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} -combine -pipe -pass-exit-codes @gol
159 -x @var{language} -v -### --help --target-help --version}
161 @item C Language Options
162 @xref{C Dialect Options,,Options Controlling C Dialect}.
163 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
164 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
165 -fhosted -ffreestanding -fms-extensions @gol
166 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
167 -fallow-single-precision -fcond-mismatch @gol
168 -fsigned-bitfields -fsigned-char @gol
169 -funsigned-bitfields -funsigned-char}
171 @item C++ Language Options
172 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
173 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
174 -fconserve-space -fno-const-strings @gol
175 -fno-elide-constructors @gol
176 -fno-enforce-eh-specs @gol
177 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
178 -fno-implicit-templates @gol
179 -fno-implicit-inline-templates @gol
180 -fno-implement-inlines -fms-extensions @gol
181 -fno-nonansi-builtins -fno-operator-names @gol
182 -fno-optional-diags -fpermissive @gol
183 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
184 -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
185 -fno-default-inline -Wabi -Wctor-dtor-privacy @gol
186 -Wnon-virtual-dtor -Wreorder @gol
187 -Weffc++ -Wno-deprecated @gol
188 -Wno-non-template-friend -Wold-style-cast @gol
189 -Woverloaded-virtual -Wno-pmf-conversions @gol
190 -Wsign-promo -Wsynth}
192 @item Objective-C Language Options
193 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
195 -fconstant-string-class=@var{class-name} @gol
196 -fgnu-runtime -fnext-runtime @gol
197 -fno-nil-receivers @gol
198 -fobjc-exceptions @gol
199 -freplace-objc-classes @gol
202 -Wno-protocol -Wselector -Wundeclared-selector}
204 @item Language Independent Options
205 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
206 @gccoptlist{-fmessage-length=@var{n} @gol
207 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
209 @item Warning Options
210 @xref{Warning Options,,Options to Request or Suppress Warnings}.
211 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
212 -w -Wextra -Wall -Waggregate-return @gol
213 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
214 -Wconversion -Wno-deprecated-declarations @gol
215 -Wdisabled-optimization -Wno-div-by-zero -Wendif-labels @gol
216 -Werror -Werror-implicit-function-declaration @gol
217 -Wfloat-equal -Wformat -Wformat=2 @gol
218 -Wno-format-extra-args -Wformat-nonliteral @gol
219 -Wformat-security -Wformat-y2k @gol
220 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
221 -Wimport -Wno-import -Winit-self -Winline @gol
222 -Wno-invalid-offsetof -Winvalid-pch @gol
223 -Wlarger-than-@var{len} -Wlong-long @gol
224 -Wmain -Wmissing-braces @gol
225 -Wmissing-format-attribute -Wmissing-noreturn @gol
226 -Wno-multichar -Wnonnull -Wpacked -Wpadded @gol
227 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
228 -Wreturn-type -Wsequence-point -Wshadow @gol
229 -Wsign-compare -Wstrict-aliasing -Wstrict-aliasing=2 @gol
230 -Wswitch -Wswitch-default -Wswitch-enum @gol
231 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
232 -Wunknown-pragmas -Wunreachable-code @gol
233 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
234 -Wunused-value -Wunused-variable -Wwrite-strings @gol
237 @item C-only Warning Options
238 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
239 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
240 -Wstrict-prototypes -Wtraditional @gol
241 -Wdeclaration-after-statement}
243 @item Debugging Options
244 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
245 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
246 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
247 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
248 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
249 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
250 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
251 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
252 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs @gol
253 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
254 -ftest-coverage -ftime-report -fvar-tracking @gol
255 -g -g@var{level} -gcoff -gdwarf-2 @gol
256 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
257 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
258 -print-multi-directory -print-multi-lib @gol
259 -print-prog-name=@var{program} -print-search-dirs -Q @gol
262 @item Optimization Options
263 @xref{Optimize Options,,Options that Control Optimization}.
264 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
265 -falign-labels=@var{n} -falign-loops=@var{n} @gol
266 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
267 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
268 -fcaller-saves -fcprop-registers @gol
269 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
270 -fdelayed-branch -fdelete-null-pointer-checks @gol
271 -fexpensive-optimizations -ffast-math -ffloat-store @gol
272 -fforce-addr -fforce-mem -ffunction-sections @gol
273 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -floop-optimize @gol
274 -fcrossjumping -fif-conversion -fif-conversion2 @gol
275 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
276 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
277 -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
278 -fno-default-inline -fno-defer-pop @gol
279 -fno-function-cse -fno-guess-branch-probability @gol
280 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
281 -funsafe-math-optimizations -ffinite-math-only @gol
282 -fno-trapping-math -fno-zero-initialized-in-bss @gol
283 -fomit-frame-pointer -foptimize-register-move @gol
284 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
285 -fprofile-generate -fprofile-use @gol
286 -freduce-all-givs -fregmove -frename-registers @gol
287 -freorder-blocks -freorder-blocks-and-partition -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 -mabi=@var{name} @gol
372 -mapcs-26 -mapcs-32 @gol
373 -mapcs-stack-check -mno-apcs-stack-check @gol
374 -mapcs-float -mno-apcs-float @gol
375 -mapcs-reentrant -mno-apcs-reentrant @gol
376 -msched-prolog -mno-sched-prolog @gol
377 -mlittle-endian -mbig-endian -mwords-little-endian @gol
378 -malignment-traps -mno-alignment-traps @gol
379 -mfloat-abi=@var{name} soft-float -mhard-float -mfpe @gol
380 -mthumb-interwork -mno-thumb-interwork @gol
381 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
382 -mstructure-size-boundary=@var{n} @gol
383 -mabort-on-noreturn @gol
384 -mlong-calls -mno-long-calls @gol
385 -msingle-pic-base -mno-single-pic-base @gol
386 -mpic-register=@var{reg} @gol
387 -mnop-fun-dllimport @gol
388 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
389 -mpoke-function-name @gol
391 -mtpcs-frame -mtpcs-leaf-frame @gol
392 -mcaller-super-interworking -mcallee-super-interworking}
394 @emph{MN10300 Options}
395 @gccoptlist{-mmult-bug -mno-mult-bug @gol
396 -mam33 -mno-am33 @gol
397 -mam33-2 -mno-am33-2 @gol
400 @emph{M32R/D Options}
401 @gccoptlist{-m32r2 -m32rx -m32r @gol
403 -malign-loops -mno-align-loops @gol
404 -missue-rate=@var{number} @gol
405 -mbranch-cost=@var{number} @gol
406 -mmodel=@var{code-size-model-type} @gol
407 -msdata=@var{sdata-type} @gol
408 -mno-flush-func -mflush-func=@var{name} @gol
409 -mno-flush-trap -mflush-trap=@var{number} @gol
412 @emph{RS/6000 and PowerPC Options}
413 @gccoptlist{-mcpu=@var{cpu-type} @gol
414 -mtune=@var{cpu-type} @gol
415 -mpower -mno-power -mpower2 -mno-power2 @gol
416 -mpowerpc -mpowerpc64 -mno-powerpc @gol
417 -maltivec -mno-altivec @gol
418 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
419 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
420 -mnew-mnemonics -mold-mnemonics @gol
421 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
422 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
423 -malign-power -malign-natural @gol
424 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
425 -mstring -mno-string -mupdate -mno-update @gol
426 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
427 -mstrict-align -mno-strict-align -mrelocatable @gol
428 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
429 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
430 -mdynamic-no-pic @gol
431 -mprioritize-restricted-insns=@var{priority} @gol
432 -msched-costly-dep=@var{dependence_type} @gol
433 -minsert-sched-nops=@var{scheme} @gol
434 -mcall-sysv -mcall-netbsd @gol
435 -maix-struct-return -msvr4-struct-return @gol
436 -mabi=altivec -mabi=no-altivec @gol
437 -mabi=spe -mabi=no-spe @gol
438 -misel=yes -misel=no @gol
439 -mspe=yes -mspe=no @gol
440 -mfloat-gprs=yes -mfloat-gprs=no @gol
441 -mprototype -mno-prototype @gol
442 -msim -mmvme -mads -myellowknife -memb -msdata @gol
443 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
445 @emph{Darwin Options}
446 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
447 -arch_only -bind_at_load -bundle -bundle_loader @gol
448 -client_name -compatibility_version -current_version @gol
449 -dependency-file -dylib_file -dylinker_install_name @gol
450 -dynamic -dynamiclib -exported_symbols_list @gol
451 -filelist -flat_namespace -force_cpusubtype_ALL @gol
452 -force_flat_namespace -headerpad_max_install_names @gol
453 -image_base -init -install_name -keep_private_externs @gol
454 -multi_module -multiply_defined -multiply_defined_unused @gol
455 -noall_load -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
456 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
457 -private_bundle -read_only_relocs -sectalign @gol
458 -sectobjectsymbols -whyload -seg1addr @gol
459 -sectcreate -sectobjectsymbols -sectorder @gol
460 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
461 -segprot -segs_read_only_addr -segs_read_write_addr @gol
462 -single_module -static -sub_library -sub_umbrella @gol
463 -twolevel_namespace -umbrella -undefined @gol
464 -unexported_symbols_list -weak_reference_mismatches @gol
468 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
469 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
470 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
471 -mxgot -mno-xgot -membedded-pic -mno-embedded-pic @gol
472 -mgp32 -mgp64 -mfp32 -mfp64 -mhard-float -msoft-float @gol
473 -msingle-float -mdouble-float -mint64 -mlong64 -mlong32 @gol
474 -G@var{num} -membedded-data -mno-embedded-data @gol
475 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
476 -msplit-addresses -mno-split-addresses @gol
477 -mexplicit-relocs -mno-explicit-relocs @gol
478 -mrnames -mno-rnames @gol
479 -mcheck-zero-division -mno-check-zero-division @gol
480 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
481 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
482 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
483 -mfix-vr4120 -mno-fix-vr4120 -mfix-sb1 -mno-fix-sb1 @gol
484 -mflush-func=@var{func} -mno-flush-func @gol
485 -mbranch-likely -mno-branch-likely}
487 @emph{i386 and x86-64 Options}
488 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
489 -mfpmath=@var{unit} @gol
490 -masm=@var{dialect} -mno-fancy-math-387 @gol
491 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
492 -mno-wide-multiply -mrtd -malign-double @gol
493 -mpreferred-stack-boundary=@var{num} @gol
494 -mmmx -msse -msse2 -msse3 -m3dnow @gol
495 -mthreads -mno-align-stringops -minline-all-stringops @gol
496 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
497 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
498 -mno-red-zone -mno-tls-direct-seg-refs @gol
499 -mcmodel=@var{code-model} @gol
503 @gccoptlist{-march=@var{architecture-type} @gol
504 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
505 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
506 -mjump-in-delay -mlinker-opt -mlong-calls @gol
507 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
508 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
509 -mno-jump-in-delay -mno-long-load-store @gol
510 -mno-portable-runtime -mno-soft-float @gol
511 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
512 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
513 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
514 -nolibdld -static -threads}
516 @emph{DEC Alpha Options}
517 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
518 -mieee -mieee-with-inexact -mieee-conformant @gol
519 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
520 -mtrap-precision=@var{mode} -mbuild-constants @gol
521 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
522 -mbwx -mmax -mfix -mcix @gol
523 -mfloat-vax -mfloat-ieee @gol
524 -mexplicit-relocs -msmall-data -mlarge-data @gol
525 -msmall-text -mlarge-text @gol
526 -mmemory-latency=@var{time}}
528 @emph{DEC Alpha/VMS Options}
529 @gccoptlist{-mvms-return-codes}
531 @emph{H8/300 Options}
532 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
535 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
536 -m4-nofpu -m4-single-only -m4-single -m4 @gol
537 -m5-64media -m5-64media-nofpu @gol
538 -m5-32media -m5-32media-nofpu @gol
539 -m5-compact -m5-compact-nofpu @gol
540 -mb -ml -mdalign -mrelax @gol
541 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
542 -mieee -misize -mpadstruct -mspace @gol
543 -mprefergot -musermode}
545 @emph{System V Options}
546 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
549 @gccoptlist{-EB -EL @gol
550 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
551 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
553 @emph{TMS320C3x/C4x Options}
554 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
555 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
556 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
557 -mparallel-insns -mparallel-mpy -mpreserve-float}
560 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
561 -mprolog-function -mno-prolog-function -mspace @gol
562 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
563 -mapp-regs -mno-app-regs @gol
564 -mdisable-callt -mno-disable-callt @gol
570 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
571 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
572 -mregparam -mnoregparam -msb -mnosb @gol
573 -mbitfield -mnobitfield -mhimem -mnohimem}
576 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
577 -mcall-prologues -mno-tablejump -mtiny-stack}
580 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
581 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
582 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
583 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
584 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
587 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
588 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
589 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
590 -mno-base-addresses -msingle-exit -mno-single-exit}
593 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
594 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
595 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
596 -minline-float-divide-max-throughput @gol
597 -minline-int-divide-min-latency @gol
598 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
599 -mfixed-range=@var{register-range}}
601 @emph{S/390 and zSeries Options}
602 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
603 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
604 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
605 -m64 -m31 -mdebug -mno-debug -mesa -mzarch -mfused-madd -mno-fused-madd}
608 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
609 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
610 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
611 -mstack-align -mdata-align -mconst-align @gol
612 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
613 -melf -maout -melinux -mlinux -sim -sim2 @gol
614 -mmul-bug-workaround -mno-mul-bug-workaround}
616 @emph{PDP-11 Options}
617 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
618 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
619 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
620 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
621 -mbranch-expensive -mbranch-cheap @gol
622 -msplit -mno-split -munix-asm -mdec-asm}
624 @emph{Xstormy16 Options}
627 @emph{Xtensa Options}
628 @gccoptlist{-mconst16 -mno-const16 @gol
629 -mfused-madd -mno-fused-madd @gol
630 -mtext-section-literals -mno-text-section-literals @gol
631 -mtarget-align -mno-target-align @gol
632 -mlongcalls -mno-longcalls}
635 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
636 -mhard-float -msoft-float @gol
637 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
638 -mdouble -mno-double @gol
639 -mmedia -mno-media -mmuladd -mno-muladd @gol
640 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic -mlinked-fp @gol
641 -mlibrary-pic -macc-4 -macc-8 @gol
642 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
643 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
644 -mvliw-branch -mno-vliw-branch @gol
645 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
646 -mno-nested-cond-exec -mtomcat-stats @gol
649 @item Code Generation Options
650 @xref{Code Gen Options,,Options for Code Generation Conventions}.
651 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
652 -ffixed-@var{reg} -fexceptions @gol
653 -fnon-call-exceptions -funwind-tables @gol
654 -fasynchronous-unwind-tables @gol
655 -finhibit-size-directive -finstrument-functions @gol
656 -fno-common -fno-ident @gol
657 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
658 -freg-struct-return -fshared-data -fshort-enums @gol
659 -fshort-double -fshort-wchar @gol
660 -fverbose-asm -fpack-struct -fstack-check @gol
661 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
662 -fargument-alias -fargument-noalias @gol
663 -fargument-noalias-global -fleading-underscore @gol
664 -ftls-model=@var{model} @gol
665 -ftrapv -fwrapv -fbounds-check}
669 * Overall Options:: Controlling the kind of output:
670 an executable, object files, assembler files,
671 or preprocessed source.
672 * C Dialect Options:: Controlling the variant of C language compiled.
673 * C++ Dialect Options:: Variations on C++.
674 * Objective-C Dialect Options:: Variations on Objective-C.
675 * Language Independent Options:: Controlling how diagnostics should be
677 * Warning Options:: How picky should the compiler be?
678 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
679 * Optimize Options:: How much optimization?
680 * Preprocessor Options:: Controlling header files and macro definitions.
681 Also, getting dependency information for Make.
682 * Assembler Options:: Passing options to the assembler.
683 * Link Options:: Specifying libraries and so on.
684 * Directory Options:: Where to find header files and libraries.
685 Where to find the compiler executable files.
686 * Spec Files:: How to pass switches to sub-processes.
687 * Target Options:: Running a cross-compiler, or an old version of GCC.
690 @node Overall Options
691 @section Options Controlling the Kind of Output
693 Compilation can involve up to four stages: preprocessing, compilation
694 proper, assembly and linking, always in that order. GCC is capable of
695 preprocessing and compiling several files either into several
696 assembler input files, or into one assembler input file; then each
697 assembler input file produces an object file, and linking combines all
698 the object files (those newly compiled, and those specified as input)
699 into an executable file.
701 @cindex file name suffix
702 For any given input file, the file name suffix determines what kind of
707 C source code which must be preprocessed.
710 C source code which should not be preprocessed.
713 C++ source code which should not be preprocessed.
716 Objective-C source code. Note that you must link with the library
717 @file{libobjc.a} to make an Objective-C program work.
720 Objective-C source code which should not be preprocessed.
723 C or C++ header file to be turned into a precompiled header.
727 @itemx @var{file}.cxx
728 @itemx @var{file}.cpp
729 @itemx @var{file}.CPP
730 @itemx @var{file}.c++
732 C++ source code which must be preprocessed. Note that in @samp{.cxx},
733 the last two letters must both be literally @samp{x}. Likewise,
734 @samp{.C} refers to a literal capital C@.
738 C++ header file to be turned into a precompiled header.
741 @itemx @var{file}.for
742 @itemx @var{file}.FOR
743 Fortran source code which should not be preprocessed.
746 @itemx @var{file}.fpp
747 @itemx @var{file}.FPP
748 Fortran source code which must be preprocessed (with the traditional
752 Fortran source code which must be preprocessed with a RATFOR
753 preprocessor (not included with GCC)@.
755 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
756 Using and Porting GNU Fortran}, for more details of the handling of
759 @c FIXME: Descriptions of Java file types.
766 Ada source code file which contains a library unit declaration (a
767 declaration of a package, subprogram, or generic, or a generic
768 instantiation), or a library unit renaming declaration (a package,
769 generic, or subprogram renaming declaration). Such files are also
772 @itemx @var{file}.adb
773 Ada source code file containing a library unit body (a subprogram or
774 package body). Such files are also called @dfn{bodies}.
776 @c GCC also knows about some suffixes for languages not yet included:
785 Assembler code which must be preprocessed.
788 An object file to be fed straight into linking.
789 Any file name with no recognized suffix is treated this way.
793 You can specify the input language explicitly with the @option{-x} option:
796 @item -x @var{language}
797 Specify explicitly the @var{language} for the following input files
798 (rather than letting the compiler choose a default based on the file
799 name suffix). This option applies to all following input files until
800 the next @option{-x} option. Possible values for @var{language} are:
802 c c-header cpp-output
803 c++ c++-header c++-cpp-output
804 objective-c objective-c-header objc-cpp-output
805 assembler assembler-with-cpp
807 f77 f77-cpp-input ratfor
813 Turn off any specification of a language, so that subsequent files are
814 handled according to their file name suffixes (as they are if @option{-x}
815 has not been used at all).
817 @item -pass-exit-codes
818 @opindex pass-exit-codes
819 Normally the @command{gcc} program will exit with the code of 1 if any
820 phase of the compiler returns a non-success return code. If you specify
821 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
822 numerically highest error produced by any phase that returned an error
826 If you only want some of the stages of compilation, you can use
827 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
828 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
829 @command{gcc} is to stop. Note that some combinations (for example,
830 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
835 Compile or assemble the source files, but do not link. The linking
836 stage simply is not done. The ultimate output is in the form of an
837 object file for each source file.
839 By default, the object file name for a source file is made by replacing
840 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
842 Unrecognized input files, not requiring compilation or assembly, are
847 Stop after the stage of compilation proper; do not assemble. The output
848 is in the form of an assembler code file for each non-assembler input
851 By default, the assembler file name for a source file is made by
852 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
854 Input files that don't require compilation are ignored.
858 Stop after the preprocessing stage; do not run the compiler proper. The
859 output is in the form of preprocessed source code, which is sent to the
862 Input files which don't require preprocessing are ignored.
864 @cindex output file option
867 Place output in file @var{file}. This applies regardless to whatever
868 sort of output is being produced, whether it be an executable file,
869 an object file, an assembler file or preprocessed C code.
871 If @option{-o} is not specified, the default is to put an executable file
872 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
873 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
874 all preprocessed C source on standard output.
878 Print (on standard error output) the commands executed to run the stages
879 of compilation. Also print the version number of the compiler driver
880 program and of the preprocessor and the compiler proper.
884 Like @option{-v} except the commands are not executed and all command
885 arguments are quoted. This is useful for shell scripts to capture the
886 driver-generated command lines.
890 Use pipes rather than temporary files for communication between the
891 various stages of compilation. This fails to work on some systems where
892 the assembler is unable to read from a pipe; but the GNU assembler has
897 If you are compiling multiple source files, this option tells the driver
898 to pass all the source files to the compiler at once (for those
899 languages for which the compiler can handle this). This will allow
900 intermodule analysis (IMA) to be performed by the compiler. Currently the only
901 language for which this is supported is C. If you pass source files for
902 multiple languages to the driver, using this option, the driver will invoke
903 the compiler(s) that support IMA once each, passing each compiler all the
904 source files appropriate for it. For those languages that do not support
905 IMA this option will be ignored, and the compiler will be invoked once for
906 each source file in that language. If you use this option in conjunction
907 with -save-temps, the compiler will generate multiple pre-processed files
908 (one for each source file), but only one (combined) .o or .s file.
912 Print (on the standard output) a description of the command line options
913 understood by @command{gcc}. If the @option{-v} option is also specified
914 then @option{--help} will also be passed on to the various processes
915 invoked by @command{gcc}, so that they can display the command line options
916 they accept. If the @option{-Wextra} option is also specified then command
917 line options which have no documentation associated with them will also
922 Print (on the standard output) a description of target specific command
923 line options for each tool.
927 Display the version number and copyrights of the invoked GCC.
931 @section Compiling C++ Programs
933 @cindex suffixes for C++ source
934 @cindex C++ source file suffixes
935 C++ source files conventionally use one of the suffixes @samp{.C},
936 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
937 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
938 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
939 files with these names and compiles them as C++ programs even if you
940 call the compiler the same way as for compiling C programs (usually
941 with the name @command{gcc}).
945 However, C++ programs often require class libraries as well as a
946 compiler that understands the C++ language---and under some
947 circumstances, you might want to compile programs or header files from
948 standard input, or otherwise without a suffix that flags them as C++
949 programs. You might also like to precompile a C header file with a
950 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
951 program that calls GCC with the default language set to C++, and
952 automatically specifies linking against the C++ library. On many
953 systems, @command{g++} is also installed with the name @command{c++}.
955 @cindex invoking @command{g++}
956 When you compile C++ programs, you may specify many of the same
957 command-line options that you use for compiling programs in any
958 language; or command-line options meaningful for C and related
959 languages; or options that are meaningful only for C++ programs.
960 @xref{C Dialect Options,,Options Controlling C Dialect}, for
961 explanations of options for languages related to C@.
962 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
963 explanations of options that are meaningful only for C++ programs.
965 @node C Dialect Options
966 @section Options Controlling C Dialect
967 @cindex dialect options
968 @cindex language dialect options
969 @cindex options, dialect
971 The following options control the dialect of C (or languages derived
972 from C, such as C++ and Objective-C) that the compiler accepts:
979 In C mode, support all ISO C90 programs. In C++ mode,
980 remove GNU extensions that conflict with ISO C++.
982 This turns off certain features of GCC that are incompatible with ISO
983 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
984 such as the @code{asm} and @code{typeof} keywords, and
985 predefined macros such as @code{unix} and @code{vax} that identify the
986 type of system you are using. It also enables the undesirable and
987 rarely used ISO trigraph feature. For the C compiler,
988 it disables recognition of C++ style @samp{//} comments as well as
989 the @code{inline} keyword.
991 The alternate keywords @code{__asm__}, @code{__extension__},
992 @code{__inline__} and @code{__typeof__} continue to work despite
993 @option{-ansi}. You would not want to use them in an ISO C program, of
994 course, but it is useful to put them in header files that might be included
995 in compilations done with @option{-ansi}. Alternate predefined macros
996 such as @code{__unix__} and @code{__vax__} are also available, with or
997 without @option{-ansi}.
999 The @option{-ansi} option does not cause non-ISO programs to be
1000 rejected gratuitously. For that, @option{-pedantic} is required in
1001 addition to @option{-ansi}. @xref{Warning Options}.
1003 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1004 option is used. Some header files may notice this macro and refrain
1005 from declaring certain functions or defining certain macros that the
1006 ISO standard doesn't call for; this is to avoid interfering with any
1007 programs that might use these names for other things.
1009 Functions which would normally be built in but do not have semantics
1010 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1011 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1012 built-in functions provided by GCC}, for details of the functions
1017 Determine the language standard. This option is currently only
1018 supported when compiling C or C++. A value for this option must be
1019 provided; possible values are
1024 ISO C90 (same as @option{-ansi}).
1026 @item iso9899:199409
1027 ISO C90 as modified in amendment 1.
1033 ISO C99. Note that this standard is not yet fully supported; see
1034 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1035 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1038 Default, ISO C90 plus GNU extensions (including some C99 features).
1042 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1043 this will become the default. The name @samp{gnu9x} is deprecated.
1046 The 1998 ISO C++ standard plus amendments.
1049 The same as @option{-std=c++98} plus GNU extensions. This is the
1050 default for C++ code.
1053 Even when this option is not specified, you can still use some of the
1054 features of newer standards in so far as they do not conflict with
1055 previous C standards. For example, you may use @code{__restrict__} even
1056 when @option{-std=c99} is not specified.
1058 The @option{-std} options specifying some version of ISO C have the same
1059 effects as @option{-ansi}, except that features that were not in ISO C90
1060 but are in the specified version (for example, @samp{//} comments and
1061 the @code{inline} keyword in ISO C99) are not disabled.
1063 @xref{Standards,,Language Standards Supported by GCC}, for details of
1064 these standard versions.
1066 @item -aux-info @var{filename}
1068 Output to the given filename prototyped declarations for all functions
1069 declared and/or defined in a translation unit, including those in header
1070 files. This option is silently ignored in any language other than C@.
1072 Besides declarations, the file indicates, in comments, the origin of
1073 each declaration (source file and line), whether the declaration was
1074 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1075 @samp{O} for old, respectively, in the first character after the line
1076 number and the colon), and whether it came from a declaration or a
1077 definition (@samp{C} or @samp{F}, respectively, in the following
1078 character). In the case of function definitions, a K&R-style list of
1079 arguments followed by their declarations is also provided, inside
1080 comments, after the declaration.
1084 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1085 keyword, so that code can use these words as identifiers. You can use
1086 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1087 instead. @option{-ansi} implies @option{-fno-asm}.
1089 In C++, this switch only affects the @code{typeof} keyword, since
1090 @code{asm} and @code{inline} are standard keywords. You may want to
1091 use the @option{-fno-gnu-keywords} flag instead, which has the same
1092 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1093 switch only affects the @code{asm} and @code{typeof} keywords, since
1094 @code{inline} is a standard keyword in ISO C99.
1097 @itemx -fno-builtin-@var{function}
1098 @opindex fno-builtin
1099 @cindex built-in functions
1100 Don't recognize built-in functions that do not begin with
1101 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1102 functions provided by GCC}, for details of the functions affected,
1103 including those which are not built-in functions when @option{-ansi} or
1104 @option{-std} options for strict ISO C conformance are used because they
1105 do not have an ISO standard meaning.
1107 GCC normally generates special code to handle certain built-in functions
1108 more efficiently; for instance, calls to @code{alloca} may become single
1109 instructions that adjust the stack directly, and calls to @code{memcpy}
1110 may become inline copy loops. The resulting code is often both smaller
1111 and faster, but since the function calls no longer appear as such, you
1112 cannot set a breakpoint on those calls, nor can you change the behavior
1113 of the functions by linking with a different library.
1115 With the @option{-fno-builtin-@var{function}} option
1116 only the built-in function @var{function} is
1117 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1118 function is named this is not built-in in this version of GCC, this
1119 option is ignored. There is no corresponding
1120 @option{-fbuiltin-@var{function}} option; if you wish to enable
1121 built-in functions selectively when using @option{-fno-builtin} or
1122 @option{-ffreestanding}, you may define macros such as:
1125 #define abs(n) __builtin_abs ((n))
1126 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1131 @cindex hosted environment
1133 Assert that compilation takes place in a hosted environment. This implies
1134 @option{-fbuiltin}. A hosted environment is one in which the
1135 entire standard library is available, and in which @code{main} has a return
1136 type of @code{int}. Examples are nearly everything except a kernel.
1137 This is equivalent to @option{-fno-freestanding}.
1139 @item -ffreestanding
1140 @opindex ffreestanding
1141 @cindex hosted environment
1143 Assert that compilation takes place in a freestanding environment. This
1144 implies @option{-fno-builtin}. A freestanding environment
1145 is one in which the standard library may not exist, and program startup may
1146 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1147 This is equivalent to @option{-fno-hosted}.
1149 @xref{Standards,,Language Standards Supported by GCC}, for details of
1150 freestanding and hosted environments.
1152 @item -fms-extensions
1153 @opindex fms-extensions
1154 Accept some non-standard constructs used in Microsoft header files.
1158 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1159 options for strict ISO C conformance) implies @option{-trigraphs}.
1161 @item -no-integrated-cpp
1162 @opindex no-integrated-cpp
1163 Performs a compilation in two passes: preprocessing and compiling. This
1164 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1165 @option{-B} option. The user supplied compilation step can then add in
1166 an additional preprocessing step after normal preprocessing but before
1167 compiling. The default is to use the integrated cpp (internal cpp)
1169 The semantics of this option will change if "cc1", "cc1plus", and
1170 "cc1obj" are merged.
1172 @cindex traditional C language
1173 @cindex C language, traditional
1175 @itemx -traditional-cpp
1176 @opindex traditional-cpp
1177 @opindex traditional
1178 Formerly, these options caused GCC to attempt to emulate a pre-standard
1179 C compiler. They are now only supported with the @option{-E} switch.
1180 The preprocessor continues to support a pre-standard mode. See the GNU
1181 CPP manual for details.
1183 @item -fcond-mismatch
1184 @opindex fcond-mismatch
1185 Allow conditional expressions with mismatched types in the second and
1186 third arguments. The value of such an expression is void. This option
1187 is not supported for C++.
1189 @item -funsigned-char
1190 @opindex funsigned-char
1191 Let the type @code{char} be unsigned, like @code{unsigned char}.
1193 Each kind of machine has a default for what @code{char} should
1194 be. It is either like @code{unsigned char} by default or like
1195 @code{signed char} by default.
1197 Ideally, a portable program should always use @code{signed char} or
1198 @code{unsigned char} when it depends on the signedness of an object.
1199 But many programs have been written to use plain @code{char} and
1200 expect it to be signed, or expect it to be unsigned, depending on the
1201 machines they were written for. This option, and its inverse, let you
1202 make such a program work with the opposite default.
1204 The type @code{char} is always a distinct type from each of
1205 @code{signed char} or @code{unsigned char}, even though its behavior
1206 is always just like one of those two.
1209 @opindex fsigned-char
1210 Let the type @code{char} be signed, like @code{signed char}.
1212 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1213 the negative form of @option{-funsigned-char}. Likewise, the option
1214 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1216 @item -fsigned-bitfields
1217 @itemx -funsigned-bitfields
1218 @itemx -fno-signed-bitfields
1219 @itemx -fno-unsigned-bitfields
1220 @opindex fsigned-bitfields
1221 @opindex funsigned-bitfields
1222 @opindex fno-signed-bitfields
1223 @opindex fno-unsigned-bitfields
1224 These options control whether a bit-field is signed or unsigned, when the
1225 declaration does not use either @code{signed} or @code{unsigned}. By
1226 default, such a bit-field is signed, because this is consistent: the
1227 basic integer types such as @code{int} are signed types.
1230 @node C++ Dialect Options
1231 @section Options Controlling C++ Dialect
1233 @cindex compiler options, C++
1234 @cindex C++ options, command line
1235 @cindex options, C++
1236 This section describes the command-line options that are only meaningful
1237 for C++ programs; but you can also use most of the GNU compiler options
1238 regardless of what language your program is in. For example, you
1239 might compile a file @code{firstClass.C} like this:
1242 g++ -g -frepo -O -c firstClass.C
1246 In this example, only @option{-frepo} is an option meant
1247 only for C++ programs; you can use the other options with any
1248 language supported by GCC@.
1250 Here is a list of options that are @emph{only} for compiling C++ programs:
1254 @item -fabi-version=@var{n}
1255 @opindex fabi-version
1256 Use version @var{n} of the C++ ABI. Version 2 is the version of the
1257 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1258 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1259 the version that conforms most closely to the C++ ABI specification.
1260 Therefore, the ABI obtained using version 0 will change as ABI bugs
1263 The default is version 2.
1265 @item -fno-access-control
1266 @opindex fno-access-control
1267 Turn off all access checking. This switch is mainly useful for working
1268 around bugs in the access control code.
1272 Check that the pointer returned by @code{operator new} is non-null
1273 before attempting to modify the storage allocated. This check is
1274 normally unnecessary because the C++ standard specifies that
1275 @code{operator new} will only return @code{0} if it is declared
1276 @samp{throw()}, in which case the compiler will always check the
1277 return value even without this option. In all other cases, when
1278 @code{operator new} has a non-empty exception specification, memory
1279 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1280 @samp{new (nothrow)}.
1282 @item -fconserve-space
1283 @opindex fconserve-space
1284 Put uninitialized or runtime-initialized global variables into the
1285 common segment, as C does. This saves space in the executable at the
1286 cost of not diagnosing duplicate definitions. If you compile with this
1287 flag and your program mysteriously crashes after @code{main()} has
1288 completed, you may have an object that is being destroyed twice because
1289 two definitions were merged.
1291 This option is no longer useful on most targets, now that support has
1292 been added for putting variables into BSS without making them common.
1294 @item -fno-const-strings
1295 @opindex fno-const-strings
1296 Give string constants type @code{char *} instead of type @code{const
1297 char *}. By default, G++ uses type @code{const char *} as required by
1298 the standard. Even if you use @option{-fno-const-strings}, you cannot
1299 actually modify the value of a string constant.
1301 This option might be removed in a future release of G++. For maximum
1302 portability, you should structure your code so that it works with
1303 string constants that have type @code{const char *}.
1305 @item -fno-elide-constructors
1306 @opindex fno-elide-constructors
1307 The C++ standard allows an implementation to omit creating a temporary
1308 which is only used to initialize another object of the same type.
1309 Specifying this option disables that optimization, and forces G++ to
1310 call the copy constructor in all cases.
1312 @item -fno-enforce-eh-specs
1313 @opindex fno-enforce-eh-specs
1314 Don't check for violation of exception specifications at runtime. This
1315 option violates the C++ standard, but may be useful for reducing code
1316 size in production builds, much like defining @samp{NDEBUG}. The compiler
1317 will still optimize based on the exception specifications.
1320 @itemx -fno-for-scope
1322 @opindex fno-for-scope
1323 If @option{-ffor-scope} is specified, the scope of variables declared in
1324 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1325 as specified by the C++ standard.
1326 If @option{-fno-for-scope} is specified, the scope of variables declared in
1327 a @i{for-init-statement} extends to the end of the enclosing scope,
1328 as was the case in old versions of G++, and other (traditional)
1329 implementations of C++.
1331 The default if neither flag is given to follow the standard,
1332 but to allow and give a warning for old-style code that would
1333 otherwise be invalid, or have different behavior.
1335 @item -fno-gnu-keywords
1336 @opindex fno-gnu-keywords
1337 Do not recognize @code{typeof} as a keyword, so that code can use this
1338 word as an identifier. You can use the keyword @code{__typeof__} instead.
1339 @option{-ansi} implies @option{-fno-gnu-keywords}.
1341 @item -fno-implicit-templates
1342 @opindex fno-implicit-templates
1343 Never emit code for non-inline templates which are instantiated
1344 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1345 @xref{Template Instantiation}, for more information.
1347 @item -fno-implicit-inline-templates
1348 @opindex fno-implicit-inline-templates
1349 Don't emit code for implicit instantiations of inline templates, either.
1350 The default is to handle inlines differently so that compiles with and
1351 without optimization will need the same set of explicit instantiations.
1353 @item -fno-implement-inlines
1354 @opindex fno-implement-inlines
1355 To save space, do not emit out-of-line copies of inline functions
1356 controlled by @samp{#pragma implementation}. This will cause linker
1357 errors if these functions are not inlined everywhere they are called.
1359 @item -fms-extensions
1360 @opindex fms-extensions
1361 Disable pedantic warnings about constructs used in MFC, such as implicit
1362 int and getting a pointer to member function via non-standard syntax.
1364 @item -fno-nonansi-builtins
1365 @opindex fno-nonansi-builtins
1366 Disable built-in declarations of functions that are not mandated by
1367 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1368 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1370 @item -fno-operator-names
1371 @opindex fno-operator-names
1372 Do not treat the operator name keywords @code{and}, @code{bitand},
1373 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1374 synonyms as keywords.
1376 @item -fno-optional-diags
1377 @opindex fno-optional-diags
1378 Disable diagnostics that the standard says a compiler does not need to
1379 issue. Currently, the only such diagnostic issued by G++ is the one for
1380 a name having multiple meanings within a class.
1383 @opindex fpermissive
1384 Downgrade some diagnostics about nonconformant code from errors to
1385 warnings. Thus, using @option{-fpermissive} will allow some
1386 nonconforming code to compile.
1390 Enable automatic template instantiation at link time. This option also
1391 implies @option{-fno-implicit-templates}. @xref{Template
1392 Instantiation}, for more information.
1396 Disable generation of information about every class with virtual
1397 functions for use by the C++ runtime type identification features
1398 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1399 of the language, you can save some space by using this flag. Note that
1400 exception handling uses the same information, but it will generate it as
1405 Emit statistics about front-end processing at the end of the compilation.
1406 This information is generally only useful to the G++ development team.
1408 @item -ftemplate-depth-@var{n}
1409 @opindex ftemplate-depth
1410 Set the maximum instantiation depth for template classes to @var{n}.
1411 A limit on the template instantiation depth is needed to detect
1412 endless recursions during template class instantiation. ANSI/ISO C++
1413 conforming programs must not rely on a maximum depth greater than 17.
1415 @item -fuse-cxa-atexit
1416 @opindex fuse-cxa-atexit
1417 Register destructors for objects with static storage duration with the
1418 @code{__cxa_atexit} function rather than the @code{atexit} function.
1419 This option is required for fully standards-compliant handling of static
1420 destructors, but will only work if your C library supports
1421 @code{__cxa_atexit}.
1425 Do not use weak symbol support, even if it is provided by the linker.
1426 By default, G++ will use weak symbols if they are available. This
1427 option exists only for testing, and should not be used by end-users;
1428 it will result in inferior code and has no benefits. This option may
1429 be removed in a future release of G++.
1433 Do not search for header files in the standard directories specific to
1434 C++, but do still search the other standard directories. (This option
1435 is used when building the C++ library.)
1438 In addition, these optimization, warning, and code generation options
1439 have meanings only for C++ programs:
1442 @item -fno-default-inline
1443 @opindex fno-default-inline
1444 Do not assume @samp{inline} for functions defined inside a class scope.
1445 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1446 functions will have linkage like inline functions; they just won't be
1449 @item -Wabi @r{(C++ only)}
1451 Warn when G++ generates code that is probably not compatible with the
1452 vendor-neutral C++ ABI. Although an effort has been made to warn about
1453 all such cases, there are probably some cases that are not warned about,
1454 even though G++ is generating incompatible code. There may also be
1455 cases where warnings are emitted even though the code that is generated
1458 You should rewrite your code to avoid these warnings if you are
1459 concerned about the fact that code generated by G++ may not be binary
1460 compatible with code generated by other compilers.
1462 The known incompatibilities at this point include:
1467 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1468 pack data into the same byte as a base class. For example:
1471 struct A @{ virtual void f(); int f1 : 1; @};
1472 struct B : public A @{ int f2 : 1; @};
1476 In this case, G++ will place @code{B::f2} into the same byte
1477 as@code{A::f1}; other compilers will not. You can avoid this problem
1478 by explicitly padding @code{A} so that its size is a multiple of the
1479 byte size on your platform; that will cause G++ and other compilers to
1480 layout @code{B} identically.
