is found is an
5006 ordinary object file, it is linked in the usual fashion. The only
5007 difference between using an @option{-l} option and specifying a file name
5008 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
5009 and searches several directories.
5013 You need this special case of the @option{-l} option in order to
5014 link an Objective-C program.
5017 @opindex nostartfiles
5018 Do not use the standard system startup files when linking.
5019 The standard system libraries are used normally, unless @option{-nostdlib}
5020 or @option{-nodefaultlibs} is used.
5022 @item -nodefaultlibs
5023 @opindex nodefaultlibs
5024 Do not use the standard system libraries when linking.
5025 Only the libraries you specify will be passed to the linker.
5026 The standard startup files are used normally, unless @option{-nostartfiles}
5027 is used. The compiler may generate calls to memcmp, memset, and memcpy
5028 for System V (and ISO C) environments or to bcopy and bzero for
5029 BSD environments. These entries are usually resolved by entries in
5030 libc. These entry points should be supplied through some other
5031 mechanism when this option is specified.
5035 Do not use the standard system startup files or libraries when linking.
5036 No startup files and only the libraries you specify will be passed to
5037 the linker. The compiler may generate calls to memcmp, memset, and memcpy
5038 for System V (and ISO C) environments or to bcopy and bzero for
5039 BSD environments. These entries are usually resolved by entries in
5040 libc. These entry points should be supplied through some other
5041 mechanism when this option is specified.
5043 @cindex @option{-lgcc}, use with @option{-nostdlib}
5044 @cindex @option{-nostdlib} and unresolved references
5045 @cindex unresolved references and @option{-nostdlib}
5046 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5047 @cindex @option{-nodefaultlibs} and unresolved references
5048 @cindex unresolved references and @option{-nodefaultlibs}
5049 One of the standard libraries bypassed by @option{-nostdlib} and
5050 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5051 that GCC uses to overcome shortcomings of particular machines, or special
5052 needs for some languages.
5053 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5054 Collection (GCC) Internals},
5055 for more discussion of @file{libgcc.a}.)
5056 In most cases, you need @file{libgcc.a} even when you want to avoid
5057 other standard libraries. In other words, when you specify @option{-nostdlib}
5058 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5059 This ensures that you have no unresolved references to internal GCC
5060 library subroutines. (For example, @samp{__main}, used to ensure C++
5061 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5062 GNU Compiler Collection (GCC) Internals}.)
5066 Produce a position independent executable on targets which support it.
5067 For predictable results, you must also specify the same set of options
5068 that were used to generate code (@option{-fpie}, @option{-fPIE},
5069 or model suboptions) when you specify this option.
5073 Remove all symbol table and relocation information from the executable.
5077 On systems that support dynamic linking, this prevents linking with the shared
5078 libraries. On other systems, this option has no effect.
5082 Produce a shared object which can then be linked with other objects to
5083 form an executable. Not all systems support this option. For predictable
5084 results, you must also specify the same set of options that were used to
5085 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5086 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5087 needs to build supplementary stub code for constructors to work. On
5088 multi-libbed systems, @samp{gcc -shared} must select the correct support
5089 libraries to link against. Failing to supply the correct flags may lead
5090 to subtle defects. Supplying them in cases where they are not necessary
5093 @item -shared-libgcc
5094 @itemx -static-libgcc
5095 @opindex shared-libgcc
5096 @opindex static-libgcc
5097 On systems that provide @file{libgcc} as a shared library, these options
5098 force the use of either the shared or static version respectively.
5099 If no shared version of @file{libgcc} was built when the compiler was
5100 configured, these options have no effect.
5102 There are several situations in which an application should use the
5103 shared @file{libgcc} instead of the static version. The most common
5104 of these is when the application wishes to throw and catch exceptions
5105 across different shared libraries. In that case, each of the libraries
5106 as well as the application itself should use the shared @file{libgcc}.
5108 Therefore, the G++ and GCJ drivers automatically add
5109 @option{-shared-libgcc} whenever you build a shared library or a main
5110 executable, because C++ and Java programs typically use exceptions, so
5111 this is the right thing to do.
5113 If, instead, you use the GCC driver to create shared libraries, you may
5114 find that they will not always be linked with the shared @file{libgcc}.
5115 If GCC finds, at its configuration time, that you have a GNU linker that
5116 does not support option @option{--eh-frame-hdr}, it will link the shared
5117 version of @file{libgcc} into shared libraries by default. Otherwise,
5118 it will take advantage of the linker and optimize away the linking with
5119 the shared version of @file{libgcc}, linking with the static version of
5120 libgcc by default. This allows exceptions to propagate through such
5121 shared libraries, without incurring relocation costs at library load
5124 However, if a library or main executable is supposed to throw or catch
5125 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5126 for the languages used in the program, or using the option
5127 @option{-shared-libgcc}, such that it is linked with the shared
5132 Bind references to global symbols when building a shared object. Warn
5133 about any unresolved references (unless overridden by the link editor
5134 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5137 @item -Xlinker @var{option}
5139 Pass @var{option} as an option to the linker. You can use this to
5140 supply system-specific linker options which GCC does not know how to
5143 If you want to pass an option that takes an argument, you must use
5144 @option{-Xlinker} twice, once for the option and once for the argument.
5145 For example, to pass @option{-assert definitions}, you must write
5146 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5147 @option{-Xlinker "-assert definitions"}, because this passes the entire
5148 string as a single argument, which is not what the linker expects.
5150 @item -Wl,@var{option}
5152 Pass @var{option} as an option to the linker. If @var{option} contains
5153 commas, it is split into multiple options at the commas.
5155 @item -u @var{symbol}
5157 Pretend the symbol @var{symbol} is undefined, to force linking of
5158 library modules to define it. You can use @option{-u} multiple times with
5159 different symbols to force loading of additional library modules.
5162 @node Directory Options
5163 @section Options for Directory Search
5164 @cindex directory options
5165 @cindex options, directory search
5168 These options specify directories to search for header files, for
5169 libraries and for parts of the compiler:
5174 Add the directory @var{dir} to the head of the list of directories to be
5175 searched for header files. This can be used to override a system header
5176 file, substituting your own version, since these directories are
5177 searched before the system header file directories. However, you should
5178 not use this option to add directories that contain vendor-supplied
5179 system header files (use @option{-isystem} for that). If you use more than
5180 one @option{-I} option, the directories are scanned in left-to-right
5181 order; the standard system directories come after.
5183 If a standard system include directory, or a directory specified with
5184 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5185 option will be ignored. The directory will still be searched but as a
5186 system directory at its normal position in the system include chain.
5187 This is to ensure that GCC's procedure to fix buggy system headers and
5188 the ordering for the include_next directive are not inadvertently changed.
5189 If you really need to change the search order for system directories,
5190 use the @option{-nostdinc} and/or @option{-isystem} options.
5194 Any directories you specify with @option{-I} options before the @option{-I-}
5195 option are searched only for the case of @samp{#include "@var{file}"};
5196 they are not searched for @samp{#include <@var{file}>}.
5198 If additional directories are specified with @option{-I} options after
5199 the @option{-I-}, these directories are searched for all @samp{#include}
5200 directives. (Ordinarily @emph{all} @option{-I} directories are used
5203 In addition, the @option{-I-} option inhibits the use of the current
5204 directory (where the current input file came from) as the first search
5205 directory for @samp{#include "@var{file}"}. There is no way to
5206 override this effect of @option{-I-}. With @option{-I.} you can specify
5207 searching the directory which was current when the compiler was
5208 invoked. That is not exactly the same as what the preprocessor does
5209 by default, but it is often satisfactory.
5211 @option{-I-} does not inhibit the use of the standard system directories
5212 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5217 Add directory @var{dir} to the list of directories to be searched
5220 @item -B@var{prefix}
5222 This option specifies where to find the executables, libraries,
5223 include files, and data files of the compiler itself.
5225 The compiler driver program runs one or more of the subprograms
5226 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5227 @var{prefix} as a prefix for each program it tries to run, both with and
5228 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5230 For each subprogram to be run, the compiler driver first tries the
5231 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5232 was not specified, the driver tries two standard prefixes, which are
5233 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5234 those results in a file name that is found, the unmodified program
5235 name is searched for using the directories specified in your
5236 @env{PATH} environment variable.
5238 The compiler will check to see if the path provided by the @option{-B}
5239 refers to a directory, and if necessary it will add a directory
5240 separator character at the end of the path.
5242 @option{-B} prefixes that effectively specify directory names also apply
5243 to libraries in the linker, because the compiler translates these
5244 options into @option{-L} options for the linker. They also apply to
5245 includes files in the preprocessor, because the compiler translates these
5246 options into @option{-isystem} options for the preprocessor. In this case,
5247 the compiler appends @samp{include} to the prefix.
5249 The run-time support file @file{libgcc.a} can also be searched for using
5250 the @option{-B} prefix, if needed. If it is not found there, the two
5251 standard prefixes above are tried, and that is all. The file is left
5252 out of the link if it is not found by those means.
5254 Another way to specify a prefix much like the @option{-B} prefix is to use
5255 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5258 As a special kludge, if the path provided by @option{-B} is
5259 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5260 9, then it will be replaced by @file{[dir/]include}. This is to help
5261 with boot-strapping the compiler.
5263 @item -specs=@var{file}
5265 Process @var{file} after the compiler reads in the standard @file{specs}
5266 file, in order to override the defaults that the @file{gcc} driver
5267 program uses when determining what switches to pass to @file{cc1},
5268 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5269 @option{-specs=@var{file}} can be specified on the command line, and they
5270 are processed in order, from left to right.
5276 @section Specifying subprocesses and the switches to pass to them
5279 @command{gcc} is a driver program. It performs its job by invoking a
5280 sequence of other programs to do the work of compiling, assembling and
5281 linking. GCC interprets its command-line parameters and uses these to
5282 deduce which programs it should invoke, and which command-line options
5283 it ought to place on their command lines. This behavior is controlled
5284 by @dfn{spec strings}. In most cases there is one spec string for each
5285 program that GCC can invoke, but a few programs have multiple spec
5286 strings to control their behavior. The spec strings built into GCC can
5287 be overridden by using the @option{-specs=} command-line switch to specify
5290 @dfn{Spec files} are plaintext files that are used to construct spec
5291 strings. They consist of a sequence of directives separated by blank
5292 lines. The type of directive is determined by the first non-whitespace
5293 character on the line and it can be one of the following:
5296 @item %@var{command}
5297 Issues a @var{command} to the spec file processor. The commands that can
5301 @item %include <@var{file}>
5303 Search for @var{file} and insert its text at the current point in the
5306 @item %include_noerr <@var{file}>
5307 @cindex %include_noerr
5308 Just like @samp{%include}, but do not generate an error message if the include
5309 file cannot be found.
5311 @item %rename @var{old_name} @var{new_name}
5313 Rename the spec string @var{old_name} to @var{new_name}.
5317 @item *[@var{spec_name}]:
5318 This tells the compiler to create, override or delete the named spec
5319 string. All lines after this directive up to the next directive or
5320 blank line are considered to be the text for the spec string. If this
5321 results in an empty string then the spec will be deleted. (Or, if the
5322 spec did not exist, then nothing will happened.) Otherwise, if the spec
5323 does not currently exist a new spec will be created. If the spec does
5324 exist then its contents will be overridden by the text of this
5325 directive, unless the first character of that text is the @samp{+}
5326 character, in which case the text will be appended to the spec.
5328 @item [@var{suffix}]:
5329 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5330 and up to the next directive or blank line are considered to make up the
5331 spec string for the indicated suffix. When the compiler encounters an
5332 input file with the named suffix, it will processes the spec string in
5333 order to work out how to compile that file. For example:
5340 This says that any input file whose name ends in @samp{.ZZ} should be
5341 passed to the program @samp{z-compile}, which should be invoked with the
5342 command-line switch @option{-input} and with the result of performing the
5343 @samp{%i} substitution. (See below.)
5345 As an alternative to providing a spec string, the text that follows a
5346 suffix directive can be one of the following:
5349 @item @@@var{language}
5350 This says that the suffix is an alias for a known @var{language}. This is
5351 similar to using the @option{-x} command-line switch to GCC to specify a
5352 language explicitly. For example:
5359 Says that .ZZ files are, in fact, C++ source files.
5362 This causes an error messages saying:
5365 @var{name} compiler not installed on this system.
5369 GCC already has an extensive list of suffixes built into it.
5370 This directive will add an entry to the end of the list of suffixes, but
5371 since the list is searched from the end backwards, it is effectively
5372 possible to override earlier entries using this technique.
5376 GCC has the following spec strings built into it. Spec files can
5377 override these strings or create their own. Note that individual
5378 targets can also add their own spec strings to this list.
5381 asm Options to pass to the assembler
5382 asm_final Options to pass to the assembler post-processor
5383 cpp Options to pass to the C preprocessor
5384 cc1 Options to pass to the C compiler
5385 cc1plus Options to pass to the C++ compiler
5386 endfile Object files to include at the end of the link
5387 link Options to pass to the linker
5388 lib Libraries to include on the command line to the linker
5389 libgcc Decides which GCC support library to pass to the linker
5390 linker Sets the name of the linker
5391 predefines Defines to be passed to the C preprocessor
5392 signed_char Defines to pass to CPP to say whether @code{char} is signed
5394 startfile Object files to include at the start of the link
5397 Here is a small example of a spec file:
5403 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5406 This example renames the spec called @samp{lib} to @samp{old_lib} and
5407 then overrides the previous definition of @samp{lib} with a new one.
5408 The new definition adds in some extra command-line options before
5409 including the text of the old definition.
5411 @dfn{Spec strings} are a list of command-line options to be passed to their
5412 corresponding program. In addition, the spec strings can contain
5413 @samp{%}-prefixed sequences to substitute variable text or to
5414 conditionally insert text into the command line. Using these constructs
5415 it is possible to generate quite complex command lines.
5417 Here is a table of all defined @samp{%}-sequences for spec
5418 strings. Note that spaces are not generated automatically around the
5419 results of expanding these sequences. Therefore you can concatenate them
5420 together or combine them with constant text in a single argument.
5424 Substitute one @samp{%} into the program name or argument.
5427 Substitute the name of the input file being processed.
5430 Substitute the basename of the input file being processed.
5431 This is the substring up to (and not including) the last period
5432 and not including the directory.
5435 This is the same as @samp{%b}, but include the file suffix (text after
5439 Marks the argument containing or following the @samp{%d} as a
5440 temporary file name, so that that file will be deleted if GCC exits
5441 successfully. Unlike @samp{%g}, this contributes no text to the
5444 @item %g@var{suffix}
5445 Substitute a file name that has suffix @var{suffix} and is chosen
5446 once per compilation, and mark the argument in the same way as
5447 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5448 name is now chosen in a way that is hard to predict even when previously
5449 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5450 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5451 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5452 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5453 was simply substituted with a file name chosen once per compilation,
5454 without regard to any appended suffix (which was therefore treated
5455 just like ordinary text), making such attacks more likely to succeed.
5457 @item %u@var{suffix}
5458 Like @samp{%g}, but generates a new temporary file name even if
5459 @samp{%u@var{suffix}} was already seen.
5461 @item %U@var{suffix}
5462 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5463 new one if there is no such last file name. In the absence of any
5464 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5465 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5466 would involve the generation of two distinct file names, one
5467 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5468 simply substituted with a file name chosen for the previous @samp{%u},
5469 without regard to any appended suffix.
5471 @item %j@var{suffix}
5472 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5473 writable, and if save-temps is off; otherwise, substitute the name
5474 of a temporary file, just like @samp{%u}. This temporary file is not
5475 meant for communication between processes, but rather as a junk
5478 @item %|@var{suffix}
5479 @itemx %m@var{suffix}
5480 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5481 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5482 all. These are the two most common ways to instruct a program that it
5483 should read from standard input or write to standard output. If you
5484 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5485 construct: see for example @file{f/lang-specs.h}.
5487 @item %.@var{SUFFIX}
5488 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5489 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5490 terminated by the next space or %.
5493 Marks the argument containing or following the @samp{%w} as the
5494 designated output file of this compilation. This puts the argument
5495 into the sequence of arguments that @samp{%o} will substitute later.
5498 Substitutes the names of all the output files, with spaces
5499 automatically placed around them. You should write spaces
5500 around the @samp{%o} as well or the results are undefined.
5501 @samp{%o} is for use in the specs for running the linker.
5502 Input files whose names have no recognized suffix are not compiled
5503 at all, but they are included among the output files, so they will
5507 Substitutes the suffix for object files. Note that this is
5508 handled specially when it immediately follows @samp{%g, %u, or %U},
5509 because of the need for those to form complete file names. The
5510 handling is such that @samp{%O} is treated exactly as if it had already
5511 been substituted, except that @samp{%g, %u, and %U} do not currently
5512 support additional @var{suffix} characters following @samp{%O} as they would
5513 following, for example, @samp{.o}.
5516 Substitutes the standard macro predefinitions for the
5517 current target machine. Use this when running @code{cpp}.
5520 Like @samp{%p}, but puts @samp{__} before and after the name of each
5521 predefined macro, except for macros that start with @samp{__} or with
5522 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5526 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5527 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5528 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5532 Current argument is the name of a library or startup file of some sort.
5533 Search for that file in a standard list of directories and substitute
5534 the full name found.
5537 Print @var{str} as an error message. @var{str} is terminated by a newline.
5538 Use this when inconsistent options are detected.
5541 Substitute the contents of spec string @var{name} at this point.
5544 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5546 @item %x@{@var{option}@}
5547 Accumulate an option for @samp{%X}.
5550 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5554 Output the accumulated assembler options specified by @option{-Wa}.
5557 Output the accumulated preprocessor options specified by @option{-Wp}.
5560 Process the @code{asm} spec. This is used to compute the
5561 switches to be passed to the assembler.
5564 Process the @code{asm_final} spec. This is a spec string for
5565 passing switches to an assembler post-processor, if such a program is
5569 Process the @code{link} spec. This is the spec for computing the
5570 command line passed to the linker. Typically it will make use of the
5571 @samp{%L %G %S %D and %E} sequences.
5574 Dump out a @option{-L} option for each directory that GCC believes might
5575 contain startup files. If the target supports multilibs then the
5576 current multilib directory will be prepended to each of these paths.
5579 Output the multilib directory with directory separators replaced with
5580 @samp{_}. If multilib directories are not set, or the multilib directory is
5581 @file{.} then this option emits nothing.
5584 Process the @code{lib} spec. This is a spec string for deciding which
5585 libraries should be included on the command line to the linker.
5588 Process the @code{libgcc} spec. This is a spec string for deciding
5589 which GCC support library should be included on the command line to the linker.
5592 Process the @code{startfile} spec. This is a spec for deciding which
5593 object files should be the first ones passed to the linker. Typically
5594 this might be a file named @file{crt0.o}.
5597 Process the @code{endfile} spec. This is a spec string that specifies
5598 the last object files that will be passed to the linker.
5601 Process the @code{cpp} spec. This is used to construct the arguments
5602 to be passed to the C preprocessor.
5605 Process the @code{signed_char} spec. This is intended to be used
5606 to tell cpp whether a char is signed. It typically has the definition:
5608 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
5612 Process the @code{cc1} spec. This is used to construct the options to be
5613 passed to the actual C compiler (@samp{cc1}).
5616 Process the @code{cc1plus} spec. This is used to construct the options to be
5617 passed to the actual C++ compiler (@samp{cc1plus}).
5620 Substitute the variable part of a matched option. See below.
5621 Note that each comma in the substituted string is replaced by
5625 Remove all occurrences of @code{-S} from the command line. Note---this
5626 command is position dependent. @samp{%} commands in the spec string
5627 before this one will see @code{-S}, @samp{%} commands in the spec string
5628 after this one will not.
5630 @item %:@var{function}(@var{args})
5631 Call the named function @var{function}, passing it @var{args}.
5632 @var{args} is first processed as a nested spec string, then split
5633 into an argument vector in the usual fashion. The function returns
5634 a string which is processed as if it had appeared literally as part
5635 of the current spec.
5637 The following built-in spec functions are provided:
5640 @item @code{if-exists}
5641 The @code{if-exists} spec function takes one argument, an absolute
5642 pathname to a file. If the file exists, @code{if-exists} returns the
5643 pathname. Here is a small example of its usage:
5647 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5650 @item @code{if-exists-else}
5651 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5652 spec function, except that it takes two arguments. The first argument is
5653 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5654 returns the pathname. If it does not exist, it returns the second argument.
5655 This way, @code{if-exists-else} can be used to select one file or another,
5656 based on the existence of the first. Here is a small example of its usage:
5660 crt0%O%s %:if-exists(crti%O%s) \
5661 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5666 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5667 If that switch was not specified, this substitutes nothing. Note that
5668 the leading dash is omitted when specifying this option, and it is
5669 automatically inserted if the substitution is performed. Thus the spec
5670 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5671 and would output the command line option @option{-foo}.
5673 @item %W@{@code{S}@}
5674 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5677 @item %@{@code{S}*@}
5678 Substitutes all the switches specified to GCC whose names start
5679 with @code{-S}, but which also take an argument. This is used for
5680 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5681 GCC considers @option{-o foo} as being
5682 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5683 text, including the space. Thus two arguments would be generated.
5685 @item %@{@code{S}*&@code{T}*@}
5686 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5687 (the order of @code{S} and @code{T} in the spec is not significant).
5688 There can be any number of ampersand-separated variables; for each the
5689 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5691 @item %@{@code{S}:@code{X}@}
5692 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
5694 @item %@{!@code{S}:@code{X}@}
5695 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
5697 @item %@{@code{S}*:@code{X}@}
5698 Substitutes @code{X} if one or more switches whose names start with
5699 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
5700 once, no matter how many such switches appeared. However, if @code{%*}
5701 appears somewhere in @code{X}, then @code{X} will be substituted once
5702 for each matching switch, with the @code{%*} replaced by the part of
5703 that switch that matched the @code{*}.
5705 @item %@{.@code{S}:@code{X}@}
5706 Substitutes @code{X}, if processing a file with suffix @code{S}.
5708 @item %@{!.@code{S}:@code{X}@}
5709 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
5711 @item %@{@code{S}|@code{P}:@code{X}@}
5712 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
5713 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
5714 although they have a stronger binding than the @samp{|}. If @code{%*}
5715 appears in @code{X}, all of the alternatives must be starred, and only
5716 the first matching alternative is substituted.
5718 For example, a spec string like this:
5721 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5724 will output the following command-line options from the following input
5725 command-line options:
5730 -d fred.c -foo -baz -boggle
5731 -d jim.d -bar -baz -boggle
5734 @item %@{S:X; T:Y; :D@}
5736 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
5737 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
5738 be as many clauses as you need. This may be combined with @code{.},
5739 @code{!}, @code{|}, and @code{*} as needed.
5744 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
5745 construct may contain other nested @samp{%} constructs or spaces, or
5746 even newlines. They are processed as usual, as described above.
5747 Trailing white space in @code{X} is ignored. White space may also
5748 appear anywhere on the left side of the colon in these constructs,
5749 except between @code{.} or @code{*} and the corresponding word.
5751 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
5752 handled specifically in these constructs. If another value of
5753 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5754 @option{-W} switch is found later in the command line, the earlier
5755 switch value is ignored, except with @{@code{S}*@} where @code{S} is
5756 just one letter, which passes all matching options.
5758 The character @samp{|} at the beginning of the predicate text is used to
5759 indicate that a command should be piped to the following command, but
5760 only if @option{-pipe} is specified.
