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 @gol
172 @item C++ Language Options
173 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
174 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
175 -fconserve-space -fno-const-strings @gol
176 -fno-elide-constructors @gol
177 -fno-enforce-eh-specs @gol
178 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
179 -fno-implicit-templates @gol
180 -fno-implicit-inline-templates @gol
181 -fno-implement-inlines -fms-extensions @gol
182 -fno-nonansi-builtins -fno-operator-names @gol
183 -fno-optional-diags -fpermissive @gol
184 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
185 -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
186 -fno-default-inline -Wabi -Wctor-dtor-privacy @gol
187 -Wnon-virtual-dtor -Wreorder @gol
188 -Weffc++ -Wno-deprecated @gol
189 -Wno-non-template-friend -Wold-style-cast @gol
190 -Woverloaded-virtual -Wno-pmf-conversions @gol
191 -Wsign-promo -Wsynth}
193 @item Objective-C Language Options
194 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
196 -fconstant-string-class=@var{class-name} @gol
197 -fgnu-runtime -fnext-runtime @gol
198 -fno-nil-receivers @gol
199 -fobjc-exceptions @gol
200 -freplace-objc-classes @gol
203 -Wno-protocol -Wselector -Wundeclared-selector}
205 @item Language Independent Options
206 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
207 @gccoptlist{-fmessage-length=@var{n} @gol
208 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
210 @item Warning Options
211 @xref{Warning Options,,Options to Request or Suppress Warnings}.
212 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
213 -w -Wextra -Wall -Waggregate-return @gol
214 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
215 -Wconversion -Wno-deprecated-declarations @gol
216 -Wdisabled-optimization -Wno-div-by-zero -Wendif-labels @gol
217 -Werror -Werror-implicit-function-declaration @gol
218 -Wfloat-equal -Wformat -Wformat=2 @gol
219 -Wno-format-extra-args -Wformat-nonliteral @gol
220 -Wformat-security -Wformat-y2k @gol
221 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
222 -Wimport -Wno-import -Winit-self -Winline @gol
223 -Wno-invalid-offsetof -Winvalid-pch @gol
224 -Wlarger-than-@var{len} -Wlong-long @gol
225 -Wmain -Wmissing-braces @gol
226 -Wmissing-format-attribute -Wmissing-noreturn @gol
227 -Wno-multichar -Wnonnull -Wpacked -Wpadded @gol
228 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
229 -Wreturn-type -Wsequence-point -Wshadow @gol
230 -Wsign-compare -Wstrict-aliasing @gol
231 -Wswitch -Wswitch-default -Wswitch-enum @gol
232 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
233 -Wunknown-pragmas -Wunreachable-code @gol
234 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
235 -Wunused-value -Wunused-variable -Wwrite-strings}
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 @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-sb1 -mno-fix-sb1 -mflush-func=@var{func} @gol
482 -mno-flush-func -mbranch-likely -mno-branch-likely}
484 @emph{i386 and x86-64 Options}
485 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
486 -mfpmath=@var{unit} @gol
487 -masm=@var{dialect} -mno-fancy-math-387 @gol
488 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
489 -mno-wide-multiply -mrtd -malign-double @gol
490 -mpreferred-stack-boundary=@var{num} @gol
491 -mmmx -msse -msse2 -mpni -m3dnow @gol
492 -mthreads -mno-align-stringops -minline-all-stringops @gol
493 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
494 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
495 -mno-red-zone -mno-tls-direct-seg-refs @gol
496 -mcmodel=@var{code-model} @gol
500 @gccoptlist{-march=@var{architecture-type} @gol
501 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
502 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
503 -mjump-in-delay -mlinker-opt -mlong-calls @gol
504 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
505 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
506 -mno-jump-in-delay -mno-long-load-store @gol
507 -mno-portable-runtime -mno-soft-float @gol
508 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
509 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
510 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
511 -nolibdld -static -threads}
513 @emph{DEC Alpha Options}
514 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
515 -mieee -mieee-with-inexact -mieee-conformant @gol
516 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
517 -mtrap-precision=@var{mode} -mbuild-constants @gol
518 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
519 -mbwx -mmax -mfix -mcix @gol
520 -mfloat-vax -mfloat-ieee @gol
521 -mexplicit-relocs -msmall-data -mlarge-data @gol
522 -msmall-text -mlarge-text @gol
523 -mmemory-latency=@var{time}}
525 @emph{DEC Alpha/VMS Options}
526 @gccoptlist{-mvms-return-codes}
528 @emph{H8/300 Options}
529 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
532 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
533 -m4-nofpu -m4-single-only -m4-single -m4 @gol
534 -m5-64media -m5-64media-nofpu @gol
535 -m5-32media -m5-32media-nofpu @gol
536 -m5-compact -m5-compact-nofpu @gol
537 -mb -ml -mdalign -mrelax @gol
538 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
539 -mieee -misize -mpadstruct -mspace @gol
540 -mprefergot -musermode}
542 @emph{System V Options}
543 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
546 @gccoptlist{-EB -EL @gol
547 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
548 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
550 @emph{TMS320C3x/C4x Options}
551 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
552 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
553 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
554 -mparallel-insns -mparallel-mpy -mpreserve-float}
557 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
558 -mprolog-function -mno-prolog-function -mspace @gol
559 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
560 -mapp-regs -mno-app-regs @gol
561 -mdisable-callt -mno-disable-callt @gol
567 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
568 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
569 -mregparam -mnoregparam -msb -mnosb @gol
570 -mbitfield -mnobitfield -mhimem -mnohimem}
573 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
574 -mcall-prologues -mno-tablejump -mtiny-stack}
577 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
578 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
579 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
580 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
581 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
584 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
585 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
586 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
587 -mno-base-addresses -msingle-exit -mno-single-exit}
590 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
591 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
592 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
593 -minline-float-divide-max-throughput @gol
594 -minline-int-divide-min-latency @gol
595 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
596 -mfixed-range=@var{register-range}}
598 @emph{S/390 and zSeries Options}
599 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
600 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
601 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
602 -m64 -m31 -mdebug -mno-debug -mesa -mzarch -mfused-madd -mno-fused-madd}
605 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
606 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
607 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
608 -mstack-align -mdata-align -mconst-align @gol
609 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
610 -melf -maout -melinux -mlinux -sim -sim2}
612 @emph{PDP-11 Options}
613 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
614 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
615 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
616 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
617 -mbranch-expensive -mbranch-cheap @gol
618 -msplit -mno-split -munix-asm -mdec-asm}
620 @emph{Xstormy16 Options}
623 @emph{Xtensa Options}
624 @gccoptlist{-mconst16 -mno-const16 @gol
625 -mfused-madd -mno-fused-madd @gol
626 -mtext-section-literals -mno-text-section-literals @gol
627 -mtarget-align -mno-target-align @gol
628 -mlongcalls -mno-longcalls}
631 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
632 -mhard-float -msoft-float @gol
633 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
634 -mdouble -mno-double @gol
635 -mmedia -mno-media -mmuladd -mno-muladd @gol
636 -mlibrary-pic -macc-4 -macc-8 @gol
637 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
638 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
639 -mvliw-branch -mno-vliw-branch @gol
640 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
641 -mno-nested-cond-exec -mtomcat-stats @gol
644 @item Code Generation Options
645 @xref{Code Gen Options,,Options for Code Generation Conventions}.
646 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
647 -ffixed-@var{reg} -fexceptions @gol
648 -fnon-call-exceptions -funwind-tables @gol
649 -fasynchronous-unwind-tables @gol
650 -finhibit-size-directive -finstrument-functions @gol
651 -fno-common -fno-ident @gol
652 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
653 -freg-struct-return -fshared-data -fshort-enums @gol
654 -fshort-double -fshort-wchar @gol
655 -fverbose-asm -fpack-struct -fstack-check @gol
656 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
657 -fargument-alias -fargument-noalias @gol
658 -fargument-noalias-global -fleading-underscore @gol
659 -ftls-model=@var{model} @gol
660 -ftrapv -fwrapv -fbounds-check}
664 * Overall Options:: Controlling the kind of output:
665 an executable, object files, assembler files,
666 or preprocessed source.
667 * C Dialect Options:: Controlling the variant of C language compiled.
668 * C++ Dialect Options:: Variations on C++.
669 * Objective-C Dialect Options:: Variations on Objective-C.
670 * Language Independent Options:: Controlling how diagnostics should be
672 * Warning Options:: How picky should the compiler be?
673 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
674 * Optimize Options:: How much optimization?
675 * Preprocessor Options:: Controlling header files and macro definitions.
676 Also, getting dependency information for Make.
677 * Assembler Options:: Passing options to the assembler.
678 * Link Options:: Specifying libraries and so on.
679 * Directory Options:: Where to find header files and libraries.
680 Where to find the compiler executable files.
681 * Spec Files:: How to pass switches to sub-processes.
682 * Target Options:: Running a cross-compiler, or an old version of GCC.
685 @node Overall Options
686 @section Options Controlling the Kind of Output
688 Compilation can involve up to four stages: preprocessing, compilation
689 proper, assembly and linking, always in that order. GCC is capable of
690 preprocessing and compiling several files either into several
691 assembler input files, or into one assembler input file; then each
692 assembler input file produces an object file, and linking combines all
693 the object files (those newly compiled, and those specified as input)
694 into an executable file.
696 @cindex file name suffix
697 For any given input file, the file name suffix determines what kind of
702 C source code which must be preprocessed.
705 C source code which should not be preprocessed.
708 C++ source code which should not be preprocessed.
711 Objective-C source code. Note that you must link with the library
712 @file{libobjc.a} to make an Objective-C program work.
715 Objective-C source code which should not be preprocessed.
718 C or C++ header file to be turned into a precompiled header.
722 @itemx @var{file}.cxx
723 @itemx @var{file}.cpp
724 @itemx @var{file}.CPP
725 @itemx @var{file}.c++
727 C++ source code which must be preprocessed. Note that in @samp{.cxx},
728 the last two letters must both be literally @samp{x}. Likewise,
729 @samp{.C} refers to a literal capital C@.
733 C++ header file to be turned into a precompiled header.
736 @itemx @var{file}.for
737 @itemx @var{file}.FOR
738 Fortran source code which should not be preprocessed.
741 @itemx @var{file}.fpp
742 @itemx @var{file}.FPP
743 Fortran source code which must be preprocessed (with the traditional
747 Fortran source code which must be preprocessed with a RATFOR
748 preprocessor (not included with GCC)@.
750 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
751 Using and Porting GNU Fortran}, for more details of the handling of
754 @c FIXME: Descriptions of Java file types.
761 Ada source code file which contains a library unit declaration (a
762 declaration of a package, subprogram, or generic, or a generic
763 instantiation), or a library unit renaming declaration (a package,
764 generic, or subprogram renaming declaration). Such files are also
767 @itemx @var{file}.adb
768 Ada source code file containing a library unit body (a subprogram or
769 package body). Such files are also called @dfn{bodies}.
771 @c GCC also knows about some suffixes for languages not yet included:
780 Assembler code which must be preprocessed.
783 An object file to be fed straight into linking.
784 Any file name with no recognized suffix is treated this way.
788 You can specify the input language explicitly with the @option{-x} option:
791 @item -x @var{language}
792 Specify explicitly the @var{language} for the following input files
793 (rather than letting the compiler choose a default based on the file
794 name suffix). This option applies to all following input files until
795 the next @option{-x} option. Possible values for @var{language} are:
797 c c-header cpp-output
798 c++ c++-header c++-cpp-output
799 objective-c objective-c-header objc-cpp-output
800 assembler assembler-with-cpp
802 f77 f77-cpp-input ratfor
808 Turn off any specification of a language, so that subsequent files are
809 handled according to their file name suffixes (as they are if @option{-x}
810 has not been used at all).
812 @item -pass-exit-codes
813 @opindex pass-exit-codes
814 Normally the @command{gcc} program will exit with the code of 1 if any
815 phase of the compiler returns a non-success return code. If you specify
816 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
817 numerically highest error produced by any phase that returned an error
821 If you only want some of the stages of compilation, you can use
822 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
823 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
824 @command{gcc} is to stop. Note that some combinations (for example,
825 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
830 Compile or assemble the source files, but do not link. The linking
831 stage simply is not done. The ultimate output is in the form of an
832 object file for each source file.
834 By default, the object file name for a source file is made by replacing
835 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
837 Unrecognized input files, not requiring compilation or assembly, are
842 Stop after the stage of compilation proper; do not assemble. The output
843 is in the form of an assembler code file for each non-assembler input
846 By default, the assembler file name for a source file is made by
847 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
849 Input files that don't require compilation are ignored.
853 Stop after the preprocessing stage; do not run the compiler proper. The
854 output is in the form of preprocessed source code, which is sent to the
857 Input files which don't require preprocessing are ignored.
859 @cindex output file option
862 Place output in file @var{file}. This applies regardless to whatever
863 sort of output is being produced, whether it be an executable file,
864 an object file, an assembler file or preprocessed C code.
866 If you specify @option{-o} when compiling more than one input file, or
867 you are producing an executable file as output, all the source files
868 on the command line will be compiled at once.
870 If @option{-o} is not specified, the default is to put an executable file
871 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
872 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
873 all preprocessed C source on standard output.
877 Print (on standard error output) the commands executed to run the stages
878 of compilation. Also print the version number of the compiler driver
879 program and of the preprocessor and the compiler proper.
883 Like @option{-v} except the commands are not executed and all command
884 arguments are quoted. This is useful for shell scripts to capture the
885 driver-generated command lines.
889 Use pipes rather than temporary files for communication between the
890 various stages of compilation. This fails to work on some systems where
891 the assembler is unable to read from a pipe; but the GNU assembler has
896 Print (on the standard output) a description of the command line options
897 understood by @command{gcc}. If the @option{-v} option is also specified
898 then @option{--help} will also be passed on to the various processes
899 invoked by @command{gcc}, so that they can display the command line options
900 they accept. If the @option{-Wextra} option is also specified then command
901 line options which have no documentation associated with them will also
906 Print (on the standard output) a description of target specific command
907 line options for each tool.
911 Display the version number and copyrights of the invoked GCC.
915 @section Compiling C++ Programs
917 @cindex suffixes for C++ source
918 @cindex C++ source file suffixes
919 C++ source files conventionally use one of the suffixes @samp{.C},
920 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
921 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
922 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
923 files with these names and compiles them as C++ programs even if you
924 call the compiler the same way as for compiling C programs (usually
925 with the name @command{gcc}).
929 However, C++ programs often require class libraries as well as a
930 compiler that understands the C++ language---and under some
931 circumstances, you might want to compile programs or header files from
932 standard input, or otherwise without a suffix that flags them as C++
933 programs. You might also like to precompile a C header file with a
934 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
935 program that calls GCC with the default language set to C++, and
936 automatically specifies linking against the C++ library. On many
937 systems, @command{g++} is also installed with the name @command{c++}.
939 @cindex invoking @command{g++}
940 When you compile C++ programs, you may specify many of the same
941 command-line options that you use for compiling programs in any
942 language; or command-line options meaningful for C and related
943 languages; or options that are meaningful only for C++ programs.
944 @xref{C Dialect Options,,Options Controlling C Dialect}, for
945 explanations of options for languages related to C@.
946 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
947 explanations of options that are meaningful only for C++ programs.
949 @node C Dialect Options
950 @section Options Controlling C Dialect
951 @cindex dialect options
952 @cindex language dialect options
953 @cindex options, dialect
955 The following options control the dialect of C (or languages derived
956 from C, such as C++ and Objective-C) that the compiler accepts:
963 In C mode, support all ISO C90 programs. In C++ mode,
964 remove GNU extensions that conflict with ISO C++.
966 This turns off certain features of GCC that are incompatible with ISO
967 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
968 such as the @code{asm} and @code{typeof} keywords, and
969 predefined macros such as @code{unix} and @code{vax} that identify the
970 type of system you are using. It also enables the undesirable and
971 rarely used ISO trigraph feature. For the C compiler,
972 it disables recognition of C++ style @samp{//} comments as well as
973 the @code{inline} keyword.
975 The alternate keywords @code{__asm__}, @code{__extension__},
976 @code{__inline__} and @code{__typeof__} continue to work despite
977 @option{-ansi}. You would not want to use them in an ISO C program, of
978 course, but it is useful to put them in header files that might be included
979 in compilations done with @option{-ansi}. Alternate predefined macros
980 such as @code{__unix__} and @code{__vax__} are also available, with or
981 without @option{-ansi}.
983 The @option{-ansi} option does not cause non-ISO programs to be
984 rejected gratuitously. For that, @option{-pedantic} is required in
985 addition to @option{-ansi}. @xref{Warning Options}.
987 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
988 option is used. Some header files may notice this macro and refrain
989 from declaring certain functions or defining certain macros that the
990 ISO standard doesn't call for; this is to avoid interfering with any
991 programs that might use these names for other things.
993 Functions which would normally be built in but do not have semantics
994 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
995 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
996 built-in functions provided by GCC}, for details of the functions
1001 Determine the language standard. This option is currently only
1002 supported when compiling C or C++. A value for this option must be
1003 provided; possible values are
1008 ISO C90 (same as @option{-ansi}).
1010 @item iso9899:199409
1011 ISO C90 as modified in amendment 1.
1017 ISO C99. Note that this standard is not yet fully supported; see
1018 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1019 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1022 Default, ISO C90 plus GNU extensions (including some C99 features).
1026 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1027 this will become the default. The name @samp{gnu9x} is deprecated.
1030 The 1998 ISO C++ standard plus amendments.
1033 The same as @option{-std=c++98} plus GNU extensions. This is the
1034 default for C++ code.
1037 Even when this option is not specified, you can still use some of the
1038 features of newer standards in so far as they do not conflict with
1039 previous C standards. For example, you may use @code{__restrict__} even
1040 when @option{-std=c99} is not specified.
1042 The @option{-std} options specifying some version of ISO C have the same
1043 effects as @option{-ansi}, except that features that were not in ISO C90
1044 but are in the specified version (for example, @samp{//} comments and
1045 the @code{inline} keyword in ISO C99) are not disabled.
1047 @xref{Standards,,Language Standards Supported by GCC}, for details of
1048 these standard versions.
1050 @item -aux-info @var{filename}
1052 Output to the given filename prototyped declarations for all functions
1053 declared and/or defined in a translation unit, including those in header
1054 files. This option is silently ignored in any language other than C@.
1056 Besides declarations, the file indicates, in comments, the origin of
1057 each declaration (source file and line), whether the declaration was
1058 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1059 @samp{O} for old, respectively, in the first character after the line
1060 number and the colon), and whether it came from a declaration or a
1061 definition (@samp{C} or @samp{F}, respectively, in the following
1062 character). In the case of function definitions, a K&R-style list of
1063 arguments followed by their declarations is also provided, inside
1064 comments, after the declaration.
1068 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1069 keyword, so that code can use these words as identifiers. You can use
1070 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1071 instead. @option{-ansi} implies @option{-fno-asm}.
1073 In C++, this switch only affects the @code{typeof} keyword, since
1074 @code{asm} and @code{inline} are standard keywords. You may want to
1075 use the @option{-fno-gnu-keywords} flag instead, which has the same
1076 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1077 switch only affects the @code{asm} and @code{typeof} keywords, since
1078 @code{inline} is a standard keyword in ISO C99.
1081 @itemx -fno-builtin-@var{function}
1082 @opindex fno-builtin
1083 @cindex built-in functions
1084 Don't recognize built-in functions that do not begin with
1085 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1086 functions provided by GCC}, for details of the functions affected,
1087 including those which are not built-in functions when @option{-ansi} or
1088 @option{-std} options for strict ISO C conformance are used because they
1089 do not have an ISO standard meaning.
1091 GCC normally generates special code to handle certain built-in functions
1092 more efficiently; for instance, calls to @code{alloca} may become single
1093 instructions that adjust the stack directly, and calls to @code{memcpy}
1094 may become inline copy loops. The resulting code is often both smaller
1095 and faster, but since the function calls no longer appear as such, you
1096 cannot set a breakpoint on those calls, nor can you change the behavior
1097 of the functions by linking with a different library.
1099 With the @option{-fno-builtin-@var{function}} option
1100 only the built-in function @var{function} is
1101 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1102 function is named this is not built-in in this version of GCC, this
1103 option is ignored. There is no corresponding
1104 @option{-fbuiltin-@var{function}} option; if you wish to enable
1105 built-in functions selectively when using @option{-fno-builtin} or
1106 @option{-ffreestanding}, you may define macros such as:
1109 #define abs(n) __builtin_abs ((n))
1110 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1115 @cindex hosted environment
1117 Assert that compilation takes place in a hosted environment. This implies
1118 @option{-fbuiltin}. A hosted environment is one in which the
1119 entire standard library is available, and in which @code{main} has a return
1120 type of @code{int}. Examples are nearly everything except a kernel.
1121 This is equivalent to @option{-fno-freestanding}.
1123 @item -ffreestanding
1124 @opindex ffreestanding
1125 @cindex hosted environment
1127 Assert that compilation takes place in a freestanding environment. This
1128 implies @option{-fno-builtin}. A freestanding environment
1129 is one in which the standard library may not exist, and program startup may
1130 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1131 This is equivalent to @option{-fno-hosted}.
1133 @xref{Standards,,Language Standards Supported by GCC}, for details of
1134 freestanding and hosted environments.
1136 @item -fms-extensions
1137 @opindex fms-extensions
1138 Accept some non-standard constructs used in Microsoft header files.
1142 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1143 options for strict ISO C conformance) implies @option{-trigraphs}.
1145 @item -no-integrated-cpp
1146 @opindex no-integrated-cpp
1147 Performs a compilation in two passes: preprocessing and compiling. This
1148 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1149 @option{-B} option. The user supplied compilation step can then add in
1150 an additional preprocessing step after normal preprocessing but before
1151 compiling. The default is to use the integrated cpp (internal cpp)
1153 The semantics of this option will change if "cc1", "cc1plus", and
1154 "cc1obj" are merged.
1156 @cindex traditional C language
1157 @cindex C language, traditional
1159 @itemx -traditional-cpp
1160 @opindex traditional-cpp
1161 @opindex traditional
1162 Formerly, these options caused GCC to attempt to emulate a pre-standard
1163 C compiler. They are now only supported with the @option{-E} switch.
1164 The preprocessor continues to support a pre-standard mode. See the GNU
1165 CPP manual for details.
1167 @item -fcond-mismatch
1168 @opindex fcond-mismatch
1169 Allow conditional expressions with mismatched types in the second and
1170 third arguments. The value of such an expression is void. This option
1171 is not supported for C++.
1173 @item -funsigned-char
1174 @opindex funsigned-char
1175 Let the type @code{char} be unsigned, like @code{unsigned char}.
1177 Each kind of machine has a default for what @code{char} should
1178 be. It is either like @code{unsigned char} by default or like
1179 @code{signed char} by default.
1181 Ideally, a portable program should always use @code{signed char} or
1182 @code{unsigned char} when it depends on the signedness of an object.
1183 But many programs have been written to use plain @code{char} and
1184 expect it to be signed, or expect it to be unsigned, depending on the
1185 machines they were written for. This option, and its inverse, let you
1186 make such a program work with the opposite default.
1188 The type @code{char} is always a distinct type from each of
1189 @code{signed char} or @code{unsigned char}, even though its behavior
1190 is always just like one of those two.
1193 @opindex fsigned-char
1194 Let the type @code{char} be signed, like @code{signed char}.
1196 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1197 the negative form of @option{-funsigned-char}. Likewise, the option
1198 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1200 @item -fsigned-bitfields
1201 @itemx -funsigned-bitfields
1202 @itemx -fno-signed-bitfields
1203 @itemx -fno-unsigned-bitfields
1204 @opindex fsigned-bitfields
1205 @opindex funsigned-bitfields
1206 @opindex fno-signed-bitfields
1207 @opindex fno-unsigned-bitfields
1208 These options control whether a bit-field is signed or unsigned, when the
1209 declaration does not use either @code{signed} or @code{unsigned}. By
1210 default, such a bit-field is signed, because this is consistent: the
1211 basic integer types such as @code{int} are signed types.
1213 @item -fwritable-strings
1214 @opindex fwritable-strings
1215 Store string constants in the writable data segment and don't uniquize
1216 them. This is for compatibility with old programs which assume they can
1217 write into string constants.
1219 Writing into string constants is a very bad idea; ``constants'' should
1222 This option is deprecated.
1225 @node C++ Dialect Options
1226 @section Options Controlling C++ Dialect
1228 @cindex compiler options, C++
1229 @cindex C++ options, command line
1230 @cindex options, C++
1231 This section describes the command-line options that are only meaningful
1232 for C++ programs; but you can also use most of the GNU compiler options
1233 regardless of what language your program is in. For example, you
1234 might compile a file @code{firstClass.C} like this:
1237 g++ -g -frepo -O -c firstClass.C
1241 In this example, only @option{-frepo} is an option meant
1242 only for C++ programs; you can use the other options with any
1243 language supported by GCC@.
1245 Here is a list of options that are @emph{only} for compiling C++ programs:
1249 @item -fabi-version=@var{n}
1250 @opindex fabi-version
1251 Use version @var{n} of the C++ ABI. Version 2 is the version of the
1252 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1253 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1254 the version that conforms most closely to the C++ ABI specification.
1255 Therefore, the ABI obtained using version 0 will change as ABI bugs
1258 The default is version 1.
1260 @item -fno-access-control
1261 @opindex fno-access-control
1262 Turn off all access checking. This switch is mainly useful for working
1263 around bugs in the access control code.
1267 Check that the pointer returned by @code{operator new} is non-null
1268 before attempting to modify the storage allocated. This check is
1269 normally unnecessary because the C++ standard specifies that
1270 @code{operator new} will only return @code{0} if it is declared
1271 @samp{throw()}, in which case the compiler will always check the
1272 return value even without this option. In all other cases, when
1273 @code{operator new} has a non-empty exception specification, memory
1274 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1275 @samp{new (nothrow)}.
1277 @item -fconserve-space
1278 @opindex fconserve-space
1279 Put uninitialized or runtime-initialized global variables into the
1280 common segment, as C does. This saves space in the executable at the
1281 cost of not diagnosing duplicate definitions. If you compile with this
1282 flag and your program mysteriously crashes after @code{main()} has
1283 completed, you may have an object that is being destroyed twice because
1284 two definitions were merged.
1286 This option is no longer useful on most targets, now that support has
1287 been added for putting variables into BSS without making them common.
1289 @item -fno-const-strings
1290 @opindex fno-const-strings
1291 Give string constants type @code{char *} instead of type @code{const
1292 char *}. By default, G++ uses type @code{const char *} as required by
1293 the standard. Even if you use @option{-fno-const-strings}, you cannot
1294 actually modify the value of a string constant, unless you also use
1295 @option{-fwritable-strings}.
1297 This option might be removed in a future release of G++. For maximum
1298 portability, you should structure your code so that it works with
1299 string constants that have type @code{const char *}.
1301 @item -fno-elide-constructors
1302 @opindex fno-elide-constructors
1303 The C++ standard allows an implementation to omit creating a temporary
1304 which is only used to initialize another object of the same type.
1305 Specifying this option disables that optimization, and forces G++ to
1306 call the copy constructor in all cases.
1308 @item -fno-enforce-eh-specs
1309 @opindex fno-enforce-eh-specs
1310 Don't check for violation of exception specifications at runtime. This
1311 option violates the C++ standard, but may be useful for reducing code
1312 size in production builds, much like defining @samp{NDEBUG}. The compiler
1313 will still optimize based on the exception specifications.
1316 @itemx -fno-for-scope
1318 @opindex fno-for-scope
1319 If @option{-ffor-scope} is specified, the scope of variables declared in
1320 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1321 as specified by the C++ standard.
1322 If @option{-fno-for-scope} is specified, the scope of variables declared in
1323 a @i{for-init-statement} extends to the end of the enclosing scope,
1324 as was the case in old versions of G++, and other (traditional)
1325 implementations of C++.
1327 The default if neither flag is given to follow the standard,
1328 but to allow and give a warning for old-style code that would
1329 otherwise be invalid, or have different behavior.
1331 @item -fno-gnu-keywords
1332 @opindex fno-gnu-keywords
1333 Do not recognize @code{typeof} as a keyword, so that code can use this
1334 word as an identifier. You can use the keyword @code{__typeof__} instead.
1335 @option{-ansi} implies @option{-fno-gnu-keywords}.
1337 @item -fno-implicit-templates
1338 @opindex fno-implicit-templates
1339 Never emit code for non-inline templates which are instantiated
1340 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1341 @xref{Template Instantiation}, for more information.
1343 @item -fno-implicit-inline-templates
1344 @opindex fno-implicit-inline-templates
1345 Don't emit code for implicit instantiations of inline templates, either.
1346 The default is to handle inlines differently so that compiles with and
1347 without optimization will need the same set of explicit instantiations.
1349 @item -fno-implement-inlines
1350 @opindex fno-implement-inlines
1351 To save space, do not emit out-of-line copies of inline functions
1352 controlled by @samp{#pragma implementation}. This will cause linker
1353 errors if these functions are not inlined everywhere they are called.
1355 @item -fms-extensions
1356 @opindex fms-extensions
1357 Disable pedantic warnings about constructs used in MFC, such as implicit
1358 int and getting a pointer to member function via non-standard syntax.
1360 @item -fno-nonansi-builtins
1361 @opindex fno-nonansi-builtins
1362 Disable built-in declarations of functions that are not mandated by
1363 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1364 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1366 @item -fno-operator-names
1367 @opindex fno-operator-names
1368 Do not treat the operator name keywords @code{and}, @code{bitand},
1369 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1370 synonyms as keywords.
1372 @item -fno-optional-diags
1373 @opindex fno-optional-diags
1374 Disable diagnostics that the standard says a compiler does not need to
1375 issue. Currently, the only such diagnostic issued by G++ is the one for
1376 a name having multiple meanings within a class.
1379 @opindex fpermissive
1380 Downgrade some diagnostics about nonconformant code from errors to
1381 warnings. Thus, using @option{-fpermissive} will allow some
1382 nonconforming code to compile.
1386 Enable automatic template instantiation at link time. This option also
1387 implies @option{-fno-implicit-templates}. @xref{Template
1388 Instantiation}, for more information.
1392 Disable generation of information about every class with virtual
1393 functions for use by the C++ runtime type identification features
1394 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1395 of the language, you can save some space by using this flag. Note that
1396 exception handling uses the same information, but it will generate it as
1401 Emit statistics about front-end processing at the end of the compilation.
1402 This information is generally only useful to the G++ development team.
1404 @item -ftemplate-depth-@var{n}
1405 @opindex ftemplate-depth
1406 Set the maximum instantiation depth for template classes to @var{n}.
1407 A limit on the template instantiation depth is needed to detect
1408 endless recursions during template class instantiation. ANSI/ISO C++
1409 conforming programs must not rely on a maximum depth greater than 17.
1411 @item -fuse-cxa-atexit
1412 @opindex fuse-cxa-atexit
1413 Register destructors for objects with static storage duration with the
1414 @code{__cxa_atexit} function rather than the @code{atexit} function.
1415 This option is required for fully standards-compliant handling of static
1416 destructors, but will only work if your C library supports
1417 @code{__cxa_atexit}.
1421 Do not use weak symbol support, even if it is provided by the linker.
1422 By default, G++ will use weak symbols if they are available. This
1423 option exists only for testing, and should not be used by end-users;
1424 it will result in inferior code and has no benefits. This option may
1425 be removed in a future release of G++.
1429 Do not search for header files in the standard directories specific to
1430 C++, but do still search the other standard directories. (This option
1431 is used when building the C++ library.)
1434 In addition, these optimization, warning, and code generation options
1435 have meanings only for C++ programs:
1438 @item -fno-default-inline
1439 @opindex fno-default-inline
1440 Do not assume @samp{inline} for functions defined inside a class scope.
1441 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1442 functions will have linkage like inline functions; they just won't be
1445 @item -Wabi @r{(C++ only)}
1447 Warn when G++ generates code that is probably not compatible with the
1448 vendor-neutral C++ ABI. Although an effort has been made to warn about
1449 all such cases, there are probably some cases that are not warned about,
1450 even though G++ is generating incompatible code. There may also be
1451 cases where warnings are emitted even though the code that is generated
1454 You should rewrite your code to avoid these warnings if you are
1455 concerned about the fact that code generated by G++ may not be binary
1456 compatible with code generated by other compilers.
1458 The known incompatibilities at this point include:
1463 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1464 pack data into the same byte as a base class. For example:
1467 struct A @{ virtual void f(); int f1 : 1; @};
1468 struct B : public A @{ int f2 : 1; @};
1472 In this case, G++ will place @code{B::f2} into the same byte
1473 as@code{A::f1}; other compilers will not. You can avoid this problem
1474 by explicitly padding @code{A} so that its size is a multiple of the
1475 byte size on your platform; that will cause G++ and other compilers to
1476 layout @code{B} identically.
1479 Incorrect handling of tail-padding for virtual bases. G++ does not use
1480 tail padding when laying out virtual bases. For example:
1483 struct A @{ virtual void f(); char c1; @};
1484 struct B @{ B(); char c2; @};
1485 struct C : public A, public virtual B @{@};
1489 In this case, G++ will not place @code{B} into the tail-padding for
1490 @code{A}; other compilers will. You can avoid this problem by
1491 explicitly padding @code{A} so that its size is a multiple of its
1492 alignment (ignoring virtual base classes); that will cause G++ and other
1493 compilers to layout @code{C} identically.
1496 Incorrect handling of bit-fields with declared widths greater than that
1497 of their underlying types, when the bit-fields appear in a union. For
1501 union U @{ int i : 4096; @};
1505 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1506 union too small by the number of bits in an @code{int}.
1509 Empty classes can be placed at incorrect offsets. For example:
1519 struct C : public B, public A @{@};
1523 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1524 it should be placed at offset zero. G++ mistakenly believes that the
1525 @code{A} data member of @code{B} is already at offset zero.
1528 Names of template functions whose types involve @code{typename} or
1529 template template parameters can be mangled incorrectly.
1532 template <typename Q>
1533 void f(typename Q::X) @{@}
1535 template <template <typename> class Q>
1536 void f(typename Q<int>::X) @{@}
1540 Instantiations of these templates may be mangled incorrectly.
1544 @item -Wctor-dtor-privacy @r{(C++ only)}
1545 @opindex Wctor-dtor-privacy
1546 Warn when a class seems unusable because all the constructors or
1547 destructors in that class are private, and it has neither friends nor
1548 public static member functions.
1550 @item -Wnon-virtual-dtor @r{(C++ only)}
1551 @opindex Wnon-virtual-dtor
1552 Warn when a class appears to be polymorphic, thereby requiring a virtual
1553 destructor, yet it declares a non-virtual one.
1554 This warning is enabled by @option{-Wall}.
1556 @item -Wreorder @r{(C++ only)}
1558 @cindex reordering, warning
1559 @cindex warning for reordering of member initializers
1560 Warn when the order of member initializers given in the code does not
1561 match the order in which they must be executed. For instance:
1567 A(): j (0), i (1) @{ @}
1571 The compiler will rearrange the member initializers for @samp{i}
1572 and @samp{j} to match the declaration order of the members, emitting
1573 a warning to that effect. This warning is enabled by @option{-Wall}.
1576 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1579 @item -Weffc++ @r{(C++ only)}
1581 Warn about violations of the following style guidelines from Scott Meyers'
1582 @cite{Effective C++} book:
1586 Item 11: Define a copy constructor and an assignment operator for classes
1587 with dynamically allocated memory.
1590 Item 12: Prefer initialization to assignment in constructors.
1593 Item 14: Make destructors virtual in base classes.
1596 Item 15: Have @code{operator=} return a reference to @code{*this}.
1599 Item 23: Don't try to return a reference when you must return an object.
1603 Also warn about violations of the following style guidelines from
1604 Scott Meyers' @cite{More Effective C++} book:
1608 Item 6: Distinguish between prefix and postfix forms of increment and
1609 decrement operators.
1612 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1616 When selecting this option, be aware that the standard library
1617 headers do not obey all of these guidelines; use @samp{grep -v}
1618 to filter out those warnings.
1620 @item -Wno-deprecated @r{(C++ only)}
1621 @opindex Wno-deprecated
1622 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1624 @item -Wno-non-template-friend @r{(C++ only)}
1625 @opindex Wno-non-template-friend
1626 Disable warnings when non-templatized friend functions are declared
1627 within a template. Since the advent of explicit template specification
1628 support in G++, if the name of the friend is an unqualified-id (i.e.,
1629 @samp{friend foo(int)}), the C++ language specification demands that the
1630 friend declare or define an ordinary, nontemplate function. (Section
1631 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1632 could be interpreted as a particular specialization of a templatized
1633 function. Because this non-conforming behavior is no longer the default
1634 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1635 check existing code for potential trouble spots and is on by default.
1636 This new compiler behavior can be turned off with
1637 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1638 but disables the helpful warning.
1640 @item -Wold-style-cast @r{(C++ only)}
1641 @opindex Wold-style-cast
1642 Warn if an old-style (C-style) cast to a non-void type is used within
1643 a C++ program. The new-style casts (@samp{static_cast},
1644 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1645 unintended effects and much easier to search for.
1647 @item -Woverloaded-virtual @r{(C++ only)}
1648 @opindex Woverloaded-virtual
1649 @cindex overloaded virtual fn, warning
1650 @cindex warning for overloaded virtual fn
1651 Warn when a function declaration hides virtual functions from a
1652 base class. For example, in:
1659 struct B: public A @{
1664 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1672 will fail to compile.
1674 @item -Wno-pmf-conversions @r{(C++ only)}
1675 @opindex Wno-pmf-conversions
1676 Disable the diagnostic for converting a bound pointer to member function
1679 @item -Wsign-promo @r{(C++ only)}
1680 @opindex Wsign-promo
1681 Warn when overload resolution chooses a promotion from unsigned or
1682 enumeral type to a signed type, over a conversion to an unsigned type of
1683 the same size. Previous versions of G++ would try to preserve
1684 unsignedness, but the standard mandates the current behavior.
1686 @item -Wsynth @r{(C++ only)}
1688 @cindex warning for synthesized methods
1689 @cindex synthesized methods, warning
1690 Warn when G++'s synthesis behavior does not match that of cfront. For
1696 A& operator = (int);
1706 In this example, G++ will synthesize a default @samp{A& operator =
1707 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1710 @node Objective-C Dialect Options
1711 @section Options Controlling Objective-C Dialect
1713 @cindex compiler options, Objective-C
1714 @cindex Objective-C options, command line
1715 @cindex options, Objective-C
1716 (NOTE: This manual does not describe the Objective-C language itself. See
1717 @w{@uref{http://gcc.gnu.org/readings.html}} for references.)
1719 This section describes the command-line options that are only meaningful
1720 for Objective-C programs, but you can also use most of the GNU compiler
1721 options regardless of what language your program is in. For example,
1722 you might compile a file @code{some_class.m} like this:
1725 gcc -g -fgnu-runtime -O -c some_class.m
1729 In this example, @option{-fgnu-runtime} is an option meant only for
1730 Objective-C programs; you can use the other options with any language
1733 Here is a list of options that are @emph{only} for compiling Objective-C
1737 @item -fconstant-string-class=@var{class-name}
1738 @opindex fconstant-string-class
1739 Use @var{class-name} as the name of the class to instantiate for each
1740 literal string specified with the syntax @code{@@"@dots{}"}. The default
1741 class name is @code{NXConstantString} if the GNU runtime is being used, and
1742 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1743 @option{-fconstant-cfstrings} option, if also present, will override the
1744 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1745 to be laid out as constant CoreFoundation strings.
1748 @opindex fgnu-runtime
1749 Generate object code compatible with the standard GNU Objective-C
1750 runtime. This is the default for most types of systems.
1752 @item -fnext-runtime
1753 @opindex fnext-runtime
1754 Generate output compatible with the NeXT runtime. This is the default
1755 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1756 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1759 @item -fno-nil-receivers
1760 @opindex -fno-nil-receivers
1761 Assume that all Objective-C message dispatches (e.g.,
1762 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1763 is not @code{nil}. This allows for more efficient entry points in the runtime to be
1764 used. Currently, this option is only available in conjunction with
1765 the NeXT runtime on Mac OS X 10.3 and later.
1767 @item -fobjc-exceptions
1768 @opindex -fobjc-exceptions
1769 Enable syntactic support for structured exception handling in Objective-C,
1770 similar to what is offered by C++ and Java. Currently, this option is only
1771 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1779 @@catch (AnObjCClass *exc) @{
1786 @@catch (AnotherClass *exc) @{
1789 @@catch (id allOthers) @{
1799 The @code{@@throw} statement may appear anywhere in an Objective-C or
1800 Objective-C++ program; when used inside of a @code{@@catch} block, the
1801 @code{@@throw} may appear without an argument (as shown above), in which case
1802 the object caught by the @code{@@catch} will be rethrown.
1804 Note that only (pointers to) Objective-C objects may be thrown and
1805 caught using this scheme. When an object is thrown, it will be caught
1806 by the nearest @code{@@catch} clause capable of handling objects of that type,
1807 analogously to how @code{catch} blocks work in C++ and Java. A
1808 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1809 any and all Objective-C exceptions not caught by previous @code{@@catch}
1812 The @code{@@finally} clause, if present, will be executed upon exit from the
1813 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1814 regardless of whether any exceptions are thrown, caught or rethrown
1815 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1816 of the @code{finally} clause in Java.