1483 Incorrect handling of tail-padding for virtual bases. G++ does not use
1484 tail padding when laying out virtual bases. For example:
1487 struct A @{ virtual void f(); char c1; @};
1488 struct B @{ B(); char c2; @};
1489 struct C : public A, public virtual B @{@};
1493 In this case, G++ will not place @code{B} into the tail-padding for
1494 @code{A}; other compilers will. You can avoid this problem by
1495 explicitly padding @code{A} so that its size is a multiple of its
1496 alignment (ignoring virtual base classes); that will cause G++ and other
1497 compilers to layout @code{C} identically.
1500 Incorrect handling of bit-fields with declared widths greater than that
1501 of their underlying types, when the bit-fields appear in a union. For
1505 union U @{ int i : 4096; @};
1509 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1510 union too small by the number of bits in an @code{int}.
1513 Empty classes can be placed at incorrect offsets. For example:
1523 struct C : public B, public A @{@};
1527 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1528 it should be placed at offset zero. G++ mistakenly believes that the
1529 @code{A} data member of @code{B} is already at offset zero.
1532 Names of template functions whose types involve @code{typename} or
1533 template template parameters can be mangled incorrectly.
1536 template <typename Q>
1537 void f(typename Q::X) @{@}
1539 template <template <typename> class Q>
1540 void f(typename Q<int>::X) @{@}
1544 Instantiations of these templates may be mangled incorrectly.
1548 @item -Wctor-dtor-privacy @r{(C++ only)}
1549 @opindex Wctor-dtor-privacy
1550 Warn when a class seems unusable because all the constructors or
1551 destructors in that class are private, and it has neither friends nor
1552 public static member functions.
1554 @item -Wnon-virtual-dtor @r{(C++ only)}
1555 @opindex Wnon-virtual-dtor
1556 Warn when a class appears to be polymorphic, thereby requiring a virtual
1557 destructor, yet it declares a non-virtual one.
1558 This warning is enabled by @option{-Wall}.
1560 @item -Wreorder @r{(C++ only)}
1562 @cindex reordering, warning
1563 @cindex warning for reordering of member initializers
1564 Warn when the order of member initializers given in the code does not
1565 match the order in which they must be executed. For instance:
1571 A(): j (0), i (1) @{ @}
1575 The compiler will rearrange the member initializers for @samp{i}
1576 and @samp{j} to match the declaration order of the members, emitting
1577 a warning to that effect. This warning is enabled by @option{-Wall}.
1580 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1583 @item -Weffc++ @r{(C++ only)}
1585 Warn about violations of the following style guidelines from Scott Meyers'
1586 @cite{Effective C++} book:
1590 Item 11: Define a copy constructor and an assignment operator for classes
1591 with dynamically allocated memory.
1594 Item 12: Prefer initialization to assignment in constructors.
1597 Item 14: Make destructors virtual in base classes.
1600 Item 15: Have @code{operator=} return a reference to @code{*this}.
1603 Item 23: Don't try to return a reference when you must return an object.
1607 Also warn about violations of the following style guidelines from
1608 Scott Meyers' @cite{More Effective C++} book:
1612 Item 6: Distinguish between prefix and postfix forms of increment and
1613 decrement operators.
1616 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1620 When selecting this option, be aware that the standard library
1621 headers do not obey all of these guidelines; use @samp{grep -v}
1622 to filter out those warnings.
1624 @item -Wno-deprecated @r{(C++ only)}
1625 @opindex Wno-deprecated
1626 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1628 @item -Wno-non-template-friend @r{(C++ only)}
1629 @opindex Wno-non-template-friend
1630 Disable warnings when non-templatized friend functions are declared
1631 within a template. Since the advent of explicit template specification
1632 support in G++, if the name of the friend is an unqualified-id (i.e.,
1633 @samp{friend foo(int)}), the C++ language specification demands that the
1634 friend declare or define an ordinary, nontemplate function. (Section
1635 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1636 could be interpreted as a particular specialization of a templatized
1637 function. Because this non-conforming behavior is no longer the default
1638 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1639 check existing code for potential trouble spots and is on by default.
1640 This new compiler behavior can be turned off with
1641 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1642 but disables the helpful warning.
1644 @item -Wold-style-cast @r{(C++ only)}
1645 @opindex Wold-style-cast
1646 Warn if an old-style (C-style) cast to a non-void type is used within
1647 a C++ program. The new-style casts (@samp{static_cast},
1648 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1649 unintended effects and much easier to search for.
1651 @item -Woverloaded-virtual @r{(C++ only)}
1652 @opindex Woverloaded-virtual
1653 @cindex overloaded virtual fn, warning
1654 @cindex warning for overloaded virtual fn
1655 Warn when a function declaration hides virtual functions from a
1656 base class. For example, in:
1663 struct B: public A @{
1668 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1676 will fail to compile.
1678 @item -Wno-pmf-conversions @r{(C++ only)}
1679 @opindex Wno-pmf-conversions
1680 Disable the diagnostic for converting a bound pointer to member function
1683 @item -Wsign-promo @r{(C++ only)}
1684 @opindex Wsign-promo
1685 Warn when overload resolution chooses a promotion from unsigned or
1686 enumeral type to a signed type, over a conversion to an unsigned type of
1687 the same size. Previous versions of G++ would try to preserve
1688 unsignedness, but the standard mandates the current behavior.
1690 @item -Wsynth @r{(C++ only)}
1692 @cindex warning for synthesized methods
1693 @cindex synthesized methods, warning
1694 Warn when G++'s synthesis behavior does not match that of cfront. For
1700 A& operator = (int);
1710 In this example, G++ will synthesize a default @samp{A& operator =
1711 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1714 @node Objective-C Dialect Options
1715 @section Options Controlling Objective-C Dialect
1717 @cindex compiler options, Objective-C
1718 @cindex Objective-C options, command line
1719 @cindex options, Objective-C
1720 (NOTE: This manual does not describe the Objective-C language itself. See
1721 @w{@uref{http://gcc.gnu.org/readings.html}} for references.)
1723 This section describes the command-line options that are only meaningful
1724 for Objective-C programs, but you can also use most of the GNU compiler
1725 options regardless of what language your program is in. For example,
1726 you might compile a file @code{some_class.m} like this:
1729 gcc -g -fgnu-runtime -O -c some_class.m
1733 In this example, @option{-fgnu-runtime} is an option meant only for
1734 Objective-C programs; you can use the other options with any language
1737 Here is a list of options that are @emph{only} for compiling Objective-C
1741 @item -fconstant-string-class=@var{class-name}
1742 @opindex fconstant-string-class
1743 Use @var{class-name} as the name of the class to instantiate for each
1744 literal string specified with the syntax @code{@@"@dots{}"}. The default
1745 class name is @code{NXConstantString} if the GNU runtime is being used, and
1746 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1747 @option{-fconstant-cfstrings} option, if also present, will override the
1748 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1749 to be laid out as constant CoreFoundation strings.
1752 @opindex fgnu-runtime
1753 Generate object code compatible with the standard GNU Objective-C
1754 runtime. This is the default for most types of systems.
1756 @item -fnext-runtime
1757 @opindex fnext-runtime
1758 Generate output compatible with the NeXT runtime. This is the default
1759 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1760 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1763 @item -fno-nil-receivers
1764 @opindex -fno-nil-receivers
1765 Assume that all Objective-C message dispatches (e.g.,
1766 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1767 is not @code{nil}. This allows for more efficient entry points in the runtime to be
1768 used. Currently, this option is only available in conjunction with
1769 the NeXT runtime on Mac OS X 10.3 and later.
1771 @item -fobjc-exceptions
1772 @opindex -fobjc-exceptions
1773 Enable syntactic support for structured exception handling in Objective-C,
1774 similar to what is offered by C++ and Java. Currently, this option is only
1775 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1783 @@catch (AnObjCClass *exc) @{
1790 @@catch (AnotherClass *exc) @{
1793 @@catch (id allOthers) @{
1803 The @code{@@throw} statement may appear anywhere in an Objective-C or
1804 Objective-C++ program; when used inside of a @code{@@catch} block, the
1805 @code{@@throw} may appear without an argument (as shown above), in which case
1806 the object caught by the @code{@@catch} will be rethrown.
1808 Note that only (pointers to) Objective-C objects may be thrown and
1809 caught using this scheme. When an object is thrown, it will be caught
1810 by the nearest @code{@@catch} clause capable of handling objects of that type,
1811 analogously to how @code{catch} blocks work in C++ and Java. A
1812 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1813 any and all Objective-C exceptions not caught by previous @code{@@catch}
1816 The @code{@@finally} clause, if present, will be executed upon exit from the
1817 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1818 regardless of whether any exceptions are thrown, caught or rethrown
1819 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1820 of the @code{finally} clause in Java.
1822 There are several caveats to using the new exception mechanism:
1826 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1827 idioms provided by the @code{NSException} class, the new
1828 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1829 systems, due to additional functionality needed in the (NeXT) Objective-C
1833 As mentioned above, the new exceptions do not support handling
1834 types other than Objective-C objects. Furthermore, when used from
1835 Objective-C++, the Objective-C exception model does not interoperate with C++
1836 exceptions at this time. This means you cannot @code{@@throw} an exception
1837 from Objective-C and @code{catch} it in C++, or vice versa
1838 (i.e., @code{throw @dots{} @@catch}).
1841 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1842 blocks for thread-safe execution:
1845 @@synchronized (ObjCClass *guard) @{
1850 Upon entering the @code{@@synchronized} block, a thread of execution shall
1851 first check whether a lock has been placed on the corresponding @code{guard}
1852 object by another thread. If it has, the current thread shall wait until
1853 the other thread relinquishes its lock. Once @code{guard} becomes available,
1854 the current thread will place its own lock on it, execute the code contained in
1855 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1856 making @code{guard} available to other threads).
1858 Unlike Java, Objective-C does not allow for entire methods to be marked
1859 @code{@@synchronized}. Note that throwing exceptions out of
1860 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1861 to be unlocked properly.
1863 @item -freplace-objc-classes
1864 @opindex -freplace-objc-classes
1865 Emit a special marker instructing @command{ld(1)} not to statically link in
1866 the resulting object file, and allow @command{dyld(1)} to load it in at
1867 run time instead. This is used in conjunction with the Fix-and-Continue
1868 debugging mode, where the object file in question may be recompiled and
1869 dynamically reloaded in the course of program execution, without the need
1870 to restart the program itself. Currently, Fix-and-Continue functionality
1871 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
1875 @opindex -fzero-link
1876 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1877 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1878 compile time) with static class references that get initialized at load time,
1879 which improves run-time performance. Specifying the @option{-fzero-link} flag
1880 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1881 to be retained. This is useful in Zero-Link debugging mode, since it allows
1882 for individual class implementations to be modified during program execution.
1886 Dump interface declarations for all classes seen in the source file to a
1887 file named @file{@var{sourcename}.decl}.
1890 @opindex Wno-protocol
1891 If a class is declared to implement a protocol, a warning is issued for
1892 every method in the protocol that is not implemented by the class. The
1893 default behavior is to issue a warning for every method not explicitly
1894 implemented in the class, even if a method implementation is inherited
1895 from the superclass. If you use the @code{-Wno-protocol} option, then
1896 methods inherited from the superclass are considered to be implemented,
1897 and no warning is issued for them.
1901 Warn if multiple methods of different types for the same selector are
1902 found during compilation. The check is performed on the list of methods
1903 in the final stage of compilation. Additionally, a check is performed
1904 for each selector appearing in a @code{@@selector(@dots{})}
1905 expression, and a corresponding method for that selector has been found
1906 during compilation. Because these checks scan the method table only at
1907 the end of compilation, these warnings are not produced if the final
1908 stage of compilation is not reached, for example because an error is
1909 found during compilation, or because the @code{-fsyntax-only} option is
1912 @item -Wundeclared-selector
1913 @opindex Wundeclared-selector
1914 Warn if a @code{@@selector(@dots{})} expression referring to an
1915 undeclared selector is found. A selector is considered undeclared if no
1916 method with that name has been declared before the
1917 @code{@@selector(@dots{})} expression, either explicitly in an
1918 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
1919 an @code{@@implementation} section. This option always performs its
1920 checks as soon as a @code{@@selector(@dots{})} expression is found,
1921 while @code{-Wselector} only performs its checks in the final stage of
1922 compilation. This also enforces the coding style convention
1923 that methods and selectors must be declared before being used.
1925 @item -print-objc-runtime-info
1926 @opindex -print-objc-runtime-info
1927 Generate C header describing the largest structure that is passed by
1932 @node Language Independent Options
1933 @section Options to Control Diagnostic Messages Formatting
1934 @cindex options to control diagnostics formatting
1935 @cindex diagnostic messages
1936 @cindex message formatting
1938 Traditionally, diagnostic messages have been formatted irrespective of
1939 the output device's aspect (e.g.@: its width, @dots{}). The options described
1940 below can be used to control the diagnostic messages formatting
1941 algorithm, e.g.@: how many characters per line, how often source location
1942 information should be reported. Right now, only the C++ front end can
1943 honor these options. However it is expected, in the near future, that
1944 the remaining front ends would be able to digest them correctly.
1947 @item -fmessage-length=@var{n}
1948 @opindex fmessage-length
1949 Try to format error messages so that they fit on lines of about @var{n}
1950 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1951 the front ends supported by GCC@. If @var{n} is zero, then no
1952 line-wrapping will be done; each error message will appear on a single
1955 @opindex fdiagnostics-show-location
1956 @item -fdiagnostics-show-location=once
1957 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1958 reporter to emit @emph{once} source location information; that is, in
1959 case the message is too long to fit on a single physical line and has to
1960 be wrapped, the source location won't be emitted (as prefix) again,
1961 over and over, in subsequent continuation lines. This is the default
1964 @item -fdiagnostics-show-location=every-line
1965 Only meaningful in line-wrapping mode. Instructs the diagnostic
1966 messages reporter to emit the same source location information (as
1967 prefix) for physical lines that result from the process of breaking
1968 a message which is too long to fit on a single line.
1972 @node Warning Options
1973 @section Options to Request or Suppress Warnings
1974 @cindex options to control warnings
1975 @cindex warning messages
1976 @cindex messages, warning
1977 @cindex suppressing warnings
1979 Warnings are diagnostic messages that report constructions which
1980 are not inherently erroneous but which are risky or suggest there
1981 may have been an error.
1983 You can request many specific warnings with options beginning @samp{-W},
1984 for example @option{-Wimplicit} to request warnings on implicit
1985 declarations. Each of these specific warning options also has a
1986 negative form beginning @samp{-Wno-} to turn off warnings;
1987 for example, @option{-Wno-implicit}. This manual lists only one of the
1988 two forms, whichever is not the default.
1990 The following options control the amount and kinds of warnings produced
1991 by GCC; for further, language-specific options also refer to
1992 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
1995 @cindex syntax checking
1997 @opindex fsyntax-only
1998 Check the code for syntax errors, but don't do anything beyond that.
2002 Issue all the warnings demanded by strict ISO C and ISO C++;
2003 reject all programs that use forbidden extensions, and some other
2004 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2005 version of the ISO C standard specified by any @option{-std} option used.
2007 Valid ISO C and ISO C++ programs should compile properly with or without
2008 this option (though a rare few will require @option{-ansi} or a
2009 @option{-std} option specifying the required version of ISO C)@. However,
2010 without this option, certain GNU extensions and traditional C and C++
2011 features are supported as well. With this option, they are rejected.
2013 @option{-pedantic} does not cause warning messages for use of the
2014 alternate keywords whose names begin and end with @samp{__}. Pedantic
2015 warnings are also disabled in the expression that follows
2016 @code{__extension__}. However, only system header files should use
2017 these escape routes; application programs should avoid them.
2018 @xref{Alternate Keywords}.
2020 Some users try to use @option{-pedantic} to check programs for strict ISO
2021 C conformance. They soon find that it does not do quite what they want:
2022 it finds some non-ISO practices, but not all---only those for which
2023 ISO C @emph{requires} a diagnostic, and some others for which
2024 diagnostics have been added.
2026 A feature to report any failure to conform to ISO C might be useful in
2027 some instances, but would require considerable additional work and would
2028 be quite different from @option{-pedantic}. We don't have plans to
2029 support such a feature in the near future.
2031 Where the standard specified with @option{-std} represents a GNU
2032 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2033 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2034 extended dialect is based. Warnings from @option{-pedantic} are given
2035 where they are required by the base standard. (It would not make sense
2036 for such warnings to be given only for features not in the specified GNU
2037 C dialect, since by definition the GNU dialects of C include all
2038 features the compiler supports with the given option, and there would be
2039 nothing to warn about.)
2041 @item -pedantic-errors
2042 @opindex pedantic-errors
2043 Like @option{-pedantic}, except that errors are produced rather than
2048 Inhibit all warning messages.
2052 Inhibit warning messages about the use of @samp{#import}.
2054 @item -Wchar-subscripts
2055 @opindex Wchar-subscripts
2056 Warn if an array subscript has type @code{char}. This is a common cause
2057 of error, as programmers often forget that this type is signed on some
2062 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2063 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2067 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2068 the arguments supplied have types appropriate to the format string
2069 specified, and that the conversions specified in the format string make
2070 sense. This includes standard functions, and others specified by format
2071 attributes (@pxref{Function Attributes}), in the @code{printf},
2072 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2073 not in the C standard) families.
2075 The formats are checked against the format features supported by GNU
2076 libc version 2.2. These include all ISO C90 and C99 features, as well
2077 as features from the Single Unix Specification and some BSD and GNU
2078 extensions. Other library implementations may not support all these
2079 features; GCC does not support warning about features that go beyond a
2080 particular library's limitations. However, if @option{-pedantic} is used
2081 with @option{-Wformat}, warnings will be given about format features not
2082 in the selected standard version (but not for @code{strfmon} formats,
2083 since those are not in any version of the C standard). @xref{C Dialect
2084 Options,,Options Controlling C Dialect}.
2086 Since @option{-Wformat} also checks for null format arguments for
2087 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2089 @option{-Wformat} is included in @option{-Wall}. For more control over some
2090 aspects of format checking, the options @option{-Wformat-y2k},
2091 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2092 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2093 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2096 @opindex Wformat-y2k
2097 If @option{-Wformat} is specified, also warn about @code{strftime}
2098 formats which may yield only a two-digit year.
2100 @item -Wno-format-extra-args
2101 @opindex Wno-format-extra-args
2102 If @option{-Wformat} is specified, do not warn about excess arguments to a
2103 @code{printf} or @code{scanf} format function. The C standard specifies
2104 that such arguments are ignored.
2106 Where the unused arguments lie between used arguments that are
2107 specified with @samp{$} operand number specifications, normally
2108 warnings are still given, since the implementation could not know what
2109 type to pass to @code{va_arg} to skip the unused arguments. However,
2110 in the case of @code{scanf} formats, this option will suppress the
2111 warning if the unused arguments are all pointers, since the Single
2112 Unix Specification says that such unused arguments are allowed.
2114 @item -Wno-format-zero-length
2115 @opindex Wno-format-zero-length
2116 If @option{-Wformat} is specified, do not warn about zero-length formats.
2117 The C standard specifies that zero-length formats are allowed.
2119 @item -Wformat-nonliteral
2120 @opindex Wformat-nonliteral
2121 If @option{-Wformat} is specified, also warn if the format string is not a
2122 string literal and so cannot be checked, unless the format function
2123 takes its format arguments as a @code{va_list}.
2125 @item -Wformat-security
2126 @opindex Wformat-security
2127 If @option{-Wformat} is specified, also warn about uses of format
2128 functions that represent possible security problems. At present, this
2129 warns about calls to @code{printf} and @code{scanf} functions where the
2130 format string is not a string literal and there are no format arguments,
2131 as in @code{printf (foo);}. This may be a security hole if the format
2132 string came from untrusted input and contains @samp{%n}. (This is
2133 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2134 in future warnings may be added to @option{-Wformat-security} that are not
2135 included in @option{-Wformat-nonliteral}.)
2139 Enable @option{-Wformat} plus format checks not included in
2140 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2141 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2145 Warn about passing a null pointer for arguments marked as
2146 requiring a non-null value by the @code{nonnull} function attribute.
2148 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2149 can be disabled with the @option{-Wno-nonnull} option.
2151 @item -Winit-self @r{(C, C++, and Objective-C only)}
2153 Warn about uninitialized variables which are initialized with themselves.
2154 Note this option can only be used with the @option{-Wuninitialized} option,
2155 which in turn only works with @option{-O1} and above.
2157 For example, GCC will warn about @code{i} being uninitialized in the
2158 following snippet only when @option{-Winit-self} has been specified:
2169 @item -Wimplicit-int
2170 @opindex Wimplicit-int
2171 Warn when a declaration does not specify a type.
2173 @item -Wimplicit-function-declaration
2174 @itemx -Werror-implicit-function-declaration
2175 @opindex Wimplicit-function-declaration
2176 @opindex Werror-implicit-function-declaration
2177 Give a warning (or error) whenever a function is used before being
2182 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2186 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2187 function with external linkage, returning int, taking either zero
2188 arguments, two, or three arguments of appropriate types.
2190 @item -Wmissing-braces
2191 @opindex Wmissing-braces
2192 Warn if an aggregate or union initializer is not fully bracketed. In
2193 the following example, the initializer for @samp{a} is not fully
2194 bracketed, but that for @samp{b} is fully bracketed.
2197 int a[2][2] = @{ 0, 1, 2, 3 @};
2198 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2202 @opindex Wparentheses
2203 Warn if parentheses are omitted in certain contexts, such
2204 as when there is an assignment in a context where a truth value
2205 is expected, or when operators are nested whose precedence people
2206 often get confused about.
2208 Also warn about constructions where there may be confusion to which
2209 @code{if} statement an @code{else} branch belongs. Here is an example of
2224 In C, every @code{else} branch belongs to the innermost possible @code{if}
2225 statement, which in this example is @code{if (b)}. This is often not
2226 what the programmer expected, as illustrated in the above example by
2227 indentation the programmer chose. When there is the potential for this
2228 confusion, GCC will issue a warning when this flag is specified.
2229 To eliminate the warning, add explicit braces around the innermost
2230 @code{if} statement so there is no way the @code{else} could belong to
2231 the enclosing @code{if}. The resulting code would look like this:
2247 @item -Wsequence-point
2248 @opindex Wsequence-point
2249 Warn about code that may have undefined semantics because of violations
2250 of sequence point rules in the C standard.
2252 The C standard defines the order in which expressions in a C program are
2253 evaluated in terms of @dfn{sequence points}, which represent a partial
2254 ordering between the execution of parts of the program: those executed
2255 before the sequence point, and those executed after it. These occur
2256 after the evaluation of a full expression (one which is not part of a
2257 larger expression), after the evaluation of the first operand of a
2258 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2259 function is called (but after the evaluation of its arguments and the
2260 expression denoting the called function), and in certain other places.
2261 Other than as expressed by the sequence point rules, the order of
2262 evaluation of subexpressions of an expression is not specified. All
2263 these rules describe only a partial order rather than a total order,
2264 since, for example, if two functions are called within one expression
2265 with no sequence point between them, the order in which the functions
2266 are called is not specified. However, the standards committee have
2267 ruled that function calls do not overlap.
2269 It is not specified when between sequence points modifications to the
2270 values of objects take effect. Programs whose behavior depends on this
2271 have undefined behavior; the C standard specifies that ``Between the
2272 previous and next sequence point an object shall have its stored value
2273 modified at most once by the evaluation of an expression. Furthermore,
2274 the prior value shall be read only to determine the value to be
2275 stored.''. If a program breaks these rules, the results on any
2276 particular implementation are entirely unpredictable.
2278 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2279 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2280 diagnosed by this option, and it may give an occasional false positive
2281 result, but in general it has been found fairly effective at detecting
2282 this sort of problem in programs.
2284 The present implementation of this option only works for C programs. A
2285 future implementation may also work for C++ programs.
2287 The C standard is worded confusingly, therefore there is some debate
2288 over the precise meaning of the sequence point rules in subtle cases.
2289 Links to discussions of the problem, including proposed formal
2290 definitions, may be found on our readings page, at
2291 @w{@uref{http://gcc.gnu.org/readings.html}}.
2294 @opindex Wreturn-type
2295 Warn whenever a function is defined with a return-type that defaults to
2296 @code{int}. Also warn about any @code{return} statement with no
2297 return-value in a function whose return-type is not @code{void}.
2299 For C++, a function without return type always produces a diagnostic
2300 message, even when @option{-Wno-return-type} is specified. The only
2301 exceptions are @samp{main} and functions defined in system headers.
2305 Warn whenever a @code{switch} statement has an index of enumeral type
2306 and lacks a @code{case} for one or more of the named codes of that
2307 enumeration. (The presence of a @code{default} label prevents this
2308 warning.) @code{case} labels outside the enumeration range also
2309 provoke warnings when this option is used.
2311 @item -Wswitch-default
2312 @opindex Wswitch-switch
2313 Warn whenever a @code{switch} statement does not have a @code{default}
2317 @opindex Wswitch-enum
2318 Warn whenever a @code{switch} statement has an index of enumeral type
2319 and lacks a @code{case} for one or more of the named codes of that
2320 enumeration. @code{case} labels outside the enumeration range also
2321 provoke warnings when this option is used.
2325 Warn if any trigraphs are encountered that might change the meaning of
2326 the program (trigraphs within comments are not warned about).
2328 @item -Wunused-function
2329 @opindex Wunused-function
2330 Warn whenever a static function is declared but not defined or a
2331 non\-inline static function is unused.
2333 @item -Wunused-label
2334 @opindex Wunused-label
2335 Warn whenever a label is declared but not used.
2337 To suppress this warning use the @samp{unused} attribute
2338 (@pxref{Variable Attributes}).
2340 @item -Wunused-parameter
2341 @opindex Wunused-parameter
2342 Warn whenever a function parameter is unused aside from its declaration.
2344 To suppress this warning use the @samp{unused} attribute
2345 (@pxref{Variable Attributes}).
2347 @item -Wunused-variable
2348 @opindex Wunused-variable
2349 Warn whenever a local variable or non-constant static variable is unused
2350 aside from its declaration
2352 To suppress this warning use the @samp{unused} attribute
2353 (@pxref{Variable Attributes}).
2355 @item -Wunused-value
2356 @opindex Wunused-value
2357 Warn whenever a statement computes a result that is explicitly not used.
2359 To suppress this warning cast the expression to @samp{void}.
2363 All the above @option{-Wunused} options combined.
2365 In order to get a warning about an unused function parameter, you must
2366 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2367 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2369 @item -Wuninitialized
2370 @opindex Wuninitialized
2371 Warn if an automatic variable is used without first being initialized or
2372 if a variable may be clobbered by a @code{setjmp} call.
2374 These warnings are possible only in optimizing compilation,
2375 because they require data flow information that is computed only
2376 when optimizing. If you don't specify @option{-O}, you simply won't
2379 If you want to warn about code which uses the uninitialized value of the
2380 variable in its own initializer, use the @option{-Winit-self} option.
2382 These warnings occur only for variables that are candidates for
2383 register allocation. Therefore, they do not occur for a variable that
2384 is declared @code{volatile}, or whose address is taken, or whose size
2385 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2386 structures, unions or arrays, even when they are in registers.
2388 Note that there may be no warning about a variable that is used only
2389 to compute a value that itself is never used, because such
2390 computations may be deleted by data flow analysis before the warnings
2393 These warnings are made optional because GCC is not smart
2394 enough to see all the reasons why the code might be correct
2395 despite appearing to have an error. Here is one example of how
2416 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2417 always initialized, but GCC doesn't know this. Here is
2418 another common case:
2423 if (change_y) save_y = y, y = new_y;
2425 if (change_y) y = save_y;
2430 This has no bug because @code{save_y} is used only if it is set.
2432 @cindex @code{longjmp} warnings
2433 This option also warns when a non-volatile automatic variable might be
2434 changed by a call to @code{longjmp}. These warnings as well are possible
2435 only in optimizing compilation.
2437 The compiler sees only the calls to @code{setjmp}. It cannot know
2438 where @code{longjmp} will be called; in fact, a signal handler could
2439 call it at any point in the code. As a result, you may get a warning
2440 even when there is in fact no problem because @code{longjmp} cannot
2441 in fact be called at the place which would cause a problem.
2443 Some spurious warnings can be avoided if you declare all the functions
2444 you use that never return as @code{noreturn}. @xref{Function
2447 @item -Wunknown-pragmas
2448 @opindex Wunknown-pragmas
2449 @cindex warning for unknown pragmas
2450 @cindex unknown pragmas, warning
2451 @cindex pragmas, warning of unknown
2452 Warn when a #pragma directive is encountered which is not understood by
2453 GCC@. If this command line option is used, warnings will even be issued
2454 for unknown pragmas in system header files. This is not the case if
2455 the warnings were only enabled by the @option{-Wall} command line option.
2457 @item -Wstrict-aliasing
2458 @opindex Wstrict-aliasing
2459 This option is only active when @option{-fstrict-aliasing} is active.
2460 It warns about code which might break the strict aliasing rules that the
2461 compiler is using for optimization. The warning does not catch all
2462 cases, but does attempt to catch the more common pitfalls. It is
2463 included in @option{-Wall}.
2465 @item -Wstrict-aliasing=2
2466 @opindex Wstrict-aliasing=2
2467 This option is only active when @option{-fstrict-aliasing} is active.
2468 It warns about all code which might break the strict aliasing rules that the
2469 compiler is using for optimization. This warning catches all cases, but
2470 it will also give a warning for some ambiguous cases that are safe.
2474 All of the above @samp{-W} options combined. This enables all the
2475 warnings about constructions that some users consider questionable, and
2476 that are easy to avoid (or modify to prevent the warning), even in
2477 conjunction with macros. This also enables some language-specific
2478 warnings described in @ref{C++ Dialect Options} and
2479 @ref{Objective-C Dialect Options}.
2482 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2483 Some of them warn about constructions that users generally do not
2484 consider questionable, but which occasionally you might wish to check
2485 for; others warn about constructions that are necessary or hard to avoid
2486 in some cases, and there is no simple way to modify the code to suppress
2493 (This option used to be called @option{-W}. The older name is still
2494 supported, but the newer name is more descriptive.) Print extra warning
2495 messages for these events:
2499 A function can return either with or without a value. (Falling
2500 off the end of the function body is considered returning without
2501 a value.) For example, this function would evoke such a
2515 An expression-statement or the left-hand side of a comma expression
2516 contains no side effects.
2517 To suppress the warning, cast the unused expression to void.
2518 For example, an expression such as @samp{x[i,j]} will cause a warning,
2519 but @samp{x[(void)i,j]} will not.
2522 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2525 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2526 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2527 that of ordinary mathematical notation.
2530 Storage-class specifiers like @code{static} are not the first things in
2531 a declaration. According to the C Standard, this usage is obsolescent.
2534 The return type of a function has a type qualifier such as @code{const}.
2535 Such a type qualifier has no effect, since the value returned by a
2536 function is not an lvalue. (But don't warn about the GNU extension of
2537 @code{volatile void} return types. That extension will be warned about
2538 if @option{-pedantic} is specified.)
2541 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2545 A comparison between signed and unsigned values could produce an
2546 incorrect result when the signed value is converted to unsigned.
2547 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2550 An aggregate has an initializer which does not initialize all members.
2551 For example, the following code would cause such a warning, because
2552 @code{x.h} would be implicitly initialized to zero:
2555 struct s @{ int f, g, h; @};
2556 struct s x = @{ 3, 4 @};
2560 A function parameter is declared without a type specifier in K&R-style
2568 An empty body occurs in an @samp{if} or @samp{else} statement.
2571 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2572 @samp{>}, or @samp{>=}.
2575 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2578 Any of several floating-point events that often indicate errors, such as
2579 overflow, underflow, loss of precision, etc.
2581 @item @r{(C++ only)}
2582 An enumerator and a non-enumerator both appear in a conditional expression.
2584 @item @r{(C++ only)}
2585 A non-static reference or non-static @samp{const} member appears in a
2586 class without constructors.
2588 @item @r{(C++ only)}
2589 Ambiguous virtual bases.
2591 @item @r{(C++ only)}
2592 Subscripting an array which has been declared @samp{register}.
2594 @item @r{(C++ only)}
2595 Taking the address of a variable which has been declared @samp{register}.
2597 @item @r{(C++ only)}
2598 A base class is not initialized in a derived class' copy constructor.
2601 @item -Wno-div-by-zero
2602 @opindex Wno-div-by-zero
2603 @opindex Wdiv-by-zero
2604 Do not warn about compile-time integer division by zero. Floating point
2605 division by zero is not warned about, as it can be a legitimate way of
2606 obtaining infinities and NaNs.
2608 @item -Wsystem-headers
2609 @opindex Wsystem-headers
2610 @cindex warnings from system headers
2611 @cindex system headers, warnings from
2612 Print warning messages for constructs found in system header files.
2613 Warnings from system headers are normally suppressed, on the assumption
2614 that they usually do not indicate real problems and would only make the
2615 compiler output harder to read. Using this command line option tells
2616 GCC to emit warnings from system headers as if they occurred in user
2617 code. However, note that using @option{-Wall} in conjunction with this
2618 option will @emph{not} warn about unknown pragmas in system
2619 headers---for that, @option{-Wunknown-pragmas} must also be used.
2622 @opindex Wfloat-equal
2623 Warn if floating point values are used in equality comparisons.
2625 The idea behind this is that sometimes it is convenient (for the
2626 programmer) to consider floating-point values as approximations to
2627 infinitely precise real numbers. If you are doing this, then you need
2628 to compute (by analyzing the code, or in some other way) the maximum or
2629 likely maximum error that the computation introduces, and allow for it
2630 when performing comparisons (and when producing output, but that's a
2631 different problem). In particular, instead of testing for equality, you
2632 would check to see whether the two values have ranges that overlap; and
2633 this is done with the relational operators, so equality comparisons are
2636 @item -Wtraditional @r{(C only)}
2637 @opindex Wtraditional
2638 Warn about certain constructs that behave differently in traditional and
2639 ISO C@. Also warn about ISO C constructs that have no traditional C
2640 equivalent, and/or problematic constructs which should be avoided.
2644 Macro parameters that appear within string literals in the macro body.
2645 In traditional C macro replacement takes place within string literals,
2646 but does not in ISO C@.
2649 In traditional C, some preprocessor directives did not exist.
2650 Traditional preprocessors would only consider a line to be a directive
2651 if the @samp{#} appeared in column 1 on the line. Therefore
2652 @option{-Wtraditional} warns about directives that traditional C
2653 understands but would ignore because the @samp{#} does not appear as the
2654 first character on the line. It also suggests you hide directives like
2655 @samp{#pragma} not understood by traditional C by indenting them. Some
2656 traditional implementations would not recognize @samp{#elif}, so it
2657 suggests avoiding it altogether.
2660 A function-like macro that appears without arguments.
2663 The unary plus operator.
2666 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2667 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2668 constants.) Note, these suffixes appear in macros defined in the system
2669 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2670 Use of these macros in user code might normally lead to spurious
2671 warnings, however GCC's integrated preprocessor has enough context to
2672 avoid warning in these cases.
2675 A function declared external in one block and then used after the end of
2679 A @code{switch} statement has an operand of type @code{long}.
2682 A non-@code{static} function declaration follows a @code{static} one.
2683 This construct is not accepted by some traditional C compilers.
2686 The ISO type of an integer constant has a different width or
2687 signedness from its traditional type. This warning is only issued if
2688 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2689 typically represent bit patterns, are not warned about.
2692 Usage of ISO string concatenation is detected.
2695 Initialization of automatic aggregates.
2698 Identifier conflicts with labels. Traditional C lacks a separate
2699 namespace for labels.
2702 Initialization of unions. If the initializer is zero, the warning is
2703 omitted. This is done under the assumption that the zero initializer in
2704 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2705 initializer warnings and relies on default initialization to zero in the
2709 Conversions by prototypes between fixed/floating point values and vice
2710 versa. The absence of these prototypes when compiling with traditional
2711 C would cause serious problems. This is a subset of the possible
2712 conversion warnings, for the full set use @option{-Wconversion}.
2715 Use of ISO C style function definitions. This warning intentionally is
2716 @emph{not} issued for prototype declarations or variadic functions
2717 because these ISO C features will appear in your code when using
2718 libiberty's traditional C compatibility macros, @code{PARAMS} and
2719 @code{VPARAMS}. This warning is also bypassed for nested functions
2720 because that feature is already a GCC extension and thus not relevant to
2721 traditional C compatibility.
2724 @item -Wdeclaration-after-statement @r{(C only)}
2725 @opindex Wdeclaration-after-statement
2726 Warn when a declaration is found after a statement in a block. This
2727 construct, known from C++, was introduced with ISO C99 and is by default
2728 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2729 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2733 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2735 @item -Wendif-labels
2736 @opindex Wendif-labels
2737 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2741 Warn whenever a local variable shadows another local variable, parameter or
2742 global variable or whenever a built-in function is shadowed.
2744 @item -Wlarger-than-@var{len}
2745 @opindex Wlarger-than
2746 Warn whenever an object of larger than @var{len} bytes is defined.
2748 @item -Wpointer-arith
2749 @opindex Wpointer-arith
2750 Warn about anything that depends on the ``size of'' a function type or
2751 of @code{void}. GNU C assigns these types a size of 1, for
2752 convenience in calculations with @code{void *} pointers and pointers
2755 @item -Wbad-function-cast @r{(C only)}
2756 @opindex Wbad-function-cast
2757 Warn whenever a function call is cast to a non-matching type.
2758 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2762 Warn whenever a pointer is cast so as to remove a type qualifier from
2763 the target type. For example, warn if a @code{const char *} is cast
2764 to an ordinary @code{char *}.
2767 @opindex Wcast-align
2768 Warn whenever a pointer is cast such that the required alignment of the
2769 target is increased. For example, warn if a @code{char *} is cast to
2770 an @code{int *} on machines where integers can only be accessed at
2771 two- or four-byte boundaries.
2773 @item -Wwrite-strings
2774 @opindex Wwrite-strings
2775 When compiling C, give string constants the type @code{const
2776 char[@var{length}]} so that
2777 copying the address of one into a non-@code{const} @code{char *}
2778 pointer will get a warning; when compiling C++, warn about the
2779 deprecated conversion from string constants to @code{char *}.
2780 These warnings will help you find at
2781 compile time code that can try to write into a string constant, but
2782 only if you have been very careful about using @code{const} in
2783 declarations and prototypes. Otherwise, it will just be a nuisance;
2784 this is why we did not make @option{-Wall} request these warnings.
2787 @opindex Wconversion
2788 Warn if a prototype causes a type conversion that is different from what
2789 would happen to the same argument in the absence of a prototype. This
2790 includes conversions of fixed point to floating and vice versa, and
2791 conversions changing the width or signedness of a fixed point argument
2792 except when the same as the default promotion.
2794 Also, warn if a negative integer constant expression is implicitly
2795 converted to an unsigned type. For example, warn about the assignment
2796 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2797 casts like @code{(unsigned) -1}.
2799 @item -Wsign-compare
2800 @opindex Wsign-compare
2801 @cindex warning for comparison of signed and unsigned values
2802 @cindex comparison of signed and unsigned values, warning
2803 @cindex signed and unsigned values, comparison warning
2804 Warn when a comparison between signed and unsigned values could produce
2805 an incorrect result when the signed value is converted to unsigned.
2806 This warning is also enabled by @option{-Wextra}; to get the other warnings
2807 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2809 @item -Waggregate-return
2810 @opindex Waggregate-return
2811 Warn if any functions that return structures or unions are defined or
2812 called. (In languages where you can return an array, this also elicits
2815 @item -Wstrict-prototypes @r{(C only)}
2816 @opindex Wstrict-prototypes
2817 Warn if a function is declared or defined without specifying the
2818 argument types. (An old-style function definition is permitted without
2819 a warning if preceded by a declaration which specifies the argument
2822 @item -Wold-style-definition @r{(C only)}
2823 @opindex Wold-style-definition
2824 Warn if an old-style function definition is used. A warning is given
2825 even if there is a previous prototype.
2827 @item -Wmissing-prototypes @r{(C only)}
2828 @opindex Wmissing-prototypes
2829 Warn if a global function is defined without a previous prototype
2830 declaration. This warning is issued even if the definition itself
2831 provides a prototype. The aim is to detect global functions that fail
2832 to be declared in header files.
2834 @item -Wmissing-declarations @r{(C only)}
2835 @opindex Wmissing-declarations
2836 Warn if a global function is defined without a previous declaration.
2837 Do so even if the definition itself provides a prototype.
2838 Use this option to detect global functions that are not declared in
2841 @item -Wmissing-noreturn
2842 @opindex Wmissing-noreturn
2843 Warn about functions which might be candidates for attribute @code{noreturn}.
2844 Note these are only possible candidates, not absolute ones. Care should
2845 be taken to manually verify functions actually do not ever return before
2846 adding the @code{noreturn} attribute, otherwise subtle code generation
2847 bugs could be introduced. You will not get a warning for @code{main} in
2848 hosted C environments.