5762 It is built into GCC which switches take arguments and which do not.
5763 (You might think it would be useful to generalize this to allow each
5764 compiler's spec to say which switches take arguments. But this cannot
5765 be done in a consistent fashion. GCC cannot even decide which input
5766 files have been specified without knowing which switches take arguments,
5767 and it must know which input files to compile in order to tell which
5770 GCC also knows implicitly that arguments starting in @option{-l} are to be
5771 treated as compiler output files, and passed to the linker in their
5772 proper position among the other output files.
5774 @c man begin OPTIONS
5776 @node Target Options
5777 @section Specifying Target Machine and Compiler Version
5778 @cindex target options
5779 @cindex cross compiling
5780 @cindex specifying machine version
5781 @cindex specifying compiler version and target machine
5782 @cindex compiler version, specifying
5783 @cindex target machine, specifying
5785 The usual way to run GCC is to run the executable called @file{gcc}, or
5786 @file{<machine>-gcc} when cross-compiling, or
5787 @file{<machine>-gcc-<version>} to run a version other than the one that
5788 was installed last. Sometimes this is inconvenient, so GCC provides
5789 options that will switch to another cross-compiler or version.
5792 @item -b @var{machine}
5794 The argument @var{machine} specifies the target machine for compilation.
5796 The value to use for @var{machine} is the same as was specified as the
5797 machine type when configuring GCC as a cross-compiler. For
5798 example, if a cross-compiler was configured with @samp{configure
5799 i386v}, meaning to compile for an 80386 running System V, then you
5800 would specify @option{-b i386v} to run that cross compiler.
5802 @item -V @var{version}
5804 The argument @var{version} specifies which version of GCC to run.
5805 This is useful when multiple versions are installed. For example,
5806 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5809 The @option{-V} and @option{-b} options work by running the
5810 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5811 use them if you can just run that directly.
5813 @node Submodel Options
5814 @section Hardware Models and Configurations
5815 @cindex submodel options
5816 @cindex specifying hardware config
5817 @cindex hardware models and configurations, specifying
5818 @cindex machine dependent options
5820 Earlier we discussed the standard option @option{-b} which chooses among
5821 different installed compilers for completely different target
5822 machines, such as VAX vs.@: 68000 vs.@: 80386.
5824 In addition, each of these target machine types can have its own
5825 special options, starting with @samp{-m}, to choose among various
5826 hardware models or configurations---for example, 68010 vs 68020,
5827 floating coprocessor or none. A single installed version of the
5828 compiler can compile for any model or configuration, according to the
5831 Some configurations of the compiler also support additional special
5832 options, usually for compatibility with other compilers on the same
5835 These options are defined by the macro @code{TARGET_SWITCHES} in the
5836 machine description. The default for the options is also defined by
5837 that macro, which enables you to change the defaults.
5847 * RS/6000 and PowerPC Options::
5850 * i386 and x86-64 Options::
5852 * DEC Alpha Options::
5853 * DEC Alpha/VMS Options::
5856 * System V Options::
5857 * TMS320C3x/C4x Options::
5864 * S/390 and zSeries Options::
5868 * Xstormy16 Options::
5873 @node M680x0 Options
5874 @subsection M680x0 Options
5875 @cindex M680x0 options
5877 These are the @samp{-m} options defined for the 68000 series. The default
5878 values for these options depends on which style of 68000 was selected when
5879 the compiler was configured; the defaults for the most common choices are
5887 Generate output for a 68000. This is the default
5888 when the compiler is configured for 68000-based systems.
5890 Use this option for microcontrollers with a 68000 or EC000 core,
5891 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5897 Generate output for a 68020. This is the default
5898 when the compiler is configured for 68020-based systems.
5902 Generate output containing 68881 instructions for floating point.
5903 This is the default for most 68020 systems unless @option{--nfp} was
5904 specified when the compiler was configured.
5908 Generate output for a 68030. This is the default when the compiler is
5909 configured for 68030-based systems.
5913 Generate output for a 68040. This is the default when the compiler is
5914 configured for 68040-based systems.
5916 This option inhibits the use of 68881/68882 instructions that have to be
5917 emulated by software on the 68040. Use this option if your 68040 does not
5918 have code to emulate those instructions.
5922 Generate output for a 68060. This is the default when the compiler is
5923 configured for 68060-based systems.
5925 This option inhibits the use of 68020 and 68881/68882 instructions that
5926 have to be emulated by software on the 68060. Use this option if your 68060
5927 does not have code to emulate those instructions.
5931 Generate output for a CPU32. This is the default
5932 when the compiler is configured for CPU32-based systems.
5934 Use this option for microcontrollers with a
5935 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5936 68336, 68340, 68341, 68349 and 68360.
5940 Generate output for a 520X ``coldfire'' family cpu. This is the default
5941 when the compiler is configured for 520X-based systems.
5943 Use this option for microcontroller with a 5200 core, including
5944 the MCF5202, MCF5203, MCF5204 and MCF5202.
5949 Generate output for a 68040, without using any of the new instructions.
5950 This results in code which can run relatively efficiently on either a
5951 68020/68881 or a 68030 or a 68040. The generated code does use the
5952 68881 instructions that are emulated on the 68040.
5956 Generate output for a 68060, without using any of the new instructions.
5957 This results in code which can run relatively efficiently on either a
5958 68020/68881 or a 68030 or a 68040. The generated code does use the
5959 68881 instructions that are emulated on the 68060.
5962 @opindex msoft-float
5963 Generate output containing library calls for floating point.
5964 @strong{Warning:} the requisite libraries are not available for all m68k
5965 targets. Normally the facilities of the machine's usual C compiler are
5966 used, but this can't be done directly in cross-compilation. You must
5967 make your own arrangements to provide suitable library functions for
5968 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5969 @samp{m68k-*-coff} do provide software floating point support.
5973 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5976 @opindex mnobitfield
5977 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5978 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5982 Do use the bit-field instructions. The @option{-m68020} option implies
5983 @option{-mbitfield}. This is the default if you use a configuration
5984 designed for a 68020.
5988 Use a different function-calling convention, in which functions
5989 that take a fixed number of arguments return with the @code{rtd}
5990 instruction, which pops their arguments while returning. This
5991 saves one instruction in the caller since there is no need to pop
5992 the arguments there.
5994 This calling convention is incompatible with the one normally
5995 used on Unix, so you cannot use it if you need to call libraries
5996 compiled with the Unix compiler.
5998 Also, you must provide function prototypes for all functions that
5999 take variable numbers of arguments (including @code{printf});
6000 otherwise incorrect code will be generated for calls to those
6003 In addition, seriously incorrect code will result if you call a
6004 function with too many arguments. (Normally, extra arguments are
6005 harmlessly ignored.)
6007 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
6008 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
6011 @itemx -mno-align-int
6013 @opindex mno-align-int
6014 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
6015 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
6016 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
6017 Aligning variables on 32-bit boundaries produces code that runs somewhat
6018 faster on processors with 32-bit busses at the expense of more memory.
6020 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
6021 align structures containing the above types differently than
6022 most published application binary interface specifications for the m68k.
6026 Use the pc-relative addressing mode of the 68000 directly, instead of
6027 using a global offset table. At present, this option implies @option{-fpic},
6028 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
6029 not presently supported with @option{-mpcrel}, though this could be supported for
6030 68020 and higher processors.
6032 @item -mno-strict-align
6033 @itemx -mstrict-align
6034 @opindex mno-strict-align
6035 @opindex mstrict-align
6036 Do not (do) assume that unaligned memory references will be handled by
6040 Generate code that allows the data segment to be located in a different
6041 area of memory from the text segment. This allows for execute in place in
6042 an environment without virtual memory management. This option implies -fPIC.
6045 Generate code that assumes that the data segment follows the text segment.
6046 This is the default.
6048 @item -mid-shared-library
6049 Generate code that supports shared libraries via the library ID method.
6050 This allows for execute in place and shared libraries in an environment
6051 without virtual memory management. This option implies -fPIC.
6053 @item -mno-id-shared-library
6054 Generate code that doesn't assume ID based shared libraries are being used.
6055 This is the default.
6057 @item -mshared-library-id=n
6058 Specified the identification number of the ID based shared library being
6059 compiled. Specifying a value of 0 will generate more compact code, specifying
6060 other values will force the allocation of that number to the current
6061 library but is no more space or time efficient than omitting this option.
6065 @node M68hc1x Options
6066 @subsection M68hc1x Options
6067 @cindex M68hc1x options
6069 These are the @samp{-m} options defined for the 68hc11 and 68hc12
6070 microcontrollers. The default values for these options depends on
6071 which style of microcontroller was selected when the compiler was configured;
6072 the defaults for the most common choices are given below.
6079 Generate output for a 68HC11. This is the default
6080 when the compiler is configured for 68HC11-based systems.
6086 Generate output for a 68HC12. This is the default
6087 when the compiler is configured for 68HC12-based systems.
6093 Generate output for a 68HCS12.
6096 @opindex mauto-incdec
6097 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
6104 Enable the use of 68HC12 min and max instructions.
6107 @itemx -mno-long-calls
6108 @opindex mlong-calls
6109 @opindex mno-long-calls
6110 Treat all calls as being far away (near). If calls are assumed to be
6111 far away, the compiler will use the @code{call} instruction to
6112 call a function and the @code{rtc} instruction for returning.
6116 Consider type @code{int} to be 16 bits wide, like @code{short int}.
6118 @item -msoft-reg-count=@var{count}
6119 @opindex msoft-reg-count
6120 Specify the number of pseudo-soft registers which are used for the
6121 code generation. The maximum number is 32. Using more pseudo-soft
6122 register may or may not result in better code depending on the program.
6123 The default is 4 for 68HC11 and 2 for 68HC12.
6128 @subsection VAX Options
6131 These @samp{-m} options are defined for the VAX:
6136 Do not output certain jump instructions (@code{aobleq} and so on)
6137 that the Unix assembler for the VAX cannot handle across long
6142 Do output those jump instructions, on the assumption that you
6143 will assemble with the GNU assembler.
6147 Output code for g-format floating point numbers instead of d-format.
6151 @subsection SPARC Options
6152 @cindex SPARC options
6154 These @samp{-m} options are supported on the SPARC:
6159 @opindex mno-app-regs
6161 Specify @option{-mapp-regs} to generate output using the global registers
6162 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
6165 To be fully SVR4 ABI compliant at the cost of some performance loss,
6166 specify @option{-mno-app-regs}. You should compile libraries and system
6167 software with this option.
6172 @opindex mhard-float
6173 Generate output containing floating point instructions. This is the
6179 @opindex msoft-float
6180 Generate output containing library calls for floating point.
6181 @strong{Warning:} the requisite libraries are not available for all SPARC
6182 targets. Normally the facilities of the machine's usual C compiler are
6183 used, but this cannot be done directly in cross-compilation. You must make
6184 your own arrangements to provide suitable library functions for
6185 cross-compilation. The embedded targets @samp{sparc-*-aout} and
6186 @samp{sparclite-*-*} do provide software floating point support.
6188 @option{-msoft-float} changes the calling convention in the output file;
6189 therefore, it is only useful if you compile @emph{all} of a program with
6190 this option. In particular, you need to compile @file{libgcc.a}, the
6191 library that comes with GCC, with @option{-msoft-float} in order for
6194 @item -mhard-quad-float
6195 @opindex mhard-quad-float
6196 Generate output containing quad-word (long double) floating point
6199 @item -msoft-quad-float
6200 @opindex msoft-quad-float
6201 Generate output containing library calls for quad-word (long double)
6202 floating point instructions. The functions called are those specified
6203 in the SPARC ABI@. This is the default.
6205 As of this writing, there are no SPARC implementations that have hardware
6206 support for the quad-word floating point instructions. They all invoke
6207 a trap handler for one of these instructions, and then the trap handler
6208 emulates the effect of the instruction. Because of the trap handler overhead,
6209 this is much slower than calling the ABI library routines. Thus the
6210 @option{-msoft-quad-float} option is the default.
6212 @item -mno-unaligned-doubles
6213 @itemx -munaligned-doubles
6214 @opindex mno-unaligned-doubles
6215 @opindex munaligned-doubles
6216 Assume that doubles have 8 byte alignment. This is the default.
6218 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
6219 alignment only if they are contained in another type, or if they have an
6220 absolute address. Otherwise, it assumes they have 4 byte alignment.
6221 Specifying this option avoids some rare compatibility problems with code
6222 generated by other compilers. It is not the default because it results
6223 in a performance loss, especially for floating point code.
6225 @item -mno-faster-structs
6226 @itemx -mfaster-structs
6227 @opindex mno-faster-structs
6228 @opindex mfaster-structs
6229 With @option{-mfaster-structs}, the compiler assumes that structures
6230 should have 8 byte alignment. This enables the use of pairs of
6231 @code{ldd} and @code{std} instructions for copies in structure
6232 assignment, in place of twice as many @code{ld} and @code{st} pairs.
6233 However, the use of this changed alignment directly violates the SPARC
6234 ABI@. Thus, it's intended only for use on targets where the developer
6235 acknowledges that their resulting code will not be directly in line with
6236 the rules of the ABI@.
6239 @opindex mimpure-text
6240 @option{-mimpure-text}, used in addition to @option{-shared}, tells
6241 the compiler to not pass @option{-z text} to the linker when linking a
6242 shared object. Using this option, you can link position-dependent
6243 code into a shared object.
6245 @option{-mimpure-text} suppresses the ``relocations remain against
6246 allocatable but non-writable sections'' linker error message.
6247 However, the necessary relocations will trigger copy-on-write, and the
6248 shared object is not actually shared across processes. Instead of
6249 using @option{-mimpure-text}, you should compile all source code with
6250 @option{-fpic} or @option{-fPIC}.
6252 This option is only available on SunOS and Solaris.
6254 @item -mcpu=@var{cpu_type}
6256 Set the instruction set, register set, and instruction scheduling parameters
6257 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
6258 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
6259 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
6260 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
6263 Default instruction scheduling parameters are used for values that select
6264 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
6265 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
6267 Here is a list of each supported architecture and their supported
6272 v8: supersparc, hypersparc
6273 sparclite: f930, f934, sparclite86x
6275 v9: ultrasparc, ultrasparc3
6278 By default (unless configured otherwise), GCC generates code for the V7
6279 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
6280 additionally optimizes it for the Cypress CY7C602 chip, as used in the
6281 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
6282 SPARCStation 1, 2, IPX etc.
6284 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
6285 architecture. The only difference from V7 code is that the compiler emits
6286 the integer multiply and integer divide instructions which exist in SPARC-V8
6287 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
6288 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
6291 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
6292 the SPARC architecture. This adds the integer multiply, integer divide step
6293 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
6294 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
6295 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU. With
6296 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
6297 MB86934 chip, which is the more recent SPARClite with FPU.
6299 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
6300 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
6301 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
6302 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
6303 optimizes it for the TEMIC SPARClet chip.
6305 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
6306 architecture. This adds 64-bit integer and floating-point move instructions,
6307 3 additional floating-point condition code registers and conditional move
6308 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
6309 optimizes it for the Sun UltraSPARC I/II chips. With
6310 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
6311 Sun UltraSPARC III chip.
6313 @item -mtune=@var{cpu_type}
6315 Set the instruction scheduling parameters for machine type
6316 @var{cpu_type}, but do not set the instruction set or register set that the
6317 option @option{-mcpu=@var{cpu_type}} would.
6319 The same values for @option{-mcpu=@var{cpu_type}} can be used for
6320 @option{-mtune=@var{cpu_type}}, but the only useful values are those
6321 that select a particular cpu implementation. Those are @samp{cypress},
6322 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
6323 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
6329 @opindex -mno-v8plus
6330 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI. The
6331 difference from the V8 ABI is that the global and out registers are
6332 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
6333 mode for all SPARC-V9 processors.
6339 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
6340 Visual Instruction Set extensions. The default is @option{-mno-vis}.
6343 These @samp{-m} options are supported in addition to the above
6344 on SPARC-V9 processors in 64-bit environments:
6347 @item -mlittle-endian
6348 @opindex mlittle-endian
6349 Generate code for a processor running in little-endian mode. It is only
6350 available for a few configurations and most notably not on Solaris.
6356 Generate code for a 32-bit or 64-bit environment.
6357 The 32-bit environment sets int, long and pointer to 32 bits.
6358 The 64-bit environment sets int to 32 bits and long and pointer
6361 @item -mcmodel=medlow
6362 @opindex mcmodel=medlow
6363 Generate code for the Medium/Low code model: the program must be linked
6364 in the low 32 bits of the address space. Pointers are 64 bits.
6365 Programs can be statically or dynamically linked.
6367 @item -mcmodel=medmid
6368 @opindex mcmodel=medmid
6369 Generate code for the Medium/Middle code model: the program must be linked
6370 in the low 44 bits of the address space, the text segment must be less than
6371 2G bytes, and data segment must be within 2G of the text segment.
6372 Pointers are 64 bits.
6374 @item -mcmodel=medany
6375 @opindex mcmodel=medany
6376 Generate code for the Medium/Anywhere code model: the program may be linked
6377 anywhere in the address space, the text segment must be less than
6378 2G bytes, and data segment must be within 2G of the text segment.
6379 Pointers are 64 bits.
6381 @item -mcmodel=embmedany
6382 @opindex mcmodel=embmedany
6383 Generate code for the Medium/Anywhere code model for embedded systems:
6384 assume a 32-bit text and a 32-bit data segment, both starting anywhere
6385 (determined at link time). Register %g4 points to the base of the
6386 data segment. Pointers are still 64 bits.
6387 Programs are statically linked, PIC is not supported.
6390 @itemx -mno-stack-bias
6391 @opindex mstack-bias
6392 @opindex mno-stack-bias
6393 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
6394 frame pointer if present, are offset by @minus{}2047 which must be added back
6395 when making stack frame references. This is the default in 64-bit mode.
6396 Otherwise, assume no such offset is present.
6400 @subsection ARM Options
6403 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6408 @opindex mapcs-frame
6409 Generate a stack frame that is compliant with the ARM Procedure Call
6410 Standard for all functions, even if this is not strictly necessary for
6411 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6412 with this option will cause the stack frames not to be generated for
6413 leaf functions. The default is @option{-mno-apcs-frame}.
6417 This is a synonym for @option{-mapcs-frame}.
6421 Generate code for a processor running with a 26-bit program counter,
6422 and conforming to the function calling standards for the APCS 26-bit
6423 option. This option replaces the @option{-m2} and @option{-m3} options
6424 of previous releases of the compiler.
6428 Generate code for a processor running with a 32-bit program counter,
6429 and conforming to the function calling standards for the APCS 32-bit
6430 option. This option replaces the @option{-m6} option of previous releases
6434 @c not currently implemented
6435 @item -mapcs-stack-check
6436 @opindex mapcs-stack-check
6437 Generate code to check the amount of stack space available upon entry to
6438 every function (that actually uses some stack space). If there is
6439 insufficient space available then either the function
6440 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6441 called, depending upon the amount of stack space required. The run time
6442 system is required to provide these functions. The default is
6443 @option{-mno-apcs-stack-check}, since this produces smaller code.
6445 @c not currently implemented
6447 @opindex mapcs-float
6448 Pass floating point arguments using the float point registers. This is
6449 one of the variants of the APCS@. This option is recommended if the
6450 target hardware has a floating point unit or if a lot of floating point
6451 arithmetic is going to be performed by the code. The default is
6452 @option{-mno-apcs-float}, since integer only code is slightly increased in
6453 size if @option{-mapcs-float} is used.
6455 @c not currently implemented
6456 @item -mapcs-reentrant
6457 @opindex mapcs-reentrant
6458 Generate reentrant, position independent code. The default is
6459 @option{-mno-apcs-reentrant}.
6462 @item -mthumb-interwork
6463 @opindex mthumb-interwork
6464 Generate code which supports calling between the ARM and Thumb
6465 instruction sets. Without this option the two instruction sets cannot
6466 be reliably used inside one program. The default is
6467 @option{-mno-thumb-interwork}, since slightly larger code is generated
6468 when @option{-mthumb-interwork} is specified.
6470 @item -mno-sched-prolog
6471 @opindex mno-sched-prolog
6472 Prevent the reordering of instructions in the function prolog, or the
6473 merging of those instruction with the instructions in the function's
6474 body. This means that all functions will start with a recognizable set
6475 of instructions (or in fact one of a choice from a small set of
6476 different function prologues), and this information can be used to
6477 locate the start if functions inside an executable piece of code. The
6478 default is @option{-msched-prolog}.
6481 @opindex mhard-float
6482 Generate output containing floating point instructions. This is the
6486 @opindex msoft-float
6487 Generate output containing library calls for floating point.
6488 @strong{Warning:} the requisite libraries are not available for all ARM
6489 targets. Normally the facilities of the machine's usual C compiler are
6490 used, but this cannot be done directly in cross-compilation. You must make
6491 your own arrangements to provide suitable library functions for
6494 @option{-msoft-float} changes the calling convention in the output file;
6495 therefore, it is only useful if you compile @emph{all} of a program with
6496 this option. In particular, you need to compile @file{libgcc.a}, the
6497 library that comes with GCC, with @option{-msoft-float} in order for
6500 @item -mfloat-abi=@var{name}
6502 Specifies which ABI to use for floating point values. Permissible values
6503 are: @samp{soft}, @samp{softfp} and @samp{hard}.
6505 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
6506 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
6507 of floating point instructions, but still uses the soft-float calling
6510 @item -mlittle-endian
6511 @opindex mlittle-endian
6512 Generate code for a processor running in little-endian mode. This is
6513 the default for all standard configurations.
6516 @opindex mbig-endian
6517 Generate code for a processor running in big-endian mode; the default is
6518 to compile code for a little-endian processor.
6520 @item -mwords-little-endian
6521 @opindex mwords-little-endian
6522 This option only applies when generating code for big-endian processors.
6523 Generate code for a little-endian word order but a big-endian byte
6524 order. That is, a byte order of the form @samp{32107654}. Note: this
6525 option should only be used if you require compatibility with code for
6526 big-endian ARM processors generated by versions of the compiler prior to
6529 @item -malignment-traps
6530 @opindex malignment-traps
6531 Generate code that will not trap if the MMU has alignment traps enabled.
6532 On ARM architectures prior to ARMv4, there were no instructions to
6533 access half-word objects stored in memory. However, when reading from
6534 memory a feature of the ARM architecture allows a word load to be used,
6535 even if the address is unaligned, and the processor core will rotate the
6536 data as it is being loaded. This option tells the compiler that such
6537 misaligned accesses will cause a MMU trap and that it should instead
6538 synthesize the access as a series of byte accesses. The compiler can
6539 still use word accesses to load half-word data if it knows that the
6540 address is aligned to a word boundary.
6542 This option is ignored when compiling for ARM architecture 4 or later,
6543 since these processors have instructions to directly access half-word
6546 @item -mno-alignment-traps
6547 @opindex mno-alignment-traps
6548 Generate code that assumes that the MMU will not trap unaligned
6549 accesses. This produces better code when the target instruction set
6550 does not have half-word memory operations (i.e.@: implementations prior to
6553 Note that you cannot use this option to access unaligned word objects,
6554 since the processor will only fetch one 32-bit aligned object from
6557 The default setting for most targets is @option{-mno-alignment-traps}, since
6558 this produces better code when there are no half-word memory
6559 instructions available.