1818 There are several caveats to using the new exception mechanism:
1822 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1823 idioms provided by the @code{NSException} class, the new
1824 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1825 systems, due to additional functionality needed in the (NeXT) Objective-C
1829 As mentioned above, the new exceptions do not support handling
1830 types other than Objective-C objects. Furthermore, when used from
1831 Objective-C++, the Objective-C exception model does not interoperate with C++
1832 exceptions at this time. This means you cannot @code{@@throw} an exception
1833 from Objective-C and @code{catch} it in C++, or vice versa
1834 (i.e., @code{throw @dots{} @@catch}).
1837 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1838 blocks for thread-safe execution:
1841 @@synchronized (ObjCClass *guard) @{
1846 Upon entering the @code{@@synchronized} block, a thread of execution shall
1847 first check whether a lock has been placed on the corresponding @code{guard}
1848 object by another thread. If it has, the current thread shall wait until
1849 the other thread relinquishes its lock. Once @code{guard} becomes available,
1850 the current thread will place its own lock on it, execute the code contained in
1851 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1852 making @code{guard} available to other threads).
1854 Unlike Java, Objective-C does not allow for entire methods to be marked
1855 @code{@@synchronized}. Note that throwing exceptions out of
1856 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1857 to be unlocked properly.
1859 @item -freplace-objc-classes
1860 @opindex -freplace-objc-classes
1861 Emit a special marker instructing @command{ld(1)} not to statically link in
1862 the resulting object file, and allow @command{dyld(1)} to load it in at
1863 run time instead. This is used in conjunction with the Fix-and-Continue
1864 debugging mode, where the object file in question may be recompiled and
1865 dynamically reloaded in the course of program execution, without the need
1866 to restart the program itself. Currently, Fix-and-Continue functionality
1867 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
1871 @opindex -fzero-link
1872 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1873 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1874 compile time) with static class references that get initialized at load time,
1875 which improves run-time performance. Specifying the @option{-fzero-link} flag
1876 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1877 to be retained. This is useful in Zero-Link debugging mode, since it allows
1878 for individual class implementations to be modified during program execution.
1882 Dump interface declarations for all classes seen in the source file to a
1883 file named @file{@var{sourcename}.decl}.
1886 @opindex Wno-protocol
1887 If a class is declared to implement a protocol, a warning is issued for
1888 every method in the protocol that is not implemented by the class. The
1889 default behavior is to issue a warning for every method not explicitly
1890 implemented in the class, even if a method implementation is inherited
1891 from the superclass. If you use the @code{-Wno-protocol} option, then
1892 methods inherited from the superclass are considered to be implemented,
1893 and no warning is issued for them.
1897 Warn if multiple methods of different types for the same selector are
1898 found during compilation. The check is performed on the list of methods
1899 in the final stage of compilation. Additionally, a check is performed
1900 for each selector appearing in a @code{@@selector(@dots{})}
1901 expression, and a corresponding method for that selector has been found
1902 during compilation. Because these checks scan the method table only at
1903 the end of compilation, these warnings are not produced if the final
1904 stage of compilation is not reached, for example because an error is
1905 found during compilation, or because the @code{-fsyntax-only} option is
1908 @item -Wundeclared-selector
1909 @opindex Wundeclared-selector
1910 Warn if a @code{@@selector(@dots{})} expression referring to an
1911 undeclared selector is found. A selector is considered undeclared if no
1912 method with that name has been declared before the
1913 @code{@@selector(@dots{})} expression, either explicitly in an
1914 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
1915 an @code{@@implementation} section. This option always performs its
1916 checks as soon as a @code{@@selector(@dots{})} expression is found,
1917 while @code{-Wselector} only performs its checks in the final stage of
1918 compilation. This also enforces the coding style convention
1919 that methods and selectors must be declared before being used.
1921 @item -print-objc-runtime-info
1922 @opindex -print-objc-runtime-info
1923 Generate C header describing the largest structure that is passed by
1928 @node Language Independent Options
1929 @section Options to Control Diagnostic Messages Formatting
1930 @cindex options to control diagnostics formatting
1931 @cindex diagnostic messages
1932 @cindex message formatting
1934 Traditionally, diagnostic messages have been formatted irrespective of
1935 the output device's aspect (e.g.@: its width, @dots{}). The options described
1936 below can be used to control the diagnostic messages formatting
1937 algorithm, e.g.@: how many characters per line, how often source location
1938 information should be reported. Right now, only the C++ front end can
1939 honor these options. However it is expected, in the near future, that
1940 the remaining front ends would be able to digest them correctly.
1943 @item -fmessage-length=@var{n}
1944 @opindex fmessage-length
1945 Try to format error messages so that they fit on lines of about @var{n}
1946 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1947 the front ends supported by GCC@. If @var{n} is zero, then no
1948 line-wrapping will be done; each error message will appear on a single
1951 @opindex fdiagnostics-show-location
1952 @item -fdiagnostics-show-location=once
1953 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1954 reporter to emit @emph{once} source location information; that is, in
1955 case the message is too long to fit on a single physical line and has to
1956 be wrapped, the source location won't be emitted (as prefix) again,
1957 over and over, in subsequent continuation lines. This is the default
1960 @item -fdiagnostics-show-location=every-line
1961 Only meaningful in line-wrapping mode. Instructs the diagnostic
1962 messages reporter to emit the same source location information (as
1963 prefix) for physical lines that result from the process of breaking
1964 a message which is too long to fit on a single line.
1968 @node Warning Options
1969 @section Options to Request or Suppress Warnings
1970 @cindex options to control warnings
1971 @cindex warning messages
1972 @cindex messages, warning
1973 @cindex suppressing warnings
1975 Warnings are diagnostic messages that report constructions which
1976 are not inherently erroneous but which are risky or suggest there
1977 may have been an error.
1979 You can request many specific warnings with options beginning @samp{-W},
1980 for example @option{-Wimplicit} to request warnings on implicit
1981 declarations. Each of these specific warning options also has a
1982 negative form beginning @samp{-Wno-} to turn off warnings;
1983 for example, @option{-Wno-implicit}. This manual lists only one of the
1984 two forms, whichever is not the default.
1986 The following options control the amount and kinds of warnings produced
1987 by GCC; for further, language-specific options also refer to
1988 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
1991 @cindex syntax checking
1993 @opindex fsyntax-only
1994 Check the code for syntax errors, but don't do anything beyond that.
1998 Issue all the warnings demanded by strict ISO C and ISO C++;
1999 reject all programs that use forbidden extensions, and some other
2000 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2001 version of the ISO C standard specified by any @option{-std} option used.
2003 Valid ISO C and ISO C++ programs should compile properly with or without
2004 this option (though a rare few will require @option{-ansi} or a
2005 @option{-std} option specifying the required version of ISO C)@. However,
2006 without this option, certain GNU extensions and traditional C and C++
2007 features are supported as well. With this option, they are rejected.
2009 @option{-pedantic} does not cause warning messages for use of the
2010 alternate keywords whose names begin and end with @samp{__}. Pedantic
2011 warnings are also disabled in the expression that follows
2012 @code{__extension__}. However, only system header files should use
2013 these escape routes; application programs should avoid them.
2014 @xref{Alternate Keywords}.
2016 Some users try to use @option{-pedantic} to check programs for strict ISO
2017 C conformance. They soon find that it does not do quite what they want:
2018 it finds some non-ISO practices, but not all---only those for which
2019 ISO C @emph{requires} a diagnostic, and some others for which
2020 diagnostics have been added.
2022 A feature to report any failure to conform to ISO C might be useful in
2023 some instances, but would require considerable additional work and would
2024 be quite different from @option{-pedantic}. We don't have plans to
2025 support such a feature in the near future.
2027 Where the standard specified with @option{-std} represents a GNU
2028 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2029 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2030 extended dialect is based. Warnings from @option{-pedantic} are given
2031 where they are required by the base standard. (It would not make sense
2032 for such warnings to be given only for features not in the specified GNU
2033 C dialect, since by definition the GNU dialects of C include all
2034 features the compiler supports with the given option, and there would be
2035 nothing to warn about.)
2037 @item -pedantic-errors
2038 @opindex pedantic-errors
2039 Like @option{-pedantic}, except that errors are produced rather than
2044 Inhibit all warning messages.
2048 Inhibit warning messages about the use of @samp{#import}.
2050 @item -Wchar-subscripts
2051 @opindex Wchar-subscripts
2052 Warn if an array subscript has type @code{char}. This is a common cause
2053 of error, as programmers often forget that this type is signed on some
2058 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2059 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2063 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2064 the arguments supplied have types appropriate to the format string
2065 specified, and that the conversions specified in the format string make
2066 sense. This includes standard functions, and others specified by format
2067 attributes (@pxref{Function Attributes}), in the @code{printf},
2068 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2069 not in the C standard) families.
2071 The formats are checked against the format features supported by GNU
2072 libc version 2.2. These include all ISO C90 and C99 features, as well
2073 as features from the Single Unix Specification and some BSD and GNU
2074 extensions. Other library implementations may not support all these
2075 features; GCC does not support warning about features that go beyond a
2076 particular library's limitations. However, if @option{-pedantic} is used
2077 with @option{-Wformat}, warnings will be given about format features not
2078 in the selected standard version (but not for @code{strfmon} formats,
2079 since those are not in any version of the C standard). @xref{C Dialect
2080 Options,,Options Controlling C Dialect}.
2082 Since @option{-Wformat} also checks for null format arguments for
2083 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2085 @option{-Wformat} is included in @option{-Wall}. For more control over some
2086 aspects of format checking, the options @option{-Wformat-y2k},
2087 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2088 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2089 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2092 @opindex Wformat-y2k
2093 If @option{-Wformat} is specified, also warn about @code{strftime}
2094 formats which may yield only a two-digit year.
2096 @item -Wno-format-extra-args
2097 @opindex Wno-format-extra-args
2098 If @option{-Wformat} is specified, do not warn about excess arguments to a
2099 @code{printf} or @code{scanf} format function. The C standard specifies
2100 that such arguments are ignored.
2102 Where the unused arguments lie between used arguments that are
2103 specified with @samp{$} operand number specifications, normally
2104 warnings are still given, since the implementation could not know what
2105 type to pass to @code{va_arg} to skip the unused arguments. However,
2106 in the case of @code{scanf} formats, this option will suppress the
2107 warning if the unused arguments are all pointers, since the Single
2108 Unix Specification says that such unused arguments are allowed.
2110 @item -Wno-format-zero-length
2111 @opindex Wno-format-zero-length
2112 If @option{-Wformat} is specified, do not warn about zero-length formats.
2113 The C standard specifies that zero-length formats are allowed.
2115 @item -Wformat-nonliteral
2116 @opindex Wformat-nonliteral
2117 If @option{-Wformat} is specified, also warn if the format string is not a
2118 string literal and so cannot be checked, unless the format function
2119 takes its format arguments as a @code{va_list}.
2121 @item -Wformat-security
2122 @opindex Wformat-security
2123 If @option{-Wformat} is specified, also warn about uses of format
2124 functions that represent possible security problems. At present, this
2125 warns about calls to @code{printf} and @code{scanf} functions where the
2126 format string is not a string literal and there are no format arguments,
2127 as in @code{printf (foo);}. This may be a security hole if the format
2128 string came from untrusted input and contains @samp{%n}. (This is
2129 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2130 in future warnings may be added to @option{-Wformat-security} that are not
2131 included in @option{-Wformat-nonliteral}.)
2135 Enable @option{-Wformat} plus format checks not included in
2136 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2137 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2141 Warn about passing a null pointer for arguments marked as
2142 requiring a non-null value by the @code{nonnull} function attribute.
2144 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2145 can be disabled with the @option{-Wno-nonnull} option.
2147 @item -Winit-self @r{(C, C++, and Objective-C only)}
2149 Warn about uninitialized variables which are initialized with themselves.
2150 Note this option can only be used with the @option{-Wuninitialized} option,
2151 which in turn only works with @option{-O1} and above.
2153 For example, GCC will warn about @code{i} being uninitialized in the
2154 following snippet only when @option{-Winit-self} has been specified:
2165 @item -Wimplicit-int
2166 @opindex Wimplicit-int
2167 Warn when a declaration does not specify a type.
2169 @item -Wimplicit-function-declaration
2170 @itemx -Werror-implicit-function-declaration
2171 @opindex Wimplicit-function-declaration
2172 @opindex Werror-implicit-function-declaration
2173 Give a warning (or error) whenever a function is used before being
2178 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2182 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2183 function with external linkage, returning int, taking either zero
2184 arguments, two, or three arguments of appropriate types.
2186 @item -Wmissing-braces
2187 @opindex Wmissing-braces
2188 Warn if an aggregate or union initializer is not fully bracketed. In
2189 the following example, the initializer for @samp{a} is not fully
2190 bracketed, but that for @samp{b} is fully bracketed.
2193 int a[2][2] = @{ 0, 1, 2, 3 @};
2194 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2198 @opindex Wparentheses
2199 Warn if parentheses are omitted in certain contexts, such
2200 as when there is an assignment in a context where a truth value
2201 is expected, or when operators are nested whose precedence people
2202 often get confused about.
2204 Also warn about constructions where there may be confusion to which
2205 @code{if} statement an @code{else} branch belongs. Here is an example of
2220 In C, every @code{else} branch belongs to the innermost possible @code{if}
2221 statement, which in this example is @code{if (b)}. This is often not
2222 what the programmer expected, as illustrated in the above example by
2223 indentation the programmer chose. When there is the potential for this
2224 confusion, GCC will issue a warning when this flag is specified.
2225 To eliminate the warning, add explicit braces around the innermost
2226 @code{if} statement so there is no way the @code{else} could belong to
2227 the enclosing @code{if}. The resulting code would look like this:
2243 @item -Wsequence-point
2244 @opindex Wsequence-point
2245 Warn about code that may have undefined semantics because of violations
2246 of sequence point rules in the C standard.
2248 The C standard defines the order in which expressions in a C program are
2249 evaluated in terms of @dfn{sequence points}, which represent a partial
2250 ordering between the execution of parts of the program: those executed
2251 before the sequence point, and those executed after it. These occur
2252 after the evaluation of a full expression (one which is not part of a
2253 larger expression), after the evaluation of the first operand of a
2254 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2255 function is called (but after the evaluation of its arguments and the
2256 expression denoting the called function), and in certain other places.
2257 Other than as expressed by the sequence point rules, the order of
2258 evaluation of subexpressions of an expression is not specified. All
2259 these rules describe only a partial order rather than a total order,
2260 since, for example, if two functions are called within one expression
2261 with no sequence point between them, the order in which the functions
2262 are called is not specified. However, the standards committee have
2263 ruled that function calls do not overlap.
2265 It is not specified when between sequence points modifications to the
2266 values of objects take effect. Programs whose behavior depends on this
2267 have undefined behavior; the C standard specifies that ``Between the
2268 previous and next sequence point an object shall have its stored value
2269 modified at most once by the evaluation of an expression. Furthermore,
2270 the prior value shall be read only to determine the value to be
2271 stored.''. If a program breaks these rules, the results on any
2272 particular implementation are entirely unpredictable.
2274 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2275 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2276 diagnosed by this option, and it may give an occasional false positive
2277 result, but in general it has been found fairly effective at detecting
2278 this sort of problem in programs.
2280 The present implementation of this option only works for C programs. A
2281 future implementation may also work for C++ programs.
2283 The C standard is worded confusingly, therefore there is some debate
2284 over the precise meaning of the sequence point rules in subtle cases.
2285 Links to discussions of the problem, including proposed formal
2286 definitions, may be found on our readings page, at
2287 @w{@uref{http://gcc.gnu.org/readings.html}}.
2290 @opindex Wreturn-type
2291 Warn whenever a function is defined with a return-type that defaults to
2292 @code{int}. Also warn about any @code{return} statement with no
2293 return-value in a function whose return-type is not @code{void}.
2295 For C++, a function without return type always produces a diagnostic
2296 message, even when @option{-Wno-return-type} is specified. The only
2297 exceptions are @samp{main} and functions defined in system headers.
2301 Warn whenever a @code{switch} statement has an index of enumeral type
2302 and lacks a @code{case} for one or more of the named codes of that
2303 enumeration. (The presence of a @code{default} label prevents this
2304 warning.) @code{case} labels outside the enumeration range also
2305 provoke warnings when this option is used.
2307 @item -Wswitch-default
2308 @opindex Wswitch-switch
2309 Warn whenever a @code{switch} statement does not have a @code{default}
2313 @opindex Wswitch-enum
2314 Warn whenever a @code{switch} statement has an index of enumeral type
2315 and lacks a @code{case} for one or more of the named codes of that
2316 enumeration. @code{case} labels outside the enumeration range also
2317 provoke warnings when this option is used.
2321 Warn if any trigraphs are encountered that might change the meaning of
2322 the program (trigraphs within comments are not warned about).
2324 @item -Wunused-function
2325 @opindex Wunused-function
2326 Warn whenever a static function is declared but not defined or a
2327 non\-inline static function is unused.
2329 @item -Wunused-label
2330 @opindex Wunused-label
2331 Warn whenever a label is declared but not used.
2333 To suppress this warning use the @samp{unused} attribute
2334 (@pxref{Variable Attributes}).
2336 @item -Wunused-parameter
2337 @opindex Wunused-parameter
2338 Warn whenever a function parameter is unused aside from its declaration.
2340 To suppress this warning use the @samp{unused} attribute
2341 (@pxref{Variable Attributes}).
2343 @item -Wunused-variable
2344 @opindex Wunused-variable
2345 Warn whenever a local variable or non-constant static variable is unused
2346 aside from its declaration
2348 To suppress this warning use the @samp{unused} attribute
2349 (@pxref{Variable Attributes}).
2351 @item -Wunused-value
2352 @opindex Wunused-value
2353 Warn whenever a statement computes a result that is explicitly not used.
2355 To suppress this warning cast the expression to @samp{void}.
2359 All the above @option{-Wunused} options combined.
2361 In order to get a warning about an unused function parameter, you must
2362 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2363 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2365 @item -Wuninitialized
2366 @opindex Wuninitialized
2367 Warn if an automatic variable is used without first being initialized or
2368 if a variable may be clobbered by a @code{setjmp} call.
2370 These warnings are possible only in optimizing compilation,
2371 because they require data flow information that is computed only
2372 when optimizing. If you don't specify @option{-O}, you simply won't
2375 If you want to warn about code which uses the uninitialized value of the
2376 variable in its own initializer, use the @option{-Winit-self} option.
2378 These warnings occur only for variables that are candidates for
2379 register allocation. Therefore, they do not occur for a variable that
2380 is declared @code{volatile}, or whose address is taken, or whose size
2381 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2382 structures, unions or arrays, even when they are in registers.
2384 Note that there may be no warning about a variable that is used only
2385 to compute a value that itself is never used, because such
2386 computations may be deleted by data flow analysis before the warnings
2389 These warnings are made optional because GCC is not smart
2390 enough to see all the reasons why the code might be correct
2391 despite appearing to have an error. Here is one example of how
2412 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2413 always initialized, but GCC doesn't know this. Here is
2414 another common case:
2419 if (change_y) save_y = y, y = new_y;
2421 if (change_y) y = save_y;
2426 This has no bug because @code{save_y} is used only if it is set.
2428 @cindex @code{longjmp} warnings
2429 This option also warns when a non-volatile automatic variable might be
2430 changed by a call to @code{longjmp}. These warnings as well are possible
2431 only in optimizing compilation.
2433 The compiler sees only the calls to @code{setjmp}. It cannot know
2434 where @code{longjmp} will be called; in fact, a signal handler could
2435 call it at any point in the code. As a result, you may get a warning
2436 even when there is in fact no problem because @code{longjmp} cannot
2437 in fact be called at the place which would cause a problem.
2439 Some spurious warnings can be avoided if you declare all the functions
2440 you use that never return as @code{noreturn}. @xref{Function
2443 @item -Wunknown-pragmas
2444 @opindex Wunknown-pragmas
2445 @cindex warning for unknown pragmas
2446 @cindex unknown pragmas, warning
2447 @cindex pragmas, warning of unknown
2448 Warn when a #pragma directive is encountered which is not understood by
2449 GCC@. If this command line option is used, warnings will even be issued
2450 for unknown pragmas in system header files. This is not the case if
2451 the warnings were only enabled by the @option{-Wall} command line option.
2453 @item -Wstrict-aliasing
2454 @opindex Wstrict-aliasing
2455 This option is only active when @option{-fstrict-aliasing} is active.
2456 It warns about code which might break the strict aliasing rules that the
2457 compiler is using for optimization. The warning does not catch all
2458 cases, but does attempt to catch the more common pitfalls. It is
2459 included in @option{-Wall}.
2463 All of the above @samp{-W} options combined. This enables all the
2464 warnings about constructions that some users consider questionable, and
2465 that are easy to avoid (or modify to prevent the warning), even in
2466 conjunction with macros. This also enables some language-specific
2467 warnings described in @ref{C++ Dialect Options} and
2468 @ref{Objective-C Dialect Options}.
2471 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2472 Some of them warn about constructions that users generally do not
2473 consider questionable, but which occasionally you might wish to check
2474 for; others warn about constructions that are necessary or hard to avoid
2475 in some cases, and there is no simple way to modify the code to suppress
2482 (This option used to be called @option{-W}. The older name is still
2483 supported, but the newer name is more descriptive.) Print extra warning
2484 messages for these events:
2488 A function can return either with or without a value. (Falling
2489 off the end of the function body is considered returning without
2490 a value.) For example, this function would evoke such a
2504 An expression-statement or the left-hand side of a comma expression
2505 contains no side effects.
2506 To suppress the warning, cast the unused expression to void.
2507 For example, an expression such as @samp{x[i,j]} will cause a warning,
2508 but @samp{x[(void)i,j]} will not.
2511 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2514 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2515 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2516 that of ordinary mathematical notation.
2519 Storage-class specifiers like @code{static} are not the first things in
2520 a declaration. According to the C Standard, this usage is obsolescent.
2523 The return type of a function has a type qualifier such as @code{const}.
2524 Such a type qualifier has no effect, since the value returned by a
2525 function is not an lvalue. (But don't warn about the GNU extension of
2526 @code{volatile void} return types. That extension will be warned about
2527 if @option{-pedantic} is specified.)
2530 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2534 A comparison between signed and unsigned values could produce an
2535 incorrect result when the signed value is converted to unsigned.
2536 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2539 An aggregate has an initializer which does not initialize all members.
2540 For example, the following code would cause such a warning, because
2541 @code{x.h} would be implicitly initialized to zero:
2544 struct s @{ int f, g, h; @};
2545 struct s x = @{ 3, 4 @};
2549 A function parameter is declared without a type specifier in K&R-style
2557 An empty body occurs in an @samp{if} or @samp{else} statement.
2560 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2561 @samp{>}, or @samp{>=}.
2564 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2567 Any of several floating-point events that often indicate errors, such as
2568 overflow, underflow, loss of precision, etc.
2570 @item @r{(C++ only)}
2571 An enumerator and a non-enumerator both appear in a conditional expression.
2573 @item @r{(C++ only)}
2574 A non-static reference or non-static @samp{const} member appears in a
2575 class without constructors.
2577 @item @r{(C++ only)}
2578 Ambiguous virtual bases.
2580 @item @r{(C++ only)}
2581 Subscripting an array which has been declared @samp{register}.
2583 @item @r{(C++ only)}
2584 Taking the address of a variable which has been declared @samp{register}.
2586 @item @r{(C++ only)}
2587 A base class is not initialized in a derived class' copy constructor.
2590 @item -Wno-div-by-zero
2591 @opindex Wno-div-by-zero
2592 @opindex Wdiv-by-zero
2593 Do not warn about compile-time integer division by zero. Floating point
2594 division by zero is not warned about, as it can be a legitimate way of
2595 obtaining infinities and NaNs.
2597 @item -Wsystem-headers
2598 @opindex Wsystem-headers
2599 @cindex warnings from system headers
2600 @cindex system headers, warnings from
2601 Print warning messages for constructs found in system header files.
2602 Warnings from system headers are normally suppressed, on the assumption
2603 that they usually do not indicate real problems and would only make the
2604 compiler output harder to read. Using this command line option tells
2605 GCC to emit warnings from system headers as if they occurred in user
2606 code. However, note that using @option{-Wall} in conjunction with this
2607 option will @emph{not} warn about unknown pragmas in system
2608 headers---for that, @option{-Wunknown-pragmas} must also be used.
2611 @opindex Wfloat-equal
2612 Warn if floating point values are used in equality comparisons.
2614 The idea behind this is that sometimes it is convenient (for the
2615 programmer) to consider floating-point values as approximations to
2616 infinitely precise real numbers. If you are doing this, then you need
2617 to compute (by analyzing the code, or in some other way) the maximum or
2618 likely maximum error that the computation introduces, and allow for it
2619 when performing comparisons (and when producing output, but that's a
2620 different problem). In particular, instead of testing for equality, you
2621 would check to see whether the two values have ranges that overlap; and
2622 this is done with the relational operators, so equality comparisons are
2625 @item -Wtraditional @r{(C only)}
2626 @opindex Wtraditional
2627 Warn about certain constructs that behave differently in traditional and
2628 ISO C@. Also warn about ISO C constructs that have no traditional C
2629 equivalent, and/or problematic constructs which should be avoided.
2633 Macro parameters that appear within string literals in the macro body.
2634 In traditional C macro replacement takes place within string literals,
2635 but does not in ISO C@.
2638 In traditional C, some preprocessor directives did not exist.
2639 Traditional preprocessors would only consider a line to be a directive
2640 if the @samp{#} appeared in column 1 on the line. Therefore
2641 @option{-Wtraditional} warns about directives that traditional C
2642 understands but would ignore because the @samp{#} does not appear as the
2643 first character on the line. It also suggests you hide directives like
2644 @samp{#pragma} not understood by traditional C by indenting them. Some
2645 traditional implementations would not recognize @samp{#elif}, so it
2646 suggests avoiding it altogether.
2649 A function-like macro that appears without arguments.
2652 The unary plus operator.
2655 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2656 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2657 constants.) Note, these suffixes appear in macros defined in the system
2658 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2659 Use of these macros in user code might normally lead to spurious
2660 warnings, however gcc's integrated preprocessor has enough context to
2661 avoid warning in these cases.
2664 A function declared external in one block and then used after the end of
2668 A @code{switch} statement has an operand of type @code{long}.
2671 A non-@code{static} function declaration follows a @code{static} one.
2672 This construct is not accepted by some traditional C compilers.
2675 The ISO type of an integer constant has a different width or
2676 signedness from its traditional type. This warning is only issued if
2677 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2678 typically represent bit patterns, are not warned about.
2681 Usage of ISO string concatenation is detected.
2684 Initialization of automatic aggregates.
2687 Identifier conflicts with labels. Traditional C lacks a separate
2688 namespace for labels.
2691 Initialization of unions. If the initializer is zero, the warning is
2692 omitted. This is done under the assumption that the zero initializer in
2693 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2694 initializer warnings and relies on default initialization to zero in the
2698 Conversions by prototypes between fixed/floating point values and vice
2699 versa. The absence of these prototypes when compiling with traditional
2700 C would cause serious problems. This is a subset of the possible
2701 conversion warnings, for the full set use @option{-Wconversion}.
2704 Use of ISO C style function definitions. This warning intentionally is
2705 @emph{not} issued for prototype declarations or variadic functions
2706 because these ISO C features will appear in your code when using
2707 libiberty's traditional C compatibility macros, @code{PARAMS} and
2708 @code{VPARAMS}. This warning is also bypassed for nested functions
2709 because that feature is already a gcc extension and thus not relevant to
2710 traditional C compatibility.
2713 @item -Wdeclaration-after-statement @r{(C only)}
2714 @opindex Wdeclaration-after-statement
2715 Warn when a declaration is found after a statement in a block. This
2716 construct, known from C++, was introduced with ISO C99 and is by default
2717 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2718 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2722 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2724 @item -Wendif-labels
2725 @opindex Wendif-labels
2726 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2730 Warn whenever a local variable shadows another local variable, parameter or
2731 global variable or whenever a built-in function is shadowed.
2733 @item -Wlarger-than-@var{len}
2734 @opindex Wlarger-than
2735 Warn whenever an object of larger than @var{len} bytes is defined.
2737 @item -Wpointer-arith
2738 @opindex Wpointer-arith
2739 Warn about anything that depends on the ``size of'' a function type or
2740 of @code{void}. GNU C assigns these types a size of 1, for
2741 convenience in calculations with @code{void *} pointers and pointers
2744 @item -Wbad-function-cast @r{(C only)}
2745 @opindex Wbad-function-cast
2746 Warn whenever a function call is cast to a non-matching type.
2747 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2751 Warn whenever a pointer is cast so as to remove a type qualifier from
2752 the target type. For example, warn if a @code{const char *} is cast
2753 to an ordinary @code{char *}.
2756 @opindex Wcast-align
2757 Warn whenever a pointer is cast such that the required alignment of the
2758 target is increased. For example, warn if a @code{char *} is cast to
2759 an @code{int *} on machines where integers can only be accessed at
2760 two- or four-byte boundaries.
2762 @item -Wwrite-strings
2763 @opindex Wwrite-strings
2764 When compiling C, give string constants the type @code{const
2765 char[@var{length}]} so that
2766 copying the address of one into a non-@code{const} @code{char *}
2767 pointer will get a warning; when compiling C++, warn about the
2768 deprecated conversion from string constants to @code{char *}.
2769 These warnings will help you find at
2770 compile time code that can try to write into a string constant, but
2771 only if you have been very careful about using @code{const} in
2772 declarations and prototypes. Otherwise, it will just be a nuisance;
2773 this is why we did not make @option{-Wall} request these warnings.
2776 @opindex Wconversion
2777 Warn if a prototype causes a type conversion that is different from what
2778 would happen to the same argument in the absence of a prototype. This
2779 includes conversions of fixed point to floating and vice versa, and
2780 conversions changing the width or signedness of a fixed point argument
2781 except when the same as the default promotion.
2783 Also, warn if a negative integer constant expression is implicitly
2784 converted to an unsigned type. For example, warn about the assignment
2785 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2786 casts like @code{(unsigned) -1}.
2788 @item -Wsign-compare
2789 @opindex Wsign-compare
2790 @cindex warning for comparison of signed and unsigned values
2791 @cindex comparison of signed and unsigned values, warning
2792 @cindex signed and unsigned values, comparison warning
2793 Warn when a comparison between signed and unsigned values could produce
2794 an incorrect result when the signed value is converted to unsigned.
2795 This warning is also enabled by @option{-Wextra}; to get the other warnings
2796 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2798 @item -Waggregate-return
2799 @opindex Waggregate-return
2800 Warn if any functions that return structures or unions are defined or
2801 called. (In languages where you can return an array, this also elicits
2804 @item -Wstrict-prototypes @r{(C only)}
2805 @opindex Wstrict-prototypes
2806 Warn if a function is declared or defined without specifying the
2807 argument types. (An old-style function definition is permitted without
2808 a warning if preceded by a declaration which specifies the argument
2811 @item -Wold-style-definition @r{(C only)}
2812 @opindex Wold-style-definition
2813 Warn if an old-style function definition is used. A warning is given
2814 even if there is a previous prototype.
2816 @item -Wmissing-prototypes @r{(C only)}
2817 @opindex Wmissing-prototypes
2818 Warn if a global function is defined without a previous prototype
2819 declaration. This warning is issued even if the definition itself
2820 provides a prototype. The aim is to detect global functions that fail
2821 to be declared in header files.
2823 @item -Wmissing-declarations @r{(C only)}
2824 @opindex Wmissing-declarations
2825 Warn if a global function is defined without a previous declaration.
2826 Do so even if the definition itself provides a prototype.
2827 Use this option to detect global functions that are not declared in
2830 @item -Wmissing-noreturn
2831 @opindex Wmissing-noreturn
2832 Warn about functions which might be candidates for attribute @code{noreturn}.
2833 Note these are only possible candidates, not absolute ones. Care should
2834 be taken to manually verify functions actually do not ever return before
2835 adding the @code{noreturn} attribute, otherwise subtle code generation
2836 bugs could be introduced. You will not get a warning for @code{main} in
2837 hosted C environments.
2839 @item -Wmissing-format-attribute
2840 @opindex Wmissing-format-attribute
2842 If @option{-Wformat} is enabled, also warn about functions which might be
2843 candidates for @code{format} attributes. Note these are only possible
2844 candidates, not absolute ones. GCC will guess that @code{format}
2845 attributes might be appropriate for any function that calls a function
2846 like @code{vprintf} or @code{vscanf}, but this might not always be the
2847 case, and some functions for which @code{format} attributes are
2848 appropriate may not be detected. This option has no effect unless
2849 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2851 @item -Wno-multichar
2852 @opindex Wno-multichar
2854 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2855 Usually they indicate a typo in the user's code, as they have
2856 implementation-defined values, and should not be used in portable code.
2858 @item -Wno-deprecated-declarations
2859 @opindex Wno-deprecated-declarations
2860 Do not warn about uses of functions, variables, and types marked as
2861 deprecated by using the @code{deprecated} attribute.
2862 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2863 @pxref{Type Attributes}.)
2867 Warn if a structure is given the packed attribute, but the packed
2868 attribute has no effect on the layout or size of the structure.
2869 Such structures may be mis-aligned for little benefit. For
2870 instance, in this code, the variable @code{f.x} in @code{struct bar}
2871 will be misaligned even though @code{struct bar} does not itself
2872 have the packed attribute:
2879 @} __attribute__((packed));
2889 Warn if padding is included in a structure, either to align an element
2890 of the structure or to align the whole structure. Sometimes when this
2891 happens it is possible to rearrange the fields of the structure to
2892 reduce the padding and so make the structure smaller.
2894 @item -Wredundant-decls
2895 @opindex Wredundant-decls
2896 Warn if anything is declared more than once in the same scope, even in
2897 cases where multiple declaration is valid and changes nothing.
2899 @item -Wnested-externs @r{(C only)}
2900 @opindex Wnested-externs
2901 Warn if an @code{extern} declaration is encountered within a function.
2903 @item -Wunreachable-code
2904 @opindex Wunreachable-code
2905 Warn if the compiler detects that code will never be executed.
2907 This option is intended to warn when the compiler detects that at
2908 least a whole line of source code will never be executed, because
2909 some condition is never satisfied or because it is after a
2910 procedure that never returns.
2912 It is possible for this option to produce a warning even though there
2913 are circumstances under which part of the affected line can be executed,
2914 so care should be taken when removing apparently-unreachable code.
2916 For instance, when a function is inlined, a warning may mean that the
2917 line is unreachable in only one inlined copy of the function.
2919 This option is not made part of @option{-Wall} because in a debugging
2920 version of a program there is often substantial code which checks
2921 correct functioning of the program and is, hopefully, unreachable
2922 because the program does work. Another common use of unreachable
2923 code is to provide behavior which is selectable at compile-time.
2927 Warn if a function can not be inlined and it was declared as inline.
2928 Even with this option, the compiler will not warn about failures to
2929 inline functions declared in system headers.
2931 The compiler uses a variety of heuristics to determine whether or not
2932 to inline a function. For example, the compiler takes into account
2933 the size of the function being inlined and the the amount of inlining
2934 that has already been done in the current function. Therefore,
2935 seemingly insignificant changes in the source program can cause the
2936 warnings produced by @option{-Winline} to appear or disappear.
2938 @item -Wno-invalid-offsetof @r{(C++ only)}
2939 @opindex Wno-invalid-offsetof
2940 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
2941 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
2942 to a non-POD type is undefined. In existing C++ implementations,
2943 however, @samp{offsetof} typically gives meaningful results even when
2944 applied to certain kinds of non-POD types. (Such as a simple
2945 @samp{struct} that fails to be a POD type only by virtue of having a
2946 constructor.) This flag is for users who are aware that they are
2947 writing nonportable code and who have deliberately chosen to ignore the
2950 The restrictions on @samp{offsetof} may be relaxed in a future version
2951 of the C++ standard.
2954 @opindex Winvalid-pch
2955 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
2956 the search path but can't be used.
2960 @opindex Wno-long-long
2961 Warn if @samp{long long} type is used. This is default. To inhibit
2962 the warning messages, use @option{-Wno-long-long}. Flags
2963 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2964 only when @option{-pedantic} flag is used.
2966 @item -Wdisabled-optimization
2967 @opindex Wdisabled-optimization
2968 Warn if a requested optimization pass is disabled. This warning does
2969 not generally indicate that there is anything wrong with your code; it
2970 merely indicates that GCC's optimizers were unable to handle the code
2971 effectively. Often, the problem is that your code is too big or too
2972 complex; GCC will refuse to optimize programs when the optimization
2973 itself is likely to take inordinate amounts of time.
2977 Make all warnings into errors.
2980 @node Debugging Options
2981 @section Options for Debugging Your Program or GCC
2982 @cindex options, debugging
2983 @cindex debugging information options
2985 GCC has various special options that are used for debugging
2986 either your program or GCC:
2991 Produce debugging information in the operating system's native format
2992 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2995 On most systems that use stabs format, @option{-g} enables use of extra
2996 debugging information that only GDB can use; this extra information
2997 makes debugging work better in GDB but will probably make other debuggers
2999 refuse to read the program. If you want to control for certain whether
3000 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3001 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3003 Unlike most other C compilers, GCC allows you to use @option{-g} with
3004 @option{-O}. The shortcuts taken by optimized code may occasionally
3005 produce surprising results: some variables you declared may not exist
3006 at all; flow of control may briefly move where you did not expect it;
3007 some statements may not be executed because they compute constant
3008 results or their values were already at hand; some statements may
3009 execute in different places because they were moved out of loops.
3011 Nevertheless it proves possible to debug optimized output. This makes
3012 it reasonable to use the optimizer for programs that might have bugs.
3014 The following options are useful when GCC is generated with the
3015 capability for more than one debugging format.
3019 Produce debugging information for use by GDB@. This means to use the
3020 most expressive format available (DWARF 2, stabs, or the native format
3021 if neither of those are supported), including GDB extensions if at all
3026 Produce debugging information in stabs format (if that is supported),
3027 without GDB extensions. This is the format used by DBX on most BSD
3028 systems. On MIPS, Alpha and System V Release 4 systems this option
3029 produces stabs debugging output which is not understood by DBX or SDB@.
3030 On System V Release 4 systems this option requires the GNU assembler.
3032 @item -feliminate-unused-debug-symbols
3033 @opindex feliminate-unused-debug-symbols
3034 Produce debugging information in stabs format (if that is supported),
3035 for only symbols that are actually used.
3039 Produce debugging information in stabs format (if that is supported),
3040 using GNU extensions understood only by the GNU debugger (GDB)@. The
3041 use of these extensions is likely to make other debuggers crash or
3042 refuse to read the program.
3046 Produce debugging information in COFF format (if that is supported).
3047 This is the format used by SDB on most System V systems prior to
3052 Produce debugging information in XCOFF format (if that is supported).
3053 This is the format used by the DBX debugger on IBM RS/6000 systems.
3057 Produce debugging information in XCOFF format (if that is supported),
3058 using GNU extensions understood only by the GNU debugger (GDB)@. The
3059 use of these extensions is likely to make other debuggers crash or
3060 refuse to read the program, and may cause assemblers other than the GNU
3061 assembler (GAS) to fail with an error.
3065 Produce debugging information in DWARF version 2 format (if that is
3066 supported). This is the format used by DBX on IRIX 6.
3070 Produce debugging information in VMS debug format (if that is
3071 supported). This is the format used by DEBUG on VMS systems.
3074 @itemx -ggdb@var{level}
3075 @itemx -gstabs@var{level}
3076 @itemx -gcoff@var{level}
3077 @itemx -gxcoff@var{level}
3078 @itemx -gvms@var{level}
3079 Request debugging information and also use @var{level} to specify how
3080 much information. The default level is 2.
3082 Level 1 produces minimal information, enough for making backtraces in
3083 parts of the program that you don't plan to debug. This includes
3084 descriptions of functions and external variables, but no information
3085 about local variables and no line numbers.
3087 Level 3 includes extra information, such as all the macro definitions
3088 present in the program. Some debuggers support macro expansion when
3089 you use @option{-g3}.
3091 Note that in order to avoid confusion between DWARF1 debug level 2,
3092 and DWARF2 @option{-gdwarf-2} does not accept a concatenated debug
3093 level. Instead use an additional @option{-g@var{level}} option to
3094 change the debug level for DWARF2.
3096 @item -feliminate-dwarf2-dups
3097 @opindex feliminate-dwarf2-dups
3098 Compress DWARF2 debugging information by eliminating duplicated
3099 information about each symbol. This option only makes sense when
3100 generating DWARF2 debugging information with @option{-gdwarf-2}.
3102 @cindex @command{prof}
3105 Generate extra code to write profile information suitable for the
3106 analysis program @command{prof}. You must use this option when compiling
3107 the source files you want data about, and you must also use it when
3110 @cindex @command{gprof}
3113 Generate extra code to write profile information suitable for the
3114 analysis program @command{gprof}. You must use this option when compiling
3115 the source files you want data about, and you must also use it when
3120 Makes the compiler print out each function name as it is compiled, and
3121 print some statistics about each pass when it finishes.