2850 @item -Wmissing-format-attribute
2851 @opindex Wmissing-format-attribute
2853 If @option{-Wformat} is enabled, also warn about functions which might be
2854 candidates for @code{format} attributes. Note these are only possible
2855 candidates, not absolute ones. GCC will guess that @code{format}
2856 attributes might be appropriate for any function that calls a function
2857 like @code{vprintf} or @code{vscanf}, but this might not always be the
2858 case, and some functions for which @code{format} attributes are
2859 appropriate may not be detected. This option has no effect unless
2860 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2862 @item -Wno-multichar
2863 @opindex Wno-multichar
2865 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2866 Usually they indicate a typo in the user's code, as they have
2867 implementation-defined values, and should not be used in portable code.
2869 @item -Wno-deprecated-declarations
2870 @opindex Wno-deprecated-declarations
2871 Do not warn about uses of functions, variables, and types marked as
2872 deprecated by using the @code{deprecated} attribute.
2873 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2874 @pxref{Type Attributes}.)
2878 Warn if a structure is given the packed attribute, but the packed
2879 attribute has no effect on the layout or size of the structure.
2880 Such structures may be mis-aligned for little benefit. For
2881 instance, in this code, the variable @code{f.x} in @code{struct bar}
2882 will be misaligned even though @code{struct bar} does not itself
2883 have the packed attribute:
2890 @} __attribute__((packed));
2900 Warn if padding is included in a structure, either to align an element
2901 of the structure or to align the whole structure. Sometimes when this
2902 happens it is possible to rearrange the fields of the structure to
2903 reduce the padding and so make the structure smaller.
2905 @item -Wredundant-decls
2906 @opindex Wredundant-decls
2907 Warn if anything is declared more than once in the same scope, even in
2908 cases where multiple declaration is valid and changes nothing.
2910 @item -Wnested-externs @r{(C only)}
2911 @opindex Wnested-externs
2912 Warn if an @code{extern} declaration is encountered within a function.
2914 @item -Wunreachable-code
2915 @opindex Wunreachable-code
2916 Warn if the compiler detects that code will never be executed.
2918 This option is intended to warn when the compiler detects that at
2919 least a whole line of source code will never be executed, because
2920 some condition is never satisfied or because it is after a
2921 procedure that never returns.
2923 It is possible for this option to produce a warning even though there
2924 are circumstances under which part of the affected line can be executed,
2925 so care should be taken when removing apparently-unreachable code.
2927 For instance, when a function is inlined, a warning may mean that the
2928 line is unreachable in only one inlined copy of the function.
2930 This option is not made part of @option{-Wall} because in a debugging
2931 version of a program there is often substantial code which checks
2932 correct functioning of the program and is, hopefully, unreachable
2933 because the program does work. Another common use of unreachable
2934 code is to provide behavior which is selectable at compile-time.
2938 Warn if a function can not be inlined and it was declared as inline.
2939 Even with this option, the compiler will not warn about failures to
2940 inline functions declared in system headers.
2942 The compiler uses a variety of heuristics to determine whether or not
2943 to inline a function. For example, the compiler takes into account
2944 the size of the function being inlined and the the amount of inlining
2945 that has already been done in the current function. Therefore,
2946 seemingly insignificant changes in the source program can cause the
2947 warnings produced by @option{-Winline} to appear or disappear.
2949 @item -Wno-invalid-offsetof @r{(C++ only)}
2950 @opindex Wno-invalid-offsetof
2951 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
2952 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
2953 to a non-POD type is undefined. In existing C++ implementations,
2954 however, @samp{offsetof} typically gives meaningful results even when
2955 applied to certain kinds of non-POD types. (Such as a simple
2956 @samp{struct} that fails to be a POD type only by virtue of having a
2957 constructor.) This flag is for users who are aware that they are
2958 writing nonportable code and who have deliberately chosen to ignore the
2961 The restrictions on @samp{offsetof} may be relaxed in a future version
2962 of the C++ standard.
2965 @opindex Winvalid-pch
2966 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
2967 the search path but can't be used.
2971 @opindex Wno-long-long
2972 Warn if @samp{long long} type is used. This is default. To inhibit
2973 the warning messages, use @option{-Wno-long-long}. Flags
2974 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2975 only when @option{-pedantic} flag is used.
2977 @item -Wvariadic-macros
2978 @opindex Wvariadic-macros
2979 @opindex Wno-variadic-macros
2980 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
2981 alternate syntax when in pedantic ISO C99 mode. This is default.
2982 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
2984 @item -Wdisabled-optimization
2985 @opindex Wdisabled-optimization
2986 Warn if a requested optimization pass is disabled. This warning does
2987 not generally indicate that there is anything wrong with your code; it
2988 merely indicates that GCC's optimizers were unable to handle the code
2989 effectively. Often, the problem is that your code is too big or too
2990 complex; GCC will refuse to optimize programs when the optimization
2991 itself is likely to take inordinate amounts of time.
2995 Make all warnings into errors.
2998 @node Debugging Options
2999 @section Options for Debugging Your Program or GCC
3000 @cindex options, debugging
3001 @cindex debugging information options
3003 GCC has various special options that are used for debugging
3004 either your program or GCC:
3009 Produce debugging information in the operating system's native format
3010 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
3013 On most systems that use stabs format, @option{-g} enables use of extra
3014 debugging information that only GDB can use; this extra information
3015 makes debugging work better in GDB but will probably make other debuggers
3017 refuse to read the program. If you want to control for certain whether
3018 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3019 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3021 Unlike most other C compilers, GCC allows you to use @option{-g} with
3022 @option{-O}. The shortcuts taken by optimized code may occasionally
3023 produce surprising results: some variables you declared may not exist
3024 at all; flow of control may briefly move where you did not expect it;
3025 some statements may not be executed because they compute constant
3026 results or their values were already at hand; some statements may
3027 execute in different places because they were moved out of loops.
3029 Nevertheless it proves possible to debug optimized output. This makes
3030 it reasonable to use the optimizer for programs that might have bugs.
3032 The following options are useful when GCC is generated with the
3033 capability for more than one debugging format.
3037 Produce debugging information for use by GDB@. This means to use the
3038 most expressive format available (DWARF 2, stabs, or the native format
3039 if neither of those are supported), including GDB extensions if at all
3044 Produce debugging information in stabs format (if that is supported),
3045 without GDB extensions. This is the format used by DBX on most BSD
3046 systems. On MIPS, Alpha and System V Release 4 systems this option
3047 produces stabs debugging output which is not understood by DBX or SDB@.
3048 On System V Release 4 systems this option requires the GNU assembler.
3050 @item -feliminate-unused-debug-symbols
3051 @opindex feliminate-unused-debug-symbols
3052 Produce debugging information in stabs format (if that is supported),
3053 for only symbols that are actually used.
3057 Produce debugging information in stabs 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.
3064 Produce debugging information in COFF format (if that is supported).
3065 This is the format used by SDB on most System V systems prior to
3070 Produce debugging information in XCOFF format (if that is supported).
3071 This is the format used by the DBX debugger on IBM RS/6000 systems.
3075 Produce debugging information in XCOFF format (if that is supported),
3076 using GNU extensions understood only by the GNU debugger (GDB)@. The
3077 use of these extensions is likely to make other debuggers crash or
3078 refuse to read the program, and may cause assemblers other than the GNU
3079 assembler (GAS) to fail with an error.
3083 Produce debugging information in DWARF version 2 format (if that is
3084 supported). This is the format used by DBX on IRIX 6.
3088 Produce debugging information in VMS debug format (if that is
3089 supported). This is the format used by DEBUG on VMS systems.
3092 @itemx -ggdb@var{level}
3093 @itemx -gstabs@var{level}
3094 @itemx -gcoff@var{level}
3095 @itemx -gxcoff@var{level}
3096 @itemx -gvms@var{level}
3097 Request debugging information and also use @var{level} to specify how
3098 much information. The default level is 2.
3100 Level 1 produces minimal information, enough for making backtraces in
3101 parts of the program that you don't plan to debug. This includes
3102 descriptions of functions and external variables, but no information
3103 about local variables and no line numbers.
3105 Level 3 includes extra information, such as all the macro definitions
3106 present in the program. Some debuggers support macro expansion when
3107 you use @option{-g3}.
3109 Note that in order to avoid confusion between DWARF1 debug level 2,
3110 and DWARF2 @option{-gdwarf-2} does not accept a concatenated debug
3111 level. Instead use an additional @option{-g@var{level}} option to
3112 change the debug level for DWARF2.
3114 @item -feliminate-dwarf2-dups
3115 @opindex feliminate-dwarf2-dups
3116 Compress DWARF2 debugging information by eliminating duplicated
3117 information about each symbol. This option only makes sense when
3118 generating DWARF2 debugging information with @option{-gdwarf-2}.
3120 @cindex @command{prof}
3123 Generate extra code to write profile information suitable for the
3124 analysis program @command{prof}. You must use this option when compiling
3125 the source files you want data about, and you must also use it when
3128 @cindex @command{gprof}
3131 Generate extra code to write profile information suitable for the
3132 analysis program @command{gprof}. You must use this option when compiling
3133 the source files you want data about, and you must also use it when
3138 Makes the compiler print out each function name as it is compiled, and
3139 print some statistics about each pass when it finishes.
3142 @opindex ftime-report
3143 Makes the compiler print some statistics about the time consumed by each
3144 pass when it finishes.
3147 @opindex fmem-report
3148 Makes the compiler print some statistics about permanent memory
3149 allocation when it finishes.
3151 @item -fprofile-arcs
3152 @opindex fprofile-arcs
3153 Add code so that program flow @dfn{arcs} are instrumented. During
3154 execution the program records how many times each branch and call is
3155 executed and how many times it is taken or returns. When the compiled
3156 program exits it saves this data to a file called
3157 @file{@var{auxname}.gcda} for each source file. The data may be used for
3158 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3159 test coverage analysis (@option{-ftest-coverage}). Each object file's
3160 @var{auxname} is generated from the name of the output file, if
3161 explicitly specified and it is not the final executable, otherwise it is
3162 the basename of the source file. In both cases any suffix is removed
3163 (e.g. @file{foo.gcda} for input file @file{dir/foo.c}, or
3164 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3169 Compile the source files with @option{-fprofile-arcs} plus optimization
3170 and code generation options. For test coverage analysis, use the
3171 additional @option{-ftest-coverage} option. You do not need to profile
3172 every source file in a program.
3175 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3176 (the latter implies the former).
3179 Run the program on a representative workload to generate the arc profile
3180 information. This may be repeated any number of times. You can run
3181 concurrent instances of your program, and provided that the file system
3182 supports locking, the data files will be correctly updated. Also
3183 @code{fork} calls are detected and correctly handled (double counting
3187 For profile-directed optimizations, compile the source files again with
3188 the same optimization and code generation options plus
3189 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3190 Control Optimization}).
3193 For test coverage analysis, use @command{gcov} to produce human readable
3194 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3195 @command{gcov} documentation for further information.
3199 With @option{-fprofile-arcs}, for each function of your program GCC
3200 creates a program flow graph, then finds a spanning tree for the graph.
3201 Only arcs that are not on the spanning tree have to be instrumented: the
3202 compiler adds code to count the number of times that these arcs are
3203 executed. When an arc is the only exit or only entrance to a block, the
3204 instrumentation code can be added to the block; otherwise, a new basic
3205 block must be created to hold the instrumentation code.
3208 @item -ftest-coverage
3209 @opindex ftest-coverage
3210 Produce a notes file that the @command{gcov} code-coverage utility
3211 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3212 show program coverage. Each source file's note file is called
3213 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3214 above for a description of @var{auxname} and instructions on how to
3215 generate test coverage data. Coverage data will match the source files
3216 more closely, if you do not optimize.
3218 @item -d@var{letters}
3220 Says to make debugging dumps during compilation at times specified by
3221 @var{letters}. This is used for debugging the compiler. The file names
3222 for most of the dumps are made by appending a pass number and a word to
3223 the @var{dumpname}. @var{dumpname} is generated from the name of the
3224 output file, if explicitly specified and it is not an executable,
3225 otherwise it is the basename of the source file. In both cases any
3226 suffix is removed (e.g. @file{foo.01.rtl} or @file{foo.02.sibling}).
3227 Here are the possible letters for use in @var{letters}, and their
3233 Annotate the assembler output with miscellaneous debugging information.
3236 Dump after computing branch probabilities, to @file{@var{file}.12.bp}.
3239 Dump after block reordering, to @file{@var{file}.31.bbro}.
3242 Dump after instruction combination, to the file @file{@var{file}.20.combine}.
3245 Dump after the first if conversion, to the file @file{@var{file}.14.ce1}.
3246 Also dump after the second if conversion, to the file @file{@var{file}.21.ce2}.
3249 Dump after branch target load optimization, to to @file{@var{file}.32.btl}.
3250 Also dump after delayed branch scheduling, to @file{@var{file}.36.dbr}.
3253 Dump all macro definitions, at the end of preprocessing, in addition to
3257 Dump after the third if conversion, to @file{@var{file}.30.ce3}.
3260 Dump after control and data flow analysis, to @file{@var{file}.11.cfg}.
3261 Also dump after life analysis, to @file{@var{file}.19.life}.
3264 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.07.addressof}.
3267 Dump after global register allocation, to @file{@var{file}.25.greg}.
3270 Dump after GCSE, to @file{@var{file}.08.gcse}.
3271 Also dump after jump bypassing and control flow optimizations, to
3272 @file{@var{file}.10.bypass}.
3275 Dump after finalization of EH handling code, to @file{@var{file}.03.eh}.
3278 Dump after sibling call optimizations, to @file{@var{file}.02.sibling}.
3281 Dump after the first jump optimization, to @file{@var{file}.04.jump}.
3284 Dump after conversion from registers to stack, to @file{@var{file}.34.stack}.
3287 Dump after local register allocation, to @file{@var{file}.24.lreg}.
3290 Dump after loop optimization passes, to @file{@var{file}.09.loop} and
3291 @file{@var{file}.16.loop2}.
3294 Dump after performing the machine dependent reorganization pass, to
3295 @file{@var{file}.35.mach}.
3298 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3301 Dump after the register move pass, to @file{@var{file}.22.regmove}.
3304 Dump after post-reload optimizations, to @file{@var{file}.26.postreload}.
3307 Dump after RTL generation, to @file{@var{file}.01.rtl}.
3310 Dump after the second scheduling pass, to @file{@var{file}.33.sched2}.
3313 Dump after CSE (including the jump optimization that sometimes follows
3314 CSE), to @file{@var{file}.06.cse}.
3317 Dump after the first scheduling pass, to @file{@var{file}.23.sched}.
3320 Dump after the second CSE pass (including the jump optimization that
3321 sometimes follows CSE), to @file{@var{file}.18.cse2}.
3324 Dump after running tracer, to @file{@var{file}.15.tracer}.
3327 Dump after null pointer elimination pass to @file{@var{file}.05.null}.
3330 Dump callgraph and unit-at-a-time optimization @file{@var{file}.00.unit}.
3333 Dump after the value profile transformations, to @file{@var{file}.13.vpt}.
3334 Also dump after variable tracking, to @file{@var{file}.35.vartrack}.
3337 Dump after the second flow pass, to @file{@var{file}.27.flow2}.
3340 Dump after the peephole pass, to @file{@var{file}.28.peephole2}.
3343 Dump after constructing the web, to @file{@var{file}.17.web}.
3346 Produce all the dumps listed above.
3349 Produce a core dump whenever an error occurs.
3352 Print statistics on memory usage, at the end of the run, to
3356 Annotate the assembler output with a comment indicating which
3357 pattern and alternative was used. The length of each instruction is
3361 Dump the RTL in the assembler output as a comment before each instruction.
3362 Also turns on @option{-dp} annotation.
3365 For each of the other indicated dump files (except for
3366 @file{@var{file}.01.rtl}), dump a representation of the control flow graph
3367 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3370 Just generate RTL for a function instead of compiling it. Usually used
3374 Dump debugging information during parsing, to standard error.
3377 @item -fdump-unnumbered
3378 @opindex fdump-unnumbered
3379 When doing debugging dumps (see @option{-d} option above), suppress instruction
3380 numbers and line number note output. This makes it more feasible to
3381 use diff on debugging dumps for compiler invocations with different
3382 options, in particular with and without @option{-g}.
3384 @item -fdump-translation-unit @r{(C and C++ only)}
3385 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3386 @opindex fdump-translation-unit
3387 Dump a representation of the tree structure for the entire translation
3388 unit to a file. The file name is made by appending @file{.tu} to the
3389 source file name. If the @samp{-@var{options}} form is used, @var{options}
3390 controls the details of the dump as described for the
3391 @option{-fdump-tree} options.
3393 @item -fdump-class-hierarchy @r{(C++ only)}
3394 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3395 @opindex fdump-class-hierarchy
3396 Dump a representation of each class's hierarchy and virtual function
3397 table layout to a file. The file name is made by appending @file{.class}
3398 to the source file name. If the @samp{-@var{options}} form is used,
3399 @var{options} controls the details of the dump as described for the
3400 @option{-fdump-tree} options.
3402 @item -fdump-tree-@var{switch} @r{(C++ only)}
3403 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3405 Control the dumping at various stages of processing the intermediate
3406 language tree to a file. The file name is generated by appending a switch
3407 specific suffix to the source file name. If the @samp{-@var{options}}
3408 form is used, @var{options} is a list of @samp{-} separated options that
3409 control the details of the dump. Not all options are applicable to all
3410 dumps, those which are not meaningful will be ignored. The following
3411 options are available
3415 Print the address of each node. Usually this is not meaningful as it
3416 changes according to the environment and source file. Its primary use
3417 is for tying up a dump file with a debug environment.
3419 Inhibit dumping of members of a scope or body of a function merely
3420 because that scope has been reached. Only dump such items when they
3421 are directly reachable by some other path.
3423 Turn on all options.
3426 The following tree dumps are possible:
3429 Dump before any tree based optimization, to @file{@var{file}.original}.
3431 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3433 Dump after function inlining, to @file{@var{file}.inlined}.
3436 @item -frandom-seed=@var{string}
3437 @opindex frandom-string
3438 This option provides a seed that GCC uses when it would otherwise use
3439 random numbers. It is used to generate certain symbol names
3440 that have to be different in every compiled file. It is also used to
3441 place unique stamps in coverage data files and the object files that
3442 produce them. You can use the @option{-frandom-seed} option to produce
3443 reproducibly identical object files.
3445 The @var{string} should be different for every file you compile.
3447 @item -fsched-verbose=@var{n}
3448 @opindex fsched-verbose
3449 On targets that use instruction scheduling, this option controls the
3450 amount of debugging output the scheduler prints. This information is
3451 written to standard error, unless @option{-dS} or @option{-dR} is
3452 specified, in which case it is output to the usual dump
3453 listing file, @file{.sched} or @file{.sched2} respectively. However
3454 for @var{n} greater than nine, the output is always printed to standard
3457 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3458 same information as @option{-dRS}. For @var{n} greater than one, it
3459 also output basic block probabilities, detailed ready list information
3460 and unit/insn info. For @var{n} greater than two, it includes RTL
3461 at abort point, control-flow and regions info. And for @var{n} over
3462 four, @option{-fsched-verbose} also includes dependence info.
3466 Store the usual ``temporary'' intermediate files permanently; place them
3467 in the current directory and name them based on the source file. Thus,
3468 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3469 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3470 preprocessed @file{foo.i} output file even though the compiler now
3471 normally uses an integrated preprocessor.
3475 Report the CPU time taken by each subprocess in the compilation
3476 sequence. For C source files, this is the compiler proper and assembler
3477 (plus the linker if linking is done). The output looks like this:
3484 The first number on each line is the ``user time,'' that is time spent
3485 executing the program itself. The second number is ``system time,''
3486 time spent executing operating system routines on behalf of the program.
3487 Both numbers are in seconds.
3489 @item -fvar-tracking
3490 @opindex fvar-tracking
3491 Run variable tracking pass. It computes where variables are stored at each
3492 position in code. Better debugging information is then generated
3493 (if the debugging information format supports this information).
3495 It is enabled by default when compiling with optimization (@option{-Os},
3496 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
3497 the debug info format supports it.
3499 @item -print-file-name=@var{library}
3500 @opindex print-file-name
3501 Print the full absolute name of the library file @var{library} that
3502 would be used when linking---and don't do anything else. With this
3503 option, GCC does not compile or link anything; it just prints the
3506 @item -print-multi-directory
3507 @opindex print-multi-directory
3508 Print the directory name corresponding to the multilib selected by any
3509 other switches present in the command line. This directory is supposed
3510 to exist in @env{GCC_EXEC_PREFIX}.
3512 @item -print-multi-lib
3513 @opindex print-multi-lib
3514 Print the mapping from multilib directory names to compiler switches
3515 that enable them. The directory name is separated from the switches by
3516 @samp{;}, and each switch starts with an @samp{@@} instead of the
3517 @samp{-}, without spaces between multiple switches. This is supposed to
3518 ease shell-processing.
3520 @item -print-prog-name=@var{program}
3521 @opindex print-prog-name
3522 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3524 @item -print-libgcc-file-name
3525 @opindex print-libgcc-file-name
3526 Same as @option{-print-file-name=libgcc.a}.
3528 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3529 but you do want to link with @file{libgcc.a}. You can do
3532 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3535 @item -print-search-dirs
3536 @opindex print-search-dirs
3537 Print the name of the configured installation directory and a list of
3538 program and library directories @command{gcc} will search---and don't do anything else.
3540 This is useful when @command{gcc} prints the error message
3541 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3542 To resolve this you either need to put @file{cpp0} and the other compiler
3543 components where @command{gcc} expects to find them, or you can set the environment
3544 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3545 Don't forget the trailing '/'.
3546 @xref{Environment Variables}.
3549 @opindex dumpmachine
3550 Print the compiler's target machine (for example,
3551 @samp{i686-pc-linux-gnu})---and don't do anything else.
3554 @opindex dumpversion
3555 Print the compiler version (for example, @samp{3.0})---and don't do
3560 Print the compiler's built-in specs---and don't do anything else. (This
3561 is used when GCC itself is being built.) @xref{Spec Files}.
3563 @item -feliminate-unused-debug-types
3564 @opindex feliminate-unused-debug-types
3565 Normally, when producing DWARF2 output, GCC will emit debugging
3566 information for all types declared in a compilation
3567 unit, regardless of whether or not they are actually used
3568 in that compilation unit. Sometimes this is useful, such as
3569 if, in the debugger, you want to cast a value to a type that is
3570 not actually used in your program (but is declared). More often,
3571 however, this results in a significant amount of wasted space.
3572 With this option, GCC will avoid producing debug symbol output
3573 for types that are nowhere used in the source file being compiled.
3576 @node Optimize Options
3577 @section Options That Control Optimization
3578 @cindex optimize options
3579 @cindex options, optimization
3581 These options control various sorts of optimizations.
3583 Without any optimization option, the compiler's goal is to reduce the
3584 cost of compilation and to make debugging produce the expected
3585 results. Statements are independent: if you stop the program with a
3586 breakpoint between statements, you can then assign a new value to any
3587 variable or change the program counter to any other statement in the
3588 function and get exactly the results you would expect from the source
3591 Turning on optimization flags makes the compiler attempt to improve
3592 the performance and/or code size at the expense of compilation time
3593 and possibly the ability to debug the program.
3595 The compiler performs optimization based on the knowledge it has of
3596 the program. Using the @option{-funit-at-a-time} flag will allow the
3597 compiler to consider information gained from later functions in the
3598 file when compiling a function. Compiling multiple files at once to a
3599 single output file (and using @option{-funit-at-a-time}) will allow
3600 the compiler to use information gained from all of the files when
3601 compiling each of them.
3603 Not all optimizations are controlled directly by a flag. Only
3604 optimizations that have a flag are listed.
3611 Optimize. Optimizing compilation takes somewhat more time, and a lot
3612 more memory for a large function.
3614 With @option{-O}, the compiler tries to reduce code size and execution
3615 time, without performing any optimizations that take a great deal of
3618 @option{-O} turns on the following optimization flags:
3619 @gccoptlist{-fdefer-pop @gol
3620 -fmerge-constants @gol
3622 -floop-optimize @gol
3623 -fif-conversion @gol
3624 -fif-conversion2 @gol
3625 -fdelayed-branch @gol
3626 -fguess-branch-probability @gol
3629 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3630 where doing so does not interfere with debugging.
3634 Optimize even more. GCC performs nearly all supported optimizations
3635 that do not involve a space-speed tradeoff. The compiler does not
3636 perform loop unrolling or function inlining when you specify @option{-O2}.
3637 As compared to @option{-O}, this option increases both compilation time
3638 and the performance of the generated code.
3640 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3641 also turns on the following optimization flags:
3642 @gccoptlist{-fforce-mem @gol
3643 -foptimize-sibling-calls @gol
3644 -fstrength-reduce @gol
3645 -fcse-follow-jumps -fcse-skip-blocks @gol
3646 -frerun-cse-after-loop -frerun-loop-opt @gol
3647 -fgcse -fgcse-lm -fgcse-sm -fgcse-las @gol
3648 -fdelete-null-pointer-checks @gol
3649 -fexpensive-optimizations @gol
3651 -fschedule-insns -fschedule-insns2 @gol
3652 -fsched-interblock -fsched-spec @gol
3655 -freorder-blocks -freorder-functions @gol
3656 -fstrict-aliasing @gol
3657 -funit-at-a-time @gol
3658 -falign-functions -falign-jumps @gol
3659 -falign-loops -falign-labels @gol
3662 Please note the warning under @option{-fgcse} about
3663 invoking @option{-O2} on programs that use computed gotos.
3667 Optimize yet more. @option{-O3} turns on all optimizations specified by
3668 @option{-O2} and also turns on the @option{-finline-functions},
3669 @option{-fweb} and @option{-frename-registers} options.
3673 Do not optimize. This is the default.
3677 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3678 do not typically increase code size. It also performs further
3679 optimizations designed to reduce code size.
3681 @option{-Os} disables the following optimization flags:
3682 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3683 -falign-labels -freorder-blocks -freorder-blocks-and-partition -fprefetch-loop-arrays}
3685 If you use multiple @option{-O} options, with or without level numbers,
3686 the last such option is the one that is effective.
3689 Options of the form @option{-f@var{flag}} specify machine-independent
3690 flags. Most flags have both positive and negative forms; the negative
3691 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3692 below, only one of the forms is listed---the one you typically will
3693 use. You can figure out the other form by either removing @samp{no-}
3696 The following options control specific optimizations. They are either
3697 activated by @option{-O} options or are related to ones that are. You
3698 can use the following flags in the rare cases when ``fine-tuning'' of
3699 optimizations to be performed is desired.
3702 @item -fno-default-inline
3703 @opindex fno-default-inline
3704 Do not make member functions inline by default merely because they are
3705 defined inside the class scope (C++ only). Otherwise, when you specify
3706 @w{@option{-O}}, member functions defined inside class scope are compiled
3707 inline by default; i.e., you don't need to add @samp{inline} in front of
3708 the member function name.
3710 @item -fno-defer-pop
3711 @opindex fno-defer-pop
3712 Always pop the arguments to each function call as soon as that function
3713 returns. For machines which must pop arguments after a function call,
3714 the compiler normally lets arguments accumulate on the stack for several
3715 function calls and pops them all at once.
3717 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3721 Force memory operands to be copied into registers before doing
3722 arithmetic on them. This produces better code by making all memory
3723 references potential common subexpressions. When they are not common
3724 subexpressions, instruction combination should eliminate the separate
3727 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3730 @opindex fforce-addr
3731 Force memory address constants to be copied into registers before
3732 doing arithmetic on them. This may produce better code just as
3733 @option{-fforce-mem} may.
3735 @item -fomit-frame-pointer
3736 @opindex fomit-frame-pointer
3737 Don't keep the frame pointer in a register for functions that
3738 don't need one. This avoids the instructions to save, set up and
3739 restore frame pointers; it also makes an extra register available
3740 in many functions. @strong{It also makes debugging impossible on
3743 On some machines, such as the VAX, this flag has no effect, because
3744 the standard calling sequence automatically handles the frame pointer
3745 and nothing is saved by pretending it doesn't exist. The
3746 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3747 whether a target machine supports this flag. @xref{Registers,,Register
3748 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3750 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3752 @item -foptimize-sibling-calls
3753 @opindex foptimize-sibling-calls
3754 Optimize sibling and tail recursive calls.
3756 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3760 Don't pay attention to the @code{inline} keyword. Normally this option
3761 is used to keep the compiler from expanding any functions inline.
3762 Note that if you are not optimizing, no functions can be expanded inline.
3764 @item -finline-functions
3765 @opindex finline-functions
3766 Integrate all simple functions into their callers. The compiler
3767 heuristically decides which functions are simple enough to be worth
3768 integrating in this way.
3770 If all calls to a given function are integrated, and the function is
3771 declared @code{static}, then the function is normally not output as
3772 assembler code in its own right.
3774 Enabled at level @option{-O3}.
3776 @item -finline-limit=@var{n}
3777 @opindex finline-limit
3778 By default, GCC limits the size of functions that can be inlined. This flag
3779 allows the control of this limit for functions that are explicitly marked as
3780 inline (i.e., marked with the inline keyword or defined within the class
3781 definition in c++). @var{n} is the size of functions that can be inlined in
3782 number of pseudo instructions (not counting parameter handling). The default
3783 value of @var{n} is 600.
3784 Increasing this value can result in more inlined code at
3785 the cost of compilation time and memory consumption. Decreasing usually makes
3786 the compilation faster and less code will be inlined (which presumably
3787 means slower programs). This option is particularly useful for programs that
3788 use inlining heavily such as those based on recursive templates with C++.
3790 Inlining is actually controlled by a number of parameters, which may be
3791 specified individually by using @option{--param @var{name}=@var{value}}.
3792 The @option{-finline-limit=@var{n}} option sets some of these parameters
3796 @item max-inline-insns-single
3797 is set to @var{n}/2.
3798 @item max-inline-insns-auto
3799 is set to @var{n}/2.
3800 @item min-inline-insns
3801 is set to 130 or @var{n}/4, whichever is smaller.
3802 @item max-inline-insns-rtl
3806 See below for a documentation of the individual
3807 parameters controlling inlining.
3809 @emph{Note:} pseudo instruction represents, in this particular context, an
3810 abstract measurement of function's size. In no way, it represents a count
3811 of assembly instructions and as such its exact meaning might change from one
3812 release to an another.
3814 @item -fkeep-inline-functions
3815 @opindex fkeep-inline-functions
3816 Even if all calls to a given function are integrated, and the function
3817 is declared @code{static}, nevertheless output a separate run-time
3818 callable version of the function. This switch does not affect
3819 @code{extern inline} functions.
3821 @item -fkeep-static-consts
3822 @opindex fkeep-static-consts
3823 Emit variables declared @code{static const} when optimization isn't turned
3824 on, even if the variables aren't referenced.
3826 GCC enables this option by default. If you want to force the compiler to
3827 check if the variable was referenced, regardless of whether or not
3828 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3830 @item -fmerge-constants
3831 Attempt to merge identical constants (string constants and floating point
3832 constants) across compilation units.
3834 This option is the default for optimized compilation if the assembler and
3835 linker support it. Use @option{-fno-merge-constants} to inhibit this
3838 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3840 @item -fmerge-all-constants
3841 Attempt to merge identical constants and identical variables.
3843 This option implies @option{-fmerge-constants}. In addition to
3844 @option{-fmerge-constants} this considers e.g. even constant initialized
3845 arrays or initialized constant variables with integral or floating point
3846 types. Languages like C or C++ require each non-automatic variable to
3847 have distinct location, so using this option will result in non-conforming
3852 Use a graph coloring register allocator. Currently this option is meant
3853 only for testing. Users should not specify this option, since it is not
3854 yet ready for production use.
3856 @item -fno-branch-count-reg
3857 @opindex fno-branch-count-reg
3858 Do not use ``decrement and branch'' instructions on a count register,
3859 but instead generate a sequence of instructions that decrement a
3860 register, compare it against zero, then branch based upon the result.
3861 This option is only meaningful on architectures that support such
3862 instructions, which include x86, PowerPC, IA-64 and S/390.
3864 The default is @option{-fbranch-count-reg}, enabled when
3865 @option{-fstrength-reduce} is enabled.
3867 @item -fno-function-cse
3868 @opindex fno-function-cse
3869 Do not put function addresses in registers; make each instruction that
3870 calls a constant function contain the function's address explicitly.
3872 This option results in less efficient code, but some strange hacks
3873 that alter the assembler output may be confused by the optimizations
3874 performed when this option is not used.
3876 The default is @option{-ffunction-cse}
3878 @item -fno-zero-initialized-in-bss
3879 @opindex fno-zero-initialized-in-bss
3880 If the target supports a BSS section, GCC by default puts variables that
3881 are initialized to zero into BSS@. This can save space in the resulting
3884 This option turns off this behavior because some programs explicitly
3885 rely on variables going to the data section. E.g., so that the
3886 resulting executable can find the beginning of that section and/or make
3887 assumptions based on that.
3889 The default is @option{-fzero-initialized-in-bss}.
3891 @item -fstrength-reduce
3892 @opindex fstrength-reduce
3893 Perform the optimizations of loop strength reduction and
3894 elimination of iteration variables.
3896 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3898 @item -fthread-jumps
3899 @opindex fthread-jumps
3900 Perform optimizations where we check to see if a jump branches to a
3901 location where another comparison subsumed by the first is found. If
3902 so, the first branch is redirected to either the destination of the
3903 second branch or a point immediately following it, depending on whether
3904 the condition is known to be true or false.
3906 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3908 @item -fcse-follow-jumps
3909 @opindex fcse-follow-jumps
3910 In common subexpression elimination, scan through jump instructions
3911 when the target of the jump is not reached by any other path. For
3912 example, when CSE encounters an @code{if} statement with an
3913 @code{else} clause, CSE will follow the jump when the condition
3916 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3918 @item -fcse-skip-blocks
3919 @opindex fcse-skip-blocks
3920 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3921 follow jumps which conditionally skip over blocks. When CSE
3922 encounters a simple @code{if} statement with no else clause,
3923 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3924 body of the @code{if}.
3926 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3928 @item -frerun-cse-after-loop
3929 @opindex frerun-cse-after-loop
3930 Re-run common subexpression elimination after loop optimizations has been
3933 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3935 @item -frerun-loop-opt
3936 @opindex frerun-loop-opt
3937 Run the loop optimizer twice.
3939 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3943 Perform a global common subexpression elimination pass.
3944 This pass also performs global constant and copy propagation.
3946 @emph{Note:} When compiling a program using computed gotos, a GCC
3947 extension, you may get better runtime performance if you disable
3948 the global common subexpression elimination pass by adding
3949 @option{-fno-gcse} to the command line.
3951 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3955 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3956 attempt to move loads which are only killed by stores into themselves. This
3957 allows a loop containing a load/store sequence to be changed to a load outside
3958 the loop, and a copy/store within the loop.
3960 Enabled by default when gcse is enabled.
3964 When @option{-fgcse-sm} is enabled, a store motion pass is run after
3965 global common subexpression elimination. This pass will attempt to move
3966 stores out of loops. When used in conjunction with @option{-fgcse-lm},
3967 loops containing a load/store sequence can be changed to a load before
3968 the loop and a store after the loop.
3970 Enabled by default when gcse is enabled.
3974 When @option{-fgcse-las} is enabled, the global common subexpression
3975 elimination pass eliminates redundant loads that come after stores to the
3976 same memory location (both partial and full redundancies).
3978 Enabled by default when gcse is enabled.
3980 @item -floop-optimize
3981 @opindex floop-optimize
3982 Perform loop optimizations: move constant expressions out of loops, simplify
3983 exit test conditions and optionally do strength-reduction and loop unrolling as
3986 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3988 @item -fcrossjumping
3989 @opindex crossjumping
3990 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
3991 resulting code may or may not perform better than without cross-jumping.
3993 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3995 @item -fif-conversion
3996 @opindex if-conversion
3997 Attempt to transform conditional jumps into branch-less equivalents. This
3998 include use of conditional moves, min, max, set flags and abs instructions, and
3999 some tricks doable by standard arithmetics. The use of conditional execution
4000 on chips where it is available is controlled by @code{if-conversion2}.
4002 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4004 @item -fif-conversion2
4005 @opindex if-conversion2
4006 Use conditional execution (where available) to transform conditional jumps into
4007 branch-less equivalents.
4009 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4011 @item -fdelete-null-pointer-checks
4012 @opindex fdelete-null-pointer-checks
4013 Use global dataflow analysis to identify and eliminate useless checks
4014 for null pointers. The compiler assumes that dereferencing a null
4015 pointer would have halted the program. If a pointer is checked after
4016 it has already been dereferenced, it cannot be null.
4018 In some environments, this assumption is not true, and programs can
4019 safely dereference null pointers. Use
4020 @option{-fno-delete-null-pointer-checks} to disable this optimization
4021 for programs which depend on that behavior.
4023 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4025 @item -fexpensive-optimizations
4026 @opindex fexpensive-optimizations
4027 Perform a number of minor optimizations that are relatively expensive.
4029 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4031 @item -foptimize-register-move
4033 @opindex foptimize-register-move
4035 Attempt to reassign register numbers in move instructions and as
4036 operands of other simple instructions in order to maximize the amount of
4037 register tying. This is especially helpful on machines with two-operand
4040 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4043 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4045 @item -fdelayed-branch
4046 @opindex fdelayed-branch
4047 If supported for the target machine, attempt to reorder instructions
4048 to exploit instruction slots available after delayed branch
4051 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4053 @item -fschedule-insns
4054 @opindex fschedule-insns
4055 If supported for the target machine, attempt to reorder instructions to
4056 eliminate execution stalls due to required data being unavailable. This
4057 helps machines that have slow floating point or memory load instructions
4058 by allowing other instructions to be issued until the result of the load
4059 or floating point instruction is required.
4061 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4063 @item -fschedule-insns2
4064 @opindex fschedule-insns2
4065 Similar to @option{-fschedule-insns}, but requests an additional pass of
4066 instruction scheduling after register allocation has been done. This is
4067 especially useful on machines with a relatively small number of
4068 registers and where memory load instructions take more than one cycle.
4070 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4072 @item -fno-sched-interblock
4073 @opindex fno-sched-interblock
4074 Don't schedule instructions across basic blocks. This is normally
4075 enabled by default when scheduling before register allocation, i.e.@:
4076 with @option{-fschedule-insns} or at @option{-O2} or higher.
4078 @item -fno-sched-spec
4079 @opindex fno-sched-spec
4080 Don't allow speculative motion of non-load instructions. This is normally
4081 enabled by default when scheduling before register allocation, i.e.@:
4082 with @option{-fschedule-insns} or at @option{-O2} or higher.
4084 @item -fsched-spec-load
4085 @opindex fsched-spec-load
4086 Allow speculative motion of some load instructions. This only makes
4087 sense when scheduling before register allocation, i.e.@: with
4088 @option{-fschedule-insns} or at @option{-O2} or higher.
4090 @item -fsched-spec-load-dangerous
4091 @opindex fsched-spec-load-dangerous
4092 Allow speculative motion of more load instructions. This only makes
4093 sense when scheduling before register allocation, i.e.@: with
4094 @option{-fschedule-insns} or at @option{-O2} or higher.
4096 @item -fsched-stalled-insns=@var{n}
4097 @opindex fsched-stalled-insns
4098 Define how many insns (if any) can be moved prematurely from the queue
4099 of stalled insns into the ready list, during the second scheduling pass.
4101 @item -fsched-stalled-insns-dep=@var{n}
4102 @opindex fsched-stalled-insns-dep
4103 Define how many insn groups (cycles) will be examined for a dependency
4104 on a stalled insn that is candidate for premature removal from the queue
4105 of stalled insns. Has an effect only during the second scheduling pass,
4106 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4108 @item -fsched2-use-superblocks
4109 @opindex fsched2-use-superblocks
4110 When scheduling after register allocation, do use superblock scheduling
4111 algorithm. Superblock scheduling allows motion across basic block boundaries
4112 resulting on faster schedules. This option is experimental, as not all machine
4113 descriptions used by GCC model the CPU closely enough to avoid unreliable
4114 results from the algorithm.
4116 This only makes sense when scheduling after register allocation, i.e.@: with
4117 @option{-fschedule-insns2} or at @option{-O2} or higher.