6561 @item -mshort-load-bytes
6562 @itemx -mno-short-load-words
6563 @opindex mshort-load-bytes
6564 @opindex mno-short-load-words
6565 These are deprecated aliases for @option{-malignment-traps}.
6567 @item -mno-short-load-bytes
6568 @itemx -mshort-load-words
6569 @opindex mno-short-load-bytes
6570 @opindex mshort-load-words
6571 This are deprecated aliases for @option{-mno-alignment-traps}.
6573 @item -mcpu=@var{name}
6575 This specifies the name of the target ARM processor. GCC uses this name
6576 to determine what kind of instructions it can emit when generating
6577 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6578 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6579 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6580 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6581 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6582 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6583 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6584 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6585 @samp{arm920t}, @samp{arm926ejs}, @samp{arm940t}, @samp{arm9tdmi},
6586 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ejs},
6587 @samp{arm1136js}, @samp{arm1136jfs} ,@samp{xscale}, @samp{iwmmxt},
6590 @itemx -mtune=@var{name}
6592 This option is very similar to the @option{-mcpu=} option, except that
6593 instead of specifying the actual target processor type, and hence
6594 restricting which instructions can be used, it specifies that GCC should
6595 tune the performance of the code as if the target were of the type
6596 specified in this option, but still choosing the instructions that it
6597 will generate based on the cpu specified by a @option{-mcpu=} option.
6598 For some ARM implementations better performance can be obtained by using
6601 @item -march=@var{name}
6603 This specifies the name of the target ARM architecture. GCC uses this
6604 name to determine what kind of instructions it can emit when generating
6605 assembly code. This option can be used in conjunction with or instead
6606 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6607 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6608 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
6609 @samp{iwmmxt}, @samp{ep9312}.
6611 @item -mfpu=@var{name}
6612 @itemx -mfpe=@var{number}
6613 @itemx -mfp=@var{number}
6617 This specifies what floating point hardware (or hardware emulation) is
6618 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
6619 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
6620 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
6621 with older versions of GCC@.
6623 If @option{-msoft-float} is specified this specifies the format of
6624 floating point values.
6626 @item -mstructure-size-boundary=@var{n}
6627 @opindex mstructure-size-boundary
6628 The size of all structures and unions will be rounded up to a multiple
6629 of the number of bits set by this option. Permissible values are 8 and
6630 32. The default value varies for different toolchains. For the COFF
6631 targeted toolchain the default value is 8. Specifying the larger number
6632 can produce faster, more efficient code, but can also increase the size
6633 of the program. The two values are potentially incompatible. Code
6634 compiled with one value cannot necessarily expect to work with code or
6635 libraries compiled with the other value, if they exchange information
6636 using structures or unions.
6638 @item -mabort-on-noreturn
6639 @opindex mabort-on-noreturn
6640 Generate a call to the function @code{abort} at the end of a
6641 @code{noreturn} function. It will be executed if the function tries to
6645 @itemx -mno-long-calls
6646 @opindex mlong-calls
6647 @opindex mno-long-calls
6648 Tells the compiler to perform function calls by first loading the
6649 address of the function into a register and then performing a subroutine
6650 call on this register. This switch is needed if the target function
6651 will lie outside of the 64 megabyte addressing range of the offset based
6652 version of subroutine call instruction.
6654 Even if this switch is enabled, not all function calls will be turned
6655 into long calls. The heuristic is that static functions, functions
6656 which have the @samp{short-call} attribute, functions that are inside
6657 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6658 definitions have already been compiled within the current compilation
6659 unit, will not be turned into long calls. The exception to this rule is
6660 that weak function definitions, functions with the @samp{long-call}
6661 attribute or the @samp{section} attribute, and functions that are within
6662 the scope of a @samp{#pragma long_calls} directive, will always be
6663 turned into long calls.
6665 This feature is not enabled by default. Specifying
6666 @option{-mno-long-calls} will restore the default behavior, as will
6667 placing the function calls within the scope of a @samp{#pragma
6668 long_calls_off} directive. Note these switches have no effect on how
6669 the compiler generates code to handle function calls via function
6672 @item -mnop-fun-dllimport
6673 @opindex mnop-fun-dllimport
6674 Disable support for the @code{dllimport} attribute.
6676 @item -msingle-pic-base
6677 @opindex msingle-pic-base
6678 Treat the register used for PIC addressing as read-only, rather than
6679 loading it in the prologue for each function. The run-time system is
6680 responsible for initializing this register with an appropriate value
6681 before execution begins.
6683 @item -mpic-register=@var{reg}
6684 @opindex mpic-register
6685 Specify the register to be used for PIC addressing. The default is R10
6686 unless stack-checking is enabled, when R9 is used.
6688 @item -mcirrus-fix-invalid-insns
6689 @opindex mcirrus-fix-invalid-insns
6690 @opindex mno-cirrus-fix-invalid-insns
6691 Insert NOPs into the instruction stream to in order to work around
6692 problems with invalid Maverick instruction combinations. This option
6693 is only valid if the @option{-mcpu=ep9312} option has been used to
6694 enable generation of instructions for the Cirrus Maverick floating
6695 point co-processor. This option is not enabled by default, since the
6696 problem is only present in older Maverick implementations. The default
6697 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6700 @item -mpoke-function-name
6701 @opindex mpoke-function-name
6702 Write the name of each function into the text section, directly
6703 preceding the function prologue. The generated code is similar to this:
6707 .ascii "arm_poke_function_name", 0
6710 .word 0xff000000 + (t1 - t0)
6711 arm_poke_function_name
6713 stmfd sp!, @{fp, ip, lr, pc@}
6717 When performing a stack backtrace, code can inspect the value of
6718 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6719 location @code{pc - 12} and the top 8 bits are set, then we know that
6720 there is a function name embedded immediately preceding this location
6721 and has length @code{((pc[-3]) & 0xff000000)}.
6725 Generate code for the 16-bit Thumb instruction set. The default is to
6726 use the 32-bit ARM instruction set.
6729 @opindex mtpcs-frame
6730 Generate a stack frame that is compliant with the Thumb Procedure Call
6731 Standard for all non-leaf functions. (A leaf function is one that does
6732 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6734 @item -mtpcs-leaf-frame
6735 @opindex mtpcs-leaf-frame
6736 Generate a stack frame that is compliant with the Thumb Procedure Call
6737 Standard for all leaf functions. (A leaf function is one that does
6738 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6740 @item -mcallee-super-interworking
6741 @opindex mcallee-super-interworking
6742 Gives all externally visible functions in the file being compiled an ARM
6743 instruction set header which switches to Thumb mode before executing the
6744 rest of the function. This allows these functions to be called from
6745 non-interworking code.
6747 @item -mcaller-super-interworking
6748 @opindex mcaller-super-interworking
6749 Allows calls via function pointers (including virtual functions) to
6750 execute correctly regardless of whether the target code has been
6751 compiled for interworking or not. There is a small overhead in the cost
6752 of executing a function pointer if this option is enabled.
6756 @node MN10300 Options
6757 @subsection MN10300 Options
6758 @cindex MN10300 options
6760 These @option{-m} options are defined for Matsushita MN10300 architectures:
6765 Generate code to avoid bugs in the multiply instructions for the MN10300
6766 processors. This is the default.
6769 @opindex mno-mult-bug
6770 Do not generate code to avoid bugs in the multiply instructions for the
6775 Generate code which uses features specific to the AM33 processor.
6779 Do not generate code which uses features specific to the AM33 processor. This
6784 Do not link in the C run-time initialization object file.
6788 Indicate to the linker that it should perform a relaxation optimization pass
6789 to shorten branches, calls and absolute memory addresses. This option only
6790 has an effect when used on the command line for the final link step.
6792 This option makes symbolic debugging impossible.
6796 @node M32R/D Options
6797 @subsection M32R/D Options
6798 @cindex M32R/D options
6800 These @option{-m} options are defined for Renesas M32R/D architectures:
6805 Generate code for the M32R/2@.
6809 Generate code for the M32R/X@.
6813 Generate code for the M32R@. This is the default.
6816 @opindex mmodel=small
6817 Assume all objects live in the lower 16MB of memory (so that their addresses
6818 can be loaded with the @code{ld24} instruction), and assume all subroutines
6819 are reachable with the @code{bl} instruction.
6820 This is the default.
6822 The addressability of a particular object can be set with the
6823 @code{model} attribute.
6825 @item -mmodel=medium
6826 @opindex mmodel=medium
6827 Assume objects may be anywhere in the 32-bit address space (the compiler
6828 will generate @code{seth/add3} instructions to load their addresses), and
6829 assume all subroutines are reachable with the @code{bl} instruction.
6832 @opindex mmodel=large
6833 Assume objects may be anywhere in the 32-bit address space (the compiler
6834 will generate @code{seth/add3} instructions to load their addresses), and
6835 assume subroutines may not be reachable with the @code{bl} instruction
6836 (the compiler will generate the much slower @code{seth/add3/jl}
6837 instruction sequence).
6840 @opindex msdata=none
6841 Disable use of the small data area. Variables will be put into
6842 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6843 @code{section} attribute has been specified).
6844 This is the default.
6846 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6847 Objects may be explicitly put in the small data area with the
6848 @code{section} attribute using one of these sections.
6851 @opindex msdata=sdata
6852 Put small global and static data in the small data area, but do not
6853 generate special code to reference them.
6857 Put small global and static data in the small data area, and generate
6858 special instructions to reference them.
6862 @cindex smaller data references
6863 Put global and static objects less than or equal to @var{num} bytes
6864 into the small data or bss sections instead of the normal data or bss
6865 sections. The default value of @var{num} is 8.
6866 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6867 for this option to have any effect.
6869 All modules should be compiled with the same @option{-G @var{num}} value.
6870 Compiling with different values of @var{num} may or may not work; if it
6871 doesn't the linker will give an error message---incorrect code will not be
6876 Makes the M32R specific code in the compiler display some statistics
6877 that might help in debugging programs.
6880 @opindex malign-loops
6881 Align all loops to a 32-byte boundary.
6883 @item -mno-align-loops
6884 @opindex mno-align-loops
6885 Do not enforce a 32-byte alignment for loops. This is the default.
6887 @item -missue-rate=@var{number}
6888 @opindex missue-rate=@var{number}
6889 Issue @var{number} instructions per cycle. @var{number} can only be 1
6892 @item -mbranch-cost=@var{number}
6893 @opindex mbranch-cost=@var{number}
6894 @var{number} can only be 1 or 2. If it is 1 then branches will be
6895 preferred over conditional code, if it is 2, then the opposite will
6898 @item -mflush-trap=@var{number}
6899 @opindex mflush-trap=@var{number}
6900 Specifies the trap number to use to flush the cache. The default is
6901 12. Valid numbers are between 0 and 15 inclusive.
6903 @item -mno-flush-trap
6904 @opindex mno-flush-trap
6905 Specifies that the cache cannot be flushed by using a trap.
6907 @item -mflush-func=@var{name}
6908 @opindex mflush-func=@var{name}
6909 Specifies the name of the operating system function to call to flush
6910 the cache. The default is @emph{_flush_cache}, but a function call
6911 will only be used if a trap is not available.
6913 @item -mno-flush-func
6914 @opindex mno-flush-func
6915 Indicates that there is no OS function for flushing the cache.
6919 @node RS/6000 and PowerPC Options
6920 @subsection IBM RS/6000 and PowerPC Options
6921 @cindex RS/6000 and PowerPC Options
6922 @cindex IBM RS/6000 and PowerPC Options
6924 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6932 @itemx -mpowerpc-gpopt
6933 @itemx -mno-powerpc-gpopt
6934 @itemx -mpowerpc-gfxopt
6935 @itemx -mno-powerpc-gfxopt
6937 @itemx -mno-powerpc64
6943 @opindex mno-powerpc
6944 @opindex mpowerpc-gpopt
6945 @opindex mno-powerpc-gpopt
6946 @opindex mpowerpc-gfxopt
6947 @opindex mno-powerpc-gfxopt
6949 @opindex mno-powerpc64
6950 GCC supports two related instruction set architectures for the
6951 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6952 instructions supported by the @samp{rios} chip set used in the original
6953 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6954 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6955 the IBM 4xx microprocessors.
6957 Neither architecture is a subset of the other. However there is a
6958 large common subset of instructions supported by both. An MQ
6959 register is included in processors supporting the POWER architecture.
6961 You use these options to specify which instructions are available on the
6962 processor you are using. The default value of these options is
6963 determined when configuring GCC@. Specifying the
6964 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6965 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6966 rather than the options listed above.
6968 The @option{-mpower} option allows GCC to generate instructions that
6969 are found only in the POWER architecture and to use the MQ register.
6970 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
6971 to generate instructions that are present in the POWER2 architecture but
6972 not the original POWER architecture.
6974 The @option{-mpowerpc} option allows GCC to generate instructions that
6975 are found only in the 32-bit subset of the PowerPC architecture.
6976 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
6977 GCC to use the optional PowerPC architecture instructions in the
6978 General Purpose group, including floating-point square root. Specifying
6979 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
6980 use the optional PowerPC architecture instructions in the Graphics
6981 group, including floating-point select.
6983 The @option{-mpowerpc64} option allows GCC to generate the additional
6984 64-bit instructions that are found in the full PowerPC64 architecture
6985 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6986 @option{-mno-powerpc64}.
6988 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
6989 will use only the instructions in the common subset of both
6990 architectures plus some special AIX common-mode calls, and will not use
6991 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
6992 permits GCC to use any instruction from either architecture and to
6993 allow use of the MQ register; specify this for the Motorola MPC601.
6995 @item -mnew-mnemonics
6996 @itemx -mold-mnemonics
6997 @opindex mnew-mnemonics
6998 @opindex mold-mnemonics
6999 Select which mnemonics to use in the generated assembler code. With
7000 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
7001 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
7002 assembler mnemonics defined for the POWER architecture. Instructions
7003 defined in only one architecture have only one mnemonic; GCC uses that
7004 mnemonic irrespective of which of these options is specified.
7006 GCC defaults to the mnemonics appropriate for the architecture in
7007 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
7008 value of these option. Unless you are building a cross-compiler, you
7009 should normally not specify either @option{-mnew-mnemonics} or
7010 @option{-mold-mnemonics}, but should instead accept the default.
7012 @item -mcpu=@var{cpu_type}
7014 Set architecture type, register usage, choice of mnemonics, and
7015 instruction scheduling parameters for machine type @var{cpu_type}.
7016 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
7017 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
7018 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
7019 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
7020 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
7021 @samp{860}, @samp{970}, @samp{common}, @samp{ec603e}, @samp{G3},
7022 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
7023 @samp{power4}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
7024 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64a}.
7026 @option{-mcpu=common} selects a completely generic processor. Code
7027 generated under this option will run on any POWER or PowerPC processor.
7028 GCC will use only the instructions in the common subset of both
7029 architectures, and will not use the MQ register. GCC assumes a generic
7030 processor model for scheduling purposes.
7032 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
7033 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
7034 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
7035 types, with an appropriate, generic processor model assumed for
7036 scheduling purposes.
7038 The other options specify a specific processor. Code generated under
7039 those options will run best on that processor, and may not run at all on
7042 The @option{-mcpu} options automatically enable or disable the
7043 following options: @option{-maltivec}, @option{-mhard-float},
7044 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
7045 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
7046 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
7047 @option{-mstring}. The particular options set for any particular CPU
7048 will vary between compiler versions, depending on what setting seems
7049 to produce optimal code for that CPU; it doesn't necessarily reflect
7050 the actual hardware's capabilities. If you wish to set an individual
7051 option to a particular value, you may specify it after the
7052 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
7054 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
7055 not enabled or disabled by the @option{-mcpu} option at present, since
7056 AIX does not have full support for these options. You may still
7057 enable or disable them individually if you're sure it'll work in your
7060 @item -mtune=@var{cpu_type}
7062 Set the instruction scheduling parameters for machine type
7063 @var{cpu_type}, but do not set the architecture type, register usage, or
7064 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
7065 values for @var{cpu_type} are used for @option{-mtune} as for
7066 @option{-mcpu}. If both are specified, the code generated will use the
7067 architecture, registers, and mnemonics set by @option{-mcpu}, but the
7068 scheduling parameters set by @option{-mtune}.
7073 @opindex mno-altivec
7074 These switches enable or disable the use of built-in functions that
7075 allow access to the AltiVec instruction set. You may also need to set
7076 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
7081 Extend the current ABI with SPE ABI extensions. This does not change
7082 the default ABI, instead it adds the SPE ABI extensions to the current
7086 @opindex mabi=no-spe
7087 Disable Booke SPE ABI extensions for the current ABI.
7089 @item -misel=@var{yes/no}
7092 This switch enables or disables the generation of ISEL instructions.
7094 @item -mspe=@var{yes/no}
7097 This switch enables or disables the generation of SPE simd
7100 @item -mfloat-gprs=@var{yes/no}
7102 @opindex mfloat-gprs
7103 This switch enables or disables the generation of floating point
7104 operations on the general purpose registers for architectures that
7105 support it. This option is currently only available on the MPC8540.
7108 @itemx -mno-fp-in-toc
7109 @itemx -mno-sum-in-toc
7110 @itemx -mminimal-toc
7112 @opindex mno-fp-in-toc
7113 @opindex mno-sum-in-toc
7114 @opindex mminimal-toc
7115 Modify generation of the TOC (Table Of Contents), which is created for
7116 every executable file. The @option{-mfull-toc} option is selected by
7117 default. In that case, GCC will allocate at least one TOC entry for
7118 each unique non-automatic variable reference in your program. GCC
7119 will also place floating-point constants in the TOC@. However, only
7120 16,384 entries are available in the TOC@.
7122 If you receive a linker error message that saying you have overflowed
7123 the available TOC space, you can reduce the amount of TOC space used
7124 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
7125 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
7126 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
7127 generate code to calculate the sum of an address and a constant at
7128 run-time instead of putting that sum into the TOC@. You may specify one
7129 or both of these options. Each causes GCC to produce very slightly
7130 slower and larger code at the expense of conserving TOC space.
7132 If you still run out of space in the TOC even when you specify both of
7133 these options, specify @option{-mminimal-toc} instead. This option causes
7134 GCC to make only one TOC entry for every file. When you specify this
7135 option, GCC will produce code that is slower and larger but which
7136 uses extremely little TOC space. You may wish to use this option
7137 only on files that contain less frequently executed code.
7143 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
7144 @code{long} type, and the infrastructure needed to support them.
7145 Specifying @option{-maix64} implies @option{-mpowerpc64} and
7146 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
7147 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
7152 @opindex mno-xl-call
7153 On AIX, pass floating-point arguments to prototyped functions beyond the
7154 register save area (RSA) on the stack in addition to argument FPRs. The
7155 AIX calling convention was extended but not initially documented to
7156 handle an obscure K&R C case of calling a function that takes the
7157 address of its arguments with fewer arguments than declared. AIX XL
7158 compilers access floating point arguments which do not fit in the
7159 RSA from the stack when a subroutine is compiled without
7160 optimization. Because always storing floating-point arguments on the
7161 stack is inefficient and rarely needed, this option is not enabled by
7162 default and only is necessary when calling subroutines compiled by AIX
7163 XL compilers without optimization.
7167 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
7168 application written to use message passing with special startup code to
7169 enable the application to run. The system must have PE installed in the
7170 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
7171 must be overridden with the @option{-specs=} option to specify the
7172 appropriate directory location. The Parallel Environment does not
7173 support threads, so the @option{-mpe} option and the @option{-pthread}
7174 option are incompatible.
7176 @item -malign-natural
7177 @itemx -malign-power
7178 @opindex malign-natural
7179 @opindex malign-power
7180 On AIX, Darwin, and 64-bit PowerPC GNU/Linux, the option
7181 @option{-malign-natural} overrides the ABI-defined alignment of larger
7182 types, such as floating-point doubles, on their natural size-based boundary.
7183 The option @option{-malign-power} instructs GCC to follow the ABI-specified
7184 alignment rules. GCC defaults to the standard alignment defined in the ABI.
7188 @opindex msoft-float
7189 @opindex mhard-float
7190 Generate code that does not use (uses) the floating-point register set.
7191 Software floating point emulation is provided if you use the
7192 @option{-msoft-float} option, and pass the option to GCC when linking.
7195 @itemx -mno-multiple
7197 @opindex mno-multiple
7198 Generate code that uses (does not use) the load multiple word
7199 instructions and the store multiple word instructions. These
7200 instructions are generated by default on POWER systems, and not
7201 generated on PowerPC systems. Do not use @option{-mmultiple} on little
7202 endian PowerPC systems, since those instructions do not work when the
7203 processor is in little endian mode. The exceptions are PPC740 and
7204 PPC750 which permit the instructions usage in little endian mode.
7210 Generate code that uses (does not use) the load string instructions
7211 and the store string word instructions to save multiple registers and
7212 do small block moves. These instructions are generated by default on
7213 POWER systems, and not generated on PowerPC systems. Do not use
7214 @option{-mstring} on little endian PowerPC systems, since those
7215 instructions do not work when the processor is in little endian mode.
7216 The exceptions are PPC740 and PPC750 which permit the instructions
7217 usage in little endian mode.
7223 Generate code that uses (does not use) the load or store instructions
7224 that update the base register to the address of the calculated memory
7225 location. These instructions are generated by default. If you use
7226 @option{-mno-update}, there is a small window between the time that the
7227 stack pointer is updated and the address of the previous frame is
7228 stored, which means code that walks the stack frame across interrupts or
7229 signals may get corrupted data.
7232 @itemx -mno-fused-madd
7233 @opindex mfused-madd
7234 @opindex mno-fused-madd
7235 Generate code that uses (does not use) the floating point multiply and
7236 accumulate instructions. These instructions are generated by default if
7237 hardware floating is used.
7239 @item -mno-bit-align
7241 @opindex mno-bit-align
7243 On System V.4 and embedded PowerPC systems do not (do) force structures
7244 and unions that contain bit-fields to be aligned to the base type of the
7247 For example, by default a structure containing nothing but 8
7248 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7249 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7250 the structure would be aligned to a 1 byte boundary and be one byte in
7253 @item -mno-strict-align
7254 @itemx -mstrict-align
7255 @opindex mno-strict-align
7256 @opindex mstrict-align
7257 On System V.4 and embedded PowerPC systems do not (do) assume that
7258 unaligned memory references will be handled by the system.
7261 @itemx -mno-relocatable
7262 @opindex mrelocatable
7263 @opindex mno-relocatable
7264 On embedded PowerPC systems generate code that allows (does not allow)
7265 the program to be relocated to a different address at runtime. If you
7266 use @option{-mrelocatable} on any module, all objects linked together must
7267 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7269 @item -mrelocatable-lib
7270 @itemx -mno-relocatable-lib
7271 @opindex mrelocatable-lib
7272 @opindex mno-relocatable-lib
7273 On embedded PowerPC systems generate code that allows (does not allow)
7274 the program to be relocated to a different address at runtime. Modules
7275 compiled with @option{-mrelocatable-lib} can be linked with either modules
7276 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7277 with modules compiled with the @option{-mrelocatable} options.
7283 On System V.4 and embedded PowerPC systems do not (do) assume that
7284 register 2 contains a pointer to a global area pointing to the addresses
7285 used in the program.
7288 @itemx -mlittle-endian
7290 @opindex mlittle-endian
7291 On System V.4 and embedded PowerPC systems compile code for the
7292 processor in little endian mode. The @option{-mlittle-endian} option is
7293 the same as @option{-mlittle}.
7298 @opindex mbig-endian
7299 On System V.4 and embedded PowerPC systems compile code for the
7300 processor in big endian mode. The @option{-mbig-endian} option is
7301 the same as @option{-mbig}.
7303 @item -mdynamic-no-pic
7304 @opindex mdynamic-no-pic
7305 On Darwin and Mac OS X systems, compile code so that it is not
7306 relocatable, but that its external references are relocatable. The
7307 resulting code is suitable for applications, but not shared
7310 @item -mprioritize-restricted-insns=@var{priority}
7311 @opindex mprioritize-restricted-insns
7312 This option controls the priority that is assigned to
7313 dispatch-slot restricted instructions during the second scheduling
7314 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
7315 @var{no/highest/second-highest} priority to dispatch slot restricted
7318 @item -msched-costly-dep=@var{dependence_type}
7319 @opindex msched-costly-dep
7320 This option controls which dependences are considered costly
7321 by the target during instruction scheduling. The argument
7322 @var{dependence_type} takes one of the following values:
7323 @var{no}: no dependence is costly,
7324 @var{all}: all dependences are costly,
7325 @var{true_store_to_load}: a true dependence from store to load is costly,
7326 @var{store_to_load}: any dependence from store to load is costly,
7327 @var{number}: any dependence which latency >= @var{number} is costly.