3124 @opindex ftime-report
3125 Makes the compiler print some statistics about the time consumed by each
3126 pass when it finishes.
3129 @opindex fmem-report
3130 Makes the compiler print some statistics about permanent memory
3131 allocation when it finishes.
3133 @item -fprofile-arcs
3134 @opindex fprofile-arcs
3135 Add code so that program flow @dfn{arcs} are instrumented. During
3136 execution the program records how many times each branch and call is
3137 executed and how many times it is taken or returns. When the compiled
3138 program exits it saves this data to a file called
3139 @file{@var{auxname}.gcda} for each source file. The data may be used for
3140 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3141 test coverage analysis (@option{-ftest-coverage}). Each object file's
3142 @var{auxname} is generated from the name of the output file, if
3143 explicitly specified and it is not the final executable, otherwise it is
3144 the basename of the source file. In both cases any suffix is removed
3145 (e.g. @file{foo.gcda} for input file @file{dir/foo.c}, or
3146 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3151 Compile the source files with @option{-fprofile-arcs} plus optimization
3152 and code generation options. For test coverage analysis, use the
3153 additional @option{-ftest-coverage} option. You do not need to profile
3154 every source file in a program.
3157 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3158 (the latter implies the former).
3161 Run the program on a representative workload to generate the arc profile
3162 information. This may be repeated any number of times. You can run
3163 concurrent instances of your program, and provided that the file system
3164 supports locking, the data files will be correctly updated. Also
3165 @code{fork} calls are detected and correctly handled (double counting
3169 For profile-directed optimizations, compile the source files again with
3170 the same optimization and code generation options plus
3171 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3172 Control Optimization}).
3175 For test coverage analysis, use @command{gcov} to produce human readable
3176 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3177 @command{gcov} documentation for further information.
3181 With @option{-fprofile-arcs}, for each function of your program GCC
3182 creates a program flow graph, then finds a spanning tree for the graph.
3183 Only arcs that are not on the spanning tree have to be instrumented: the
3184 compiler adds code to count the number of times that these arcs are
3185 executed. When an arc is the only exit or only entrance to a block, the
3186 instrumentation code can be added to the block; otherwise, a new basic
3187 block must be created to hold the instrumentation code.
3190 @item -ftest-coverage
3191 @opindex ftest-coverage
3192 Produce a notes file that the @command{gcov} code-coverage utility
3193 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3194 show program coverage. Each source file's note file is called
3195 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3196 above for a description of @var{auxname} and instructions on how to
3197 generate test coverage data. Coverage data will match the source files
3198 more closely, if you do not optimize.
3200 @item -d@var{letters}
3202 Says to make debugging dumps during compilation at times specified by
3203 @var{letters}. This is used for debugging the compiler. The file names
3204 for most of the dumps are made by appending a pass number and a word to
3205 the @var{dumpname}. @var{dumpname} is generated from the name of the
3206 output file, if explicitly specified and it is not an executable,
3207 otherwise it is the basename of the source file. In both cases any
3208 suffix is removed (e.g. @file{foo.01.rtl} or @file{foo.02.sibling}).
3209 Here are the possible letters for use in @var{letters}, and their
3215 Annotate the assembler output with miscellaneous debugging information.
3218 Dump after computing branch probabilities, to @file{@var{file}.12.bp}.
3221 Dump after block reordering, to @file{@var{file}.31.bbro}.
3224 Dump after instruction combination, to the file @file{@var{file}.20.combine}.
3227 Dump after the first if conversion, to the file @file{@var{file}.14.ce1}.
3228 Also dump after the second if conversion, to the file @file{@var{file}.21.ce2}.
3231 Dump after branch target load optimization, to to @file{@var{file}.32.btl}.
3232 Also dump after delayed branch scheduling, to @file{@var{file}.36.dbr}.
3235 Dump all macro definitions, at the end of preprocessing, in addition to
3239 Dump after the third if conversion, to @file{@var{file}.30.ce3}.
3242 Dump after control and data flow analysis, to @file{@var{file}.11.cfg}.
3243 Also dump after life analysis, to @file{@var{file}.19.life}.
3246 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.07.addressof}.
3249 Dump after global register allocation, to @file{@var{file}.25.greg}.
3252 Dump after GCSE, to @file{@var{file}.08.gcse}.
3253 Also dump after jump bypassing and control flow optimizations, to
3254 @file{@var{file}.10.bypass}.
3257 Dump after finalization of EH handling code, to @file{@var{file}.03.eh}.
3260 Dump after sibling call optimizations, to @file{@var{file}.02.sibling}.
3263 Dump after the first jump optimization, to @file{@var{file}.04.jump}.
3266 Dump after conversion from registers to stack, to @file{@var{file}.34.stack}.
3269 Dump after local register allocation, to @file{@var{file}.24.lreg}.
3272 Dump after loop optimization passes, to @file{@var{file}.09.loop} and
3273 @file{@var{file}.16.loop2}.
3276 Dump after performing the machine dependent reorganization pass, to
3277 @file{@var{file}.35.mach}.
3280 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3283 Dump after the register move pass, to @file{@var{file}.22.regmove}.
3286 Dump after post-reload optimizations, to @file{@var{file}.26.postreload}.
3289 Dump after RTL generation, to @file{@var{file}.01.rtl}.
3292 Dump after the second scheduling pass, to @file{@var{file}.33.sched2}.
3295 Dump after CSE (including the jump optimization that sometimes follows
3296 CSE), to @file{@var{file}.06.cse}.
3299 Dump after the first scheduling pass, to @file{@var{file}.23.sched}.
3302 Dump after the second CSE pass (including the jump optimization that
3303 sometimes follows CSE), to @file{@var{file}.18.cse2}.
3306 Dump after running tracer, to @file{@var{file}.15.tracer}.
3309 Dump after null pointer elimination pass to @file{@var{file}.05.null}.
3312 Dump callgraph and unit-at-a-time optimization @file{@var{file}.00.unit}.
3315 Dump after the value profile transformations, to @file{@var{file}.13.vpt}.
3318 Dump after the second flow pass, to @file{@var{file}.27.flow2}.
3321 Dump after the peephole pass, to @file{@var{file}.28.peephole2}.
3324 Dump after constructing the web, to @file{@var{file}.17.web}.
3327 Produce all the dumps listed above.
3330 Produce a core dump whenever an error occurs.
3333 Print statistics on memory usage, at the end of the run, to
3337 Annotate the assembler output with a comment indicating which
3338 pattern and alternative was used. The length of each instruction is
3342 Dump the RTL in the assembler output as a comment before each instruction.
3343 Also turns on @option{-dp} annotation.
3346 For each of the other indicated dump files (except for
3347 @file{@var{file}.01.rtl}), dump a representation of the control flow graph
3348 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3351 Just generate RTL for a function instead of compiling it. Usually used
3355 Dump debugging information during parsing, to standard error.
3358 @item -fdump-unnumbered
3359 @opindex fdump-unnumbered
3360 When doing debugging dumps (see @option{-d} option above), suppress instruction
3361 numbers and line number note output. This makes it more feasible to
3362 use diff on debugging dumps for compiler invocations with different
3363 options, in particular with and without @option{-g}.
3365 @item -fdump-translation-unit @r{(C and C++ only)}
3366 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3367 @opindex fdump-translation-unit
3368 Dump a representation of the tree structure for the entire translation
3369 unit to a file. The file name is made by appending @file{.tu} to the
3370 source file name. If the @samp{-@var{options}} form is used, @var{options}
3371 controls the details of the dump as described for the
3372 @option{-fdump-tree} options.
3374 @item -fdump-class-hierarchy @r{(C++ only)}
3375 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3376 @opindex fdump-class-hierarchy
3377 Dump a representation of each class's hierarchy and virtual function
3378 table layout to a file. The file name is made by appending @file{.class}
3379 to the source file name. If the @samp{-@var{options}} form is used,
3380 @var{options} controls the details of the dump as described for the
3381 @option{-fdump-tree} options.
3383 @item -fdump-tree-@var{switch} @r{(C++ only)}
3384 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3386 Control the dumping at various stages of processing the intermediate
3387 language tree to a file. The file name is generated by appending a switch
3388 specific suffix to the source file name. If the @samp{-@var{options}}
3389 form is used, @var{options} is a list of @samp{-} separated options that
3390 control the details of the dump. Not all options are applicable to all
3391 dumps, those which are not meaningful will be ignored. The following
3392 options are available
3396 Print the address of each node. Usually this is not meaningful as it
3397 changes according to the environment and source file. Its primary use
3398 is for tying up a dump file with a debug environment.
3400 Inhibit dumping of members of a scope or body of a function merely
3401 because that scope has been reached. Only dump such items when they
3402 are directly reachable by some other path.
3404 Turn on all options.
3407 The following tree dumps are possible:
3410 Dump before any tree based optimization, to @file{@var{file}.original}.
3412 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3414 Dump after function inlining, to @file{@var{file}.inlined}.
3417 @item -frandom-seed=@var{string}
3418 @opindex frandom-string
3419 This option provides a seed that GCC uses when it would otherwise use
3420 random numbers. It is used to generate certain symbol names
3421 that have to be different in every compiled file. It is also used to
3422 place unique stamps in coverage data files and the object files that
3423 produce them. You can use the @option{-frandom-seed} option to produce
3424 reproducibly identical object files.
3426 The @var{string} should be different for every file you compile.
3428 @item -fsched-verbose=@var{n}
3429 @opindex fsched-verbose
3430 On targets that use instruction scheduling, this option controls the
3431 amount of debugging output the scheduler prints. This information is
3432 written to standard error, unless @option{-dS} or @option{-dR} is
3433 specified, in which case it is output to the usual dump
3434 listing file, @file{.sched} or @file{.sched2} respectively. However
3435 for @var{n} greater than nine, the output is always printed to standard
3438 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3439 same information as @option{-dRS}. For @var{n} greater than one, it
3440 also output basic block probabilities, detailed ready list information
3441 and unit/insn info. For @var{n} greater than two, it includes RTL
3442 at abort point, control-flow and regions info. And for @var{n} over
3443 four, @option{-fsched-verbose} also includes dependence info.
3447 Store the usual ``temporary'' intermediate files permanently; place them
3448 in the current directory and name them based on the source file. Thus,
3449 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3450 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3451 preprocessed @file{foo.i} output file even though the compiler now
3452 normally uses an integrated preprocessor.
3456 Report the CPU time taken by each subprocess in the compilation
3457 sequence. For C source files, this is the compiler proper and assembler
3458 (plus the linker if linking is done). The output looks like this:
3465 The first number on each line is the ``user time,'' that is time spent
3466 executing the program itself. The second number is ``system time,''
3467 time spent executing operating system routines on behalf of the program.
3468 Both numbers are in seconds.
3470 @item -print-file-name=@var{library}
3471 @opindex print-file-name
3472 Print the full absolute name of the library file @var{library} that
3473 would be used when linking---and don't do anything else. With this
3474 option, GCC does not compile or link anything; it just prints the
3477 @item -print-multi-directory
3478 @opindex print-multi-directory
3479 Print the directory name corresponding to the multilib selected by any
3480 other switches present in the command line. This directory is supposed
3481 to exist in @env{GCC_EXEC_PREFIX}.
3483 @item -print-multi-lib
3484 @opindex print-multi-lib
3485 Print the mapping from multilib directory names to compiler switches
3486 that enable them. The directory name is separated from the switches by
3487 @samp{;}, and each switch starts with an @samp{@@} instead of the
3488 @samp{-}, without spaces between multiple switches. This is supposed to
3489 ease shell-processing.
3491 @item -print-prog-name=@var{program}
3492 @opindex print-prog-name
3493 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3495 @item -print-libgcc-file-name
3496 @opindex print-libgcc-file-name
3497 Same as @option{-print-file-name=libgcc.a}.
3499 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3500 but you do want to link with @file{libgcc.a}. You can do
3503 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3506 @item -print-search-dirs
3507 @opindex print-search-dirs
3508 Print the name of the configured installation directory and a list of
3509 program and library directories gcc will search---and don't do anything else.
3511 This is useful when gcc prints the error message
3512 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3513 To resolve this you either need to put @file{cpp0} and the other compiler
3514 components where gcc expects to find them, or you can set the environment
3515 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3516 Don't forget the trailing '/'.
3517 @xref{Environment Variables}.
3520 @opindex dumpmachine
3521 Print the compiler's target machine (for example,
3522 @samp{i686-pc-linux-gnu})---and don't do anything else.
3525 @opindex dumpversion
3526 Print the compiler version (for example, @samp{3.0})---and don't do
3531 Print the compiler's built-in specs---and don't do anything else. (This
3532 is used when GCC itself is being built.) @xref{Spec Files}.
3534 @item -feliminate-unused-debug-types
3535 @opindex feliminate-unused-debug-types
3536 Normally, when producing DWARF2 output, GCC will emit debugging
3537 information for all types declared in a compilation
3538 unit, regardless of whether or not they are actually used
3539 in that compilation unit. Sometimes this is useful, such as
3540 if, in the debugger, you want to cast a value to a type that is
3541 not actually used in your program (but is declared). More often,
3542 however, this results in a significant amount of wasted space.
3543 With this option, GCC will avoid producing debug symbol output
3544 for types that are nowhere used in the source file being compiled.
3547 @node Optimize Options
3548 @section Options That Control Optimization
3549 @cindex optimize options
3550 @cindex options, optimization
3552 These options control various sorts of optimizations.
3554 Without any optimization option, the compiler's goal is to reduce the
3555 cost of compilation and to make debugging produce the expected
3556 results. Statements are independent: if you stop the program with a
3557 breakpoint between statements, you can then assign a new value to any
3558 variable or change the program counter to any other statement in the
3559 function and get exactly the results you would expect from the source
3562 Turning on optimization flags makes the compiler attempt to improve
3563 the performance and/or code size at the expense of compilation time
3564 and possibly the ability to debug the program.
3566 The compiler performs optimization based on the knowledge it has of
3567 the program. Using the @option{-funit-at-a-time} flag will allow the
3568 compiler to consider information gained from later functions in the
3569 file when compiling a function. Compiling multiple files at once to a
3570 single output file (and using @option{-funit-at-a-time}) will allow
3571 the compiler to use information gained from all of the files when
3572 compiling each of them.
3574 Not all optimizations are controlled directly by a flag. Only
3575 optimizations that have a flag are listed.
3582 Optimize. Optimizing compilation takes somewhat more time, and a lot
3583 more memory for a large function.
3585 With @option{-O}, the compiler tries to reduce code size and execution
3586 time, without performing any optimizations that take a great deal of
3589 @option{-O} turns on the following optimization flags:
3590 @gccoptlist{-fdefer-pop @gol
3591 -fmerge-constants @gol
3593 -floop-optimize @gol
3594 -fif-conversion @gol
3595 -fif-conversion2 @gol
3596 -fdelayed-branch @gol
3597 -fguess-branch-probability @gol
3600 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3601 where doing so does not interfere with debugging.
3605 Optimize even more. GCC performs nearly all supported optimizations
3606 that do not involve a space-speed tradeoff. The compiler does not
3607 perform loop unrolling or function inlining when you specify @option{-O2}.
3608 As compared to @option{-O}, this option increases both compilation time
3609 and the performance of the generated code.
3611 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3612 also turns on the following optimization flags:
3613 @gccoptlist{-fforce-mem @gol
3614 -foptimize-sibling-calls @gol
3615 -fstrength-reduce @gol
3616 -fcse-follow-jumps -fcse-skip-blocks @gol
3617 -frerun-cse-after-loop -frerun-loop-opt @gol
3618 -fgcse -fgcse-lm -fgcse-sm -fgcse-las @gol
3619 -fdelete-null-pointer-checks @gol
3620 -fexpensive-optimizations @gol
3622 -fschedule-insns -fschedule-insns2 @gol
3623 -fsched-interblock -fsched-spec @gol
3626 -freorder-blocks -freorder-functions @gol
3627 -fstrict-aliasing @gol
3628 -funit-at-a-time @gol
3629 -falign-functions -falign-jumps @gol
3630 -falign-loops -falign-labels @gol
3633 Please note the warning under @option{-fgcse} about
3634 invoking @option{-O2} on programs that use computed gotos.
3638 Optimize yet more. @option{-O3} turns on all optimizations specified by
3639 @option{-O2} and also turns on the @option{-finline-functions},
3640 @option{-fweb} and @option{-frename-registers} options.
3644 Do not optimize. This is the default.
3648 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3649 do not typically increase code size. It also performs further
3650 optimizations designed to reduce code size.
3652 @option{-Os} disables the following optimization flags:
3653 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3654 -falign-labels -freorder-blocks -fprefetch-loop-arrays}
3656 If you use multiple @option{-O} options, with or without level numbers,
3657 the last such option is the one that is effective.
3660 Options of the form @option{-f@var{flag}} specify machine-independent
3661 flags. Most flags have both positive and negative forms; the negative
3662 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3663 below, only one of the forms is listed---the one you typically will
3664 use. You can figure out the other form by either removing @samp{no-}
3667 The following options control specific optimizations. They are either
3668 activated by @option{-O} options or are related to ones that are. You
3669 can use the following flags in the rare cases when ``fine-tuning'' of
3670 optimizations to be performed is desired.
3673 @item -fno-default-inline
3674 @opindex fno-default-inline
3675 Do not make member functions inline by default merely because they are
3676 defined inside the class scope (C++ only). Otherwise, when you specify
3677 @w{@option{-O}}, member functions defined inside class scope are compiled
3678 inline by default; i.e., you don't need to add @samp{inline} in front of
3679 the member function name.
3681 @item -fno-defer-pop
3682 @opindex fno-defer-pop
3683 Always pop the arguments to each function call as soon as that function
3684 returns. For machines which must pop arguments after a function call,
3685 the compiler normally lets arguments accumulate on the stack for several
3686 function calls and pops them all at once.
3688 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3692 Force memory operands to be copied into registers before doing
3693 arithmetic on them. This produces better code by making all memory
3694 references potential common subexpressions. When they are not common
3695 subexpressions, instruction combination should eliminate the separate
3698 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3701 @opindex fforce-addr
3702 Force memory address constants to be copied into registers before
3703 doing arithmetic on them. This may produce better code just as
3704 @option{-fforce-mem} may.
3706 @item -fomit-frame-pointer
3707 @opindex fomit-frame-pointer
3708 Don't keep the frame pointer in a register for functions that
3709 don't need one. This avoids the instructions to save, set up and
3710 restore frame pointers; it also makes an extra register available
3711 in many functions. @strong{It also makes debugging impossible on
3714 On some machines, such as the VAX, this flag has no effect, because
3715 the standard calling sequence automatically handles the frame pointer
3716 and nothing is saved by pretending it doesn't exist. The
3717 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3718 whether a target machine supports this flag. @xref{Registers,,Register
3719 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3721 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3723 @item -foptimize-sibling-calls
3724 @opindex foptimize-sibling-calls
3725 Optimize sibling and tail recursive calls.
3727 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3731 Don't pay attention to the @code{inline} keyword. Normally this option
3732 is used to keep the compiler from expanding any functions inline.
3733 Note that if you are not optimizing, no functions can be expanded inline.
3735 @item -finline-functions
3736 @opindex finline-functions
3737 Integrate all simple functions into their callers. The compiler
3738 heuristically decides which functions are simple enough to be worth
3739 integrating in this way.
3741 If all calls to a given function are integrated, and the function is
3742 declared @code{static}, then the function is normally not output as
3743 assembler code in its own right.
3745 Enabled at level @option{-O3}.
3747 @item -finline-limit=@var{n}
3748 @opindex finline-limit
3749 By default, gcc limits the size of functions that can be inlined. This flag
3750 allows the control of this limit for functions that are explicitly marked as
3751 inline (i.e., marked with the inline keyword or defined within the class
3752 definition in c++). @var{n} is the size of functions that can be inlined in
3753 number of pseudo instructions (not counting parameter handling). The default
3754 value of @var{n} is 600.
3755 Increasing this value can result in more inlined code at
3756 the cost of compilation time and memory consumption. Decreasing usually makes
3757 the compilation faster and less code will be inlined (which presumably
3758 means slower programs). This option is particularly useful for programs that
3759 use inlining heavily such as those based on recursive templates with C++.
3761 Inlining is actually controlled by a number of parameters, which may be
3762 specified individually by using @option{--param @var{name}=@var{value}}.
3763 The @option{-finline-limit=@var{n}} option sets some of these parameters
3767 @item max-inline-insns-single
3768 is set to @var{n}/2.
3769 @item max-inline-insns-auto
3770 is set to @var{n}/2.
3771 @item min-inline-insns
3772 is set to 130 or @var{n}/4, whichever is smaller.
3773 @item max-inline-insns-rtl
3777 See below for a documentation of the individual
3778 parameters controlling inlining.
3780 @emph{Note:} pseudo instruction represents, in this particular context, an
3781 abstract measurement of function's size. In no way, it represents a count
3782 of assembly instructions and as such its exact meaning might change from one
3783 release to an another.
3785 @item -fkeep-inline-functions
3786 @opindex fkeep-inline-functions
3787 Even if all calls to a given function are integrated, and the function
3788 is declared @code{static}, nevertheless output a separate run-time
3789 callable version of the function. This switch does not affect
3790 @code{extern inline} functions.
3792 @item -fkeep-static-consts
3793 @opindex fkeep-static-consts
3794 Emit variables declared @code{static const} when optimization isn't turned
3795 on, even if the variables aren't referenced.
3797 GCC enables this option by default. If you want to force the compiler to
3798 check if the variable was referenced, regardless of whether or not
3799 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3801 @item -fmerge-constants
3802 Attempt to merge identical constants (string constants and floating point
3803 constants) across compilation units.
3805 This option is the default for optimized compilation if the assembler and
3806 linker support it. Use @option{-fno-merge-constants} to inhibit this
3809 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3811 @item -fmerge-all-constants
3812 Attempt to merge identical constants and identical variables.
3814 This option implies @option{-fmerge-constants}. In addition to
3815 @option{-fmerge-constants} this considers e.g. even constant initialized
3816 arrays or initialized constant variables with integral or floating point
3817 types. Languages like C or C++ require each non-automatic variable to
3818 have distinct location, so using this option will result in non-conforming
3823 Use a graph coloring register allocator. Currently this option is meant
3824 for testing, so we are interested to hear about miscompilations with
3827 @item -fno-branch-count-reg
3828 @opindex fno-branch-count-reg
3829 Do not use ``decrement and branch'' instructions on a count register,
3830 but instead generate a sequence of instructions that decrement a
3831 register, compare it against zero, then branch based upon the result.
3832 This option is only meaningful on architectures that support such
3833 instructions, which include x86, PowerPC, IA-64 and S/390.
3835 The default is @option{-fbranch-count-reg}, enabled when
3836 @option{-fstrength-reduce} is enabled.
3838 @item -fno-function-cse
3839 @opindex fno-function-cse
3840 Do not put function addresses in registers; make each instruction that
3841 calls a constant function contain the function's address explicitly.
3843 This option results in less efficient code, but some strange hacks
3844 that alter the assembler output may be confused by the optimizations
3845 performed when this option is not used.
3847 The default is @option{-ffunction-cse}
3849 @item -fno-zero-initialized-in-bss
3850 @opindex fno-zero-initialized-in-bss
3851 If the target supports a BSS section, GCC by default puts variables that
3852 are initialized to zero into BSS@. This can save space in the resulting
3855 This option turns off this behavior because some programs explicitly
3856 rely on variables going to the data section. E.g., so that the
3857 resulting executable can find the beginning of that section and/or make
3858 assumptions based on that.
3860 The default is @option{-fzero-initialized-in-bss}.
3862 @item -fstrength-reduce
3863 @opindex fstrength-reduce
3864 Perform the optimizations of loop strength reduction and
3865 elimination of iteration variables.
3867 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3869 @item -fthread-jumps
3870 @opindex fthread-jumps
3871 Perform optimizations where we check to see if a jump branches to a
3872 location where another comparison subsumed by the first is found. If
3873 so, the first branch is redirected to either the destination of the
3874 second branch or a point immediately following it, depending on whether
3875 the condition is known to be true or false.
3877 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3879 @item -fcse-follow-jumps
3880 @opindex fcse-follow-jumps
3881 In common subexpression elimination, scan through jump instructions
3882 when the target of the jump is not reached by any other path. For
3883 example, when CSE encounters an @code{if} statement with an
3884 @code{else} clause, CSE will follow the jump when the condition
3887 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3889 @item -fcse-skip-blocks
3890 @opindex fcse-skip-blocks
3891 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3892 follow jumps which conditionally skip over blocks. When CSE
3893 encounters a simple @code{if} statement with no else clause,
3894 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3895 body of the @code{if}.
3897 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3899 @item -frerun-cse-after-loop
3900 @opindex frerun-cse-after-loop
3901 Re-run common subexpression elimination after loop optimizations has been
3904 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3906 @item -frerun-loop-opt
3907 @opindex frerun-loop-opt
3908 Run the loop optimizer twice.
3910 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3914 Perform a global common subexpression elimination pass.
3915 This pass also performs global constant and copy propagation.
3917 @emph{Note:} When compiling a program using computed gotos, a GCC
3918 extension, you may get better runtime performance if you disable
3919 the global common subexpression elimination pass by adding
3920 @option{-fno-gcse} to the command line.
3922 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3926 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3927 attempt to move loads which are only killed by stores into themselves. This
3928 allows a loop containing a load/store sequence to be changed to a load outside
3929 the loop, and a copy/store within the loop.
3931 Enabled by default when gcse is enabled.
3935 When @option{-fgcse-sm} is enabled, a store motion pass is run after
3936 global common subexpression elimination. This pass will attempt to move
3937 stores out of loops. When used in conjunction with @option{-fgcse-lm},
3938 loops containing a load/store sequence can be changed to a load before
3939 the loop and a store after the loop.
3941 Enabled by default when gcse is enabled.
3945 When @option{-fgcse-las} is enabled, the global common subexpression
3946 elimination pass eliminates redundant loads that come after stores to the
3947 same memory location (both partial and full redundancies).
3949 Enabled by default when gcse is enabled.
3951 @item -floop-optimize
3952 @opindex floop-optimize
3953 Perform loop optimizations: move constant expressions out of loops, simplify
3954 exit test conditions and optionally do strength-reduction and loop unrolling as
3957 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3959 @item -fcrossjumping
3960 @opindex crossjumping
3961 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
3962 resulting code may or may not perform better than without cross-jumping.
3964 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3966 @item -fif-conversion
3967 @opindex if-conversion
3968 Attempt to transform conditional jumps into branch-less equivalents. This
3969 include use of conditional moves, min, max, set flags and abs instructions, and
3970 some tricks doable by standard arithmetics. The use of conditional execution
3971 on chips where it is available is controlled by @code{if-conversion2}.
3973 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3975 @item -fif-conversion2
3976 @opindex if-conversion2
3977 Use conditional execution (where available) to transform conditional jumps into
3978 branch-less equivalents.
3980 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3982 @item -fdelete-null-pointer-checks
3983 @opindex fdelete-null-pointer-checks
3984 Use global dataflow analysis to identify and eliminate useless checks
3985 for null pointers. The compiler assumes that dereferencing a null
3986 pointer would have halted the program. If a pointer is checked after
3987 it has already been dereferenced, it cannot be null.
3989 In some environments, this assumption is not true, and programs can
3990 safely dereference null pointers. Use
3991 @option{-fno-delete-null-pointer-checks} to disable this optimization
3992 for programs which depend on that behavior.
3994 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3996 @item -fexpensive-optimizations
3997 @opindex fexpensive-optimizations
3998 Perform a number of minor optimizations that are relatively expensive.
4000 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4002 @item -foptimize-register-move
4004 @opindex foptimize-register-move
4006 Attempt to reassign register numbers in move instructions and as
4007 operands of other simple instructions in order to maximize the amount of
4008 register tying. This is especially helpful on machines with two-operand
4011 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4014 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4016 @item -fdelayed-branch
4017 @opindex fdelayed-branch
4018 If supported for the target machine, attempt to reorder instructions
4019 to exploit instruction slots available after delayed branch
4022 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4024 @item -fschedule-insns
4025 @opindex fschedule-insns
4026 If supported for the target machine, attempt to reorder instructions to
4027 eliminate execution stalls due to required data being unavailable. This
4028 helps machines that have slow floating point or memory load instructions
4029 by allowing other instructions to be issued until the result of the load
4030 or floating point instruction is required.
4032 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4034 @item -fschedule-insns2
4035 @opindex fschedule-insns2
4036 Similar to @option{-fschedule-insns}, but requests an additional pass of
4037 instruction scheduling after register allocation has been done. This is
4038 especially useful on machines with a relatively small number of
4039 registers and where memory load instructions take more than one cycle.
4041 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4043 @item -fno-sched-interblock
4044 @opindex fno-sched-interblock
4045 Don't schedule instructions across basic blocks. This is normally
4046 enabled by default when scheduling before register allocation, i.e.@:
4047 with @option{-fschedule-insns} or at @option{-O2} or higher.
4049 @item -fno-sched-spec
4050 @opindex fno-sched-spec
4051 Don't allow speculative motion of non-load instructions. This is normally
4052 enabled by default when scheduling before register allocation, i.e.@:
4053 with @option{-fschedule-insns} or at @option{-O2} or higher.
4055 @item -fsched-spec-load
4056 @opindex fsched-spec-load
4057 Allow speculative motion of some load instructions. This only makes
4058 sense when scheduling before register allocation, i.e.@: with
4059 @option{-fschedule-insns} or at @option{-O2} or higher.
4061 @item -fsched-spec-load-dangerous
4062 @opindex fsched-spec-load-dangerous
4063 Allow speculative motion of more load instructions. This only makes
4064 sense when scheduling before register allocation, i.e.@: with
4065 @option{-fschedule-insns} or at @option{-O2} or higher.
4067 @item -fsched-stalled-insns=@var{n}
4068 @opindex fsched-stalled-insns
4069 Define how many insns (if any) can be moved prematurely from the queue
4070 of stalled insns into the ready list, during the second scheduling pass.
4072 @item -fsched-stalled-insns-dep=@var{n}
4073 @opindex fsched-stalled-insns-dep
4074 Define how many insn groups (cycles) will be examined for a dependency
4075 on a stalled insn that is candidate for premature removal from the queue
4076 of stalled insns. Has an effect only during the second scheduling pass,
4077 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4079 @item -fsched2-use-superblocks
4080 @opindex fsched2-use-superblocks
4081 When scheduling after register allocation, do use superblock scheduling
4082 algorithm. Superblock scheduling allows motion across basic block boundaries
4083 resulting on faster schedules. This option is experimental, as not all machine
4084 descriptions used by GCC model the CPU closely enough to avoid unreliable
4085 results from the algorithm.
4087 This only makes sense when scheduling after register allocation, i.e.@: with
4088 @option{-fschedule-insns2} or at @option{-O2} or higher.
4090 @item -fsched2-use-traces
4091 @opindex fsched2-use-traces
4092 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4093 allocation and additionally perform code duplication in order to increase the
4094 size of superblocks using tracer pass. See @option{-ftracer} for details on
4097 This mode should produce faster but significantly longer programs. Also
4098 without @code{-fbranch-probabilities} the traces constructed may not match the
4099 reality and hurt the performance. This only makes
4100 sense when scheduling after register allocation, i.e.@: with
4101 @option{-fschedule-insns2} or at @option{-O2} or higher.
4103 @item -fcaller-saves
4104 @opindex fcaller-saves
4105 Enable values to be allocated in registers that will be clobbered by
4106 function calls, by emitting extra instructions to save and restore the
4107 registers around such calls. Such allocation is done only when it
4108 seems to result in better code than would otherwise be produced.
4110 This option is always enabled by default on certain machines, usually
4111 those which have no call-preserved registers to use instead.
4113 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4115 @item -fmove-all-movables
4116 @opindex fmove-all-movables
4117 Forces all invariant computations in loops to be moved
4120 @item -freduce-all-givs
4121 @opindex freduce-all-givs
4122 Forces all general-induction variables in loops to be
4125 @emph{Note:} When compiling programs written in Fortran,
4126 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4127 by default when you use the optimizer.
4129 These options may generate better or worse code; results are highly
4130 dependent on the structure of loops within the source code.
4132 These two options are intended to be removed someday, once
4133 they have helped determine the efficacy of various
4134 approaches to improving loop optimizations.
4136 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
4137 know how use of these options affects
4138 the performance of your production code.
4139 We're very interested in code that runs @emph{slower}
4140 when these options are @emph{enabled}.
4143 @itemx -fno-peephole2
4144 @opindex fno-peephole
4145 @opindex fno-peephole2
4146 Disable any machine-specific peephole optimizations. The difference
4147 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4148 are implemented in the compiler; some targets use one, some use the
4149 other, a few use both.
4151 @option{-fpeephole} is enabled by default.
4152 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4154 @item -fno-guess-branch-probability
4155 @opindex fno-guess-branch-probability
4156 Do not guess branch probabilities using a randomized model.
4158 Sometimes gcc will opt to use a randomized model to guess branch
4159 probabilities, when none are available from either profiling feedback
4160 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4161 different runs of the compiler on the same program may produce different
4164 In a hard real-time system, people don't want different runs of the
4165 compiler to produce code that has different behavior; minimizing
4166 non-determinism is of paramount import. This switch allows users to
4167 reduce non-determinism, possibly at the expense of inferior
4170 The default is @option{-fguess-branch-probability} at levels
4171 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4173 @item -freorder-blocks
4174 @opindex freorder-blocks
4175 Reorder basic blocks in the compiled function in order to reduce number of
4176 taken branches and improve code locality.
4178 Enabled at levels @option{-O2}, @option{-O3}.
4180 @item -freorder-functions
4181 @opindex freorder-functions
4182 Reorder basic blocks in the compiled function in order to reduce number of
4183 taken branches and improve code locality. This is implemented by using special
4184 subsections @code{text.hot} for most frequently executed functions and
4185 @code{text.unlikely} for unlikely executed functions. Reordering is done by
4186 the linker so object file format must support named sections and linker must
4187 place them in a reasonable way.
4189 Also profile feedback must be available in to make this option effective. See
4190 @option{-fprofile-arcs} for details.
4192 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4194 @item -fstrict-aliasing
4195 @opindex fstrict-aliasing
4196 Allows the compiler to assume the strictest aliasing rules applicable to
4197 the language being compiled. For C (and C++), this activates
4198 optimizations based on the type of expressions. In particular, an
4199 object of one type is assumed never to reside at the same address as an
4200 object of a different type, unless the types are almost the same. For
4201 example, an @code{unsigned int} can alias an @code{int}, but not a
4202 @code{void*} or a @code{double}. A character type may alias any other
4205 Pay special attention to code like this:
4218 The practice of reading from a different union member than the one most
4219 recently written to (called ``type-punning'') is common. Even with
4220 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4221 is accessed through the union type. So, the code above will work as
4222 expected. However, this code might not:
4233 Every language that wishes to perform language-specific alias analysis
4234 should define a function that computes, given an @code{tree}
4235 node, an alias set for the node. Nodes in different alias sets are not
4236 allowed to alias. For an example, see the C front-end function
4237 @code{c_get_alias_set}.
4239 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4241 @item -falign-functions
4242 @itemx -falign-functions=@var{n}
4243 @opindex falign-functions
4244 Align the start of functions to the next power-of-two greater than
4245 @var{n}, skipping up to @var{n} bytes. For instance,
4246 @option{-falign-functions=32} aligns functions to the next 32-byte
4247 boundary, but @option{-falign-functions=24} would align to the next
4248 32-byte boundary only if this can be done by skipping 23 bytes or less.
4250 @option{-fno-align-functions} and @option{-falign-functions=1} are
4251 equivalent and mean that functions will not be aligned.
4253 Some assemblers only support this flag when @var{n} is a power of two;
4254 in that case, it is rounded up.
4256 If @var{n} is not specified or is zero, use a machine-dependent default.
4258 Enabled at levels @option{-O2}, @option{-O3}.
4260 @item -falign-labels
4261 @itemx -falign-labels=@var{n}
4262 @opindex falign-labels
4263 Align all branch targets to a power-of-two boundary, skipping up to
4264 @var{n} bytes like @option{-falign-functions}. This option can easily
4265 make code slower, because it must insert dummy operations for when the
4266 branch target is reached in the usual flow of the code.
4268 @option{-fno-align-labels} and @option{-falign-labels=1} are
4269 equivalent and mean that labels will not be aligned.
4271 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4272 are greater than this value, then their values are used instead.
4274 If @var{n} is not specified or is zero, use a machine-dependent default
4275 which is very likely to be @samp{1}, meaning no alignment.
4277 Enabled at levels @option{-O2}, @option{-O3}.
4280 @itemx -falign-loops=@var{n}
4281 @opindex falign-loops
4282 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4283 like @option{-falign-functions}. The hope is that the loop will be
4284 executed many times, which will make up for any execution of the dummy
4287 @option{-fno-align-loops} and @option{-falign-loops=1} are
4288 equivalent and mean that loops will not be aligned.
4290 If @var{n} is not specified or is zero, use a machine-dependent default.
4292 Enabled at levels @option{-O2}, @option{-O3}.
4295 @itemx -falign-jumps=@var{n}
4296 @opindex falign-jumps
4297 Align branch targets to a power-of-two boundary, for branch targets
4298 where the targets can only be reached by jumping, skipping up to @var{n}
4299 bytes like @option{-falign-functions}. In this case, no dummy operations
4302 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4303 equivalent and mean that loops will not be aligned.
4305 If @var{n} is not specified or is zero, use a machine-dependent default.
4307 Enabled at levels @option{-O2}, @option{-O3}.
4309 @item -frename-registers
4310 @opindex frename-registers
4311 Attempt to avoid false dependencies in scheduled code by making use
4312 of registers left over after register allocation. This optimization
4313 will most benefit processors with lots of registers. It can, however,
4314 make debugging impossible, since variables will no longer stay in
4315 a ``home register''.
4319 Constructs webs as commonly used for register allocation purposes and assign
4320 each web individual pseudo register. This allows our register allocation pass
4321 to operate on pseudos directly, but also strengthens several other optimization
4322 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
4323 however, make debugging impossible, since variables will no longer stay in a
4326 Enabled at levels @option{-O3}.
4328 @item -fno-cprop-registers
4329 @opindex fno-cprop-registers
4330 After register allocation and post-register allocation instruction splitting,
4331 we perform a copy-propagation pass to try to reduce scheduling dependencies
4332 and occasionally eliminate the copy.
4334 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4336 @item -fprofile-generate
4337 @opindex fprofile-generate
4338 Enable options usually used for instrumenting application to produce profile usefull
4339 for later recompilation profile feedback based optimization.
4341 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}
4344 @opindex fprofile-use
4345 Enable profile feedback directed optimizations, and optimizations
4346 generally profitable only with profile feedback available.
4348 The following options are enabled: @code{-fbranch-probabilities},
4349 @code{-fvpt}, @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}.
4353 The following options control compiler behavior regarding floating
4354 point arithmetic. These options trade off between speed and
4355 correctness. All must be specifically enabled.
4359 @opindex ffloat-store
4360 Do not store floating point variables in registers, and inhibit other
4361 options that might change whether a floating point value is taken from a
4364 @cindex floating point precision
4365 This option prevents undesirable excess precision on machines such as
4366 the 68000 where the floating registers (of the 68881) keep more
4367 precision than a @code{double} is supposed to have. Similarly for the
4368 x86 architecture. For most programs, the excess precision does only
4369 good, but a few programs rely on the precise definition of IEEE floating
4370 point. Use @option{-ffloat-store} for such programs, after modifying
4371 them to store all pertinent intermediate computations into variables.
4375 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4376 @option{-fno-trapping-math}, @option{-ffinite-math-only},
4377 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
4379 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4381 This option should never be turned on by any @option{-O} option since
4382 it can result in incorrect output for programs which depend on
4383 an exact implementation of IEEE or ISO rules/specifications for
4386 @item -fno-math-errno
4387 @opindex fno-math-errno
4388 Do not set ERRNO after calling math functions that are executed
4389 with a single instruction, e.g., sqrt. A program that relies on
4390 IEEE exceptions for math error handling may want to use this flag
4391 for speed while maintaining IEEE arithmetic compatibility.
4393 This option should never be turned on by any @option{-O} option since
4394 it can result in incorrect output for programs which depend on
4395 an exact implementation of IEEE or ISO rules/specifications for
4398 The default is @option{-fmath-errno}.
4400 @item -funsafe-math-optimizations
4401 @opindex funsafe-math-optimizations
4402 Allow optimizations for floating-point arithmetic that (a) assume
4403 that arguments and results are valid and (b) may violate IEEE or
4404 ANSI standards. When used at link-time, it may include libraries
4405 or startup files that change the default FPU control word or other
4406 similar optimizations.
4408 This option should never be turned on by any @option{-O} option since
4409 it can result in incorrect output for programs which depend on
4410 an exact implementation of IEEE or ISO rules/specifications for
4413 The default is @option{-fno-unsafe-math-optimizations}.
4415 @item -ffinite-math-only
4416 @opindex ffinite-math-only
4417 Allow optimizations for floating-point arithmetic that assume
4418 that arguments and results are not NaNs or +-Infs.