4119 @item -fsched2-use-traces
4120 @opindex fsched2-use-traces
4121 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4122 allocation and additionally perform code duplication in order to increase the
4123 size of superblocks using tracer pass. See @option{-ftracer} for details on
4126 This mode should produce faster but significantly longer programs. Also
4127 without @code{-fbranch-probabilities} the traces constructed may not match the
4128 reality and hurt the performance. This only makes
4129 sense when scheduling after register allocation, i.e.@: with
4130 @option{-fschedule-insns2} or at @option{-O2} or higher.
4132 @item -fcaller-saves
4133 @opindex fcaller-saves
4134 Enable values to be allocated in registers that will be clobbered by
4135 function calls, by emitting extra instructions to save and restore the
4136 registers around such calls. Such allocation is done only when it
4137 seems to result in better code than would otherwise be produced.
4139 This option is always enabled by default on certain machines, usually
4140 those which have no call-preserved registers to use instead.
4142 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4144 @item -fmove-all-movables
4145 @opindex fmove-all-movables
4146 Forces all invariant computations in loops to be moved
4149 @item -freduce-all-givs
4150 @opindex freduce-all-givs
4151 Forces all general-induction variables in loops to be
4154 @emph{Note:} When compiling programs written in Fortran,
4155 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4156 by default when you use the optimizer.
4158 These options may generate better or worse code; results are highly
4159 dependent on the structure of loops within the source code.
4161 These two options are intended to be removed someday, once
4162 they have helped determine the efficacy of various
4163 approaches to improving loop optimizations.
4165 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
4166 know how use of these options affects
4167 the performance of your production code.
4168 We're very interested in code that runs @emph{slower}
4169 when these options are @emph{enabled}.
4172 @itemx -fno-peephole2
4173 @opindex fno-peephole
4174 @opindex fno-peephole2
4175 Disable any machine-specific peephole optimizations. The difference
4176 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4177 are implemented in the compiler; some targets use one, some use the
4178 other, a few use both.
4180 @option{-fpeephole} is enabled by default.
4181 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4183 @item -fno-guess-branch-probability
4184 @opindex fno-guess-branch-probability
4185 Do not guess branch probabilities using a randomized model.
4187 Sometimes GCC will opt to use a randomized model to guess branch
4188 probabilities, when none are available from either profiling feedback
4189 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4190 different runs of the compiler on the same program may produce different
4193 In a hard real-time system, people don't want different runs of the
4194 compiler to produce code that has different behavior; minimizing
4195 non-determinism is of paramount import. This switch allows users to
4196 reduce non-determinism, possibly at the expense of inferior
4199 The default is @option{-fguess-branch-probability} at levels
4200 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4202 @item -freorder-blocks
4203 @opindex freorder-blocks
4204 Reorder basic blocks in the compiled function in order to reduce number of
4205 taken branches and improve code locality.
4207 Enabled at levels @option{-O2}, @option{-O3}.
4209 @item -freorder-blocks-and-partition
4210 @opindex freorder-blocks-and-partition
4211 In addition to reordering basic blocks in the compiled function, in order
4212 to reduce number of taken branches, partitions hot and cold basic blocks
4213 into separate sections of the assembly and .o files, to improve
4214 paging and cache locality performance.
4216 @item -freorder-functions
4217 @opindex freorder-functions
4218 Reorder basic blocks in the compiled function in order to reduce number of
4219 taken branches and improve code locality. This is implemented by using special
4220 subsections @code{.text.hot} for most frequently executed functions and
4221 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
4222 the linker so object file format must support named sections and linker must
4223 place them in a reasonable way.
4225 Also profile feedback must be available in to make this option effective. See
4226 @option{-fprofile-arcs} for details.
4228 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4230 @item -fstrict-aliasing
4231 @opindex fstrict-aliasing
4232 Allows the compiler to assume the strictest aliasing rules applicable to
4233 the language being compiled. For C (and C++), this activates
4234 optimizations based on the type of expressions. In particular, an
4235 object of one type is assumed never to reside at the same address as an
4236 object of a different type, unless the types are almost the same. For
4237 example, an @code{unsigned int} can alias an @code{int}, but not a
4238 @code{void*} or a @code{double}. A character type may alias any other
4241 Pay special attention to code like this:
4254 The practice of reading from a different union member than the one most
4255 recently written to (called ``type-punning'') is common. Even with
4256 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4257 is accessed through the union type. So, the code above will work as
4258 expected. However, this code might not:
4269 Every language that wishes to perform language-specific alias analysis
4270 should define a function that computes, given an @code{tree}
4271 node, an alias set for the node. Nodes in different alias sets are not
4272 allowed to alias. For an example, see the C front-end function
4273 @code{c_get_alias_set}.
4275 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4277 @item -falign-functions
4278 @itemx -falign-functions=@var{n}
4279 @opindex falign-functions
4280 Align the start of functions to the next power-of-two greater than
4281 @var{n}, skipping up to @var{n} bytes. For instance,
4282 @option{-falign-functions=32} aligns functions to the next 32-byte
4283 boundary, but @option{-falign-functions=24} would align to the next
4284 32-byte boundary only if this can be done by skipping 23 bytes or less.
4286 @option{-fno-align-functions} and @option{-falign-functions=1} are
4287 equivalent and mean that functions will not be aligned.
4289 Some assemblers only support this flag when @var{n} is a power of two;
4290 in that case, it is rounded up.
4292 If @var{n} is not specified or is zero, use a machine-dependent default.
4294 Enabled at levels @option{-O2}, @option{-O3}.
4296 @item -falign-labels
4297 @itemx -falign-labels=@var{n}
4298 @opindex falign-labels
4299 Align all branch targets to a power-of-two boundary, skipping up to
4300 @var{n} bytes like @option{-falign-functions}. This option can easily
4301 make code slower, because it must insert dummy operations for when the
4302 branch target is reached in the usual flow of the code.
4304 @option{-fno-align-labels} and @option{-falign-labels=1} are
4305 equivalent and mean that labels will not be aligned.
4307 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4308 are greater than this value, then their values are used instead.
4310 If @var{n} is not specified or is zero, use a machine-dependent default
4311 which is very likely to be @samp{1}, meaning no alignment.
4313 Enabled at levels @option{-O2}, @option{-O3}.
4316 @itemx -falign-loops=@var{n}
4317 @opindex falign-loops
4318 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4319 like @option{-falign-functions}. The hope is that the loop will be
4320 executed many times, which will make up for any execution of the dummy
4323 @option{-fno-align-loops} and @option{-falign-loops=1} are
4324 equivalent and mean that loops will not be aligned.
4326 If @var{n} is not specified or is zero, use a machine-dependent default.
4328 Enabled at levels @option{-O2}, @option{-O3}.
4331 @itemx -falign-jumps=@var{n}
4332 @opindex falign-jumps
4333 Align branch targets to a power-of-two boundary, for branch targets
4334 where the targets can only be reached by jumping, skipping up to @var{n}
4335 bytes like @option{-falign-functions}. In this case, no dummy operations
4338 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4339 equivalent and mean that loops will not be aligned.
4341 If @var{n} is not specified or is zero, use a machine-dependent default.
4343 Enabled at levels @option{-O2}, @option{-O3}.
4347 Constructs webs as commonly used for register allocation purposes and assign
4348 each web individual pseudo register. This allows our register allocation pass
4349 to operate on pseudos directly, but also strengthens several other optimization
4350 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
4351 however, make debugging impossible, since variables will no longer stay in a
4354 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os},
4355 on targets where the default format for debugging information supports
4358 @item -fno-cprop-registers
4359 @opindex fno-cprop-registers
4360 After register allocation and post-register allocation instruction splitting,
4361 we perform a copy-propagation pass to try to reduce scheduling dependencies
4362 and occasionally eliminate the copy.
4364 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4366 @item -fprofile-generate
4367 @opindex fprofile-generate
4369 Enable options usually used for instrumenting application to produce
4370 profile useful for later recompilation with profile feedback based
4371 optimization. You must use @code{-fprofile-generate} both when
4372 compiling and when linking your program.
4374 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
4377 @opindex fprofile-use
4378 Enable profile feedback directed optimizations, and optimizations
4379 generally profitable only with profile feedback available.
4381 The following options are enabled: @code{-fbranch-probabilities},
4382 @code{-fvpt}, @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}.
4386 The following options control compiler behavior regarding floating
4387 point arithmetic. These options trade off between speed and
4388 correctness. All must be specifically enabled.
4392 @opindex ffloat-store
4393 Do not store floating point variables in registers, and inhibit other
4394 options that might change whether a floating point value is taken from a
4397 @cindex floating point precision
4398 This option prevents undesirable excess precision on machines such as
4399 the 68000 where the floating registers (of the 68881) keep more
4400 precision than a @code{double} is supposed to have. Similarly for the
4401 x86 architecture. For most programs, the excess precision does only
4402 good, but a few programs rely on the precise definition of IEEE floating
4403 point. Use @option{-ffloat-store} for such programs, after modifying
4404 them to store all pertinent intermediate computations into variables.
4408 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4409 @option{-fno-trapping-math}, @option{-ffinite-math-only},
4410 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
4412 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4414 This option should never be turned on by any @option{-O} option since
4415 it can result in incorrect output for programs which depend on
4416 an exact implementation of IEEE or ISO rules/specifications for
4419 @item -fno-math-errno
4420 @opindex fno-math-errno
4421 Do not set ERRNO after calling math functions that are executed
4422 with a single instruction, e.g., sqrt. A program that relies on
4423 IEEE exceptions for math error handling may want to use this flag
4424 for speed while maintaining IEEE arithmetic compatibility.
4426 This option should never be turned on by any @option{-O} option since
4427 it can result in incorrect output for programs which depend on
4428 an exact implementation of IEEE or ISO rules/specifications for
4431 The default is @option{-fmath-errno}.
4433 @item -funsafe-math-optimizations
4434 @opindex funsafe-math-optimizations
4435 Allow optimizations for floating-point arithmetic that (a) assume
4436 that arguments and results are valid and (b) may violate IEEE or
4437 ANSI standards. When used at link-time, it may include libraries
4438 or startup files that change the default FPU control word or other
4439 similar optimizations.
4441 This option should never be turned on by any @option{-O} option since
4442 it can result in incorrect output for programs which depend on
4443 an exact implementation of IEEE or ISO rules/specifications for
4446 The default is @option{-fno-unsafe-math-optimizations}.
4448 @item -ffinite-math-only
4449 @opindex ffinite-math-only
4450 Allow optimizations for floating-point arithmetic that assume
4451 that arguments and results are not NaNs or +-Infs.
4453 This option should never be turned on by any @option{-O} option since
4454 it can result in incorrect output for programs which depend on
4455 an exact implementation of IEEE or ISO rules/specifications.
4457 The default is @option{-fno-finite-math-only}.
4459 @item -fno-trapping-math
4460 @opindex fno-trapping-math
4461 Compile code assuming that floating-point operations cannot generate
4462 user-visible traps. These traps include division by zero, overflow,
4463 underflow, inexact result and invalid operation. This option implies
4464 @option{-fno-signaling-nans}. Setting this option may allow faster
4465 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4467 This option should never be turned on by any @option{-O} option since
4468 it can result in incorrect output for programs which depend on
4469 an exact implementation of IEEE or ISO rules/specifications for
4472 The default is @option{-ftrapping-math}.
4474 @item -frounding-math
4475 @opindex frounding-math
4476 Disable transformations and optimizations that assume default floating
4477 point rounding behavior. This is round-to-zero for all floating point
4478 to integer conversions, and round-to-nearest for all other arithmetic
4479 truncations. This option should be specified for programs that change
4480 the FP rounding mode dynamically, or that may be executed with a
4481 non-default rounding mode. This option disables constant folding of
4482 floating point expressions at compile-time (which may be affected by
4483 rounding mode) and arithmetic transformations that are unsafe in the
4484 presence of sign-dependent rounding modes.
4486 The default is @option{-fno-rounding-math}.
4488 This option is experimental and does not currently guarantee to
4489 disable all GCC optimizations that are affected by rounding mode.
4490 Future versions of GCC may provide finer control of this setting
4491 using C99's @code{FENV_ACCESS} pragma. This command line option
4492 will be used to specify the default state for @code{FENV_ACCESS}.
4494 @item -fsignaling-nans
4495 @opindex fsignaling-nans
4496 Compile code assuming that IEEE signaling NaNs may generate user-visible
4497 traps during floating-point operations. Setting this option disables
4498 optimizations that may change the number of exceptions visible with
4499 signaling NaNs. This option implies @option{-ftrapping-math}.
4501 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4504 The default is @option{-fno-signaling-nans}.
4506 This option is experimental and does not currently guarantee to
4507 disable all GCC optimizations that affect signaling NaN behavior.
4509 @item -fsingle-precision-constant
4510 @opindex fsingle-precision-constant
4511 Treat floating point constant as single precision constant instead of
4512 implicitly converting it to double precision constant.
4517 The following options control optimizations that may improve
4518 performance, but are not enabled by any @option{-O} options. This
4519 section includes experimental options that may produce broken code.
4522 @item -fbranch-probabilities
4523 @opindex fbranch-probabilities
4524 After running a program compiled with @option{-fprofile-arcs}
4525 (@pxref{Debugging Options,, Options for Debugging Your Program or
4526 @command{gcc}}), you can compile it a second time using
4527 @option{-fbranch-probabilities}, to improve optimizations based on
4528 the number of times each branch was taken. When the program
4529 compiled with @option{-fprofile-arcs} exits it saves arc execution
4530 counts to a file called @file{@var{sourcename}.gcda} for each source
4531 file The information in this data file is very dependent on the
4532 structure of the generated code, so you must use the same source code
4533 and the same optimization options for both compilations.
4535 With @option{-fbranch-probabilities}, GCC puts a
4536 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4537 These can be used to improve optimization. Currently, they are only
4538 used in one place: in @file{reorg.c}, instead of guessing which path a
4539 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4540 exactly determine which path is taken more often.
4542 @item -fprofile-values
4543 @opindex fprofile-values
4544 If combined with @option{-fprofile-arcs}, it adds code so that some
4545 data about values of expressions in the program is gathered.
4547 With @option{-fbranch-probabilities}, it reads back the data gathered
4548 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
4549 notes to instructions for their later usage in optimizations.
4551 Enabled with @option{-profile-generate} and @option{-profile-use}.
4555 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
4556 a code to gather information about values of expressions.
4558 With @option{-fbranch-probabilities}, it reads back the data gathered
4559 and actually performs the optimizations based on them.
4560 Currently the optimizations include specialization of division operation
4561 using the knowledge about the value of the denominator.
4563 Enabled with @option{-profile-generate} and @option{-profile-use}.
4565 @item -frename-registers
4566 @opindex frename-registers
4567 Attempt to avoid false dependencies in scheduled code by making use
4568 of registers left over after register allocation. This optimization
4569 will most benefit processors with lots of registers. Depending on the
4570 debug information format adopted by the target, however, it can
4571 make debugging impossible, since variables will no longer stay in
4572 a ``home register''.
4574 Not enabled by default at any level because it has known bugs.
4578 Use a graph coloring register allocator. Currently this option is meant
4579 for testing, so we are interested to hear about miscompilations with
4584 Perform tail duplication to enlarge superblock size. This transformation
4585 simplifies the control flow of the function allowing other optimizations to do
4588 Enabled with @option{-profile-use}.
4590 @item -funit-at-a-time
4591 @opindex funit-at-a-time
4592 Parse the whole compilation unit before starting to produce code.
4593 This allows some extra optimizations to take place but consumes more
4596 Enabled at levels @option{-O2}, @option{-O3}.
4598 @item -funroll-loops
4599 @opindex funroll-loops
4600 Unroll loops whose number of iterations can be determined at compile time or
4601 upon entry to the loop. @option{-funroll-loops} implies
4602 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4603 (i.e. complete removal of loops with small constant number of iterations).
4604 This option makes code larger, and may or may not make it run faster.
4606 Enabled with @option{-profile-use}.
4608 @item -funroll-all-loops
4609 @opindex funroll-all-loops
4610 Unroll all loops, even if their number of iterations is uncertain when
4611 the loop is entered. This usually makes programs run more slowly.
4612 @option{-funroll-all-loops} implies the same options as
4613 @option{-funroll-loops}.
4616 @opindex fpeel-loops
4617 Peels the loops for that there is enough information that they do not
4618 roll much (from profile feedback). It also turns on complete loop peeling
4619 (i.e. complete removal of loops with small constant number of iterations).
4621 Enabled with @option{-profile-use}.
4623 @item -funswitch-loops
4624 @opindex funswitch-loops
4625 Move branches with loop invariant conditions out of the loop, with duplicates
4626 of the loop on both branches (modified according to result of the condition).
4628 @item -fold-unroll-loops
4629 @opindex fold-unroll-loops
4630 Unroll loops whose number of iterations can be determined at compile
4631 time or upon entry to the loop, using the old loop unroller whose loop
4632 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4633 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4634 option makes code larger, and may or may not make it run faster.
4636 @item -fold-unroll-all-loops
4637 @opindex fold-unroll-all-loops
4638 Unroll all loops, even if their number of iterations is uncertain when
4639 the loop is entered. This is done using the old loop unroller whose loop
4640 recognition is based on notes from frontend. This usually makes programs run more slowly.
4641 @option{-fold-unroll-all-loops} implies the same options as
4642 @option{-fold-unroll-loops}.
4644 @item -fprefetch-loop-arrays
4645 @opindex fprefetch-loop-arrays
4646 If supported by the target machine, generate instructions to prefetch
4647 memory to improve the performance of loops that access large arrays.
4649 Disabled at level @option{-Os}.
4651 @item -ffunction-sections
4652 @itemx -fdata-sections
4653 @opindex ffunction-sections
4654 @opindex fdata-sections
4655 Place each function or data item into its own section in the output
4656 file if the target supports arbitrary sections. The name of the
4657 function or the name of the data item determines the section's name
4660 Use these options on systems where the linker can perform optimizations
4661 to improve locality of reference in the instruction space. Most systems
4662 using the ELF object format and SPARC processors running Solaris 2 have
4663 linkers with such optimizations. AIX may have these optimizations in
4666 Only use these options when there are significant benefits from doing
4667 so. When you specify these options, the assembler and linker will
4668 create larger object and executable files and will also be slower.
4669 You will not be able to use @code{gprof} on all systems if you
4670 specify this option and you may have problems with debugging if
4671 you specify both this option and @option{-g}.
4673 @item -fbranch-target-load-optimize
4674 @opindex fbranch-target-load-optimize
4675 Perform branch target register load optimization before prologue / epilogue
4677 The use of target registers can typically be exposed only during reload,
4678 thus hoisting loads out of loops and doing inter-block scheduling needs
4679 a separate optimization pass.
4681 @item -fbranch-target-load-optimize2
4682 @opindex fbranch-target-load-optimize2
4683 Perform branch target register load optimization after prologue / epilogue
4686 @item -fbtr-bb-exclusive
4687 @opindex fbtr-bb-exclusive
4688 When performing branch target register load optimization, don't reuse
4689 branch target registers in within any basic block.
4691 @item --param @var{name}=@var{value}
4693 In some places, GCC uses various constants to control the amount of
4694 optimization that is done. For example, GCC will not inline functions
4695 that contain more that a certain number of instructions. You can
4696 control some of these constants on the command-line using the
4697 @option{--param} option.
4699 The names of specific parameters, and the meaning of the values, are
4700 tied to the internals of the compiler, and are subject to change
4701 without notice in future releases.
4703 In each case, the @var{value} is an integer. The allowable choices for
4704 @var{name} are given in the following table:
4707 @item max-crossjump-edges
4708 The maximum number of incoming edges to consider for crossjumping.
4709 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
4710 the number of edges incoming to each block. Increasing values mean
4711 more aggressive optimization, making the compile time increase with
4712 probably small improvement in executable size.
4714 @item max-delay-slot-insn-search
4715 The maximum number of instructions to consider when looking for an
4716 instruction to fill a delay slot. If more than this arbitrary number of
4717 instructions is searched, the time savings from filling the delay slot
4718 will be minimal so stop searching. Increasing values mean more
4719 aggressive optimization, making the compile time increase with probably
4720 small improvement in executable run time.
4722 @item max-delay-slot-live-search
4723 When trying to fill delay slots, the maximum number of instructions to
4724 consider when searching for a block with valid live register
4725 information. Increasing this arbitrarily chosen value means more
4726 aggressive optimization, increasing the compile time. This parameter
4727 should be removed when the delay slot code is rewritten to maintain the
4730 @item max-gcse-memory
4731 The approximate maximum amount of memory that will be allocated in
4732 order to perform the global common subexpression elimination
4733 optimization. If more memory than specified is required, the
4734 optimization will not be done.
4736 @item max-gcse-passes
4737 The maximum number of passes of GCSE to run. The default is 1.
4739 @item max-pending-list-length
4740 The maximum number of pending dependencies scheduling will allow
4741 before flushing the current state and starting over. Large functions
4742 with few branches or calls can create excessively large lists which
4743 needlessly consume memory and resources.
4745 @item max-inline-insns-single
4746 Several parameters control the tree inliner used in gcc.
4747 This number sets the maximum number of instructions (counted in GCC's
4748 internal representation) in a single function that the tree inliner
4749 will consider for inlining. This only affects functions declared
4750 inline and methods implemented in a class declaration (C++).
4751 The default value is 500.
4753 @item max-inline-insns-auto
4754 When you use @option{-finline-functions} (included in @option{-O3}),
4755 a lot of functions that would otherwise not be considered for inlining
4756 by the compiler will be investigated. To those functions, a different
4757 (more restrictive) limit compared to functions declared inline can
4759 The default value is 120.
4761 @item large-function-insns
4762 The limit specifying really large functions. For functions greater than this
4763 limit inlining is constrained by @option{--param large-function-growth}.
4764 This parameter is useful primarily to avoid extreme compilation time caused by non-linear
4765 algorithms used by the backend.
4766 This parameter is ignored when @option{-funit-at-a-time} is not used.
4767 The default value is 3000.
4769 @item large-function-growth
4770 Specifies maximal growth of large function caused by inlining in percents.
4771 This parameter is ignored when @option{-funit-at-a-time} is not used.
4772 The default value is 200.
4774 @item inline-unit-growth
4775 Specifies maximal overall growth of the compilation unit caused by inlining.
4776 This parameter is ignored when @option{-funit-at-a-time} is not used.
4777 The default value is 150.
4779 @item max-inline-insns-rtl
4780 For languages that use the RTL inliner (this happens at a later stage
4781 than tree inlining), you can set the maximum allowable size (counted
4782 in RTL instructions) for the RTL inliner with this parameter.
4783 The default value is 600.
4785 @item max-unrolled-insns
4786 The maximum number of instructions that a loop should have if that loop
4787 is unrolled, and if the loop is unrolled, it determines how many times
4788 the loop code is unrolled.
4790 @item max-average-unrolled-insns
4791 The maximum number of instructions biased by probabilities of their execution
4792 that a loop should have if that loop is unrolled, and if the loop is unrolled,
4793 it determines how many times the loop code is unrolled.
4795 @item max-unroll-times
4796 The maximum number of unrollings of a single loop.
4798 @item max-peeled-insns
4799 The maximum number of instructions that a loop should have if that loop
4800 is peeled, and if the loop is peeled, it determines how many times
4801 the loop code is peeled.
4803 @item max-peel-times
4804 The maximum number of peelings of a single loop.
4806 @item max-completely-peeled-insns
4807 The maximum number of insns of a completely peeled loop.
4809 @item max-completely-peel-times
4810 The maximum number of iterations of a loop to be suitable for complete peeling.
4812 @item max-unswitch-insns
4813 The maximum number of insns of an unswitched loop.
4815 @item max-unswitch-level
4816 The maximum number of branches unswitched in a single loop.
4818 @item hot-bb-count-fraction
4819 Select fraction of the maximal count of repetitions of basic block in program
4820 given basic block needs to have to be considered hot.
4822 @item hot-bb-frequency-fraction
4823 Select fraction of the maximal frequency of executions of basic block in
4824 function given basic block needs to have to be considered hot
4826 @item tracer-dynamic-coverage
4827 @itemx tracer-dynamic-coverage-feedback
4829 This value is used to limit superblock formation once the given percentage of
4830 executed instructions is covered. This limits unnecessary code size
4833 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4834 feedback is available. The real profiles (as opposed to statically estimated
4835 ones) are much less balanced allowing the threshold to be larger value.
4837 @item tracer-max-code-growth
4838 Stop tail duplication once code growth has reached given percentage. This is
4839 rather hokey argument, as most of the duplicates will be eliminated later in
4840 cross jumping, so it may be set to much higher values than is the desired code
4843 @item tracer-min-branch-ratio
4845 Stop reverse growth when the reverse probability of best edge is less than this
4846 threshold (in percent).
4848 @item tracer-min-branch-ratio
4849 @itemx tracer-min-branch-ratio-feedback
4851 Stop forward growth if the best edge do have probability lower than this
4854 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
4855 compilation for profile feedback and one for compilation without. The value
4856 for compilation with profile feedback needs to be more conservative (higher) in
4857 order to make tracer effective.
4859 @item max-cse-path-length
4861 Maximum number of basic blocks on path that cse considers. The default is 10.
4863 @item ggc-min-expand
4865 GCC uses a garbage collector to manage its own memory allocation. This
4866 parameter specifies the minimum percentage by which the garbage
4867 collector's heap should be allowed to expand between collections.
4868 Tuning this may improve compilation speed; it has no effect on code
4871 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
4872 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
4873 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
4874 GCC is not able to calculate RAM on a particular platform, the lower
4875 bound of 30% is used. Setting this parameter and
4876 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4877 every opportunity. This is extremely slow, but can be useful for
4880 @item ggc-min-heapsize
4882 Minimum size of the garbage collector's heap before it begins bothering
4883 to collect garbage. The first collection occurs after the heap expands
4884 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4885 tuning this may improve compilation speed, and has no effect on code
4888 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
4889 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
4890 available, the notion of "RAM" is the smallest of actual RAM,
4891 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
4892 RAM on a particular platform, the lower bound is used. Setting this
4893 parameter very large effectively disables garbage collection. Setting
4894 this parameter and @option{ggc-min-expand} to zero causes a full
4895 collection to occur at every opportunity.
4897 @item max-reload-search-insns
4898 The maximum number of instruction reload should look backward for equivalent
4899 register. Increasing values mean more aggressive optimization, making the
4900 compile time increase with probably slightly better performance. The default
4903 @item max-cselib-memory-location
4904 The maximum number of memory locations cselib should take into acount.
4905 Increasing values mean more aggressive optimization, making the compile time
4906 increase with probably slightly better performance. The default value is 500.
4908 @item reorder-blocks-duplicate
4909 @itemx reorder-blocks-duplicate-feedback
4911 Used by basic block reordering pass to decide whether to use unconditional
4912 branch or duplicate the code on its destination. Code is duplicated when its
4913 estimated size is smaller than this value multiplied by the estimated size of
4914 unconditional jump in the hot spots of the program.
4916 The @option{reorder-block-duplicate-feedback} is used only when profile
4917 feedback is available and may be set to higher values than
4918 @option{reorder-block-duplicate} since information about the hot spots is more
4921 @item max-sched-region-blocks
4922 The maximum number of blocks in a region to be considered for
4923 interblock scheduling. The default value is 10.
4925 @item max-sched-region-insns
4926 The maximum number of insns in a region to be considered for
4927 interblock scheduling. The default value is 100.
4931 @node Preprocessor Options
4932 @section Options Controlling the Preprocessor
4933 @cindex preprocessor options
4934 @cindex options, preprocessor
4936 These options control the C preprocessor, which is run on each C source
4937 file before actual compilation.
4939 If you use the @option{-E} option, nothing is done except preprocessing.
4940 Some of these options make sense only together with @option{-E} because
4941 they cause the preprocessor output to be unsuitable for actual
4946 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4947 and pass @var{option} directly through to the preprocessor. If
4948 @var{option} contains commas, it is split into multiple options at the
4949 commas. However, many options are modified, translated or interpreted
4950 by the compiler driver before being passed to the preprocessor, and
4951 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4952 interface is undocumented and subject to change, so whenever possible
4953 you should avoid using @option{-Wp} and let the driver handle the
4956 @item -Xpreprocessor @var{option}
4957 @opindex preprocessor
4958 Pass @var{option} as an option to the preprocessor. You can use this to
4959 supply system-specific preprocessor options which GCC does not know how to
4962 If you want to pass an option that takes an argument, you must use
4963 @option{-Xpreprocessor} twice, once for the option and once for the argument.
4966 @include cppopts.texi
4968 @node Assembler Options
4969 @section Passing Options to the Assembler
4971 @c prevent bad page break with this line
4972 You can pass options to the assembler.
4975 @item -Wa,@var{option}
4977 Pass @var{option} as an option to the assembler. If @var{option}
4978 contains commas, it is split into multiple options at the commas.
4980 @item -Xassembler @var{option}
4982 Pass @var{option} as an option to the assembler. You can use this to
4983 supply system-specific assembler options which GCC does not know how to
4986 If you want to pass an option that takes an argument, you must use
4987 @option{-Xassembler} twice, once for the option and once for the argument.
4992 @section Options for Linking
4993 @cindex link options
4994 @cindex options, linking
4996 These options come into play when the compiler links object files into
4997 an executable output file. They are meaningless if the compiler is
4998 not doing a link step.
5002 @item @var{object-file-name}
5003 A file name that does not end in a special recognized suffix is
5004 considered to name an object file or library. (Object files are
5005 distinguished from libraries by the linker according to the file
5006 contents.) If linking is done, these object files are used as input
5015 If any of these options is used, then the linker is not run, and
5016 object file names should not be used as arguments. @xref{Overall
5020 @item -l@var{library}
5021 @itemx -l @var{library}
5023 Search the library named @var{library} when linking. (The second
5024 alternative with the library as a separate argument is only for
5025 POSIX compliance and is not recommended.)
5027 It makes a difference where in the command you write this option; the
5028 linker searches and processes libraries and object files in the order they
5029 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
5030 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
5031 to functions in @samp{z}, those functions may not be loaded.
5033 The linker searches a standard list of directories for the library,
5034 which is actually a file named @file{lib@var{library}.a}. The linker
5035 then uses this file as if it had been specified precisely by name.
5037 The directories searched include several standard system directories
5038 plus any that you specify with @option{-L}.
5040 Normally the files found this way are library files---archive files
5041 whose members are object files. The linker handles an archive file by
5042 scanning through it for members which define symbols that have so far
5043 been referenced but not defined. But if the file that is found is an
5044 ordinary object file, it is linked in the usual fashion. The only
5045 difference between using an @option{-l} option and specifying a file name
5046 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
5047 and searches several directories.
5051 You need this special case of the @option{-l} option in order to
5052 link an Objective-C program.
5055 @opindex nostartfiles
5056 Do not use the standard system startup files when linking.
5057 The standard system libraries are used normally, unless @option{-nostdlib}
5058 or @option{-nodefaultlibs} is used.
5060 @item -nodefaultlibs
5061 @opindex nodefaultlibs
5062 Do not use the standard system libraries when linking.
5063 Only the libraries you specify will be passed to the linker.
5064 The standard startup files are used normally, unless @option{-nostartfiles}
5065 is used. The compiler may generate calls to memcmp, memset, and memcpy
5066 for System V (and ISO C) environments or to bcopy and bzero for
5067 BSD environments. These entries are usually resolved by entries in
5068 libc. These entry points should be supplied through some other
5069 mechanism when this option is specified.
5073 Do not use the standard system startup files or libraries when linking.
5074 No startup files and only the libraries you specify will be passed to
5075 the linker. The compiler may generate calls to memcmp, memset, and memcpy
5076 for System V (and ISO C) environments or to bcopy and bzero for
5077 BSD environments. These entries are usually resolved by entries in
5078 libc. These entry points should be supplied through some other
5079 mechanism when this option is specified.
5081 @cindex @option{-lgcc}, use with @option{-nostdlib}
5082 @cindex @option{-nostdlib} and unresolved references
5083 @cindex unresolved references and @option{-nostdlib}
5084 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5085 @cindex @option{-nodefaultlibs} and unresolved references
5086 @cindex unresolved references and @option{-nodefaultlibs}
5087 One of the standard libraries bypassed by @option{-nostdlib} and
5088 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5089 that GCC uses to overcome shortcomings of particular machines, or special
5090 needs for some languages.
5091 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5092 Collection (GCC) Internals},
5093 for more discussion of @file{libgcc.a}.)
5094 In most cases, you need @file{libgcc.a} even when you want to avoid
5095 other standard libraries. In other words, when you specify @option{-nostdlib}
5096 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5097 This ensures that you have no unresolved references to internal GCC
5098 library subroutines. (For example, @samp{__main}, used to ensure C++
5099 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5100 GNU Compiler Collection (GCC) Internals}.)
5104 Produce a position independent executable on targets which support it.
5105 For predictable results, you must also specify the same set of options
5106 that were used to generate code (@option{-fpie}, @option{-fPIE},
5107 or model suboptions) when you specify this option.
5111 Remove all symbol table and relocation information from the executable.
5115 On systems that support dynamic linking, this prevents linking with the shared
5116 libraries. On other systems, this option has no effect.
5120 Produce a shared object which can then be linked with other objects to
5121 form an executable. Not all systems support this option. For predictable
5122 results, you must also specify the same set of options that were used to
5123 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5124 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5125 needs to build supplementary stub code for constructors to work. On
5126 multi-libbed systems, @samp{gcc -shared} must select the correct support
5127 libraries to link against. Failing to supply the correct flags may lead
5128 to subtle defects. Supplying them in cases where they are not necessary
5131 @item -shared-libgcc
5132 @itemx -static-libgcc
5133 @opindex shared-libgcc
5134 @opindex static-libgcc
5135 On systems that provide @file{libgcc} as a shared library, these options
5136 force the use of either the shared or static version respectively.
5137 If no shared version of @file{libgcc} was built when the compiler was
5138 configured, these options have no effect.
5140 There are several situations in which an application should use the
5141 shared @file{libgcc} instead of the static version. The most common
5142 of these is when the application wishes to throw and catch exceptions
5143 across different shared libraries. In that case, each of the libraries
5144 as well as the application itself should use the shared @file{libgcc}.
5146 Therefore, the G++ and GCJ drivers automatically add
5147 @option{-shared-libgcc} whenever you build a shared library or a main
5148 executable, because C++ and Java programs typically use exceptions, so
5149 this is the right thing to do.
5151 If, instead, you use the GCC driver to create shared libraries, you may
5152 find that they will not always be linked with the shared @file{libgcc}.
5153 If GCC finds, at its configuration time, that you have a GNU linker that
5154 does not support option @option{--eh-frame-hdr}, it will link the shared
5155 version of @file{libgcc} into shared libraries by default. Otherwise,
5156 it will take advantage of the linker and optimize away the linking with
5157 the shared version of @file{libgcc}, linking with the static version of
5158 libgcc by default. This allows exceptions to propagate through such
5159 shared libraries, without incurring relocation costs at library load
5162 However, if a library or main executable is supposed to throw or catch
5163 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5164 for the languages used in the program, or using the option
5165 @option{-shared-libgcc}, such that it is linked with the shared
5170 Bind references to global symbols when building a shared object. Warn
5171 about any unresolved references (unless overridden by the link editor
5172 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5175 @item -Xlinker @var{option}
5177 Pass @var{option} as an option to the linker. You can use this to
5178 supply system-specific linker options which GCC does not know how to
5181 If you want to pass an option that takes an argument, you must use
5182 @option{-Xlinker} twice, once for the option and once for the argument.
5183 For example, to pass @option{-assert definitions}, you must write
5184 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5185 @option{-Xlinker "-assert definitions"}, because this passes the entire
5186 string as a single argument, which is not what the linker expects.
5188 @item -Wl,@var{option}
5190 Pass @var{option} as an option to the linker. If @var{option} contains
5191 commas, it is split into multiple options at the commas.
5193 @item -u @var{symbol}
5195 Pretend the symbol @var{symbol} is undefined, to force linking of
5196 library modules to define it. You can use @option{-u} multiple times with
5197 different symbols to force loading of additional library modules.
5200 @node Directory Options
5201 @section Options for Directory Search
5202 @cindex directory options
5203 @cindex options, directory search
5206 These options specify directories to search for header files, for
5207 libraries and for parts of the compiler:
5212 Add the directory @var{dir} to the head of the list of directories to be
5213 searched for header files. This can be used to override a system header
5214 file, substituting your own version, since these directories are
5215 searched before the system header file directories. However, you should
5216 not use this option to add directories that contain vendor-supplied
5217 system header files (use @option{-isystem} for that). If you use more than
5218 one @option{-I} option, the directories are scanned in left-to-right
5219 order; the standard system directories come after.
5221 If a standard system include directory, or a directory specified with
5222 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5223 option will be ignored. The directory will still be searched but as a
5224 system directory at its normal position in the system include chain.
5225 This is to ensure that GCC's procedure to fix buggy system headers and
5226 the ordering for the include_next directive are not inadvertently changed.
5227 If you really need to change the search order for system directories,
5228 use the @option{-nostdinc} and/or @option{-isystem} options.
5232 Any directories you specify with @option{-I} options before the @option{-I-}
5233 option are searched only for the case of @samp{#include "@var{file}"};
5234 they are not searched for @samp{#include <@var{file}>}.
5236 If additional directories are specified with @option{-I} options after
5237 the @option{-I-}, these directories are searched for all @samp{#include}
5238 directives. (Ordinarily @emph{all} @option{-I} directories are used
5241 In addition, the @option{-I-} option inhibits the use of the current
5242 directory (where the current input file came from) as the first search
5243 directory for @samp{#include "@var{file}"}. There is no way to
5244 override this effect of @option{-I-}. With @option{-I.} you can specify
5245 searching the directory which was current when the compiler was
5246 invoked. That is not exactly the same as what the preprocessor does
5247 by default, but it is often satisfactory.
5249 @option{-I-} does not inhibit the use of the standard system directories
5250 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5255 Add directory @var{dir} to the list of directories to be searched
5258 @item -B@var{prefix}
5260 This option specifies where to find the executables, libraries,
5261 include files, and data files of the compiler itself.
5263 The compiler driver program runs one or more of the subprograms
5264 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5265 @var{prefix} as a prefix for each program it tries to run, both with and
5266 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5268 For each subprogram to be run, the compiler driver first tries the
5269 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5270 was not specified, the driver tries two standard prefixes, which are
5271 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5272 those results in a file name that is found, the unmodified program
5273 name is searched for using the directories specified in your
5274 @env{PATH} environment variable.
5276 The compiler will check to see if the path provided by the @option{-B}
5277 refers to a directory, and if necessary it will add a directory
5278 separator character at the end of the path.
5280 @option{-B} prefixes that effectively specify directory names also apply
5281 to libraries in the linker, because the compiler translates these
5282 options into @option{-L} options for the linker. They also apply to
5283 includes files in the preprocessor, because the compiler translates these
5284 options into @option{-isystem} options for the preprocessor. In this case,
5285 the compiler appends @samp{include} to the prefix.
5287 The run-time support file @file{libgcc.a} can also be searched for using
5288 the @option{-B} prefix, if needed. If it is not found there, the two
5289 standard prefixes above are tried, and that is all. The file is left
5290 out of the link if it is not found by those means.
5292 Another way to specify a prefix much like the @option{-B} prefix is to use
5293 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5296 As a special kludge, if the path provided by @option{-B} is
5297 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5298 9, then it will be replaced by @file{[dir/]include}. This is to help
5299 with boot-strapping the compiler.
5301 @item -specs=@var{file}
5303 Process @var{file} after the compiler reads in the standard @file{specs}
5304 file, in order to override the defaults that the @file{gcc} driver
5305 program uses when determining what switches to pass to @file{cc1},
5306 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5307 @option{-specs=@var{file}} can be specified on the command line, and they
5308 are processed in order, from left to right.
5314 @section Specifying subprocesses and the switches to pass to them
5317 @command{gcc} is a driver program. It performs its job by invoking a
5318 sequence of other programs to do the work of compiling, assembling and
5319 linking. GCC interprets its command-line parameters and uses these to
5320 deduce which programs it should invoke, and which command-line options
5321 it ought to place on their command lines. This behavior is controlled
5322 by @dfn{spec strings}. In most cases there is one spec string for each
5323 program that GCC can invoke, but a few programs have multiple spec
5324 strings to control their behavior. The spec strings built into GCC can
5325 be overridden by using the @option{-specs=} command-line switch to specify
5328 @dfn{Spec files} are plaintext files that are used to construct spec
5329 strings. They consist of a sequence of directives separated by blank
5330 lines. The type of directive is determined by the first non-whitespace
5331 character on the line and it can be one of the following:
5334 @item %@var{command}
5335 Issues a @var{command} to the spec file processor. The commands that can
5339 @item %include <@var{file}>
5341 Search for @var{file} and insert its text at the current point in the
5344 @item %include_noerr <@var{file}>
5345 @cindex %include_noerr
5346 Just like @samp{%include}, but do not generate an error message if the include
5347 file cannot be found.