7329 @item -minsert-sched-nops=@var{scheme}
7330 @opindex minsert-sched-nops
7331 This option controls which nop insertion scheme will be used during
7332 the second scheduling pass. The argument @var{scheme} takes one of the
7334 @var{no}: Don't insert nops.
7335 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
7336 according to the scheduler's grouping.
7337 @var{regroup_exact}: Insert nops to force costly dependent insns into
7338 separate groups. Insert exactly as many nops as needed to force an insn
7339 to a new group, according to the estimated processor grouping.
7340 @var{number}: Insert nops to force costly dependent insns into
7341 separate groups. Insert @var{number} nops to force an insn to a new group.
7345 On System V.4 and embedded PowerPC systems compile code using calling
7346 conventions that adheres to the March 1995 draft of the System V
7347 Application Binary Interface, PowerPC processor supplement. This is the
7348 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7350 @item -mcall-sysv-eabi
7351 @opindex mcall-sysv-eabi
7352 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7354 @item -mcall-sysv-noeabi
7355 @opindex mcall-sysv-noeabi
7356 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7358 @item -mcall-solaris
7359 @opindex mcall-solaris
7360 On System V.4 and embedded PowerPC systems compile code for the Solaris
7364 @opindex mcall-linux
7365 On System V.4 and embedded PowerPC systems compile code for the
7366 Linux-based GNU system.
7370 On System V.4 and embedded PowerPC systems compile code for the
7371 Hurd-based GNU system.
7374 @opindex mcall-netbsd
7375 On System V.4 and embedded PowerPC systems compile code for the
7376 NetBSD operating system.
7378 @item -maix-struct-return
7379 @opindex maix-struct-return
7380 Return all structures in memory (as specified by the AIX ABI)@.
7382 @item -msvr4-struct-return
7383 @opindex msvr4-struct-return
7384 Return structures smaller than 8 bytes in registers (as specified by the
7388 @opindex mabi=altivec
7389 Extend the current ABI with AltiVec ABI extensions. This does not
7390 change the default ABI, instead it adds the AltiVec ABI extensions to
7393 @item -mabi=no-altivec
7394 @opindex mabi=no-altivec
7395 Disable AltiVec ABI extensions for the current ABI.
7398 @itemx -mno-prototype
7400 @opindex mno-prototype
7401 On System V.4 and embedded PowerPC systems assume that all calls to
7402 variable argument functions are properly prototyped. Otherwise, the
7403 compiler must insert an instruction before every non prototyped call to
7404 set or clear bit 6 of the condition code register (@var{CR}) to
7405 indicate whether floating point values were passed in the floating point
7406 registers in case the function takes a variable arguments. With
7407 @option{-mprototype}, only calls to prototyped variable argument functions
7408 will set or clear the bit.
7412 On embedded PowerPC systems, assume that the startup module is called
7413 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7414 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7419 On embedded PowerPC systems, assume that the startup module is called
7420 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7425 On embedded PowerPC systems, assume that the startup module is called
7426 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7430 @opindex myellowknife
7431 On embedded PowerPC systems, assume that the startup module is called
7432 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7437 On System V.4 and embedded PowerPC systems, specify that you are
7438 compiling for a VxWorks system.
7442 Specify that you are compiling for the WindISS simulation environment.
7446 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7447 header to indicate that @samp{eabi} extended relocations are used.
7453 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7454 Embedded Applications Binary Interface (eabi) which is a set of
7455 modifications to the System V.4 specifications. Selecting @option{-meabi}
7456 means that the stack is aligned to an 8 byte boundary, a function
7457 @code{__eabi} is called to from @code{main} to set up the eabi
7458 environment, and the @option{-msdata} option can use both @code{r2} and
7459 @code{r13} to point to two separate small data areas. Selecting
7460 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7461 do not call an initialization function from @code{main}, and the
7462 @option{-msdata} option will only use @code{r13} to point to a single
7463 small data area. The @option{-meabi} option is on by default if you
7464 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7467 @opindex msdata=eabi
7468 On System V.4 and embedded PowerPC systems, put small initialized
7469 @code{const} global and static data in the @samp{.sdata2} section, which
7470 is pointed to by register @code{r2}. Put small initialized
7471 non-@code{const} global and static data in the @samp{.sdata} section,
7472 which is pointed to by register @code{r13}. Put small uninitialized
7473 global and static data in the @samp{.sbss} section, which is adjacent to
7474 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7475 incompatible with the @option{-mrelocatable} option. The
7476 @option{-msdata=eabi} option also sets the @option{-memb} option.
7479 @opindex msdata=sysv
7480 On System V.4 and embedded PowerPC systems, put small global and static
7481 data in the @samp{.sdata} section, which is pointed to by register
7482 @code{r13}. Put small uninitialized global and static data in the
7483 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7484 The @option{-msdata=sysv} option is incompatible with the
7485 @option{-mrelocatable} option.
7487 @item -msdata=default
7489 @opindex msdata=default
7491 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7492 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7493 same as @option{-msdata=sysv}.
7496 @opindex msdata-data
7497 On System V.4 and embedded PowerPC systems, put small global and static
7498 data in the @samp{.sdata} section. Put small uninitialized global and
7499 static data in the @samp{.sbss} section. Do not use register @code{r13}
7500 to address small data however. This is the default behavior unless
7501 other @option{-msdata} options are used.
7505 @opindex msdata=none
7507 On embedded PowerPC systems, put all initialized global and static data
7508 in the @samp{.data} section, and all uninitialized data in the
7509 @samp{.bss} section.
7513 @cindex smaller data references (PowerPC)
7514 @cindex .sdata/.sdata2 references (PowerPC)
7515 On embedded PowerPC systems, put global and static items less than or
7516 equal to @var{num} bytes into the small data or bss sections instead of
7517 the normal data or bss section. By default, @var{num} is 8. The
7518 @option{-G @var{num}} switch is also passed to the linker.
7519 All modules should be compiled with the same @option{-G @var{num}} value.
7522 @itemx -mno-regnames
7524 @opindex mno-regnames
7525 On System V.4 and embedded PowerPC systems do (do not) emit register
7526 names in the assembly language output using symbolic forms.
7529 @itemx -mno-longcall
7531 @opindex mno-longcall
7532 Default to making all function calls indirectly, using a register, so
7533 that functions which reside further than 32 megabytes (33,554,432
7534 bytes) from the current location can be called. This setting can be
7535 overridden by the @code{shortcall} function attribute, or by
7536 @code{#pragma longcall(0)}.
7538 Some linkers are capable of detecting out-of-range calls and generating
7539 glue code on the fly. On these systems, long calls are unnecessary and
7540 generate slower code. As of this writing, the AIX linker can do this,
7541 as can the GNU linker for PowerPC/64. It is planned to add this feature
7542 to the GNU linker for 32-bit PowerPC systems as well.
7544 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
7545 callee, L42'', plus a ``branch island'' (glue code). The two target
7546 addresses represent the callee and the ``branch island.'' The
7547 Darwin/PPC linker will prefer the first address and generate a ``bl
7548 callee'' if the PPC ``bl'' instruction will reach the callee directly;
7549 otherwise, the linker will generate ``bl L42'' to call the ``branch
7550 island.'' The ``branch island'' is appended to the body of the
7551 calling function; it computes the full 32-bit address of the callee
7554 On Mach-O (Darwin) systems, this option directs the compiler emit to
7555 the glue for every direct call, and the Darwin linker decides whether
7556 to use or discard it.
7558 In the future, we may cause GCC to ignore all longcall specifications
7559 when the linker is known to generate glue.
7563 Adds support for multithreading with the @dfn{pthreads} library.
7564 This option sets flags for both the preprocessor and linker.
7568 @node Darwin Options
7569 @subsection Darwin Options
7570 @cindex Darwin options
7572 These options are defined for all architectures running the Darwin operating
7573 system. They are useful for compatibility with other Mac OS compilers.
7578 Loads all members of static archive libraries.
7579 See man ld(1) for more information.
7581 @item -arch_errors_fatal
7582 @opindex arch_errors_fatal
7583 Cause the errors having to do with files that have the wrong architecture
7587 @opindex bind_at_load
7588 Causes the output file to be marked such that the dynamic linker will
7589 bind all undefined references when the file is loaded or launched.
7593 Produce a Mach-o bundle format file.
7594 See man ld(1) for more information.
7596 @item -bundle_loader @var{executable}
7597 @opindex bundle_loader
7598 This specifies the @var{executable} that will be loading the build
7599 output file being linked. See man ld(1) for more information.
7601 @item -allowable_client @var{client_name}
7605 @itemx -compatibility_version
7606 @itemx -current_version
7607 @itemx -dependency-file
7609 @itemx -dylinker_install_name
7612 @itemx -exported_symbols_list
7614 @itemx -flat_namespace
7615 @itemx -force_cpusubtype_ALL
7616 @itemx -force_flat_namespace
7617 @itemx -headerpad_max_install_names
7620 @itemx -install_name
7621 @itemx -keep_private_externs
7622 @itemx -multi_module
7623 @itemx -multiply_defined
7624 @itemx -multiply_defined_unused
7626 @itemx -nofixprebinding
7629 @itemx -noseglinkedit
7630 @itemx -pagezero_size
7632 @itemx -prebind_all_twolevel_modules
7633 @itemx -private_bundle
7634 @itemx -read_only_relocs
7636 @itemx -sectobjectsymbols
7640 @itemx -sectobjectsymbols
7642 @itemx -seg_addr_table
7643 @itemx -seg_addr_table_filename
7646 @itemx -segs_read_only_addr
7647 @itemx -segs_read_write_addr
7648 @itemx -single_module
7651 @itemx -sub_umbrella
7652 @itemx -twolevel_namespace
7655 @itemx -unexported_symbols_list
7656 @itemx -weak_reference_mismatches
7659 @opindex allowable_client
7661 @opindex client_name
7662 @opindex compatibility_version
7663 @opindex current_version
7664 @opindex dependency-file
7666 @opindex dylinker_install_name
7669 @opindex exported_symbols_list
7671 @opindex flat_namespace
7672 @opindex force_cpusubtype_ALL
7673 @opindex force_flat_namespace
7674 @opindex headerpad_max_install_names
7677 @opindex install_name
7678 @opindex keep_private_externs
7679 @opindex multi_module
7680 @opindex multiply_defined
7681 @opindex multiply_defined_unused
7683 @opindex nofixprebinding
7684 @opindex nomultidefs
7686 @opindex noseglinkedit
7687 @opindex pagezero_size
7689 @opindex prebind_all_twolevel_modules
7690 @opindex private_bundle
7691 @opindex read_only_relocs
7693 @opindex sectobjectsymbols
7697 @opindex sectobjectsymbols
7699 @opindex seg_addr_table
7700 @opindex seg_addr_table_filename
7701 @opindex seglinkedit
7703 @opindex segs_read_only_addr
7704 @opindex segs_read_write_addr
7705 @opindex single_module
7707 @opindex sub_library
7708 @opindex sub_umbrella
7709 @opindex twolevel_namespace
7712 @opindex unexported_symbols_list
7713 @opindex weak_reference_mismatches
7714 @opindex whatsloaded
7716 These options are available for Darwin linker. Darwin linker man page
7717 describes them in detail.
7722 @subsection MIPS Options
7723 @cindex MIPS options
7729 Generate big-endian code.
7733 Generate little-endian code. This is the default for @samp{mips*el-*-*}
7736 @item -march=@var{arch}
7738 Generate code that will run on @var{arch}, which can be the name of a
7739 generic MIPS ISA, or the name of a particular processor.
7741 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
7742 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
7743 The processor names are:
7744 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
7746 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
7747 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
7751 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4300},
7752 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
7753 The special value @samp{from-abi} selects the
7754 most compatible architecture for the selected ABI (that is,
7755 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
7757 In processor names, a final @samp{000} can be abbreviated as @samp{k}
7758 (for example, @samp{-march=r2k}). Prefixes are optional, and
7759 @samp{vr} may be written @samp{r}.
7761 GCC defines two macros based on the value of this option. The first
7762 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
7763 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
7764 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
7765 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
7766 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
7768 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
7769 above. In other words, it will have the full prefix and will not
7770 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
7771 the macro names the resolved architecture (either @samp{"mips1"} or
7772 @samp{"mips3"}). It names the default architecture when no
7773 @option{-march} option is given.
7775 @item -mtune=@var{arch}
7777 Optimize for @var{arch}. Among other things, this option controls
7778 the way instructions are scheduled, and the perceived cost of arithmetic
7779 operations. The list of @var{arch} values is the same as for
7782 When this option is not used, GCC will optimize for the processor
7783 specified by @option{-march}. By using @option{-march} and
7784 @option{-mtune} together, it is possible to generate code that will
7785 run on a family of processors, but optimize the code for one
7786 particular member of that family.
7788 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
7789 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
7790 @samp{-march} ones described above.
7794 Equivalent to @samp{-march=mips1}.
7798 Equivalent to @samp{-march=mips2}.
7802 Equivalent to @samp{-march=mips3}.
7806 Equivalent to @samp{-march=mips4}.
7810 Equivalent to @samp{-march=mips32}.
7814 Equivalent to @samp{-march=mips32r2}.
7818 Equivalent to @samp{-march=mips64}.
7824 Use (do not use) the MIPS16 ISA.
7836 Generate code for the given ABI@.
7838 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
7839 generates 64-bit code when you select a 64-bit architecture, but you
7840 can use @option{-mgp32} to get 32-bit code instead.
7843 @itemx -mno-abicalls
7845 @opindex mno-abicalls
7846 Generate (do not generate) SVR4-style position-independent code.
7847 @option{-mabicalls} is the default for SVR4-based systems.
7853 Lift (do not lift) the usual restrictions on the size of the global
7856 GCC normally uses a single instruction to load values from the GOT.
7857 While this is relatively efficient, it will only work if the GOT
7858 is smaller than about 64k. Anything larger will cause the linker
7859 to report an error such as:
7861 @cindex relocation truncated to fit (MIPS)
7863 relocation truncated to fit: R_MIPS_GOT16 foobar
7866 If this happens, you should recompile your code with @option{-mxgot}.
7867 It should then work with very large GOTs, although it will also be
7868 less efficient, since it will take three instructions to fetch the
7869 value of a global symbol.
7871 Note that some linkers can create multiple GOTs. If you have such a
7872 linker, you should only need to use @option{-mxgot} when a single object
7873 file accesses more than 64k's worth of GOT entries. Very few do.
7875 These options have no effect unless GCC is generating position
7878 @item -membedded-pic
7879 @itemx -mno-embedded-pic
7880 @opindex membedded-pic
7881 @opindex mno-embedded-pic
7882 Generate (do not generate) position-independent code suitable for some
7883 embedded systems. All calls are made using PC relative addresses, and
7884 all data is addressed using the $gp register. No more than 65536
7885 bytes of global data may be used. This requires GNU as and GNU ld,
7886 which do most of the work.
7890 Assume that general-purpose registers are 32 bits wide.
7894 Assume that general-purpose registers are 64 bits wide.
7898 Assume that floating-point registers are 32 bits wide.
7902 Assume that floating-point registers are 64 bits wide.
7905 @opindex mhard-float
7906 Use floating-point coprocessor instructions.
7909 @opindex msoft-float
7910 Do not use floating-point coprocessor instructions. Implement
7911 floating-point calculations using library calls instead.
7913 @item -msingle-float
7914 @opindex msingle-float
7915 Assume that the floating-point coprocessor only supports single-precision
7918 @itemx -mdouble-float
7919 @opindex mdouble-float
7920 Assume that the floating-point coprocessor supports double-precision
7921 operations. This is the default.
7925 Force @code{int} and @code{long} types to be 64 bits wide. See
7926 @option{-mlong32} for an explanation of the default and the way
7927 that the pointer size is determined.
7931 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
7932 an explanation of the default and the way that the pointer size is
7937 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
7939 The default size of @code{int}s, @code{long}s and pointers depends on
7940 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
7941 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
7942 32-bit @code{long}s. Pointers are the same size as @code{long}s,
7943 or the same size as integer registers, whichever is smaller.
7947 @cindex smaller data references (MIPS)
7948 @cindex gp-relative references (MIPS)
7949 Put global and static items less than or equal to @var{num} bytes into
7950 the small data or bss section instead of the normal data or bss section.
7951 This allows the data to be accessed using a single instruction.
7953 All modules should be compiled with the same @option{-G @var{num}}
7956 @item -membedded-data
7957 @itemx -mno-embedded-data
7958 @opindex membedded-data
7959 @opindex mno-embedded-data
7960 Allocate variables to the read-only data section first if possible, then
7961 next in the small data section if possible, otherwise in data. This gives
7962 slightly slower code than the default, but reduces the amount of RAM required
7963 when executing, and thus may be preferred for some embedded systems.
7965 @item -muninit-const-in-rodata
7966 @itemx -mno-uninit-const-in-rodata
7967 @opindex muninit-const-in-rodata
7968 @opindex mno-uninit-const-in-rodata
7969 Put uninitialized @code{const} variables in the read-only data section.
7970 This option is only meaningful in conjunction with @option{-membedded-data}.
7972 @item -msplit-addresses
7973 @itemx -mno-split-addresses
7974 @opindex msplit-addresses
7975 @opindex mno-split-addresses
7976 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
7977 relocation operators. This option has been superceded by
7978 @option{-mexplicit-relocs} but is retained for backwards compatibility.
7980 @item -mexplicit-relocs
7981 @itemx -mno-explicit-relocs
7982 @opindex mexplicit-relocs
7983 @opindex mno-explicit-relocs
7984 Use (do not use) assembler relocation operators when dealing with symbolic
7985 addresses. The alternative, selected by @option{-mno-explicit-relocs},
7986 is to use assembler macros instead.
7988 @option{-mexplicit-relocs} is usually the default if GCC was configured
7989 to use an assembler that supports relocation operators. However, the
7990 combination of @option{-mabicalls} and @option{-fno-unit-at-a-time}
7991 implies @option{-mno-explicit-relocs} unless explicitly overridden.
7992 This is because, when generating abicalls, the choice of relocation
7993 depends on whether a symbol is local or global. In some rare cases,
7994 GCC will not be able to decide this until the whole compilation unit
8001 Generate (do not generate) code that refers to registers using their
8002 software names. The default is @option{-mno-rnames}, which tells GCC
8003 to use hardware names like @samp{$4} instead of software names like
8004 @samp{a0}. The only assembler known to support @option{-rnames} is
8005 the Algorithmics assembler.
8007 @item -mcheck-zero-division
8008 @itemx -mno-check-zero-division
8009 @opindex mcheck-zero-division
8010 @opindex mno-check-zero-division
8011 Trap (do not trap) on integer division by zero. The default is
8012 @option{-mcheck-zero-division}.
8018 Force (do not force) the use of @code{memcpy()} for non-trivial block
8019 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
8020 most constant-sized copies.
8023 @itemx -mno-long-calls
8024 @opindex mlong-calls
8025 @opindex mno-long-calls
8026 Disable (do not disable) use of the @code{jal} instruction. Calling
8027 functions using @code{jal} is more efficient but requires the caller
8028 and callee to be in the same 256 megabyte segment.
8030 This option has no effect on abicalls code. The default is
8031 @option{-mno-long-calls}.
8037 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
8038 instructions, as provided by the R4650 ISA.
8041 @itemx -mno-fused-madd
8042 @opindex mfused-madd
8043 @opindex mno-fused-madd
8044 Enable (disable) use of the floating point multiply-accumulate
8045 instructions, when they are available. The default is
8046 @option{-mfused-madd}.
8048 When multiply-accumulate instructions are used, the intermediate
8049 product is calculated to infinite precision and is not subject to
8050 the FCSR Flush to Zero bit. This may be undesirable in some
8055 Tell the MIPS assembler to not run its preprocessor over user
8056 assembler files (with a @samp{.s} suffix) when assembling them.
8059 @itemx -mno-fix-r4000
8061 @opindex mno-fix-r4000
8062 Work around certain R4000 CPU errata:
8065 A double-word or a variable shift may give an incorrect result if executed
8066 immediately after starting an integer division.
8068 A double-word or a variable shift may give an incorrect result if executed
8069 while an integer multiplication is in progress.
8075 Work around certain SB-1 CPU core errata.
8076 (This flag currently works around the SB-1 revision 2
8077 ``F1'' and ``F2'' floating point errata.)
8079 @item -mflush-func=@var{func}
8080 @itemx -mno-flush-func
8081 @opindex mflush-func
8082 Specifies the function to call to flush the I and D caches, or to not
8083 call any such function. If called, the function must take the same
8084 arguments as the common @code{_flush_func()}, that is, the address of the
8085 memory range for which the cache is being flushed, the size of the
8086 memory range, and the number 3 (to flush both caches). The default
8087 depends on the target GCC was configured for, but commonly is either
8088 @samp{_flush_func} or @samp{__cpu_flush}.
8090 @item -mbranch-likely
8091 @itemx -mno-branch-likely
8092 @opindex mbranch-likely
8093 @opindex mno-branch-likely
8094 Enable or disable use of Branch Likely instructions, regardless of the
8095 default for the selected architecture. By default, Branch Likely
8096 instructions may be generated if they are supported by the selected
8097 architecture. An exception is for the MIPS32 and MIPS64 architectures
8098 and processors which implement those architectures; for those, Branch
8099 Likely instructions will not be generated by default because the MIPS32
8100 and MIPS64 architectures specifically deprecate their use.
8103 @node i386 and x86-64 Options
8104 @subsection Intel 386 and AMD x86-64 Options
8105 @cindex i386 Options
8106 @cindex x86-64 Options
8107 @cindex Intel 386 Options
8108 @cindex AMD x86-64 Options
8110 These @samp{-m} options are defined for the i386 and x86-64 family of
8114 @item -mtune=@var{cpu-type}
8116 Tune to @var{cpu-type} everything applicable about the generated code, except
8117 for the ABI and the set of available instructions. The choices for
8121 Original Intel's i386 CPU.
8123 Intel's i486 CPU. (No scheduling is implemented for this chip.)
8125 Intel Pentium CPU with no MMX support.
8127 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
8128 @item i686, pentiumpro
8129 Intel PentiumPro CPU.
8131 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
8132 @item pentium3, pentium3m
8133 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
8136 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
8137 support. Used by Centrino notebooks.
8138 @item pentium4, pentium4m
8139 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
8141 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
8144 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
8145 SSE2 and SSE3 instruction set support.
8147 AMD K6 CPU with MMX instruction set support.
8149 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
8150 @item athlon, athlon-tbird
8151 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
8153 @item athlon-4, athlon-xp, athlon-mp
8154 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
8155 instruction set support.
8156 @item k8, opteron, athlon64, athlon-fx
8157 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
8158 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
8160 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
8163 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
8164 instruction set support.
8166 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
8167 implemented for this chip.)
8169 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
8170 implemented for this chip.)
8173 While picking a specific @var{cpu-type} will schedule things appropriately
8174 for that particular chip, the compiler will not generate any code that
8175 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8178 @item -march=@var{cpu-type}
8180 Generate instructions for the machine type @var{cpu-type}. The choices
8181 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8182 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8184 @item -mcpu=@var{cpu-type}
8186 A deprecated synonym for @option{-mtune}.
8195 @opindex mpentiumpro
8196 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8197 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8198 These synonyms are deprecated.