4420 This option should never be turned on by any @option{-O} option since
4421 it can result in incorrect output for programs which depend on
4422 an exact implementation of IEEE or ISO rules/specifications.
4424 The default is @option{-fno-finite-math-only}.
4426 @item -fno-trapping-math
4427 @opindex fno-trapping-math
4428 Compile code assuming that floating-point operations cannot generate
4429 user-visible traps. These traps include division by zero, overflow,
4430 underflow, inexact result and invalid operation. This option implies
4431 @option{-fno-signaling-nans}. Setting this option may allow faster
4432 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4434 This option should never be turned on by any @option{-O} option since
4435 it can result in incorrect output for programs which depend on
4436 an exact implementation of IEEE or ISO rules/specifications for
4439 The default is @option{-ftrapping-math}.
4441 @item -frounding-math
4442 @opindex frounding-math
4443 Disable transformations and optimizations that assume default floating
4444 point rounding behavior. This is round-to-zero for all floating point
4445 to integer conversions, and round-to-nearest for all other arithmetic
4446 truncations. This option should be specified for programs that change
4447 the FP rounding mode dynamically, or that may be executed with a
4448 non-default rounding mode. This option disables constant folding of
4449 floating point expressions at compile-time (which may be affected by
4450 rounding mode) and arithmetic transformations that are unsafe in the
4451 presence of sign-dependent rounding modes.
4453 The default is @option{-fno-rounding-math}.
4455 This option is experimental and does not currently guarantee to
4456 disable all GCC optimizations that are affected by rounding mode.
4457 Future versions of gcc may provide finer control of this setting
4458 using C99's @code{FENV_ACCESS} pragma. This command line option
4459 will be used to specify the default state for @code{FENV_ACCESS}.
4461 @item -fsignaling-nans
4462 @opindex fsignaling-nans
4463 Compile code assuming that IEEE signaling NaNs may generate user-visible
4464 traps during floating-point operations. Setting this option disables
4465 optimizations that may change the number of exceptions visible with
4466 signaling NaNs. This option implies @option{-ftrapping-math}.
4468 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4471 The default is @option{-fno-signaling-nans}.
4473 This option is experimental and does not currently guarantee to
4474 disable all GCC optimizations that affect signaling NaN behavior.
4476 @item -fsingle-precision-constant
4477 @opindex fsingle-precision-constant
4478 Treat floating point constant as single precision constant instead of
4479 implicitly converting it to double precision constant.
4484 The following options control optimizations that may improve
4485 performance, but are not enabled by any @option{-O} options. This
4486 section includes experimental options that may produce broken code.
4489 @item -fbranch-probabilities
4490 @opindex fbranch-probabilities
4491 After running a program compiled with @option{-fprofile-arcs}
4492 (@pxref{Debugging Options,, Options for Debugging Your Program or
4493 @command{gcc}}), you can compile it a second time using
4494 @option{-fbranch-probabilities}, to improve optimizations based on
4495 the number of times each branch was taken. When the program
4496 compiled with @option{-fprofile-arcs} exits it saves arc execution
4497 counts to a file called @file{@var{sourcename}.gcda} for each source
4498 file The information in this data file is very dependent on the
4499 structure of the generated code, so you must use the same source code
4500 and the same optimization options for both compilations.
4502 With @option{-fbranch-probabilities}, GCC puts a
4503 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4504 These can be used to improve optimization. Currently, they are only
4505 used in one place: in @file{reorg.c}, instead of guessing which path a
4506 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4507 exactly determine which path is taken more often.
4509 @item -fprofile-values
4510 @opindex fprofile-values
4511 If combined with @option{-fprofile-arcs}, it adds code so that some
4512 data about values of expressions in the program is gathered.
4514 With @option{-fbranch-probabilities}, it reads back the data gathered
4515 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
4516 notes to instructions for their later usage in optimizations.
4520 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
4521 a code to gather information about values of expressions.
4523 With @option{-fbranch-probabilities}, it reads back the data gathered
4524 and actually performs the optimizations based on them.
4525 Currently the optimizations include specialization of division operation
4526 using the knowledge about the value of the denominator.
4530 Use a graph coloring register allocator. Currently this option is meant
4531 for testing, so we are interested to hear about miscompilations with
4536 Perform tail duplication to enlarge superblock size. This transformation
4537 simplifies the control flow of the function allowing other optimizations to do
4540 @item -funit-at-a-time
4541 @opindex funit-at-a-time
4542 Parse the whole compilation unit before starting to produce code.
4543 This allows some extra optimizations to take place but consumes more
4546 @item -funroll-loops
4547 @opindex funroll-loops
4548 Unroll loops whose number of iterations can be determined at compile time or
4549 upon entry to the loop. @option{-funroll-loops} implies
4550 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4551 (i.e. complete removal of loops with small constant number of iterations).
4552 This option makes code larger, and may or may not make it run faster.
4554 @item -funroll-all-loops
4555 @opindex funroll-all-loops
4556 Unroll all loops, even if their number of iterations is uncertain when
4557 the loop is entered. This usually makes programs run more slowly.
4558 @option{-funroll-all-loops} implies the same options as
4559 @option{-funroll-loops}.
4562 @opindex fpeel-loops
4563 Peels the loops for that there is enough information that they do not
4564 roll much (from profile feedback). It also turns on complete loop peeling
4565 (i.e. complete removal of loops with small constant number of iterations).
4567 @item -funswitch-loops
4568 @opindex funswitch-loops
4569 Move branches with loop invariant conditions out of the loop, with duplicates
4570 of the loop on both branches (modified according to result of the condition).
4572 @item -fold-unroll-loops
4573 @opindex fold-unroll-loops
4574 Unroll loops whose number of iterations can be determined at compile
4575 time or upon entry to the loop, using the old loop unroller whose loop
4576 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4577 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4578 option makes code larger, and may or may not make it run faster.
4580 @item -fold-unroll-all-loops
4581 @opindex fold-unroll-all-loops
4582 Unroll all loops, even if their number of iterations is uncertain when
4583 the loop is entered. This is done using the old loop unroller whose loop
4584 recognition is based on notes from frontend. This usually makes programs run more slowly.
4585 @option{-fold-unroll-all-loops} implies the same options as
4586 @option{-fold-unroll-loops}.
4588 @item -funswitch-loops
4589 @opindex funswitch-loops
4590 Move branches with loop invariant conditions out of the loop, with duplicates
4591 of the loop on both branches (modified according to result of the condition).
4593 @item -funswitch-loops
4594 @opindex funswitch-loops
4595 Move branches with loop invariant conditions out of the loop, with duplicates
4596 of the loop on both branches (modified according to result of the condition).
4598 @item -fprefetch-loop-arrays
4599 @opindex fprefetch-loop-arrays
4600 If supported by the target machine, generate instructions to prefetch
4601 memory to improve the performance of loops that access large arrays.
4603 Disabled at level @option{-Os}.
4605 @item -ffunction-sections
4606 @itemx -fdata-sections
4607 @opindex ffunction-sections
4608 @opindex fdata-sections
4609 Place each function or data item into its own section in the output
4610 file if the target supports arbitrary sections. The name of the
4611 function or the name of the data item determines the section's name
4614 Use these options on systems where the linker can perform optimizations
4615 to improve locality of reference in the instruction space. Most systems
4616 using the ELF object format and SPARC processors running Solaris 2 have
4617 linkers with such optimizations. AIX may have these optimizations in
4620 Only use these options when there are significant benefits from doing
4621 so. When you specify these options, the assembler and linker will
4622 create larger object and executable files and will also be slower.
4623 You will not be able to use @code{gprof} on all systems if you
4624 specify this option and you may have problems with debugging if
4625 you specify both this option and @option{-g}.
4627 @item -fbranch-target-load-optimize
4628 @opindex fbranch-target-load-optimize
4629 Perform branch target register load optimization before prologue / epilogue
4631 The use of target registers can typically be exposed only during reload,
4632 thus hoisting loads out of loops and doing inter-block scheduling needs
4633 a separate optimization pass.
4635 @item -fbranch-target-load-optimize2
4636 @opindex fbranch-target-load-optimize2
4637 Perform branch target register load optimization after prologue / epilogue
4640 @item -fbtr-bb-exclusive
4641 @opindex fbtr-bb-exclusive
4642 WHen performing branch target register load optimization, don't reuse
4643 branch target registers in within any basic block.
4645 @item --param @var{name}=@var{value}
4647 In some places, GCC uses various constants to control the amount of
4648 optimization that is done. For example, GCC will not inline functions
4649 that contain more that a certain number of instructions. You can
4650 control some of these constants on the command-line using the
4651 @option{--param} option.
4653 The names of specific parameters, and the meaning of the values, are
4654 tied to the internals of the compiler, and are subject to change
4655 without notice in future releases.
4657 In each case, the @var{value} is an integer. The allowable choices for
4658 @var{name} are given in the following table:
4661 @item max-crossjump-edges
4662 The maximum number of incoming edges to consider for crossjumping.
4663 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
4664 the number of edges incoming to each block. Increasing values mean
4665 more aggressive optimization, making the compile time increase with
4666 probably small improvement in executable size.
4668 @item max-delay-slot-insn-search
4669 The maximum number of instructions to consider when looking for an
4670 instruction to fill a delay slot. If more than this arbitrary number of
4671 instructions is searched, the time savings from filling the delay slot
4672 will be minimal so stop searching. Increasing values mean more
4673 aggressive optimization, making the compile time increase with probably
4674 small improvement in executable run time.
4676 @item max-delay-slot-live-search
4677 When trying to fill delay slots, the maximum number of instructions to
4678 consider when searching for a block with valid live register
4679 information. Increasing this arbitrarily chosen value means more
4680 aggressive optimization, increasing the compile time. This parameter
4681 should be removed when the delay slot code is rewritten to maintain the
4684 @item max-gcse-memory
4685 The approximate maximum amount of memory that will be allocated in
4686 order to perform the global common subexpression elimination
4687 optimization. If more memory than specified is required, the
4688 optimization will not be done.
4690 @item max-gcse-passes
4691 The maximum number of passes of GCSE to run.
4693 @item max-pending-list-length
4694 The maximum number of pending dependencies scheduling will allow
4695 before flushing the current state and starting over. Large functions
4696 with few branches or calls can create excessively large lists which
4697 needlessly consume memory and resources.
4699 @item max-inline-insns-single
4700 Several parameters control the tree inliner used in gcc.
4701 This number sets the maximum number of instructions (counted in gcc's
4702 internal representation) in a single function that the tree inliner
4703 will consider for inlining. This only affects functions declared
4704 inline and methods implemented in a class declaration (C++).
4705 The default value is 500.
4707 @item max-inline-insns-auto
4708 When you use @option{-finline-functions} (included in @option{-O3}),
4709 a lot of functions that would otherwise not be considered for inlining
4710 by the compiler will be investigated. To those functions, a different
4711 (more restrictive) limit compared to functions declared inline can
4713 The default value is 120.
4715 @item large-function-insns
4716 The limit specifying really large functions. For functions greater than this
4717 limit inlining is constrained by @option{--param large-function-growth}.
4718 This parameter is useful primarily to avoid extreme compilation time caused by non-linear
4719 algorithms used by the backend.
4720 This parameter is ignored when @option{-funit-at-a-time} is not used.
4721 The default value is 3000.
4723 @item large-function-growth
4724 Specifies maximal growth of large function caused by inlining in percents.
4725 This parameter is ignored when @option{-funit-at-a-time} is not used.
4726 The default value is 200.
4728 @item inline-unit-growth
4729 Specifies maximal overall growth of the compilation unit caused by inlining.
4730 This parameter is ignored when @option{-funit-at-a-time} is not used.
4731 The default value is 150.
4733 @item max-inline-insns-rtl
4734 For languages that use the RTL inliner (this happens at a later stage
4735 than tree inlining), you can set the maximum allowable size (counted
4736 in RTL instructions) for the RTL inliner with this parameter.
4737 The default value is 600.
4739 @item max-unrolled-insns
4740 The maximum number of instructions that a loop should have if that loop
4741 is unrolled, and if the loop is unrolled, it determines how many times
4742 the loop code is unrolled.
4744 @item max-average-unrolled-insns
4745 The maximum number of instructions biased by probabilities of their execution
4746 that a loop should have if that loop is unrolled, and if the loop is unrolled,
4747 it determines how many times the loop code is unrolled.
4749 @item max-unroll-times
4750 The maximum number of unrollings of a single loop.
4752 @item max-peeled-insns
4753 The maximum number of instructions that a loop should have if that loop
4754 is peeled, and if the loop is peeled, it determines how many times
4755 the loop code is peeled.
4757 @item max-peel-times
4758 The maximum number of peelings of a single loop.
4760 @item max-completely-peeled-insns
4761 The maximum number of insns of a completely peeled loop.
4763 @item max-completely-peel-times
4764 The maximum number of iterations of a loop to be suitable for complete peeling.
4766 @item max-unswitch-insns
4767 The maximum number of insns of an unswitched loop.
4769 @item max-unswitch-level
4770 The maximum number of branches unswitched in a single loop.
4772 @item hot-bb-count-fraction
4773 Select fraction of the maximal count of repetitions of basic block in program
4774 given basic block needs to have to be considered hot.
4776 @item hot-bb-frequency-fraction
4777 Select fraction of the maximal frequency of executions of basic block in
4778 function given basic block needs to have to be considered hot
4780 @item tracer-dynamic-coverage
4781 @itemx tracer-dynamic-coverage-feedback
4783 This value is used to limit superblock formation once the given percentage of
4784 executed instructions is covered. This limits unnecessary code size
4787 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4788 feedback is available. The real profiles (as opposed to statically estimated
4789 ones) are much less balanced allowing the threshold to be larger value.
4791 @item tracer-max-code-growth
4792 Stop tail duplication once code growth has reached given percentage. This is
4793 rather hokey argument, as most of the duplicates will be eliminated later in
4794 cross jumping, so it may be set to much higher values than is the desired code
4797 @item tracer-min-branch-ratio
4799 Stop reverse growth when the reverse probability of best edge is less than this
4800 threshold (in percent).
4802 @item tracer-min-branch-ratio
4803 @itemx tracer-min-branch-ratio-feedback
4805 Stop forward growth if the best edge do have probability lower than this
4808 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
4809 compilation for profile feedback and one for compilation without. The value
4810 for compilation with profile feedback needs to be more conservative (higher) in
4811 order to make tracer effective.
4813 @item max-cse-path-length
4815 Maximum number of basic blocks on path that cse considers.
4817 @item ggc-min-expand
4819 GCC uses a garbage collector to manage its own memory allocation. This
4820 parameter specifies the minimum percentage by which the garbage
4821 collector's heap should be allowed to expand between collections.
4822 Tuning this may improve compilation speed; it has no effect on code
4825 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
4826 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
4827 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
4828 GCC is not able to calculate RAM on a particular platform, the lower
4829 bound of 30% is used. Setting this parameter and
4830 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4831 every opportunity. This is extremely slow, but can be useful for
4834 @item ggc-min-heapsize
4836 Minimum size of the garbage collector's heap before it begins bothering
4837 to collect garbage. The first collection occurs after the heap expands
4838 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4839 tuning this may improve compilation speed, and has no effect on code
4842 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
4843 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
4844 available, the notion of "RAM" is the smallest of actual RAM,
4845 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
4846 RAM on a particular platform, the lower bound is used. Setting this
4847 parameter very large effectively disables garbage collection. Setting
4848 this parameter and @option{ggc-min-expand} to zero causes a full
4849 collection to occur at every opportunity.
4851 @item max-reload-search-insns
4852 The maximum number of instruction reload should look backward for equivalent
4853 register. Increasing values mean more aggressive optimization, making the
4854 compile time increase with probably slightly better performance. The default
4857 @item max-cselib-memory-location
4858 The maximum number of memory locations cselib should take into acount.
4859 Increasing values mean more aggressive optimization, making the compile time
4860 increase with probably slightly better performance. The default value is 500.
4862 @item reorder-blocks-duplicate
4863 @itemx reorder-blocks-duplicate-feedback
4865 Used by basic block reordering pass to decide whether to use unconditional
4866 branch or duplicate the code on its destination. Code is duplicated when its
4867 estimated size is smaller than this value multiplied by the estimated size of
4868 unconditional jump in the hot spots of the program.
4870 The @option{reorder-block-duplicate-feedback} is used only when profile
4871 feedback is available and may be set to higher values than
4872 @option{reorder-block-duplicate} since information about the hot spots is more
4877 @node Preprocessor Options
4878 @section Options Controlling the Preprocessor
4879 @cindex preprocessor options
4880 @cindex options, preprocessor
4882 These options control the C preprocessor, which is run on each C source
4883 file before actual compilation.
4885 If you use the @option{-E} option, nothing is done except preprocessing.
4886 Some of these options make sense only together with @option{-E} because
4887 they cause the preprocessor output to be unsuitable for actual
4892 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4893 and pass @var{option} directly through to the preprocessor. If
4894 @var{option} contains commas, it is split into multiple options at the
4895 commas. However, many options are modified, translated or interpreted
4896 by the compiler driver before being passed to the preprocessor, and
4897 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4898 interface is undocumented and subject to change, so whenever possible
4899 you should avoid using @option{-Wp} and let the driver handle the
4902 @item -Xpreprocessor @var{option}
4903 @opindex preprocessor
4904 Pass @var{option} as an option to the preprocessor. You can use this to
4905 supply system-specific preprocessor options which GCC does not know how to
4908 If you want to pass an option that takes an argument, you must use
4909 @option{-Xpreprocessor} twice, once for the option and once for the argument.
4912 @include cppopts.texi
4914 @node Assembler Options
4915 @section Passing Options to the Assembler
4917 @c prevent bad page break with this line
4918 You can pass options to the assembler.
4921 @item -Wa,@var{option}
4923 Pass @var{option} as an option to the assembler. If @var{option}
4924 contains commas, it is split into multiple options at the commas.
4926 @item -Xassembler @var{option}
4928 Pass @var{option} as an option to the assembler. You can use this to
4929 supply system-specific assembler options which GCC does not know how to
4932 If you want to pass an option that takes an argument, you must use
4933 @option{-Xassembler} twice, once for the option and once for the argument.
4938 @section Options for Linking
4939 @cindex link options
4940 @cindex options, linking
4942 These options come into play when the compiler links object files into
4943 an executable output file. They are meaningless if the compiler is
4944 not doing a link step.
4948 @item @var{object-file-name}
4949 A file name that does not end in a special recognized suffix is
4950 considered to name an object file or library. (Object files are
4951 distinguished from libraries by the linker according to the file
4952 contents.) If linking is done, these object files are used as input
4961 If any of these options is used, then the linker is not run, and
4962 object file names should not be used as arguments. @xref{Overall
4966 @item -l@var{library}
4967 @itemx -l @var{library}
4969 Search the library named @var{library} when linking. (The second
4970 alternative with the library as a separate argument is only for
4971 POSIX compliance and is not recommended.)
4973 It makes a difference where in the command you write this option; the
4974 linker searches and processes libraries and object files in the order they
4975 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4976 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4977 to functions in @samp{z}, those functions may not be loaded.
4979 The linker searches a standard list of directories for the library,
4980 which is actually a file named @file{lib@var{library}.a}. The linker
4981 then uses this file as if it had been specified precisely by name.
4983 The directories searched include several standard system directories
4984 plus any that you specify with @option{-L}.
4986 Normally the files found this way are library files---archive files
4987 whose members are object files. The linker handles an archive file by
4988 scanning through it for members which define symbols that have so far
4989 been referenced but not defined. But if the file that is found is an
4990 ordinary object file, it is linked in the usual fashion. The only
4991 difference between using an @option{-l} option and specifying a file name
4992 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4993 and searches several directories.
4997 You need this special case of the @option{-l} option in order to
4998 link an Objective-C program.
5001 @opindex nostartfiles
5002 Do not use the standard system startup files when linking.
5003 The standard system libraries are used normally, unless @option{-nostdlib}
5004 or @option{-nodefaultlibs} is used.
5006 @item -nodefaultlibs
5007 @opindex nodefaultlibs
5008 Do not use the standard system libraries when linking.
5009 Only the libraries you specify will be passed to the linker.
5010 The standard startup files are used normally, unless @option{-nostartfiles}
5011 is used. The compiler may generate calls to memcmp, memset, and memcpy
5012 for System V (and ISO C) environments or to bcopy and bzero for
5013 BSD environments. These entries are usually resolved by entries in
5014 libc. These entry points should be supplied through some other
5015 mechanism when this option is specified.
5019 Do not use the standard system startup files or libraries when linking.
5020 No startup files and only the libraries you specify will be passed to
5021 the linker. The compiler may generate calls to memcmp, memset, and memcpy
5022 for System V (and ISO C) environments or to bcopy and bzero for
5023 BSD environments. These entries are usually resolved by entries in
5024 libc. These entry points should be supplied through some other
5025 mechanism when this option is specified.
5027 @cindex @option{-lgcc}, use with @option{-nostdlib}
5028 @cindex @option{-nostdlib} and unresolved references
5029 @cindex unresolved references and @option{-nostdlib}
5030 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5031 @cindex @option{-nodefaultlibs} and unresolved references
5032 @cindex unresolved references and @option{-nodefaultlibs}
5033 One of the standard libraries bypassed by @option{-nostdlib} and
5034 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5035 that GCC uses to overcome shortcomings of particular machines, or special
5036 needs for some languages.
5037 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5038 Collection (GCC) Internals},
5039 for more discussion of @file{libgcc.a}.)
5040 In most cases, you need @file{libgcc.a} even when you want to avoid
5041 other standard libraries. In other words, when you specify @option{-nostdlib}
5042 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5043 This ensures that you have no unresolved references to internal GCC
5044 library subroutines. (For example, @samp{__main}, used to ensure C++
5045 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5046 GNU Compiler Collection (GCC) Internals}.)
5050 Produce a position independent executable on targets which support it.
5051 For predictable results, you must also specify the same set of options
5052 that were used to generate code (@option{-fpie}, @option{-fPIE},
5053 or model suboptions) when you specify this option.
5057 Remove all symbol table and relocation information from the executable.
5061 On systems that support dynamic linking, this prevents linking with the shared
5062 libraries. On other systems, this option has no effect.
5066 Produce a shared object which can then be linked with other objects to
5067 form an executable. Not all systems support this option. For predictable
5068 results, you must also specify the same set of options that were used to
5069 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5070 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5071 needs to build supplementary stub code for constructors to work. On
5072 multi-libbed systems, @samp{gcc -shared} must select the correct support
5073 libraries to link against. Failing to supply the correct flags may lead
5074 to subtle defects. Supplying them in cases where they are not necessary
5077 @item -shared-libgcc
5078 @itemx -static-libgcc
5079 @opindex shared-libgcc
5080 @opindex static-libgcc
5081 On systems that provide @file{libgcc} as a shared library, these options
5082 force the use of either the shared or static version respectively.
5083 If no shared version of @file{libgcc} was built when the compiler was
5084 configured, these options have no effect.
5086 There are several situations in which an application should use the
5087 shared @file{libgcc} instead of the static version. The most common
5088 of these is when the application wishes to throw and catch exceptions
5089 across different shared libraries. In that case, each of the libraries
5090 as well as the application itself should use the shared @file{libgcc}.
5092 Therefore, the G++ and GCJ drivers automatically add
5093 @option{-shared-libgcc} whenever you build a shared library or a main
5094 executable, because C++ and Java programs typically use exceptions, so
5095 this is the right thing to do.
5097 If, instead, you use the GCC driver to create shared libraries, you may
5098 find that they will not always be linked with the shared @file{libgcc}.
5099 If GCC finds, at its configuration time, that you have a GNU linker that
5100 does not support option @option{--eh-frame-hdr}, it will link the shared
5101 version of @file{libgcc} into shared libraries by default. Otherwise,
5102 it will take advantage of the linker and optimize away the linking with
5103 the shared version of @file{libgcc}, linking with the static version of
5104 libgcc by default. This allows exceptions to propagate through such
5105 shared libraries, without incurring relocation costs at library load
5108 However, if a library or main executable is supposed to throw or catch
5109 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5110 for the languages used in the program, or using the option
5111 @option{-shared-libgcc}, such that it is linked with the shared
5116 Bind references to global symbols when building a shared object. Warn
5117 about any unresolved references (unless overridden by the link editor
5118 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5121 @item -Xlinker @var{option}
5123 Pass @var{option} as an option to the linker. You can use this to
5124 supply system-specific linker options which GCC does not know how to
5127 If you want to pass an option that takes an argument, you must use
5128 @option{-Xlinker} twice, once for the option and once for the argument.
5129 For example, to pass @option{-assert definitions}, you must write
5130 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5131 @option{-Xlinker "-assert definitions"}, because this passes the entire
5132 string as a single argument, which is not what the linker expects.
5134 @item -Wl,@var{option}
5136 Pass @var{option} as an option to the linker. If @var{option} contains
5137 commas, it is split into multiple options at the commas.
5139 @item -u @var{symbol}
5141 Pretend the symbol @var{symbol} is undefined, to force linking of
5142 library modules to define it. You can use @option{-u} multiple times with
5143 different symbols to force loading of additional library modules.
5146 @node Directory Options
5147 @section Options for Directory Search
5148 @cindex directory options
5149 @cindex options, directory search
5152 These options specify directories to search for header files, for
5153 libraries and for parts of the compiler:
5158 Add the directory @var{dir} to the head of the list of directories to be
5159 searched for header files. This can be used to override a system header
5160 file, substituting your own version, since these directories are
5161 searched before the system header file directories. However, you should
5162 not use this option to add directories that contain vendor-supplied
5163 system header files (use @option{-isystem} for that). If you use more than
5164 one @option{-I} option, the directories are scanned in left-to-right
5165 order; the standard system directories come after.
5167 If a standard system include directory, or a directory specified with
5168 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5169 option will be ignored. The directory will still be searched but as a
5170 system directory at its normal position in the system include chain.
5171 This is to ensure that GCC's procedure to fix buggy system headers and
5172 the ordering for the include_next directive are not inadvertently changed.
5173 If you really need to change the search order for system directories,
5174 use the @option{-nostdinc} and/or @option{-isystem} options.
5178 Any directories you specify with @option{-I} options before the @option{-I-}
5179 option are searched only for the case of @samp{#include "@var{file}"};
5180 they are not searched for @samp{#include <@var{file}>}.
5182 If additional directories are specified with @option{-I} options after
5183 the @option{-I-}, these directories are searched for all @samp{#include}
5184 directives. (Ordinarily @emph{all} @option{-I} directories are used
5187 In addition, the @option{-I-} option inhibits the use of the current
5188 directory (where the current input file came from) as the first search
5189 directory for @samp{#include "@var{file}"}. There is no way to
5190 override this effect of @option{-I-}. With @option{-I.} you can specify
5191 searching the directory which was current when the compiler was
5192 invoked. That is not exactly the same as what the preprocessor does
5193 by default, but it is often satisfactory.
5195 @option{-I-} does not inhibit the use of the standard system directories
5196 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5201 Add directory @var{dir} to the list of directories to be searched
5204 @item -B@var{prefix}
5206 This option specifies where to find the executables, libraries,
5207 include files, and data files of the compiler itself.
5209 The compiler driver program runs one or more of the subprograms
5210 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5211 @var{prefix} as a prefix for each program it tries to run, both with and
5212 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5214 For each subprogram to be run, the compiler driver first tries the
5215 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5216 was not specified, the driver tries two standard prefixes, which are
5217 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5218 those results in a file name that is found, the unmodified program
5219 name is searched for using the directories specified in your
5220 @env{PATH} environment variable.
5222 The compiler will check to see if the path provided by the @option{-B}
5223 refers to a directory, and if necessary it will add a directory
5224 separator character at the end of the path.
5226 @option{-B} prefixes that effectively specify directory names also apply
5227 to libraries in the linker, because the compiler translates these
5228 options into @option{-L} options for the linker. They also apply to
5229 includes files in the preprocessor, because the compiler translates these
5230 options into @option{-isystem} options for the preprocessor. In this case,
5231 the compiler appends @samp{include} to the prefix.
5233 The run-time support file @file{libgcc.a} can also be searched for using
5234 the @option{-B} prefix, if needed. If it is not found there, the two
5235 standard prefixes above are tried, and that is all. The file is left
5236 out of the link if it is not found by those means.
5238 Another way to specify a prefix much like the @option{-B} prefix is to use
5239 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5242 As a special kludge, if the path provided by @option{-B} is
5243 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5244 9, then it will be replaced by @file{[dir/]include}. This is to help
5245 with boot-strapping the compiler.
5247 @item -specs=@var{file}
5249 Process @var{file} after the compiler reads in the standard @file{specs}
5250 file, in order to override the defaults that the @file{gcc} driver
5251 program uses when determining what switches to pass to @file{cc1},
5252 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5253 @option{-specs=@var{file}} can be specified on the command line, and they
5254 are processed in order, from left to right.
5260 @section Specifying subprocesses and the switches to pass to them
5263 @command{gcc} is a driver program. It performs its job by invoking a
5264 sequence of other programs to do the work of compiling, assembling and
5265 linking. GCC interprets its command-line parameters and uses these to
5266 deduce which programs it should invoke, and which command-line options
5267 it ought to place on their command lines. This behavior is controlled
5268 by @dfn{spec strings}. In most cases there is one spec string for each
5269 program that GCC can invoke, but a few programs have multiple spec
5270 strings to control their behavior. The spec strings built into GCC can
5271 be overridden by using the @option{-specs=} command-line switch to specify
5274 @dfn{Spec files} are plaintext files that are used to construct spec
5275 strings. They consist of a sequence of directives separated by blank
5276 lines. The type of directive is determined by the first non-whitespace
5277 character on the line and it can be one of the following:
5280 @item %@var{command}
5281 Issues a @var{command} to the spec file processor. The commands that can
5285 @item %include <@var{file}>
5287 Search for @var{file} and insert its text at the current point in the
5290 @item %include_noerr <@var{file}>
5291 @cindex %include_noerr
5292 Just like @samp{%include}, but do not generate an error message if the include
5293 file cannot be found.
5295 @item %rename @var{old_name} @var{new_name}
5297 Rename the spec string @var{old_name} to @var{new_name}.
5301 @item *[@var{spec_name}]:
5302 This tells the compiler to create, override or delete the named spec
5303 string. All lines after this directive up to the next directive or
5304 blank line are considered to be the text for the spec string. If this
5305 results in an empty string then the spec will be deleted. (Or, if the
5306 spec did not exist, then nothing will happened.) Otherwise, if the spec
5307 does not currently exist a new spec will be created. If the spec does
5308 exist then its contents will be overridden by the text of this
5309 directive, unless the first character of that text is the @samp{+}
5310 character, in which case the text will be appended to the spec.
5312 @item [@var{suffix}]:
5313 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5314 and up to the next directive or blank line are considered to make up the
5315 spec string for the indicated suffix. When the compiler encounters an
5316 input file with the named suffix, it will processes the spec string in
5317 order to work out how to compile that file. For example:
5324 This says that any input file whose name ends in @samp{.ZZ} should be
5325 passed to the program @samp{z-compile}, which should be invoked with the
5326 command-line switch @option{-input} and with the result of performing the
5327 @samp{%i} substitution. (See below.)
5329 As an alternative to providing a spec string, the text that follows a
5330 suffix directive can be one of the following:
5333 @item @@@var{language}
5334 This says that the suffix is an alias for a known @var{language}. This is
5335 similar to using the @option{-x} command-line switch to GCC to specify a
5336 language explicitly. For example:
5343 Says that .ZZ files are, in fact, C++ source files.
5346 This causes an error messages saying:
5349 @var{name} compiler not installed on this system.
5353 GCC already has an extensive list of suffixes built into it.
5354 This directive will add an entry to the end of the list of suffixes, but
5355 since the list is searched from the end backwards, it is effectively
5356 possible to override earlier entries using this technique.
5360 GCC has the following spec strings built into it. Spec files can
5361 override these strings or create their own. Note that individual
5362 targets can also add their own spec strings to this list.
5365 asm Options to pass to the assembler
5366 asm_final Options to pass to the assembler post-processor
5367 cpp Options to pass to the C preprocessor
5368 cc1 Options to pass to the C compiler
5369 cc1plus Options to pass to the C++ compiler
5370 endfile Object files to include at the end of the link
5371 link Options to pass to the linker
5372 lib Libraries to include on the command line to the linker
5373 libgcc Decides which GCC support library to pass to the linker
5374 linker Sets the name of the linker
5375 predefines Defines to be passed to the C preprocessor
5376 signed_char Defines to pass to CPP to say whether @code{char} is signed
5378 startfile Object files to include at the start of the link
5381 Here is a small example of a spec file:
5387 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5390 This example renames the spec called @samp{lib} to @samp{old_lib} and
5391 then overrides the previous definition of @samp{lib} with a new one.
5392 The new definition adds in some extra command-line options before
5393 including the text of the old definition.
5395 @dfn{Spec strings} are a list of command-line options to be passed to their
5396 corresponding program. In addition, the spec strings can contain
5397 @samp{%}-prefixed sequences to substitute variable text or to
5398 conditionally insert text into the command line. Using these constructs
5399 it is possible to generate quite complex command lines.
5401 Here is a table of all defined @samp{%}-sequences for spec
5402 strings. Note that spaces are not generated automatically around the
5403 results of expanding these sequences. Therefore you can concatenate them
5404 together or combine them with constant text in a single argument.
5408 Substitute one @samp{%} into the program name or argument.
5411 Substitute the name of the input file being processed.
5414 Substitute the basename of the input file being processed.
5415 This is the substring up to (and not including) the last period
5416 and not including the directory.
5419 This is the same as @samp{%b}, but include the file suffix (text after
5423 Marks the argument containing or following the @samp{%d} as a
5424 temporary file name, so that that file will be deleted if GCC exits
5425 successfully. Unlike @samp{%g}, this contributes no text to the
5428 @item %g@var{suffix}
5429 Substitute a file name that has suffix @var{suffix} and is chosen
5430 once per compilation, and mark the argument in the same way as
5431 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5432 name is now chosen in a way that is hard to predict even when previously
5433 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5434 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5435 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5436 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5437 was simply substituted with a file name chosen once per compilation,
5438 without regard to any appended suffix (which was therefore treated
5439 just like ordinary text), making such attacks more likely to succeed.
5441 @item %u@var{suffix}
5442 Like @samp{%g}, but generates a new temporary file name even if
5443 @samp{%u@var{suffix}} was already seen.
5445 @item %U@var{suffix}
5446 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5447 new one if there is no such last file name. In the absence of any
5448 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5449 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5450 would involve the generation of two distinct file names, one
5451 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5452 simply substituted with a file name chosen for the previous @samp{%u},
5453 without regard to any appended suffix.
5455 @item %j@var{suffix}
5456 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5457 writable, and if save-temps is off; otherwise, substitute the name
5458 of a temporary file, just like @samp{%u}. This temporary file is not
5459 meant for communication between processes, but rather as a junk
5462 @item %|@var{suffix}
5463 @itemx %m@var{suffix}
5464 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5465 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5466 all. These are the two most common ways to instruct a program that it
5467 should read from standard input or write to standard output. If you
5468 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5469 construct: see for example @file{f/lang-specs.h}.
5471 @item %.@var{SUFFIX}
5472 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5473 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5474 terminated by the next space or %.
5477 Marks the argument containing or following the @samp{%w} as the
5478 designated output file of this compilation. This puts the argument
5479 into the sequence of arguments that @samp{%o} will substitute later.
5482 Substitutes the names of all the output files, with spaces
5483 automatically placed around them. You should write spaces
5484 around the @samp{%o} as well or the results are undefined.
5485 @samp{%o} is for use in the specs for running the linker.
5486 Input files whose names have no recognized suffix are not compiled
5487 at all, but they are included among the output files, so they will
5491 Substitutes the suffix for object files. Note that this is
5492 handled specially when it immediately follows @samp{%g, %u, or %U},
5493 because of the need for those to form complete file names. The
5494 handling is such that @samp{%O} is treated exactly as if it had already
5495 been substituted, except that @samp{%g, %u, and %U} do not currently
5496 support additional @var{suffix} characters following @samp{%O} as they would
5497 following, for example, @samp{.o}.
5500 Substitutes the standard macro predefinitions for the
5501 current target machine. Use this when running @code{cpp}.
5504 Like @samp{%p}, but puts @samp{__} before and after the name of each
5505 predefined macro, except for macros that start with @samp{__} or with
5506 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5510 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5511 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5512 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5516 Current argument is the name of a library or startup file of some sort.
5517 Search for that file in a standard list of directories and substitute
5518 the full name found.
5521 Print @var{str} as an error message. @var{str} is terminated by a newline.
5522 Use this when inconsistent options are detected.
5525 Substitute the contents of spec string @var{name} at this point.
5528 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5530 @item %x@{@var{option}@}
5531 Accumulate an option for @samp{%X}.
5534 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5538 Output the accumulated assembler options specified by @option{-Wa}.
5541 Output the accumulated preprocessor options specified by @option{-Wp}.
5544 Process the @code{asm} spec. This is used to compute the
5545 switches to be passed to the assembler.
5548 Process the @code{asm_final} spec. This is a spec string for
5549 passing switches to an assembler post-processor, if such a program is
5553 Process the @code{link} spec. This is the spec for computing the
5554 command line passed to the linker. Typically it will make use of the
5555 @samp{%L %G %S %D and %E} sequences.
5558 Dump out a @option{-L} option for each directory that GCC believes might
5559 contain startup files. If the target supports multilibs then the
5560 current multilib directory will be prepended to each of these paths.
5563 Output the multilib directory with directory separators replaced with
5564 @samp{_}. If multilib directories are not set, or the multilib directory is
5565 @file{.} then this option emits nothing.
5568 Process the @code{lib} spec. This is a spec string for deciding which
5569 libraries should be included on the command line to the linker.
5572 Process the @code{libgcc} spec. This is a spec string for deciding
5573 which GCC support library should be included on the command line to the linker.
5576 Process the @code{startfile} spec. This is a spec for deciding which
5577 object files should be the first ones passed to the linker. Typically
5578 this might be a file named @file{crt0.o}.
5581 Process the @code{endfile} spec. This is a spec string that specifies
5582 the last object files that will be passed to the linker.
5585 Process the @code{cpp} spec. This is used to construct the arguments
5586 to be passed to the C preprocessor.
5589 Process the @code{signed_char} spec. This is intended to be used
5590 to tell cpp whether a char is signed. It typically has the definition:
5592 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
5596 Process the @code{cc1} spec. This is used to construct the options to be
5597 passed to the actual C compiler (@samp{cc1}).
5600 Process the @code{cc1plus} spec. This is used to construct the options to be
5601 passed to the actual C++ compiler (@samp{cc1plus}).
5604 Substitute the variable part of a matched option. See below.
5605 Note that each comma in the substituted string is replaced by
5609 Remove all occurrences of @code{-S} from the command line. Note---this
5610 command is position dependent. @samp{%} commands in the spec string
5611 before this one will see @code{-S}, @samp{%} commands in the spec string
5612 after this one will not.
5614 @item %:@var{function}(@var{args})
5615 Call the named function @var{function}, passing it @var{args}.
5616 @var{args} is first processed as a nested spec string, then split
5617 into an argument vector in the usual fashion. The function returns
5618 a string which is processed as if it had appeared literally as part
5619 of the current spec.
5621 The following built-in spec functions are provided:
5624 @item @code{if-exists}
5625 The @code{if-exists} spec function takes one argument, an absolute
5626 pathname to a file. If the file exists, @code{if-exists} returns the
5627 pathname. Here is a small example of its usage:
5631 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5634 @item @code{if-exists-else}
5635 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5636 spec function, except that it takes two arguments. The first argument is
5637 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5638 returns the pathname. If it does not exist, it returns the second argument.
5639 This way, @code{if-exists-else} can be used to select one file or another,
5640 based on the existence of the first. Here is a small example of its usage:
5644 crt0%O%s %:if-exists(crti%O%s) \
5645 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5650 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5651 If that switch was not specified, this substitutes nothing. Note that
5652 the leading dash is omitted when specifying this option, and it is
5653 automatically inserted if the substitution is performed. Thus the spec
5654 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5655 and would output the command line option @option{-foo}.
5657 @item %W@{@code{S}@}
5658 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5661 @item %@{@code{S}*@}
5662 Substitutes all the switches specified to GCC whose names start
5663 with @code{-S}, but which also take an argument. This is used for
5664 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5665 GCC considers @option{-o foo} as being
5666 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5667 text, including the space. Thus two arguments would be generated.
5669 @item %@{@code{S}*&@code{T}*@}
5670 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5671 (the order of @code{S} and @code{T} in the spec is not significant).
5672 There can be any number of ampersand-separated variables; for each the
5673 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5675 @item %@{@code{S}:@code{X}@}
5676 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
5678 @item %@{!@code{S}:@code{X}@}
5679 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
5681 @item %@{@code{S}*:@code{X}@}
5682 Substitutes @code{X} if one or more switches whose names start with
5683 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
5684 once, no matter how many such switches appeared. However, if @code{%*}
5685 appears somewhere in @code{X}, then @code{X} will be substituted once
5686 for each matching switch, with the @code{%*} replaced by the part of
5687 that switch that matched the @code{*}.