5349 @item %rename @var{old_name} @var{new_name}
5351 Rename the spec string @var{old_name} to @var{new_name}.
5355 @item *[@var{spec_name}]:
5356 This tells the compiler to create, override or delete the named spec
5357 string. All lines after this directive up to the next directive or
5358 blank line are considered to be the text for the spec string. If this
5359 results in an empty string then the spec will be deleted. (Or, if the
5360 spec did not exist, then nothing will happened.) Otherwise, if the spec
5361 does not currently exist a new spec will be created. If the spec does
5362 exist then its contents will be overridden by the text of this
5363 directive, unless the first character of that text is the @samp{+}
5364 character, in which case the text will be appended to the spec.
5366 @item [@var{suffix}]:
5367 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5368 and up to the next directive or blank line are considered to make up the
5369 spec string for the indicated suffix. When the compiler encounters an
5370 input file with the named suffix, it will processes the spec string in
5371 order to work out how to compile that file. For example:
5378 This says that any input file whose name ends in @samp{.ZZ} should be
5379 passed to the program @samp{z-compile}, which should be invoked with the
5380 command-line switch @option{-input} and with the result of performing the
5381 @samp{%i} substitution. (See below.)
5383 As an alternative to providing a spec string, the text that follows a
5384 suffix directive can be one of the following:
5387 @item @@@var{language}
5388 This says that the suffix is an alias for a known @var{language}. This is
5389 similar to using the @option{-x} command-line switch to GCC to specify a
5390 language explicitly. For example:
5397 Says that .ZZ files are, in fact, C++ source files.
5400 This causes an error messages saying:
5403 @var{name} compiler not installed on this system.
5407 GCC already has an extensive list of suffixes built into it.
5408 This directive will add an entry to the end of the list of suffixes, but
5409 since the list is searched from the end backwards, it is effectively
5410 possible to override earlier entries using this technique.
5414 GCC has the following spec strings built into it. Spec files can
5415 override these strings or create their own. Note that individual
5416 targets can also add their own spec strings to this list.
5419 asm Options to pass to the assembler
5420 asm_final Options to pass to the assembler post-processor
5421 cpp Options to pass to the C preprocessor
5422 cc1 Options to pass to the C compiler
5423 cc1plus Options to pass to the C++ compiler
5424 endfile Object files to include at the end of the link
5425 link Options to pass to the linker
5426 lib Libraries to include on the command line to the linker
5427 libgcc Decides which GCC support library to pass to the linker
5428 linker Sets the name of the linker
5429 predefines Defines to be passed to the C preprocessor
5430 signed_char Defines to pass to CPP to say whether @code{char} is signed
5432 startfile Object files to include at the start of the link
5435 Here is a small example of a spec file:
5441 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5444 This example renames the spec called @samp{lib} to @samp{old_lib} and
5445 then overrides the previous definition of @samp{lib} with a new one.
5446 The new definition adds in some extra command-line options before
5447 including the text of the old definition.
5449 @dfn{Spec strings} are a list of command-line options to be passed to their
5450 corresponding program. In addition, the spec strings can contain
5451 @samp{%}-prefixed sequences to substitute variable text or to
5452 conditionally insert text into the command line. Using these constructs
5453 it is possible to generate quite complex command lines.
5455 Here is a table of all defined @samp{%}-sequences for spec
5456 strings. Note that spaces are not generated automatically around the
5457 results of expanding these sequences. Therefore you can concatenate them
5458 together or combine them with constant text in a single argument.
5462 Substitute one @samp{%} into the program name or argument.
5465 Substitute the name of the input file being processed.
5468 Substitute the basename of the input file being processed.
5469 This is the substring up to (and not including) the last period
5470 and not including the directory.
5473 This is the same as @samp{%b}, but include the file suffix (text after
5477 Marks the argument containing or following the @samp{%d} as a
5478 temporary file name, so that that file will be deleted if GCC exits
5479 successfully. Unlike @samp{%g}, this contributes no text to the
5482 @item %g@var{suffix}
5483 Substitute a file name that has suffix @var{suffix} and is chosen
5484 once per compilation, and mark the argument in the same way as
5485 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5486 name is now chosen in a way that is hard to predict even when previously
5487 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5488 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5489 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5490 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5491 was simply substituted with a file name chosen once per compilation,
5492 without regard to any appended suffix (which was therefore treated
5493 just like ordinary text), making such attacks more likely to succeed.
5495 @item %u@var{suffix}
5496 Like @samp{%g}, but generates a new temporary file name even if
5497 @samp{%u@var{suffix}} was already seen.
5499 @item %U@var{suffix}
5500 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5501 new one if there is no such last file name. In the absence of any
5502 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5503 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5504 would involve the generation of two distinct file names, one
5505 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5506 simply substituted with a file name chosen for the previous @samp{%u},
5507 without regard to any appended suffix.
5509 @item %j@var{suffix}
5510 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5511 writable, and if save-temps is off; otherwise, substitute the name
5512 of a temporary file, just like @samp{%u}. This temporary file is not
5513 meant for communication between processes, but rather as a junk
5516 @item %|@var{suffix}
5517 @itemx %m@var{suffix}
5518 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5519 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5520 all. These are the two most common ways to instruct a program that it
5521 should read from standard input or write to standard output. If you
5522 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5523 construct: see for example @file{f/lang-specs.h}.
5525 @item %.@var{SUFFIX}
5526 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5527 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5528 terminated by the next space or %.
5531 Marks the argument containing or following the @samp{%w} as the
5532 designated output file of this compilation. This puts the argument
5533 into the sequence of arguments that @samp{%o} will substitute later.
5536 Substitutes the names of all the output files, with spaces
5537 automatically placed around them. You should write spaces
5538 around the @samp{%o} as well or the results are undefined.
5539 @samp{%o} is for use in the specs for running the linker.
5540 Input files whose names have no recognized suffix are not compiled
5541 at all, but they are included among the output files, so they will
5545 Substitutes the suffix for object files. Note that this is
5546 handled specially when it immediately follows @samp{%g, %u, or %U},
5547 because of the need for those to form complete file names. The
5548 handling is such that @samp{%O} is treated exactly as if it had already
5549 been substituted, except that @samp{%g, %u, and %U} do not currently
5550 support additional @var{suffix} characters following @samp{%O} as they would
5551 following, for example, @samp{.o}.
5554 Substitutes the standard macro predefinitions for the
5555 current target machine. Use this when running @code{cpp}.
5558 Like @samp{%p}, but puts @samp{__} before and after the name of each
5559 predefined macro, except for macros that start with @samp{__} or with
5560 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5564 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5565 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5566 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5570 Current argument is the name of a library or startup file of some sort.
5571 Search for that file in a standard list of directories and substitute
5572 the full name found.
5575 Print @var{str} as an error message. @var{str} is terminated by a newline.
5576 Use this when inconsistent options are detected.
5579 Substitute the contents of spec string @var{name} at this point.
5582 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5584 @item %x@{@var{option}@}
5585 Accumulate an option for @samp{%X}.
5588 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5592 Output the accumulated assembler options specified by @option{-Wa}.
5595 Output the accumulated preprocessor options specified by @option{-Wp}.
5598 Process the @code{asm} spec. This is used to compute the
5599 switches to be passed to the assembler.
5602 Process the @code{asm_final} spec. This is a spec string for
5603 passing switches to an assembler post-processor, if such a program is
5607 Process the @code{link} spec. This is the spec for computing the
5608 command line passed to the linker. Typically it will make use of the
5609 @samp{%L %G %S %D and %E} sequences.
5612 Dump out a @option{-L} option for each directory that GCC believes might
5613 contain startup files. If the target supports multilibs then the
5614 current multilib directory will be prepended to each of these paths.
5617 Output the multilib directory with directory separators replaced with
5618 @samp{_}. If multilib directories are not set, or the multilib directory is
5619 @file{.} then this option emits nothing.
5622 Process the @code{lib} spec. This is a spec string for deciding which
5623 libraries should be included on the command line to the linker.
5626 Process the @code{libgcc} spec. This is a spec string for deciding
5627 which GCC support library should be included on the command line to the linker.
5630 Process the @code{startfile} spec. This is a spec for deciding which
5631 object files should be the first ones passed to the linker. Typically
5632 this might be a file named @file{crt0.o}.
5635 Process the @code{endfile} spec. This is a spec string that specifies
5636 the last object files that will be passed to the linker.
5639 Process the @code{cpp} spec. This is used to construct the arguments
5640 to be passed to the C preprocessor.
5643 Process the @code{signed_char} spec. This is intended to be used
5644 to tell cpp whether a char is signed. It typically has the definition:
5646 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
5650 Process the @code{cc1} spec. This is used to construct the options to be
5651 passed to the actual C compiler (@samp{cc1}).
5654 Process the @code{cc1plus} spec. This is used to construct the options to be
5655 passed to the actual C++ compiler (@samp{cc1plus}).
5658 Substitute the variable part of a matched option. See below.
5659 Note that each comma in the substituted string is replaced by
5663 Remove all occurrences of @code{-S} from the command line. Note---this
5664 command is position dependent. @samp{%} commands in the spec string
5665 before this one will see @code{-S}, @samp{%} commands in the spec string
5666 after this one will not.
5668 @item %:@var{function}(@var{args})
5669 Call the named function @var{function}, passing it @var{args}.
5670 @var{args} is first processed as a nested spec string, then split
5671 into an argument vector in the usual fashion. The function returns
5672 a string which is processed as if it had appeared literally as part
5673 of the current spec.
5675 The following built-in spec functions are provided:
5678 @item @code{if-exists}
5679 The @code{if-exists} spec function takes one argument, an absolute
5680 pathname to a file. If the file exists, @code{if-exists} returns the
5681 pathname. Here is a small example of its usage:
5685 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5688 @item @code{if-exists-else}
5689 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5690 spec function, except that it takes two arguments. The first argument is
5691 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5692 returns the pathname. If it does not exist, it returns the second argument.
5693 This way, @code{if-exists-else} can be used to select one file or another,
5694 based on the existence of the first. Here is a small example of its usage:
5698 crt0%O%s %:if-exists(crti%O%s) \
5699 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5704 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5705 If that switch was not specified, this substitutes nothing. Note that
5706 the leading dash is omitted when specifying this option, and it is
5707 automatically inserted if the substitution is performed. Thus the spec
5708 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5709 and would output the command line option @option{-foo}.
5711 @item %W@{@code{S}@}
5712 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5715 @item %@{@code{S}*@}
5716 Substitutes all the switches specified to GCC whose names start
5717 with @code{-S}, but which also take an argument. This is used for
5718 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5719 GCC considers @option{-o foo} as being
5720 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5721 text, including the space. Thus two arguments would be generated.
5723 @item %@{@code{S}*&@code{T}*@}
5724 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5725 (the order of @code{S} and @code{T} in the spec is not significant).
5726 There can be any number of ampersand-separated variables; for each the
5727 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5729 @item %@{@code{S}:@code{X}@}
5730 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
5732 @item %@{!@code{S}:@code{X}@}
5733 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
5735 @item %@{@code{S}*:@code{X}@}
5736 Substitutes @code{X} if one or more switches whose names start with
5737 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
5738 once, no matter how many such switches appeared. However, if @code{%*}
5739 appears somewhere in @code{X}, then @code{X} will be substituted once
5740 for each matching switch, with the @code{%*} replaced by the part of
5741 that switch that matched the @code{*}.
5743 @item %@{.@code{S}:@code{X}@}
5744 Substitutes @code{X}, if processing a file with suffix @code{S}.
5746 @item %@{!.@code{S}:@code{X}@}
5747 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
5749 @item %@{@code{S}|@code{P}:@code{X}@}
5750 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
5751 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
5752 although they have a stronger binding than the @samp{|}. If @code{%*}
5753 appears in @code{X}, all of the alternatives must be starred, and only
5754 the first matching alternative is substituted.
5756 For example, a spec string like this:
5759 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5762 will output the following command-line options from the following input
5763 command-line options:
5768 -d fred.c -foo -baz -boggle
5769 -d jim.d -bar -baz -boggle
5772 @item %@{S:X; T:Y; :D@}
5774 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
5775 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
5776 be as many clauses as you need. This may be combined with @code{.},
5777 @code{!}, @code{|}, and @code{*} as needed.
5782 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
5783 construct may contain other nested @samp{%} constructs or spaces, or
5784 even newlines. They are processed as usual, as described above.
5785 Trailing white space in @code{X} is ignored. White space may also
5786 appear anywhere on the left side of the colon in these constructs,
5787 except between @code{.} or @code{*} and the corresponding word.
5789 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
5790 handled specifically in these constructs. If another value of
5791 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5792 @option{-W} switch is found later in the command line, the earlier
5793 switch value is ignored, except with @{@code{S}*@} where @code{S} is
5794 just one letter, which passes all matching options.
5796 The character @samp{|} at the beginning of the predicate text is used to
5797 indicate that a command should be piped to the following command, but
5798 only if @option{-pipe} is specified.
5800 It is built into GCC which switches take arguments and which do not.
5801 (You might think it would be useful to generalize this to allow each
5802 compiler's spec to say which switches take arguments. But this cannot
5803 be done in a consistent fashion. GCC cannot even decide which input
5804 files have been specified without knowing which switches take arguments,
5805 and it must know which input files to compile in order to tell which
5808 GCC also knows implicitly that arguments starting in @option{-l} are to be
5809 treated as compiler output files, and passed to the linker in their
5810 proper position among the other output files.
5812 @c man begin OPTIONS
5814 @node Target Options
5815 @section Specifying Target Machine and Compiler Version
5816 @cindex target options
5817 @cindex cross compiling
5818 @cindex specifying machine version
5819 @cindex specifying compiler version and target machine
5820 @cindex compiler version, specifying
5821 @cindex target machine, specifying
5823 The usual way to run GCC is to run the executable called @file{gcc}, or
5824 @file{<machine>-gcc} when cross-compiling, or
5825 @file{<machine>-gcc-<version>} to run a version other than the one that
5826 was installed last. Sometimes this is inconvenient, so GCC provides
5827 options that will switch to another cross-compiler or version.
5830 @item -b @var{machine}
5832 The argument @var{machine} specifies the target machine for compilation.
5834 The value to use for @var{machine} is the same as was specified as the
5835 machine type when configuring GCC as a cross-compiler. For
5836 example, if a cross-compiler was configured with @samp{configure
5837 i386v}, meaning to compile for an 80386 running System V, then you
5838 would specify @option{-b i386v} to run that cross compiler.
5840 @item -V @var{version}
5842 The argument @var{version} specifies which version of GCC to run.
5843 This is useful when multiple versions are installed. For example,
5844 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5847 The @option{-V} and @option{-b} options work by running the
5848 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5849 use them if you can just run that directly.
5851 @node Submodel Options
5852 @section Hardware Models and Configurations
5853 @cindex submodel options
5854 @cindex specifying hardware config
5855 @cindex hardware models and configurations, specifying
5856 @cindex machine dependent options
5858 Earlier we discussed the standard option @option{-b} which chooses among
5859 different installed compilers for completely different target
5860 machines, such as VAX vs.@: 68000 vs.@: 80386.
5862 In addition, each of these target machine types can have its own
5863 special options, starting with @samp{-m}, to choose among various
5864 hardware models or configurations---for example, 68010 vs 68020,
5865 floating coprocessor or none. A single installed version of the
5866 compiler can compile for any model or configuration, according to the
5869 Some configurations of the compiler also support additional special
5870 options, usually for compatibility with other compilers on the same
5873 These options are defined by the macro @code{TARGET_SWITCHES} in the
5874 machine description. The default for the options is also defined by
5875 that macro, which enables you to change the defaults.
5885 * RS/6000 and PowerPC Options::
5888 * i386 and x86-64 Options::
5890 * DEC Alpha Options::
5891 * DEC Alpha/VMS Options::
5894 * System V Options::
5895 * TMS320C3x/C4x Options::
5902 * S/390 and zSeries Options::
5906 * Xstormy16 Options::
5911 @node M680x0 Options
5912 @subsection M680x0 Options
5913 @cindex M680x0 options
5915 These are the @samp{-m} options defined for the 68000 series. The default
5916 values for these options depends on which style of 68000 was selected when
5917 the compiler was configured; the defaults for the most common choices are
5925 Generate output for a 68000. This is the default
5926 when the compiler is configured for 68000-based systems.
5928 Use this option for microcontrollers with a 68000 or EC000 core,
5929 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5935 Generate output for a 68020. This is the default
5936 when the compiler is configured for 68020-based systems.
5940 Generate output containing 68881 instructions for floating point.
5941 This is the default for most 68020 systems unless @option{--nfp} was
5942 specified when the compiler was configured.
5946 Generate output for a 68030. This is the default when the compiler is
5947 configured for 68030-based systems.
5951 Generate output for a 68040. This is the default when the compiler is
5952 configured for 68040-based systems.
5954 This option inhibits the use of 68881/68882 instructions that have to be
5955 emulated by software on the 68040. Use this option if your 68040 does not
5956 have code to emulate those instructions.
5960 Generate output for a 68060. This is the default when the compiler is
5961 configured for 68060-based systems.
5963 This option inhibits the use of 68020 and 68881/68882 instructions that
5964 have to be emulated by software on the 68060. Use this option if your 68060
5965 does not have code to emulate those instructions.
5969 Generate output for a CPU32. This is the default
5970 when the compiler is configured for CPU32-based systems.
5972 Use this option for microcontrollers with a
5973 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5974 68336, 68340, 68341, 68349 and 68360.
5978 Generate output for a 520X ``coldfire'' family cpu. This is the default
5979 when the compiler is configured for 520X-based systems.
5981 Use this option for microcontroller with a 5200 core, including
5982 the MCF5202, MCF5203, MCF5204 and MCF5202.
5987 Generate output for a 68040, without using any of the new instructions.
5988 This results in code which can run relatively efficiently on either a
5989 68020/68881 or a 68030 or a 68040. The generated code does use the
5990 68881 instructions that are emulated on the 68040.
5994 Generate output for a 68060, without using any of the new instructions.
5995 This results in code which can run relatively efficiently on either a
5996 68020/68881 or a 68030 or a 68040. The generated code does use the
5997 68881 instructions that are emulated on the 68060.
6000 @opindex msoft-float
6001 Generate output containing library calls for floating point.
6002 @strong{Warning:} the requisite libraries are not available for all m68k
6003 targets. Normally the facilities of the machine's usual C compiler are
6004 used, but this can't be done directly in cross-compilation. You must
6005 make your own arrangements to provide suitable library functions for
6006 cross-compilation. The embedded targets @samp{m68k-*-aout} and
6007 @samp{m68k-*-coff} do provide software floating point support.
6011 Consider type @code{int} to be 16 bits wide, like @code{short int}.
6014 @opindex mnobitfield
6015 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
6016 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
6020 Do use the bit-field instructions. The @option{-m68020} option implies
6021 @option{-mbitfield}. This is the default if you use a configuration
6022 designed for a 68020.
6026 Use a different function-calling convention, in which functions
6027 that take a fixed number of arguments return with the @code{rtd}
6028 instruction, which pops their arguments while returning. This
6029 saves one instruction in the caller since there is no need to pop
6030 the arguments there.
6032 This calling convention is incompatible with the one normally
6033 used on Unix, so you cannot use it if you need to call libraries
6034 compiled with the Unix compiler.
6036 Also, you must provide function prototypes for all functions that
6037 take variable numbers of arguments (including @code{printf});
6038 otherwise incorrect code will be generated for calls to those
6041 In addition, seriously incorrect code will result if you call a
6042 function with too many arguments. (Normally, extra arguments are
6043 harmlessly ignored.)
6045 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
6046 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
6049 @itemx -mno-align-int
6051 @opindex mno-align-int
6052 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
6053 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
6054 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
6055 Aligning variables on 32-bit boundaries produces code that runs somewhat
6056 faster on processors with 32-bit busses at the expense of more memory.
6058 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
6059 align structures containing the above types differently than
6060 most published application binary interface specifications for the m68k.
6064 Use the pc-relative addressing mode of the 68000 directly, instead of
6065 using a global offset table. At present, this option implies @option{-fpic},
6066 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
6067 not presently supported with @option{-mpcrel}, though this could be supported for
6068 68020 and higher processors.
6070 @item -mno-strict-align
6071 @itemx -mstrict-align
6072 @opindex mno-strict-align
6073 @opindex mstrict-align
6074 Do not (do) assume that unaligned memory references will be handled by
6078 Generate code that allows the data segment to be located in a different
6079 area of memory from the text segment. This allows for execute in place in
6080 an environment without virtual memory management. This option implies -fPIC.
6083 Generate code that assumes that the data segment follows the text segment.
6084 This is the default.
6086 @item -mid-shared-library
6087 Generate code that supports shared libraries via the library ID method.
6088 This allows for execute in place and shared libraries in an environment
6089 without virtual memory management. This option implies -fPIC.
6091 @item -mno-id-shared-library
6092 Generate code that doesn't assume ID based shared libraries are being used.
6093 This is the default.
6095 @item -mshared-library-id=n
6096 Specified the identification number of the ID based shared library being
6097 compiled. Specifying a value of 0 will generate more compact code, specifying
6098 other values will force the allocation of that number to the current
6099 library but is no more space or time efficient than omitting this option.
6103 @node M68hc1x Options
6104 @subsection M68hc1x Options
6105 @cindex M68hc1x options
6107 These are the @samp{-m} options defined for the 68hc11 and 68hc12
6108 microcontrollers. The default values for these options depends on
6109 which style of microcontroller was selected when the compiler was configured;
6110 the defaults for the most common choices are given below.
6117 Generate output for a 68HC11. This is the default
6118 when the compiler is configured for 68HC11-based systems.
6124 Generate output for a 68HC12. This is the default
6125 when the compiler is configured for 68HC12-based systems.
6131 Generate output for a 68HCS12.
6134 @opindex mauto-incdec
6135 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
6142 Enable the use of 68HC12 min and max instructions.
6145 @itemx -mno-long-calls
6146 @opindex mlong-calls
6147 @opindex mno-long-calls
6148 Treat all calls as being far away (near). If calls are assumed to be
6149 far away, the compiler will use the @code{call} instruction to
6150 call a function and the @code{rtc} instruction for returning.
6154 Consider type @code{int} to be 16 bits wide, like @code{short int}.
6156 @item -msoft-reg-count=@var{count}
6157 @opindex msoft-reg-count
6158 Specify the number of pseudo-soft registers which are used for the
6159 code generation. The maximum number is 32. Using more pseudo-soft
6160 register may or may not result in better code depending on the program.
6161 The default is 4 for 68HC11 and 2 for 68HC12.
6166 @subsection VAX Options
6169 These @samp{-m} options are defined for the VAX:
6174 Do not output certain jump instructions (@code{aobleq} and so on)
6175 that the Unix assembler for the VAX cannot handle across long
6180 Do output those jump instructions, on the assumption that you
6181 will assemble with the GNU assembler.
6185 Output code for g-format floating point numbers instead of d-format.
6189 @subsection SPARC Options
6190 @cindex SPARC options
6192 These @samp{-m} options are supported on the SPARC:
6197 @opindex mno-app-regs
6199 Specify @option{-mapp-regs} to generate output using the global registers
6200 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
6203 To be fully SVR4 ABI compliant at the cost of some performance loss,
6204 specify @option{-mno-app-regs}. You should compile libraries and system
6205 software with this option.
6210 @opindex mhard-float
6211 Generate output containing floating point instructions. This is the
6217 @opindex msoft-float
6218 Generate output containing library calls for floating point.
6219 @strong{Warning:} the requisite libraries are not available for all SPARC
6220 targets. Normally the facilities of the machine's usual C compiler are
6221 used, but this cannot be done directly in cross-compilation. You must make
6222 your own arrangements to provide suitable library functions for
6223 cross-compilation. The embedded targets @samp{sparc-*-aout} and
6224 @samp{sparclite-*-*} do provide software floating point support.
6226 @option{-msoft-float} changes the calling convention in the output file;
6227 therefore, it is only useful if you compile @emph{all} of a program with
6228 this option. In particular, you need to compile @file{libgcc.a}, the
6229 library that comes with GCC, with @option{-msoft-float} in order for
6232 @item -mhard-quad-float
6233 @opindex mhard-quad-float
6234 Generate output containing quad-word (long double) floating point
6237 @item -msoft-quad-float
6238 @opindex msoft-quad-float
6239 Generate output containing library calls for quad-word (long double)
6240 floating point instructions. The functions called are those specified
6241 in the SPARC ABI@. This is the default.
6243 As of this writing, there are no SPARC implementations that have hardware
6244 support for the quad-word floating point instructions. They all invoke
6245 a trap handler for one of these instructions, and then the trap handler
6246 emulates the effect of the instruction. Because of the trap handler overhead,
6247 this is much slower than calling the ABI library routines. Thus the
6248 @option{-msoft-quad-float} option is the default.
6250 @item -mno-unaligned-doubles
6251 @itemx -munaligned-doubles
6252 @opindex mno-unaligned-doubles
6253 @opindex munaligned-doubles
6254 Assume that doubles have 8 byte alignment. This is the default.
6256 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
6257 alignment only if they are contained in another type, or if they have an
6258 absolute address. Otherwise, it assumes they have 4 byte alignment.
6259 Specifying this option avoids some rare compatibility problems with code
6260 generated by other compilers. It is not the default because it results
6261 in a performance loss, especially for floating point code.
6263 @item -mno-faster-structs
6264 @itemx -mfaster-structs
6265 @opindex mno-faster-structs
6266 @opindex mfaster-structs
6267 With @option{-mfaster-structs}, the compiler assumes that structures
6268 should have 8 byte alignment. This enables the use of pairs of
6269 @code{ldd} and @code{std} instructions for copies in structure
6270 assignment, in place of twice as many @code{ld} and @code{st} pairs.
6271 However, the use of this changed alignment directly violates the SPARC
6272 ABI@. Thus, it's intended only for use on targets where the developer
6273 acknowledges that their resulting code will not be directly in line with
6274 the rules of the ABI@.
6277 @opindex mimpure-text
6278 @option{-mimpure-text}, used in addition to @option{-shared}, tells
6279 the compiler to not pass @option{-z text} to the linker when linking a
6280 shared object. Using this option, you can link position-dependent
6281 code into a shared object.
6283 @option{-mimpure-text} suppresses the ``relocations remain against
6284 allocatable but non-writable sections'' linker error message.
6285 However, the necessary relocations will trigger copy-on-write, and the
6286 shared object is not actually shared across processes. Instead of
6287 using @option{-mimpure-text}, you should compile all source code with
6288 @option{-fpic} or @option{-fPIC}.
6290 This option is only available on SunOS and Solaris.
6292 @item -mcpu=@var{cpu_type}
6294 Set the instruction set, register set, and instruction scheduling parameters
6295 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
6296 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
6297 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
6298 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
6301 Default instruction scheduling parameters are used for values that select
6302 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
6303 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
6305 Here is a list of each supported architecture and their supported
6310 v8: supersparc, hypersparc
6311 sparclite: f930, f934, sparclite86x
6313 v9: ultrasparc, ultrasparc3
6316 By default (unless configured otherwise), GCC generates code for the V7
6317 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
6318 additionally optimizes it for the Cypress CY7C602 chip, as used in the
6319 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
6320 SPARCStation 1, 2, IPX etc.
6322 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
6323 architecture. The only difference from V7 code is that the compiler emits
6324 the integer multiply and integer divide instructions which exist in SPARC-V8
6325 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
6326 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
6329 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
6330 the SPARC architecture. This adds the integer multiply, integer divide step
6331 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
6332 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
6333 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU. With
6334 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
6335 MB86934 chip, which is the more recent SPARClite with FPU.
6337 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
6338 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
6339 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
6340 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
6341 optimizes it for the TEMIC SPARClet chip.
6343 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
6344 architecture. This adds 64-bit integer and floating-point move instructions,
6345 3 additional floating-point condition code registers and conditional move
6346 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
6347 optimizes it for the Sun UltraSPARC I/II chips. With
6348 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
6349 Sun UltraSPARC III chip.
6351 @item -mtune=@var{cpu_type}
6353 Set the instruction scheduling parameters for machine type
6354 @var{cpu_type}, but do not set the instruction set or register set that the
6355 option @option{-mcpu=@var{cpu_type}} would.
6357 The same values for @option{-mcpu=@var{cpu_type}} can be used for
6358 @option{-mtune=@var{cpu_type}}, but the only useful values are those
6359 that select a particular cpu implementation. Those are @samp{cypress},
6360 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
6361 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
6367 @opindex -mno-v8plus
6368 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI. The
6369 difference from the V8 ABI is that the global and out registers are
6370 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
6371 mode for all SPARC-V9 processors.
6377 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
6378 Visual Instruction Set extensions. The default is @option{-mno-vis}.
6381 These @samp{-m} options are supported in addition to the above
6382 on SPARC-V9 processors in 64-bit environments:
6385 @item -mlittle-endian
6386 @opindex mlittle-endian
6387 Generate code for a processor running in little-endian mode. It is only
6388 available for a few configurations and most notably not on Solaris.
6394 Generate code for a 32-bit or 64-bit environment.
6395 The 32-bit environment sets int, long and pointer to 32 bits.
6396 The 64-bit environment sets int to 32 bits and long and pointer
6399 @item -mcmodel=medlow
6400 @opindex mcmodel=medlow
6401 Generate code for the Medium/Low code model: 64-bit addresses, programs
6402 must be linked in the low 32 bits of memory. Programs can be statically
6403 or dynamically linked.
6405 @item -mcmodel=medmid
6406 @opindex mcmodel=medmid
6407 Generate code for the Medium/Middle code model: 64-bit addresses, programs
6408 must be linked in the low 44 bits of memory, the text and data segments must
6409 be less than 2GB in size and the data segment must be located within 2GB of
6412 @item -mcmodel=medany
6413 @opindex mcmodel=medany
6414 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
6415 may be linked anywhere in memory, the text and data segments must be less
6416 than 2GB in size and the data segment must be located within 2GB of the
6419 @item -mcmodel=embmedany
6420 @opindex mcmodel=embmedany
6421 Generate code for the Medium/Anywhere code model for embedded systems:
6422 64-bit addresses, the text and data segments must be less than 2GB in
6423 size, both starting anywhere in memory (determined at link time). The
6424 global register %g4 points to the base of the data segment. Programs
6425 are statically linked and PIC is not supported.
6428 @itemx -mno-stack-bias
6429 @opindex mstack-bias
6430 @opindex mno-stack-bias
6431 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
6432 frame pointer if present, are offset by @minus{}2047 which must be added back
6433 when making stack frame references. This is the default in 64-bit mode.
6434 Otherwise, assume no such offset is present.
6438 @subsection ARM Options
6441 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6445 @item -mabi=@var{name}
6447 Generate code for the specified ABI. Permissible values are: @samp{apcs-gnu},
6448 @samp{atpcs}, @samp{aapcs} and @samp{iwmmxt}.
6451 @opindex mapcs-frame
6452 Generate a stack frame that is compliant with the ARM Procedure Call
6453 Standard for all functions, even if this is not strictly necessary for
6454 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6455 with this option will cause the stack frames not to be generated for
6456 leaf functions. The default is @option{-mno-apcs-frame}.
6460 This is a synonym for @option{-mapcs-frame}.
6464 Generate code for a processor running with a 26-bit program counter,
6465 and conforming to the function calling standards for the APCS 26-bit
6466 option. This option replaces the @option{-m2} and @option{-m3} options
6467 of previous releases of the compiler.
6471 Generate code for a processor running with a 32-bit program counter,
6472 and conforming to the function calling standards for the APCS 32-bit
6473 option. This option replaces the @option{-m6} option of previous releases
6477 @c not currently implemented
6478 @item -mapcs-stack-check
6479 @opindex mapcs-stack-check
6480 Generate code to check the amount of stack space available upon entry to
6481 every function (that actually uses some stack space). If there is
6482 insufficient space available then either the function
6483 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6484 called, depending upon the amount of stack space required. The run time
6485 system is required to provide these functions. The default is
6486 @option{-mno-apcs-stack-check}, since this produces smaller code.
6488 @c not currently implemented
6490 @opindex mapcs-float
6491 Pass floating point arguments using the float point registers. This is
6492 one of the variants of the APCS@. This option is recommended if the
6493 target hardware has a floating point unit or if a lot of floating point
6494 arithmetic is going to be performed by the code. The default is
6495 @option{-mno-apcs-float}, since integer only code is slightly increased in
6496 size if @option{-mapcs-float} is used.
6498 @c not currently implemented
6499 @item -mapcs-reentrant
6500 @opindex mapcs-reentrant
6501 Generate reentrant, position independent code. The default is
6502 @option{-mno-apcs-reentrant}.
6505 @item -mthumb-interwork
6506 @opindex mthumb-interwork
6507 Generate code which supports calling between the ARM and Thumb
6508 instruction sets. Without this option the two instruction sets cannot
6509 be reliably used inside one program. The default is
6510 @option{-mno-thumb-interwork}, since slightly larger code is generated
6511 when @option{-mthumb-interwork} is specified.
6513 @item -mno-sched-prolog
6514 @opindex mno-sched-prolog
6515 Prevent the reordering of instructions in the function prolog, or the
6516 merging of those instruction with the instructions in the function's
6517 body. This means that all functions will start with a recognizable set
6518 of instructions (or in fact one of a choice from a small set of
6519 different function prologues), and this information can be used to
6520 locate the start if functions inside an executable piece of code. The
6521 default is @option{-msched-prolog}.
6524 @opindex mhard-float
6525 Generate output containing floating point instructions. This is the
6529 @opindex msoft-float
6530 Generate output containing library calls for floating point.
6531 @strong{Warning:} the requisite libraries are not available for all ARM
6532 targets. Normally the facilities of the machine's usual C compiler are
6533 used, but this cannot be done directly in cross-compilation. You must make
6534 your own arrangements to provide suitable library functions for
6537 @option{-msoft-float} changes the calling convention in the output file;
6538 therefore, it is only useful if you compile @emph{all} of a program with
6539 this option. In particular, you need to compile @file{libgcc.a}, the
6540 library that comes with GCC, with @option{-msoft-float} in order for
6543 @item -mfloat-abi=@var{name}
6545 Specifies which ABI to use for floating point values. Permissible values
6546 are: @samp{soft}, @samp{softfp} and @samp{hard}.
6548 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
6549 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
6550 of floating point instructions, but still uses the soft-float calling
6553 @item -mlittle-endian
6554 @opindex mlittle-endian
6555 Generate code for a processor running in little-endian mode. This is
6556 the default for all standard configurations.
6559 @opindex mbig-endian
6560 Generate code for a processor running in big-endian mode; the default is
6561 to compile code for a little-endian processor.
6563 @item -mwords-little-endian
6564 @opindex mwords-little-endian
6565 This option only applies when generating code for big-endian processors.
6566 Generate code for a little-endian word order but a big-endian byte
6567 order. That is, a byte order of the form @samp{32107654}. Note: this
6568 option should only be used if you require compatibility with code for
6569 big-endian ARM processors generated by versions of the compiler prior to
6572 @item -malignment-traps
6573 @opindex malignment-traps
6574 Generate code that will not trap if the MMU has alignment traps enabled.
6575 On ARM architectures prior to ARMv4, there were no instructions to
6576 access half-word objects stored in memory. However, when reading from
6577 memory a feature of the ARM architecture allows a word load to be used,
6578 even if the address is unaligned, and the processor core will rotate the
6579 data as it is being loaded. This option tells the compiler that such
6580 misaligned accesses will cause a MMU trap and that it should instead
6581 synthesize the access as a series of byte accesses. The compiler can
6582 still use word accesses to load half-word data if it knows that the
6583 address is aligned to a word boundary.
6585 This option is ignored when compiling for ARM architecture 4 or later,
6586 since these processors have instructions to directly access half-word
6589 @item -mno-alignment-traps
6590 @opindex mno-alignment-traps
6591 Generate code that assumes that the MMU will not trap unaligned
6592 accesses. This produces better code when the target instruction set
6593 does not have half-word memory operations (i.e.@: implementations prior to
6596 Note that you cannot use this option to access unaligned word objects,
6597 since the processor will only fetch one 32-bit aligned object from
6600 The default setting for most targets is @option{-mno-alignment-traps}, since
6601 this produces better code when there are no half-word memory
6602 instructions available.
6604 @item -mshort-load-bytes
6605 @itemx -mno-short-load-words
6606 @opindex mshort-load-bytes
6607 @opindex mno-short-load-words
6608 These are deprecated aliases for @option{-malignment-traps}.
6610 @item -mno-short-load-bytes
6611 @itemx -mshort-load-words
6612 @opindex mno-short-load-bytes
6613 @opindex mshort-load-words
6614 This are deprecated aliases for @option{-mno-alignment-traps}.
6616 @item -mcpu=@var{name}
6618 This specifies the name of the target ARM processor. GCC uses this name
6619 to determine what kind of instructions it can emit when generating
6620 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6621 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6622 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6623 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6624 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6625 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6626 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6627 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6628 @samp{arm920t}, @samp{arm926ejs}, @samp{arm940t}, @samp{arm9tdmi},
6629 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ejs},
6630 @samp{arm1136js}, @samp{arm1136jfs} ,@samp{xscale}, @samp{iwmmxt},
6633 @itemx -mtune=@var{name}
6635 This option is very similar to the @option{-mcpu=} option, except that
6636 instead of specifying the actual target processor type, and hence
6637 restricting which instructions can be used, it specifies that GCC should
6638 tune the performance of the code as if the target were of the type
6639 specified in this option, but still choosing the instructions that it
6640 will generate based on the cpu specified by a @option{-mcpu=} option.
6641 For some ARM implementations better performance can be obtained by using
6644 @item -march=@var{name}
6646 This specifies the name of the target ARM architecture. GCC uses this
6647 name to determine what kind of instructions it can emit when generating
6648 assembly code. This option can be used in conjunction with or instead
6649 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6650 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6651 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
6652 @samp{iwmmxt}, @samp{ep9312}.
6654 @item -mfpu=@var{name}
6655 @itemx -mfpe=@var{number}
6656 @itemx -mfp=@var{number}
6660 This specifies what floating point hardware (or hardware emulation) is
6661 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
6662 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
6663 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
6664 with older versions of GCC@.
6666 If @option{-msoft-float} is specified this specifies the format of
6667 floating point values.
6669 @item -mstructure-size-boundary=@var{n}
6670 @opindex mstructure-size-boundary
6671 The size of all structures and unions will be rounded up to a multiple
6672 of the number of bits set by this option. Permissible values are 8, 32
6673 and 64. The default value varies for different toolchains. For the COFF
6674 targeted toolchain the default value is 8. A value of 64 is only allowed
6675 if the underlying ABI supports it.
6677 Specifying the larger number can produce faster, more efficient code, but
6678 can also increase the size of the program. Different values are potentially
6679 incompatible. Code compiled with one value cannot necessarily expect to
6680 work with code or libraries compiled with another value, if they exchange
6681 information using structures or unions.
6683 @item -mabort-on-noreturn
6684 @opindex mabort-on-noreturn
6685 Generate a call to the function @code{abort} at the end of a
6686 @code{noreturn} function. It will be executed if the function tries to
6690 @itemx -mno-long-calls
6691 @opindex mlong-calls
6692 @opindex mno-long-calls
6693 Tells the compiler to perform function calls by first loading the
6694 address of the function into a register and then performing a subroutine
6695 call on this register. This switch is needed if the target function
6696 will lie outside of the 64 megabyte addressing range of the offset based
6697 version of subroutine call instruction.
6699 Even if this switch is enabled, not all function calls will be turned
6700 into long calls. The heuristic is that static functions, functions
6701 which have the @samp{short-call} attribute, functions that are inside
6702 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6703 definitions have already been compiled within the current compilation
6704 unit, will not be turned into long calls. The exception to this rule is
6705 that weak function definitions, functions with the @samp{long-call}
6706 attribute or the @samp{section} attribute, and functions that are within
6707 the scope of a @samp{#pragma long_calls} directive, will always be
6708 turned into long calls.