8200 @item -mfpmath=@var{unit}
8202 Generate floating point arithmetics for selected unit @var{unit}. The choices
8207 Use the standard 387 floating point coprocessor present majority of chips and
8208 emulated otherwise. Code compiled with this option will run almost everywhere.
8209 The temporary results are computed in 80bit precision instead of precision
8210 specified by the type resulting in slightly different results compared to most
8211 of other chips. See @option{-ffloat-store} for more detailed description.
8213 This is the default choice for i386 compiler.
8216 Use scalar floating point instructions present in the SSE instruction set.
8217 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8218 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8219 instruction set supports only single precision arithmetics, thus the double and
8220 extended precision arithmetics is still done using 387. Later version, present
8221 only in Pentium4 and the future AMD x86-64 chips supports double precision
8224 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8225 @option{-msse2} switches to enable SSE extensions and make this option
8226 effective. For x86-64 compiler, these extensions are enabled by default.
8228 The resulting code should be considerably faster in the majority of cases and avoid
8229 the numerical instability problems of 387 code, but may break some existing
8230 code that expects temporaries to be 80bit.
8232 This is the default choice for the x86-64 compiler.
8235 Attempt to utilize both instruction sets at once. This effectively double the
8236 amount of available registers and on chips with separate execution units for
8237 387 and SSE the execution resources too. Use this option with care, as it is
8238 still experimental, because the GCC register allocator does not model separate
8239 functional units well resulting in instable performance.
8242 @item -masm=@var{dialect}
8243 @opindex masm=@var{dialect}
8244 Output asm instructions using selected @var{dialect}. Supported choices are
8245 @samp{intel} or @samp{att} (the default one).
8250 @opindex mno-ieee-fp
8251 Control whether or not the compiler uses IEEE floating point
8252 comparisons. These handle correctly the case where the result of a
8253 comparison is unordered.
8256 @opindex msoft-float
8257 Generate output containing library calls for floating point.
8258 @strong{Warning:} the requisite libraries are not part of GCC@.
8259 Normally the facilities of the machine's usual C compiler are used, but
8260 this can't be done directly in cross-compilation. You must make your
8261 own arrangements to provide suitable library functions for
8264 On machines where a function returns floating point results in the 80387
8265 register stack, some floating point opcodes may be emitted even if
8266 @option{-msoft-float} is used.
8268 @item -mno-fp-ret-in-387
8269 @opindex mno-fp-ret-in-387
8270 Do not use the FPU registers for return values of functions.
8272 The usual calling convention has functions return values of types
8273 @code{float} and @code{double} in an FPU register, even if there
8274 is no FPU@. The idea is that the operating system should emulate
8277 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8278 in ordinary CPU registers instead.
8280 @item -mno-fancy-math-387
8281 @opindex mno-fancy-math-387
8282 Some 387 emulators do not support the @code{sin}, @code{cos} and
8283 @code{sqrt} instructions for the 387. Specify this option to avoid
8284 generating those instructions. This option is the default on FreeBSD,
8285 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8286 indicates that the target cpu will always have an FPU and so the
8287 instruction will not need emulation. As of revision 2.6.1, these
8288 instructions are not generated unless you also use the
8289 @option{-funsafe-math-optimizations} switch.
8291 @item -malign-double
8292 @itemx -mno-align-double
8293 @opindex malign-double
8294 @opindex mno-align-double
8295 Control whether GCC aligns @code{double}, @code{long double}, and
8296 @code{long long} variables on a two word boundary or a one word
8297 boundary. Aligning @code{double} variables on a two word boundary will
8298 produce code that runs somewhat faster on a @samp{Pentium} at the
8299 expense of more memory.
8301 @strong{Warning:} if you use the @option{-malign-double} switch,
8302 structures containing the above types will be aligned differently than
8303 the published application binary interface specifications for the 386
8304 and will not be binary compatible with structures in code compiled
8305 without that switch.
8307 @item -m96bit-long-double
8308 @itemx -m128bit-long-double
8309 @opindex m96bit-long-double
8310 @opindex m128bit-long-double
8311 These switches control the size of @code{long double} type. The i386
8312 application binary interface specifies the size to be 96 bits,
8313 so @option{-m96bit-long-double} is the default in 32 bit mode.
8315 Modern architectures (Pentium and newer) would prefer @code{long double}
8316 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8317 conforming to the ABI, this would not be possible. So specifying a
8318 @option{-m128bit-long-double} will align @code{long double}
8319 to a 16 byte boundary by padding the @code{long double} with an additional
8322 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8323 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8325 Notice that neither of these options enable any extra precision over the x87
8326 standard of 80 bits for a @code{long double}.
8328 @strong{Warning:} if you override the default value for your target ABI, the
8329 structures and arrays containing @code{long double} variables will change
8330 their size as well as function calling convention for function taking
8331 @code{long double} will be modified. Hence they will not be binary
8332 compatible with arrays or structures in code compiled without that switch.
8336 @itemx -mno-svr3-shlib
8337 @opindex msvr3-shlib
8338 @opindex mno-svr3-shlib
8339 Control whether GCC places uninitialized local variables into the
8340 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8341 into @code{bss}. These options are meaningful only on System V Release 3.
8345 Use a different function-calling convention, in which functions that
8346 take a fixed number of arguments return with the @code{ret} @var{num}
8347 instruction, which pops their arguments while returning. This saves one
8348 instruction in the caller since there is no need to pop the arguments
8351 You can specify that an individual function is called with this calling
8352 sequence with the function attribute @samp{stdcall}. You can also
8353 override the @option{-mrtd} option by using the function attribute
8354 @samp{cdecl}. @xref{Function Attributes}.
8356 @strong{Warning:} this calling convention is incompatible with the one
8357 normally used on Unix, so you cannot use it if you need to call
8358 libraries compiled with the Unix compiler.
8360 Also, you must provide function prototypes for all functions that
8361 take variable numbers of arguments (including @code{printf});
8362 otherwise incorrect code will be generated for calls to those
8365 In addition, seriously incorrect code will result if you call a
8366 function with too many arguments. (Normally, extra arguments are
8367 harmlessly ignored.)
8369 @item -mregparm=@var{num}
8371 Control how many registers are used to pass integer arguments. By
8372 default, no registers are used to pass arguments, and at most 3
8373 registers can be used. You can control this behavior for a specific
8374 function by using the function attribute @samp{regparm}.
8375 @xref{Function Attributes}.
8377 @strong{Warning:} if you use this switch, and
8378 @var{num} is nonzero, then you must build all modules with the same
8379 value, including any libraries. This includes the system libraries and
8382 @item -mpreferred-stack-boundary=@var{num}
8383 @opindex mpreferred-stack-boundary
8384 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8385 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8386 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8387 size (@option{-Os}), in which case the default is the minimum correct
8388 alignment (4 bytes for x86, and 8 bytes for x86-64).
8390 On Pentium and PentiumPro, @code{double} and @code{long double} values
8391 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8392 suffer significant run time performance penalties. On Pentium III, the
8393 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8394 penalties if it is not 16 byte aligned.
8396 To ensure proper alignment of this values on the stack, the stack boundary
8397 must be as aligned as that required by any value stored on the stack.
8398 Further, every function must be generated such that it keeps the stack
8399 aligned. Thus calling a function compiled with a higher preferred
8400 stack boundary from a function compiled with a lower preferred stack
8401 boundary will most likely misalign the stack. It is recommended that
8402 libraries that use callbacks always use the default setting.
8404 This extra alignment does consume extra stack space, and generally
8405 increases code size. Code that is sensitive to stack space usage, such
8406 as embedded systems and operating system kernels, may want to reduce the
8407 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8425 These switches enable or disable the use of built-in functions that allow
8426 direct access to the MMX, SSE, SSE2, SSE3 and 3Dnow extensions of the
8429 @xref{X86 Built-in Functions}, for details of the functions enabled
8430 and disabled by these switches.
8432 To have SSE/SSE2 instructions generated automatically from floating-point
8433 code, see @option{-mfpmath=sse}.
8436 @itemx -mno-push-args
8438 @opindex mno-push-args
8439 Use PUSH operations to store outgoing parameters. This method is shorter
8440 and usually equally fast as method using SUB/MOV operations and is enabled
8441 by default. In some cases disabling it may improve performance because of
8442 improved scheduling and reduced dependencies.
8444 @item -maccumulate-outgoing-args
8445 @opindex maccumulate-outgoing-args
8446 If enabled, the maximum amount of space required for outgoing arguments will be
8447 computed in the function prologue. This is faster on most modern CPUs
8448 because of reduced dependencies, improved scheduling and reduced stack usage
8449 when preferred stack boundary is not equal to 2. The drawback is a notable
8450 increase in code size. This switch implies @option{-mno-push-args}.
8454 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8455 on thread-safe exception handling must compile and link all code with the
8456 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8457 @option{-D_MT}; when linking, it links in a special thread helper library
8458 @option{-lmingwthrd} which cleans up per thread exception handling data.
8460 @item -mno-align-stringops
8461 @opindex mno-align-stringops
8462 Do not align destination of inlined string operations. This switch reduces
8463 code size and improves performance in case the destination is already aligned,
8464 but GCC doesn't know about it.
8466 @item -minline-all-stringops
8467 @opindex minline-all-stringops
8468 By default GCC inlines string operations only when destination is known to be
8469 aligned at least to 4 byte boundary. This enables more inlining, increase code
8470 size, but may improve performance of code that depends on fast memcpy, strlen
8471 and memset for short lengths.
8473 @item -momit-leaf-frame-pointer
8474 @opindex momit-leaf-frame-pointer
8475 Don't keep the frame pointer in a register for leaf functions. This
8476 avoids the instructions to save, set up and restore frame pointers and
8477 makes an extra register available in leaf functions. The option
8478 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8479 which might make debugging harder.
8481 @item -mtls-direct-seg-refs
8482 @itemx -mno-tls-direct-seg-refs
8483 @opindex mtls-direct-seg-refs
8484 Controls whether TLS variables may be accessed with offsets from the
8485 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8486 or whether the thread base pointer must be added. Whether or not this
8487 is legal depends on the operating system, and whether it maps the
8488 segment to cover the entire TLS area.
8490 For systems that use GNU libc, the default is on.
8493 These @samp{-m} switches are supported in addition to the above
8494 on AMD x86-64 processors in 64-bit environments.
8501 Generate code for a 32-bit or 64-bit environment.
8502 The 32-bit environment sets int, long and pointer to 32 bits and
8503 generates code that runs on any i386 system.
8504 The 64-bit environment sets int to 32 bits and long and pointer
8505 to 64 bits and generates code for AMD's x86-64 architecture.
8508 @opindex no-red-zone
8509 Do not use a so called red zone for x86-64 code. The red zone is mandated
8510 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8511 stack pointer that will not be modified by signal or interrupt handlers
8512 and therefore can be used for temporary data without adjusting the stack
8513 pointer. The flag @option{-mno-red-zone} disables this red zone.
8515 @item -mcmodel=small
8516 @opindex mcmodel=small
8517 Generate code for the small code model: the program and its symbols must
8518 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8519 Programs can be statically or dynamically linked. This is the default
8522 @item -mcmodel=kernel
8523 @opindex mcmodel=kernel
8524 Generate code for the kernel code model. The kernel runs in the
8525 negative 2 GB of the address space.
8526 This model has to be used for Linux kernel code.
8528 @item -mcmodel=medium
8529 @opindex mcmodel=medium
8530 Generate code for the medium model: The program is linked in the lower 2
8531 GB of the address space but symbols can be located anywhere in the
8532 address space. Programs can be statically or dynamically linked, but
8533 building of shared libraries are not supported with the medium model.
8535 @item -mcmodel=large
8536 @opindex mcmodel=large
8537 Generate code for the large model: This model makes no assumptions
8538 about addresses and sizes of sections. Currently GCC does not implement
8543 @subsection HPPA Options
8544 @cindex HPPA Options
8546 These @samp{-m} options are defined for the HPPA family of computers:
8549 @item -march=@var{architecture-type}
8551 Generate code for the specified architecture. The choices for
8552 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8553 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8554 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8555 architecture option for your machine. Code compiled for lower numbered
8556 architectures will run on higher numbered architectures, but not the
8559 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8560 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8564 @itemx -mpa-risc-1-1
8565 @itemx -mpa-risc-2-0
8566 @opindex mpa-risc-1-0
8567 @opindex mpa-risc-1-1
8568 @opindex mpa-risc-2-0
8569 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8572 @opindex mbig-switch
8573 Generate code suitable for big switch tables. Use this option only if
8574 the assembler/linker complain about out of range branches within a switch
8577 @item -mjump-in-delay
8578 @opindex mjump-in-delay
8579 Fill delay slots of function calls with unconditional jump instructions
8580 by modifying the return pointer for the function call to be the target
8581 of the conditional jump.
8583 @item -mdisable-fpregs
8584 @opindex mdisable-fpregs
8585 Prevent floating point registers from being used in any manner. This is
8586 necessary for compiling kernels which perform lazy context switching of
8587 floating point registers. If you use this option and attempt to perform
8588 floating point operations, the compiler will abort.
8590 @item -mdisable-indexing
8591 @opindex mdisable-indexing
8592 Prevent the compiler from using indexing address modes. This avoids some
8593 rather obscure problems when compiling MIG generated code under MACH@.
8595 @item -mno-space-regs
8596 @opindex mno-space-regs
8597 Generate code that assumes the target has no space registers. This allows
8598 GCC to generate faster indirect calls and use unscaled index address modes.
8600 Such code is suitable for level 0 PA systems and kernels.
8602 @item -mfast-indirect-calls
8603 @opindex mfast-indirect-calls
8604 Generate code that assumes calls never cross space boundaries. This
8605 allows GCC to emit code which performs faster indirect calls.
8607 This option will not work in the presence of shared libraries or nested
8610 @item -mlong-load-store
8611 @opindex mlong-load-store
8612 Generate 3-instruction load and store sequences as sometimes required by
8613 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8616 @item -mportable-runtime
8617 @opindex mportable-runtime
8618 Use the portable calling conventions proposed by HP for ELF systems.
8622 Enable the use of assembler directives only GAS understands.
8624 @item -mschedule=@var{cpu-type}
8626 Schedule code according to the constraints for the machine type
8627 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8628 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8629 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8630 proper scheduling option for your machine. The default scheduling is
8634 @opindex mlinker-opt
8635 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8636 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8637 linkers in which they give bogus error messages when linking some programs.
8640 @opindex msoft-float
8641 Generate output containing library calls for floating point.
8642 @strong{Warning:} the requisite libraries are not available for all HPPA
8643 targets. Normally the facilities of the machine's usual C compiler are
8644 used, but this cannot be done directly in cross-compilation. You must make
8645 your own arrangements to provide suitable library functions for
8646 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8647 does provide software floating point support.
8649 @option{-msoft-float} changes the calling convention in the output file;
8650 therefore, it is only useful if you compile @emph{all} of a program with
8651 this option. In particular, you need to compile @file{libgcc.a}, the
8652 library that comes with GCC, with @option{-msoft-float} in order for
8657 Generate the predefine, @code{_SIO}, for server IO. The default is
8658 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8659 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8660 options are available under HP-UX and HI-UX.
8664 Use GNU ld specific options. This passes @option{-shared} to ld when
8665 building a shared library. It is the default when GCC is configured,
8666 explicitly or implicitly, with the GNU linker. This option does not
8667 have any affect on which ld is called, it only changes what parameters
8668 are passed to that ld. The ld that is called is determined by the
8669 @option{--with-ld} configure option, GCC's program search path, and
8670 finally by the user's @env{PATH}. The linker used by GCC can be printed
8671 using @samp{which `gcc -print-prog-name=ld`}.
8675 Use HP ld specific options. This passes @option{-b} to ld when building
8676 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8677 links. It is the default when GCC is configured, explicitly or
8678 implicitly, with the HP linker. This option does not have any affect on
8679 which ld is called, it only changes what parameters are passed to that
8680 ld. The ld that is called is determined by the @option{--with-ld}
8681 configure option, GCC's program search path, and finally by the user's
8682 @env{PATH}. The linker used by GCC can be printed using @samp{which
8683 `gcc -print-prog-name=ld`}.
8688 Select the FDPIC ABI, that uses function descriptors to represent
8689 pointers to functions. Without any PIC/PIE-related options, it
8690 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
8691 assumes GOT entries and small data are within a 12-bit range from the
8692 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
8693 are computed with 32 bits.
8696 @opindex minline-plt
8698 Enable inlining of PLT entries in function calls to functions that are
8699 not known to bind locally. It has no effect without @option{-mfdpic}.
8700 It's enabled by default if optimizing for speed and compiling for
8701 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
8702 optimization option such as @option{-O3} or above is present in the
8708 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
8709 that is known to be in read-only sections. It's enabled by default,
8710 except for @option{-fpic} or @option{-fpie}: even though it may help
8711 make the global offset table smaller, it trades 1 instruction for 4.
8712 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
8713 one of which may be shared by multiple symbols, and it avoids the need
8714 for a GOT entry for the referenced symbol, so it's more likely to be a
8715 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
8717 @item -multilib-library-pic
8718 @opindex multilib-library-pic
8720 Link with the (library, not FD) pic libraries. It's implied by
8721 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
8722 @option{-fpic} without @option{-mfdpic}. You should never have to use
8728 Follow the EABI requirement of always creating a frame pointer whenever
8729 a stack frame is allocated. This option is enabled by default and can
8730 be disabled with @option{-mno-linked-fp}.
8733 @opindex mno-long-calls
8734 Generate code that uses long call sequences. This ensures that a call
8735 is always able to reach linker generated stubs. The default is to generate
8736 long calls only when the distance from the call site to the beginning
8737 of the function or translation unit, as the case may be, exceeds a
8738 predefined limit set by the branch type being used. The limits for
8739 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8740 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8743 Distances are measured from the beginning of functions when using the
8744 @option{-ffunction-sections} option, or when using the @option{-mgas}
8745 and @option{-mno-portable-runtime} options together under HP-UX with
8748 It is normally not desirable to use this option as it will degrade
8749 performance. However, it may be useful in large applications,
8750 particularly when partial linking is used to build the application.
8752 The types of long calls used depends on the capabilities of the
8753 assembler and linker, and the type of code being generated. The
8754 impact on systems that support long absolute calls, and long pic
8755 symbol-difference or pc-relative calls should be relatively small.
8756 However, an indirect call is used on 32-bit ELF systems in pic code
8757 and it is quite long.
8761 Suppress the generation of link options to search libdld.sl when the
8762 @option{-static} option is specified on HP-UX 10 and later.
8766 The HP-UX implementation of setlocale in libc has a dependency on
8767 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8768 when the @option{-static} option is specified, special link options
8769 are needed to resolve this dependency.
8771 On HP-UX 10 and later, the GCC driver adds the necessary options to
8772 link with libdld.sl when the @option{-static} option is specified.
8773 This causes the resulting binary to be dynamic. On the 64-bit port,
8774 the linkers generate dynamic binaries by default in any case. The
8775 @option{-nolibdld} option can be used to prevent the GCC driver from
8776 adding these link options.
8780 Add support for multithreading with the @dfn{dce thread} library
8781 under HP-UX. This option sets flags for both the preprocessor and
8785 @node DEC Alpha Options
8786 @subsection DEC Alpha Options
8788 These @samp{-m} options are defined for the DEC Alpha implementations:
8791 @item -mno-soft-float
8793 @opindex mno-soft-float
8794 @opindex msoft-float
8795 Use (do not use) the hardware floating-point instructions for
8796 floating-point operations. When @option{-msoft-float} is specified,
8797 functions in @file{libgcc.a} will be used to perform floating-point
8798 operations. Unless they are replaced by routines that emulate the
8799 floating-point operations, or compiled in such a way as to call such
8800 emulations routines, these routines will issue floating-point
8801 operations. If you are compiling for an Alpha without floating-point
8802 operations, you must ensure that the library is built so as not to call
8805 Note that Alpha implementations without floating-point operations are
8806 required to have floating-point registers.
8811 @opindex mno-fp-regs
8812 Generate code that uses (does not use) the floating-point register set.
8813 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8814 register set is not used, floating point operands are passed in integer
8815 registers as if they were integers and floating-point results are passed
8816 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8817 so any function with a floating-point argument or return value called by code
8818 compiled with @option{-mno-fp-regs} must also be compiled with that
8821 A typical use of this option is building a kernel that does not use,
8822 and hence need not save and restore, any floating-point registers.
8826 The Alpha architecture implements floating-point hardware optimized for
8827 maximum performance. It is mostly compliant with the IEEE floating
8828 point standard. However, for full compliance, software assistance is
8829 required. This option generates code fully IEEE compliant code
8830 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8831 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8832 defined during compilation. The resulting code is less efficient but is
8833 able to correctly support denormalized numbers and exceptional IEEE
8834 values such as not-a-number and plus/minus infinity. Other Alpha
8835 compilers call this option @option{-ieee_with_no_inexact}.
8837 @item -mieee-with-inexact
8838 @opindex mieee-with-inexact
8839 This is like @option{-mieee} except the generated code also maintains
8840 the IEEE @var{inexact-flag}. Turning on this option causes the
8841 generated code to implement fully-compliant IEEE math. In addition to
8842 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8843 macro. On some Alpha implementations the resulting code may execute
8844 significantly slower than the code generated by default. Since there is
8845 very little code that depends on the @var{inexact-flag}, you should
8846 normally not specify this option. Other Alpha compilers call this
8847 option @option{-ieee_with_inexact}.
8849 @item -mfp-trap-mode=@var{trap-mode}
8850 @opindex mfp-trap-mode
8851 This option controls what floating-point related traps are enabled.
8852 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8853 The trap mode can be set to one of four values:
8857 This is the default (normal) setting. The only traps that are enabled
8858 are the ones that cannot be disabled in software (e.g., division by zero
8862 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8866 Like @samp{su}, but the instructions are marked to be safe for software
8867 completion (see Alpha architecture manual for details).
8870 Like @samp{su}, but inexact traps are enabled as well.
8873 @item -mfp-rounding-mode=@var{rounding-mode}
8874 @opindex mfp-rounding-mode
8875 Selects the IEEE rounding mode. Other Alpha compilers call this option
8876 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8881 Normal IEEE rounding mode. Floating point numbers are rounded towards
8882 the nearest machine number or towards the even machine number in case
8886 Round towards minus infinity.
8889 Chopped rounding mode. Floating point numbers are rounded towards zero.
8892 Dynamic rounding mode. A field in the floating point control register
8893 (@var{fpcr}, see Alpha architecture reference manual) controls the
8894 rounding mode in effect. The C library initializes this register for
8895 rounding towards plus infinity. Thus, unless your program modifies the
8896 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8899 @item -mtrap-precision=@var{trap-precision}
8900 @opindex mtrap-precision
8901 In the Alpha architecture, floating point traps are imprecise. This
8902 means without software assistance it is impossible to recover from a
8903 floating trap and program execution normally needs to be terminated.
8904 GCC can generate code that can assist operating system trap handlers
8905 in determining the exact location that caused a floating point trap.
8906 Depending on the requirements of an application, different levels of
8907 precisions can be selected:
8911 Program precision. This option is the default and means a trap handler
8912 can only identify which program caused a floating point exception.
8915 Function precision. The trap handler can determine the function that
8916 caused a floating point exception.
8919 Instruction precision. The trap handler can determine the exact
8920 instruction that caused a floating point exception.
8923 Other Alpha compilers provide the equivalent options called
8924 @option{-scope_safe} and @option{-resumption_safe}.