5689 @item %@{.@code{S}:@code{X}@}
5690 Substitutes @code{X}, if processing a file with suffix @code{S}.
5692 @item %@{!.@code{S}:@code{X}@}
5693 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
5695 @item %@{@code{S}|@code{P}:@code{X}@}
5696 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
5697 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
5698 although they have a stronger binding than the @samp{|}. If @code{%*}
5699 appears in @code{X}, all of the alternatives must be starred, and only
5700 the first matching alternative is substituted.
5702 For example, a spec string like this:
5705 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5708 will output the following command-line options from the following input
5709 command-line options:
5714 -d fred.c -foo -baz -boggle
5715 -d jim.d -bar -baz -boggle
5718 @item %@{S:X; T:Y; :D@}
5720 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
5721 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
5722 be as many clauses as you need. This may be combined with @code{.},
5723 @code{!}, @code{|}, and @code{*} as needed.
5728 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
5729 construct may contain other nested @samp{%} constructs or spaces, or
5730 even newlines. They are processed as usual, as described above.
5731 Trailing white space in @code{X} is ignored. White space may also
5732 appear anywhere on the left side of the colon in these constructs,
5733 except between @code{.} or @code{*} and the corresponding word.
5735 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
5736 handled specifically in these constructs. If another value of
5737 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5738 @option{-W} switch is found later in the command line, the earlier
5739 switch value is ignored, except with @{@code{S}*@} where @code{S} is
5740 just one letter, which passes all matching options.
5742 The character @samp{|} at the beginning of the predicate text is used to
5743 indicate that a command should be piped to the following command, but
5744 only if @option{-pipe} is specified.
5746 It is built into GCC which switches take arguments and which do not.
5747 (You might think it would be useful to generalize this to allow each
5748 compiler's spec to say which switches take arguments. But this cannot
5749 be done in a consistent fashion. GCC cannot even decide which input
5750 files have been specified without knowing which switches take arguments,
5751 and it must know which input files to compile in order to tell which
5754 GCC also knows implicitly that arguments starting in @option{-l} are to be
5755 treated as compiler output files, and passed to the linker in their
5756 proper position among the other output files.
5758 @c man begin OPTIONS
5760 @node Target Options
5761 @section Specifying Target Machine and Compiler Version
5762 @cindex target options
5763 @cindex cross compiling
5764 @cindex specifying machine version
5765 @cindex specifying compiler version and target machine
5766 @cindex compiler version, specifying
5767 @cindex target machine, specifying
5769 The usual way to run GCC is to run the executable called @file{gcc}, or
5770 @file{<machine>-gcc} when cross-compiling, or
5771 @file{<machine>-gcc-<version>} to run a version other than the one that
5772 was installed last. Sometimes this is inconvenient, so GCC provides
5773 options that will switch to another cross-compiler or version.
5776 @item -b @var{machine}
5778 The argument @var{machine} specifies the target machine for compilation.
5780 The value to use for @var{machine} is the same as was specified as the
5781 machine type when configuring GCC as a cross-compiler. For
5782 example, if a cross-compiler was configured with @samp{configure
5783 i386v}, meaning to compile for an 80386 running System V, then you
5784 would specify @option{-b i386v} to run that cross compiler.
5786 @item -V @var{version}
5788 The argument @var{version} specifies which version of GCC to run.
5789 This is useful when multiple versions are installed. For example,
5790 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5793 The @option{-V} and @option{-b} options work by running the
5794 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5795 use them if you can just run that directly.
5797 @node Submodel Options
5798 @section Hardware Models and Configurations
5799 @cindex submodel options
5800 @cindex specifying hardware config
5801 @cindex hardware models and configurations, specifying
5802 @cindex machine dependent options
5804 Earlier we discussed the standard option @option{-b} which chooses among
5805 different installed compilers for completely different target
5806 machines, such as VAX vs.@: 68000 vs.@: 80386.
5808 In addition, each of these target machine types can have its own
5809 special options, starting with @samp{-m}, to choose among various
5810 hardware models or configurations---for example, 68010 vs 68020,
5811 floating coprocessor or none. A single installed version of the
5812 compiler can compile for any model or configuration, according to the
5815 Some configurations of the compiler also support additional special
5816 options, usually for compatibility with other compilers on the same
5819 These options are defined by the macro @code{TARGET_SWITCHES} in the
5820 machine description. The default for the options is also defined by
5821 that macro, which enables you to change the defaults.
5831 * RS/6000 and PowerPC Options::
5834 * i386 and x86-64 Options::
5836 * DEC Alpha Options::
5837 * DEC Alpha/VMS Options::
5840 * System V Options::
5841 * TMS320C3x/C4x Options::
5848 * S/390 and zSeries Options::
5852 * Xstormy16 Options::
5857 @node M680x0 Options
5858 @subsection M680x0 Options
5859 @cindex M680x0 options
5861 These are the @samp{-m} options defined for the 68000 series. The default
5862 values for these options depends on which style of 68000 was selected when
5863 the compiler was configured; the defaults for the most common choices are
5871 Generate output for a 68000. This is the default
5872 when the compiler is configured for 68000-based systems.
5874 Use this option for microcontrollers with a 68000 or EC000 core,
5875 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5881 Generate output for a 68020. This is the default
5882 when the compiler is configured for 68020-based systems.
5886 Generate output containing 68881 instructions for floating point.
5887 This is the default for most 68020 systems unless @option{--nfp} was
5888 specified when the compiler was configured.
5892 Generate output for a 68030. This is the default when the compiler is
5893 configured for 68030-based systems.
5897 Generate output for a 68040. This is the default when the compiler is
5898 configured for 68040-based systems.
5900 This option inhibits the use of 68881/68882 instructions that have to be
5901 emulated by software on the 68040. Use this option if your 68040 does not
5902 have code to emulate those instructions.
5906 Generate output for a 68060. This is the default when the compiler is
5907 configured for 68060-based systems.
5909 This option inhibits the use of 68020 and 68881/68882 instructions that
5910 have to be emulated by software on the 68060. Use this option if your 68060
5911 does not have code to emulate those instructions.
5915 Generate output for a CPU32. This is the default
5916 when the compiler is configured for CPU32-based systems.
5918 Use this option for microcontrollers with a
5919 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5920 68336, 68340, 68341, 68349 and 68360.
5924 Generate output for a 520X ``coldfire'' family cpu. This is the default
5925 when the compiler is configured for 520X-based systems.
5927 Use this option for microcontroller with a 5200 core, including
5928 the MCF5202, MCF5203, MCF5204 and MCF5202.
5933 Generate output for a 68040, without using any of the new instructions.
5934 This results in code which can run relatively efficiently on either a
5935 68020/68881 or a 68030 or a 68040. The generated code does use the
5936 68881 instructions that are emulated on the 68040.
5940 Generate output for a 68060, without using any of the new instructions.
5941 This results in code which can run relatively efficiently on either a
5942 68020/68881 or a 68030 or a 68040. The generated code does use the
5943 68881 instructions that are emulated on the 68060.
5946 @opindex msoft-float
5947 Generate output containing library calls for floating point.
5948 @strong{Warning:} the requisite libraries are not available for all m68k
5949 targets. Normally the facilities of the machine's usual C compiler are
5950 used, but this can't be done directly in cross-compilation. You must
5951 make your own arrangements to provide suitable library functions for
5952 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5953 @samp{m68k-*-coff} do provide software floating point support.
5957 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5960 @opindex mnobitfield
5961 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5962 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5966 Do use the bit-field instructions. The @option{-m68020} option implies
5967 @option{-mbitfield}. This is the default if you use a configuration
5968 designed for a 68020.
5972 Use a different function-calling convention, in which functions
5973 that take a fixed number of arguments return with the @code{rtd}
5974 instruction, which pops their arguments while returning. This
5975 saves one instruction in the caller since there is no need to pop
5976 the arguments there.
5978 This calling convention is incompatible with the one normally
5979 used on Unix, so you cannot use it if you need to call libraries
5980 compiled with the Unix compiler.
5982 Also, you must provide function prototypes for all functions that
5983 take variable numbers of arguments (including @code{printf});
5984 otherwise incorrect code will be generated for calls to those
5987 In addition, seriously incorrect code will result if you call a
5988 function with too many arguments. (Normally, extra arguments are
5989 harmlessly ignored.)
5991 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5992 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5995 @itemx -mno-align-int
5997 @opindex mno-align-int
5998 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5999 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
6000 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
6001 Aligning variables on 32-bit boundaries produces code that runs somewhat
6002 faster on processors with 32-bit busses at the expense of more memory.
6004 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
6005 align structures containing the above types differently than
6006 most published application binary interface specifications for the m68k.
6010 Use the pc-relative addressing mode of the 68000 directly, instead of
6011 using a global offset table. At present, this option implies @option{-fpic},
6012 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
6013 not presently supported with @option{-mpcrel}, though this could be supported for
6014 68020 and higher processors.
6016 @item -mno-strict-align
6017 @itemx -mstrict-align
6018 @opindex mno-strict-align
6019 @opindex mstrict-align
6020 Do not (do) assume that unaligned memory references will be handled by
6024 Generate code that allows the data segment to be located in a different
6025 area of memory from the text segment. This allows for execute in place in
6026 an environment without virtual memory management. This option implies -fPIC.
6029 Generate code that assumes that the data segment follows the text segment.
6030 This is the default.
6032 @item -mid-shared-library
6033 Generate code that supports shared libraries via the library ID method.
6034 This allows for execute in place and shared libraries in an environment
6035 without virtual memory management. This option implies -fPIC.
6037 @item -mno-id-shared-library
6038 Generate code that doesn't assume ID based shared libraries are being used.
6039 This is the default.
6041 @item -mshared-library-id=n
6042 Specified the identification number of the ID based shared library being
6043 compiled. Specifying a value of 0 will generate more compact code, specifying
6044 other values will force the allocation of that number to the current
6045 library but is no more space or time efficient than omitting this option.
6049 @node M68hc1x Options
6050 @subsection M68hc1x Options
6051 @cindex M68hc1x options
6053 These are the @samp{-m} options defined for the 68hc11 and 68hc12
6054 microcontrollers. The default values for these options depends on
6055 which style of microcontroller was selected when the compiler was configured;
6056 the defaults for the most common choices are given below.
6063 Generate output for a 68HC11. This is the default
6064 when the compiler is configured for 68HC11-based systems.
6070 Generate output for a 68HC12. This is the default
6071 when the compiler is configured for 68HC12-based systems.
6077 Generate output for a 68HCS12.
6080 @opindex mauto-incdec
6081 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
6088 Enable the use of 68HC12 min and max instructions.
6091 @itemx -mno-long-calls
6092 @opindex mlong-calls
6093 @opindex mno-long-calls
6094 Treat all calls as being far away (near). If calls are assumed to be
6095 far away, the compiler will use the @code{call} instruction to
6096 call a function and the @code{rtc} instruction for returning.
6100 Consider type @code{int} to be 16 bits wide, like @code{short int}.
6102 @item -msoft-reg-count=@var{count}
6103 @opindex msoft-reg-count
6104 Specify the number of pseudo-soft registers which are used for the
6105 code generation. The maximum number is 32. Using more pseudo-soft
6106 register may or may not result in better code depending on the program.
6107 The default is 4 for 68HC11 and 2 for 68HC12.
6112 @subsection VAX Options
6115 These @samp{-m} options are defined for the VAX:
6120 Do not output certain jump instructions (@code{aobleq} and so on)
6121 that the Unix assembler for the VAX cannot handle across long
6126 Do output those jump instructions, on the assumption that you
6127 will assemble with the GNU assembler.
6131 Output code for g-format floating point numbers instead of d-format.
6135 @subsection SPARC Options
6136 @cindex SPARC options
6138 These @samp{-m} options are supported on the SPARC:
6143 @opindex mno-app-regs
6145 Specify @option{-mapp-regs} to generate output using the global registers
6146 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
6149 To be fully SVR4 ABI compliant at the cost of some performance loss,
6150 specify @option{-mno-app-regs}. You should compile libraries and system
6151 software with this option.
6156 @opindex mhard-float
6157 Generate output containing floating point instructions. This is the
6163 @opindex msoft-float
6164 Generate output containing library calls for floating point.
6165 @strong{Warning:} the requisite libraries are not available for all SPARC
6166 targets. Normally the facilities of the machine's usual C compiler are
6167 used, but this cannot be done directly in cross-compilation. You must make
6168 your own arrangements to provide suitable library functions for
6169 cross-compilation. The embedded targets @samp{sparc-*-aout} and
6170 @samp{sparclite-*-*} do provide software floating point support.
6172 @option{-msoft-float} changes the calling convention in the output file;
6173 therefore, it is only useful if you compile @emph{all} of a program with
6174 this option. In particular, you need to compile @file{libgcc.a}, the
6175 library that comes with GCC, with @option{-msoft-float} in order for
6178 @item -mhard-quad-float
6179 @opindex mhard-quad-float
6180 Generate output containing quad-word (long double) floating point
6183 @item -msoft-quad-float
6184 @opindex msoft-quad-float
6185 Generate output containing library calls for quad-word (long double)
6186 floating point instructions. The functions called are those specified
6187 in the SPARC ABI@. This is the default.
6189 As of this writing, there are no SPARC implementations that have hardware
6190 support for the quad-word floating point instructions. They all invoke
6191 a trap handler for one of these instructions, and then the trap handler
6192 emulates the effect of the instruction. Because of the trap handler overhead,
6193 this is much slower than calling the ABI library routines. Thus the
6194 @option{-msoft-quad-float} option is the default.
6196 @item -mno-unaligned-doubles
6197 @itemx -munaligned-doubles
6198 @opindex mno-unaligned-doubles
6199 @opindex munaligned-doubles
6200 Assume that doubles have 8 byte alignment. This is the default.
6202 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
6203 alignment only if they are contained in another type, or if they have an
6204 absolute address. Otherwise, it assumes they have 4 byte alignment.
6205 Specifying this option avoids some rare compatibility problems with code
6206 generated by other compilers. It is not the default because it results
6207 in a performance loss, especially for floating point code.
6209 @item -mno-faster-structs
6210 @itemx -mfaster-structs
6211 @opindex mno-faster-structs
6212 @opindex mfaster-structs
6213 With @option{-mfaster-structs}, the compiler assumes that structures
6214 should have 8 byte alignment. This enables the use of pairs of
6215 @code{ldd} and @code{std} instructions for copies in structure
6216 assignment, in place of twice as many @code{ld} and @code{st} pairs.
6217 However, the use of this changed alignment directly violates the SPARC
6218 ABI@. Thus, it's intended only for use on targets where the developer
6219 acknowledges that their resulting code will not be directly in line with
6220 the rules of the ABI@.
6223 @opindex mimpure-text
6224 @option{-mimpure-text}, used in addition to @option{-shared}, tells
6225 the compiler to not pass @option{-z text} to the linker when linking a
6226 shared object. Using this option, you can link position-dependent
6227 code into a shared object.
6229 @option{-mimpure-text} suppresses the ``relocations remain against
6230 allocatable but non-writable sections'' linker error message.
6231 However, the necessary relocations will trigger copy-on-write, and the
6232 shared object is not actually shared across processes. Instead of
6233 using @option{-mimpure-text}, you should compile all source code with
6234 @option{-fpic} or @option{-fPIC}.
6236 This option is only available on SunOS and Solaris.
6238 @item -mcpu=@var{cpu_type}
6240 Set the instruction set, register set, and instruction scheduling parameters
6241 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
6242 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
6243 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
6244 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
6247 Default instruction scheduling parameters are used for values that select
6248 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
6249 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
6251 Here is a list of each supported architecture and their supported
6256 v8: supersparc, hypersparc
6257 sparclite: f930, f934, sparclite86x
6259 v9: ultrasparc, ultrasparc3
6262 By default (unless configured otherwise), GCC generates code for the V7
6263 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
6264 additionally optimizes it for the Cypress CY7C602 chip, as used in the
6265 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
6266 SPARCStation 1, 2, IPX etc.
6268 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
6269 architecture. The only difference from V7 code is that the compiler emits
6270 the integer multiply and integer divide instructions which exist in SPARC-V8
6271 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
6272 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
6275 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
6276 the SPARC architecture. This adds the integer multiply, integer divide step
6277 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
6278 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
6279 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU. With
6280 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
6281 MB86934 chip, which is the more recent SPARClite with FPU.
6283 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
6284 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
6285 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
6286 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
6287 optimizes it for the TEMIC SPARClet chip.
6289 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
6290 architecture. This adds 64-bit integer and floating-point move instructions,
6291 3 additional floating-point condition code registers and conditional move
6292 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
6293 optimizes it for the Sun UltraSPARC I/II chips. With
6294 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
6295 Sun UltraSPARC III chip.
6297 @item -mtune=@var{cpu_type}
6299 Set the instruction scheduling parameters for machine type
6300 @var{cpu_type}, but do not set the instruction set or register set that the
6301 option @option{-mcpu=@var{cpu_type}} would.
6303 The same values for @option{-mcpu=@var{cpu_type}} can be used for
6304 @option{-mtune=@var{cpu_type}}, but the only useful values are those
6305 that select a particular cpu implementation. Those are @samp{cypress},
6306 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
6307 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
6313 @opindex -mno-v8plus
6314 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI. The
6315 difference from the V8 ABI is that the global and out registers are
6316 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
6317 mode for all SPARC-V9 processors.
6323 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
6324 Visual Instruction Set extensions. The default is @option{-mno-vis}.
6327 These @samp{-m} options are supported in addition to the above
6328 on SPARC-V9 processors in 64-bit environments:
6331 @item -mlittle-endian
6332 @opindex mlittle-endian
6333 Generate code for a processor running in little-endian mode. It is only
6334 available for a few configurations and most notably not on Solaris.
6340 Generate code for a 32-bit or 64-bit environment.
6341 The 32-bit environment sets int, long and pointer to 32 bits.
6342 The 64-bit environment sets int to 32 bits and long and pointer
6345 @item -mcmodel=medlow
6346 @opindex mcmodel=medlow
6347 Generate code for the Medium/Low code model: the program must be linked
6348 in the low 32 bits of the address space. Pointers are 64 bits.
6349 Programs can be statically or dynamically linked.
6351 @item -mcmodel=medmid
6352 @opindex mcmodel=medmid
6353 Generate code for the Medium/Middle code model: the program must be linked
6354 in the low 44 bits of the address space, the text segment must be less than
6355 2G bytes, and data segment must be within 2G of the text segment.
6356 Pointers are 64 bits.
6358 @item -mcmodel=medany
6359 @opindex mcmodel=medany
6360 Generate code for the Medium/Anywhere code model: the program may be linked
6361 anywhere in the address space, the text segment must be less than
6362 2G bytes, and data segment must be within 2G of the text segment.
6363 Pointers are 64 bits.
6365 @item -mcmodel=embmedany
6366 @opindex mcmodel=embmedany
6367 Generate code for the Medium/Anywhere code model for embedded systems:
6368 assume a 32-bit text and a 32-bit data segment, both starting anywhere
6369 (determined at link time). Register %g4 points to the base of the
6370 data segment. Pointers are still 64 bits.
6371 Programs are statically linked, PIC is not supported.
6374 @itemx -mno-stack-bias
6375 @opindex mstack-bias
6376 @opindex mno-stack-bias
6377 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
6378 frame pointer if present, are offset by @minus{}2047 which must be added back
6379 when making stack frame references. This is the default in 64-bit mode.
6380 Otherwise, assume no such offset is present.
6384 @subsection ARM Options
6387 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6392 @opindex mapcs-frame
6393 Generate a stack frame that is compliant with the ARM Procedure Call
6394 Standard for all functions, even if this is not strictly necessary for
6395 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6396 with this option will cause the stack frames not to be generated for
6397 leaf functions. The default is @option{-mno-apcs-frame}.
6401 This is a synonym for @option{-mapcs-frame}.
6405 Generate code for a processor running with a 26-bit program counter,
6406 and conforming to the function calling standards for the APCS 26-bit
6407 option. This option replaces the @option{-m2} and @option{-m3} options
6408 of previous releases of the compiler.
6412 Generate code for a processor running with a 32-bit program counter,
6413 and conforming to the function calling standards for the APCS 32-bit
6414 option. This option replaces the @option{-m6} option of previous releases
6418 @c not currently implemented
6419 @item -mapcs-stack-check
6420 @opindex mapcs-stack-check
6421 Generate code to check the amount of stack space available upon entry to
6422 every function (that actually uses some stack space). If there is
6423 insufficient space available then either the function
6424 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6425 called, depending upon the amount of stack space required. The run time
6426 system is required to provide these functions. The default is
6427 @option{-mno-apcs-stack-check}, since this produces smaller code.
6429 @c not currently implemented
6431 @opindex mapcs-float
6432 Pass floating point arguments using the float point registers. This is
6433 one of the variants of the APCS@. This option is recommended if the
6434 target hardware has a floating point unit or if a lot of floating point
6435 arithmetic is going to be performed by the code. The default is
6436 @option{-mno-apcs-float}, since integer only code is slightly increased in
6437 size if @option{-mapcs-float} is used.
6439 @c not currently implemented
6440 @item -mapcs-reentrant
6441 @opindex mapcs-reentrant
6442 Generate reentrant, position independent code. The default is
6443 @option{-mno-apcs-reentrant}.
6446 @item -mthumb-interwork
6447 @opindex mthumb-interwork
6448 Generate code which supports calling between the ARM and Thumb
6449 instruction sets. Without this option the two instruction sets cannot
6450 be reliably used inside one program. The default is
6451 @option{-mno-thumb-interwork}, since slightly larger code is generated
6452 when @option{-mthumb-interwork} is specified.
6454 @item -mno-sched-prolog
6455 @opindex mno-sched-prolog
6456 Prevent the reordering of instructions in the function prolog, or the
6457 merging of those instruction with the instructions in the function's
6458 body. This means that all functions will start with a recognizable set
6459 of instructions (or in fact one of a choice from a small set of
6460 different function prologues), and this information can be used to
6461 locate the start if functions inside an executable piece of code. The
6462 default is @option{-msched-prolog}.
6465 @opindex mhard-float
6466 Generate output containing floating point instructions. This is the
6470 @opindex msoft-float
6471 Generate output containing library calls for floating point.
6472 @strong{Warning:} the requisite libraries are not available for all ARM
6473 targets. Normally the facilities of the machine's usual C compiler are
6474 used, but this cannot be done directly in cross-compilation. You must make
6475 your own arrangements to provide suitable library functions for
6478 @option{-msoft-float} changes the calling convention in the output file;
6479 therefore, it is only useful if you compile @emph{all} of a program with
6480 this option. In particular, you need to compile @file{libgcc.a}, the
6481 library that comes with GCC, with @option{-msoft-float} in order for
6484 @item -mfloat-abi=@var{name}
6486 Specifies which ABI to use for floating point values. Permissible values
6487 are: @samp{soft}, @samp{softfp} and @samp{hard}.
6489 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
6490 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
6491 of floating point instructions, but still uses the soft-float calling
6494 @item -mlittle-endian
6495 @opindex mlittle-endian
6496 Generate code for a processor running in little-endian mode. This is
6497 the default for all standard configurations.
6500 @opindex mbig-endian
6501 Generate code for a processor running in big-endian mode; the default is
6502 to compile code for a little-endian processor.
6504 @item -mwords-little-endian
6505 @opindex mwords-little-endian
6506 This option only applies when generating code for big-endian processors.
6507 Generate code for a little-endian word order but a big-endian byte
6508 order. That is, a byte order of the form @samp{32107654}. Note: this
6509 option should only be used if you require compatibility with code for
6510 big-endian ARM processors generated by versions of the compiler prior to
6513 @item -malignment-traps
6514 @opindex malignment-traps
6515 Generate code that will not trap if the MMU has alignment traps enabled.
6516 On ARM architectures prior to ARMv4, there were no instructions to
6517 access half-word objects stored in memory. However, when reading from
6518 memory a feature of the ARM architecture allows a word load to be used,
6519 even if the address is unaligned, and the processor core will rotate the
6520 data as it is being loaded. This option tells the compiler that such
6521 misaligned accesses will cause a MMU trap and that it should instead
6522 synthesize the access as a series of byte accesses. The compiler can
6523 still use word accesses to load half-word data if it knows that the
6524 address is aligned to a word boundary.
6526 This option is ignored when compiling for ARM architecture 4 or later,
6527 since these processors have instructions to directly access half-word
6530 @item -mno-alignment-traps
6531 @opindex mno-alignment-traps
6532 Generate code that assumes that the MMU will not trap unaligned
6533 accesses. This produces better code when the target instruction set
6534 does not have half-word memory operations (i.e.@: implementations prior to
6537 Note that you cannot use this option to access unaligned word objects,
6538 since the processor will only fetch one 32-bit aligned object from
6541 The default setting for most targets is @option{-mno-alignment-traps}, since
6542 this produces better code when there are no half-word memory
6543 instructions available.
6545 @item -mshort-load-bytes
6546 @itemx -mno-short-load-words
6547 @opindex mshort-load-bytes
6548 @opindex mno-short-load-words
6549 These are deprecated aliases for @option{-malignment-traps}.
6551 @item -mno-short-load-bytes
6552 @itemx -mshort-load-words
6553 @opindex mno-short-load-bytes
6554 @opindex mshort-load-words
6555 This are deprecated aliases for @option{-mno-alignment-traps}.
6557 @item -mcpu=@var{name}
6559 This specifies the name of the target ARM processor. GCC uses this name
6560 to determine what kind of instructions it can emit when generating
6561 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6562 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6563 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6564 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6565 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6566 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6567 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6568 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6569 @samp{arm920t}, @samp{arm926ejs}, @samp{arm940t}, @samp{arm9tdmi},
6570 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ejs},
6571 @samp{arm1136js}, @samp{arm1136jfs} ,@samp{xscale}, @samp{iwmmxt},
6574 @itemx -mtune=@var{name}
6576 This option is very similar to the @option{-mcpu=} option, except that
6577 instead of specifying the actual target processor type, and hence
6578 restricting which instructions can be used, it specifies that GCC should
6579 tune the performance of the code as if the target were of the type
6580 specified in this option, but still choosing the instructions that it
6581 will generate based on the cpu specified by a @option{-mcpu=} option.
6582 For some ARM implementations better performance can be obtained by using
6585 @item -march=@var{name}
6587 This specifies the name of the target ARM architecture. GCC uses this
6588 name to determine what kind of instructions it can emit when generating
6589 assembly code. This option can be used in conjunction with or instead
6590 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6591 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6592 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
6593 @samp{iwmmxt}, @samp{ep9312}.
6595 @item -mfpu=@var{name}
6596 @itemx -mfpe=@var{number}
6597 @itemx -mfp=@var{number}
6601 This specifies what floating point hardware (or hardware emulation) is
6602 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
6603 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
6604 are synonyms for @option{-mpfu}=@samp{fpe}@var{number}, for compatibility
6605 with older versions of GCC@.
6607 If @option{-msoft-float} is specified this specifies the format of
6608 floating point values.
6610 @item -mstructure-size-boundary=@var{n}
6611 @opindex mstructure-size-boundary
6612 The size of all structures and unions will be rounded up to a multiple
6613 of the number of bits set by this option. Permissible values are 8 and
6614 32. The default value varies for different toolchains. For the COFF
6615 targeted toolchain the default value is 8. Specifying the larger number
6616 can produce faster, more efficient code, but can also increase the size
6617 of the program. The two values are potentially incompatible. Code
6618 compiled with one value cannot necessarily expect to work with code or
6619 libraries compiled with the other value, if they exchange information
6620 using structures or unions.
6622 @item -mabort-on-noreturn
6623 @opindex mabort-on-noreturn
6624 Generate a call to the function @code{abort} at the end of a
6625 @code{noreturn} function. It will be executed if the function tries to
6629 @itemx -mno-long-calls
6630 @opindex mlong-calls
6631 @opindex mno-long-calls
6632 Tells the compiler to perform function calls by first loading the
6633 address of the function into a register and then performing a subroutine
6634 call on this register. This switch is needed if the target function
6635 will lie outside of the 64 megabyte addressing range of the offset based
6636 version of subroutine call instruction.
6638 Even if this switch is enabled, not all function calls will be turned
6639 into long calls. The heuristic is that static functions, functions
6640 which have the @samp{short-call} attribute, functions that are inside
6641 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6642 definitions have already been compiled within the current compilation
6643 unit, will not be turned into long calls. The exception to this rule is
6644 that weak function definitions, functions with the @samp{long-call}
6645 attribute or the @samp{section} attribute, and functions that are within
6646 the scope of a @samp{#pragma long_calls} directive, will always be
6647 turned into long calls.
6649 This feature is not enabled by default. Specifying
6650 @option{-mno-long-calls} will restore the default behavior, as will
6651 placing the function calls within the scope of a @samp{#pragma
6652 long_calls_off} directive. Note these switches have no effect on how
6653 the compiler generates code to handle function calls via function
6656 @item -mnop-fun-dllimport
6657 @opindex mnop-fun-dllimport
6658 Disable support for the @code{dllimport} attribute.
6660 @item -msingle-pic-base
6661 @opindex msingle-pic-base
6662 Treat the register used for PIC addressing as read-only, rather than
6663 loading it in the prologue for each function. The run-time system is
6664 responsible for initializing this register with an appropriate value
6665 before execution begins.
6667 @item -mpic-register=@var{reg}
6668 @opindex mpic-register
6669 Specify the register to be used for PIC addressing. The default is R10
6670 unless stack-checking is enabled, when R9 is used.
6672 @item -mcirrus-fix-invalid-insns
6673 @opindex mcirrus-fix-invalid-insns
6674 @opindex mno-cirrus-fix-invalid-insns
6675 Insert NOPs into the instruction stream to in order to work around
6676 problems with invalid Maverick instruction combinations. This option
6677 is only valid if the @option{-mcpu=ep9312} option has been used to
6678 enable generation of instructions for the Cirrus Maverick floating
6679 point co-processor. This option is not enabled by default, since the
6680 problem is only present in older Maverick implementations. The default
6681 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6684 @item -mpoke-function-name
6685 @opindex mpoke-function-name
6686 Write the name of each function into the text section, directly
6687 preceding the function prologue. The generated code is similar to this:
6691 .ascii "arm_poke_function_name", 0
6694 .word 0xff000000 + (t1 - t0)
6695 arm_poke_function_name
6697 stmfd sp!, @{fp, ip, lr, pc@}
6701 When performing a stack backtrace, code can inspect the value of
6702 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6703 location @code{pc - 12} and the top 8 bits are set, then we know that
6704 there is a function name embedded immediately preceding this location
6705 and has length @code{((pc[-3]) & 0xff000000)}.
6709 Generate code for the 16-bit Thumb instruction set. The default is to
6710 use the 32-bit ARM instruction set.
6713 @opindex mtpcs-frame
6714 Generate a stack frame that is compliant with the Thumb Procedure Call
6715 Standard for all non-leaf functions. (A leaf function is one that does
6716 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6718 @item -mtpcs-leaf-frame
6719 @opindex mtpcs-leaf-frame
6720 Generate a stack frame that is compliant with the Thumb Procedure Call
6721 Standard for all leaf functions. (A leaf function is one that does
6722 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6724 @item -mcallee-super-interworking
6725 @opindex mcallee-super-interworking
6726 Gives all externally visible functions in the file being compiled an ARM
6727 instruction set header which switches to Thumb mode before executing the
6728 rest of the function. This allows these functions to be called from
6729 non-interworking code.
6731 @item -mcaller-super-interworking
6732 @opindex mcaller-super-interworking
6733 Allows calls via function pointers (including virtual functions) to
6734 execute correctly regardless of whether the target code has been
6735 compiled for interworking or not. There is a small overhead in the cost
6736 of executing a function pointer if this option is enabled.
6740 @node MN10300 Options
6741 @subsection MN10300 Options
6742 @cindex MN10300 options
6744 These @option{-m} options are defined for Matsushita MN10300 architectures:
6749 Generate code to avoid bugs in the multiply instructions for the MN10300
6750 processors. This is the default.
6753 @opindex mno-mult-bug
6754 Do not generate code to avoid bugs in the multiply instructions for the
6759 Generate code which uses features specific to the AM33 processor.
6763 Do not generate code which uses features specific to the AM33 processor. This
6768 Do not link in the C run-time initialization object file.
6772 Indicate to the linker that it should perform a relaxation optimization pass
6773 to shorten branches, calls and absolute memory addresses. This option only
6774 has an effect when used on the command line for the final link step.
6776 This option makes symbolic debugging impossible.
6780 @node M32R/D Options
6781 @subsection M32R/D Options
6782 @cindex M32R/D options
6784 These @option{-m} options are defined for Renesas M32R/D architectures:
6789 Generate code for the M32R/2@.
6793 Generate code for the M32R/X@.
6797 Generate code for the M32R@. This is the default.
6800 @opindex mmodel=small
6801 Assume all objects live in the lower 16MB of memory (so that their addresses
6802 can be loaded with the @code{ld24} instruction), and assume all subroutines
6803 are reachable with the @code{bl} instruction.
6804 This is the default.
6806 The addressability of a particular object can be set with the
6807 @code{model} attribute.
6809 @item -mmodel=medium
6810 @opindex mmodel=medium
6811 Assume objects may be anywhere in the 32-bit address space (the compiler
6812 will generate @code{seth/add3} instructions to load their addresses), and
6813 assume all subroutines are reachable with the @code{bl} instruction.
6816 @opindex mmodel=large
6817 Assume objects may be anywhere in the 32-bit address space (the compiler
6818 will generate @code{seth/add3} instructions to load their addresses), and
6819 assume subroutines may not be reachable with the @code{bl} instruction
6820 (the compiler will generate the much slower @code{seth/add3/jl}
6821 instruction sequence).
6824 @opindex msdata=none
6825 Disable use of the small data area. Variables will be put into
6826 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6827 @code{section} attribute has been specified).
6828 This is the default.
6830 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6831 Objects may be explicitly put in the small data area with the
6832 @code{section} attribute using one of these sections.
6835 @opindex msdata=sdata
6836 Put small global and static data in the small data area, but do not
6837 generate special code to reference them.
6841 Put small global and static data in the small data area, and generate
6842 special instructions to reference them.
6846 @cindex smaller data references
6847 Put global and static objects less than or equal to @var{num} bytes
6848 into the small data or bss sections instead of the normal data or bss
6849 sections. The default value of @var{num} is 8.
6850 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6851 for this option to have any effect.
6853 All modules should be compiled with the same @option{-G @var{num}} value.
6854 Compiling with different values of @var{num} may or may not work; if it
6855 doesn't the linker will give an error message---incorrect code will not be
6860 Makes the M32R specific code in the compiler display some statistics
6861 that might help in debugging programs.
6864 @opindex malign-loops
6865 Align all loops to a 32-byte boundary.
6867 @item -mno-align-loops
6868 @opindex mno-align-loops
6869 Do not enforce a 32-byte alignment for loops. This is the default.
6871 @item -missue-rate=@var{number}
6872 @opindex missue-rate=@var{number}
6873 Issue @var{number} instructions per cycle. @var{number} can only be 1
6876 @item -mbranch-cost=@var{number}
6877 @opindex mbranch-cost=@var{number}
6878 @var{number} can only be 1 or 2. If it is 1 then branches will be
6879 preferred over conditional code, if it is 2, then the opposite will
6882 @item -mflush-trap=@var{number}
6883 @opindex mflush-trap=@var{number}
6884 Specifies the trap number to use to flush the cache. The default is
6885 12. Valid numbers are between 0 and 15 inclusive.
6887 @item -mno-flush-trap
6888 @opindex mno-flush-trap
6889 Specifies that the cache cannot be flushed by using a trap.
6891 @item -mflush-func=@var{name}
6892 @opindex mflush-func=@var{name}
6893 Specifies the name of the operating system function to call to flush
6894 the cache. The default is @emph{_flush_cache}, but a function call
6895 will only be used if a trap is not available.
6897 @item -mno-flush-func
6898 @opindex mno-flush-func
6899 Indicates that there is no OS function for flushing the cache.
6903 @node RS/6000 and PowerPC Options
6904 @subsection IBM RS/6000 and PowerPC Options
6905 @cindex RS/6000 and PowerPC Options
6906 @cindex IBM RS/6000 and PowerPC Options
6908 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6916 @itemx -mpowerpc-gpopt
6917 @itemx -mno-powerpc-gpopt
6918 @itemx -mpowerpc-gfxopt
6919 @itemx -mno-powerpc-gfxopt
6921 @itemx -mno-powerpc64
6927 @opindex mno-powerpc
6928 @opindex mpowerpc-gpopt
6929 @opindex mno-powerpc-gpopt
6930 @opindex mpowerpc-gfxopt
6931 @opindex mno-powerpc-gfxopt
6933 @opindex mno-powerpc64
6934 GCC supports two related instruction set architectures for the
6935 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6936 instructions supported by the @samp{rios} chip set used in the original
6937 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6938 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6939 the IBM 4xx microprocessors.
6941 Neither architecture is a subset of the other. However there is a
6942 large common subset of instructions supported by both. An MQ
6943 register is included in processors supporting the POWER architecture.
6945 You use these options to specify which instructions are available on the
6946 processor you are using. The default value of these options is
6947 determined when configuring GCC@. Specifying the
6948 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6949 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6950 rather than the options listed above.
6952 The @option{-mpower} option allows GCC to generate instructions that
6953 are found only in the POWER architecture and to use the MQ register.
6954 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
6955 to generate instructions that are present in the POWER2 architecture but
6956 not the original POWER architecture.
6958 The @option{-mpowerpc} option allows GCC to generate instructions that
6959 are found only in the 32-bit subset of the PowerPC architecture.
6960 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
6961 GCC to use the optional PowerPC architecture instructions in the
6962 General Purpose group, including floating-point square root. Specifying
6963 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
6964 use the optional PowerPC architecture instructions in the Graphics
6965 group, including floating-point select.
6967 The @option{-mpowerpc64} option allows GCC to generate the additional
6968 64-bit instructions that are found in the full PowerPC64 architecture
6969 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6970 @option{-mno-powerpc64}.
6972 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
6973 will use only the instructions in the common subset of both
6974 architectures plus some special AIX common-mode calls, and will not use
6975 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
6976 permits GCC to use any instruction from either architecture and to
6977 allow use of the MQ register; specify this for the Motorola MPC601.
6979 @item -mnew-mnemonics
6980 @itemx -mold-mnemonics
6981 @opindex mnew-mnemonics
6982 @opindex mold-mnemonics
6983 Select which mnemonics to use in the generated assembler code. With
6984 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
6985 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
6986 assembler mnemonics defined for the POWER architecture. Instructions
6987 defined in only one architecture have only one mnemonic; GCC uses that
6988 mnemonic irrespective of which of these options is specified.
6990 GCC defaults to the mnemonics appropriate for the architecture in
6991 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
6992 value of these option. Unless you are building a cross-compiler, you
6993 should normally not specify either @option{-mnew-mnemonics} or
6994 @option{-mold-mnemonics}, but should instead accept the default.
6996 @item -mcpu=@var{cpu_type}
6998 Set architecture type, register usage, choice of mnemonics, and
6999 instruction scheduling parameters for machine type @var{cpu_type}.
7000 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
7001 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
7002 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
7003 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
7004 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
7005 @samp{860}, @samp{970}, @samp{common}, @samp{ec603e}, @samp{G3},
7006 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
7007 @samp{power4}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
7008 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64a}.
7010 @option{-mcpu=common} selects a completely generic processor. Code
7011 generated under this option will run on any POWER or PowerPC processor.
7012 GCC will use only the instructions in the common subset of both
7013 architectures, and will not use the MQ register. GCC assumes a generic
7014 processor model for scheduling purposes.
7016 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
7017 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
7018 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
7019 types, with an appropriate, generic processor model assumed for
7020 scheduling purposes.
7022 The other options specify a specific processor. Code generated under
7023 those options will run best on that processor, and may not run at all on
7026 The @option{-mcpu} options automatically enable or disable the
7027 following options: @option{-maltivec}, @option{-mhard-float},
7028 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
7029 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
7030 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
7031 @option{-mstring}. The particular options set for any particular CPU
7032 will vary between compiler versions, depending on what setting seems
7033 to produce optimal code for that CPU; it doesn't necessarily reflect
7034 the actual hardware's capabilities. If you wish to set an individual
7035 option to a particular value, you may specify it after the
7036 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
7038 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
7039 not enabled or disabled by the @option{-mcpu} option at present, since
7040 AIX does not have full support for these options. You may still
7041 enable or disable them individually if you're sure it'll work in your
7044 @item -mtune=@var{cpu_type}
7046 Set the instruction scheduling parameters for machine type
7047 @var{cpu_type}, but do not set the architecture type, register usage, or
7048 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
7049 values for @var{cpu_type} are used for @option{-mtune} as for
7050 @option{-mcpu}. If both are specified, the code generated will use the
7051 architecture, registers, and mnemonics set by @option{-mcpu}, but the
7052 scheduling parameters set by @option{-mtune}.