6710 This feature is not enabled by default. Specifying
6711 @option{-mno-long-calls} will restore the default behavior, as will
6712 placing the function calls within the scope of a @samp{#pragma
6713 long_calls_off} directive. Note these switches have no effect on how
6714 the compiler generates code to handle function calls via function
6717 @item -mnop-fun-dllimport
6718 @opindex mnop-fun-dllimport
6719 Disable support for the @code{dllimport} attribute.
6721 @item -msingle-pic-base
6722 @opindex msingle-pic-base
6723 Treat the register used for PIC addressing as read-only, rather than
6724 loading it in the prologue for each function. The run-time system is
6725 responsible for initializing this register with an appropriate value
6726 before execution begins.
6728 @item -mpic-register=@var{reg}
6729 @opindex mpic-register
6730 Specify the register to be used for PIC addressing. The default is R10
6731 unless stack-checking is enabled, when R9 is used.
6733 @item -mcirrus-fix-invalid-insns
6734 @opindex mcirrus-fix-invalid-insns
6735 @opindex mno-cirrus-fix-invalid-insns
6736 Insert NOPs into the instruction stream to in order to work around
6737 problems with invalid Maverick instruction combinations. This option
6738 is only valid if the @option{-mcpu=ep9312} option has been used to
6739 enable generation of instructions for the Cirrus Maverick floating
6740 point co-processor. This option is not enabled by default, since the
6741 problem is only present in older Maverick implementations. The default
6742 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6745 @item -mpoke-function-name
6746 @opindex mpoke-function-name
6747 Write the name of each function into the text section, directly
6748 preceding the function prologue. The generated code is similar to this:
6752 .ascii "arm_poke_function_name", 0
6755 .word 0xff000000 + (t1 - t0)
6756 arm_poke_function_name
6758 stmfd sp!, @{fp, ip, lr, pc@}
6762 When performing a stack backtrace, code can inspect the value of
6763 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6764 location @code{pc - 12} and the top 8 bits are set, then we know that
6765 there is a function name embedded immediately preceding this location
6766 and has length @code{((pc[-3]) & 0xff000000)}.
6770 Generate code for the 16-bit Thumb instruction set. The default is to
6771 use the 32-bit ARM instruction set.
6774 @opindex mtpcs-frame
6775 Generate a stack frame that is compliant with the Thumb Procedure Call
6776 Standard for all non-leaf functions. (A leaf function is one that does
6777 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6779 @item -mtpcs-leaf-frame
6780 @opindex mtpcs-leaf-frame
6781 Generate a stack frame that is compliant with the Thumb Procedure Call
6782 Standard for all leaf functions. (A leaf function is one that does
6783 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6785 @item -mcallee-super-interworking
6786 @opindex mcallee-super-interworking
6787 Gives all externally visible functions in the file being compiled an ARM
6788 instruction set header which switches to Thumb mode before executing the
6789 rest of the function. This allows these functions to be called from
6790 non-interworking code.
6792 @item -mcaller-super-interworking
6793 @opindex mcaller-super-interworking
6794 Allows calls via function pointers (including virtual functions) to
6795 execute correctly regardless of whether the target code has been
6796 compiled for interworking or not. There is a small overhead in the cost
6797 of executing a function pointer if this option is enabled.
6801 @node MN10300 Options
6802 @subsection MN10300 Options
6803 @cindex MN10300 options
6805 These @option{-m} options are defined for Matsushita MN10300 architectures:
6810 Generate code to avoid bugs in the multiply instructions for the MN10300
6811 processors. This is the default.
6814 @opindex mno-mult-bug
6815 Do not generate code to avoid bugs in the multiply instructions for the
6820 Generate code which uses features specific to the AM33 processor.
6824 Do not generate code which uses features specific to the AM33 processor. This
6829 Do not link in the C run-time initialization object file.
6833 Indicate to the linker that it should perform a relaxation optimization pass
6834 to shorten branches, calls and absolute memory addresses. This option only
6835 has an effect when used on the command line for the final link step.
6837 This option makes symbolic debugging impossible.
6841 @node M32R/D Options
6842 @subsection M32R/D Options
6843 @cindex M32R/D options
6845 These @option{-m} options are defined for Renesas M32R/D architectures:
6850 Generate code for the M32R/2@.
6854 Generate code for the M32R/X@.
6858 Generate code for the M32R@. This is the default.
6861 @opindex mmodel=small
6862 Assume all objects live in the lower 16MB of memory (so that their addresses
6863 can be loaded with the @code{ld24} instruction), and assume all subroutines
6864 are reachable with the @code{bl} instruction.
6865 This is the default.
6867 The addressability of a particular object can be set with the
6868 @code{model} attribute.
6870 @item -mmodel=medium
6871 @opindex mmodel=medium
6872 Assume objects may be anywhere in the 32-bit address space (the compiler
6873 will generate @code{seth/add3} instructions to load their addresses), and
6874 assume all subroutines are reachable with the @code{bl} instruction.
6877 @opindex mmodel=large
6878 Assume objects may be anywhere in the 32-bit address space (the compiler
6879 will generate @code{seth/add3} instructions to load their addresses), and
6880 assume subroutines may not be reachable with the @code{bl} instruction
6881 (the compiler will generate the much slower @code{seth/add3/jl}
6882 instruction sequence).
6885 @opindex msdata=none
6886 Disable use of the small data area. Variables will be put into
6887 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6888 @code{section} attribute has been specified).
6889 This is the default.
6891 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6892 Objects may be explicitly put in the small data area with the
6893 @code{section} attribute using one of these sections.
6896 @opindex msdata=sdata
6897 Put small global and static data in the small data area, but do not
6898 generate special code to reference them.
6902 Put small global and static data in the small data area, and generate
6903 special instructions to reference them.
6907 @cindex smaller data references
6908 Put global and static objects less than or equal to @var{num} bytes
6909 into the small data or bss sections instead of the normal data or bss
6910 sections. The default value of @var{num} is 8.
6911 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6912 for this option to have any effect.
6914 All modules should be compiled with the same @option{-G @var{num}} value.
6915 Compiling with different values of @var{num} may or may not work; if it
6916 doesn't the linker will give an error message---incorrect code will not be
6921 Makes the M32R specific code in the compiler display some statistics
6922 that might help in debugging programs.
6925 @opindex malign-loops
6926 Align all loops to a 32-byte boundary.
6928 @item -mno-align-loops
6929 @opindex mno-align-loops
6930 Do not enforce a 32-byte alignment for loops. This is the default.
6932 @item -missue-rate=@var{number}
6933 @opindex missue-rate=@var{number}
6934 Issue @var{number} instructions per cycle. @var{number} can only be 1
6937 @item -mbranch-cost=@var{number}
6938 @opindex mbranch-cost=@var{number}
6939 @var{number} can only be 1 or 2. If it is 1 then branches will be
6940 preferred over conditional code, if it is 2, then the opposite will
6943 @item -mflush-trap=@var{number}
6944 @opindex mflush-trap=@var{number}
6945 Specifies the trap number to use to flush the cache. The default is
6946 12. Valid numbers are between 0 and 15 inclusive.
6948 @item -mno-flush-trap
6949 @opindex mno-flush-trap
6950 Specifies that the cache cannot be flushed by using a trap.
6952 @item -mflush-func=@var{name}
6953 @opindex mflush-func=@var{name}
6954 Specifies the name of the operating system function to call to flush
6955 the cache. The default is @emph{_flush_cache}, but a function call
6956 will only be used if a trap is not available.
6958 @item -mno-flush-func
6959 @opindex mno-flush-func
6960 Indicates that there is no OS function for flushing the cache.
6964 @node RS/6000 and PowerPC Options
6965 @subsection IBM RS/6000 and PowerPC Options
6966 @cindex RS/6000 and PowerPC Options
6967 @cindex IBM RS/6000 and PowerPC Options
6969 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6977 @itemx -mpowerpc-gpopt
6978 @itemx -mno-powerpc-gpopt
6979 @itemx -mpowerpc-gfxopt
6980 @itemx -mno-powerpc-gfxopt
6982 @itemx -mno-powerpc64
6988 @opindex mno-powerpc
6989 @opindex mpowerpc-gpopt
6990 @opindex mno-powerpc-gpopt
6991 @opindex mpowerpc-gfxopt
6992 @opindex mno-powerpc-gfxopt
6994 @opindex mno-powerpc64
6995 GCC supports two related instruction set architectures for the
6996 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6997 instructions supported by the @samp{rios} chip set used in the original
6998 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6999 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
7000 the IBM 4xx microprocessors.
7002 Neither architecture is a subset of the other. However there is a
7003 large common subset of instructions supported by both. An MQ
7004 register is included in processors supporting the POWER architecture.
7006 You use these options to specify which instructions are available on the
7007 processor you are using. The default value of these options is
7008 determined when configuring GCC@. Specifying the
7009 @option{-mcpu=@var{cpu_type}} overrides the specification of these
7010 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
7011 rather than the options listed above.
7013 The @option{-mpower} option allows GCC to generate instructions that
7014 are found only in the POWER architecture and to use the MQ register.
7015 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
7016 to generate instructions that are present in the POWER2 architecture but
7017 not the original POWER architecture.
7019 The @option{-mpowerpc} option allows GCC to generate instructions that
7020 are found only in the 32-bit subset of the PowerPC architecture.
7021 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
7022 GCC to use the optional PowerPC architecture instructions in the
7023 General Purpose group, including floating-point square root. Specifying
7024 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
7025 use the optional PowerPC architecture instructions in the Graphics
7026 group, including floating-point select.
7028 The @option{-mpowerpc64} option allows GCC to generate the additional
7029 64-bit instructions that are found in the full PowerPC64 architecture
7030 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
7031 @option{-mno-powerpc64}.
7033 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
7034 will use only the instructions in the common subset of both
7035 architectures plus some special AIX common-mode calls, and will not use
7036 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
7037 permits GCC to use any instruction from either architecture and to
7038 allow use of the MQ register; specify this for the Motorola MPC601.
7040 @item -mnew-mnemonics
7041 @itemx -mold-mnemonics
7042 @opindex mnew-mnemonics
7043 @opindex mold-mnemonics
7044 Select which mnemonics to use in the generated assembler code. With
7045 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
7046 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
7047 assembler mnemonics defined for the POWER architecture. Instructions
7048 defined in only one architecture have only one mnemonic; GCC uses that
7049 mnemonic irrespective of which of these options is specified.
7051 GCC defaults to the mnemonics appropriate for the architecture in
7052 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
7053 value of these option. Unless you are building a cross-compiler, you
7054 should normally not specify either @option{-mnew-mnemonics} or
7055 @option{-mold-mnemonics}, but should instead accept the default.
7057 @item -mcpu=@var{cpu_type}
7059 Set architecture type, register usage, choice of mnemonics, and
7060 instruction scheduling parameters for machine type @var{cpu_type}.
7061 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
7062 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
7063 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
7064 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
7065 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
7066 @samp{860}, @samp{970}, @samp{common}, @samp{ec603e}, @samp{G3},
7067 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
7068 @samp{power4}, @samp{power5}, @samp{powerpc}, @samp{powerpc64},
7069 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64a}.
7071 @option{-mcpu=common} selects a completely generic processor. Code
7072 generated under this option will run on any POWER or PowerPC processor.
7073 GCC will use only the instructions in the common subset of both
7074 architectures, and will not use the MQ register. GCC assumes a generic
7075 processor model for scheduling purposes.
7077 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
7078 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
7079 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
7080 types, with an appropriate, generic processor model assumed for
7081 scheduling purposes.
7083 The other options specify a specific processor. Code generated under
7084 those options will run best on that processor, and may not run at all on
7087 The @option{-mcpu} options automatically enable or disable the
7088 following options: @option{-maltivec}, @option{-mhard-float},
7089 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
7090 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
7091 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
7092 @option{-mstring}. The particular options set for any particular CPU
7093 will vary between compiler versions, depending on what setting seems
7094 to produce optimal code for that CPU; it doesn't necessarily reflect
7095 the actual hardware's capabilities. If you wish to set an individual
7096 option to a particular value, you may specify it after the
7097 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
7099 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
7100 not enabled or disabled by the @option{-mcpu} option at present, since
7101 AIX does not have full support for these options. You may still
7102 enable or disable them individually if you're sure it'll work in your
7105 @item -mtune=@var{cpu_type}
7107 Set the instruction scheduling parameters for machine type
7108 @var{cpu_type}, but do not set the architecture type, register usage, or
7109 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
7110 values for @var{cpu_type} are used for @option{-mtune} as for
7111 @option{-mcpu}. If both are specified, the code generated will use the
7112 architecture, registers, and mnemonics set by @option{-mcpu}, but the
7113 scheduling parameters set by @option{-mtune}.
7118 @opindex mno-altivec
7119 These switches enable or disable the use of built-in functions that
7120 allow access to the AltiVec instruction set. You may also need to set
7121 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
7126 Extend the current ABI with SPE ABI extensions. This does not change
7127 the default ABI, instead it adds the SPE ABI extensions to the current
7131 @opindex mabi=no-spe
7132 Disable Booke SPE ABI extensions for the current ABI.
7134 @item -misel=@var{yes/no}
7137 This switch enables or disables the generation of ISEL instructions.
7139 @item -mspe=@var{yes/no}
7142 This switch enables or disables the generation of SPE simd
7145 @item -mfloat-gprs=@var{yes/no}
7147 @opindex mfloat-gprs
7148 This switch enables or disables the generation of floating point
7149 operations on the general purpose registers for architectures that
7150 support it. This option is currently only available on the MPC8540.
7153 @itemx -mno-fp-in-toc
7154 @itemx -mno-sum-in-toc
7155 @itemx -mminimal-toc
7157 @opindex mno-fp-in-toc
7158 @opindex mno-sum-in-toc
7159 @opindex mminimal-toc
7160 Modify generation of the TOC (Table Of Contents), which is created for
7161 every executable file. The @option{-mfull-toc} option is selected by
7162 default. In that case, GCC will allocate at least one TOC entry for
7163 each unique non-automatic variable reference in your program. GCC
7164 will also place floating-point constants in the TOC@. However, only
7165 16,384 entries are available in the TOC@.
7167 If you receive a linker error message that saying you have overflowed
7168 the available TOC space, you can reduce the amount of TOC space used
7169 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
7170 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
7171 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
7172 generate code to calculate the sum of an address and a constant at
7173 run-time instead of putting that sum into the TOC@. You may specify one
7174 or both of these options. Each causes GCC to produce very slightly
7175 slower and larger code at the expense of conserving TOC space.
7177 If you still run out of space in the TOC even when you specify both of
7178 these options, specify @option{-mminimal-toc} instead. This option causes
7179 GCC to make only one TOC entry for every file. When you specify this
7180 option, GCC will produce code that is slower and larger but which
7181 uses extremely little TOC space. You may wish to use this option
7182 only on files that contain less frequently executed code.
7188 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
7189 @code{long} type, and the infrastructure needed to support them.
7190 Specifying @option{-maix64} implies @option{-mpowerpc64} and
7191 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
7192 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
7197 @opindex mno-xl-call
7198 On AIX, pass floating-point arguments to prototyped functions beyond the
7199 register save area (RSA) on the stack in addition to argument FPRs. The
7200 AIX calling convention was extended but not initially documented to
7201 handle an obscure K&R C case of calling a function that takes the
7202 address of its arguments with fewer arguments than declared. AIX XL
7203 compilers access floating point arguments which do not fit in the
7204 RSA from the stack when a subroutine is compiled without
7205 optimization. Because always storing floating-point arguments on the
7206 stack is inefficient and rarely needed, this option is not enabled by
7207 default and only is necessary when calling subroutines compiled by AIX
7208 XL compilers without optimization.
7212 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
7213 application written to use message passing with special startup code to
7214 enable the application to run. The system must have PE installed in the
7215 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
7216 must be overridden with the @option{-specs=} option to specify the
7217 appropriate directory location. The Parallel Environment does not
7218 support threads, so the @option{-mpe} option and the @option{-pthread}
7219 option are incompatible.
7221 @item -malign-natural
7222 @itemx -malign-power
7223 @opindex malign-natural
7224 @opindex malign-power
7225 On AIX, Darwin, and 64-bit PowerPC GNU/Linux, the option
7226 @option{-malign-natural} overrides the ABI-defined alignment of larger
7227 types, such as floating-point doubles, on their natural size-based boundary.
7228 The option @option{-malign-power} instructs GCC to follow the ABI-specified
7229 alignment rules. GCC defaults to the standard alignment defined in the ABI.
7233 @opindex msoft-float
7234 @opindex mhard-float
7235 Generate code that does not use (uses) the floating-point register set.
7236 Software floating point emulation is provided if you use the
7237 @option{-msoft-float} option, and pass the option to GCC when linking.
7240 @itemx -mno-multiple
7242 @opindex mno-multiple
7243 Generate code that uses (does not use) the load multiple word
7244 instructions and the store multiple word instructions. These
7245 instructions are generated by default on POWER systems, and not
7246 generated on PowerPC systems. Do not use @option{-mmultiple} on little
7247 endian PowerPC systems, since those instructions do not work when the
7248 processor is in little endian mode. The exceptions are PPC740 and
7249 PPC750 which permit the instructions usage in little endian mode.
7255 Generate code that uses (does not use) the load string instructions
7256 and the store string word instructions to save multiple registers and
7257 do small block moves. These instructions are generated by default on
7258 POWER systems, and not generated on PowerPC systems. Do not use
7259 @option{-mstring} on little endian PowerPC systems, since those
7260 instructions do not work when the processor is in little endian mode.
7261 The exceptions are PPC740 and PPC750 which permit the instructions
7262 usage in little endian mode.
7268 Generate code that uses (does not use) the load or store instructions
7269 that update the base register to the address of the calculated memory
7270 location. These instructions are generated by default. If you use
7271 @option{-mno-update}, there is a small window between the time that the
7272 stack pointer is updated and the address of the previous frame is
7273 stored, which means code that walks the stack frame across interrupts or
7274 signals may get corrupted data.
7277 @itemx -mno-fused-madd
7278 @opindex mfused-madd
7279 @opindex mno-fused-madd
7280 Generate code that uses (does not use) the floating point multiply and
7281 accumulate instructions. These instructions are generated by default if
7282 hardware floating is used.
7284 @item -mno-bit-align
7286 @opindex mno-bit-align
7288 On System V.4 and embedded PowerPC systems do not (do) force structures
7289 and unions that contain bit-fields to be aligned to the base type of the
7292 For example, by default a structure containing nothing but 8
7293 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7294 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7295 the structure would be aligned to a 1 byte boundary and be one byte in
7298 @item -mno-strict-align
7299 @itemx -mstrict-align
7300 @opindex mno-strict-align
7301 @opindex mstrict-align
7302 On System V.4 and embedded PowerPC systems do not (do) assume that
7303 unaligned memory references will be handled by the system.
7306 @itemx -mno-relocatable
7307 @opindex mrelocatable
7308 @opindex mno-relocatable
7309 On embedded PowerPC systems generate code that allows (does not allow)
7310 the program to be relocated to a different address at runtime. If you
7311 use @option{-mrelocatable} on any module, all objects linked together must
7312 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7314 @item -mrelocatable-lib
7315 @itemx -mno-relocatable-lib
7316 @opindex mrelocatable-lib
7317 @opindex mno-relocatable-lib
7318 On embedded PowerPC systems generate code that allows (does not allow)
7319 the program to be relocated to a different address at runtime. Modules
7320 compiled with @option{-mrelocatable-lib} can be linked with either modules
7321 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7322 with modules compiled with the @option{-mrelocatable} options.
7328 On System V.4 and embedded PowerPC systems do not (do) assume that
7329 register 2 contains a pointer to a global area pointing to the addresses
7330 used in the program.
7333 @itemx -mlittle-endian
7335 @opindex mlittle-endian
7336 On System V.4 and embedded PowerPC systems compile code for the
7337 processor in little endian mode. The @option{-mlittle-endian} option is
7338 the same as @option{-mlittle}.
7343 @opindex mbig-endian
7344 On System V.4 and embedded PowerPC systems compile code for the
7345 processor in big endian mode. The @option{-mbig-endian} option is
7346 the same as @option{-mbig}.
7348 @item -mdynamic-no-pic
7349 @opindex mdynamic-no-pic
7350 On Darwin and Mac OS X systems, compile code so that it is not
7351 relocatable, but that its external references are relocatable. The
7352 resulting code is suitable for applications, but not shared
7355 @item -mprioritize-restricted-insns=@var{priority}
7356 @opindex mprioritize-restricted-insns
7357 This option controls the priority that is assigned to
7358 dispatch-slot restricted instructions during the second scheduling
7359 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
7360 @var{no/highest/second-highest} priority to dispatch slot restricted
7363 @item -msched-costly-dep=@var{dependence_type}
7364 @opindex msched-costly-dep
7365 This option controls which dependences are considered costly
7366 by the target during instruction scheduling. The argument
7367 @var{dependence_type} takes one of the following values:
7368 @var{no}: no dependence is costly,
7369 @var{all}: all dependences are costly,
7370 @var{true_store_to_load}: a true dependence from store to load is costly,
7371 @var{store_to_load}: any dependence from store to load is costly,
7372 @var{number}: any dependence which latency >= @var{number} is costly.
7374 @item -minsert-sched-nops=@var{scheme}
7375 @opindex minsert-sched-nops
7376 This option controls which nop insertion scheme will be used during
7377 the second scheduling pass. The argument @var{scheme} takes one of the
7379 @var{no}: Don't insert nops.
7380 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
7381 according to the scheduler's grouping.
7382 @var{regroup_exact}: Insert nops to force costly dependent insns into
7383 separate groups. Insert exactly as many nops as needed to force an insn
7384 to a new group, according to the estimated processor grouping.
7385 @var{number}: Insert nops to force costly dependent insns into
7386 separate groups. Insert @var{number} nops to force an insn to a new group.
7390 On System V.4 and embedded PowerPC systems compile code using calling
7391 conventions that adheres to the March 1995 draft of the System V
7392 Application Binary Interface, PowerPC processor supplement. This is the
7393 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7395 @item -mcall-sysv-eabi
7396 @opindex mcall-sysv-eabi
7397 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7399 @item -mcall-sysv-noeabi
7400 @opindex mcall-sysv-noeabi
7401 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7403 @item -mcall-solaris
7404 @opindex mcall-solaris
7405 On System V.4 and embedded PowerPC systems compile code for the Solaris
7409 @opindex mcall-linux
7410 On System V.4 and embedded PowerPC systems compile code for the
7411 Linux-based GNU system.
7415 On System V.4 and embedded PowerPC systems compile code for the
7416 Hurd-based GNU system.
7419 @opindex mcall-netbsd
7420 On System V.4 and embedded PowerPC systems compile code for the
7421 NetBSD operating system.
7423 @item -maix-struct-return
7424 @opindex maix-struct-return
7425 Return all structures in memory (as specified by the AIX ABI)@.
7427 @item -msvr4-struct-return
7428 @opindex msvr4-struct-return
7429 Return structures smaller than 8 bytes in registers (as specified by the
7433 @opindex mabi=altivec
7434 Extend the current ABI with AltiVec ABI extensions. This does not
7435 change the default ABI, instead it adds the AltiVec ABI extensions to
7438 @item -mabi=no-altivec
7439 @opindex mabi=no-altivec
7440 Disable AltiVec ABI extensions for the current ABI.
7443 @itemx -mno-prototype
7445 @opindex mno-prototype
7446 On System V.4 and embedded PowerPC systems assume that all calls to
7447 variable argument functions are properly prototyped. Otherwise, the
7448 compiler must insert an instruction before every non prototyped call to
7449 set or clear bit 6 of the condition code register (@var{CR}) to
7450 indicate whether floating point values were passed in the floating point
7451 registers in case the function takes a variable arguments. With
7452 @option{-mprototype}, only calls to prototyped variable argument functions
7453 will set or clear the bit.
7457 On embedded PowerPC systems, assume that the startup module is called
7458 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7459 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7464 On embedded PowerPC systems, assume that the startup module is called
7465 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7470 On embedded PowerPC systems, assume that the startup module is called
7471 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7475 @opindex myellowknife
7476 On embedded PowerPC systems, assume that the startup module is called
7477 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7482 On System V.4 and embedded PowerPC systems, specify that you are
7483 compiling for a VxWorks system.
7487 Specify that you are compiling for the WindISS simulation environment.
7491 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7492 header to indicate that @samp{eabi} extended relocations are used.
7498 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7499 Embedded Applications Binary Interface (eabi) which is a set of
7500 modifications to the System V.4 specifications. Selecting @option{-meabi}
7501 means that the stack is aligned to an 8 byte boundary, a function
7502 @code{__eabi} is called to from @code{main} to set up the eabi
7503 environment, and the @option{-msdata} option can use both @code{r2} and
7504 @code{r13} to point to two separate small data areas. Selecting
7505 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7506 do not call an initialization function from @code{main}, and the
7507 @option{-msdata} option will only use @code{r13} to point to a single
7508 small data area. The @option{-meabi} option is on by default if you
7509 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7512 @opindex msdata=eabi
7513 On System V.4 and embedded PowerPC systems, put small initialized
7514 @code{const} global and static data in the @samp{.sdata2} section, which
7515 is pointed to by register @code{r2}. Put small initialized
7516 non-@code{const} global and static data in the @samp{.sdata} section,
7517 which is pointed to by register @code{r13}. Put small uninitialized
7518 global and static data in the @samp{.sbss} section, which is adjacent to
7519 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7520 incompatible with the @option{-mrelocatable} option. The
7521 @option{-msdata=eabi} option also sets the @option{-memb} option.
7524 @opindex msdata=sysv
7525 On System V.4 and embedded PowerPC systems, put small global and static
7526 data in the @samp{.sdata} section, which is pointed to by register
7527 @code{r13}. Put small uninitialized global and static data in the
7528 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7529 The @option{-msdata=sysv} option is incompatible with the
7530 @option{-mrelocatable} option.
7532 @item -msdata=default
7534 @opindex msdata=default
7536 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7537 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7538 same as @option{-msdata=sysv}.
7541 @opindex msdata-data
7542 On System V.4 and embedded PowerPC systems, put small global and static
7543 data in the @samp{.sdata} section. Put small uninitialized global and
7544 static data in the @samp{.sbss} section. Do not use register @code{r13}
7545 to address small data however. This is the default behavior unless
7546 other @option{-msdata} options are used.
7550 @opindex msdata=none
7552 On embedded PowerPC systems, put all initialized global and static data
7553 in the @samp{.data} section, and all uninitialized data in the
7554 @samp{.bss} section.
7558 @cindex smaller data references (PowerPC)
7559 @cindex .sdata/.sdata2 references (PowerPC)
7560 On embedded PowerPC systems, put global and static items less than or
7561 equal to @var{num} bytes into the small data or bss sections instead of
7562 the normal data or bss section. By default, @var{num} is 8. The
7563 @option{-G @var{num}} switch is also passed to the linker.
7564 All modules should be compiled with the same @option{-G @var{num}} value.
7567 @itemx -mno-regnames
7569 @opindex mno-regnames
7570 On System V.4 and embedded PowerPC systems do (do not) emit register
7571 names in the assembly language output using symbolic forms.
7574 @itemx -mno-longcall
7576 @opindex mno-longcall
7577 Default to making all function calls indirectly, using a register, so
7578 that functions which reside further than 32 megabytes (33,554,432
7579 bytes) from the current location can be called. This setting can be
7580 overridden by the @code{shortcall} function attribute, or by
7581 @code{#pragma longcall(0)}.
7583 Some linkers are capable of detecting out-of-range calls and generating
7584 glue code on the fly. On these systems, long calls are unnecessary and
7585 generate slower code. As of this writing, the AIX linker can do this,
7586 as can the GNU linker for PowerPC/64. It is planned to add this feature
7587 to the GNU linker for 32-bit PowerPC systems as well.
7589 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
7590 callee, L42'', plus a ``branch island'' (glue code). The two target
7591 addresses represent the callee and the ``branch island.'' The
7592 Darwin/PPC linker will prefer the first address and generate a ``bl
7593 callee'' if the PPC ``bl'' instruction will reach the callee directly;
7594 otherwise, the linker will generate ``bl L42'' to call the ``branch
7595 island.'' The ``branch island'' is appended to the body of the
7596 calling function; it computes the full 32-bit address of the callee
7599 On Mach-O (Darwin) systems, this option directs the compiler emit to
7600 the glue for every direct call, and the Darwin linker decides whether
7601 to use or discard it.
7603 In the future, we may cause GCC to ignore all longcall specifications
7604 when the linker is known to generate glue.
7608 Adds support for multithreading with the @dfn{pthreads} library.
7609 This option sets flags for both the preprocessor and linker.
7613 @node Darwin Options
7614 @subsection Darwin Options
7615 @cindex Darwin options
7617 These options are defined for all architectures running the Darwin operating
7618 system. They are useful for compatibility with other Mac OS compilers.
7623 Loads all members of static archive libraries.
7624 See man ld(1) for more information.
7626 @item -arch_errors_fatal
7627 @opindex arch_errors_fatal
7628 Cause the errors having to do with files that have the wrong architecture
7632 @opindex bind_at_load
7633 Causes the output file to be marked such that the dynamic linker will
7634 bind all undefined references when the file is loaded or launched.
7638 Produce a Mach-o bundle format file.
7639 See man ld(1) for more information.
7641 @item -bundle_loader @var{executable}
7642 @opindex bundle_loader
7643 This specifies the @var{executable} that will be loading the build
7644 output file being linked. See man ld(1) for more information.
7646 @item -allowable_client @var{client_name}
7650 @itemx -compatibility_version
7651 @itemx -current_version
7652 @itemx -dependency-file
7654 @itemx -dylinker_install_name
7657 @itemx -exported_symbols_list
7659 @itemx -flat_namespace
7660 @itemx -force_cpusubtype_ALL
7661 @itemx -force_flat_namespace
7662 @itemx -headerpad_max_install_names
7665 @itemx -install_name
7666 @itemx -keep_private_externs
7667 @itemx -multi_module
7668 @itemx -multiply_defined
7669 @itemx -multiply_defined_unused
7671 @itemx -nofixprebinding
7674 @itemx -noseglinkedit
7675 @itemx -pagezero_size
7677 @itemx -prebind_all_twolevel_modules
7678 @itemx -private_bundle
7679 @itemx -read_only_relocs
7681 @itemx -sectobjectsymbols
7685 @itemx -sectobjectsymbols
7687 @itemx -seg_addr_table
7688 @itemx -seg_addr_table_filename
7691 @itemx -segs_read_only_addr
7692 @itemx -segs_read_write_addr
7693 @itemx -single_module
7696 @itemx -sub_umbrella
7697 @itemx -twolevel_namespace
7700 @itemx -unexported_symbols_list
7701 @itemx -weak_reference_mismatches
7704 @opindex allowable_client
7706 @opindex client_name
7707 @opindex compatibility_version
7708 @opindex current_version
7709 @opindex dependency-file
7711 @opindex dylinker_install_name
7714 @opindex exported_symbols_list
7716 @opindex flat_namespace
7717 @opindex force_cpusubtype_ALL
7718 @opindex force_flat_namespace
7719 @opindex headerpad_max_install_names
7722 @opindex install_name
7723 @opindex keep_private_externs
7724 @opindex multi_module
7725 @opindex multiply_defined
7726 @opindex multiply_defined_unused
7728 @opindex nofixprebinding
7729 @opindex nomultidefs
7731 @opindex noseglinkedit
7732 @opindex pagezero_size
7734 @opindex prebind_all_twolevel_modules
7735 @opindex private_bundle
7736 @opindex read_only_relocs
7738 @opindex sectobjectsymbols
7742 @opindex sectobjectsymbols
7744 @opindex seg_addr_table
7745 @opindex seg_addr_table_filename
7746 @opindex seglinkedit
7748 @opindex segs_read_only_addr
7749 @opindex segs_read_write_addr
7750 @opindex single_module
7752 @opindex sub_library
7753 @opindex sub_umbrella
7754 @opindex twolevel_namespace
7757 @opindex unexported_symbols_list
7758 @opindex weak_reference_mismatches
7759 @opindex whatsloaded
7761 These options are available for Darwin linker. Darwin linker man page
7762 describes them in detail.
7767 @subsection MIPS Options
7768 @cindex MIPS options
7774 Generate big-endian code.
7778 Generate little-endian code. This is the default for @samp{mips*el-*-*}
7781 @item -march=@var{arch}
7783 Generate code that will run on @var{arch}, which can be the name of a
7784 generic MIPS ISA, or the name of a particular processor.
7786 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
7787 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
7788 The processor names are:
7789 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
7791 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
7792 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
7796 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
7797 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
7798 The special value @samp{from-abi} selects the
7799 most compatible architecture for the selected ABI (that is,
7800 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
7802 In processor names, a final @samp{000} can be abbreviated as @samp{k}
7803 (for example, @samp{-march=r2k}). Prefixes are optional, and
7804 @samp{vr} may be written @samp{r}.
7806 GCC defines two macros based on the value of this option. The first
7807 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
7808 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
7809 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
7810 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
7811 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
7813 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
7814 above. In other words, it will have the full prefix and will not
7815 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
7816 the macro names the resolved architecture (either @samp{"mips1"} or
7817 @samp{"mips3"}). It names the default architecture when no
7818 @option{-march} option is given.
7820 @item -mtune=@var{arch}
7822 Optimize for @var{arch}. Among other things, this option controls
7823 the way instructions are scheduled, and the perceived cost of arithmetic
7824 operations. The list of @var{arch} values is the same as for
7827 When this option is not used, GCC will optimize for the processor
7828 specified by @option{-march}. By using @option{-march} and
7829 @option{-mtune} together, it is possible to generate code that will
7830 run on a family of processors, but optimize the code for one
7831 particular member of that family.
7833 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
7834 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
7835 @samp{-march} ones described above.
7839 Equivalent to @samp{-march=mips1}.
7843 Equivalent to @samp{-march=mips2}.
7847 Equivalent to @samp{-march=mips3}.
7851 Equivalent to @samp{-march=mips4}.
7855 Equivalent to @samp{-march=mips32}.
7859 Equivalent to @samp{-march=mips32r2}.
7863 Equivalent to @samp{-march=mips64}.
7869 Use (do not use) the MIPS16 ISA.
7881 Generate code for the given ABI@.
7883 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
7884 generates 64-bit code when you select a 64-bit architecture, but you
7885 can use @option{-mgp32} to get 32-bit code instead.
7887 For information about the O64 ABI, see
7888 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
7891 @itemx -mno-abicalls
7893 @opindex mno-abicalls
7894 Generate (do not generate) SVR4-style position-independent code.
7895 @option{-mabicalls} is the default for SVR4-based systems.
7901 Lift (do not lift) the usual restrictions on the size of the global
7904 GCC normally uses a single instruction to load values from the GOT.
7905 While this is relatively efficient, it will only work if the GOT
7906 is smaller than about 64k. Anything larger will cause the linker
7907 to report an error such as:
7909 @cindex relocation truncated to fit (MIPS)
7911 relocation truncated to fit: R_MIPS_GOT16 foobar
7914 If this happens, you should recompile your code with @option{-mxgot}.
7915 It should then work with very large GOTs, although it will also be
7916 less efficient, since it will take three instructions to fetch the
7917 value of a global symbol.
7919 Note that some linkers can create multiple GOTs. If you have such a
7920 linker, you should only need to use @option{-mxgot} when a single object
7921 file accesses more than 64k's worth of GOT entries. Very few do.
7923 These options have no effect unless GCC is generating position
7926 @item -membedded-pic
7927 @itemx -mno-embedded-pic
7928 @opindex membedded-pic
7929 @opindex mno-embedded-pic
7930 Generate (do not generate) position-independent code suitable for some
7931 embedded systems. All calls are made using PC relative addresses, and
7932 all data is addressed using the $gp register. No more than 65536
7933 bytes of global data may be used. This requires GNU as and GNU ld,
7934 which do most of the work.
7938 Assume that general-purpose registers are 32 bits wide.
7942 Assume that general-purpose registers are 64 bits wide.
7946 Assume that floating-point registers are 32 bits wide.
7950 Assume that floating-point registers are 64 bits wide.
7953 @opindex mhard-float
7954 Use floating-point coprocessor instructions.
7957 @opindex msoft-float
7958 Do not use floating-point coprocessor instructions. Implement
7959 floating-point calculations using library calls instead.
7961 @item -msingle-float
7962 @opindex msingle-float
7963 Assume that the floating-point coprocessor only supports single-precision
7966 @itemx -mdouble-float
7967 @opindex mdouble-float
7968 Assume that the floating-point coprocessor supports double-precision
7969 operations. This is the default.
7973 Force @code{int} and @code{long} types to be 64 bits wide. See
7974 @option{-mlong32} for an explanation of the default and the way
7975 that the pointer size is determined.
7979 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
7980 an explanation of the default and the way that the pointer size is
7985 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
7987 The default size of @code{int}s, @code{long}s and pointers depends on
7988 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
7989 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
7990 32-bit @code{long}s. Pointers are the same size as @code{long}s,
7991 or the same size as integer registers, whichever is smaller.
7995 @cindex smaller data references (MIPS)
7996 @cindex gp-relative references (MIPS)
7997 Put global and static items less than or equal to @var{num} bytes into
7998 the small data or bss section instead of the normal data or bss section.
7999 This allows the data to be accessed using a single instruction.
8001 All modules should be compiled with the same @option{-G @var{num}}
8004 @item -membedded-data
8005 @itemx -mno-embedded-data
8006 @opindex membedded-data
8007 @opindex mno-embedded-data
8008 Allocate variables to the read-only data section first if possible, then
8009 next in the small data section if possible, otherwise in data. This gives
8010 slightly slower code than the default, but reduces the amount of RAM required
8011 when executing, and thus may be preferred for some embedded systems.
8013 @item -muninit-const-in-rodata
8014 @itemx -mno-uninit-const-in-rodata
8015 @opindex muninit-const-in-rodata
8016 @opindex mno-uninit-const-in-rodata
8017 Put uninitialized @code{const} variables in the read-only data section.
8018 This option is only meaningful in conjunction with @option{-membedded-data}.
8020 @item -msplit-addresses
8021 @itemx -mno-split-addresses
8022 @opindex msplit-addresses
8023 @opindex mno-split-addresses
8024 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
8025 relocation operators. This option has been superceded by
8026 @option{-mexplicit-relocs} but is retained for backwards compatibility.
8028 @item -mexplicit-relocs
8029 @itemx -mno-explicit-relocs
8030 @opindex mexplicit-relocs
8031 @opindex mno-explicit-relocs
8032 Use (do not use) assembler relocation operators when dealing with symbolic
8033 addresses. The alternative, selected by @option{-mno-explicit-relocs},
8034 is to use assembler macros instead.
8036 @option{-mexplicit-relocs} is usually the default if GCC was configured
8037 to use an assembler that supports relocation operators. However, the
8038 combination of @option{-mabicalls} and @option{-fno-unit-at-a-time}
8039 implies @option{-mno-explicit-relocs} unless explicitly overridden.
8040 This is because, when generating abicalls, the choice of relocation
8041 depends on whether a symbol is local or global. In some rare cases,
8042 GCC will not be able to decide this until the whole compilation unit
8049 Generate (do not generate) code that refers to registers using their
8050 software names. The default is @option{-mno-rnames}, which tells GCC
8051 to use hardware names like @samp{$4} instead of software names like
8052 @samp{a0}. The only assembler known to support @option{-rnames} is
8053 the Algorithmics assembler.
8055 @item -mcheck-zero-division
8056 @itemx -mno-check-zero-division
8057 @opindex mcheck-zero-division
8058 @opindex mno-check-zero-division
8059 Trap (do not trap) on integer division by zero. The default is
8060 @option{-mcheck-zero-division}.
8066 Force (do not force) the use of @code{memcpy()} for non-trivial block
8067 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
8068 most constant-sized copies.
8071 @itemx -mno-long-calls
8072 @opindex mlong-calls
8073 @opindex mno-long-calls
8074 Disable (do not disable) use of the @code{jal} instruction. Calling
8075 functions using @code{jal} is more efficient but requires the caller
8076 and callee to be in the same 256 megabyte segment.
8078 This option has no effect on abicalls code. The default is
8079 @option{-mno-long-calls}.
8085 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
8086 instructions, as provided by the R4650 ISA.
8089 @itemx -mno-fused-madd
8090 @opindex mfused-madd
8091 @opindex mno-fused-madd
8092 Enable (disable) use of the floating point multiply-accumulate
8093 instructions, when they are available. The default is
8094 @option{-mfused-madd}.