8926 @item -mieee-conformant
8927 @opindex mieee-conformant
8928 This option marks the generated code as IEEE conformant. You must not
8929 use this option unless you also specify @option{-mtrap-precision=i} and either
8930 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8931 is to emit the line @samp{.eflag 48} in the function prologue of the
8932 generated assembly file. Under DEC Unix, this has the effect that
8933 IEEE-conformant math library routines will be linked in.
8935 @item -mbuild-constants
8936 @opindex mbuild-constants
8937 Normally GCC examines a 32- or 64-bit integer constant to
8938 see if it can construct it from smaller constants in two or three
8939 instructions. If it cannot, it will output the constant as a literal and
8940 generate code to load it from the data segment at runtime.
8942 Use this option to require GCC to construct @emph{all} integer constants
8943 using code, even if it takes more instructions (the maximum is six).
8945 You would typically use this option to build a shared library dynamic
8946 loader. Itself a shared library, it must relocate itself in memory
8947 before it can find the variables and constants in its own data segment.
8953 Select whether to generate code to be assembled by the vendor-supplied
8954 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8972 Indicate whether GCC should generate code to use the optional BWX,
8973 CIX, FIX and MAX instruction sets. The default is to use the instruction
8974 sets supported by the CPU type specified via @option{-mcpu=} option or that
8975 of the CPU on which GCC was built if none was specified.
8980 @opindex mfloat-ieee
8981 Generate code that uses (does not use) VAX F and G floating point
8982 arithmetic instead of IEEE single and double precision.
8984 @item -mexplicit-relocs
8985 @itemx -mno-explicit-relocs
8986 @opindex mexplicit-relocs
8987 @opindex mno-explicit-relocs
8988 Older Alpha assemblers provided no way to generate symbol relocations
8989 except via assembler macros. Use of these macros does not allow
8990 optimal instruction scheduling. GNU binutils as of version 2.12
8991 supports a new syntax that allows the compiler to explicitly mark
8992 which relocations should apply to which instructions. This option
8993 is mostly useful for debugging, as GCC detects the capabilities of
8994 the assembler when it is built and sets the default accordingly.
8998 @opindex msmall-data
8999 @opindex mlarge-data
9000 When @option{-mexplicit-relocs} is in effect, static data is
9001 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9002 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9003 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9004 16-bit relocations off of the @code{$gp} register. This limits the
9005 size of the small data area to 64KB, but allows the variables to be
9006 directly accessed via a single instruction.
9008 The default is @option{-mlarge-data}. With this option the data area
9009 is limited to just below 2GB. Programs that require more than 2GB of
9010 data must use @code{malloc} or @code{mmap} to allocate the data in the
9011 heap instead of in the program's data segment.
9013 When generating code for shared libraries, @option{-fpic} implies
9014 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9018 @opindex msmall-text
9019 @opindex mlarge-text
9020 When @option{-msmall-text} is used, the compiler assumes that the
9021 code of the entire program (or shared library) fits in 4MB, and is
9022 thus reachable with a branch instruction. When @option{-msmall-data}
9023 is used, the compiler can assume that all local symbols share the
9024 same @code{$gp} value, and thus reduce the number of instructions
9025 required for a function call from 4 to 1.
9027 The default is @option{-mlarge-text}.
9029 @item -mcpu=@var{cpu_type}
9031 Set the instruction set and instruction scheduling parameters for
9032 machine type @var{cpu_type}. You can specify either the @samp{EV}
9033 style name or the corresponding chip number. GCC supports scheduling
9034 parameters for the EV4, EV5 and EV6 family of processors and will
9035 choose the default values for the instruction set from the processor
9036 you specify. If you do not specify a processor type, GCC will default
9037 to the processor on which the compiler was built.
9039 Supported values for @var{cpu_type} are
9045 Schedules as an EV4 and has no instruction set extensions.
9049 Schedules as an EV5 and has no instruction set extensions.
9053 Schedules as an EV5 and supports the BWX extension.
9058 Schedules as an EV5 and supports the BWX and MAX extensions.
9062 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9066 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9069 @item -mtune=@var{cpu_type}
9071 Set only the instruction scheduling parameters for machine type
9072 @var{cpu_type}. The instruction set is not changed.
9074 @item -mmemory-latency=@var{time}
9075 @opindex mmemory-latency
9076 Sets the latency the scheduler should assume for typical memory
9077 references as seen by the application. This number is highly
9078 dependent on the memory access patterns used by the application
9079 and the size of the external cache on the machine.
9081 Valid options for @var{time} are
9085 A decimal number representing clock cycles.
9091 The compiler contains estimates of the number of clock cycles for
9092 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9093 (also called Dcache, Scache, and Bcache), as well as to main memory.
9094 Note that L3 is only valid for EV5.
9099 @node DEC Alpha/VMS Options
9100 @subsection DEC Alpha/VMS Options
9102 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9105 @item -mvms-return-codes
9106 @opindex mvms-return-codes
9107 Return VMS condition codes from main. The default is to return POSIX
9108 style condition (e.g.@ error) codes.
9111 @node H8/300 Options
9112 @subsection H8/300 Options
9114 These @samp{-m} options are defined for the H8/300 implementations:
9119 Shorten some address references at link time, when possible; uses the
9120 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9121 ld, Using ld}, for a fuller description.
9125 Generate code for the H8/300H@.
9129 Generate code for the H8S@.
9133 Generate code for the H8S and H8/300H in the normal mode. This switch
9134 must be used either with -mh or -ms.
9138 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9142 Make @code{int} data 32 bits by default.
9146 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9147 The default for the H8/300H and H8S is to align longs and floats on 4
9149 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9150 This option has no effect on the H8/300.
9154 @subsection SH Options
9156 These @samp{-m} options are defined for the SH implementations:
9161 Generate code for the SH1.
9165 Generate code for the SH2.
9168 Generate code for the SH2e.
9172 Generate code for the SH3.
9176 Generate code for the SH3e.
9180 Generate code for the SH4 without a floating-point unit.
9182 @item -m4-single-only
9183 @opindex m4-single-only
9184 Generate code for the SH4 with a floating-point unit that only
9185 supports single-precision arithmetic.
9189 Generate code for the SH4 assuming the floating-point unit is in
9190 single-precision mode by default.
9194 Generate code for the SH4.
9198 Compile code for the processor in big endian mode.
9202 Compile code for the processor in little endian mode.
9206 Align doubles at 64-bit boundaries. Note that this changes the calling
9207 conventions, and thus some functions from the standard C library will
9208 not work unless you recompile it first with @option{-mdalign}.
9212 Shorten some address references at link time, when possible; uses the
9213 linker option @option{-relax}.
9217 Use 32-bit offsets in @code{switch} tables. The default is to use
9222 Enable the use of the instruction @code{fmovd}.
9226 Comply with the calling conventions defined by Renesas.
9230 Mark the @code{MAC} register as call-clobbered, even if
9231 @option{-mhitachi} is given.
9235 Increase IEEE-compliance of floating-point code.
9239 Dump instruction size and location in the assembly code.
9243 This option is deprecated. It pads structures to multiple of 4 bytes,
9244 which is incompatible with the SH ABI@.
9248 Optimize for space instead of speed. Implied by @option{-Os}.
9252 When generating position-independent code, emit function calls using
9253 the Global Offset Table instead of the Procedure Linkage Table.
9257 Generate a library function call to invalidate instruction cache
9258 entries, after fixing up a trampoline. This library function call
9259 doesn't assume it can write to the whole memory address space. This
9260 is the default when the target is @code{sh-*-linux*}.
9263 @node System V Options
9264 @subsection Options for System V
9266 These additional options are available on System V Release 4 for
9267 compatibility with other compilers on those systems:
9272 Create a shared object.
9273 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9277 Identify the versions of each tool used by the compiler, in a
9278 @code{.ident} assembler directive in the output.
9282 Refrain from adding @code{.ident} directives to the output file (this is
9285 @item -YP,@var{dirs}
9287 Search the directories @var{dirs}, and no others, for libraries
9288 specified with @option{-l}.
9292 Look in the directory @var{dir} to find the M4 preprocessor.
9293 The assembler uses this option.
9294 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9295 @c the generic assembler that comes with Solaris takes just -Ym.
9298 @node TMS320C3x/C4x Options
9299 @subsection TMS320C3x/C4x Options
9300 @cindex TMS320C3x/C4x Options
9302 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9306 @item -mcpu=@var{cpu_type}
9308 Set the instruction set, register set, and instruction scheduling
9309 parameters for machine type @var{cpu_type}. Supported values for
9310 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9311 @samp{c44}. The default is @samp{c40} to generate code for the
9316 @itemx -msmall-memory
9318 @opindex mbig-memory
9320 @opindex msmall-memory
9322 Generates code for the big or small memory model. The small memory
9323 model assumed that all data fits into one 64K word page. At run-time
9324 the data page (DP) register must be set to point to the 64K page
9325 containing the .bss and .data program sections. The big memory model is
9326 the default and requires reloading of the DP register for every direct
9333 Allow (disallow) allocation of general integer operands into the block
9340 Enable (disable) generation of code using decrement and branch,
9341 DBcond(D), instructions. This is enabled by default for the C4x. To be
9342 on the safe side, this is disabled for the C3x, since the maximum
9343 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9344 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9345 that it can utilize the decrement and branch instruction, but will give
9346 up if there is more than one memory reference in the loop. Thus a loop
9347 where the loop counter is decremented can generate slightly more
9348 efficient code, in cases where the RPTB instruction cannot be utilized.
9350 @item -mdp-isr-reload
9352 @opindex mdp-isr-reload
9354 Force the DP register to be saved on entry to an interrupt service
9355 routine (ISR), reloaded to point to the data section, and restored on
9356 exit from the ISR@. This should not be required unless someone has
9357 violated the small memory model by modifying the DP register, say within
9364 For the C3x use the 24-bit MPYI instruction for integer multiplies
9365 instead of a library call to guarantee 32-bit results. Note that if one
9366 of the operands is a constant, then the multiplication will be performed
9367 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9368 then squaring operations are performed inline instead of a library call.
9371 @itemx -mno-fast-fix
9373 @opindex mno-fast-fix
9374 The C3x/C4x FIX instruction to convert a floating point value to an
9375 integer value chooses the nearest integer less than or equal to the
9376 floating point value rather than to the nearest integer. Thus if the
9377 floating point number is negative, the result will be incorrectly
9378 truncated an additional code is necessary to detect and correct this
9379 case. This option can be used to disable generation of the additional
9380 code required to correct the result.
9386 Enable (disable) generation of repeat block sequences using the RPTB
9387 instruction for zero overhead looping. The RPTB construct is only used
9388 for innermost loops that do not call functions or jump across the loop
9389 boundaries. There is no advantage having nested RPTB loops due to the
9390 overhead required to save and restore the RC, RS, and RE registers.
9391 This is enabled by default with @option{-O2}.
9393 @item -mrpts=@var{count}
9397 Enable (disable) the use of the single instruction repeat instruction
9398 RPTS@. If a repeat block contains a single instruction, and the loop
9399 count can be guaranteed to be less than the value @var{count}, GCC will
9400 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9401 then a RPTS will be emitted even if the loop count cannot be determined
9402 at compile time. Note that the repeated instruction following RPTS does
9403 not have to be reloaded from memory each iteration, thus freeing up the
9404 CPU buses for operands. However, since interrupts are blocked by this
9405 instruction, it is disabled by default.
9407 @item -mloop-unsigned
9408 @itemx -mno-loop-unsigned
9409 @opindex mloop-unsigned
9410 @opindex mno-loop-unsigned
9411 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9412 is @math{2^{31} + 1} since these instructions test if the iteration count is
9413 negative to terminate the loop. If the iteration count is unsigned
9414 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9415 exceeded. This switch allows an unsigned iteration count.
9419 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9420 with. This also enforces compatibility with the API employed by the TI
9421 C3x C compiler. For example, long doubles are passed as structures
9422 rather than in floating point registers.
9428 Generate code that uses registers (stack) for passing arguments to functions.
9429 By default, arguments are passed in registers where possible rather
9430 than by pushing arguments on to the stack.
9432 @item -mparallel-insns
9433 @itemx -mno-parallel-insns
9434 @opindex mparallel-insns
9435 @opindex mno-parallel-insns
9436 Allow the generation of parallel instructions. This is enabled by
9437 default with @option{-O2}.
9439 @item -mparallel-mpy
9440 @itemx -mno-parallel-mpy
9441 @opindex mparallel-mpy
9442 @opindex mno-parallel-mpy
9443 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9444 provided @option{-mparallel-insns} is also specified. These instructions have
9445 tight register constraints which can pessimize the code generation
9451 @subsection V850 Options
9452 @cindex V850 Options
9454 These @samp{-m} options are defined for V850 implementations:
9458 @itemx -mno-long-calls
9459 @opindex mlong-calls
9460 @opindex mno-long-calls
9461 Treat all calls as being far away (near). If calls are assumed to be
9462 far away, the compiler will always load the functions address up into a
9463 register, and call indirect through the pointer.
9469 Do not optimize (do optimize) basic blocks that use the same index
9470 pointer 4 or more times to copy pointer into the @code{ep} register, and
9471 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9472 option is on by default if you optimize.
9474 @item -mno-prolog-function
9475 @itemx -mprolog-function
9476 @opindex mno-prolog-function
9477 @opindex mprolog-function
9478 Do not use (do use) external functions to save and restore registers
9479 at the prologue and epilogue of a function. The external functions
9480 are slower, but use less code space if more than one function saves
9481 the same number of registers. The @option{-mprolog-function} option
9482 is on by default if you optimize.
9486 Try to make the code as small as possible. At present, this just turns
9487 on the @option{-mep} and @option{-mprolog-function} options.
9491 Put static or global variables whose size is @var{n} bytes or less into
9492 the tiny data area that register @code{ep} points to. The tiny data
9493 area can hold up to 256 bytes in total (128 bytes for byte references).
9497 Put static or global variables whose size is @var{n} bytes or less into
9498 the small data area that register @code{gp} points to. The small data
9499 area can hold up to 64 kilobytes.
9503 Put static or global variables whose size is @var{n} bytes or less into
9504 the first 32 kilobytes of memory.
9508 Specify that the target processor is the V850.
9511 @opindex mbig-switch
9512 Generate code suitable for big switch tables. Use this option only if
9513 the assembler/linker complain about out of range branches within a switch
9518 This option will cause r2 and r5 to be used in the code generated by
9519 the compiler. This setting is the default.
9522 @opindex mno-app-regs
9523 This option will cause r2 and r5 to be treated as fixed registers.
9527 Specify that the target processor is the V850E1. The preprocessor
9528 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
9529 this option is used.
9533 Specify that the target processor is the V850E. The preprocessor
9534 constant @samp{__v850e__} will be defined if this option is used.
9536 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
9537 are defined then a default target processor will be chosen and the
9538 relevant @samp{__v850*__} preprocessor constant will be defined.
9540 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9541 defined, regardless of which processor variant is the target.
9543 @item -mdisable-callt
9544 @opindex mdisable-callt
9545 This option will suppress generation of the CALLT instruction for the
9546 v850e and v850e1 flavors of the v850 architecture. The default is
9547 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9552 @subsection ARC Options
9555 These options are defined for ARC implementations:
9560 Compile code for little endian mode. This is the default.
9564 Compile code for big endian mode.
9567 @opindex mmangle-cpu
9568 Prepend the name of the cpu to all public symbol names.
9569 In multiple-processor systems, there are many ARC variants with different
9570 instruction and register set characteristics. This flag prevents code
9571 compiled for one cpu to be linked with code compiled for another.
9572 No facility exists for handling variants that are ``almost identical''.
9573 This is an all or nothing option.
9575 @item -mcpu=@var{cpu}
9577 Compile code for ARC variant @var{cpu}.
9578 Which variants are supported depend on the configuration.
9579 All variants support @option{-mcpu=base}, this is the default.
9581 @item -mtext=@var{text-section}
9582 @itemx -mdata=@var{data-section}
9583 @itemx -mrodata=@var{readonly-data-section}
9587 Put functions, data, and readonly data in @var{text-section},
9588 @var{data-section}, and @var{readonly-data-section} respectively
9589 by default. This can be overridden with the @code{section} attribute.
9590 @xref{Variable Attributes}.
9595 @subsection NS32K Options
9596 @cindex NS32K options
9598 These are the @samp{-m} options defined for the 32000 series. The default
9599 values for these options depends on which style of 32000 was selected when
9600 the compiler was configured; the defaults for the most common choices are
9608 Generate output for a 32032. This is the default
9609 when the compiler is configured for 32032 and 32016 based systems.
9615 Generate output for a 32332. This is the default
9616 when the compiler is configured for 32332-based systems.
9622 Generate output for a 32532. This is the default
9623 when the compiler is configured for 32532-based systems.
9627 Generate output containing 32081 instructions for floating point.
9628 This is the default for all systems.
9632 Generate output containing 32381 instructions for floating point. This
9633 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9634 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9638 Try and generate multiply-add floating point instructions @code{polyF}
9639 and @code{dotF}. This option is only available if the @option{-m32381}
9640 option is in effect. Using these instructions requires changes to
9641 register allocation which generally has a negative impact on
9642 performance. This option should only be enabled when compiling code
9643 particularly likely to make heavy use of multiply-add instructions.
9646 @opindex mnomulti-add
9647 Do not try and generate multiply-add floating point instructions
9648 @code{polyF} and @code{dotF}. This is the default on all platforms.
9651 @opindex msoft-float
9652 Generate output containing library calls for floating point.
9653 @strong{Warning:} the requisite libraries may not be available.
9655 @item -mieee-compare
9656 @itemx -mno-ieee-compare
9657 @opindex mieee-compare
9658 @opindex mno-ieee-compare
9659 Control whether or not the compiler uses IEEE floating point
9660 comparisons. These handle correctly the case where the result of a
9661 comparison is unordered.
9662 @strong{Warning:} the requisite kernel support may not be available.
9665 @opindex mnobitfield
9666 Do not use the bit-field instructions. On some machines it is faster to
9667 use shifting and masking operations. This is the default for the pc532.
9671 Do use the bit-field instructions. This is the default for all platforms
9676 Use a different function-calling convention, in which functions
9677 that take a fixed number of arguments return pop their
9678 arguments on return with the @code{ret} instruction.
9680 This calling convention is incompatible with the one normally
9681 used on Unix, so you cannot use it if you need to call libraries
9682 compiled with the Unix compiler.
9684 Also, you must provide function prototypes for all functions that
9685 take variable numbers of arguments (including @code{printf});
9686 otherwise incorrect code will be generated for calls to those
9689 In addition, seriously incorrect code will result if you call a
9690 function with too many arguments. (Normally, extra arguments are
9691 harmlessly ignored.)
9693 This option takes its name from the 680x0 @code{rtd} instruction.
9698 Use a different function-calling convention where the first two arguments
9699 are passed in registers.
9701 This calling convention is incompatible with the one normally
9702 used on Unix, so you cannot use it if you need to call libraries
9703 compiled with the Unix compiler.
9706 @opindex mnoregparam
9707 Do not pass any arguments in registers. This is the default for all
9712 It is OK to use the sb as an index register which is always loaded with
9713 zero. This is the default for the pc532-netbsd target.
9717 The sb register is not available for use or has not been initialized to
9718 zero by the run time system. This is the default for all targets except
9719 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9720 @option{-fpic} is set.
9724 Many ns32000 series addressing modes use displacements of up to 512MB@.
9725 If an address is above 512MB then displacements from zero can not be used.
9726 This option causes code to be generated which can be loaded above 512MB@.
9727 This may be useful for operating systems or ROM code.
9731 Assume code will be loaded in the first 512MB of virtual address space.
9732 This is the default for all platforms.
9738 @subsection AVR Options
9741 These options are defined for AVR implementations:
9744 @item -mmcu=@var{mcu}
9746 Specify ATMEL AVR instruction set or MCU type.
9748 Instruction set avr1 is for the minimal AVR core, not supported by the C
9749 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9750 attiny11, attiny12, attiny15, attiny28).
9752 Instruction set avr2 (default) is for the classic AVR core with up to
9753 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9754 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9755 at90c8534, at90s8535).
9757 Instruction set avr3 is for the classic AVR core with up to 128K program
9758 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9760 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9761 memory space (MCU types: atmega8, atmega83, atmega85).
9763 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9764 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9765 atmega64, atmega128, at43usb355, at94k).
9769 Output instruction sizes to the asm file.
9771 @item -minit-stack=@var{N}
9772 @opindex minit-stack
9773 Specify the initial stack address, which may be a symbol or numeric value,
9774 @samp{__stack} is the default.
9776 @item -mno-interrupts
9777 @opindex mno-interrupts
9778 Generated code is not compatible with hardware interrupts.
9779 Code size will be smaller.
9781 @item -mcall-prologues
9782 @opindex mcall-prologues
9783 Functions prologues/epilogues expanded as call to appropriate
9784 subroutines. Code size will be smaller.
9786 @item -mno-tablejump
9787 @opindex mno-tablejump
9788 Do not generate tablejump insns which sometimes increase code size.
9791 @opindex mtiny-stack
9792 Change only the low 8 bits of the stack pointer.
9796 @subsection MCore Options
9797 @cindex MCore options
9799 These are the @samp{-m} options defined for the Motorola M*Core
9807 @opindex mno-hardlit
9808 Inline constants into the code stream if it can be done in two
9809 instructions or less.
9815 Use the divide instruction. (Enabled by default).
9817 @item -mrelax-immediate
9818 @itemx -mno-relax-immediate
9819 @opindex mrelax-immediate
9820 @opindex mno-relax-immediate
9821 Allow arbitrary sized immediates in bit operations.
9823 @item -mwide-bitfields
9824 @itemx -mno-wide-bitfields
9825 @opindex mwide-bitfields
9826 @opindex mno-wide-bitfields
9827 Always treat bit-fields as int-sized.
9829 @item -m4byte-functions
9830 @itemx -mno-4byte-functions
9831 @opindex m4byte-functions
9832 @opindex mno-4byte-functions
9833 Force all functions to be aligned to a four byte boundary.
9835 @item -mcallgraph-data
9836 @itemx -mno-callgraph-data
9837 @opindex mcallgraph-data
9838 @opindex mno-callgraph-data
9839 Emit callgraph information.
9842 @itemx -mno-slow-bytes
9843 @opindex mslow-bytes
9844 @opindex mno-slow-bytes
9845 Prefer word access when reading byte quantities.
9847 @item -mlittle-endian
9849 @opindex mlittle-endian
9850 @opindex mbig-endian
9851 Generate code for a little endian target.
9857 Generate code for the 210 processor.
9861 @subsection IA-64 Options
9862 @cindex IA-64 Options
9864 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9868 @opindex mbig-endian
9869 Generate code for a big endian target. This is the default for HP-UX@.
9871 @item -mlittle-endian
9872 @opindex mlittle-endian
9873 Generate code for a little endian target. This is the default for AIX5
9880 Generate (or don't) code for the GNU assembler. This is the default.
9881 @c Also, this is the default if the configure option @option{--with-gnu-as}
9888 Generate (or don't) code for the GNU linker. This is the default.
9889 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9894 Generate code that does not use a global pointer register. The result
9895 is not position independent code, and violates the IA-64 ABI@.
9897 @item -mvolatile-asm-stop
9898 @itemx -mno-volatile-asm-stop
9899 @opindex mvolatile-asm-stop
9900 @opindex mno-volatile-asm-stop
9901 Generate (or don't) a stop bit immediately before and after volatile asm
9906 Generate code that works around Itanium B step errata.
9908 @item -mregister-names
9909 @itemx -mno-register-names
9910 @opindex mregister-names
9911 @opindex mno-register-names
9912 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9913 the stacked registers. This may make assembler output more readable.
9919 Disable (or enable) optimizations that use the small data section. This may
9920 be useful for working around optimizer bugs.