7057 @opindex mno-altivec
7058 These switches enable or disable the use of built-in functions that
7059 allow access to the AltiVec instruction set. You may also need to set
7060 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
7065 Extend the current ABI with SPE ABI extensions. This does not change
7066 the default ABI, instead it adds the SPE ABI extensions to the current
7070 @opindex mabi=no-spe
7071 Disable Booke SPE ABI extensions for the current ABI.
7073 @item -misel=@var{yes/no}
7076 This switch enables or disables the generation of ISEL instructions.
7078 @item -mspe=@var{yes/no}
7081 This switch enables or disables the generation of SPE simd
7084 @item -mfloat-gprs=@var{yes/no}
7086 @opindex mfloat-gprs
7087 This switch enables or disables the generation of floating point
7088 operations on the general purpose registers for architectures that
7089 support it. This option is currently only available on the MPC8540.
7092 @itemx -mno-fp-in-toc
7093 @itemx -mno-sum-in-toc
7094 @itemx -mminimal-toc
7096 @opindex mno-fp-in-toc
7097 @opindex mno-sum-in-toc
7098 @opindex mminimal-toc
7099 Modify generation of the TOC (Table Of Contents), which is created for
7100 every executable file. The @option{-mfull-toc} option is selected by
7101 default. In that case, GCC will allocate at least one TOC entry for
7102 each unique non-automatic variable reference in your program. GCC
7103 will also place floating-point constants in the TOC@. However, only
7104 16,384 entries are available in the TOC@.
7106 If you receive a linker error message that saying you have overflowed
7107 the available TOC space, you can reduce the amount of TOC space used
7108 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
7109 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
7110 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
7111 generate code to calculate the sum of an address and a constant at
7112 run-time instead of putting that sum into the TOC@. You may specify one
7113 or both of these options. Each causes GCC to produce very slightly
7114 slower and larger code at the expense of conserving TOC space.
7116 If you still run out of space in the TOC even when you specify both of
7117 these options, specify @option{-mminimal-toc} instead. This option causes
7118 GCC to make only one TOC entry for every file. When you specify this
7119 option, GCC will produce code that is slower and larger but which
7120 uses extremely little TOC space. You may wish to use this option
7121 only on files that contain less frequently executed code.
7127 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
7128 @code{long} type, and the infrastructure needed to support them.
7129 Specifying @option{-maix64} implies @option{-mpowerpc64} and
7130 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
7131 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
7136 @opindex mno-xl-call
7137 On AIX, pass floating-point arguments to prototyped functions beyond the
7138 register save area (RSA) on the stack in addition to argument FPRs. The
7139 AIX calling convention was extended but not initially documented to
7140 handle an obscure K&R C case of calling a function that takes the
7141 address of its arguments with fewer arguments than declared. AIX XL
7142 compilers access floating point arguments which do not fit in the
7143 RSA from the stack when a subroutine is compiled without
7144 optimization. Because always storing floating-point arguments on the
7145 stack is inefficient and rarely needed, this option is not enabled by
7146 default and only is necessary when calling subroutines compiled by AIX
7147 XL compilers without optimization.
7151 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
7152 application written to use message passing with special startup code to
7153 enable the application to run. The system must have PE installed in the
7154 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
7155 must be overridden with the @option{-specs=} option to specify the
7156 appropriate directory location. The Parallel Environment does not
7157 support threads, so the @option{-mpe} option and the @option{-pthread}
7158 option are incompatible.
7160 @item -malign-natural
7161 @itemx -malign-power
7162 @opindex malign-natural
7163 @opindex malign-power
7164 On AIX, Darwin, and 64-bit PowerPC GNU/Linux, the option
7165 @option{-malign-natural} overrides the ABI-defined alignment of larger
7166 types, such as floating-point doubles, on their natural size-based boundary.
7167 The option @option{-malign-power} instructs GCC to follow the ABI-specified
7168 alignment rules. GCC defaults to the standard alignment defined in the ABI.
7172 @opindex msoft-float
7173 @opindex mhard-float
7174 Generate code that does not use (uses) the floating-point register set.
7175 Software floating point emulation is provided if you use the
7176 @option{-msoft-float} option, and pass the option to GCC when linking.
7179 @itemx -mno-multiple
7181 @opindex mno-multiple
7182 Generate code that uses (does not use) the load multiple word
7183 instructions and the store multiple word instructions. These
7184 instructions are generated by default on POWER systems, and not
7185 generated on PowerPC systems. Do not use @option{-mmultiple} on little
7186 endian PowerPC systems, since those instructions do not work when the
7187 processor is in little endian mode. The exceptions are PPC740 and
7188 PPC750 which permit the instructions usage in little endian mode.
7194 Generate code that uses (does not use) the load string instructions
7195 and the store string word instructions to save multiple registers and
7196 do small block moves. These instructions are generated by default on
7197 POWER systems, and not generated on PowerPC systems. Do not use
7198 @option{-mstring} on little endian PowerPC systems, since those
7199 instructions do not work when the processor is in little endian mode.
7200 The exceptions are PPC740 and PPC750 which permit the instructions
7201 usage in little endian mode.
7207 Generate code that uses (does not use) the load or store instructions
7208 that update the base register to the address of the calculated memory
7209 location. These instructions are generated by default. If you use
7210 @option{-mno-update}, there is a small window between the time that the
7211 stack pointer is updated and the address of the previous frame is
7212 stored, which means code that walks the stack frame across interrupts or
7213 signals may get corrupted data.
7216 @itemx -mno-fused-madd
7217 @opindex mfused-madd
7218 @opindex mno-fused-madd
7219 Generate code that uses (does not use) the floating point multiply and
7220 accumulate instructions. These instructions are generated by default if
7221 hardware floating is used.
7223 @item -mno-bit-align
7225 @opindex mno-bit-align
7227 On System V.4 and embedded PowerPC systems do not (do) force structures
7228 and unions that contain bit-fields to be aligned to the base type of the
7231 For example, by default a structure containing nothing but 8
7232 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7233 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7234 the structure would be aligned to a 1 byte boundary and be one byte in
7237 @item -mno-strict-align
7238 @itemx -mstrict-align
7239 @opindex mno-strict-align
7240 @opindex mstrict-align
7241 On System V.4 and embedded PowerPC systems do not (do) assume that
7242 unaligned memory references will be handled by the system.
7245 @itemx -mno-relocatable
7246 @opindex mrelocatable
7247 @opindex mno-relocatable
7248 On embedded PowerPC systems generate code that allows (does not allow)
7249 the program to be relocated to a different address at runtime. If you
7250 use @option{-mrelocatable} on any module, all objects linked together must
7251 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7253 @item -mrelocatable-lib
7254 @itemx -mno-relocatable-lib
7255 @opindex mrelocatable-lib
7256 @opindex mno-relocatable-lib
7257 On embedded PowerPC systems generate code that allows (does not allow)
7258 the program to be relocated to a different address at runtime. Modules
7259 compiled with @option{-mrelocatable-lib} can be linked with either modules
7260 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7261 with modules compiled with the @option{-mrelocatable} options.
7267 On System V.4 and embedded PowerPC systems do not (do) assume that
7268 register 2 contains a pointer to a global area pointing to the addresses
7269 used in the program.
7272 @itemx -mlittle-endian
7274 @opindex mlittle-endian
7275 On System V.4 and embedded PowerPC systems compile code for the
7276 processor in little endian mode. The @option{-mlittle-endian} option is
7277 the same as @option{-mlittle}.
7282 @opindex mbig-endian
7283 On System V.4 and embedded PowerPC systems compile code for the
7284 processor in big endian mode. The @option{-mbig-endian} option is
7285 the same as @option{-mbig}.
7287 @item -mdynamic-no-pic
7288 @opindex mdynamic-no-pic
7289 On Darwin and Mac OS X systems, compile code so that it is not
7290 relocatable, but that its external references are relocatable. The
7291 resulting code is suitable for applications, but not shared
7294 @item -mprioritize-restricted-insns=@var{priority}
7295 @opindex mprioritize-restricted-insns
7296 This option controls the priority that is assigned to
7297 dispatch-slot restricted instructions during the second scheduling
7298 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
7299 @var{no/highest/second-highest} priority to dispatch slot restricted
7302 @item -msched-costly-dep=@var{dependence_type}
7303 @opindex msched-costly-dep
7304 This option controls which dependences are considered costly
7305 by the target during instruction scheduling. The argument
7306 @var{dependence_type} takes one of the following values:
7307 @var{no}: no dependence is costly,
7308 @var{all}: all dependences are costly,
7309 @var{true_store_to_load}: a true dependence from store to load is costly,
7310 @var{store_to_load}: any dependence from store to load is costly,
7311 @var{number}: any dependence which latency >= @var{number} is costly.
7313 @item -minsert-sched-nops=@var{scheme}
7314 @opindex minsert-sched-nops
7315 This option controls which nop insertion scheme will be used during
7316 the second scheduling pass. The argument @var{scheme} takes one of the
7318 @var{no}: Don't insert nops.
7319 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
7320 according to the scheduler's grouping.
7321 @var{regroup_exact}: Insert nops to force costly dependent insns into
7322 separate groups. Insert exactly as many nops as needed to force an insn
7323 to a new group, according to the estimated processor grouping.
7324 @var{number}: Insert nops to force costly dependent insns into
7325 separate groups. Insert @var{number} nops to force an insn to a new group.
7329 On System V.4 and embedded PowerPC systems compile code using calling
7330 conventions that adheres to the March 1995 draft of the System V
7331 Application Binary Interface, PowerPC processor supplement. This is the
7332 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7334 @item -mcall-sysv-eabi
7335 @opindex mcall-sysv-eabi
7336 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7338 @item -mcall-sysv-noeabi
7339 @opindex mcall-sysv-noeabi
7340 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7342 @item -mcall-solaris
7343 @opindex mcall-solaris
7344 On System V.4 and embedded PowerPC systems compile code for the Solaris
7348 @opindex mcall-linux
7349 On System V.4 and embedded PowerPC systems compile code for the
7350 Linux-based GNU system.
7354 On System V.4 and embedded PowerPC systems compile code for the
7355 Hurd-based GNU system.
7358 @opindex mcall-netbsd
7359 On System V.4 and embedded PowerPC systems compile code for the
7360 NetBSD operating system.
7362 @item -maix-struct-return
7363 @opindex maix-struct-return
7364 Return all structures in memory (as specified by the AIX ABI)@.
7366 @item -msvr4-struct-return
7367 @opindex msvr4-struct-return
7368 Return structures smaller than 8 bytes in registers (as specified by the
7372 @opindex mabi=altivec
7373 Extend the current ABI with AltiVec ABI extensions. This does not
7374 change the default ABI, instead it adds the AltiVec ABI extensions to
7377 @item -mabi=no-altivec
7378 @opindex mabi=no-altivec
7379 Disable AltiVec ABI extensions for the current ABI.
7382 @itemx -mno-prototype
7384 @opindex mno-prototype
7385 On System V.4 and embedded PowerPC systems assume that all calls to
7386 variable argument functions are properly prototyped. Otherwise, the
7387 compiler must insert an instruction before every non prototyped call to
7388 set or clear bit 6 of the condition code register (@var{CR}) to
7389 indicate whether floating point values were passed in the floating point
7390 registers in case the function takes a variable arguments. With
7391 @option{-mprototype}, only calls to prototyped variable argument functions
7392 will set or clear the bit.
7396 On embedded PowerPC systems, assume that the startup module is called
7397 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7398 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7403 On embedded PowerPC systems, assume that the startup module is called
7404 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7409 On embedded PowerPC systems, assume that the startup module is called
7410 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7414 @opindex myellowknife
7415 On embedded PowerPC systems, assume that the startup module is called
7416 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7421 On System V.4 and embedded PowerPC systems, specify that you are
7422 compiling for a VxWorks system.
7426 Specify that you are compiling for the WindISS simulation environment.
7430 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7431 header to indicate that @samp{eabi} extended relocations are used.
7437 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7438 Embedded Applications Binary Interface (eabi) which is a set of
7439 modifications to the System V.4 specifications. Selecting @option{-meabi}
7440 means that the stack is aligned to an 8 byte boundary, a function
7441 @code{__eabi} is called to from @code{main} to set up the eabi
7442 environment, and the @option{-msdata} option can use both @code{r2} and
7443 @code{r13} to point to two separate small data areas. Selecting
7444 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7445 do not call an initialization function from @code{main}, and the
7446 @option{-msdata} option will only use @code{r13} to point to a single
7447 small data area. The @option{-meabi} option is on by default if you
7448 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7451 @opindex msdata=eabi
7452 On System V.4 and embedded PowerPC systems, put small initialized
7453 @code{const} global and static data in the @samp{.sdata2} section, which
7454 is pointed to by register @code{r2}. Put small initialized
7455 non-@code{const} global and static data in the @samp{.sdata} section,
7456 which is pointed to by register @code{r13}. Put small uninitialized
7457 global and static data in the @samp{.sbss} section, which is adjacent to
7458 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7459 incompatible with the @option{-mrelocatable} option. The
7460 @option{-msdata=eabi} option also sets the @option{-memb} option.
7463 @opindex msdata=sysv
7464 On System V.4 and embedded PowerPC systems, put small global and static
7465 data in the @samp{.sdata} section, which is pointed to by register
7466 @code{r13}. Put small uninitialized global and static data in the
7467 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7468 The @option{-msdata=sysv} option is incompatible with the
7469 @option{-mrelocatable} option.
7471 @item -msdata=default
7473 @opindex msdata=default
7475 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7476 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7477 same as @option{-msdata=sysv}.
7480 @opindex msdata-data
7481 On System V.4 and embedded PowerPC systems, put small global and static
7482 data in the @samp{.sdata} section. Put small uninitialized global and
7483 static data in the @samp{.sbss} section. Do not use register @code{r13}
7484 to address small data however. This is the default behavior unless
7485 other @option{-msdata} options are used.
7489 @opindex msdata=none
7491 On embedded PowerPC systems, put all initialized global and static data
7492 in the @samp{.data} section, and all uninitialized data in the
7493 @samp{.bss} section.
7497 @cindex smaller data references (PowerPC)
7498 @cindex .sdata/.sdata2 references (PowerPC)
7499 On embedded PowerPC systems, put global and static items less than or
7500 equal to @var{num} bytes into the small data or bss sections instead of
7501 the normal data or bss section. By default, @var{num} is 8. The
7502 @option{-G @var{num}} switch is also passed to the linker.
7503 All modules should be compiled with the same @option{-G @var{num}} value.
7506 @itemx -mno-regnames
7508 @opindex mno-regnames
7509 On System V.4 and embedded PowerPC systems do (do not) emit register
7510 names in the assembly language output using symbolic forms.
7513 @itemx -mno-longcall
7515 @opindex mno-longcall
7516 Default to making all function calls via pointers, so that functions
7517 which reside further than 64 megabytes (67,108,864 bytes) from the
7518 current location can be called. This setting can be overridden by the
7519 @code{shortcall} function attribute, or by @code{#pragma longcall(0)}.
7521 Some linkers are capable of detecting out-of-range calls and generating
7522 glue code on the fly. On these systems, long calls are unnecessary and
7523 generate slower code. As of this writing, the AIX linker can do this,
7524 as can the GNU linker for PowerPC/64. It is planned to add this feature
7525 to the GNU linker for 32-bit PowerPC systems as well.
7527 On Mach-O (Darwin) systems, this option directs the compiler emit to
7528 the glue for every direct call, and the Darwin linker decides whether
7529 to use or discard it.
7531 In the future, we may cause GCC to ignore all longcall specifications
7532 when the linker is known to generate glue.
7536 Adds support for multithreading with the @dfn{pthreads} library.
7537 This option sets flags for both the preprocessor and linker.
7541 @node Darwin Options
7542 @subsection Darwin Options
7543 @cindex Darwin options
7545 These options are defined for all architectures running the Darwin operating
7546 system. They are useful for compatibility with other Mac OS compilers.
7551 Loads all members of static archive libraries.
7552 See man ld(1) for more information.
7554 @item -arch_errors_fatal
7555 @opindex arch_errors_fatal
7556 Cause the errors having to do with files that have the wrong architecture
7560 @opindex bind_at_load
7561 Causes the output file to be marked such that the dynamic linker will
7562 bind all undefined references when the file is loaded or launched.
7566 Produce a Mach-o bundle format file.
7567 See man ld(1) for more information.
7569 @item -bundle_loader @var{executable}
7570 @opindex bundle_loader
7571 This specifies the @var{executable} that will be loading the build
7572 output file being linked. See man ld(1) for more information.
7574 @item -allowable_client @var{client_name}
7578 @itemx -compatibility_version
7579 @itemx -current_version
7580 @itemx -dependency-file
7582 @itemx -dylinker_install_name
7585 @itemx -exported_symbols_list
7587 @itemx -flat_namespace
7588 @itemx -force_cpusubtype_ALL
7589 @itemx -force_flat_namespace
7590 @itemx -headerpad_max_install_names
7593 @itemx -install_name
7594 @itemx -keep_private_externs
7595 @itemx -multi_module
7596 @itemx -multiply_defined
7597 @itemx -multiply_defined_unused
7599 @itemx -nofixprebinding
7602 @itemx -noseglinkedit
7603 @itemx -pagezero_size
7605 @itemx -prebind_all_twolevel_modules
7606 @itemx -private_bundle
7607 @itemx -read_only_relocs
7609 @itemx -sectobjectsymbols
7613 @itemx -sectobjectsymbols
7615 @itemx -seg_addr_table
7616 @itemx -seg_addr_table_filename
7619 @itemx -segs_read_only_addr
7620 @itemx -segs_read_write_addr
7621 @itemx -single_module
7624 @itemx -sub_umbrella
7625 @itemx -twolevel_namespace
7628 @itemx -unexported_symbols_list
7629 @itemx -weak_reference_mismatches
7632 @opindex allowable_client
7634 @opindex client_name
7635 @opindex compatibility_version
7636 @opindex current_version
7637 @opindex dependency-file
7639 @opindex dylinker_install_name
7642 @opindex exported_symbols_list
7644 @opindex flat_namespace
7645 @opindex force_cpusubtype_ALL
7646 @opindex force_flat_namespace
7647 @opindex headerpad_max_install_names
7650 @opindex install_name
7651 @opindex keep_private_externs
7652 @opindex multi_module
7653 @opindex multiply_defined
7654 @opindex multiply_defined_unused
7656 @opindex nofixprebinding
7657 @opindex nomultidefs
7659 @opindex noseglinkedit
7660 @opindex pagezero_size
7662 @opindex prebind_all_twolevel_modules
7663 @opindex private_bundle
7664 @opindex read_only_relocs
7666 @opindex sectobjectsymbols
7670 @opindex sectobjectsymbols
7672 @opindex seg_addr_table
7673 @opindex seg_addr_table_filename
7674 @opindex seglinkedit
7676 @opindex segs_read_only_addr
7677 @opindex segs_read_write_addr
7678 @opindex single_module
7680 @opindex sub_library
7681 @opindex sub_umbrella
7682 @opindex twolevel_namespace
7685 @opindex unexported_symbols_list
7686 @opindex weak_reference_mismatches
7687 @opindex whatsloaded
7689 These options are available for Darwin linker. Darwin linker man page
7690 describes them in detail.
7695 @subsection MIPS Options
7696 @cindex MIPS options
7702 Generate big-endian code.
7706 Generate little-endian code. This is the default for @samp{mips*el-*-*}
7709 @item -march=@var{arch}
7711 Generate code that will run on @var{arch}, which can be the name of a
7712 generic MIPS ISA, or the name of a particular processor.
7714 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
7715 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
7716 The processor names are:
7717 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
7719 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
7720 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
7724 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4300},
7725 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
7726 The special value @samp{from-abi} selects the
7727 most compatible architecture for the selected ABI (that is,
7728 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
7730 In processor names, a final @samp{000} can be abbreviated as @samp{k}
7731 (for example, @samp{-march=r2k}). Prefixes are optional, and
7732 @samp{vr} may be written @samp{r}.
7734 GCC defines two macros based on the value of this option. The first
7735 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
7736 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
7737 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
7738 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
7739 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
7741 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
7742 above. In other words, it will have the full prefix and will not
7743 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
7744 the macro names the resolved architecture (either @samp{"mips1"} or
7745 @samp{"mips3"}). It names the default architecture when no
7746 @option{-march} option is given.
7748 @item -mtune=@var{arch}
7750 Optimize for @var{arch}. Among other things, this option controls
7751 the way instructions are scheduled, and the perceived cost of arithmetic
7752 operations. The list of @var{arch} values is the same as for
7755 When this option is not used, GCC will optimize for the processor
7756 specified by @option{-march}. By using @option{-march} and
7757 @option{-mtune} together, it is possible to generate code that will
7758 run on a family of processors, but optimize the code for one
7759 particular member of that family.
7761 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
7762 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
7763 @samp{-march} ones described above.
7767 Equivalent to @samp{-march=mips1}.
7771 Equivalent to @samp{-march=mips2}.
7775 Equivalent to @samp{-march=mips3}.
7779 Equivalent to @samp{-march=mips4}.
7783 Equivalent to @samp{-march=mips32}.
7787 Equivalent to @samp{-march=mips32r2}.
7791 Equivalent to @samp{-march=mips64}.
7797 Use (do not use) the MIPS16 ISA.
7809 Generate code for the given ABI@.
7811 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
7812 generates 64-bit code when you select a 64-bit architecture, but you
7813 can use @option{-mgp32} to get 32-bit code instead.
7816 @itemx -mno-abicalls
7818 @opindex mno-abicalls
7819 Generate (do not generate) SVR4-style position-independent code.
7820 @option{-mabicalls} is the default for SVR4-based systems.
7826 Lift (do not lift) the usual restrictions on the size of the global
7829 GCC normally uses a single instruction to load values from the GOT.
7830 While this is relatively efficient, it will only work if the GOT
7831 is smaller than about 64k. Anything larger will cause the linker
7832 to report an error such as:
7834 @cindex relocation truncated to fit (MIPS)
7836 relocation truncated to fit: R_MIPS_GOT16 foobar
7839 If this happens, you should recompile your code with @option{-mxgot}.
7840 It should then work with very large GOTs, although it will also be
7841 less efficient, since it will take three instructions to fetch the
7842 value of a global symbol.
7844 Note that some linkers can create multiple GOTs. If you have such a
7845 linker, you should only need to use @option{-mxgot} when a single object
7846 file accesses more than 64k's worth of GOT entries. Very few do.
7848 These options have no effect unless GCC is generating position
7851 @item -membedded-pic
7852 @itemx -mno-embedded-pic
7853 @opindex membedded-pic
7854 @opindex mno-embedded-pic
7855 Generate (do not generate) position-independent code suitable for some
7856 embedded systems. All calls are made using PC relative addresses, and
7857 all data is addressed using the $gp register. No more than 65536
7858 bytes of global data may be used. This requires GNU as and GNU ld,
7859 which do most of the work.
7863 Assume that general-purpose registers are 32 bits wide.
7867 Assume that general-purpose registers are 64 bits wide.
7871 Assume that floating-point registers are 32 bits wide.
7875 Assume that floating-point registers are 64 bits wide.
7878 @opindex mhard-float
7879 Use floating-point coprocessor instructions.
7882 @opindex msoft-float
7883 Do not use floating-point coprocessor instructions. Implement
7884 floating-point calculations using library calls instead.
7886 @item -msingle-float
7887 @opindex msingle-float
7888 Assume that the floating-point coprocessor only supports single-precision
7891 @itemx -mdouble-float
7892 @opindex mdouble-float
7893 Assume that the floating-point coprocessor supports double-precision
7894 operations. This is the default.
7898 Force @code{int} and @code{long} types to be 64 bits wide. See
7899 @option{-mlong32} for an explanation of the default and the way
7900 that the pointer size is determined.
7904 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
7905 an explanation of the default and the way that the pointer size is
7910 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
7912 The default size of @code{int}s, @code{long}s and pointers depends on
7913 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
7914 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
7915 32-bit @code{long}s. Pointers are the same size as @code{long}s,
7916 or the same size as integer registers, whichever is smaller.
7920 @cindex smaller data references (MIPS)
7921 @cindex gp-relative references (MIPS)
7922 Put global and static items less than or equal to @var{num} bytes into
7923 the small data or bss section instead of the normal data or bss section.
7924 This allows the data to be accessed using a single instruction.
7926 All modules should be compiled with the same @option{-G @var{num}}
7929 @item -membedded-data
7930 @itemx -mno-embedded-data
7931 @opindex membedded-data
7932 @opindex mno-embedded-data
7933 Allocate variables to the read-only data section first if possible, then
7934 next in the small data section if possible, otherwise in data. This gives
7935 slightly slower code than the default, but reduces the amount of RAM required
7936 when executing, and thus may be preferred for some embedded systems.
7938 @item -muninit-const-in-rodata
7939 @itemx -mno-uninit-const-in-rodata
7940 @opindex muninit-const-in-rodata
7941 @opindex mno-uninit-const-in-rodata
7942 Put uninitialized @code{const} variables in the read-only data section.
7943 This option is only meaningful in conjunction with @option{-membedded-data}.
7945 @item -msplit-addresses
7946 @itemx -mno-split-addresses
7947 @opindex msplit-addresses
7948 @opindex mno-split-addresses
7949 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
7950 relocation operators. This option has been superceded by
7951 @option{-mexplicit-relocs} but is retained for backwards compatibility.
7953 @item -mexplicit-relocs
7954 @itemx -mno-explicit-relocs
7955 @opindex mexplicit-relocs
7956 @opindex mno-explicit-relocs
7957 Use (do not use) assembler relocation operators when dealing with symbolic
7958 addresses. The alternative, selected by @option{-mno-explicit-relocs},
7959 is to use assembler macros instead.
7961 @option{-mexplicit-relocs} is usually the default if GCC was
7962 configured to use an assembler that supports relocation operators.
7963 However, there are two exceptions:
7967 GCC is not yet able to generate explicit relocations for the combination
7968 of @option{-mabi=64} and @option{-mno-abicalls}. This will be addressed
7969 in a future release.
7972 The combination of @option{-mabicalls} and @option{-fno-unit-at-a-time}
7973 implies @option{-mno-explicit-relocs} unless explicitly overridden.
7974 This is because, when generating abicalls, the choice of relocation
7975 depends on whether a symbol is local or global. In some rare cases,
7976 GCC will not be able to decide this until the whole compilation unit
7984 Generate (do not generate) code that refers to registers using their
7985 software names. The default is @option{-mno-rnames}, which tells GCC
7986 to use hardware names like @samp{$4} instead of software names like
7987 @samp{a0}. The only assembler known to support @option{-rnames} is
7988 the Algorithmics assembler.
7990 @item -mcheck-zero-division
7991 @itemx -mno-check-zero-division
7992 @opindex mcheck-zero-division
7993 @opindex mno-check-zero-division
7994 Trap (do not trap) on integer division by zero. The default is
7995 @option{-mcheck-zero-division}.
8001 Force (do not force) the use of @code{memcpy()} for non-trivial block
8002 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
8003 most constant-sized copies.
8006 @itemx -mno-long-calls
8007 @opindex mlong-calls
8008 @opindex mno-long-calls
8009 Disable (do not disable) use of the @code{jal} instruction. Calling
8010 functions using @code{jal} is more efficient but requires the caller
8011 and callee to be in the same 256 megabyte segment.
8013 This option has no effect on abicalls code. The default is
8014 @option{-mno-long-calls}.
8020 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
8021 instructions, as provided by the R4650 ISA.
8024 @itemx -mno-fused-madd
8025 @opindex mfused-madd
8026 @opindex mno-fused-madd
8027 Enable (disable) use of the floating point multiply-accumulate
8028 instructions, when they are available. The default is
8029 @option{-mfused-madd}.
8031 When multiply-accumulate instructions are used, the intermediate
8032 product is calculated to infinite precision and is not subject to
8033 the FCSR Flush to Zero bit. This may be undesirable in some
8038 Tell the MIPS assembler to not run its preprocessor over user
8039 assembler files (with a @samp{.s} suffix) when assembling them.
8044 Work around certain SB-1 CPU core errata.
8045 (This flag currently works around the SB-1 revision 2
8046 ``F1'' and ``F2'' floating point errata.)
8048 @item -mflush-func=@var{func}
8049 @itemx -mno-flush-func
8050 @opindex mflush-func
8051 Specifies the function to call to flush the I and D caches, or to not
8052 call any such function. If called, the function must take the same
8053 arguments as the common @code{_flush_func()}, that is, the address of the
8054 memory range for which the cache is being flushed, the size of the
8055 memory range, and the number 3 (to flush both caches). The default
8056 depends on the target gcc was configured for, but commonly is either
8057 @samp{_flush_func} or @samp{__cpu_flush}.
8059 @item -mbranch-likely
8060 @itemx -mno-branch-likely
8061 @opindex mbranch-likely
8062 @opindex mno-branch-likely
8063 Enable or disable use of Branch Likely instructions, regardless of the
8064 default for the selected architecture. By default, Branch Likely
8065 instructions may be generated if they are supported by the selected
8066 architecture. An exception is for the MIPS32 and MIPS64 architectures
8067 and processors which implement those architectures; for those, Branch
8068 Likely instructions will not be generated by default because the MIPS32
8069 and MIPS64 architectures specifically deprecate their use.
8072 @node i386 and x86-64 Options
8073 @subsection Intel 386 and AMD x86-64 Options
8074 @cindex i386 Options
8075 @cindex x86-64 Options
8076 @cindex Intel 386 Options
8077 @cindex AMD x86-64 Options
8079 These @samp{-m} options are defined for the i386 and x86-64 family of
8083 @item -mtune=@var{cpu-type}
8085 Tune to @var{cpu-type} everything applicable about the generated code, except
8086 for the ABI and the set of available instructions. The choices for
8090 Original Intel's i386 CPU.
8092 Intel's i486 CPU. (No scheduling is implemented for this chip.)
8094 Intel Pentium CPU with no MMX support.
8096 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
8097 @item i686, pentiumpro
8098 Intel PentiumPro CPU.
8100 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
8102 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
8105 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
8107 AMD K6 CPU with MMX instruction set support.
8109 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
8110 @item athlon, athlon-tbird
8111 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
8113 @item athlon-4, athlon-xp, athlon-mp
8114 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
8115 instruction set support.
8116 @item k8, opteron, athlon64, athlon-fx
8117 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
8118 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
8120 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
8123 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
8124 instruction set support.
8126 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
8127 implemented for this chip.)
8129 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
8130 implemented for this chip.)
8133 While picking a specific @var{cpu-type} will schedule things appropriately
8134 for that particular chip, the compiler will not generate any code that
8135 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8138 @item -march=@var{cpu-type}
8140 Generate instructions for the machine type @var{cpu-type}. The choices
8141 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8142 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8144 @item -mcpu=@var{cpu-type}
8146 A deprecated synonym for @option{-mtune}.
8155 @opindex mpentiumpro
8156 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8157 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8158 These synonyms are deprecated.
8160 @item -mfpmath=@var{unit}
8162 Generate floating point arithmetics for selected unit @var{unit}. The choices
8167 Use the standard 387 floating point coprocessor present majority of chips and
8168 emulated otherwise. Code compiled with this option will run almost everywhere.
8169 The temporary results are computed in 80bit precision instead of precision
8170 specified by the type resulting in slightly different results compared to most
8171 of other chips. See @option{-ffloat-store} for more detailed description.
8173 This is the default choice for i386 compiler.
8176 Use scalar floating point instructions present in the SSE instruction set.
8177 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8178 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8179 instruction set supports only single precision arithmetics, thus the double and
8180 extended precision arithmetics is still done using 387. Later version, present
8181 only in Pentium4 and the future AMD x86-64 chips supports double precision
8184 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8185 @option{-msse2} switches to enable SSE extensions and make this option
8186 effective. For x86-64 compiler, these extensions are enabled by default.
8188 The resulting code should be considerably faster in the majority of cases and avoid
8189 the numerical instability problems of 387 code, but may break some existing
8190 code that expects temporaries to be 80bit.
8192 This is the default choice for the x86-64 compiler.
8195 Use all SSE extensions enabled by @option{-msse2} as well as the new
8196 SSE extensions in Prescott New Instructions. @option{-mpni} also
8197 enables 2 builtin functions, @code{__builtin_ia32_monitor} and
8198 @code{__builtin_ia32_mwait}, for new instructions @code{monitor} and
8202 Attempt to utilize both instruction sets at once. This effectively double the
8203 amount of available registers and on chips with separate execution units for
8204 387 and SSE the execution resources too. Use this option with care, as it is
8205 still experimental, because the gcc register allocator does not model separate
8206 functional units well resulting in instable performance.
8209 @item -masm=@var{dialect}
8210 @opindex masm=@var{dialect}
8211 Output asm instructions using selected @var{dialect}. Supported choices are
8212 @samp{intel} or @samp{att} (the default one).
8217 @opindex mno-ieee-fp
8218 Control whether or not the compiler uses IEEE floating point
8219 comparisons. These handle correctly the case where the result of a
8220 comparison is unordered.
8223 @opindex msoft-float
8224 Generate output containing library calls for floating point.
8225 @strong{Warning:} the requisite libraries are not part of GCC@.
8226 Normally the facilities of the machine's usual C compiler are used, but
8227 this can't be done directly in cross-compilation. You must make your
8228 own arrangements to provide suitable library functions for
8231 On machines where a function returns floating point results in the 80387
8232 register stack, some floating point opcodes may be emitted even if
8233 @option{-msoft-float} is used.
8235 @item -mno-fp-ret-in-387
8236 @opindex mno-fp-ret-in-387
8237 Do not use the FPU registers for return values of functions.
8239 The usual calling convention has functions return values of types
8240 @code{float} and @code{double} in an FPU register, even if there
8241 is no FPU@. The idea is that the operating system should emulate
8244 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8245 in ordinary CPU registers instead.
8247 @item -mno-fancy-math-387
8248 @opindex mno-fancy-math-387
8249 Some 387 emulators do not support the @code{sin}, @code{cos} and
8250 @code{sqrt} instructions for the 387. Specify this option to avoid
8251 generating those instructions. This option is the default on FreeBSD,
8252 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8253 indicates that the target cpu will always have an FPU and so the
8254 instruction will not need emulation. As of revision 2.6.1, these
8255 instructions are not generated unless you also use the
8256 @option{-funsafe-math-optimizations} switch.
8258 @item -malign-double
8259 @itemx -mno-align-double
8260 @opindex malign-double
8261 @opindex mno-align-double
8262 Control whether GCC aligns @code{double}, @code{long double}, and
8263 @code{long long} variables on a two word boundary or a one word
8264 boundary. Aligning @code{double} variables on a two word boundary will
8265 produce code that runs somewhat faster on a @samp{Pentium} at the
8266 expense of more memory.
8268 @strong{Warning:} if you use the @option{-malign-double} switch,
8269 structures containing the above types will be aligned differently than
8270 the published application binary interface specifications for the 386
8271 and will not be binary compatible with structures in code compiled
8272 without that switch.
8274 @item -m96bit-long-double
8275 @itemx -m128bit-long-double
8276 @opindex m96bit-long-double
8277 @opindex m128bit-long-double
8278 These switches control the size of @code{long double} type. The i386
8279 application binary interface specifies the size to be 96 bits,
8280 so @option{-m96bit-long-double} is the default in 32 bit mode.
8282 Modern architectures (Pentium and newer) would prefer @code{long double}
8283 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8284 conforming to the ABI, this would not be possible. So specifying a
8285 @option{-m128bit-long-double} will align @code{long double}
8286 to a 16 byte boundary by padding the @code{long double} with an additional
8289 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8290 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8292 Notice that neither of these options enable any extra precision over the x87
8293 standard of 80 bits for a @code{long double}.
8295 @strong{Warning:} if you override the default value for your target ABI, the
8296 structures and arrays containing @code{long double} variables will change
8297 their size as well as function calling convention for function taking
8298 @code{long double} will be modified. Hence they will not be binary
8299 compatible with arrays or structures in code compiled without that switch.
8303 @itemx -mno-svr3-shlib
8304 @opindex msvr3-shlib
8305 @opindex mno-svr3-shlib
8306 Control whether GCC places uninitialized local variables into the
8307 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8308 into @code{bss}. These options are meaningful only on System V Release 3.
8312 Use a different function-calling convention, in which functions that
8313 take a fixed number of arguments return with the @code{ret} @var{num}
8314 instruction, which pops their arguments while returning. This saves one
8315 instruction in the caller since there is no need to pop the arguments
8318 You can specify that an individual function is called with this calling
8319 sequence with the function attribute @samp{stdcall}. You can also
8320 override the @option{-mrtd} option by using the function attribute
8321 @samp{cdecl}. @xref{Function Attributes}.
8323 @strong{Warning:} this calling convention is incompatible with the one
8324 normally used on Unix, so you cannot use it if you need to call
8325 libraries compiled with the Unix compiler.
8327 Also, you must provide function prototypes for all functions that
8328 take variable numbers of arguments (including @code{printf});
8329 otherwise incorrect code will be generated for calls to those
8332 In addition, seriously incorrect code will result if you call a
8333 function with too many arguments. (Normally, extra arguments are
8334 harmlessly ignored.)
8336 @item -mregparm=@var{num}
8338 Control how many registers are used to pass integer arguments. By
8339 default, no registers are used to pass arguments, and at most 3
8340 registers can be used. You can control this behavior for a specific
8341 function by using the function attribute @samp{regparm}.
8342 @xref{Function Attributes}.
8344 @strong{Warning:} if you use this switch, and
8345 @var{num} is nonzero, then you must build all modules with the same
8346 value, including any libraries. This includes the system libraries and
8349 @item -mpreferred-stack-boundary=@var{num}
8350 @opindex mpreferred-stack-boundary
8351 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8352 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8353 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8354 size (@option{-Os}), in which case the default is the minimum correct
8355 alignment (4 bytes for x86, and 8 bytes for x86-64).
8357 On Pentium and PentiumPro, @code{double} and @code{long double} values
8358 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8359 suffer significant run time performance penalties. On Pentium III, the
8360 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8361 penalties if it is not 16 byte aligned.
8363 To ensure proper alignment of this values on the stack, the stack boundary
8364 must be as aligned as that required by any value stored on the stack.
8365 Further, every function must be generated such that it keeps the stack
8366 aligned. Thus calling a function compiled with a higher preferred
8367 stack boundary from a function compiled with a lower preferred stack
8368 boundary will most likely misalign the stack. It is recommended that
8369 libraries that use callbacks always use the default setting.
8371 This extra alignment does consume extra stack space, and generally
8372 increases code size. Code that is sensitive to stack space usage, such
8373 as embedded systems and operating system kernels, may want to reduce the
8374 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8392 These switches enable or disable the use of built-in functions that allow
8393 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
8395 @xref{X86 Built-in Functions}, for details of the functions enabled
8396 and disabled by these switches.
8398 To have SSE/SSE2 instructions generated automatically from floating-point
8399 code, see @option{-mfpmath=sse}.
8402 @itemx -mno-push-args
8404 @opindex mno-push-args
8405 Use PUSH operations to store outgoing parameters. This method is shorter
8406 and usually equally fast as method using SUB/MOV operations and is enabled
8407 by default. In some cases disabling it may improve performance because of
8408 improved scheduling and reduced dependencies.
8410 @item -maccumulate-outgoing-args
8411 @opindex maccumulate-outgoing-args
8412 If enabled, the maximum amount of space required for outgoing arguments will be
8413 computed in the function prologue. This is faster on most modern CPUs
8414 because of reduced dependencies, improved scheduling and reduced stack usage
8415 when preferred stack boundary is not equal to 2. The drawback is a notable
8416 increase in code size. This switch implies @option{-mno-push-args}.
8420 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8421 on thread-safe exception handling must compile and link all code with the
8422 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8423 @option{-D_MT}; when linking, it links in a special thread helper library
8424 @option{-lmingwthrd} which cleans up per thread exception handling data.
8426 @item -mno-align-stringops
8427 @opindex mno-align-stringops
8428 Do not align destination of inlined string operations. This switch reduces
8429 code size and improves performance in case the destination is already aligned,
8430 but gcc don't know about it.
8432 @item -minline-all-stringops
8433 @opindex minline-all-stringops
8434 By default GCC inlines string operations only when destination is known to be
8435 aligned at least to 4 byte boundary. This enables more inlining, increase code
8436 size, but may improve performance of code that depends on fast memcpy, strlen
8437 and memset for short lengths.
8439 @item -momit-leaf-frame-pointer
8440 @opindex momit-leaf-frame-pointer
8441 Don't keep the frame pointer in a register for leaf functions. This
8442 avoids the instructions to save, set up and restore frame pointers and
8443 makes an extra register available in leaf functions. The option
8444 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8445 which might make debugging harder.
8447 @item -mtls-direct-seg-refs
8448 @itemx -mno-tls-direct-seg-refs
8449 @opindex mtls-direct-seg-refs
8450 Controls whether TLS variables may be accessed with offsets from the
8451 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8452 or whether the thread base pointer must be added. Whether or not this
8453 is legal depends on the operating system, and whether it maps the
8454 segment to cover the entire TLS area.
8456 For systems that use GNU libc, the default is on.
8459 These @samp{-m} switches are supported in addition to the above
8460 on AMD x86-64 processors in 64-bit environments.
8467 Generate code for a 32-bit or 64-bit environment.
8468 The 32-bit environment sets int, long and pointer to 32 bits and
8469 generates code that runs on any i386 system.