8096 When multiply-accumulate instructions are used, the intermediate
8097 product is calculated to infinite precision and is not subject to
8098 the FCSR Flush to Zero bit. This may be undesirable in some
8103 Tell the MIPS assembler to not run its preprocessor over user
8104 assembler files (with a @samp{.s} suffix) when assembling them.
8107 @itemx -mno-fix-r4000
8109 @opindex mno-fix-r4000
8110 Work around certain R4000 CPU errata:
8113 A double-word or a variable shift may give an incorrect result if executed
8114 immediately after starting an integer division.
8116 A double-word or a variable shift may give an incorrect result if executed
8117 while an integer multiplication is in progress.
8119 An integer division may give an incorrect result if started in a delay slot
8120 of a taken branch or a jump.
8124 @itemx -mno-fix-r4400
8126 @opindex mno-fix-r4400
8127 Work around certain R4400 CPU errata:
8130 A double-word or a variable shift may give an incorrect result if executed
8131 immediately after starting an integer division.
8135 @itemx -mno-fix-vr4120
8136 @opindex mfix-vr4120
8137 Work around certain VR4120 errata:
8140 @code{dmultu} does not always produce the correct result.
8142 @code{div} and @code{ddiv} do not always produce the correct result if one
8143 of the operands is negative.
8145 The workarounds for the division errata rely on special functions in
8146 @file{libgcc.a}. At present, these functions are only provided by
8147 the @code{mips64vr*-elf} configurations.
8149 Other VR4120 errata require a nop to be inserted between certain pairs of
8150 instructions. These errata are handled by the assembler, not by GCC itself.
8155 Work around certain SB-1 CPU core errata.
8156 (This flag currently works around the SB-1 revision 2
8157 ``F1'' and ``F2'' floating point errata.)
8159 @item -mflush-func=@var{func}
8160 @itemx -mno-flush-func
8161 @opindex mflush-func
8162 Specifies the function to call to flush the I and D caches, or to not
8163 call any such function. If called, the function must take the same
8164 arguments as the common @code{_flush_func()}, that is, the address of the
8165 memory range for which the cache is being flushed, the size of the
8166 memory range, and the number 3 (to flush both caches). The default
8167 depends on the target GCC was configured for, but commonly is either
8168 @samp{_flush_func} or @samp{__cpu_flush}.
8170 @item -mbranch-likely
8171 @itemx -mno-branch-likely
8172 @opindex mbranch-likely
8173 @opindex mno-branch-likely
8174 Enable or disable use of Branch Likely instructions, regardless of the
8175 default for the selected architecture. By default, Branch Likely
8176 instructions may be generated if they are supported by the selected
8177 architecture. An exception is for the MIPS32 and MIPS64 architectures
8178 and processors which implement those architectures; for those, Branch
8179 Likely instructions will not be generated by default because the MIPS32
8180 and MIPS64 architectures specifically deprecate their use.
8183 @node i386 and x86-64 Options
8184 @subsection Intel 386 and AMD x86-64 Options
8185 @cindex i386 Options
8186 @cindex x86-64 Options
8187 @cindex Intel 386 Options
8188 @cindex AMD x86-64 Options
8190 These @samp{-m} options are defined for the i386 and x86-64 family of
8194 @item -mtune=@var{cpu-type}
8196 Tune to @var{cpu-type} everything applicable about the generated code, except
8197 for the ABI and the set of available instructions. The choices for
8201 Original Intel's i386 CPU.
8203 Intel's i486 CPU. (No scheduling is implemented for this chip.)
8205 Intel Pentium CPU with no MMX support.
8207 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
8208 @item i686, pentiumpro
8209 Intel PentiumPro CPU.
8211 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
8212 @item pentium3, pentium3m
8213 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
8216 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
8217 support. Used by Centrino notebooks.
8218 @item pentium4, pentium4m
8219 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
8221 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
8224 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
8225 SSE2 and SSE3 instruction set support.
8227 AMD K6 CPU with MMX instruction set support.
8229 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
8230 @item athlon, athlon-tbird
8231 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
8233 @item athlon-4, athlon-xp, athlon-mp
8234 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
8235 instruction set support.
8236 @item k8, opteron, athlon64, athlon-fx
8237 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
8238 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
8240 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
8243 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
8244 instruction set support.
8246 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
8247 implemented for this chip.)
8249 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
8250 implemented for this chip.)
8253 While picking a specific @var{cpu-type} will schedule things appropriately
8254 for that particular chip, the compiler will not generate any code that
8255 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8258 @item -march=@var{cpu-type}
8260 Generate instructions for the machine type @var{cpu-type}. The choices
8261 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8262 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8264 @item -mcpu=@var{cpu-type}
8266 A deprecated synonym for @option{-mtune}.
8275 @opindex mpentiumpro
8276 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8277 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8278 These synonyms are deprecated.
8280 @item -mfpmath=@var{unit}
8282 Generate floating point arithmetics for selected unit @var{unit}. The choices
8287 Use the standard 387 floating point coprocessor present majority of chips and
8288 emulated otherwise. Code compiled with this option will run almost everywhere.
8289 The temporary results are computed in 80bit precision instead of precision
8290 specified by the type resulting in slightly different results compared to most
8291 of other chips. See @option{-ffloat-store} for more detailed description.
8293 This is the default choice for i386 compiler.
8296 Use scalar floating point instructions present in the SSE instruction set.
8297 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8298 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8299 instruction set supports only single precision arithmetics, thus the double and
8300 extended precision arithmetics is still done using 387. Later version, present
8301 only in Pentium4 and the future AMD x86-64 chips supports double precision
8304 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8305 @option{-msse2} switches to enable SSE extensions and make this option
8306 effective. For x86-64 compiler, these extensions are enabled by default.
8308 The resulting code should be considerably faster in the majority of cases and avoid
8309 the numerical instability problems of 387 code, but may break some existing
8310 code that expects temporaries to be 80bit.
8312 This is the default choice for the x86-64 compiler.
8315 Attempt to utilize both instruction sets at once. This effectively double the
8316 amount of available registers and on chips with separate execution units for
8317 387 and SSE the execution resources too. Use this option with care, as it is
8318 still experimental, because the GCC register allocator does not model separate
8319 functional units well resulting in instable performance.
8322 @item -masm=@var{dialect}
8323 @opindex masm=@var{dialect}
8324 Output asm instructions using selected @var{dialect}. Supported choices are
8325 @samp{intel} or @samp{att} (the default one).
8330 @opindex mno-ieee-fp
8331 Control whether or not the compiler uses IEEE floating point
8332 comparisons. These handle correctly the case where the result of a
8333 comparison is unordered.
8336 @opindex msoft-float
8337 Generate output containing library calls for floating point.
8338 @strong{Warning:} the requisite libraries are not part of GCC@.
8339 Normally the facilities of the machine's usual C compiler are used, but
8340 this can't be done directly in cross-compilation. You must make your
8341 own arrangements to provide suitable library functions for
8344 On machines where a function returns floating point results in the 80387
8345 register stack, some floating point opcodes may be emitted even if
8346 @option{-msoft-float} is used.
8348 @item -mno-fp-ret-in-387
8349 @opindex mno-fp-ret-in-387
8350 Do not use the FPU registers for return values of functions.
8352 The usual calling convention has functions return values of types
8353 @code{float} and @code{double} in an FPU register, even if there
8354 is no FPU@. The idea is that the operating system should emulate
8357 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8358 in ordinary CPU registers instead.
8360 @item -mno-fancy-math-387
8361 @opindex mno-fancy-math-387
8362 Some 387 emulators do not support the @code{sin}, @code{cos} and
8363 @code{sqrt} instructions for the 387. Specify this option to avoid
8364 generating those instructions. This option is the default on FreeBSD,
8365 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8366 indicates that the target cpu will always have an FPU and so the
8367 instruction will not need emulation. As of revision 2.6.1, these
8368 instructions are not generated unless you also use the
8369 @option{-funsafe-math-optimizations} switch.
8371 @item -malign-double
8372 @itemx -mno-align-double
8373 @opindex malign-double
8374 @opindex mno-align-double
8375 Control whether GCC aligns @code{double}, @code{long double}, and
8376 @code{long long} variables on a two word boundary or a one word
8377 boundary. Aligning @code{double} variables on a two word boundary will
8378 produce code that runs somewhat faster on a @samp{Pentium} at the
8379 expense of more memory.
8381 @strong{Warning:} if you use the @option{-malign-double} switch,
8382 structures containing the above types will be aligned differently than
8383 the published application binary interface specifications for the 386
8384 and will not be binary compatible with structures in code compiled
8385 without that switch.
8387 @item -m96bit-long-double
8388 @itemx -m128bit-long-double
8389 @opindex m96bit-long-double
8390 @opindex m128bit-long-double
8391 These switches control the size of @code{long double} type. The i386
8392 application binary interface specifies the size to be 96 bits,
8393 so @option{-m96bit-long-double} is the default in 32 bit mode.
8395 Modern architectures (Pentium and newer) would prefer @code{long double}
8396 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8397 conforming to the ABI, this would not be possible. So specifying a
8398 @option{-m128bit-long-double} will align @code{long double}
8399 to a 16 byte boundary by padding the @code{long double} with an additional
8402 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8403 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8405 Notice that neither of these options enable any extra precision over the x87
8406 standard of 80 bits for a @code{long double}.
8408 @strong{Warning:} if you override the default value for your target ABI, the
8409 structures and arrays containing @code{long double} variables will change
8410 their size as well as function calling convention for function taking
8411 @code{long double} will be modified. Hence they will not be binary
8412 compatible with arrays or structures in code compiled without that switch.
8416 @itemx -mno-svr3-shlib
8417 @opindex msvr3-shlib
8418 @opindex mno-svr3-shlib
8419 Control whether GCC places uninitialized local variables into the
8420 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8421 into @code{bss}. These options are meaningful only on System V Release 3.
8425 Use a different function-calling convention, in which functions that
8426 take a fixed number of arguments return with the @code{ret} @var{num}
8427 instruction, which pops their arguments while returning. This saves one
8428 instruction in the caller since there is no need to pop the arguments
8431 You can specify that an individual function is called with this calling
8432 sequence with the function attribute @samp{stdcall}. You can also
8433 override the @option{-mrtd} option by using the function attribute
8434 @samp{cdecl}. @xref{Function Attributes}.
8436 @strong{Warning:} this calling convention is incompatible with the one
8437 normally used on Unix, so you cannot use it if you need to call
8438 libraries compiled with the Unix compiler.
8440 Also, you must provide function prototypes for all functions that
8441 take variable numbers of arguments (including @code{printf});
8442 otherwise incorrect code will be generated for calls to those
8445 In addition, seriously incorrect code will result if you call a
8446 function with too many arguments. (Normally, extra arguments are
8447 harmlessly ignored.)
8449 @item -mregparm=@var{num}
8451 Control how many registers are used to pass integer arguments. By
8452 default, no registers are used to pass arguments, and at most 3
8453 registers can be used. You can control this behavior for a specific
8454 function by using the function attribute @samp{regparm}.
8455 @xref{Function Attributes}.
8457 @strong{Warning:} if you use this switch, and
8458 @var{num} is nonzero, then you must build all modules with the same
8459 value, including any libraries. This includes the system libraries and
8462 @item -mpreferred-stack-boundary=@var{num}
8463 @opindex mpreferred-stack-boundary
8464 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8465 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8466 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8467 size (@option{-Os}), in which case the default is the minimum correct
8468 alignment (4 bytes for x86, and 8 bytes for x86-64).
8470 On Pentium and PentiumPro, @code{double} and @code{long double} values
8471 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8472 suffer significant run time performance penalties. On Pentium III, the
8473 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8474 penalties if it is not 16 byte aligned.
8476 To ensure proper alignment of this values on the stack, the stack boundary
8477 must be as aligned as that required by any value stored on the stack.
8478 Further, every function must be generated such that it keeps the stack
8479 aligned. Thus calling a function compiled with a higher preferred
8480 stack boundary from a function compiled with a lower preferred stack
8481 boundary will most likely misalign the stack. It is recommended that
8482 libraries that use callbacks always use the default setting.
8484 This extra alignment does consume extra stack space, and generally
8485 increases code size. Code that is sensitive to stack space usage, such
8486 as embedded systems and operating system kernels, may want to reduce the
8487 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8505 These switches enable or disable the use of built-in functions that allow
8506 direct access to the MMX, SSE, SSE2, SSE3 and 3Dnow extensions of the
8509 @xref{X86 Built-in Functions}, for details of the functions enabled
8510 and disabled by these switches.
8512 To have SSE/SSE2 instructions generated automatically from floating-point
8513 code, see @option{-mfpmath=sse}.
8516 @itemx -mno-push-args
8518 @opindex mno-push-args
8519 Use PUSH operations to store outgoing parameters. This method is shorter
8520 and usually equally fast as method using SUB/MOV operations and is enabled
8521 by default. In some cases disabling it may improve performance because of
8522 improved scheduling and reduced dependencies.
8524 @item -maccumulate-outgoing-args
8525 @opindex maccumulate-outgoing-args
8526 If enabled, the maximum amount of space required for outgoing arguments will be
8527 computed in the function prologue. This is faster on most modern CPUs
8528 because of reduced dependencies, improved scheduling and reduced stack usage
8529 when preferred stack boundary is not equal to 2. The drawback is a notable
8530 increase in code size. This switch implies @option{-mno-push-args}.
8534 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8535 on thread-safe exception handling must compile and link all code with the
8536 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8537 @option{-D_MT}; when linking, it links in a special thread helper library
8538 @option{-lmingwthrd} which cleans up per thread exception handling data.
8540 @item -mno-align-stringops
8541 @opindex mno-align-stringops
8542 Do not align destination of inlined string operations. This switch reduces
8543 code size and improves performance in case the destination is already aligned,
8544 but GCC doesn't know about it.
8546 @item -minline-all-stringops
8547 @opindex minline-all-stringops
8548 By default GCC inlines string operations only when destination is known to be
8549 aligned at least to 4 byte boundary. This enables more inlining, increase code
8550 size, but may improve performance of code that depends on fast memcpy, strlen
8551 and memset for short lengths.
8553 @item -momit-leaf-frame-pointer
8554 @opindex momit-leaf-frame-pointer
8555 Don't keep the frame pointer in a register for leaf functions. This
8556 avoids the instructions to save, set up and restore frame pointers and
8557 makes an extra register available in leaf functions. The option
8558 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8559 which might make debugging harder.
8561 @item -mtls-direct-seg-refs
8562 @itemx -mno-tls-direct-seg-refs
8563 @opindex mtls-direct-seg-refs
8564 Controls whether TLS variables may be accessed with offsets from the
8565 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8566 or whether the thread base pointer must be added. Whether or not this
8567 is legal depends on the operating system, and whether it maps the
8568 segment to cover the entire TLS area.
8570 For systems that use GNU libc, the default is on.
8573 These @samp{-m} switches are supported in addition to the above
8574 on AMD x86-64 processors in 64-bit environments.
8581 Generate code for a 32-bit or 64-bit environment.
8582 The 32-bit environment sets int, long and pointer to 32 bits and
8583 generates code that runs on any i386 system.
8584 The 64-bit environment sets int to 32 bits and long and pointer
8585 to 64 bits and generates code for AMD's x86-64 architecture.
8588 @opindex no-red-zone
8589 Do not use a so called red zone for x86-64 code. The red zone is mandated
8590 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8591 stack pointer that will not be modified by signal or interrupt handlers
8592 and therefore can be used for temporary data without adjusting the stack
8593 pointer. The flag @option{-mno-red-zone} disables this red zone.
8595 @item -mcmodel=small
8596 @opindex mcmodel=small
8597 Generate code for the small code model: the program and its symbols must
8598 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8599 Programs can be statically or dynamically linked. This is the default
8602 @item -mcmodel=kernel
8603 @opindex mcmodel=kernel
8604 Generate code for the kernel code model. The kernel runs in the
8605 negative 2 GB of the address space.
8606 This model has to be used for Linux kernel code.
8608 @item -mcmodel=medium
8609 @opindex mcmodel=medium
8610 Generate code for the medium model: The program is linked in the lower 2
8611 GB of the address space but symbols can be located anywhere in the
8612 address space. Programs can be statically or dynamically linked, but
8613 building of shared libraries are not supported with the medium model.
8615 @item -mcmodel=large
8616 @opindex mcmodel=large
8617 Generate code for the large model: This model makes no assumptions
8618 about addresses and sizes of sections. Currently GCC does not implement
8623 @subsection HPPA Options
8624 @cindex HPPA Options
8626 These @samp{-m} options are defined for the HPPA family of computers:
8629 @item -march=@var{architecture-type}
8631 Generate code for the specified architecture. The choices for
8632 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8633 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8634 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8635 architecture option for your machine. Code compiled for lower numbered
8636 architectures will run on higher numbered architectures, but not the
8639 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8640 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8644 @itemx -mpa-risc-1-1
8645 @itemx -mpa-risc-2-0
8646 @opindex mpa-risc-1-0
8647 @opindex mpa-risc-1-1
8648 @opindex mpa-risc-2-0
8649 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8652 @opindex mbig-switch
8653 Generate code suitable for big switch tables. Use this option only if
8654 the assembler/linker complain about out of range branches within a switch
8657 @item -mjump-in-delay
8658 @opindex mjump-in-delay
8659 Fill delay slots of function calls with unconditional jump instructions
8660 by modifying the return pointer for the function call to be the target
8661 of the conditional jump.
8663 @item -mdisable-fpregs
8664 @opindex mdisable-fpregs
8665 Prevent floating point registers from being used in any manner. This is
8666 necessary for compiling kernels which perform lazy context switching of
8667 floating point registers. If you use this option and attempt to perform
8668 floating point operations, the compiler will abort.
8670 @item -mdisable-indexing
8671 @opindex mdisable-indexing
8672 Prevent the compiler from using indexing address modes. This avoids some
8673 rather obscure problems when compiling MIG generated code under MACH@.
8675 @item -mno-space-regs
8676 @opindex mno-space-regs
8677 Generate code that assumes the target has no space registers. This allows
8678 GCC to generate faster indirect calls and use unscaled index address modes.
8680 Such code is suitable for level 0 PA systems and kernels.
8682 @item -mfast-indirect-calls
8683 @opindex mfast-indirect-calls
8684 Generate code that assumes calls never cross space boundaries. This
8685 allows GCC to emit code which performs faster indirect calls.
8687 This option will not work in the presence of shared libraries or nested
8690 @item -mlong-load-store
8691 @opindex mlong-load-store
8692 Generate 3-instruction load and store sequences as sometimes required by
8693 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8696 @item -mportable-runtime
8697 @opindex mportable-runtime
8698 Use the portable calling conventions proposed by HP for ELF systems.
8702 Enable the use of assembler directives only GAS understands.
8704 @item -mschedule=@var{cpu-type}
8706 Schedule code according to the constraints for the machine type
8707 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8708 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8709 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8710 proper scheduling option for your machine. The default scheduling is
8714 @opindex mlinker-opt
8715 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8716 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8717 linkers in which they give bogus error messages when linking some programs.
8720 @opindex msoft-float
8721 Generate output containing library calls for floating point.
8722 @strong{Warning:} the requisite libraries are not available for all HPPA
8723 targets. Normally the facilities of the machine's usual C compiler are
8724 used, but this cannot be done directly in cross-compilation. You must make
8725 your own arrangements to provide suitable library functions for
8726 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8727 does provide software floating point support.
8729 @option{-msoft-float} changes the calling convention in the output file;
8730 therefore, it is only useful if you compile @emph{all} of a program with
8731 this option. In particular, you need to compile @file{libgcc.a}, the
8732 library that comes with GCC, with @option{-msoft-float} in order for
8737 Generate the predefine, @code{_SIO}, for server IO. The default is
8738 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8739 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8740 options are available under HP-UX and HI-UX.
8744 Use GNU ld specific options. This passes @option{-shared} to ld when
8745 building a shared library. It is the default when GCC is configured,
8746 explicitly or implicitly, with the GNU linker. This option does not
8747 have any affect on which ld is called, it only changes what parameters
8748 are passed to that ld. The ld that is called is determined by the
8749 @option{--with-ld} configure option, GCC's program search path, and
8750 finally by the user's @env{PATH}. The linker used by GCC can be printed
8751 using @samp{which `gcc -print-prog-name=ld`}.
8755 Use HP ld specific options. This passes @option{-b} to ld when building
8756 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8757 links. It is the default when GCC is configured, explicitly or
8758 implicitly, with the HP linker. This option does not have any affect on
8759 which ld is called, it only changes what parameters are passed to that
8760 ld. The ld that is called is determined by the @option{--with-ld}
8761 configure option, GCC's program search path, and finally by the user's
8762 @env{PATH}. The linker used by GCC can be printed using @samp{which
8763 `gcc -print-prog-name=ld`}.
8768 Select the FDPIC ABI, that uses function descriptors to represent
8769 pointers to functions. Without any PIC/PIE-related options, it
8770 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
8771 assumes GOT entries and small data are within a 12-bit range from the
8772 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
8773 are computed with 32 bits.
8776 @opindex minline-plt
8778 Enable inlining of PLT entries in function calls to functions that are
8779 not known to bind locally. It has no effect without @option{-mfdpic}.
8780 It's enabled by default if optimizing for speed and compiling for
8781 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
8782 optimization option such as @option{-O3} or above is present in the
8788 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
8789 that is known to be in read-only sections. It's enabled by default,
8790 except for @option{-fpic} or @option{-fpie}: even though it may help
8791 make the global offset table smaller, it trades 1 instruction for 4.
8792 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
8793 one of which may be shared by multiple symbols, and it avoids the need
8794 for a GOT entry for the referenced symbol, so it's more likely to be a
8795 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
8797 @item -multilib-library-pic
8798 @opindex multilib-library-pic
8800 Link with the (library, not FD) pic libraries. It's implied by
8801 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
8802 @option{-fpic} without @option{-mfdpic}. You should never have to use
8808 Follow the EABI requirement of always creating a frame pointer whenever
8809 a stack frame is allocated. This option is enabled by default and can
8810 be disabled with @option{-mno-linked-fp}.
8813 @opindex mno-long-calls
8814 Generate code that uses long call sequences. This ensures that a call
8815 is always able to reach linker generated stubs. The default is to generate
8816 long calls only when the distance from the call site to the beginning
8817 of the function or translation unit, as the case may be, exceeds a
8818 predefined limit set by the branch type being used. The limits for
8819 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8820 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8823 Distances are measured from the beginning of functions when using the
8824 @option{-ffunction-sections} option, or when using the @option{-mgas}
8825 and @option{-mno-portable-runtime} options together under HP-UX with
8828 It is normally not desirable to use this option as it will degrade
8829 performance. However, it may be useful in large applications,
8830 particularly when partial linking is used to build the application.
8832 The types of long calls used depends on the capabilities of the
8833 assembler and linker, and the type of code being generated. The
8834 impact on systems that support long absolute calls, and long pic
8835 symbol-difference or pc-relative calls should be relatively small.
8836 However, an indirect call is used on 32-bit ELF systems in pic code
8837 and it is quite long.
8841 Suppress the generation of link options to search libdld.sl when the
8842 @option{-static} option is specified on HP-UX 10 and later.
8846 The HP-UX implementation of setlocale in libc has a dependency on
8847 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8848 when the @option{-static} option is specified, special link options
8849 are needed to resolve this dependency.
8851 On HP-UX 10 and later, the GCC driver adds the necessary options to
8852 link with libdld.sl when the @option{-static} option is specified.
8853 This causes the resulting binary to be dynamic. On the 64-bit port,
8854 the linkers generate dynamic binaries by default in any case. The
8855 @option{-nolibdld} option can be used to prevent the GCC driver from
8856 adding these link options.
8860 Add support for multithreading with the @dfn{dce thread} library
8861 under HP-UX. This option sets flags for both the preprocessor and
8865 @node DEC Alpha Options
8866 @subsection DEC Alpha Options
8868 These @samp{-m} options are defined for the DEC Alpha implementations:
8871 @item -mno-soft-float
8873 @opindex mno-soft-float
8874 @opindex msoft-float
8875 Use (do not use) the hardware floating-point instructions for
8876 floating-point operations. When @option{-msoft-float} is specified,
8877 functions in @file{libgcc.a} will be used to perform floating-point
8878 operations. Unless they are replaced by routines that emulate the
8879 floating-point operations, or compiled in such a way as to call such
8880 emulations routines, these routines will issue floating-point
8881 operations. If you are compiling for an Alpha without floating-point
8882 operations, you must ensure that the library is built so as not to call
8885 Note that Alpha implementations without floating-point operations are
8886 required to have floating-point registers.
8891 @opindex mno-fp-regs
8892 Generate code that uses (does not use) the floating-point register set.
8893 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8894 register set is not used, floating point operands are passed in integer
8895 registers as if they were integers and floating-point results are passed
8896 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8897 so any function with a floating-point argument or return value called by code
8898 compiled with @option{-mno-fp-regs} must also be compiled with that
8901 A typical use of this option is building a kernel that does not use,
8902 and hence need not save and restore, any floating-point registers.
8906 The Alpha architecture implements floating-point hardware optimized for
8907 maximum performance. It is mostly compliant with the IEEE floating
8908 point standard. However, for full compliance, software assistance is
8909 required. This option generates code fully IEEE compliant code
8910 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8911 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8912 defined during compilation. The resulting code is less efficient but is
8913 able to correctly support denormalized numbers and exceptional IEEE
8914 values such as not-a-number and plus/minus infinity. Other Alpha
8915 compilers call this option @option{-ieee_with_no_inexact}.
8917 @item -mieee-with-inexact
8918 @opindex mieee-with-inexact
8919 This is like @option{-mieee} except the generated code also maintains
8920 the IEEE @var{inexact-flag}. Turning on this option causes the
8921 generated code to implement fully-compliant IEEE math. In addition to
8922 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8923 macro. On some Alpha implementations the resulting code may execute
8924 significantly slower than the code generated by default. Since there is
8925 very little code that depends on the @var{inexact-flag}, you should
8926 normally not specify this option. Other Alpha compilers call this
8927 option @option{-ieee_with_inexact}.
8929 @item -mfp-trap-mode=@var{trap-mode}
8930 @opindex mfp-trap-mode
8931 This option controls what floating-point related traps are enabled.
8932 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8933 The trap mode can be set to one of four values:
8937 This is the default (normal) setting. The only traps that are enabled
8938 are the ones that cannot be disabled in software (e.g., division by zero
8942 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8946 Like @samp{su}, but the instructions are marked to be safe for software
8947 completion (see Alpha architecture manual for details).
8950 Like @samp{su}, but inexact traps are enabled as well.
8953 @item -mfp-rounding-mode=@var{rounding-mode}
8954 @opindex mfp-rounding-mode
8955 Selects the IEEE rounding mode. Other Alpha compilers call this option
8956 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8961 Normal IEEE rounding mode. Floating point numbers are rounded towards
8962 the nearest machine number or towards the even machine number in case
8966 Round towards minus infinity.
8969 Chopped rounding mode. Floating point numbers are rounded towards zero.
8972 Dynamic rounding mode. A field in the floating point control register
8973 (@var{fpcr}, see Alpha architecture reference manual) controls the
8974 rounding mode in effect. The C library initializes this register for
8975 rounding towards plus infinity. Thus, unless your program modifies the
8976 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8979 @item -mtrap-precision=@var{trap-precision}
8980 @opindex mtrap-precision
8981 In the Alpha architecture, floating point traps are imprecise. This
8982 means without software assistance it is impossible to recover from a
8983 floating trap and program execution normally needs to be terminated.
8984 GCC can generate code that can assist operating system trap handlers
8985 in determining the exact location that caused a floating point trap.
8986 Depending on the requirements of an application, different levels of
8987 precisions can be selected:
8991 Program precision. This option is the default and means a trap handler
8992 can only identify which program caused a floating point exception.
8995 Function precision. The trap handler can determine the function that
8996 caused a floating point exception.
8999 Instruction precision. The trap handler can determine the exact
9000 instruction that caused a floating point exception.
9003 Other Alpha compilers provide the equivalent options called
9004 @option{-scope_safe} and @option{-resumption_safe}.
9006 @item -mieee-conformant
9007 @opindex mieee-conformant
9008 This option marks the generated code as IEEE conformant. You must not
9009 use this option unless you also specify @option{-mtrap-precision=i} and either
9010 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9011 is to emit the line @samp{.eflag 48} in the function prologue of the
9012 generated assembly file. Under DEC Unix, this has the effect that
9013 IEEE-conformant math library routines will be linked in.
9015 @item -mbuild-constants
9016 @opindex mbuild-constants
9017 Normally GCC examines a 32- or 64-bit integer constant to
9018 see if it can construct it from smaller constants in two or three
9019 instructions. If it cannot, it will output the constant as a literal and
9020 generate code to load it from the data segment at runtime.
9022 Use this option to require GCC to construct @emph{all} integer constants
9023 using code, even if it takes more instructions (the maximum is six).
9025 You would typically use this option to build a shared library dynamic
9026 loader. Itself a shared library, it must relocate itself in memory
9027 before it can find the variables and constants in its own data segment.
9033 Select whether to generate code to be assembled by the vendor-supplied
9034 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9052 Indicate whether GCC should generate code to use the optional BWX,
9053 CIX, FIX and MAX instruction sets. The default is to use the instruction
9054 sets supported by the CPU type specified via @option{-mcpu=} option or that
9055 of the CPU on which GCC was built if none was specified.
9060 @opindex mfloat-ieee
9061 Generate code that uses (does not use) VAX F and G floating point
9062 arithmetic instead of IEEE single and double precision.
9064 @item -mexplicit-relocs
9065 @itemx -mno-explicit-relocs
9066 @opindex mexplicit-relocs
9067 @opindex mno-explicit-relocs
9068 Older Alpha assemblers provided no way to generate symbol relocations
9069 except via assembler macros. Use of these macros does not allow
9070 optimal instruction scheduling. GNU binutils as of version 2.12
9071 supports a new syntax that allows the compiler to explicitly mark
9072 which relocations should apply to which instructions. This option
9073 is mostly useful for debugging, as GCC detects the capabilities of
9074 the assembler when it is built and sets the default accordingly.
9078 @opindex msmall-data
9079 @opindex mlarge-data
9080 When @option{-mexplicit-relocs} is in effect, static data is
9081 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9082 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9083 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9084 16-bit relocations off of the @code{$gp} register. This limits the
9085 size of the small data area to 64KB, but allows the variables to be
9086 directly accessed via a single instruction.
9088 The default is @option{-mlarge-data}. With this option the data area
9089 is limited to just below 2GB. Programs that require more than 2GB of
9090 data must use @code{malloc} or @code{mmap} to allocate the data in the
9091 heap instead of in the program's data segment.
9093 When generating code for shared libraries, @option{-fpic} implies
9094 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9098 @opindex msmall-text
9099 @opindex mlarge-text
9100 When @option{-msmall-text} is used, the compiler assumes that the
9101 code of the entire program (or shared library) fits in 4MB, and is
9102 thus reachable with a branch instruction. When @option{-msmall-data}
9103 is used, the compiler can assume that all local symbols share the
9104 same @code{$gp} value, and thus reduce the number of instructions
9105 required for a function call from 4 to 1.
9107 The default is @option{-mlarge-text}.
9109 @item -mcpu=@var{cpu_type}
9111 Set the instruction set and instruction scheduling parameters for
9112 machine type @var{cpu_type}. You can specify either the @samp{EV}
9113 style name or the corresponding chip number. GCC supports scheduling
9114 parameters for the EV4, EV5 and EV6 family of processors and will
9115 choose the default values for the instruction set from the processor
9116 you specify. If you do not specify a processor type, GCC will default
9117 to the processor on which the compiler was built.
9119 Supported values for @var{cpu_type} are
9125 Schedules as an EV4 and has no instruction set extensions.
9129 Schedules as an EV5 and has no instruction set extensions.
9133 Schedules as an EV5 and supports the BWX extension.
9138 Schedules as an EV5 and supports the BWX and MAX extensions.
9142 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9146 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9149 @item -mtune=@var{cpu_type}
9151 Set only the instruction scheduling parameters for machine type
9152 @var{cpu_type}. The instruction set is not changed.
9154 @item -mmemory-latency=@var{time}
9155 @opindex mmemory-latency
9156 Sets the latency the scheduler should assume for typical memory
9157 references as seen by the application. This number is highly
9158 dependent on the memory access patterns used by the application
9159 and the size of the external cache on the machine.
9161 Valid options for @var{time} are
9165 A decimal number representing clock cycles.
9171 The compiler contains estimates of the number of clock cycles for
9172 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9173 (also called Dcache, Scache, and Bcache), as well as to main memory.
9174 Note that L3 is only valid for EV5.
9179 @node DEC Alpha/VMS Options
9180 @subsection DEC Alpha/VMS Options
9182 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9185 @item -mvms-return-codes
9186 @opindex mvms-return-codes
9187 Return VMS condition codes from main. The default is to return POSIX
9188 style condition (e.g.@ error) codes.
9191 @node H8/300 Options
9192 @subsection H8/300 Options
9194 These @samp{-m} options are defined for the H8/300 implementations:
9199 Shorten some address references at link time, when possible; uses the
9200 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9201 ld, Using ld}, for a fuller description.
9205 Generate code for the H8/300H@.
9209 Generate code for the H8S@.
9213 Generate code for the H8S and H8/300H in the normal mode. This switch
9214 must be used either with -mh or -ms.
9218 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9222 Make @code{int} data 32 bits by default.
9226 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9227 The default for the H8/300H and H8S is to align longs and floats on 4
9229 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9230 This option has no effect on the H8/300.
9234 @subsection SH Options
9236 These @samp{-m} options are defined for the SH implementations:
9241 Generate code for the SH1.
9245 Generate code for the SH2.
9248 Generate code for the SH2e.
9252 Generate code for the SH3.
9256 Generate code for the SH3e.
9260 Generate code for the SH4 without a floating-point unit.
9262 @item -m4-single-only
9263 @opindex m4-single-only
9264 Generate code for the SH4 with a floating-point unit that only
9265 supports single-precision arithmetic.
9269 Generate code for the SH4 assuming the floating-point unit is in
9270 single-precision mode by default.
9274 Generate code for the SH4.
9278 Compile code for the processor in big endian mode.
9282 Compile code for the processor in little endian mode.
9286 Align doubles at 64-bit boundaries. Note that this changes the calling
9287 conventions, and thus some functions from the standard C library will
9288 not work unless you recompile it first with @option{-mdalign}.
9292 Shorten some address references at link time, when possible; uses the
9293 linker option @option{-relax}.
9297 Use 32-bit offsets in @code{switch} tables. The default is to use
9302 Enable the use of the instruction @code{fmovd}.
9306 Comply with the calling conventions defined by Renesas.
9310 Mark the @code{MAC} register as call-clobbered, even if
9311 @option{-mhitachi} is given.
9315 Increase IEEE-compliance of floating-point code.
9319 Dump instruction size and location in the assembly code.
9323 This option is deprecated. It pads structures to multiple of 4 bytes,
9324 which is incompatible with the SH ABI@.
9328 Optimize for space instead of speed. Implied by @option{-Os}.
9332 When generating position-independent code, emit function calls using
9333 the Global Offset Table instead of the Procedure Linkage Table.
9337 Generate a library function call to invalidate instruction cache
9338 entries, after fixing up a trampoline. This library function call
9339 doesn't assume it can write to the whole memory address space. This
9340 is the default when the target is @code{sh-*-linux*}.
9343 @node System V Options
9344 @subsection Options for System V
9346 These additional options are available on System V Release 4 for
9347 compatibility with other compilers on those systems:
9352 Create a shared object.
9353 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9357 Identify the versions of each tool used by the compiler, in a
9358 @code{.ident} assembler directive in the output.
9362 Refrain from adding @code{.ident} directives to the output file (this is
9365 @item -YP,@var{dirs}
9367 Search the directories @var{dirs}, and no others, for libraries
9368 specified with @option{-l}.
9372 Look in the directory @var{dir} to find the M4 preprocessor.
9373 The assembler uses this option.
9374 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9375 @c the generic assembler that comes with Solaris takes just -Ym.
9378 @node TMS320C3x/C4x Options
9379 @subsection TMS320C3x/C4x Options
9380 @cindex TMS320C3x/C4x Options
9382 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9386 @item -mcpu=@var{cpu_type}
9388 Set the instruction set, register set, and instruction scheduling
9389 parameters for machine type @var{cpu_type}. Supported values for
9390 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9391 @samp{c44}. The default is @samp{c40} to generate code for the
9396 @itemx -msmall-memory
9398 @opindex mbig-memory
9400 @opindex msmall-memory
9402 Generates code for the big or small memory model. The small memory
9403 model assumed that all data fits into one 64K word page. At run-time
9404 the data page (DP) register must be set to point to the 64K page
9405 containing the .bss and .data program sections. The big memory model is
9406 the default and requires reloading of the DP register for every direct
9413 Allow (disallow) allocation of general integer operands into the block
9420 Enable (disable) generation of code using decrement and branch,
9421 DBcond(D), instructions. This is enabled by default for the C4x. To be
9422 on the safe side, this is disabled for the C3x, since the maximum
9423 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9424 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9425 that it can utilize the decrement and branch instruction, but will give
9426 up if there is more than one memory reference in the loop. Thus a loop
9427 where the loop counter is decremented can generate slightly more
9428 efficient code, in cases where the RPTB instruction cannot be utilized.
9430 @item -mdp-isr-reload
9432 @opindex mdp-isr-reload
9434 Force the DP register to be saved on entry to an interrupt service
9435 routine (ISR), reloaded to point to the data section, and restored on
9436 exit from the ISR@. This should not be required unless someone has
9437 violated the small memory model by modifying the DP register, say within
9444 For the C3x use the 24-bit MPYI instruction for integer multiplies
9445 instead of a library call to guarantee 32-bit results. Note that if one
9446 of the operands is a constant, then the multiplication will be performed
9447 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9448 then squaring operations are performed inline instead of a library call.
9451 @itemx -mno-fast-fix
9453 @opindex mno-fast-fix
9454 The C3x/C4x FIX instruction to convert a floating point value to an
9455 integer value chooses the nearest integer less than or equal to the
9456 floating point value rather than to the nearest integer. Thus if the
9457 floating point number is negative, the result will be incorrectly
9458 truncated an additional code is necessary to detect and correct this
9459 case. This option can be used to disable generation of the additional
9460 code required to correct the result.
9466 Enable (disable) generation of repeat block sequences using the RPTB
9467 instruction for zero overhead looping. The RPTB construct is only used
9468 for innermost loops that do not call functions or jump across the loop
9469 boundaries. There is no advantage having nested RPTB loops due to the
9470 overhead required to save and restore the RC, RS, and RE registers.
9471 This is enabled by default with @option{-O2}.
9473 @item -mrpts=@var{count}
9477 Enable (disable) the use of the single instruction repeat instruction
9478 RPTS@. If a repeat block contains a single instruction, and the loop
9479 count can be guaranteed to be less than the value @var{count}, GCC will
9480 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9481 then a RPTS will be emitted even if the loop count cannot be determined
9482 at compile time. Note that the repeated instruction following RPTS does
9483 not have to be reloaded from memory each iteration, thus freeing up the
9484 CPU buses for operands. However, since interrupts are blocked by this
9485 instruction, it is disabled by default.
9487 @item -mloop-unsigned
9488 @itemx -mno-loop-unsigned
9489 @opindex mloop-unsigned
9490 @opindex mno-loop-unsigned
9491 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9492 is @math{2^{31} + 1} since these instructions test if the iteration count is
9493 negative to terminate the loop. If the iteration count is unsigned
9494 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9495 exceeded. This switch allows an unsigned iteration count.
9499 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9500 with. This also enforces compatibility with the API employed by the TI
9501 C3x C compiler. For example, long doubles are passed as structures
9502 rather than in floating point registers.
9508 Generate code that uses registers (stack) for passing arguments to functions.
9509 By default, arguments are passed in registers where possible rather
9510 than by pushing arguments on to the stack.
9512 @item -mparallel-insns
9513 @itemx -mno-parallel-insns
9514 @opindex mparallel-insns
9515 @opindex mno-parallel-insns
9516 Allow the generation of parallel instructions. This is enabled by
9517 default with @option{-O2}.
9519 @item -mparallel-mpy
9520 @itemx -mno-parallel-mpy
9521 @opindex mparallel-mpy
9522 @opindex mno-parallel-mpy
9523 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9524 provided @option{-mparallel-insns} is also specified. These instructions have
9525 tight register constraints which can pessimize the code generation
9531 @subsection V850 Options
9532 @cindex V850 Options
9534 These @samp{-m} options are defined for V850 implementations:
9538 @itemx -mno-long-calls
9539 @opindex mlong-calls
9540 @opindex mno-long-calls
9541 Treat all calls as being far away (near). If calls are assumed to be
9542 far away, the compiler will always load the functions address up into a
9543 register, and call indirect through the pointer.