9923 @opindex mconstant-gp
9924 Generate code that uses a single constant global pointer value. This is
9925 useful when compiling kernel code.
9929 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9930 This is useful when compiling firmware code.
9932 @item -minline-float-divide-min-latency
9933 @opindex minline-float-divide-min-latency
9934 Generate code for inline divides of floating point values
9935 using the minimum latency algorithm.
9937 @item -minline-float-divide-max-throughput
9938 @opindex minline-float-divide-max-throughput
9939 Generate code for inline divides of floating point values
9940 using the maximum throughput algorithm.
9942 @item -minline-int-divide-min-latency
9943 @opindex minline-int-divide-min-latency
9944 Generate code for inline divides of integer values
9945 using the minimum latency algorithm.
9947 @item -minline-int-divide-max-throughput
9948 @opindex minline-int-divide-max-throughput
9949 Generate code for inline divides of integer values
9950 using the maximum throughput algorithm.
9952 @item -mno-dwarf2-asm
9954 @opindex mno-dwarf2-asm
9955 @opindex mdwarf2-asm
9956 Don't (or do) generate assembler code for the DWARF2 line number debugging
9957 info. This may be useful when not using the GNU assembler.
9959 @item -mfixed-range=@var{register-range}
9960 @opindex mfixed-range
9961 Generate code treating the given register range as fixed registers.
9962 A fixed register is one that the register allocator can not use. This is
9963 useful when compiling kernel code. A register range is specified as
9964 two registers separated by a dash. Multiple register ranges can be
9965 specified separated by a comma.
9967 @item -mearly-stop-bits
9968 @itemx -mno-early-stop-bits
9969 @opindex mearly-stop-bits
9970 @opindex mno-early-stop-bits
9971 Allow stop bits to be placed earlier than immediately preceding the
9972 instruction that triggered the stop bit. This can improve instruction
9973 scheduling, but does not always do so.
9976 @node S/390 and zSeries Options
9977 @subsection S/390 and zSeries Options
9978 @cindex S/390 and zSeries Options
9980 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
9985 @opindex mhard-float
9986 @opindex msoft-float
9987 Use (do not use) the hardware floating-point instructions and registers
9988 for floating-point operations. When @option{-msoft-float} is specified,
9989 functions in @file{libgcc.a} will be used to perform floating-point
9990 operations. When @option{-mhard-float} is specified, the compiler
9991 generates IEEE floating-point instructions. This is the default.
9994 @itemx -mno-backchain
9996 @opindex mno-backchain
9997 Generate (or do not generate) code which maintains an explicit
9998 backchain within the stack frame that points to the caller's frame.
9999 This may be needed to allow debugging using tools that do not understand
10000 DWARF-2 call frame information. The default is not to generate the
10004 @itemx -mno-small-exec
10005 @opindex msmall-exec
10006 @opindex mno-small-exec
10007 Generate (or do not generate) code using the @code{bras} instruction
10008 to do subroutine calls.
10009 This only works reliably if the total executable size does not
10010 exceed 64k. The default is to use the @code{basr} instruction instead,
10011 which does not have this limitation.
10017 When @option{-m31} is specified, generate code compliant to the
10018 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
10019 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
10020 particular to generate 64-bit instructions. For the @samp{s390}
10021 targets, the default is @option{-m31}, while the @samp{s390x}
10022 targets default to @option{-m64}.
10028 When @option{-mzarch} is specified, generate code using the
10029 instructions available on z/Architecture.
10030 When @option{-mesa} is specified, generate code using the
10031 instructions available on ESA/390. Note that @option{-mesa} is
10032 not possible with @option{-m64}.
10033 When generating code compliant to the GNU/Linux for S/390 ABI,
10034 the default is @option{-mesa}. When generating code compliant
10035 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
10041 Generate (or do not generate) code using the @code{mvcle} instruction
10042 to perform block moves. When @option{-mno-mvcle} is specified,
10043 use a @code{mvc} loop instead. This is the default.
10049 Print (or do not print) additional debug information when compiling.
10050 The default is to not print debug information.
10052 @item -march=@var{cpu-type}
10054 Generate code that will run on @var{cpu-type}, which is the name of a system
10055 representing a certain processor type. Possible values for
10056 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10057 When generating code using the instructions available on z/Architecture,
10058 the default is @option{-march=z900}. Otherwise, the default is
10059 @option{-march=g5}.
10061 @item -mtune=@var{cpu-type}
10063 Tune to @var{cpu-type} everything applicable about the generated code,
10064 except for the ABI and the set of available instructions.
10065 The list of @var{cpu-type} values is the same as for @option{-march}.
10066 The default is the value used for @option{-march}.
10069 @itemx -mno-fused-madd
10070 @opindex mfused-madd
10071 @opindex mno-fused-madd
10072 Generate code that uses (does not use) the floating point multiply and
10073 accumulate instructions. These instructions are generated by default if
10074 hardware floating point is used.
10078 @subsection CRIS Options
10079 @cindex CRIS Options
10081 These options are defined specifically for the CRIS ports.
10084 @item -march=@var{architecture-type}
10085 @itemx -mcpu=@var{architecture-type}
10088 Generate code for the specified architecture. The choices for
10089 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10090 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
10091 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10094 @item -mtune=@var{architecture-type}
10096 Tune to @var{architecture-type} everything applicable about the generated
10097 code, except for the ABI and the set of available instructions. The
10098 choices for @var{architecture-type} are the same as for
10099 @option{-march=@var{architecture-type}}.
10101 @item -mmax-stack-frame=@var{n}
10102 @opindex mmax-stack-frame
10103 Warn when the stack frame of a function exceeds @var{n} bytes.
10105 @item -melinux-stacksize=@var{n}
10106 @opindex melinux-stacksize
10107 Only available with the @samp{cris-axis-aout} target. Arranges for
10108 indications in the program to the kernel loader that the stack of the
10109 program should be set to @var{n} bytes.
10115 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10116 @option{-march=v3} and @option{-march=v8} respectively.
10118 @item -mmul-bug-workaround
10119 @itemx -mno-mul-bug-workaround
10120 @opindex mmul-bug-workaround
10121 @opindex mno-mul-bug-workaround
10122 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
10123 models where it applies. This option is active by default.
10127 Enable CRIS-specific verbose debug-related information in the assembly
10128 code. This option also has the effect to turn off the @samp{#NO_APP}
10129 formatted-code indicator to the assembler at the beginning of the
10134 Do not use condition-code results from previous instruction; always emit
10135 compare and test instructions before use of condition codes.
10137 @item -mno-side-effects
10138 @opindex mno-side-effects
10139 Do not emit instructions with side-effects in addressing modes other than
10142 @item -mstack-align
10143 @itemx -mno-stack-align
10144 @itemx -mdata-align
10145 @itemx -mno-data-align
10146 @itemx -mconst-align
10147 @itemx -mno-const-align
10148 @opindex mstack-align
10149 @opindex mno-stack-align
10150 @opindex mdata-align
10151 @opindex mno-data-align
10152 @opindex mconst-align
10153 @opindex mno-const-align
10154 These options (no-options) arranges (eliminate arrangements) for the
10155 stack-frame, individual data and constants to be aligned for the maximum
10156 single data access size for the chosen CPU model. The default is to
10157 arrange for 32-bit alignment. ABI details such as structure layout are
10158 not affected by these options.
10166 Similar to the stack- data- and const-align options above, these options
10167 arrange for stack-frame, writable data and constants to all be 32-bit,
10168 16-bit or 8-bit aligned. The default is 32-bit alignment.
10170 @item -mno-prologue-epilogue
10171 @itemx -mprologue-epilogue
10172 @opindex mno-prologue-epilogue
10173 @opindex mprologue-epilogue
10174 With @option{-mno-prologue-epilogue}, the normal function prologue and
10175 epilogue that sets up the stack-frame are omitted and no return
10176 instructions or return sequences are generated in the code. Use this
10177 option only together with visual inspection of the compiled code: no
10178 warnings or errors are generated when call-saved registers must be saved,
10179 or storage for local variable needs to be allocated.
10183 @opindex mno-gotplt
10185 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10186 instruction sequences that load addresses for functions from the PLT part
10187 of the GOT rather than (traditional on other architectures) calls to the
10188 PLT. The default is @option{-mgotplt}.
10192 Legacy no-op option only recognized with the cris-axis-aout target.
10196 Legacy no-op option only recognized with the cris-axis-elf and
10197 cris-axis-linux-gnu targets.
10201 Only recognized with the cris-axis-aout target, where it selects a
10202 GNU/linux-like multilib, include files and instruction set for
10203 @option{-march=v8}.
10207 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10211 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
10212 to link with input-output functions from a simulator library. Code,
10213 initialized data and zero-initialized data are allocated consecutively.
10217 Like @option{-sim}, but pass linker options to locate initialized data at
10218 0x40000000 and zero-initialized data at 0x80000000.
10222 @subsection MMIX Options
10223 @cindex MMIX Options
10225 These options are defined for the MMIX:
10229 @itemx -mno-libfuncs
10231 @opindex mno-libfuncs
10232 Specify that intrinsic library functions are being compiled, passing all
10233 values in registers, no matter the size.
10236 @itemx -mno-epsilon
10238 @opindex mno-epsilon
10239 Generate floating-point comparison instructions that compare with respect
10240 to the @code{rE} epsilon register.
10242 @item -mabi=mmixware
10244 @opindex mabi-mmixware
10246 Generate code that passes function parameters and return values that (in
10247 the called function) are seen as registers @code{$0} and up, as opposed to
10248 the GNU ABI which uses global registers @code{$231} and up.
10250 @item -mzero-extend
10251 @itemx -mno-zero-extend
10252 @opindex mzero-extend
10253 @opindex mno-zero-extend
10254 When reading data from memory in sizes shorter than 64 bits, use (do not
10255 use) zero-extending load instructions by default, rather than
10256 sign-extending ones.
10259 @itemx -mno-knuthdiv
10261 @opindex mno-knuthdiv
10262 Make the result of a division yielding a remainder have the same sign as
10263 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10264 remainder follows the sign of the dividend. Both methods are
10265 arithmetically valid, the latter being almost exclusively used.
10267 @item -mtoplevel-symbols
10268 @itemx -mno-toplevel-symbols
10269 @opindex mtoplevel-symbols
10270 @opindex mno-toplevel-symbols
10271 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10272 code can be used with the @code{PREFIX} assembly directive.
10276 Generate an executable in the ELF format, rather than the default
10277 @samp{mmo} format used by the @command{mmix} simulator.
10279 @item -mbranch-predict
10280 @itemx -mno-branch-predict
10281 @opindex mbranch-predict
10282 @opindex mno-branch-predict
10283 Use (do not use) the probable-branch instructions, when static branch
10284 prediction indicates a probable branch.
10286 @item -mbase-addresses
10287 @itemx -mno-base-addresses
10288 @opindex mbase-addresses
10289 @opindex mno-base-addresses
10290 Generate (do not generate) code that uses @emph{base addresses}. Using a
10291 base address automatically generates a request (handled by the assembler
10292 and the linker) for a constant to be set up in a global register. The
10293 register is used for one or more base address requests within the range 0
10294 to 255 from the value held in the register. The generally leads to short
10295 and fast code, but the number of different data items that can be
10296 addressed is limited. This means that a program that uses lots of static
10297 data may require @option{-mno-base-addresses}.
10299 @item -msingle-exit
10300 @itemx -mno-single-exit
10301 @opindex msingle-exit
10302 @opindex mno-single-exit
10303 Force (do not force) generated code to have a single exit point in each
10307 @node PDP-11 Options
10308 @subsection PDP-11 Options
10309 @cindex PDP-11 Options
10311 These options are defined for the PDP-11:
10316 Use hardware FPP floating point. This is the default. (FIS floating
10317 point on the PDP-11/40 is not supported.)
10320 @opindex msoft-float
10321 Do not use hardware floating point.
10325 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10329 Return floating-point results in memory. This is the default.
10333 Generate code for a PDP-11/40.
10337 Generate code for a PDP-11/45. This is the default.
10341 Generate code for a PDP-11/10.
10343 @item -mbcopy-builtin
10344 @opindex bcopy-builtin
10345 Use inline @code{movstrhi} patterns for copying memory. This is the
10350 Do not use inline @code{movstrhi} patterns for copying memory.
10356 Use 16-bit @code{int}. This is the default.
10362 Use 32-bit @code{int}.
10365 @itemx -mno-float32
10367 @opindex mno-float32
10368 Use 64-bit @code{float}. This is the default.
10371 @itemx -mno-float64
10373 @opindex mno-float64
10374 Use 32-bit @code{float}.
10378 Use @code{abshi2} pattern. This is the default.
10382 Do not use @code{abshi2} pattern.
10384 @item -mbranch-expensive
10385 @opindex mbranch-expensive
10386 Pretend that branches are expensive. This is for experimenting with
10387 code generation only.
10389 @item -mbranch-cheap
10390 @opindex mbranch-cheap
10391 Do not pretend that branches are expensive. This is the default.
10395 Generate code for a system with split I&D.
10399 Generate code for a system without split I&D. This is the default.
10403 Use Unix assembler syntax. This is the default when configured for
10404 @samp{pdp11-*-bsd}.
10408 Use DEC assembler syntax. This is the default when configured for any
10409 PDP-11 target other than @samp{pdp11-*-bsd}.
10412 @node Xstormy16 Options
10413 @subsection Xstormy16 Options
10414 @cindex Xstormy16 Options
10416 These options are defined for Xstormy16:
10421 Choose startup files and linker script suitable for the simulator.
10425 @subsection FRV Options
10426 @cindex FRV Options
10432 Only use the first 32 general purpose registers.
10437 Use all 64 general purpose registers.
10442 Use only the first 32 floating point registers.
10447 Use all 64 floating point registers
10450 @opindex mhard-float
10452 Use hardware instructions for floating point operations.
10455 @opindex msoft-float
10457 Use library routines for floating point operations.
10462 Dynamically allocate condition code registers.
10467 Do not try to dynamically allocate condition code registers, only
10468 use @code{icc0} and @code{fcc0}.
10473 Change ABI to use double word insns.
10478 Do not use double word instructions.
10483 Use floating point double instructions.
10486 @opindex mno-double
10488 Do not use floating point double instructions.
10493 Use media instructions.
10498 Do not use media instructions.
10503 Use multiply and add/subtract instructions.
10506 @opindex mno-muladd
10508 Do not use multiply and add/subtract instructions.
10510 @item -mlibrary-pic
10511 @opindex mlibrary-pic
10513 Generate position-independent EABI code.
10518 Use only the first four media accumulator registers.
10523 Use all eight media accumulator registers.
10528 Pack VLIW instructions.
10533 Do not pack VLIW instructions.
10536 @opindex mno-eflags
10538 Do not mark ABI switches in e_flags.
10541 @opindex mcond-move
10543 Enable the use of conditional-move instructions (default).
10545 This switch is mainly for debugging the compiler and will likely be removed
10546 in a future version.
10548 @item -mno-cond-move
10549 @opindex mno-cond-move
10551 Disable the use of conditional-move instructions.
10553 This switch is mainly for debugging the compiler and will likely be removed
10554 in a future version.
10559 Enable the use of conditional set instructions (default).
10561 This switch is mainly for debugging the compiler and will likely be removed
10562 in a future version.
10567 Disable the use of conditional set instructions.
10569 This switch is mainly for debugging the compiler and will likely be removed
10570 in a future version.
10573 @opindex mcond-exec
10575 Enable the use of conditional execution (default).
10577 This switch is mainly for debugging the compiler and will likely be removed
10578 in a future version.
10580 @item -mno-cond-exec
10581 @opindex mno-cond-exec
10583 Disable the use of conditional execution.
10585 This switch is mainly for debugging the compiler and will likely be removed
10586 in a future version.
10588 @item -mvliw-branch
10589 @opindex mvliw-branch
10591 Run a pass to pack branches into VLIW instructions (default).
10593 This switch is mainly for debugging the compiler and will likely be removed
10594 in a future version.
10596 @item -mno-vliw-branch
10597 @opindex mno-vliw-branch
10599 Do not run a pass to pack branches into VLIW instructions.
10601 This switch is mainly for debugging the compiler and will likely be removed
10602 in a future version.
10604 @item -mmulti-cond-exec
10605 @opindex mmulti-cond-exec
10607 Enable optimization of @code{&&} and @code{||} in conditional execution
10610 This switch is mainly for debugging the compiler and will likely be removed
10611 in a future version.
10613 @item -mno-multi-cond-exec
10614 @opindex mno-multi-cond-exec
10616 Disable optimization of @code{&&} and @code{||} in conditional execution.
10618 This switch is mainly for debugging the compiler and will likely be removed
10619 in a future version.
10621 @item -mnested-cond-exec
10622 @opindex mnested-cond-exec
10624 Enable nested conditional execution optimizations (default).
10626 This switch is mainly for debugging the compiler and will likely be removed
10627 in a future version.
10629 @item -mno-nested-cond-exec
10630 @opindex mno-nested-cond-exec
10632 Disable nested conditional execution optimizations.
10634 This switch is mainly for debugging the compiler and will likely be removed
10635 in a future version.
10637 @item -mtomcat-stats
10638 @opindex mtomcat-stats
10640 Cause gas to print out tomcat statistics.
10642 @item -mcpu=@var{cpu}
10645 Select the processor type for which to generate code. Possible values are
10646 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
10651 @node Xtensa Options
10652 @subsection Xtensa Options
10653 @cindex Xtensa Options
10655 These options are supported for Xtensa targets:
10659 @itemx -mno-const16
10661 @opindex mno-const16
10662 Enable or disable use of @code{CONST16} instructions for loading
10663 constant values. The @code{CONST16} instruction is currently not a
10664 standard option from Tensilica. When enabled, @code{CONST16}
10665 instructions are always used in place of the standard @code{L32R}
10666 instructions. The use of @code{CONST16} is enabled by default only if
10667 the @code{L32R} instruction is not available.
10670 @itemx -mno-fused-madd
10671 @opindex mfused-madd
10672 @opindex mno-fused-madd
10673 Enable or disable use of fused multiply/add and multiply/subtract
10674 instructions in the floating-point option. This has no effect if the
10675 floating-point option is not also enabled. Disabling fused multiply/add
10676 and multiply/subtract instructions forces the compiler to use separate
10677 instructions for the multiply and add/subtract operations. This may be
10678 desirable in some cases where strict IEEE 754-compliant results are
10679 required: the fused multiply add/subtract instructions do not round the
10680 intermediate result, thereby producing results with @emph{more} bits of
10681 precision than specified by the IEEE standard. Disabling fused multiply
10682 add/subtract instructions also ensures that the program output is not
10683 sensitive to the compiler's ability to combine multiply and add/subtract
10686 @item -mtext-section-literals
10687 @itemx -mno-text-section-literals
10688 @opindex mtext-section-literals
10689 @opindex mno-text-section-literals
10690 Control the treatment of literal pools. The default is
10691 @option{-mno-text-section-literals}, which places literals in a separate
10692 section in the output file. This allows the literal pool to be placed
10693 in a data RAM/ROM, and it also allows the linker to combine literal
10694 pools from separate object files to remove redundant literals and
10695 improve code size. With @option{-mtext-section-literals}, the literals
10696 are interspersed in the text section in order to keep them as close as
10697 possible to their references. This may be necessary for large assembly
10700 @item -mtarget-align
10701 @itemx -mno-target-align
10702 @opindex mtarget-align
10703 @opindex mno-target-align
10704 When this option is enabled, GCC instructs the assembler to
10705 automatically align instructions to reduce branch penalties at the
10706 expense of some code density. The assembler attempts to widen density
10707 instructions to align branch targets and the instructions following call
10708 instructions. If there are not enough preceding safe density
10709 instructions to align a target, no widening will be performed. The
10710 default is @option{-mtarget-align}. These options do not affect the
10711 treatment of auto-aligned instructions like @code{LOOP}, which the
10712 assembler will always align, either by widening density instructions or
10713 by inserting no-op instructions.
10716 @itemx -mno-longcalls
10717 @opindex mlongcalls
10718 @opindex mno-longcalls
10719 When this option is enabled, GCC instructs the assembler to translate
10720 direct calls to indirect calls unless it can determine that the target
10721 of a direct call is in the range allowed by the call instruction. This
10722 translation typically occurs for calls to functions in other source
10723 files. Specifically, the assembler translates a direct @code{CALL}
10724 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
10725 The default is @option{-mno-longcalls}. This option should be used in
10726 programs where the call target can potentially be out of range. This
10727 option is implemented in the assembler, not the compiler, so the
10728 assembly code generated by GCC will still show direct call
10729 instructions---look at the disassembled object code to see the actual
10730 instructions. Note that the assembler will use an indirect call for
10731 every cross-file call, not just those that really will be out of range.
10734 @node Code Gen Options
10735 @section Options for Code Generation Conventions
10736 @cindex code generation conventions
10737 @cindex options, code generation
10738 @cindex run-time options
10740 These machine-independent options control the interface conventions
10741 used in code generation.
10743 Most of them have both positive and negative forms; the negative form
10744 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
10745 one of the forms is listed---the one which is not the default. You
10746 can figure out the other form by either removing @samp{no-} or adding
10750 @item -fbounds-check
10751 @opindex fbounds-check
10752 For front-ends that support it, generate additional code to check that
10753 indices used to access arrays are within the declared range. This is
10754 currently only supported by the Java and Fortran 77 front-ends, where
10755 this option defaults to true and false respectively.
10759 This option generates traps for signed overflow on addition, subtraction,
10760 multiplication operations.
10764 This option instructs the compiler to assume that signed arithmetic
10765 overflow of addition, subtraction and multiplication wraps around
10766 using twos-complement representation. This flag enables some optimizations
10767 and disables other. This option is enabled by default for the Java
10768 front-end, as required by the Java language specification.
10771 @opindex fexceptions
10772 Enable exception handling. Generates extra code needed to propagate
10773 exceptions. For some targets, this implies GCC will generate frame
10774 unwind information for all functions, which can produce significant data
10775 size overhead, although it does not affect execution. If you do not
10776 specify this option, GCC will enable it by default for languages like
10777 C++ which normally require exception handling, and disable it for
10778 languages like C that do not normally require it. However, you may need
10779 to enable this option when compiling C code that needs to interoperate
10780 properly with exception handlers written in C++. You may also wish to
10781 disable this option if you are compiling older C++ programs that don't
10782 use exception handling.
10784 @item -fnon-call-exceptions
10785 @opindex fnon-call-exceptions
10786 Generate code that allows trapping instructions to throw exceptions.
10787 Note that this requires platform-specific runtime support that does
10788 not exist everywhere. Moreover, it only allows @emph{trapping}
10789 instructions to throw exceptions, i.e.@: memory references or floating
10790 point instructions. It does not allow exceptions to be thrown from
10791 arbitrary signal handlers such as @code{SIGALRM}.
10793 @item -funwind-tables
10794 @opindex funwind-tables
10795 Similar to @option{-fexceptions}, except that it will just generate any needed
10796 static data, but will not affect the generated code in any other way.
10797 You will normally not enable this option; instead, a language processor
10798 that needs this handling would enable it on your behalf.
10800 @item -fasynchronous-unwind-tables
10801 @opindex funwind-tables
10802 Generate unwind table in dwarf2 format, if supported by target machine. The
10803 table is exact at each instruction boundary, so it can be used for stack
10804 unwinding from asynchronous events (such as debugger or garbage collector).
10806 @item -fpcc-struct-return
10807 @opindex fpcc-struct-return
10808 Return ``short'' @code{struct} and @code{union} values in memory like
10809 longer ones, rather than in registers. This convention is less
10810 efficient, but it has the advantage of allowing intercallability between
10811 GCC-compiled files and files compiled with other compilers, particularly
10812 the Portable C Compiler (pcc).