8470 The 64-bit environment sets int to 32 bits and long and pointer
8471 to 64 bits and generates code for AMD's x86-64 architecture.
8474 @opindex no-red-zone
8475 Do not use a so called red zone for x86-64 code. The red zone is mandated
8476 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8477 stack pointer that will not be modified by signal or interrupt handlers
8478 and therefore can be used for temporary data without adjusting the stack
8479 pointer. The flag @option{-mno-red-zone} disables this red zone.
8481 @item -mcmodel=small
8482 @opindex mcmodel=small
8483 Generate code for the small code model: the program and its symbols must
8484 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8485 Programs can be statically or dynamically linked. This is the default
8488 @item -mcmodel=kernel
8489 @opindex mcmodel=kernel
8490 Generate code for the kernel code model. The kernel runs in the
8491 negative 2 GB of the address space.
8492 This model has to be used for Linux kernel code.
8494 @item -mcmodel=medium
8495 @opindex mcmodel=medium
8496 Generate code for the medium model: The program is linked in the lower 2
8497 GB of the address space but symbols can be located anywhere in the
8498 address space. Programs can be statically or dynamically linked, but
8499 building of shared libraries are not supported with the medium model.
8501 @item -mcmodel=large
8502 @opindex mcmodel=large
8503 Generate code for the large model: This model makes no assumptions
8504 about addresses and sizes of sections. Currently GCC does not implement
8509 @subsection HPPA Options
8510 @cindex HPPA Options
8512 These @samp{-m} options are defined for the HPPA family of computers:
8515 @item -march=@var{architecture-type}
8517 Generate code for the specified architecture. The choices for
8518 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8519 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8520 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8521 architecture option for your machine. Code compiled for lower numbered
8522 architectures will run on higher numbered architectures, but not the
8525 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8526 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8530 @itemx -mpa-risc-1-1
8531 @itemx -mpa-risc-2-0
8532 @opindex mpa-risc-1-0
8533 @opindex mpa-risc-1-1
8534 @opindex mpa-risc-2-0
8535 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8538 @opindex mbig-switch
8539 Generate code suitable for big switch tables. Use this option only if
8540 the assembler/linker complain about out of range branches within a switch
8543 @item -mjump-in-delay
8544 @opindex mjump-in-delay
8545 Fill delay slots of function calls with unconditional jump instructions
8546 by modifying the return pointer for the function call to be the target
8547 of the conditional jump.
8549 @item -mdisable-fpregs
8550 @opindex mdisable-fpregs
8551 Prevent floating point registers from being used in any manner. This is
8552 necessary for compiling kernels which perform lazy context switching of
8553 floating point registers. If you use this option and attempt to perform
8554 floating point operations, the compiler will abort.
8556 @item -mdisable-indexing
8557 @opindex mdisable-indexing
8558 Prevent the compiler from using indexing address modes. This avoids some
8559 rather obscure problems when compiling MIG generated code under MACH@.
8561 @item -mno-space-regs
8562 @opindex mno-space-regs
8563 Generate code that assumes the target has no space registers. This allows
8564 GCC to generate faster indirect calls and use unscaled index address modes.
8566 Such code is suitable for level 0 PA systems and kernels.
8568 @item -mfast-indirect-calls
8569 @opindex mfast-indirect-calls
8570 Generate code that assumes calls never cross space boundaries. This
8571 allows GCC to emit code which performs faster indirect calls.
8573 This option will not work in the presence of shared libraries or nested
8576 @item -mlong-load-store
8577 @opindex mlong-load-store
8578 Generate 3-instruction load and store sequences as sometimes required by
8579 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8582 @item -mportable-runtime
8583 @opindex mportable-runtime
8584 Use the portable calling conventions proposed by HP for ELF systems.
8588 Enable the use of assembler directives only GAS understands.
8590 @item -mschedule=@var{cpu-type}
8592 Schedule code according to the constraints for the machine type
8593 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8594 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8595 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8596 proper scheduling option for your machine. The default scheduling is
8600 @opindex mlinker-opt
8601 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8602 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8603 linkers in which they give bogus error messages when linking some programs.
8606 @opindex msoft-float
8607 Generate output containing library calls for floating point.
8608 @strong{Warning:} the requisite libraries are not available for all HPPA
8609 targets. Normally the facilities of the machine's usual C compiler are
8610 used, but this cannot be done directly in cross-compilation. You must make
8611 your own arrangements to provide suitable library functions for
8612 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8613 does provide software floating point support.
8615 @option{-msoft-float} changes the calling convention in the output file;
8616 therefore, it is only useful if you compile @emph{all} of a program with
8617 this option. In particular, you need to compile @file{libgcc.a}, the
8618 library that comes with GCC, with @option{-msoft-float} in order for
8623 Generate the predefine, @code{_SIO}, for server IO. The default is
8624 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8625 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8626 options are available under HP-UX and HI-UX.
8630 Use GNU ld specific options. This passes @option{-shared} to ld when
8631 building a shared library. It is the default when GCC is configured,
8632 explicitly or implicitly, with the GNU linker. This option does not
8633 have any affect on which ld is called, it only changes what parameters
8634 are passed to that ld. The ld that is called is determined by the
8635 @option{--with-ld} configure option, gcc's program search path, and
8636 finally by the user's @env{PATH}. The linker used by GCC can be printed
8637 using @samp{which `gcc -print-prog-name=ld`}.
8641 Use HP ld specific options. This passes @option{-b} to ld when building
8642 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8643 links. It is the default when GCC is configured, explicitly or
8644 implicitly, with the HP linker. This option does not have any affect on
8645 which ld is called, it only changes what parameters are passed to that
8646 ld. The ld that is called is determined by the @option{--with-ld}
8647 configure option, gcc's program search path, and finally by the user's
8648 @env{PATH}. The linker used by GCC can be printed using @samp{which
8649 `gcc -print-prog-name=ld`}.
8652 @opindex mno-long-calls
8653 Generate code that uses long call sequences. This ensures that a call
8654 is always able to reach linker generated stubs. The default is to generate
8655 long calls only when the distance from the call site to the beginning
8656 of the function or translation unit, as the case may be, exceeds a
8657 predefined limit set by the branch type being used. The limits for
8658 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8659 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8662 Distances are measured from the beginning of functions when using the
8663 @option{-ffunction-sections} option, or when using the @option{-mgas}
8664 and @option{-mno-portable-runtime} options together under HP-UX with
8667 It is normally not desirable to use this option as it will degrade
8668 performance. However, it may be useful in large applications,
8669 particularly when partial linking is used to build the application.
8671 The types of long calls used depends on the capabilities of the
8672 assembler and linker, and the type of code being generated. The
8673 impact on systems that support long absolute calls, and long pic
8674 symbol-difference or pc-relative calls should be relatively small.
8675 However, an indirect call is used on 32-bit ELF systems in pic code
8676 and it is quite long.
8680 Suppress the generation of link options to search libdld.sl when the
8681 @option{-static} option is specified on HP-UX 10 and later.
8685 The HP-UX implementation of setlocale in libc has a dependency on
8686 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8687 when the @option{-static} option is specified, special link options
8688 are needed to resolve this dependency.
8690 On HP-UX 10 and later, the GCC driver adds the necessary options to
8691 link with libdld.sl when the @option{-static} option is specified.
8692 This causes the resulting binary to be dynamic. On the 64-bit port,
8693 the linkers generate dynamic binaries by default in any case. The
8694 @option{-nolibdld} option can be used to prevent the GCC driver from
8695 adding these link options.
8699 Add support for multithreading with the @dfn{dce thread} library
8700 under HP-UX. This option sets flags for both the preprocessor and
8704 @node DEC Alpha Options
8705 @subsection DEC Alpha Options
8707 These @samp{-m} options are defined for the DEC Alpha implementations:
8710 @item -mno-soft-float
8712 @opindex mno-soft-float
8713 @opindex msoft-float
8714 Use (do not use) the hardware floating-point instructions for
8715 floating-point operations. When @option{-msoft-float} is specified,
8716 functions in @file{libgcc.a} will be used to perform floating-point
8717 operations. Unless they are replaced by routines that emulate the
8718 floating-point operations, or compiled in such a way as to call such
8719 emulations routines, these routines will issue floating-point
8720 operations. If you are compiling for an Alpha without floating-point
8721 operations, you must ensure that the library is built so as not to call
8724 Note that Alpha implementations without floating-point operations are
8725 required to have floating-point registers.
8730 @opindex mno-fp-regs
8731 Generate code that uses (does not use) the floating-point register set.
8732 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8733 register set is not used, floating point operands are passed in integer
8734 registers as if they were integers and floating-point results are passed
8735 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8736 so any function with a floating-point argument or return value called by code
8737 compiled with @option{-mno-fp-regs} must also be compiled with that
8740 A typical use of this option is building a kernel that does not use,
8741 and hence need not save and restore, any floating-point registers.
8745 The Alpha architecture implements floating-point hardware optimized for
8746 maximum performance. It is mostly compliant with the IEEE floating
8747 point standard. However, for full compliance, software assistance is
8748 required. This option generates code fully IEEE compliant code
8749 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8750 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8751 defined during compilation. The resulting code is less efficient but is
8752 able to correctly support denormalized numbers and exceptional IEEE
8753 values such as not-a-number and plus/minus infinity. Other Alpha
8754 compilers call this option @option{-ieee_with_no_inexact}.
8756 @item -mieee-with-inexact
8757 @opindex mieee-with-inexact
8758 This is like @option{-mieee} except the generated code also maintains
8759 the IEEE @var{inexact-flag}. Turning on this option causes the
8760 generated code to implement fully-compliant IEEE math. In addition to
8761 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8762 macro. On some Alpha implementations the resulting code may execute
8763 significantly slower than the code generated by default. Since there is
8764 very little code that depends on the @var{inexact-flag}, you should
8765 normally not specify this option. Other Alpha compilers call this
8766 option @option{-ieee_with_inexact}.
8768 @item -mfp-trap-mode=@var{trap-mode}
8769 @opindex mfp-trap-mode
8770 This option controls what floating-point related traps are enabled.
8771 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8772 The trap mode can be set to one of four values:
8776 This is the default (normal) setting. The only traps that are enabled
8777 are the ones that cannot be disabled in software (e.g., division by zero
8781 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8785 Like @samp{su}, but the instructions are marked to be safe for software
8786 completion (see Alpha architecture manual for details).
8789 Like @samp{su}, but inexact traps are enabled as well.
8792 @item -mfp-rounding-mode=@var{rounding-mode}
8793 @opindex mfp-rounding-mode
8794 Selects the IEEE rounding mode. Other Alpha compilers call this option
8795 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8800 Normal IEEE rounding mode. Floating point numbers are rounded towards
8801 the nearest machine number or towards the even machine number in case
8805 Round towards minus infinity.
8808 Chopped rounding mode. Floating point numbers are rounded towards zero.
8811 Dynamic rounding mode. A field in the floating point control register
8812 (@var{fpcr}, see Alpha architecture reference manual) controls the
8813 rounding mode in effect. The C library initializes this register for
8814 rounding towards plus infinity. Thus, unless your program modifies the
8815 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8818 @item -mtrap-precision=@var{trap-precision}
8819 @opindex mtrap-precision
8820 In the Alpha architecture, floating point traps are imprecise. This
8821 means without software assistance it is impossible to recover from a
8822 floating trap and program execution normally needs to be terminated.
8823 GCC can generate code that can assist operating system trap handlers
8824 in determining the exact location that caused a floating point trap.
8825 Depending on the requirements of an application, different levels of
8826 precisions can be selected:
8830 Program precision. This option is the default and means a trap handler
8831 can only identify which program caused a floating point exception.
8834 Function precision. The trap handler can determine the function that
8835 caused a floating point exception.
8838 Instruction precision. The trap handler can determine the exact
8839 instruction that caused a floating point exception.
8842 Other Alpha compilers provide the equivalent options called
8843 @option{-scope_safe} and @option{-resumption_safe}.
8845 @item -mieee-conformant
8846 @opindex mieee-conformant
8847 This option marks the generated code as IEEE conformant. You must not
8848 use this option unless you also specify @option{-mtrap-precision=i} and either
8849 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8850 is to emit the line @samp{.eflag 48} in the function prologue of the
8851 generated assembly file. Under DEC Unix, this has the effect that
8852 IEEE-conformant math library routines will be linked in.
8854 @item -mbuild-constants
8855 @opindex mbuild-constants
8856 Normally GCC examines a 32- or 64-bit integer constant to
8857 see if it can construct it from smaller constants in two or three
8858 instructions. If it cannot, it will output the constant as a literal and
8859 generate code to load it from the data segment at runtime.
8861 Use this option to require GCC to construct @emph{all} integer constants
8862 using code, even if it takes more instructions (the maximum is six).
8864 You would typically use this option to build a shared library dynamic
8865 loader. Itself a shared library, it must relocate itself in memory
8866 before it can find the variables and constants in its own data segment.
8872 Select whether to generate code to be assembled by the vendor-supplied
8873 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8891 Indicate whether GCC should generate code to use the optional BWX,
8892 CIX, FIX and MAX instruction sets. The default is to use the instruction
8893 sets supported by the CPU type specified via @option{-mcpu=} option or that
8894 of the CPU on which GCC was built if none was specified.
8899 @opindex mfloat-ieee
8900 Generate code that uses (does not use) VAX F and G floating point
8901 arithmetic instead of IEEE single and double precision.
8903 @item -mexplicit-relocs
8904 @itemx -mno-explicit-relocs
8905 @opindex mexplicit-relocs
8906 @opindex mno-explicit-relocs
8907 Older Alpha assemblers provided no way to generate symbol relocations
8908 except via assembler macros. Use of these macros does not allow
8909 optimal instruction scheduling. GNU binutils as of version 2.12
8910 supports a new syntax that allows the compiler to explicitly mark
8911 which relocations should apply to which instructions. This option
8912 is mostly useful for debugging, as GCC detects the capabilities of
8913 the assembler when it is built and sets the default accordingly.
8917 @opindex msmall-data
8918 @opindex mlarge-data
8919 When @option{-mexplicit-relocs} is in effect, static data is
8920 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8921 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8922 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8923 16-bit relocations off of the @code{$gp} register. This limits the
8924 size of the small data area to 64KB, but allows the variables to be
8925 directly accessed via a single instruction.
8927 The default is @option{-mlarge-data}. With this option the data area
8928 is limited to just below 2GB. Programs that require more than 2GB of
8929 data must use @code{malloc} or @code{mmap} to allocate the data in the
8930 heap instead of in the program's data segment.
8932 When generating code for shared libraries, @option{-fpic} implies
8933 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8937 @opindex msmall-text
8938 @opindex mlarge-text
8939 When @option{-msmall-text} is used, the compiler assumes that the
8940 code of the entire program (or shared library) fits in 4MB, and is
8941 thus reachable with a branch instruction. When @option{-msmall-data}
8942 is used, the compiler can assume that all local symbols share the
8943 same @code{$gp} value, and thus reduce the number of instructions
8944 required for a function call from 4 to 1.
8946 The default is @option{-mlarge-text}.
8948 @item -mcpu=@var{cpu_type}
8950 Set the instruction set and instruction scheduling parameters for
8951 machine type @var{cpu_type}. You can specify either the @samp{EV}
8952 style name or the corresponding chip number. GCC supports scheduling
8953 parameters for the EV4, EV5 and EV6 family of processors and will
8954 choose the default values for the instruction set from the processor
8955 you specify. If you do not specify a processor type, GCC will default
8956 to the processor on which the compiler was built.
8958 Supported values for @var{cpu_type} are
8964 Schedules as an EV4 and has no instruction set extensions.
8968 Schedules as an EV5 and has no instruction set extensions.
8972 Schedules as an EV5 and supports the BWX extension.
8977 Schedules as an EV5 and supports the BWX and MAX extensions.
8981 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8985 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8988 @item -mtune=@var{cpu_type}
8990 Set only the instruction scheduling parameters for machine type
8991 @var{cpu_type}. The instruction set is not changed.
8993 @item -mmemory-latency=@var{time}
8994 @opindex mmemory-latency
8995 Sets the latency the scheduler should assume for typical memory
8996 references as seen by the application. This number is highly
8997 dependent on the memory access patterns used by the application
8998 and the size of the external cache on the machine.
9000 Valid options for @var{time} are
9004 A decimal number representing clock cycles.
9010 The compiler contains estimates of the number of clock cycles for
9011 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9012 (also called Dcache, Scache, and Bcache), as well as to main memory.
9013 Note that L3 is only valid for EV5.
9018 @node DEC Alpha/VMS Options
9019 @subsection DEC Alpha/VMS Options
9021 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9024 @item -mvms-return-codes
9025 @opindex mvms-return-codes
9026 Return VMS condition codes from main. The default is to return POSIX
9027 style condition (e.g.@ error) codes.
9030 @node H8/300 Options
9031 @subsection H8/300 Options
9033 These @samp{-m} options are defined for the H8/300 implementations:
9038 Shorten some address references at link time, when possible; uses the
9039 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9040 ld, Using ld}, for a fuller description.
9044 Generate code for the H8/300H@.
9048 Generate code for the H8S@.
9052 Generate code for the H8S and H8/300H in the normal mode. This switch
9053 must be used either with -mh or -ms.
9057 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9061 Make @code{int} data 32 bits by default.
9065 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9066 The default for the H8/300H and H8S is to align longs and floats on 4
9068 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9069 This option has no effect on the H8/300.
9073 @subsection SH Options
9075 These @samp{-m} options are defined for the SH implementations:
9080 Generate code for the SH1.
9084 Generate code for the SH2.
9087 Generate code for the SH2e.
9091 Generate code for the SH3.
9095 Generate code for the SH3e.
9099 Generate code for the SH4 without a floating-point unit.
9101 @item -m4-single-only
9102 @opindex m4-single-only
9103 Generate code for the SH4 with a floating-point unit that only
9104 supports single-precision arithmetic.
9108 Generate code for the SH4 assuming the floating-point unit is in
9109 single-precision mode by default.
9113 Generate code for the SH4.
9117 Compile code for the processor in big endian mode.
9121 Compile code for the processor in little endian mode.
9125 Align doubles at 64-bit boundaries. Note that this changes the calling
9126 conventions, and thus some functions from the standard C library will
9127 not work unless you recompile it first with @option{-mdalign}.
9131 Shorten some address references at link time, when possible; uses the
9132 linker option @option{-relax}.
9136 Use 32-bit offsets in @code{switch} tables. The default is to use
9141 Enable the use of the instruction @code{fmovd}.
9145 Comply with the calling conventions defined by Renesas.
9149 Mark the @code{MAC} register as call-clobbered, even if
9150 @option{-mhitachi} is given.
9154 Increase IEEE-compliance of floating-point code.
9158 Dump instruction size and location in the assembly code.
9162 This option is deprecated. It pads structures to multiple of 4 bytes,
9163 which is incompatible with the SH ABI@.
9167 Optimize for space instead of speed. Implied by @option{-Os}.
9171 When generating position-independent code, emit function calls using
9172 the Global Offset Table instead of the Procedure Linkage Table.
9176 Generate a library function call to invalidate instruction cache
9177 entries, after fixing up a trampoline. This library function call
9178 doesn't assume it can write to the whole memory address space. This
9179 is the default when the target is @code{sh-*-linux*}.
9182 @node System V Options
9183 @subsection Options for System V
9185 These additional options are available on System V Release 4 for
9186 compatibility with other compilers on those systems:
9191 Create a shared object.
9192 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9196 Identify the versions of each tool used by the compiler, in a
9197 @code{.ident} assembler directive in the output.
9201 Refrain from adding @code{.ident} directives to the output file (this is
9204 @item -YP,@var{dirs}
9206 Search the directories @var{dirs}, and no others, for libraries
9207 specified with @option{-l}.
9211 Look in the directory @var{dir} to find the M4 preprocessor.
9212 The assembler uses this option.
9213 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9214 @c the generic assembler that comes with Solaris takes just -Ym.
9217 @node TMS320C3x/C4x Options
9218 @subsection TMS320C3x/C4x Options
9219 @cindex TMS320C3x/C4x Options
9221 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9225 @item -mcpu=@var{cpu_type}
9227 Set the instruction set, register set, and instruction scheduling
9228 parameters for machine type @var{cpu_type}. Supported values for
9229 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9230 @samp{c44}. The default is @samp{c40} to generate code for the
9235 @itemx -msmall-memory
9237 @opindex mbig-memory
9239 @opindex msmall-memory
9241 Generates code for the big or small memory model. The small memory
9242 model assumed that all data fits into one 64K word page. At run-time
9243 the data page (DP) register must be set to point to the 64K page
9244 containing the .bss and .data program sections. The big memory model is
9245 the default and requires reloading of the DP register for every direct
9252 Allow (disallow) allocation of general integer operands into the block
9259 Enable (disable) generation of code using decrement and branch,
9260 DBcond(D), instructions. This is enabled by default for the C4x. To be
9261 on the safe side, this is disabled for the C3x, since the maximum
9262 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9263 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9264 that it can utilize the decrement and branch instruction, but will give
9265 up if there is more than one memory reference in the loop. Thus a loop
9266 where the loop counter is decremented can generate slightly more
9267 efficient code, in cases where the RPTB instruction cannot be utilized.
9269 @item -mdp-isr-reload
9271 @opindex mdp-isr-reload
9273 Force the DP register to be saved on entry to an interrupt service
9274 routine (ISR), reloaded to point to the data section, and restored on
9275 exit from the ISR@. This should not be required unless someone has
9276 violated the small memory model by modifying the DP register, say within
9283 For the C3x use the 24-bit MPYI instruction for integer multiplies
9284 instead of a library call to guarantee 32-bit results. Note that if one
9285 of the operands is a constant, then the multiplication will be performed
9286 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9287 then squaring operations are performed inline instead of a library call.
9290 @itemx -mno-fast-fix
9292 @opindex mno-fast-fix
9293 The C3x/C4x FIX instruction to convert a floating point value to an
9294 integer value chooses the nearest integer less than or equal to the
9295 floating point value rather than to the nearest integer. Thus if the
9296 floating point number is negative, the result will be incorrectly
9297 truncated an additional code is necessary to detect and correct this
9298 case. This option can be used to disable generation of the additional
9299 code required to correct the result.
9305 Enable (disable) generation of repeat block sequences using the RPTB
9306 instruction for zero overhead looping. The RPTB construct is only used
9307 for innermost loops that do not call functions or jump across the loop
9308 boundaries. There is no advantage having nested RPTB loops due to the
9309 overhead required to save and restore the RC, RS, and RE registers.
9310 This is enabled by default with @option{-O2}.
9312 @item -mrpts=@var{count}
9316 Enable (disable) the use of the single instruction repeat instruction
9317 RPTS@. If a repeat block contains a single instruction, and the loop
9318 count can be guaranteed to be less than the value @var{count}, GCC will
9319 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9320 then a RPTS will be emitted even if the loop count cannot be determined
9321 at compile time. Note that the repeated instruction following RPTS does
9322 not have to be reloaded from memory each iteration, thus freeing up the
9323 CPU buses for operands. However, since interrupts are blocked by this
9324 instruction, it is disabled by default.
9326 @item -mloop-unsigned
9327 @itemx -mno-loop-unsigned
9328 @opindex mloop-unsigned
9329 @opindex mno-loop-unsigned
9330 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9331 is @math{2^{31} + 1} since these instructions test if the iteration count is
9332 negative to terminate the loop. If the iteration count is unsigned
9333 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9334 exceeded. This switch allows an unsigned iteration count.
9338 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9339 with. This also enforces compatibility with the API employed by the TI
9340 C3x C compiler. For example, long doubles are passed as structures
9341 rather than in floating point registers.
9347 Generate code that uses registers (stack) for passing arguments to functions.
9348 By default, arguments are passed in registers where possible rather
9349 than by pushing arguments on to the stack.
9351 @item -mparallel-insns
9352 @itemx -mno-parallel-insns
9353 @opindex mparallel-insns
9354 @opindex mno-parallel-insns
9355 Allow the generation of parallel instructions. This is enabled by
9356 default with @option{-O2}.
9358 @item -mparallel-mpy
9359 @itemx -mno-parallel-mpy
9360 @opindex mparallel-mpy
9361 @opindex mno-parallel-mpy
9362 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9363 provided @option{-mparallel-insns} is also specified. These instructions have
9364 tight register constraints which can pessimize the code generation
9370 @subsection V850 Options
9371 @cindex V850 Options
9373 These @samp{-m} options are defined for V850 implementations:
9377 @itemx -mno-long-calls
9378 @opindex mlong-calls
9379 @opindex mno-long-calls
9380 Treat all calls as being far away (near). If calls are assumed to be
9381 far away, the compiler will always load the functions address up into a
9382 register, and call indirect through the pointer.
9388 Do not optimize (do optimize) basic blocks that use the same index
9389 pointer 4 or more times to copy pointer into the @code{ep} register, and
9390 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9391 option is on by default if you optimize.
9393 @item -mno-prolog-function
9394 @itemx -mprolog-function
9395 @opindex mno-prolog-function
9396 @opindex mprolog-function
9397 Do not use (do use) external functions to save and restore registers
9398 at the prologue and epilogue of a function. The external functions
9399 are slower, but use less code space if more than one function saves
9400 the same number of registers. The @option{-mprolog-function} option
9401 is on by default if you optimize.
9405 Try to make the code as small as possible. At present, this just turns
9406 on the @option{-mep} and @option{-mprolog-function} options.
9410 Put static or global variables whose size is @var{n} bytes or less into
9411 the tiny data area that register @code{ep} points to. The tiny data
9412 area can hold up to 256 bytes in total (128 bytes for byte references).
9416 Put static or global variables whose size is @var{n} bytes or less into
9417 the small data area that register @code{gp} points to. The small data
9418 area can hold up to 64 kilobytes.
9422 Put static or global variables whose size is @var{n} bytes or less into
9423 the first 32 kilobytes of memory.
9427 Specify that the target processor is the V850.
9430 @opindex mbig-switch
9431 Generate code suitable for big switch tables. Use this option only if
9432 the assembler/linker complain about out of range branches within a switch
9437 This option will cause r2 and r5 to be used in the code generated by
9438 the compiler. This setting is the default.
9441 @opindex mno-app-regs
9442 This option will cause r2 and r5 to be treated as fixed registers.
9446 Specify that the target processor is the V850E1. The preprocessor
9447 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
9448 this option is used.
9452 Specify that the target processor is the V850E. The preprocessor
9453 constant @samp{__v850e__} will be defined if this option is used.
9455 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
9456 are defined then a default target processor will be chosen and the
9457 relevant @samp{__v850*__} preprocessor constant will be defined.
9459 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9460 defined, regardless of which processor variant is the target.
9462 @item -mdisable-callt
9463 @opindex mdisable-callt
9464 This option will suppress generation of the CALLT instruction for the
9465 v850e and v850e1 flavors of the v850 architecture. The default is
9466 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9471 @subsection ARC Options
9474 These options are defined for ARC implementations:
9479 Compile code for little endian mode. This is the default.
9483 Compile code for big endian mode.
9486 @opindex mmangle-cpu
9487 Prepend the name of the cpu to all public symbol names.
9488 In multiple-processor systems, there are many ARC variants with different
9489 instruction and register set characteristics. This flag prevents code
9490 compiled for one cpu to be linked with code compiled for another.
9491 No facility exists for handling variants that are ``almost identical''.
9492 This is an all or nothing option.
9494 @item -mcpu=@var{cpu}
9496 Compile code for ARC variant @var{cpu}.
9497 Which variants are supported depend on the configuration.
9498 All variants support @option{-mcpu=base}, this is the default.
9500 @item -mtext=@var{text-section}
9501 @itemx -mdata=@var{data-section}
9502 @itemx -mrodata=@var{readonly-data-section}
9506 Put functions, data, and readonly data in @var{text-section},
9507 @var{data-section}, and @var{readonly-data-section} respectively
9508 by default. This can be overridden with the @code{section} attribute.
9509 @xref{Variable Attributes}.
9514 @subsection NS32K Options
9515 @cindex NS32K options
9517 These are the @samp{-m} options defined for the 32000 series. The default
9518 values for these options depends on which style of 32000 was selected when
9519 the compiler was configured; the defaults for the most common choices are
9527 Generate output for a 32032. This is the default
9528 when the compiler is configured for 32032 and 32016 based systems.
9534 Generate output for a 32332. This is the default
9535 when the compiler is configured for 32332-based systems.
9541 Generate output for a 32532. This is the default
9542 when the compiler is configured for 32532-based systems.
9546 Generate output containing 32081 instructions for floating point.
9547 This is the default for all systems.
9551 Generate output containing 32381 instructions for floating point. This
9552 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9553 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9557 Try and generate multiply-add floating point instructions @code{polyF}
9558 and @code{dotF}. This option is only available if the @option{-m32381}
9559 option is in effect. Using these instructions requires changes to
9560 register allocation which generally has a negative impact on
9561 performance. This option should only be enabled when compiling code
9562 particularly likely to make heavy use of multiply-add instructions.
9565 @opindex mnomulti-add
9566 Do not try and generate multiply-add floating point instructions
9567 @code{polyF} and @code{dotF}. This is the default on all platforms.
9570 @opindex msoft-float
9571 Generate output containing library calls for floating point.
9572 @strong{Warning:} the requisite libraries may not be available.
9574 @item -mieee-compare
9575 @itemx -mno-ieee-compare
9576 @opindex mieee-compare
9577 @opindex mno-ieee-compare
9578 Control whether or not the compiler uses IEEE floating point
9579 comparisons. These handle correctly the case where the result of a
9580 comparison is unordered.
9581 @strong{Warning:} the requisite kernel support may not be available.
9584 @opindex mnobitfield
9585 Do not use the bit-field instructions. On some machines it is faster to
9586 use shifting and masking operations. This is the default for the pc532.
9590 Do use the bit-field instructions. This is the default for all platforms
9595 Use a different function-calling convention, in which functions
9596 that take a fixed number of arguments return pop their
9597 arguments on return with the @code{ret} instruction.
9599 This calling convention is incompatible with the one normally
9600 used on Unix, so you cannot use it if you need to call libraries
9601 compiled with the Unix compiler.
9603 Also, you must provide function prototypes for all functions that
9604 take variable numbers of arguments (including @code{printf});
9605 otherwise incorrect code will be generated for calls to those
9608 In addition, seriously incorrect code will result if you call a
9609 function with too many arguments. (Normally, extra arguments are
9610 harmlessly ignored.)
9612 This option takes its name from the 680x0 @code{rtd} instruction.
9617 Use a different function-calling convention where the first two arguments
9618 are passed in registers.
9620 This calling convention is incompatible with the one normally
9621 used on Unix, so you cannot use it if you need to call libraries
9622 compiled with the Unix compiler.
9625 @opindex mnoregparam
9626 Do not pass any arguments in registers. This is the default for all
9631 It is OK to use the sb as an index register which is always loaded with
9632 zero. This is the default for the pc532-netbsd target.
9636 The sb register is not available for use or has not been initialized to
9637 zero by the run time system. This is the default for all targets except
9638 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9639 @option{-fpic} is set.
9643 Many ns32000 series addressing modes use displacements of up to 512MB@.
9644 If an address is above 512MB then displacements from zero can not be used.
9645 This option causes code to be generated which can be loaded above 512MB@.
9646 This may be useful for operating systems or ROM code.
9650 Assume code will be loaded in the first 512MB of virtual address space.
9651 This is the default for all platforms.
9657 @subsection AVR Options
9660 These options are defined for AVR implementations:
9663 @item -mmcu=@var{mcu}
9665 Specify ATMEL AVR instruction set or MCU type.
9667 Instruction set avr1 is for the minimal AVR core, not supported by the C
9668 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9669 attiny11, attiny12, attiny15, attiny28).
9671 Instruction set avr2 (default) is for the classic AVR core with up to
9672 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9673 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9674 at90c8534, at90s8535).
9676 Instruction set avr3 is for the classic AVR core with up to 128K program
9677 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9679 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9680 memory space (MCU types: atmega8, atmega83, atmega85).
9682 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9683 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9684 atmega64, atmega128, at43usb355, at94k).
9688 Output instruction sizes to the asm file.
9690 @item -minit-stack=@var{N}
9691 @opindex minit-stack
9692 Specify the initial stack address, which may be a symbol or numeric value,
9693 @samp{__stack} is the default.
9695 @item -mno-interrupts
9696 @opindex mno-interrupts
9697 Generated code is not compatible with hardware interrupts.
9698 Code size will be smaller.
9700 @item -mcall-prologues
9701 @opindex mcall-prologues
9702 Functions prologues/epilogues expanded as call to appropriate
9703 subroutines. Code size will be smaller.
9705 @item -mno-tablejump
9706 @opindex mno-tablejump
9707 Do not generate tablejump insns which sometimes increase code size.
9710 @opindex mtiny-stack
9711 Change only the low 8 bits of the stack pointer.
9715 @subsection MCore Options
9716 @cindex MCore options
9718 These are the @samp{-m} options defined for the Motorola M*Core
9726 @opindex mno-hardlit
9727 Inline constants into the code stream if it can be done in two
9728 instructions or less.
9734 Use the divide instruction. (Enabled by default).
9736 @item -mrelax-immediate
9737 @itemx -mno-relax-immediate
9738 @opindex mrelax-immediate
9739 @opindex mno-relax-immediate
9740 Allow arbitrary sized immediates in bit operations.
9742 @item -mwide-bitfields
9743 @itemx -mno-wide-bitfields
9744 @opindex mwide-bitfields
9745 @opindex mno-wide-bitfields
9746 Always treat bit-fields as int-sized.
9748 @item -m4byte-functions
9749 @itemx -mno-4byte-functions
9750 @opindex m4byte-functions
9751 @opindex mno-4byte-functions
9752 Force all functions to be aligned to a four byte boundary.
9754 @item -mcallgraph-data
9755 @itemx -mno-callgraph-data
9756 @opindex mcallgraph-data
9757 @opindex mno-callgraph-data
9758 Emit callgraph information.
9761 @itemx -mno-slow-bytes
9762 @opindex mslow-bytes
9763 @opindex mno-slow-bytes
9764 Prefer word access when reading byte quantities.
9766 @item -mlittle-endian
9768 @opindex mlittle-endian
9769 @opindex mbig-endian
9770 Generate code for a little endian target.
9776 Generate code for the 210 processor.
9780 @subsection IA-64 Options
9781 @cindex IA-64 Options
9783 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9787 @opindex mbig-endian
9788 Generate code for a big endian target. This is the default for HP-UX@.
9790 @item -mlittle-endian
9791 @opindex mlittle-endian
9792 Generate code for a little endian target. This is the default for AIX5
9799 Generate (or don't) code for the GNU assembler. This is the default.
9800 @c Also, this is the default if the configure option @option{--with-gnu-as}
9807 Generate (or don't) code for the GNU linker. This is the default.
9808 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9813 Generate code that does not use a global pointer register. The result
9814 is not position independent code, and violates the IA-64 ABI@.
9816 @item -mvolatile-asm-stop
9817 @itemx -mno-volatile-asm-stop
9818 @opindex mvolatile-asm-stop
9819 @opindex mno-volatile-asm-stop
9820 Generate (or don't) a stop bit immediately before and after volatile asm
9825 Generate code that works around Itanium B step errata.
9827 @item -mregister-names
9828 @itemx -mno-register-names
9829 @opindex mregister-names
9830 @opindex mno-register-names
9831 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9832 the stacked registers. This may make assembler output more readable.
9838 Disable (or enable) optimizations that use the small data section. This may
9839 be useful for working around optimizer bugs.
9842 @opindex mconstant-gp
9843 Generate code that uses a single constant global pointer value. This is
9844 useful when compiling kernel code.
9848 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9849 This is useful when compiling firmware code.
9851 @item -minline-float-divide-min-latency
9852 @opindex minline-float-divide-min-latency
9853 Generate code for inline divides of floating point values
9854 using the minimum latency algorithm.
9856 @item -minline-float-divide-max-throughput
9857 @opindex minline-float-divide-max-throughput
9858 Generate code for inline divides of floating point values
9859 using the maximum throughput algorithm.
9861 @item -minline-int-divide-min-latency
9862 @opindex minline-int-divide-min-latency
9863 Generate code for inline divides of integer values
9864 using the minimum latency algorithm.
9866 @item -minline-int-divide-max-throughput
9867 @opindex minline-int-divide-max-throughput
9868 Generate code for inline divides of integer values
9869 using the maximum throughput algorithm.
9871 @item -mno-dwarf2-asm
9873 @opindex mno-dwarf2-asm
9874 @opindex mdwarf2-asm
9875 Don't (or do) generate assembler code for the DWARF2 line number debugging
9876 info. This may be useful when not using the GNU assembler.
9878 @item -mfixed-range=@var{register-range}
9879 @opindex mfixed-range
9880 Generate code treating the given register range as fixed registers.
9881 A fixed register is one that the register allocator can not use. This is
9882 useful when compiling kernel code. A register range is specified as
9883 two registers separated by a dash. Multiple register ranges can be
9884 specified separated by a comma.
9886 @item -mearly-stop-bits
9887 @itemx -mno-early-stop-bits
9888 @opindex mearly-stop-bits
9889 @opindex mno-early-stop-bits
9890 Allow stop bits to be placed earlier than immediately preceding the
9891 instruction that triggered the stop bit. This can improve instruction
9892 scheduling, but does not always do so.
9895 @node S/390 and zSeries Options
9896 @subsection S/390 and zSeries Options
9897 @cindex S/390 and zSeries Options
9899 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
9904 @opindex mhard-float
9905 @opindex msoft-float
9906 Use (do not use) the hardware floating-point instructions and registers
9907 for floating-point operations. When @option{-msoft-float} is specified,
9908 functions in @file{libgcc.a} will be used to perform floating-point
9909 operations. When @option{-mhard-float} is specified, the compiler
9910 generates IEEE floating-point instructions. This is the default.
9913 @itemx -mno-backchain
9915 @opindex mno-backchain
9916 Generate (or do not generate) code which maintains an explicit
9917 backchain within the stack frame that points to the caller's frame.
9918 This may be needed to allow debugging using tools that do not understand
9919 DWARF-2 call frame information. The default is not to generate the
9923 @itemx -mno-small-exec
9924 @opindex msmall-exec
9925 @opindex mno-small-exec
9926 Generate (or do not generate) code using the @code{bras} instruction
9927 to do subroutine calls.
9928 This only works reliably if the total executable size does not
9929 exceed 64k. The default is to use the @code{basr} instruction instead,
9930 which does not have this limitation.
9936 When @option{-m31} is specified, generate code compliant to the
9937 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
9938 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
9939 particular to generate 64-bit instructions. For the @samp{s390}
9940 targets, the default is @option{-m31}, while the @samp{s390x}
9941 targets default to @option{-m64}.
9947 When @option{-mzarch} is specified, generate code using the
9948 instructions available on z/Architecture.
9949 When @option{-mesa} is specified, generate code using the
9950 instructions available on ESA/390. Note that @option{-mesa} is
9951 not possible with @option{-m64}.
9952 When generating code compliant to the GNU/Linux for S/390 ABI,
9953 the default is @option{-mesa}. When generating code compliant
9954 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
9960 Generate (or do not generate) code using the @code{mvcle} instruction
9961 to perform block moves. When @option{-mno-mvcle} is specified,
9962 use a @code{mvc} loop instead. This is the default.
9968 Print (or do not print) additional debug information when compiling.
9969 The default is to not print debug information.
9971 @item -march=@var{cpu-type}
9973 Generate code that will run on @var{cpu-type}, which is the name of a system
9974 representing a certain processor type. Possible values for
9975 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
9976 When generating code using the instructions available on z/Architecture,
9977 the default is @option{-march=z900}. Otherwise, the default is
9980 @item -mtune=@var{cpu-type}
9982 Tune to @var{cpu-type} everything applicable about the generated code,
9983 except for the ABI and the set of available instructions.
9984 The list of @var{cpu-type} values is the same as for @option{-march}.
9985 The default is the value used for @option{-march}.
9988 @itemx -mno-fused-madd
9989 @opindex mfused-madd
9990 @opindex mno-fused-madd
9991 Generate code that uses (does not use) the floating point multiply and
9992 accumulate instructions. These instructions are generated by default if
9993 hardware floating point is used.
9997 @subsection CRIS Options
9998 @cindex CRIS Options
10000 These options are defined specifically for the CRIS ports.
10003 @item -march=@var{architecture-type}
10004 @itemx -mcpu=@var{architecture-type}
10007 Generate code for the specified architecture. The choices for
10008 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10009 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
10010 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10013 @item -mtune=@var{architecture-type}
10015 Tune to @var{architecture-type} everything applicable about the generated
10016 code, except for the ABI and the set of available instructions. The
10017 choices for @var{architecture-type} are the same as for
10018 @option{-march=@var{architecture-type}}.
10020 @item -mmax-stack-frame=@var{n}
10021 @opindex mmax-stack-frame
10022 Warn when the stack frame of a function exceeds @var{n} bytes.
10024 @item -melinux-stacksize=@var{n}
10025 @opindex melinux-stacksize
10026 Only available with the @samp{cris-axis-aout} target. Arranges for
10027 indications in the program to the kernel loader that the stack of the
10028 program should be set to @var{n} bytes.