10814 The precise convention for returning structures in memory depends
10815 on the target configuration macros.
10817 Short structures and unions are those whose size and alignment match
10818 that of some integer type.
10820 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
10821 switch is not binary compatible with code compiled with the
10822 @option{-freg-struct-return} switch.
10823 Use it to conform to a non-default application binary interface.
10825 @item -freg-struct-return
10826 @opindex freg-struct-return
10827 Return @code{struct} and @code{union} values in registers when possible.
10828 This is more efficient for small structures than
10829 @option{-fpcc-struct-return}.
10831 If you specify neither @option{-fpcc-struct-return} nor
10832 @option{-freg-struct-return}, GCC defaults to whichever convention is
10833 standard for the target. If there is no standard convention, GCC
10834 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
10835 the principal compiler. In those cases, we can choose the standard, and
10836 we chose the more efficient register return alternative.
10838 @strong{Warning:} code compiled with the @option{-freg-struct-return}
10839 switch is not binary compatible with code compiled with the
10840 @option{-fpcc-struct-return} switch.
10841 Use it to conform to a non-default application binary interface.
10843 @item -fshort-enums
10844 @opindex fshort-enums
10845 Allocate to an @code{enum} type only as many bytes as it needs for the
10846 declared range of possible values. Specifically, the @code{enum} type
10847 will be equivalent to the smallest integer type which has enough room.
10849 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
10850 code that is not binary compatible with code generated without that switch.
10851 Use it to conform to a non-default application binary interface.
10853 @item -fshort-double
10854 @opindex fshort-double
10855 Use the same size for @code{double} as for @code{float}.
10857 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
10858 code that is not binary compatible with code generated without that switch.
10859 Use it to conform to a non-default application binary interface.
10861 @item -fshort-wchar
10862 @opindex fshort-wchar
10863 Override the underlying type for @samp{wchar_t} to be @samp{short
10864 unsigned int} instead of the default for the target. This option is
10865 useful for building programs to run under WINE@.
10867 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
10868 code that is not binary compatible with code generated without that switch.
10869 Use it to conform to a non-default application binary interface.
10871 @item -fshared-data
10872 @opindex fshared-data
10873 Requests that the data and non-@code{const} variables of this
10874 compilation be shared data rather than private data. The distinction
10875 makes sense only on certain operating systems, where shared data is
10876 shared between processes running the same program, while private data
10877 exists in one copy per process.
10880 @opindex fno-common
10881 In C, allocate even uninitialized global variables in the data section of the
10882 object file, rather than generating them as common blocks. This has the
10883 effect that if the same variable is declared (without @code{extern}) in
10884 two different compilations, you will get an error when you link them.
10885 The only reason this might be useful is if you wish to verify that the
10886 program will work on other systems which always work this way.
10890 Ignore the @samp{#ident} directive.
10892 @item -finhibit-size-directive
10893 @opindex finhibit-size-directive
10894 Don't output a @code{.size} assembler directive, or anything else that
10895 would cause trouble if the function is split in the middle, and the
10896 two halves are placed at locations far apart in memory. This option is
10897 used when compiling @file{crtstuff.c}; you should not need to use it
10900 @item -fverbose-asm
10901 @opindex fverbose-asm
10902 Put extra commentary information in the generated assembly code to
10903 make it more readable. This option is generally only of use to those
10904 who actually need to read the generated assembly code (perhaps while
10905 debugging the compiler itself).
10907 @option{-fno-verbose-asm}, the default, causes the
10908 extra information to be omitted and is useful when comparing two assembler
10913 @cindex global offset table
10915 Generate position-independent code (PIC) suitable for use in a shared
10916 library, if supported for the target machine. Such code accesses all
10917 constant addresses through a global offset table (GOT)@. The dynamic
10918 loader resolves the GOT entries when the program starts (the dynamic
10919 loader is not part of GCC; it is part of the operating system). If
10920 the GOT size for the linked executable exceeds a machine-specific
10921 maximum size, you get an error message from the linker indicating that
10922 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
10923 instead. (These maximums are 8k on the SPARC and 32k
10924 on the m68k and RS/6000. The 386 has no such limit.)
10926 Position-independent code requires special support, and therefore works
10927 only on certain machines. For the 386, GCC supports PIC for System V
10928 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
10929 position-independent.
10933 If supported for the target machine, emit position-independent code,
10934 suitable for dynamic linking and avoiding any limit on the size of the
10935 global offset table. This option makes a difference on the m68k
10938 Position-independent code requires special support, and therefore works
10939 only on certain machines.
10945 These options are similar to @option{-fpic} and @option{-fPIC}, but
10946 generated position independent code can be only linked into executables.
10947 Usually these options are used when @option{-pie} GCC option will be
10948 used during linking.
10950 @item -ffixed-@var{reg}
10952 Treat the register named @var{reg} as a fixed register; generated code
10953 should never refer to it (except perhaps as a stack pointer, frame
10954 pointer or in some other fixed role).
10956 @var{reg} must be the name of a register. The register names accepted
10957 are machine-specific and are defined in the @code{REGISTER_NAMES}
10958 macro in the machine description macro file.
10960 This flag does not have a negative form, because it specifies a
10963 @item -fcall-used-@var{reg}
10964 @opindex fcall-used
10965 Treat the register named @var{reg} as an allocable register that is
10966 clobbered by function calls. It may be allocated for temporaries or
10967 variables that do not live across a call. Functions compiled this way
10968 will not save and restore the register @var{reg}.
10970 It is an error to used this flag with the frame pointer or stack pointer.
10971 Use of this flag for other registers that have fixed pervasive roles in
10972 the machine's execution model will produce disastrous results.
10974 This flag does not have a negative form, because it specifies a
10977 @item -fcall-saved-@var{reg}
10978 @opindex fcall-saved
10979 Treat the register named @var{reg} as an allocable register saved by
10980 functions. It may be allocated even for temporaries or variables that
10981 live across a call. Functions compiled this way will save and restore
10982 the register @var{reg} if they use it.
10984 It is an error to used this flag with the frame pointer or stack pointer.
10985 Use of this flag for other registers that have fixed pervasive roles in
10986 the machine's execution model will produce disastrous results.
10988 A different sort of disaster will result from the use of this flag for
10989 a register in which function values may be returned.
10991 This flag does not have a negative form, because it specifies a
10994 @item -fpack-struct
10995 @opindex fpack-struct
10996 Pack all structure members together without holes.
10998 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
10999 code that is not binary compatible with code generated without that switch.
11000 Additionally, it makes the code suboptimal.
11001 Use it to conform to a non-default application binary interface.
11003 @item -finstrument-functions
11004 @opindex finstrument-functions
11005 Generate instrumentation calls for entry and exit to functions. Just
11006 after function entry and just before function exit, the following
11007 profiling functions will be called with the address of the current
11008 function and its call site. (On some platforms,
11009 @code{__builtin_return_address} does not work beyond the current
11010 function, so the call site information may not be available to the
11011 profiling functions otherwise.)
11014 void __cyg_profile_func_enter (void *this_fn,
11016 void __cyg_profile_func_exit (void *this_fn,
11020 The first argument is the address of the start of the current function,
11021 which may be looked up exactly in the symbol table.
11023 This instrumentation is also done for functions expanded inline in other
11024 functions. The profiling calls will indicate where, conceptually, the
11025 inline function is entered and exited. This means that addressable
11026 versions of such functions must be available. If all your uses of a
11027 function are expanded inline, this may mean an additional expansion of
11028 code size. If you use @samp{extern inline} in your C code, an
11029 addressable version of such functions must be provided. (This is
11030 normally the case anyways, but if you get lucky and the optimizer always
11031 expands the functions inline, you might have gotten away without
11032 providing static copies.)
11034 A function may be given the attribute @code{no_instrument_function}, in
11035 which case this instrumentation will not be done. This can be used, for
11036 example, for the profiling functions listed above, high-priority
11037 interrupt routines, and any functions from which the profiling functions
11038 cannot safely be called (perhaps signal handlers, if the profiling
11039 routines generate output or allocate memory).
11041 @item -fstack-check
11042 @opindex fstack-check
11043 Generate code to verify that you do not go beyond the boundary of the
11044 stack. You should specify this flag if you are running in an
11045 environment with multiple threads, but only rarely need to specify it in
11046 a single-threaded environment since stack overflow is automatically
11047 detected on nearly all systems if there is only one stack.
11049 Note that this switch does not actually cause checking to be done; the
11050 operating system must do that. The switch causes generation of code
11051 to ensure that the operating system sees the stack being extended.
11053 @item -fstack-limit-register=@var{reg}
11054 @itemx -fstack-limit-symbol=@var{sym}
11055 @itemx -fno-stack-limit
11056 @opindex fstack-limit-register
11057 @opindex fstack-limit-symbol
11058 @opindex fno-stack-limit
11059 Generate code to ensure that the stack does not grow beyond a certain value,
11060 either the value of a register or the address of a symbol. If the stack
11061 would grow beyond the value, a signal is raised. For most targets,
11062 the signal is raised before the stack overruns the boundary, so
11063 it is possible to catch the signal without taking special precautions.
11065 For instance, if the stack starts at absolute address @samp{0x80000000}
11066 and grows downwards, you can use the flags
11067 @option{-fstack-limit-symbol=__stack_limit} and
11068 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11069 of 128KB@. Note that this may only work with the GNU linker.
11071 @cindex aliasing of parameters
11072 @cindex parameters, aliased
11073 @item -fargument-alias
11074 @itemx -fargument-noalias
11075 @itemx -fargument-noalias-global
11076 @opindex fargument-alias
11077 @opindex fargument-noalias
11078 @opindex fargument-noalias-global
11079 Specify the possible relationships among parameters and between
11080 parameters and global data.
11082 @option{-fargument-alias} specifies that arguments (parameters) may
11083 alias each other and may alias global storage.@*
11084 @option{-fargument-noalias} specifies that arguments do not alias
11085 each other, but may alias global storage.@*
11086 @option{-fargument-noalias-global} specifies that arguments do not
11087 alias each other and do not alias global storage.
11089 Each language will automatically use whatever option is required by
11090 the language standard. You should not need to use these options yourself.
11092 @item -fleading-underscore
11093 @opindex fleading-underscore
11094 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11095 change the way C symbols are represented in the object file. One use
11096 is to help link with legacy assembly code.
11098 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11099 generate code that is not binary compatible with code generated without that
11100 switch. Use it to conform to a non-default application binary interface.
11101 Not all targets provide complete support for this switch.
11103 @item -ftls-model=@var{model}
11104 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11105 The @var{model} argument should be one of @code{global-dynamic},
11106 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11108 The default without @option{-fpic} is @code{initial-exec}; with
11109 @option{-fpic} the default is @code{global-dynamic}.
11114 @node Environment Variables
11115 @section Environment Variables Affecting GCC
11116 @cindex environment variables
11118 @c man begin ENVIRONMENT
11119 This section describes several environment variables that affect how GCC
11120 operates. Some of them work by specifying directories or prefixes to use
11121 when searching for various kinds of files. Some are used to specify other
11122 aspects of the compilation environment.
11124 Note that you can also specify places to search using options such as
11125 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11126 take precedence over places specified using environment variables, which
11127 in turn take precedence over those specified by the configuration of GCC@.
11128 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11129 GNU Compiler Collection (GCC) Internals}.
11134 @c @itemx LC_COLLATE
11136 @c @itemx LC_MONETARY
11137 @c @itemx LC_NUMERIC
11142 @c @findex LC_COLLATE
11143 @findex LC_MESSAGES
11144 @c @findex LC_MONETARY
11145 @c @findex LC_NUMERIC
11149 These environment variables control the way that GCC uses
11150 localization information that allow GCC to work with different
11151 national conventions. GCC inspects the locale categories
11152 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11153 so. These locale categories can be set to any value supported by your
11154 installation. A typical value is @samp{en_UK} for English in the United
11157 The @env{LC_CTYPE} environment variable specifies character
11158 classification. GCC uses it to determine the character boundaries in
11159 a string; this is needed for some multibyte encodings that contain quote
11160 and escape characters that would otherwise be interpreted as a string
11163 The @env{LC_MESSAGES} environment variable specifies the language to
11164 use in diagnostic messages.
11166 If the @env{LC_ALL} environment variable is set, it overrides the value
11167 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11168 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11169 environment variable. If none of these variables are set, GCC
11170 defaults to traditional C English behavior.
11174 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11175 files. GCC uses temporary files to hold the output of one stage of
11176 compilation which is to be used as input to the next stage: for example,
11177 the output of the preprocessor, which is the input to the compiler
11180 @item GCC_EXEC_PREFIX
11181 @findex GCC_EXEC_PREFIX
11182 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11183 names of the subprograms executed by the compiler. No slash is added
11184 when this prefix is combined with the name of a subprogram, but you can
11185 specify a prefix that ends with a slash if you wish.
11187 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11188 an appropriate prefix to use based on the pathname it was invoked with.
11190 If GCC cannot find the subprogram using the specified prefix, it
11191 tries looking in the usual places for the subprogram.
11193 The default value of @env{GCC_EXEC_PREFIX} is
11194 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
11195 of @code{prefix} when you ran the @file{configure} script.
11197 Other prefixes specified with @option{-B} take precedence over this prefix.
11199 This prefix is also used for finding files such as @file{crt0.o} that are
11202 In addition, the prefix is used in an unusual way in finding the
11203 directories to search for header files. For each of the standard
11204 directories whose name normally begins with @samp{/usr/local/lib/gcc}
11205 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11206 replacing that beginning with the specified prefix to produce an
11207 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11208 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11209 These alternate directories are searched first; the standard directories
11212 @item COMPILER_PATH
11213 @findex COMPILER_PATH
11214 The value of @env{COMPILER_PATH} is a colon-separated list of
11215 directories, much like @env{PATH}. GCC tries the directories thus
11216 specified when searching for subprograms, if it can't find the
11217 subprograms using @env{GCC_EXEC_PREFIX}.
11220 @findex LIBRARY_PATH
11221 The value of @env{LIBRARY_PATH} is a colon-separated list of
11222 directories, much like @env{PATH}. When configured as a native compiler,
11223 GCC tries the directories thus specified when searching for special
11224 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11225 using GCC also uses these directories when searching for ordinary
11226 libraries for the @option{-l} option (but directories specified with
11227 @option{-L} come first).
11231 @cindex locale definition
11232 This variable is used to pass locale information to the compiler. One way in
11233 which this information is used is to determine the character set to be used
11234 when character literals, string literals and comments are parsed in C and C++.
11235 When the compiler is configured to allow multibyte characters,
11236 the following values for @env{LANG} are recognized:
11240 Recognize JIS characters.
11242 Recognize SJIS characters.
11244 Recognize EUCJP characters.
11247 If @env{LANG} is not defined, or if it has some other value, then the
11248 compiler will use mblen and mbtowc as defined by the default locale to
11249 recognize and translate multibyte characters.
11253 Some additional environments variables affect the behavior of the
11256 @include cppenv.texi
11260 @node Precompiled Headers
11261 @section Using Precompiled Headers
11262 @cindex precompiled headers
11263 @cindex speed of compilation
11265 Often large projects have many header files that are included in every
11266 source file. The time the compiler takes to process these header files
11267 over and over again can account for nearly all of the time required to
11268 build the project. To make builds faster, GCC allows users to
11269 `precompile' a header file; then, if builds can use the precompiled
11270 header file they will be much faster.
11272 To create a precompiled header file, simply compile it as you would any
11273 other file, if necessary using the @option{-x} option to make the driver
11274 treat it as a C or C++ header file. You will probably want to use a
11275 tool like @command{make} to keep the precompiled header up-to-date when
11276 the headers it contains change.
11278 A precompiled header file will be searched for when @code{#include} is
11279 seen in the compilation. As it searches for the included file
11280 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
11281 compiler looks for a precompiled header in each directory just before it
11282 looks for the include file in that directory. The name searched for is
11283 the name specified in the @code{#include} with @samp{.gch} appended. If
11284 the precompiled header file can't be used, it is ignored.
11286 For instance, if you have @code{#include "all.h"}, and you have
11287 @file{all.h.gch} in the same directory as @file{all.h}, then the
11288 precompiled header file will be used if possible, and the original
11289 header will be used otherwise.
11291 Alternatively, you might decide to put the precompiled header file in a
11292 directory and use @option{-I} to ensure that directory is searched
11293 before (or instead of) the directory containing the original header.
11294 Then, if you want to check that the precompiled header file is always
11295 used, you can put a file of the same name as the original header in this
11296 directory containing an @code{#error} command.
11298 This also works with @option{-include}. So yet another way to use
11299 precompiled headers, good for projects not designed with precompiled
11300 header files in mind, is to simply take most of the header files used by
11301 a project, include them from another header file, precompile that header
11302 file, and @option{-include} the precompiled header. If the header files
11303 have guards against multiple inclusion, they will be skipped because
11304 they've already been included (in the precompiled header).
11306 If you need to precompile the same header file for different
11307 languages, targets, or compiler options, you can instead make a
11308 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11309 header in the directory. (It doesn't matter what you call the files
11310 in the directory, every precompiled header in the directory will be
11311 considered.) The first precompiled header encountered in the
11312 directory that is valid for this compilation will be used; they're
11313 searched in no particular order.
11315 There are many other possibilities, limited only by your imagination,
11316 good sense, and the constraints of your build system.
11318 A precompiled header file can be used only when these conditions apply:
11322 Only one precompiled header can be used in a particular compilation.
11324 A precompiled header can't be used once the first C token is seen. You
11325 can have preprocessor directives before a precompiled header; you can
11326 even include a precompiled header from inside another header, so long as
11327 there are no C tokens before the @code{#include}.
11329 The precompiled header file must be produced for the same language as
11330 the current compilation. You can't use a C precompiled header for a C++
11333 The precompiled header file must be produced by the same compiler
11334 version and configuration as the current compilation is using.
11335 The easiest way to guarantee this is to use the same compiler binary
11336 for creating and using precompiled headers.
11338 Any macros defined before the precompiled header (including with
11339 @option{-D}) must either be defined in the same way as when the
11340 precompiled header was generated, or must not affect the precompiled
11341 header, which usually means that the they don't appear in the
11342 precompiled header at all.
11344 Certain command-line options must be defined in the same way as when the
11345 precompiled header was generated. At present, it's not clear which
11346 options are safe to change and which are not; the safest choice is to
11347 use exactly the same options when generating and using the precompiled
11351 For all of these but the last, the compiler will automatically ignore
11352 the precompiled header if the conditions aren't met. For the last item,
11353 some option changes will cause the precompiled header to be rejected,
11354 but not all incompatible option combinations have yet been found. If
11355 you find a new incompatible combination, please consider filing a bug
11356 report, see @ref{Bugs}.
11358 @node Running Protoize
11359 @section Running Protoize
11361 The program @code{protoize} is an optional part of GCC@. You can use
11362 it to add prototypes to a program, thus converting the program to ISO
11363 C in one respect. The companion program @code{unprotoize} does the
11364 reverse: it removes argument types from any prototypes that are found.
11366 When you run these programs, you must specify a set of source files as
11367 command line arguments. The conversion programs start out by compiling
11368 these files to see what functions they define. The information gathered
11369 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11371 After scanning comes actual conversion. The specified files are all
11372 eligible to be converted; any files they include (whether sources or
11373 just headers) are eligible as well.
11375 But not all the eligible files are converted. By default,
11376 @code{protoize} and @code{unprotoize} convert only source and header
11377 files in the current directory. You can specify additional directories
11378 whose files should be converted with the @option{-d @var{directory}}
11379 option. You can also specify particular files to exclude with the
11380 @option{-x @var{file}} option. A file is converted if it is eligible, its
11381 directory name matches one of the specified directory names, and its
11382 name within the directory has not been excluded.
11384 Basic conversion with @code{protoize} consists of rewriting most
11385 function definitions and function declarations to specify the types of
11386 the arguments. The only ones not rewritten are those for varargs
11389 @code{protoize} optionally inserts prototype declarations at the
11390 beginning of the source file, to make them available for any calls that
11391 precede the function's definition. Or it can insert prototype
11392 declarations with block scope in the blocks where undeclared functions
11395 Basic conversion with @code{unprotoize} consists of rewriting most
11396 function declarations to remove any argument types, and rewriting
11397 function definitions to the old-style pre-ISO form.
11399 Both conversion programs print a warning for any function declaration or
11400 definition that they can't convert. You can suppress these warnings
11403 The output from @code{protoize} or @code{unprotoize} replaces the
11404 original source file. The original file is renamed to a name ending
11405 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11406 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11407 for DOS) file already exists, then the source file is simply discarded.
11409 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11410 scan the program and collect information about the functions it uses.
11411 So neither of these programs will work until GCC is installed.
11413 Here is a table of the options you can use with @code{protoize} and
11414 @code{unprotoize}. Each option works with both programs unless
11418 @item -B @var{directory}
11419 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11420 usual directory (normally @file{/usr/local/lib}). This file contains
11421 prototype information about standard system functions. This option
11422 applies only to @code{protoize}.
11424 @item -c @var{compilation-options}
11425 Use @var{compilation-options} as the options when running @command{gcc} to
11426 produce the @samp{.X} files. The special option @option{-aux-info} is
11427 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11429 Note that the compilation options must be given as a single argument to
11430 @code{protoize} or @code{unprotoize}. If you want to specify several
11431 @command{gcc} options, you must quote the entire set of compilation options
11432 to make them a single word in the shell.
11434 There are certain @command{gcc} arguments that you cannot use, because they
11435 would produce the wrong kind of output. These include @option{-g},
11436 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11437 the @var{compilation-options}, they are ignored.
11440 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11441 systems) instead of @samp{.c}. This is convenient if you are converting
11442 a C program to C++. This option applies only to @code{protoize}.
11445 Add explicit global declarations. This means inserting explicit
11446 declarations at the beginning of each source file for each function
11447 that is called in the file and was not declared. These declarations
11448 precede the first function definition that contains a call to an
11449 undeclared function. This option applies only to @code{protoize}.
11451 @item -i @var{string}
11452 Indent old-style parameter declarations with the string @var{string}.
11453 This option applies only to @code{protoize}.
11455 @code{unprotoize} converts prototyped function definitions to old-style
11456 function definitions, where the arguments are declared between the
11457 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11458 uses five spaces as the indentation. If you want to indent with just
11459 one space instead, use @option{-i " "}.
11462 Keep the @samp{.X} files. Normally, they are deleted after conversion
11466 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11467 a prototype declaration for each function in each block which calls the
11468 function without any declaration. This option applies only to
11472 Make no real changes. This mode just prints information about the conversions
11473 that would have been done without @option{-n}.
11476 Make no @samp{.save} files. The original files are simply deleted.
11477 Use this option with caution.
11479 @item -p @var{program}
11480 Use the program @var{program} as the compiler. Normally, the name
11481 @file{gcc} is used.
11484 Work quietly. Most warnings are suppressed.
11487 Print the version number, just like @option{-v} for @command{gcc}.
11490 If you need special compiler options to compile one of your program's
11491 source files, then you should generate that file's @samp{.X} file
11492 specially, by running @command{gcc} on that source file with the
11493 appropriate options and the option @option{-aux-info}. Then run
11494 @code{protoize} on the entire set of files. @code{protoize} will use
11495 the existing @samp{.X} file because it is newer than the source file.
11499 gcc -Dfoo=bar file1.c -aux-info file1.X
11504 You need to include the special files along with the rest in the
11505 @code{protoize} command, even though their @samp{.X} files already
11506 exist, because otherwise they won't get converted.
11508 @xref{Protoize Caveats}, for more information on how to use
11509 @code{protoize} successfully.