10034 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10035 @option{-march=v3} and @option{-march=v8} respectively.
10039 Enable CRIS-specific verbose debug-related information in the assembly
10040 code. This option also has the effect to turn off the @samp{#NO_APP}
10041 formatted-code indicator to the assembler at the beginning of the
10046 Do not use condition-code results from previous instruction; always emit
10047 compare and test instructions before use of condition codes.
10049 @item -mno-side-effects
10050 @opindex mno-side-effects
10051 Do not emit instructions with side-effects in addressing modes other than
10054 @item -mstack-align
10055 @itemx -mno-stack-align
10056 @itemx -mdata-align
10057 @itemx -mno-data-align
10058 @itemx -mconst-align
10059 @itemx -mno-const-align
10060 @opindex mstack-align
10061 @opindex mno-stack-align
10062 @opindex mdata-align
10063 @opindex mno-data-align
10064 @opindex mconst-align
10065 @opindex mno-const-align
10066 These options (no-options) arranges (eliminate arrangements) for the
10067 stack-frame, individual data and constants to be aligned for the maximum
10068 single data access size for the chosen CPU model. The default is to
10069 arrange for 32-bit alignment. ABI details such as structure layout are
10070 not affected by these options.
10078 Similar to the stack- data- and const-align options above, these options
10079 arrange for stack-frame, writable data and constants to all be 32-bit,
10080 16-bit or 8-bit aligned. The default is 32-bit alignment.
10082 @item -mno-prologue-epilogue
10083 @itemx -mprologue-epilogue
10084 @opindex mno-prologue-epilogue
10085 @opindex mprologue-epilogue
10086 With @option{-mno-prologue-epilogue}, the normal function prologue and
10087 epilogue that sets up the stack-frame are omitted and no return
10088 instructions or return sequences are generated in the code. Use this
10089 option only together with visual inspection of the compiled code: no
10090 warnings or errors are generated when call-saved registers must be saved,
10091 or storage for local variable needs to be allocated.
10095 @opindex mno-gotplt
10097 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10098 instruction sequences that load addresses for functions from the PLT part
10099 of the GOT rather than (traditional on other architectures) calls to the
10100 PLT. The default is @option{-mgotplt}.
10104 Legacy no-op option only recognized with the cris-axis-aout target.
10108 Legacy no-op option only recognized with the cris-axis-elf and
10109 cris-axis-linux-gnu targets.
10113 Only recognized with the cris-axis-aout target, where it selects a
10114 GNU/linux-like multilib, include files and instruction set for
10115 @option{-march=v8}.
10119 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10123 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
10124 to link with input-output functions from a simulator library. Code,
10125 initialized data and zero-initialized data are allocated consecutively.
10129 Like @option{-sim}, but pass linker options to locate initialized data at
10130 0x40000000 and zero-initialized data at 0x80000000.
10134 @subsection MMIX Options
10135 @cindex MMIX Options
10137 These options are defined for the MMIX:
10141 @itemx -mno-libfuncs
10143 @opindex mno-libfuncs
10144 Specify that intrinsic library functions are being compiled, passing all
10145 values in registers, no matter the size.
10148 @itemx -mno-epsilon
10150 @opindex mno-epsilon
10151 Generate floating-point comparison instructions that compare with respect
10152 to the @code{rE} epsilon register.
10154 @item -mabi=mmixware
10156 @opindex mabi-mmixware
10158 Generate code that passes function parameters and return values that (in
10159 the called function) are seen as registers @code{$0} and up, as opposed to
10160 the GNU ABI which uses global registers @code{$231} and up.
10162 @item -mzero-extend
10163 @itemx -mno-zero-extend
10164 @opindex mzero-extend
10165 @opindex mno-zero-extend
10166 When reading data from memory in sizes shorter than 64 bits, use (do not
10167 use) zero-extending load instructions by default, rather than
10168 sign-extending ones.
10171 @itemx -mno-knuthdiv
10173 @opindex mno-knuthdiv
10174 Make the result of a division yielding a remainder have the same sign as
10175 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10176 remainder follows the sign of the dividend. Both methods are
10177 arithmetically valid, the latter being almost exclusively used.
10179 @item -mtoplevel-symbols
10180 @itemx -mno-toplevel-symbols
10181 @opindex mtoplevel-symbols
10182 @opindex mno-toplevel-symbols
10183 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10184 code can be used with the @code{PREFIX} assembly directive.
10188 Generate an executable in the ELF format, rather than the default
10189 @samp{mmo} format used by the @command{mmix} simulator.
10191 @item -mbranch-predict
10192 @itemx -mno-branch-predict
10193 @opindex mbranch-predict
10194 @opindex mno-branch-predict
10195 Use (do not use) the probable-branch instructions, when static branch
10196 prediction indicates a probable branch.
10198 @item -mbase-addresses
10199 @itemx -mno-base-addresses
10200 @opindex mbase-addresses
10201 @opindex mno-base-addresses
10202 Generate (do not generate) code that uses @emph{base addresses}. Using a
10203 base address automatically generates a request (handled by the assembler
10204 and the linker) for a constant to be set up in a global register. The
10205 register is used for one or more base address requests within the range 0
10206 to 255 from the value held in the register. The generally leads to short
10207 and fast code, but the number of different data items that can be
10208 addressed is limited. This means that a program that uses lots of static
10209 data may require @option{-mno-base-addresses}.
10211 @item -msingle-exit
10212 @itemx -mno-single-exit
10213 @opindex msingle-exit
10214 @opindex mno-single-exit
10215 Force (do not force) generated code to have a single exit point in each
10219 @node PDP-11 Options
10220 @subsection PDP-11 Options
10221 @cindex PDP-11 Options
10223 These options are defined for the PDP-11:
10228 Use hardware FPP floating point. This is the default. (FIS floating
10229 point on the PDP-11/40 is not supported.)
10232 @opindex msoft-float
10233 Do not use hardware floating point.
10237 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10241 Return floating-point results in memory. This is the default.
10245 Generate code for a PDP-11/40.
10249 Generate code for a PDP-11/45. This is the default.
10253 Generate code for a PDP-11/10.
10255 @item -mbcopy-builtin
10256 @opindex bcopy-builtin
10257 Use inline @code{movstrhi} patterns for copying memory. This is the
10262 Do not use inline @code{movstrhi} patterns for copying memory.
10268 Use 16-bit @code{int}. This is the default.
10274 Use 32-bit @code{int}.
10277 @itemx -mno-float32
10279 @opindex mno-float32
10280 Use 64-bit @code{float}. This is the default.
10283 @itemx -mno-float64
10285 @opindex mno-float64
10286 Use 32-bit @code{float}.
10290 Use @code{abshi2} pattern. This is the default.
10294 Do not use @code{abshi2} pattern.
10296 @item -mbranch-expensive
10297 @opindex mbranch-expensive
10298 Pretend that branches are expensive. This is for experimenting with
10299 code generation only.
10301 @item -mbranch-cheap
10302 @opindex mbranch-cheap
10303 Do not pretend that branches are expensive. This is the default.
10307 Generate code for a system with split I&D.
10311 Generate code for a system without split I&D. This is the default.
10315 Use Unix assembler syntax. This is the default when configured for
10316 @samp{pdp11-*-bsd}.
10320 Use DEC assembler syntax. This is the default when configured for any
10321 PDP-11 target other than @samp{pdp11-*-bsd}.
10324 @node Xstormy16 Options
10325 @subsection Xstormy16 Options
10326 @cindex Xstormy16 Options
10328 These options are defined for Xstormy16:
10333 Choose startup files and linker script suitable for the simulator.
10337 @subsection FRV Options
10338 @cindex FRV Options
10344 Only use the first 32 general purpose registers.
10349 Use all 64 general purpose registers.
10354 Use only the first 32 floating point registers.
10359 Use all 64 floating point registers
10362 @opindex mhard-float
10364 Use hardware instructions for floating point operations.
10367 @opindex msoft-float
10369 Use library routines for floating point operations.
10374 Dynamically allocate condition code registers.
10379 Do not try to dynamically allocate condition code registers, only
10380 use @code{icc0} and @code{fcc0}.
10385 Change ABI to use double word insns.
10390 Do not use double word instructions.
10395 Use floating point double instructions.
10398 @opindex mno-double
10400 Do not use floating point double instructions.
10405 Use media instructions.
10410 Do not use media instructions.
10415 Use multiply and add/subtract instructions.
10418 @opindex mno-muladd
10420 Do not use multiply and add/subtract instructions.
10422 @item -mlibrary-pic
10423 @opindex mlibrary-pic
10425 Enable PIC support for building libraries
10430 Use only the first four media accumulator registers.
10435 Use all eight media accumulator registers.
10440 Pack VLIW instructions.
10445 Do not pack VLIW instructions.
10448 @opindex mno-eflags
10450 Do not mark ABI switches in e_flags.
10453 @opindex mcond-move
10455 Enable the use of conditional-move instructions (default).
10457 This switch is mainly for debugging the compiler and will likely be removed
10458 in a future version.
10460 @item -mno-cond-move
10461 @opindex mno-cond-move
10463 Disable the use of conditional-move instructions.
10465 This switch is mainly for debugging the compiler and will likely be removed
10466 in a future version.
10471 Enable the use of conditional set instructions (default).
10473 This switch is mainly for debugging the compiler and will likely be removed
10474 in a future version.
10479 Disable the use of conditional set instructions.
10481 This switch is mainly for debugging the compiler and will likely be removed
10482 in a future version.
10485 @opindex mcond-exec
10487 Enable the use of conditional execution (default).
10489 This switch is mainly for debugging the compiler and will likely be removed
10490 in a future version.
10492 @item -mno-cond-exec
10493 @opindex mno-cond-exec
10495 Disable the use of conditional execution.
10497 This switch is mainly for debugging the compiler and will likely be removed
10498 in a future version.
10500 @item -mvliw-branch
10501 @opindex mvliw-branch
10503 Run a pass to pack branches into VLIW instructions (default).
10505 This switch is mainly for debugging the compiler and will likely be removed
10506 in a future version.
10508 @item -mno-vliw-branch
10509 @opindex mno-vliw-branch
10511 Do not run a pass to pack branches into VLIW instructions.
10513 This switch is mainly for debugging the compiler and will likely be removed
10514 in a future version.
10516 @item -mmulti-cond-exec
10517 @opindex mmulti-cond-exec
10519 Enable optimization of @code{&&} and @code{||} in conditional execution
10522 This switch is mainly for debugging the compiler and will likely be removed
10523 in a future version.
10525 @item -mno-multi-cond-exec
10526 @opindex mno-multi-cond-exec
10528 Disable optimization of @code{&&} and @code{||} in conditional execution.
10530 This switch is mainly for debugging the compiler and will likely be removed
10531 in a future version.
10533 @item -mnested-cond-exec
10534 @opindex mnested-cond-exec
10536 Enable nested conditional execution optimizations (default).
10538 This switch is mainly for debugging the compiler and will likely be removed
10539 in a future version.
10541 @item -mno-nested-cond-exec
10542 @opindex mno-nested-cond-exec
10544 Disable nested conditional execution optimizations.
10546 This switch is mainly for debugging the compiler and will likely be removed
10547 in a future version.
10549 @item -mtomcat-stats
10550 @opindex mtomcat-stats
10552 Cause gas to print out tomcat statistics.
10554 @item -mcpu=@var{cpu}
10557 Select the processor type for which to generate code. Possible values are
10558 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
10563 @node Xtensa Options
10564 @subsection Xtensa Options
10565 @cindex Xtensa Options
10567 These options are supported for Xtensa targets:
10571 @itemx -mno-const16
10573 @opindex mno-const16
10574 Enable or disable use of @code{CONST16} instructions for loading
10575 constant values. The @code{CONST16} instruction is currently not a
10576 standard option from Tensilica. When enabled, @code{CONST16}
10577 instructions are always used in place of the standard @code{L32R}
10578 instructions. The use of @code{CONST16} is enabled by default only if
10579 the @code{L32R} instruction is not available.
10582 @itemx -mno-fused-madd
10583 @opindex mfused-madd
10584 @opindex mno-fused-madd
10585 Enable or disable use of fused multiply/add and multiply/subtract
10586 instructions in the floating-point option. This has no effect if the
10587 floating-point option is not also enabled. Disabling fused multiply/add
10588 and multiply/subtract instructions forces the compiler to use separate
10589 instructions for the multiply and add/subtract operations. This may be
10590 desirable in some cases where strict IEEE 754-compliant results are
10591 required: the fused multiply add/subtract instructions do not round the
10592 intermediate result, thereby producing results with @emph{more} bits of
10593 precision than specified by the IEEE standard. Disabling fused multiply
10594 add/subtract instructions also ensures that the program output is not
10595 sensitive to the compiler's ability to combine multiply and add/subtract
10598 @item -mtext-section-literals
10599 @itemx -mno-text-section-literals
10600 @opindex mtext-section-literals
10601 @opindex mno-text-section-literals
10602 Control the treatment of literal pools. The default is
10603 @option{-mno-text-section-literals}, which places literals in a separate
10604 section in the output file. This allows the literal pool to be placed
10605 in a data RAM/ROM, and it also allows the linker to combine literal
10606 pools from separate object files to remove redundant literals and
10607 improve code size. With @option{-mtext-section-literals}, the literals
10608 are interspersed in the text section in order to keep them as close as
10609 possible to their references. This may be necessary for large assembly
10612 @item -mtarget-align
10613 @itemx -mno-target-align
10614 @opindex mtarget-align
10615 @opindex mno-target-align
10616 When this option is enabled, GCC instructs the assembler to
10617 automatically align instructions to reduce branch penalties at the
10618 expense of some code density. The assembler attempts to widen density
10619 instructions to align branch targets and the instructions following call
10620 instructions. If there are not enough preceding safe density
10621 instructions to align a target, no widening will be performed. The
10622 default is @option{-mtarget-align}. These options do not affect the
10623 treatment of auto-aligned instructions like @code{LOOP}, which the
10624 assembler will always align, either by widening density instructions or
10625 by inserting no-op instructions.
10628 @itemx -mno-longcalls
10629 @opindex mlongcalls
10630 @opindex mno-longcalls
10631 When this option is enabled, GCC instructs the assembler to translate
10632 direct calls to indirect calls unless it can determine that the target
10633 of a direct call is in the range allowed by the call instruction. This
10634 translation typically occurs for calls to functions in other source
10635 files. Specifically, the assembler translates a direct @code{CALL}
10636 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
10637 The default is @option{-mno-longcalls}. This option should be used in
10638 programs where the call target can potentially be out of range. This
10639 option is implemented in the assembler, not the compiler, so the
10640 assembly code generated by GCC will still show direct call
10641 instructions---look at the disassembled object code to see the actual
10642 instructions. Note that the assembler will use an indirect call for
10643 every cross-file call, not just those that really will be out of range.
10646 @node Code Gen Options
10647 @section Options for Code Generation Conventions
10648 @cindex code generation conventions
10649 @cindex options, code generation
10650 @cindex run-time options
10652 These machine-independent options control the interface conventions
10653 used in code generation.
10655 Most of them have both positive and negative forms; the negative form
10656 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
10657 one of the forms is listed---the one which is not the default. You
10658 can figure out the other form by either removing @samp{no-} or adding
10662 @item -fbounds-check
10663 @opindex fbounds-check
10664 For front-ends that support it, generate additional code to check that
10665 indices used to access arrays are within the declared range. This is
10666 currently only supported by the Java and Fortran 77 front-ends, where
10667 this option defaults to true and false respectively.
10671 This option generates traps for signed overflow on addition, subtraction,
10672 multiplication operations.
10676 This option instructs the compiler to assume that signed arithmetic
10677 overflow of addition, subtraction and multiplication wraps around
10678 using twos-complement representation. This flag enables some optimizations
10679 and disables other. This option is enabled by default for the Java
10680 front-end, as required by the Java language specification.
10683 @opindex fexceptions
10684 Enable exception handling. Generates extra code needed to propagate
10685 exceptions. For some targets, this implies GCC will generate frame
10686 unwind information for all functions, which can produce significant data
10687 size overhead, although it does not affect execution. If you do not
10688 specify this option, GCC will enable it by default for languages like
10689 C++ which normally require exception handling, and disable it for
10690 languages like C that do not normally require it. However, you may need
10691 to enable this option when compiling C code that needs to interoperate
10692 properly with exception handlers written in C++. You may also wish to
10693 disable this option if you are compiling older C++ programs that don't
10694 use exception handling.
10696 @item -fnon-call-exceptions
10697 @opindex fnon-call-exceptions
10698 Generate code that allows trapping instructions to throw exceptions.
10699 Note that this requires platform-specific runtime support that does
10700 not exist everywhere. Moreover, it only allows @emph{trapping}
10701 instructions to throw exceptions, i.e.@: memory references or floating
10702 point instructions. It does not allow exceptions to be thrown from
10703 arbitrary signal handlers such as @code{SIGALRM}.
10705 @item -funwind-tables
10706 @opindex funwind-tables
10707 Similar to @option{-fexceptions}, except that it will just generate any needed
10708 static data, but will not affect the generated code in any other way.
10709 You will normally not enable this option; instead, a language processor
10710 that needs this handling would enable it on your behalf.
10712 @item -fasynchronous-unwind-tables
10713 @opindex funwind-tables
10714 Generate unwind table in dwarf2 format, if supported by target machine. The
10715 table is exact at each instruction boundary, so it can be used for stack
10716 unwinding from asynchronous events (such as debugger or garbage collector).
10718 @item -fpcc-struct-return
10719 @opindex fpcc-struct-return
10720 Return ``short'' @code{struct} and @code{union} values in memory like
10721 longer ones, rather than in registers. This convention is less
10722 efficient, but it has the advantage of allowing intercallability between
10723 GCC-compiled files and files compiled with other compilers, particularly
10724 the Portable C Compiler (pcc).
10726 The precise convention for returning structures in memory depends
10727 on the target configuration macros.
10729 Short structures and unions are those whose size and alignment match
10730 that of some integer type.
10732 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
10733 switch is not binary compatible with code compiled with the
10734 @option{-freg-struct-return} switch.
10735 Use it to conform to a non-default application binary interface.
10737 @item -freg-struct-return
10738 @opindex freg-struct-return
10739 Return @code{struct} and @code{union} values in registers when possible.
10740 This is more efficient for small structures than
10741 @option{-fpcc-struct-return}.
10743 If you specify neither @option{-fpcc-struct-return} nor
10744 @option{-freg-struct-return}, GCC defaults to whichever convention is
10745 standard for the target. If there is no standard convention, GCC
10746 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
10747 the principal compiler. In those cases, we can choose the standard, and
10748 we chose the more efficient register return alternative.
10750 @strong{Warning:} code compiled with the @option{-freg-struct-return}
10751 switch is not binary compatible with code compiled with the
10752 @option{-fpcc-struct-return} switch.
10753 Use it to conform to a non-default application binary interface.
10755 @item -fshort-enums
10756 @opindex fshort-enums
10757 Allocate to an @code{enum} type only as many bytes as it needs for the
10758 declared range of possible values. Specifically, the @code{enum} type
10759 will be equivalent to the smallest integer type which has enough room.
10761 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
10762 code that is not binary compatible with code generated without that switch.
10763 Use it to conform to a non-default application binary interface.
10765 @item -fshort-double
10766 @opindex fshort-double
10767 Use the same size for @code{double} as for @code{float}.
10769 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
10770 code that is not binary compatible with code generated without that switch.
10771 Use it to conform to a non-default application binary interface.
10773 @item -fshort-wchar
10774 @opindex fshort-wchar
10775 Override the underlying type for @samp{wchar_t} to be @samp{short
10776 unsigned int} instead of the default for the target. This option is
10777 useful for building programs to run under WINE@.
10779 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
10780 code that is not binary compatible with code generated without that switch.
10781 Use it to conform to a non-default application binary interface.
10783 @item -fshared-data
10784 @opindex fshared-data
10785 Requests that the data and non-@code{const} variables of this
10786 compilation be shared data rather than private data. The distinction
10787 makes sense only on certain operating systems, where shared data is
10788 shared between processes running the same program, while private data
10789 exists in one copy per process.
10792 @opindex fno-common
10793 In C, allocate even uninitialized global variables in the data section of the
10794 object file, rather than generating them as common blocks. This has the
10795 effect that if the same variable is declared (without @code{extern}) in
10796 two different compilations, you will get an error when you link them.
10797 The only reason this might be useful is if you wish to verify that the
10798 program will work on other systems which always work this way.
10802 Ignore the @samp{#ident} directive.
10804 @item -finhibit-size-directive
10805 @opindex finhibit-size-directive
10806 Don't output a @code{.size} assembler directive, or anything else that
10807 would cause trouble if the function is split in the middle, and the
10808 two halves are placed at locations far apart in memory. This option is
10809 used when compiling @file{crtstuff.c}; you should not need to use it
10812 @item -fverbose-asm
10813 @opindex fverbose-asm
10814 Put extra commentary information in the generated assembly code to
10815 make it more readable. This option is generally only of use to those
10816 who actually need to read the generated assembly code (perhaps while
10817 debugging the compiler itself).
10819 @option{-fno-verbose-asm}, the default, causes the
10820 extra information to be omitted and is useful when comparing two assembler
10825 @cindex global offset table
10827 Generate position-independent code (PIC) suitable for use in a shared
10828 library, if supported for the target machine. Such code accesses all
10829 constant addresses through a global offset table (GOT)@. The dynamic
10830 loader resolves the GOT entries when the program starts (the dynamic
10831 loader is not part of GCC; it is part of the operating system). If
10832 the GOT size for the linked executable exceeds a machine-specific
10833 maximum size, you get an error message from the linker indicating that
10834 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
10835 instead. (These maximums are 8k on the SPARC and 32k
10836 on the m68k and RS/6000. The 386 has no such limit.)
10838 Position-independent code requires special support, and therefore works
10839 only on certain machines. For the 386, GCC supports PIC for System V
10840 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
10841 position-independent.
10845 If supported for the target machine, emit position-independent code,
10846 suitable for dynamic linking and avoiding any limit on the size of the
10847 global offset table. This option makes a difference on the m68k
10850 Position-independent code requires special support, and therefore works
10851 only on certain machines.
10857 These options are similar to @option{-fpic} and @option{-fPIC}, but
10858 generated position independent code can be only linked into executables.
10859 Usually these options are used when @option{-pie} GCC option will be
10860 used during linking.
10862 @item -ffixed-@var{reg}
10864 Treat the register named @var{reg} as a fixed register; generated code
10865 should never refer to it (except perhaps as a stack pointer, frame
10866 pointer or in some other fixed role).
10868 @var{reg} must be the name of a register. The register names accepted
10869 are machine-specific and are defined in the @code{REGISTER_NAMES}
10870 macro in the machine description macro file.
10872 This flag does not have a negative form, because it specifies a
10875 @item -fcall-used-@var{reg}
10876 @opindex fcall-used
10877 Treat the register named @var{reg} as an allocable register that is
10878 clobbered by function calls. It may be allocated for temporaries or
10879 variables that do not live across a call. Functions compiled this way
10880 will not save and restore the register @var{reg}.
10882 It is an error to used this flag with the frame pointer or stack pointer.
10883 Use of this flag for other registers that have fixed pervasive roles in
10884 the machine's execution model will produce disastrous results.
10886 This flag does not have a negative form, because it specifies a
10889 @item -fcall-saved-@var{reg}
10890 @opindex fcall-saved
10891 Treat the register named @var{reg} as an allocable register saved by
10892 functions. It may be allocated even for temporaries or variables that
10893 live across a call. Functions compiled this way will save and restore
10894 the register @var{reg} if they use it.
10896 It is an error to used this flag with the frame pointer or stack pointer.
10897 Use of this flag for other registers that have fixed pervasive roles in
10898 the machine's execution model will produce disastrous results.
10900 A different sort of disaster will result from the use of this flag for
10901 a register in which function values may be returned.
10903 This flag does not have a negative form, because it specifies a
10906 @item -fpack-struct
10907 @opindex fpack-struct
10908 Pack all structure members together without holes.
10910 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
10911 code that is not binary compatible with code generated without that switch.
10912 Additionally, it makes the code suboptimal.
10913 Use it to conform to a non-default application binary interface.
10915 @item -finstrument-functions
10916 @opindex finstrument-functions
10917 Generate instrumentation calls for entry and exit to functions. Just
10918 after function entry and just before function exit, the following
10919 profiling functions will be called with the address of the current
10920 function and its call site. (On some platforms,
10921 @code{__builtin_return_address} does not work beyond the current
10922 function, so the call site information may not be available to the
10923 profiling functions otherwise.)
10926 void __cyg_profile_func_enter (void *this_fn,
10928 void __cyg_profile_func_exit (void *this_fn,
10932 The first argument is the address of the start of the current function,
10933 which may be looked up exactly in the symbol table.
10935 This instrumentation is also done for functions expanded inline in other
10936 functions. The profiling calls will indicate where, conceptually, the
10937 inline function is entered and exited. This means that addressable
10938 versions of such functions must be available. If all your uses of a
10939 function are expanded inline, this may mean an additional expansion of
10940 code size. If you use @samp{extern inline} in your C code, an
10941 addressable version of such functions must be provided. (This is
10942 normally the case anyways, but if you get lucky and the optimizer always
10943 expands the functions inline, you might have gotten away without
10944 providing static copies.)
10946 A function may be given the attribute @code{no_instrument_function}, in
10947 which case this instrumentation will not be done. This can be used, for
10948 example, for the profiling functions listed above, high-priority
10949 interrupt routines, and any functions from which the profiling functions
10950 cannot safely be called (perhaps signal handlers, if the profiling
10951 routines generate output or allocate memory).
10953 @item -fstack-check
10954 @opindex fstack-check
10955 Generate code to verify that you do not go beyond the boundary of the
10956 stack. You should specify this flag if you are running in an
10957 environment with multiple threads, but only rarely need to specify it in
10958 a single-threaded environment since stack overflow is automatically
10959 detected on nearly all systems if there is only one stack.
10961 Note that this switch does not actually cause checking to be done; the
10962 operating system must do that. The switch causes generation of code
10963 to ensure that the operating system sees the stack being extended.
10965 @item -fstack-limit-register=@var{reg}
10966 @itemx -fstack-limit-symbol=@var{sym}
10967 @itemx -fno-stack-limit
10968 @opindex fstack-limit-register
10969 @opindex fstack-limit-symbol
10970 @opindex fno-stack-limit
10971 Generate code to ensure that the stack does not grow beyond a certain value,
10972 either the value of a register or the address of a symbol. If the stack
10973 would grow beyond the value, a signal is raised. For most targets,
10974 the signal is raised before the stack overruns the boundary, so
10975 it is possible to catch the signal without taking special precautions.
10977 For instance, if the stack starts at absolute address @samp{0x80000000}
10978 and grows downwards, you can use the flags
10979 @option{-fstack-limit-symbol=__stack_limit} and
10980 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
10981 of 128KB@. Note that this may only work with the GNU linker.
10983 @cindex aliasing of parameters
10984 @cindex parameters, aliased
10985 @item -fargument-alias
10986 @itemx -fargument-noalias
10987 @itemx -fargument-noalias-global
10988 @opindex fargument-alias
10989 @opindex fargument-noalias
10990 @opindex fargument-noalias-global
10991 Specify the possible relationships among parameters and between
10992 parameters and global data.
10994 @option{-fargument-alias} specifies that arguments (parameters) may
10995 alias each other and may alias global storage.@*
10996 @option{-fargument-noalias} specifies that arguments do not alias
10997 each other, but may alias global storage.@*
10998 @option{-fargument-noalias-global} specifies that arguments do not
10999 alias each other and do not alias global storage.
11001 Each language will automatically use whatever option is required by
11002 the language standard. You should not need to use these options yourself.
11004 @item -fleading-underscore
11005 @opindex fleading-underscore
11006 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11007 change the way C symbols are represented in the object file. One use
11008 is to help link with legacy assembly code.
11010 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11011 generate code that is not binary compatible with code generated without that
11012 switch. Use it to conform to a non-default application binary interface.
11013 Not all targets provide complete support for this switch.
11015 @item -ftls-model=@var{model}
11016 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11017 The @var{model} argument should be one of @code{global-dynamic},
11018 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11020 The default without @option{-fpic} is @code{initial-exec}; with
11021 @option{-fpic} the default is @code{global-dynamic}.
11026 @node Environment Variables
11027 @section Environment Variables Affecting GCC
11028 @cindex environment variables
11030 @c man begin ENVIRONMENT
11031 This section describes several environment variables that affect how GCC
11032 operates. Some of them work by specifying directories or prefixes to use
11033 when searching for various kinds of files. Some are used to specify other
11034 aspects of the compilation environment.
11036 Note that you can also specify places to search using options such as
11037 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11038 take precedence over places specified using environment variables, which
11039 in turn take precedence over those specified by the configuration of GCC@.
11040 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11041 GNU Compiler Collection (GCC) Internals}.
11046 @c @itemx LC_COLLATE
11048 @c @itemx LC_MONETARY
11049 @c @itemx LC_NUMERIC
11054 @c @findex LC_COLLATE
11055 @findex LC_MESSAGES
11056 @c @findex LC_MONETARY
11057 @c @findex LC_NUMERIC
11061 These environment variables control the way that GCC uses
11062 localization information that allow GCC to work with different
11063 national conventions. GCC inspects the locale categories
11064 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11065 so. These locale categories can be set to any value supported by your
11066 installation. A typical value is @samp{en_UK} for English in the United
11069 The @env{LC_CTYPE} environment variable specifies character
11070 classification. GCC uses it to determine the character boundaries in
11071 a string; this is needed for some multibyte encodings that contain quote
11072 and escape characters that would otherwise be interpreted as a string
11075 The @env{LC_MESSAGES} environment variable specifies the language to
11076 use in diagnostic messages.
11078 If the @env{LC_ALL} environment variable is set, it overrides the value
11079 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11080 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11081 environment variable. If none of these variables are set, GCC
11082 defaults to traditional C English behavior.
11086 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11087 files. GCC uses temporary files to hold the output of one stage of
11088 compilation which is to be used as input to the next stage: for example,
11089 the output of the preprocessor, which is the input to the compiler
11092 @item GCC_EXEC_PREFIX
11093 @findex GCC_EXEC_PREFIX
11094 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11095 names of the subprograms executed by the compiler. No slash is added
11096 when this prefix is combined with the name of a subprogram, but you can
11097 specify a prefix that ends with a slash if you wish.
11099 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11100 an appropriate prefix to use based on the pathname it was invoked with.
11102 If GCC cannot find the subprogram using the specified prefix, it
11103 tries looking in the usual places for the subprogram.
11105 The default value of @env{GCC_EXEC_PREFIX} is
11106 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
11107 of @code{prefix} when you ran the @file{configure} script.
11109 Other prefixes specified with @option{-B} take precedence over this prefix.
11111 This prefix is also used for finding files such as @file{crt0.o} that are
11114 In addition, the prefix is used in an unusual way in finding the
11115 directories to search for header files. For each of the standard
11116 directories whose name normally begins with @samp{/usr/local/lib/gcc}
11117 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11118 replacing that beginning with the specified prefix to produce an
11119 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11120 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11121 These alternate directories are searched first; the standard directories
11124 @item COMPILER_PATH
11125 @findex COMPILER_PATH
11126 The value of @env{COMPILER_PATH} is a colon-separated list of
11127 directories, much like @env{PATH}. GCC tries the directories thus
11128 specified when searching for subprograms, if it can't find the
11129 subprograms using @env{GCC_EXEC_PREFIX}.
11132 @findex LIBRARY_PATH
11133 The value of @env{LIBRARY_PATH} is a colon-separated list of
11134 directories, much like @env{PATH}. When configured as a native compiler,
11135 GCC tries the directories thus specified when searching for special
11136 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11137 using GCC also uses these directories when searching for ordinary
11138 libraries for the @option{-l} option (but directories specified with
11139 @option{-L} come first).
11143 @cindex locale definition
11144 This variable is used to pass locale information to the compiler. One way in
11145 which this information is used is to determine the character set to be used
11146 when character literals, string literals and comments are parsed in C and C++.
11147 When the compiler is configured to allow multibyte characters,
11148 the following values for @env{LANG} are recognized:
11152 Recognize JIS characters.
11154 Recognize SJIS characters.
11156 Recognize EUCJP characters.
11159 If @env{LANG} is not defined, or if it has some other value, then the
11160 compiler will use mblen and mbtowc as defined by the default locale to
11161 recognize and translate multibyte characters.
11165 Some additional environments variables affect the behavior of the
11168 @include cppenv.texi
11172 @node Precompiled Headers
11173 @section Using Precompiled Headers
11174 @cindex precompiled headers
11175 @cindex speed of compilation
11177 Often large projects have many header files that are included in every
11178 source file. The time the compiler takes to process these header files
11179 over and over again can account for nearly all of the time required to
11180 build the project. To make builds faster, GCC allows users to
11181 `precompile' a header file; then, if builds can use the precompiled
11182 header file they will be much faster.
11184 To create a precompiled header file, simply compile it as you would any
11185 other file, if necessary using the @option{-x} option to make the driver
11186 treat it as a C or C++ header file. You will probably want to use a
11187 tool like @command{make} to keep the precompiled header up-to-date when
11188 the headers it contains change.
11190 A precompiled header file will be searched for when @code{#include} is
11191 seen in the compilation. As it searches for the included file
11192 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
11193 compiler looks for a precompiled header in each directory just before it
11194 looks for the include file in that directory. The name searched for is
11195 the name specified in the @code{#include} with @samp{.gch} appended. If
11196 the precompiled header file can't be used, it is ignored.
11198 For instance, if you have @code{#include "all.h"}, and you have
11199 @file{all.h.gch} in the same directory as @file{all.h}, then the
11200 precompiled header file will be used if possible, and the original
11201 header will be used otherwise.
11203 Alternatively, you might decide to put the precompiled header file in a
11204 directory and use @option{-I} to ensure that directory is searched
11205 before (or instead of) the directory containing the original header.
11206 Then, if you want to check that the precompiled header file is always
11207 used, you can put a file of the same name as the original header in this
11208 directory containing an @code{#error} command.
11210 This also works with @option{-include}. So yet another way to use
11211 precompiled headers, good for projects not designed with precompiled
11212 header files in mind, is to simply take most of the header files used by
11213 a project, include them from another header file, precompile that header
11214 file, and @option{-include} the precompiled header. If the header files
11215 have guards against multiple inclusion, they will be skipped because
11216 they've already been included (in the precompiled header).
11218 If you need to precompile the same header file for different
11219 languages, targets, or compiler options, you can instead make a
11220 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11221 header in the directory. (It doesn't matter what you call the files
11222 in the directory, every precompiled header in the directory will be
11223 considered.) The first precompiled header encountered in the
11224 directory that is valid for this compilation will be used; they're
11225 searched in no particular order.
11227 There are many other possibilities, limited only by your imagination,
11228 good sense, and the constraints of your build system.
11230 A precompiled header file can be used only when these conditions apply:
11234 Only one precompiled header can be used in a particular compilation.
11236 A precompiled header can't be used once the first C token is seen. You
11237 can have preprocessor directives before a precompiled header; you can
11238 even include a precompiled header from inside another header, so long as
11239 there are no C tokens before the @code{#include}.
11241 The precompiled header file must be produced for the same language as
11242 the current compilation. You can't use a C precompiled header for a C++
11245 The precompiled header file must be produced by the same compiler
11246 version and configuration as the current compilation is using.
11247 The easiest way to guarantee this is to use the same compiler binary
11248 for creating and using precompiled headers.
11250 Any macros defined before the precompiled header (including with
11251 @option{-D}) must either be defined in the same way as when the
11252 precompiled header was generated, or must not affect the precompiled
11253 header, which usually means that the they don't appear in the
11254 precompiled header at all.
11256 Certain command-line options must be defined in the same way as when the
11257 precompiled header was generated. At present, it's not clear which
11258 options are safe to change and which are not; the safest choice is to
11259 use exactly the same options when generating and using the precompiled
11263 For all of these but the last, the compiler will automatically ignore
11264 the precompiled header if the conditions aren't met. For the last item,
11265 some option changes will cause the precompiled header to be rejected,
11266 but not all incompatible option combinations have yet been found. If
11267 you find a new incompatible combination, please consider filing a bug
11268 report, see @ref{Bugs}.
11270 @node Running Protoize
11271 @section Running Protoize
11273 The program @code{protoize} is an optional part of GCC@. You can use
11274 it to add prototypes to a program, thus converting the program to ISO
11275 C in one respect. The companion program @code{unprotoize} does the
11276 reverse: it removes argument types from any prototypes that are found.
11278 When you run these programs, you must specify a set of source files as
11279 command line arguments. The conversion programs start out by compiling
11280 these files to see what functions they define. The information gathered
11281 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11283 After scanning comes actual conversion. The specified files are all
11284 eligible to be converted; any files they include (whether sources or
11285 just headers) are eligible as well.
11287 But not all the eligible files are converted. By default,
11288 @code{protoize} and @code{unprotoize} convert only source and header
11289 files in the current directory. You can specify additional directories
11290 whose files should be converted with the @option{-d @var{directory}}
11291 option. You can also specify particular files to exclude with the
11292 @option{-x @var{file}} option. A file is converted if it is eligible, its
11293 directory name matches one of the specified directory names, and its
11294 name within the directory has not been excluded.
11296 Basic conversion with @code{protoize} consists of rewriting most
11297 function definitions and function declarations to specify the types of
11298 the arguments. The only ones not rewritten are those for varargs
11301 @code{protoize} optionally inserts prototype declarations at the
11302 beginning of the source file, to make them available for any calls that
11303 precede the function's definition. Or it can insert prototype
11304 declarations with block scope in the blocks where undeclared functions
11307 Basic conversion with @code{unprotoize} consists of rewriting most
11308 function declarations to remove any argument types, and rewriting
11309 function definitions to the old-style pre-ISO form.
11311 Both conversion programs print a warning for any function declaration or
11312 definition that they can't convert. You can suppress these warnings
11315 The output from @code{protoize} or @code{unprotoize} replaces the
11316 original source file. The original file is renamed to a name ending
11317 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11318 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11319 for DOS) file already exists, then the source file is simply discarded.
11321 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11322 scan the program and collect information about the functions it uses.
11323 So neither of these programs will work until GCC is installed.
11325 Here is a table of the options you can use with @code{protoize} and
11326 @code{unprotoize}. Each option works with both programs unless
11330 @item -B @var{directory}
11331 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11332 usual directory (normally @file{/usr/local/lib}). This file contains
11333 prototype information about standard system functions. This option
11334 applies only to @code{protoize}.
11336 @item -c @var{compilation-options}
11337 Use @var{compilation-options} as the options when running @command{gcc} to
11338 produce the @samp{.X} files. The special option @option{-aux-info} is
11339 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11341 Note that the compilation options must be given as a single argument to
11342 @code{protoize} or @code{unprotoize}. If you want to specify several
11343 @command{gcc} options, you must quote the entire set of compilation options
11344 to make them a single word in the shell.
11346 There are certain @command{gcc} arguments that you cannot use, because they
11347 would produce the wrong kind of output. These include @option{-g},
11348 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11349 the @var{compilation-options}, they are ignored.
11352 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11353 systems) instead of @samp{.c}. This is convenient if you are converting
11354 a C program to C++. This option applies only to @code{protoize}.
11357 Add explicit global declarations. This means inserting explicit
11358 declarations at the beginning of each source file for each function
11359 that is called in the file and was not declared. These declarations
11360 precede the first function definition that contains a call to an
11361 undeclared function. This option applies only to @code{protoize}.
11363 @item -i @var{string}
11364 Indent old-style parameter declarations with the string @var{string}.
11365 This option applies only to @code{protoize}.
11367 @code{unprotoize} converts prototyped function definitions to old-style
11368 function definitions, where the arguments are declared between the
11369 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11370 uses five spaces as the indentation. If you want to indent with just
11371 one space instead, use @option{-i " "}.
11374 Keep the @samp{.X} files. Normally, they are deleted after conversion
11378 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11379 a prototype declaration for each function in each block which calls the
11380 function without any declaration. This option applies only to
11384 Make no real changes. This mode just prints information about the conversions
11385 that would have been done without @option{-n}.
11388 Make no @samp{.save} files. The original files are simply deleted.
11389 Use this option with caution.
11391 @item -p @var{program}
11392 Use the program @var{program} as the compiler. Normally, the name
11393 @file{gcc} is used.
11396 Work quietly. Most warnings are suppressed.
11399 Print the version number, just like @option{-v} for @command{gcc}.
11402 If you need special compiler options to compile one of your program's
11403 source files, then you should generate that file's @samp{.X} file
11404 specially, by running @command{gcc} on that source file with the
11405 appropriate options and the option @option{-aux-info}. Then run
11406 @code{protoize} on the entire set of files. @code{protoize} will use
11407 the existing @samp{.X} file because it is newer than the source file.
11411 gcc -Dfoo=bar file1.c -aux-info file1.X
11416 You need to include the special files along with the rest in the
11417 @code{protoize} command, even though their @samp{.X} files already
11418 exist, because otherwise they won't get converted.
11420 @xref{Protoize Caveats}, for more information on how to use
11421 @code{protoize} successfully.