9549 Do not optimize (do optimize) basic blocks that use the same index
9550 pointer 4 or more times to copy pointer into the @code{ep} register, and
9551 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9552 option is on by default if you optimize.
9554 @item -mno-prolog-function
9555 @itemx -mprolog-function
9556 @opindex mno-prolog-function
9557 @opindex mprolog-function
9558 Do not use (do use) external functions to save and restore registers
9559 at the prologue and epilogue of a function. The external functions
9560 are slower, but use less code space if more than one function saves
9561 the same number of registers. The @option{-mprolog-function} option
9562 is on by default if you optimize.
9566 Try to make the code as small as possible. At present, this just turns
9567 on the @option{-mep} and @option{-mprolog-function} options.
9571 Put static or global variables whose size is @var{n} bytes or less into
9572 the tiny data area that register @code{ep} points to. The tiny data
9573 area can hold up to 256 bytes in total (128 bytes for byte references).
9577 Put static or global variables whose size is @var{n} bytes or less into
9578 the small data area that register @code{gp} points to. The small data
9579 area can hold up to 64 kilobytes.
9583 Put static or global variables whose size is @var{n} bytes or less into
9584 the first 32 kilobytes of memory.
9588 Specify that the target processor is the V850.
9591 @opindex mbig-switch
9592 Generate code suitable for big switch tables. Use this option only if
9593 the assembler/linker complain about out of range branches within a switch
9598 This option will cause r2 and r5 to be used in the code generated by
9599 the compiler. This setting is the default.
9602 @opindex mno-app-regs
9603 This option will cause r2 and r5 to be treated as fixed registers.
9607 Specify that the target processor is the V850E1. The preprocessor
9608 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
9609 this option is used.
9613 Specify that the target processor is the V850E. The preprocessor
9614 constant @samp{__v850e__} will be defined if this option is used.
9616 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
9617 are defined then a default target processor will be chosen and the
9618 relevant @samp{__v850*__} preprocessor constant will be defined.
9620 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9621 defined, regardless of which processor variant is the target.
9623 @item -mdisable-callt
9624 @opindex mdisable-callt
9625 This option will suppress generation of the CALLT instruction for the
9626 v850e and v850e1 flavors of the v850 architecture. The default is
9627 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9632 @subsection ARC Options
9635 These options are defined for ARC implementations:
9640 Compile code for little endian mode. This is the default.
9644 Compile code for big endian mode.
9647 @opindex mmangle-cpu
9648 Prepend the name of the cpu to all public symbol names.
9649 In multiple-processor systems, there are many ARC variants with different
9650 instruction and register set characteristics. This flag prevents code
9651 compiled for one cpu to be linked with code compiled for another.
9652 No facility exists for handling variants that are ``almost identical''.
9653 This is an all or nothing option.
9655 @item -mcpu=@var{cpu}
9657 Compile code for ARC variant @var{cpu}.
9658 Which variants are supported depend on the configuration.
9659 All variants support @option{-mcpu=base}, this is the default.
9661 @item -mtext=@var{text-section}
9662 @itemx -mdata=@var{data-section}
9663 @itemx -mrodata=@var{readonly-data-section}
9667 Put functions, data, and readonly data in @var{text-section},
9668 @var{data-section}, and @var{readonly-data-section} respectively
9669 by default. This can be overridden with the @code{section} attribute.
9670 @xref{Variable Attributes}.
9675 @subsection NS32K Options
9676 @cindex NS32K options
9678 These are the @samp{-m} options defined for the 32000 series. The default
9679 values for these options depends on which style of 32000 was selected when
9680 the compiler was configured; the defaults for the most common choices are
9688 Generate output for a 32032. This is the default
9689 when the compiler is configured for 32032 and 32016 based systems.
9695 Generate output for a 32332. This is the default
9696 when the compiler is configured for 32332-based systems.
9702 Generate output for a 32532. This is the default
9703 when the compiler is configured for 32532-based systems.
9707 Generate output containing 32081 instructions for floating point.
9708 This is the default for all systems.
9712 Generate output containing 32381 instructions for floating point. This
9713 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9714 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9718 Try and generate multiply-add floating point instructions @code{polyF}
9719 and @code{dotF}. This option is only available if the @option{-m32381}
9720 option is in effect. Using these instructions requires changes to
9721 register allocation which generally has a negative impact on
9722 performance. This option should only be enabled when compiling code
9723 particularly likely to make heavy use of multiply-add instructions.
9726 @opindex mnomulti-add
9727 Do not try and generate multiply-add floating point instructions
9728 @code{polyF} and @code{dotF}. This is the default on all platforms.
9731 @opindex msoft-float
9732 Generate output containing library calls for floating point.
9733 @strong{Warning:} the requisite libraries may not be available.
9735 @item -mieee-compare
9736 @itemx -mno-ieee-compare
9737 @opindex mieee-compare
9738 @opindex mno-ieee-compare
9739 Control whether or not the compiler uses IEEE floating point
9740 comparisons. These handle correctly the case where the result of a
9741 comparison is unordered.
9742 @strong{Warning:} the requisite kernel support may not be available.
9745 @opindex mnobitfield
9746 Do not use the bit-field instructions. On some machines it is faster to
9747 use shifting and masking operations. This is the default for the pc532.
9751 Do use the bit-field instructions. This is the default for all platforms
9756 Use a different function-calling convention, in which functions
9757 that take a fixed number of arguments return pop their
9758 arguments on return with the @code{ret} instruction.
9760 This calling convention is incompatible with the one normally
9761 used on Unix, so you cannot use it if you need to call libraries
9762 compiled with the Unix compiler.
9764 Also, you must provide function prototypes for all functions that
9765 take variable numbers of arguments (including @code{printf});
9766 otherwise incorrect code will be generated for calls to those
9769 In addition, seriously incorrect code will result if you call a
9770 function with too many arguments. (Normally, extra arguments are
9771 harmlessly ignored.)
9773 This option takes its name from the 680x0 @code{rtd} instruction.
9778 Use a different function-calling convention where the first two arguments
9779 are passed in registers.
9781 This calling convention is incompatible with the one normally
9782 used on Unix, so you cannot use it if you need to call libraries
9783 compiled with the Unix compiler.
9786 @opindex mnoregparam
9787 Do not pass any arguments in registers. This is the default for all
9792 It is OK to use the sb as an index register which is always loaded with
9793 zero. This is the default for the pc532-netbsd target.
9797 The sb register is not available for use or has not been initialized to
9798 zero by the run time system. This is the default for all targets except
9799 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9800 @option{-fpic} is set.
9804 Many ns32000 series addressing modes use displacements of up to 512MB@.
9805 If an address is above 512MB then displacements from zero can not be used.
9806 This option causes code to be generated which can be loaded above 512MB@.
9807 This may be useful for operating systems or ROM code.
9811 Assume code will be loaded in the first 512MB of virtual address space.
9812 This is the default for all platforms.
9818 @subsection AVR Options
9821 These options are defined for AVR implementations:
9824 @item -mmcu=@var{mcu}
9826 Specify ATMEL AVR instruction set or MCU type.
9828 Instruction set avr1 is for the minimal AVR core, not supported by the C
9829 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9830 attiny11, attiny12, attiny15, attiny28).
9832 Instruction set avr2 (default) is for the classic AVR core with up to
9833 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9834 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9835 at90c8534, at90s8535).
9837 Instruction set avr3 is for the classic AVR core with up to 128K program
9838 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9840 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9841 memory space (MCU types: atmega8, atmega83, atmega85).
9843 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9844 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9845 atmega64, atmega128, at43usb355, at94k).
9849 Output instruction sizes to the asm file.
9851 @item -minit-stack=@var{N}
9852 @opindex minit-stack
9853 Specify the initial stack address, which may be a symbol or numeric value,
9854 @samp{__stack} is the default.
9856 @item -mno-interrupts
9857 @opindex mno-interrupts
9858 Generated code is not compatible with hardware interrupts.
9859 Code size will be smaller.
9861 @item -mcall-prologues
9862 @opindex mcall-prologues
9863 Functions prologues/epilogues expanded as call to appropriate
9864 subroutines. Code size will be smaller.
9866 @item -mno-tablejump
9867 @opindex mno-tablejump
9868 Do not generate tablejump insns which sometimes increase code size.
9871 @opindex mtiny-stack
9872 Change only the low 8 bits of the stack pointer.
9876 @subsection MCore Options
9877 @cindex MCore options
9879 These are the @samp{-m} options defined for the Motorola M*Core
9887 @opindex mno-hardlit
9888 Inline constants into the code stream if it can be done in two
9889 instructions or less.
9895 Use the divide instruction. (Enabled by default).
9897 @item -mrelax-immediate
9898 @itemx -mno-relax-immediate
9899 @opindex mrelax-immediate
9900 @opindex mno-relax-immediate
9901 Allow arbitrary sized immediates in bit operations.
9903 @item -mwide-bitfields
9904 @itemx -mno-wide-bitfields
9905 @opindex mwide-bitfields
9906 @opindex mno-wide-bitfields
9907 Always treat bit-fields as int-sized.
9909 @item -m4byte-functions
9910 @itemx -mno-4byte-functions
9911 @opindex m4byte-functions
9912 @opindex mno-4byte-functions
9913 Force all functions to be aligned to a four byte boundary.
9915 @item -mcallgraph-data
9916 @itemx -mno-callgraph-data
9917 @opindex mcallgraph-data
9918 @opindex mno-callgraph-data
9919 Emit callgraph information.
9922 @itemx -mno-slow-bytes
9923 @opindex mslow-bytes
9924 @opindex mno-slow-bytes
9925 Prefer word access when reading byte quantities.
9927 @item -mlittle-endian
9929 @opindex mlittle-endian
9930 @opindex mbig-endian
9931 Generate code for a little endian target.
9937 Generate code for the 210 processor.
9941 @subsection IA-64 Options
9942 @cindex IA-64 Options
9944 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9948 @opindex mbig-endian
9949 Generate code for a big endian target. This is the default for HP-UX@.
9951 @item -mlittle-endian
9952 @opindex mlittle-endian
9953 Generate code for a little endian target. This is the default for AIX5
9960 Generate (or don't) code for the GNU assembler. This is the default.
9961 @c Also, this is the default if the configure option @option{--with-gnu-as}
9968 Generate (or don't) code for the GNU linker. This is the default.
9969 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9974 Generate code that does not use a global pointer register. The result
9975 is not position independent code, and violates the IA-64 ABI@.
9977 @item -mvolatile-asm-stop
9978 @itemx -mno-volatile-asm-stop
9979 @opindex mvolatile-asm-stop
9980 @opindex mno-volatile-asm-stop
9981 Generate (or don't) a stop bit immediately before and after volatile asm
9986 Generate code that works around Itanium B step errata.
9988 @item -mregister-names
9989 @itemx -mno-register-names
9990 @opindex mregister-names
9991 @opindex mno-register-names
9992 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9993 the stacked registers. This may make assembler output more readable.
9999 Disable (or enable) optimizations that use the small data section. This may
10000 be useful for working around optimizer bugs.
10002 @item -mconstant-gp
10003 @opindex mconstant-gp
10004 Generate code that uses a single constant global pointer value. This is
10005 useful when compiling kernel code.
10009 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10010 This is useful when compiling firmware code.
10012 @item -minline-float-divide-min-latency
10013 @opindex minline-float-divide-min-latency
10014 Generate code for inline divides of floating point values
10015 using the minimum latency algorithm.
10017 @item -minline-float-divide-max-throughput
10018 @opindex minline-float-divide-max-throughput
10019 Generate code for inline divides of floating point values
10020 using the maximum throughput algorithm.
10022 @item -minline-int-divide-min-latency
10023 @opindex minline-int-divide-min-latency
10024 Generate code for inline divides of integer values
10025 using the minimum latency algorithm.
10027 @item -minline-int-divide-max-throughput
10028 @opindex minline-int-divide-max-throughput
10029 Generate code for inline divides of integer values
10030 using the maximum throughput algorithm.
10032 @item -mno-dwarf2-asm
10033 @itemx -mdwarf2-asm
10034 @opindex mno-dwarf2-asm
10035 @opindex mdwarf2-asm
10036 Don't (or do) generate assembler code for the DWARF2 line number debugging
10037 info. This may be useful when not using the GNU assembler.
10039 @item -mfixed-range=@var{register-range}
10040 @opindex mfixed-range
10041 Generate code treating the given register range as fixed registers.
10042 A fixed register is one that the register allocator can not use. This is
10043 useful when compiling kernel code. A register range is specified as
10044 two registers separated by a dash. Multiple register ranges can be
10045 specified separated by a comma.
10047 @item -mearly-stop-bits
10048 @itemx -mno-early-stop-bits
10049 @opindex mearly-stop-bits
10050 @opindex mno-early-stop-bits
10051 Allow stop bits to be placed earlier than immediately preceding the
10052 instruction that triggered the stop bit. This can improve instruction
10053 scheduling, but does not always do so.
10056 @node S/390 and zSeries Options
10057 @subsection S/390 and zSeries Options
10058 @cindex S/390 and zSeries Options
10060 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10064 @itemx -msoft-float
10065 @opindex mhard-float
10066 @opindex msoft-float
10067 Use (do not use) the hardware floating-point instructions and registers
10068 for floating-point operations. When @option{-msoft-float} is specified,
10069 functions in @file{libgcc.a} will be used to perform floating-point
10070 operations. When @option{-mhard-float} is specified, the compiler
10071 generates IEEE floating-point instructions. This is the default.
10074 @itemx -mno-backchain
10075 @opindex mbackchain
10076 @opindex mno-backchain
10077 Generate (or do not generate) code which maintains an explicit
10078 backchain within the stack frame that points to the caller's frame.
10079 This may be needed to allow debugging using tools that do not understand
10080 DWARF-2 call frame information. The default is not to generate the
10084 @itemx -mno-small-exec
10085 @opindex msmall-exec
10086 @opindex mno-small-exec
10087 Generate (or do not generate) code using the @code{bras} instruction
10088 to do subroutine calls.
10089 This only works reliably if the total executable size does not
10090 exceed 64k. The default is to use the @code{basr} instruction instead,
10091 which does not have this limitation.
10097 When @option{-m31} is specified, generate code compliant to the
10098 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
10099 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
10100 particular to generate 64-bit instructions. For the @samp{s390}
10101 targets, the default is @option{-m31}, while the @samp{s390x}
10102 targets default to @option{-m64}.
10108 When @option{-mzarch} is specified, generate code using the
10109 instructions available on z/Architecture.
10110 When @option{-mesa} is specified, generate code using the
10111 instructions available on ESA/390. Note that @option{-mesa} is
10112 not possible with @option{-m64}.
10113 When generating code compliant to the GNU/Linux for S/390 ABI,
10114 the default is @option{-mesa}. When generating code compliant
10115 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
10121 Generate (or do not generate) code using the @code{mvcle} instruction
10122 to perform block moves. When @option{-mno-mvcle} is specified,
10123 use a @code{mvc} loop instead. This is the default.
10129 Print (or do not print) additional debug information when compiling.
10130 The default is to not print debug information.
10132 @item -march=@var{cpu-type}
10134 Generate code that will run on @var{cpu-type}, which is the name of a system
10135 representing a certain processor type. Possible values for
10136 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10137 When generating code using the instructions available on z/Architecture,
10138 the default is @option{-march=z900}. Otherwise, the default is
10139 @option{-march=g5}.
10141 @item -mtune=@var{cpu-type}
10143 Tune to @var{cpu-type} everything applicable about the generated code,
10144 except for the ABI and the set of available instructions.
10145 The list of @var{cpu-type} values is the same as for @option{-march}.
10146 The default is the value used for @option{-march}.
10149 @itemx -mno-fused-madd
10150 @opindex mfused-madd
10151 @opindex mno-fused-madd
10152 Generate code that uses (does not use) the floating point multiply and
10153 accumulate instructions. These instructions are generated by default if
10154 hardware floating point is used.
10158 @subsection CRIS Options
10159 @cindex CRIS Options
10161 These options are defined specifically for the CRIS ports.
10164 @item -march=@var{architecture-type}
10165 @itemx -mcpu=@var{architecture-type}
10168 Generate code for the specified architecture. The choices for
10169 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10170 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
10171 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10174 @item -mtune=@var{architecture-type}
10176 Tune to @var{architecture-type} everything applicable about the generated
10177 code, except for the ABI and the set of available instructions. The
10178 choices for @var{architecture-type} are the same as for
10179 @option{-march=@var{architecture-type}}.
10181 @item -mmax-stack-frame=@var{n}
10182 @opindex mmax-stack-frame
10183 Warn when the stack frame of a function exceeds @var{n} bytes.
10185 @item -melinux-stacksize=@var{n}
10186 @opindex melinux-stacksize
10187 Only available with the @samp{cris-axis-aout} target. Arranges for
10188 indications in the program to the kernel loader that the stack of the
10189 program should be set to @var{n} bytes.
10195 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10196 @option{-march=v3} and @option{-march=v8} respectively.
10198 @item -mmul-bug-workaround
10199 @itemx -mno-mul-bug-workaround
10200 @opindex mmul-bug-workaround
10201 @opindex mno-mul-bug-workaround
10202 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
10203 models where it applies. This option is active by default.
10207 Enable CRIS-specific verbose debug-related information in the assembly
10208 code. This option also has the effect to turn off the @samp{#NO_APP}
10209 formatted-code indicator to the assembler at the beginning of the
10214 Do not use condition-code results from previous instruction; always emit
10215 compare and test instructions before use of condition codes.
10217 @item -mno-side-effects
10218 @opindex mno-side-effects
10219 Do not emit instructions with side-effects in addressing modes other than
10222 @item -mstack-align
10223 @itemx -mno-stack-align
10224 @itemx -mdata-align
10225 @itemx -mno-data-align
10226 @itemx -mconst-align
10227 @itemx -mno-const-align
10228 @opindex mstack-align
10229 @opindex mno-stack-align
10230 @opindex mdata-align
10231 @opindex mno-data-align
10232 @opindex mconst-align
10233 @opindex mno-const-align
10234 These options (no-options) arranges (eliminate arrangements) for the
10235 stack-frame, individual data and constants to be aligned for the maximum
10236 single data access size for the chosen CPU model. The default is to
10237 arrange for 32-bit alignment. ABI details such as structure layout are
10238 not affected by these options.
10246 Similar to the stack- data- and const-align options above, these options
10247 arrange for stack-frame, writable data and constants to all be 32-bit,
10248 16-bit or 8-bit aligned. The default is 32-bit alignment.
10250 @item -mno-prologue-epilogue
10251 @itemx -mprologue-epilogue
10252 @opindex mno-prologue-epilogue
10253 @opindex mprologue-epilogue
10254 With @option{-mno-prologue-epilogue}, the normal function prologue and
10255 epilogue that sets up the stack-frame are omitted and no return
10256 instructions or return sequences are generated in the code. Use this
10257 option only together with visual inspection of the compiled code: no
10258 warnings or errors are generated when call-saved registers must be saved,
10259 or storage for local variable needs to be allocated.
10263 @opindex mno-gotplt
10265 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10266 instruction sequences that load addresses for functions from the PLT part
10267 of the GOT rather than (traditional on other architectures) calls to the
10268 PLT. The default is @option{-mgotplt}.
10272 Legacy no-op option only recognized with the cris-axis-aout target.
10276 Legacy no-op option only recognized with the cris-axis-elf and
10277 cris-axis-linux-gnu targets.
10281 Only recognized with the cris-axis-aout target, where it selects a
10282 GNU/linux-like multilib, include files and instruction set for
10283 @option{-march=v8}.
10287 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10291 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
10292 to link with input-output functions from a simulator library. Code,
10293 initialized data and zero-initialized data are allocated consecutively.
10297 Like @option{-sim}, but pass linker options to locate initialized data at
10298 0x40000000 and zero-initialized data at 0x80000000.
10302 @subsection MMIX Options
10303 @cindex MMIX Options
10305 These options are defined for the MMIX:
10309 @itemx -mno-libfuncs
10311 @opindex mno-libfuncs
10312 Specify that intrinsic library functions are being compiled, passing all
10313 values in registers, no matter the size.
10316 @itemx -mno-epsilon
10318 @opindex mno-epsilon
10319 Generate floating-point comparison instructions that compare with respect
10320 to the @code{rE} epsilon register.
10322 @item -mabi=mmixware
10324 @opindex mabi-mmixware
10326 Generate code that passes function parameters and return values that (in
10327 the called function) are seen as registers @code{$0} and up, as opposed to
10328 the GNU ABI which uses global registers @code{$231} and up.
10330 @item -mzero-extend
10331 @itemx -mno-zero-extend
10332 @opindex mzero-extend
10333 @opindex mno-zero-extend
10334 When reading data from memory in sizes shorter than 64 bits, use (do not
10335 use) zero-extending load instructions by default, rather than
10336 sign-extending ones.
10339 @itemx -mno-knuthdiv
10341 @opindex mno-knuthdiv
10342 Make the result of a division yielding a remainder have the same sign as
10343 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10344 remainder follows the sign of the dividend. Both methods are
10345 arithmetically valid, the latter being almost exclusively used.
10347 @item -mtoplevel-symbols
10348 @itemx -mno-toplevel-symbols
10349 @opindex mtoplevel-symbols
10350 @opindex mno-toplevel-symbols
10351 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10352 code can be used with the @code{PREFIX} assembly directive.
10356 Generate an executable in the ELF format, rather than the default
10357 @samp{mmo} format used by the @command{mmix} simulator.
10359 @item -mbranch-predict
10360 @itemx -mno-branch-predict
10361 @opindex mbranch-predict
10362 @opindex mno-branch-predict
10363 Use (do not use) the probable-branch instructions, when static branch
10364 prediction indicates a probable branch.
10366 @item -mbase-addresses
10367 @itemx -mno-base-addresses
10368 @opindex mbase-addresses
10369 @opindex mno-base-addresses
10370 Generate (do not generate) code that uses @emph{base addresses}. Using a
10371 base address automatically generates a request (handled by the assembler
10372 and the linker) for a constant to be set up in a global register. The
10373 register is used for one or more base address requests within the range 0
10374 to 255 from the value held in the register. The generally leads to short
10375 and fast code, but the number of different data items that can be
10376 addressed is limited. This means that a program that uses lots of static
10377 data may require @option{-mno-base-addresses}.
10379 @item -msingle-exit
10380 @itemx -mno-single-exit
10381 @opindex msingle-exit
10382 @opindex mno-single-exit
10383 Force (do not force) generated code to have a single exit point in each
10387 @node PDP-11 Options
10388 @subsection PDP-11 Options
10389 @cindex PDP-11 Options
10391 These options are defined for the PDP-11:
10396 Use hardware FPP floating point. This is the default. (FIS floating
10397 point on the PDP-11/40 is not supported.)
10400 @opindex msoft-float
10401 Do not use hardware floating point.
10405 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10409 Return floating-point results in memory. This is the default.
10413 Generate code for a PDP-11/40.
10417 Generate code for a PDP-11/45. This is the default.
10421 Generate code for a PDP-11/10.
10423 @item -mbcopy-builtin
10424 @opindex bcopy-builtin
10425 Use inline @code{movstrhi} patterns for copying memory. This is the
10430 Do not use inline @code{movstrhi} patterns for copying memory.
10436 Use 16-bit @code{int}. This is the default.
10442 Use 32-bit @code{int}.
10445 @itemx -mno-float32
10447 @opindex mno-float32
10448 Use 64-bit @code{float}. This is the default.
10451 @itemx -mno-float64
10453 @opindex mno-float64
10454 Use 32-bit @code{float}.
10458 Use @code{abshi2} pattern. This is the default.
10462 Do not use @code{abshi2} pattern.
10464 @item -mbranch-expensive
10465 @opindex mbranch-expensive
10466 Pretend that branches are expensive. This is for experimenting with
10467 code generation only.
10469 @item -mbranch-cheap
10470 @opindex mbranch-cheap
10471 Do not pretend that branches are expensive. This is the default.
10475 Generate code for a system with split I&D.
10479 Generate code for a system without split I&D. This is the default.
10483 Use Unix assembler syntax. This is the default when configured for
10484 @samp{pdp11-*-bsd}.
10488 Use DEC assembler syntax. This is the default when configured for any
10489 PDP-11 target other than @samp{pdp11-*-bsd}.
10492 @node Xstormy16 Options
10493 @subsection Xstormy16 Options
10494 @cindex Xstormy16 Options
10496 These options are defined for Xstormy16:
10501 Choose startup files and linker script suitable for the simulator.
10505 @subsection FRV Options
10506 @cindex FRV Options
10512 Only use the first 32 general purpose registers.
10517 Use all 64 general purpose registers.
10522 Use only the first 32 floating point registers.
10527 Use all 64 floating point registers
10530 @opindex mhard-float
10532 Use hardware instructions for floating point operations.
10535 @opindex msoft-float
10537 Use library routines for floating point operations.
10542 Dynamically allocate condition code registers.
10547 Do not try to dynamically allocate condition code registers, only
10548 use @code{icc0} and @code{fcc0}.
10553 Change ABI to use double word insns.
10558 Do not use double word instructions.
10563 Use floating point double instructions.
10566 @opindex mno-double
10568 Do not use floating point double instructions.
10573 Use media instructions.
10578 Do not use media instructions.
10583 Use multiply and add/subtract instructions.
10586 @opindex mno-muladd
10588 Do not use multiply and add/subtract instructions.
10590 @item -mlibrary-pic
10591 @opindex mlibrary-pic
10593 Generate position-independent EABI code.
10598 Use only the first four media accumulator registers.
10603 Use all eight media accumulator registers.
10608 Pack VLIW instructions.
10613 Do not pack VLIW instructions.
10616 @opindex mno-eflags
10618 Do not mark ABI switches in e_flags.
10621 @opindex mcond-move
10623 Enable the use of conditional-move instructions (default).
10625 This switch is mainly for debugging the compiler and will likely be removed
10626 in a future version.
10628 @item -mno-cond-move
10629 @opindex mno-cond-move
10631 Disable the use of conditional-move instructions.
10633 This switch is mainly for debugging the compiler and will likely be removed
10634 in a future version.
10639 Enable the use of conditional set instructions (default).
10641 This switch is mainly for debugging the compiler and will likely be removed
10642 in a future version.
10647 Disable the use of conditional set instructions.
10649 This switch is mainly for debugging the compiler and will likely be removed
10650 in a future version.
10653 @opindex mcond-exec
10655 Enable the use of conditional execution (default).
10657 This switch is mainly for debugging the compiler and will likely be removed
10658 in a future version.
10660 @item -mno-cond-exec
10661 @opindex mno-cond-exec
10663 Disable the use of conditional execution.
10665 This switch is mainly for debugging the compiler and will likely be removed
10666 in a future version.
10668 @item -mvliw-branch
10669 @opindex mvliw-branch
10671 Run a pass to pack branches into VLIW instructions (default).
10673 This switch is mainly for debugging the compiler and will likely be removed
10674 in a future version.
10676 @item -mno-vliw-branch
10677 @opindex mno-vliw-branch
10679 Do not run a pass to pack branches into VLIW instructions.
10681 This switch is mainly for debugging the compiler and will likely be removed
10682 in a future version.
10684 @item -mmulti-cond-exec
10685 @opindex mmulti-cond-exec
10687 Enable optimization of @code{&&} and @code{||} in conditional execution
10690 This switch is mainly for debugging the compiler and will likely be removed
10691 in a future version.
10693 @item -mno-multi-cond-exec
10694 @opindex mno-multi-cond-exec
10696 Disable optimization of @code{&&} and @code{||} in conditional execution.
10698 This switch is mainly for debugging the compiler and will likely be removed
10699 in a future version.
10701 @item -mnested-cond-exec
10702 @opindex mnested-cond-exec
10704 Enable nested conditional execution optimizations (default).
10706 This switch is mainly for debugging the compiler and will likely be removed
10707 in a future version.
10709 @item -mno-nested-cond-exec
10710 @opindex mno-nested-cond-exec
10712 Disable nested conditional execution optimizations.
10714 This switch is mainly for debugging the compiler and will likely be removed
10715 in a future version.
10717 @item -mtomcat-stats
10718 @opindex mtomcat-stats
10720 Cause gas to print out tomcat statistics.
10722 @item -mcpu=@var{cpu}
10725 Select the processor type for which to generate code. Possible values are
10726 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
10731 @node Xtensa Options
10732 @subsection Xtensa Options
10733 @cindex Xtensa Options
10735 These options are supported for Xtensa targets:
10739 @itemx -mno-const16
10741 @opindex mno-const16
10742 Enable or disable use of @code{CONST16} instructions for loading
10743 constant values. The @code{CONST16} instruction is currently not a
10744 standard option from Tensilica. When enabled, @code{CONST16}
10745 instructions are always used in place of the standard @code{L32R}
10746 instructions. The use of @code{CONST16} is enabled by default only if
10747 the @code{L32R} instruction is not available.
10750 @itemx -mno-fused-madd
10751 @opindex mfused-madd
10752 @opindex mno-fused-madd
10753 Enable or disable use of fused multiply/add and multiply/subtract
10754 instructions in the floating-point option. This has no effect if the
10755 floating-point option is not also enabled. Disabling fused multiply/add
10756 and multiply/subtract instructions forces the compiler to use separate
10757 instructions for the multiply and add/subtract operations. This may be
10758 desirable in some cases where strict IEEE 754-compliant results are
10759 required: the fused multiply add/subtract instructions do not round the
10760 intermediate result, thereby producing results with @emph{more} bits of
10761 precision than specified by the IEEE standard. Disabling fused multiply
10762 add/subtract instructions also ensures that the program output is not
10763 sensitive to the compiler's ability to combine multiply and add/subtract
10766 @item -mtext-section-literals
10767 @itemx -mno-text-section-literals
10768 @opindex mtext-section-literals
10769 @opindex mno-text-section-literals
10770 Control the treatment of literal pools. The default is
10771 @option{-mno-text-section-literals}, which places literals in a separate
10772 section in the output file. This allows the literal pool to be placed
10773 in a data RAM/ROM, and it also allows the linker to combine literal
10774 pools from separate object files to remove redundant literals and
10775 improve code size. With @option{-mtext-section-literals}, the literals
10776 are interspersed in the text section in order to keep them as close as
10777 possible to their references. This may be necessary for large assembly
10780 @item -mtarget-align
10781 @itemx -mno-target-align
10782 @opindex mtarget-align
10783 @opindex mno-target-align
10784 When this option is enabled, GCC instructs the assembler to
10785 automatically align instructions to reduce branch penalties at the
10786 expense of some code density. The assembler attempts to widen density
10787 instructions to align branch targets and the instructions following call
10788 instructions. If there are not enough preceding safe density
10789 instructions to align a target, no widening will be performed. The
10790 default is @option{-mtarget-align}. These options do not affect the
10791 treatment of auto-aligned instructions like @code{LOOP}, which the
10792 assembler will always align, either by widening density instructions or
10793 by inserting no-op instructions.
10796 @itemx -mno-longcalls
10797 @opindex mlongcalls
10798 @opindex mno-longcalls
10799 When this option is enabled, GCC instructs the assembler to translate
10800 direct calls to indirect calls unless it can determine that the target
10801 of a direct call is in the range allowed by the call instruction. This
10802 translation typically occurs for calls to functions in other source
10803 files. Specifically, the assembler translates a direct @code{CALL}
10804 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
10805 The default is @option{-mno-longcalls}. This option should be used in
10806 programs where the call target can potentially be out of range. This
10807 option is implemented in the assembler, not the compiler, so the
10808 assembly code generated by GCC will still show direct call
10809 instructions---look at the disassembled object code to see the actual
10810 instructions. Note that the assembler will use an indirect call for
10811 every cross-file call, not just those that really will be out of range.
10814 @node Code Gen Options
10815 @section Options for Code Generation Conventions
10816 @cindex code generation conventions
10817 @cindex options, code generation
10818 @cindex run-time options
10820 These machine-independent options control the interface conventions
10821 used in code generation.
10823 Most of them have both positive and negative forms; the negative form
10824 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
10825 one of the forms is listed---the one which is not the default. You
10826 can figure out the other form by either removing @samp{no-} or adding
10830 @item -fbounds-check
10831 @opindex fbounds-check
10832 For front-ends that support it, generate additional code to check that
10833 indices used to access arrays are within the declared range. This is
10834 currently only supported by the Java and Fortran 77 front-ends, where
10835 this option defaults to true and false respectively.
10839 This option generates traps for signed overflow on addition, subtraction,
10840 multiplication operations.
10844 This option instructs the compiler to assume that signed arithmetic
10845 overflow of addition, subtraction and multiplication wraps around
10846 using twos-complement representation. This flag enables some optimizations
10847 and disables other. This option is enabled by default for the Java
10848 front-end, as required by the Java language specification.
10851 @opindex fexceptions
10852 Enable exception handling. Generates extra code needed to propagate
10853 exceptions. For some targets, this implies GCC will generate frame
10854 unwind information for all functions, which can produce significant data
10855 size overhead, although it does not affect execution. If you do not
10856 specify this option, GCC will enable it by default for languages like
10857 C++ which normally require exception handling, and disable it for
10858 languages like C that do not normally require it. However, you may need
10859 to enable this option when compiling C code that needs to interoperate
10860 properly with exception handlers written in C++. You may also wish to
10861 disable this option if you are compiling older C++ programs that don't
10862 use exception handling.
10864 @item -fnon-call-exceptions
10865 @opindex fnon-call-exceptions
10866 Generate code that allows trapping instructions to throw exceptions.
10867 Note that this requires platform-specific runtime support that does
10868 not exist everywhere. Moreover, it only allows @emph{trapping}
10869 instructions to throw exceptions, i.e.@: memory references or floating
10870 point instructions. It does not allow exceptions to be thrown from
10871 arbitrary signal handlers such as @code{SIGALRM}.
10873 @item -funwind-tables
10874 @opindex funwind-tables
10875 Similar to @option{-fexceptions}, except that it will just generate any needed
10876 static data, but will not affect the generated code in any other way.
10877 You will normally not enable this option; instead, a language processor
10878 that needs this handling would enable it on your behalf.
10880 @item -fasynchronous-unwind-tables
10881 @opindex funwind-tables
10882 Generate unwind table in dwarf2 format, if supported by target machine. The
10883 table is exact at each instruction boundary, so it can be used for stack
10884 unwinding from asynchronous events (such as debugger or garbage collector).
10886 @item -fpcc-struct-return
10887 @opindex fpcc-struct-return
10888 Return ``short'' @code{struct} and @code{union} values in memory like
10889 longer ones, rather than in registers. This convention is less
10890 efficient, but it has the advantage of allowing intercallability between
10891 GCC-compiled files and files compiled with other compilers, particularly
10892 the Portable C Compiler (pcc).
10894 The precise convention for returning structures in memory depends
10895 on the target configuration macros.
10897 Short structures and unions are those whose size and alignment match
10898 that of some integer type.
10900 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
10901 switch is not binary compatible with code compiled with the
10902 @option{-freg-struct-return} switch.
10903 Use it to conform to a non-default application binary interface.
10905 @item -freg-struct-return
10906 @opindex freg-struct-return
10907 Return @code{struct} and @code{union} values in registers when possible.
10908 This is more efficient for small structures than
10909 @option{-fpcc-struct-return}.
10911 If you specify neither @option{-fpcc-struct-return} nor
10912 @option{-freg-struct-return}, GCC defaults to whichever convention is
10913 standard for the target. If there is no standard convention, GCC
10914 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
10915 the principal compiler. In those cases, we can choose the standard, and
10916 we chose the more efficient register return alternative.
10918 @strong{Warning:} code compiled with the @option{-freg-struct-return}
10919 switch is not binary compatible with code compiled with the
10920 @option{-fpcc-struct-return} switch.
10921 Use it to conform to a non-default application binary interface.
10923 @item -fshort-enums
10924 @opindex fshort-enums
10925 Allocate to an @code{enum} type only as many bytes as it needs for the
10926 declared range of possible values. Specifically, the @code{enum} type
10927 will be equivalent to the smallest integer type which has enough room.
10929 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
10930 code that is not binary compatible with code generated without that switch.
10931 Use it to conform to a non-default application binary interface.
10933 @item -fshort-double
10934 @opindex fshort-double
10935 Use the same size for @code{double} as for @code{float}.
10937 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
10938 code that is not binary compatible with code generated without that switch.
10939 Use it to conform to a non-default application binary interface.
10941 @item -fshort-wchar
10942 @opindex fshort-wchar
10943 Override the underlying type for @samp{wchar_t} to be @samp{short
10944 unsigned int} instead of the default for the target. This option is
10945 useful for building programs to run under WINE@.
10947 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
10948 code that is not binary compatible with code generated without that switch.
10949 Use it to conform to a non-default application binary interface.
10951 @item -fshared-data
10952 @opindex fshared-data
10953 Requests that the data and non-@code{const} variables of this
10954 compilation be shared data rather than private data. The distinction
10955 makes sense only on certain operating systems, where shared data is
10956 shared between processes running the same program, while private data
10957 exists in one copy per process.
10960 @opindex fno-common
10961 In C, allocate even uninitialized global variables in the data section of the
10962 object file, rather than generating them as common blocks. This has the
10963 effect that if the same variable is declared (without @code{extern}) in
10964 two different compilations, you will get an error when you link them.
10965 The only reason this might be useful is if you wish to verify that the
10966 program will work on other systems which always work this way.
10970 Ignore the @samp{#ident} directive.
10972 @item -finhibit-size-directive
10973 @opindex finhibit-size-directive
10974 Don't output a @code{.size} assembler directive, or anything else that
10975 would cause trouble if the function is split in the middle, and the
10976 two halves are placed at locations far apart in memory. This option is
10977 used when compiling @file{crtstuff.c}; you should not need to use it
10980 @item -fverbose-asm
10981 @opindex fverbose-asm
10982 Put extra commentary information in the generated assembly code to
10983 make it more readable. This option is generally only of use to those
10984 who actually need to read the generated assembly code (perhaps while
10985 debugging the compiler itself).
10987 @option{-fno-verbose-asm}, the default, causes the
10988 extra information to be omitted and is useful when comparing two assembler
10993 @cindex global offset table
10995 Generate position-independent code (PIC) suitable for use in a shared
10996 library, if supported for the target machine. Such code accesses all
10997 constant addresses through a global offset table (GOT)@. The dynamic
10998 loader resolves the GOT entries when the program starts (the dynamic
10999 loader is not part of GCC; it is part of the operating system). If
11000 the GOT size for the linked executable exceeds a machine-specific
11001 maximum size, you get an error message from the linker indicating that
11002 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11003 instead. (These maximums are 8k on the SPARC and 32k
11004 on the m68k and RS/6000. The 386 has no such limit.)
11006 Position-independent code requires special support, and therefore works
11007 only on certain machines. For the 386, GCC supports PIC for System V
11008 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11009 position-independent.
11013 If supported for the target machine, emit position-independent code,
11014 suitable for dynamic linking and avoiding any limit on the size of the
11015 global offset table. This option makes a difference on the m68k
11018 Position-independent code requires special support, and therefore works
11019 only on certain machines.
11025 These options are similar to @option{-fpic} and @option{-fPIC}, but
11026 generated position independent code can be only linked into executables.
11027 Usually these options are used when @option{-pie} GCC option will be
11028 used during linking.
11030 @item -ffixed-@var{reg}
11032 Treat the register named @var{reg} as a fixed register; generated code
11033 should never refer to it (except perhaps as a stack pointer, frame
11034 pointer or in some other fixed role).
11036 @var{reg} must be the name of a register. The register names accepted
11037 are machine-specific and are defined in the @code{REGISTER_NAMES}
11038 macro in the machine description macro file.
11040 This flag does not have a negative form, because it specifies a
11043 @item -fcall-used-@var{reg}
11044 @opindex fcall-used
11045 Treat the register named @var{reg} as an allocable register that is
11046 clobbered by function calls. It may be allocated for temporaries or
11047 variables that do not live across a call. Functions compiled this way
11048 will not save and restore the register @var{reg}.
11050 It is an error to used this flag with the frame pointer or stack pointer.
11051 Use of this flag for other registers that have fixed pervasive roles in
11052 the machine's execution model will produce disastrous results.
11054 This flag does not have a negative form, because it specifies a
11057 @item -fcall-saved-@var{reg}
11058 @opindex fcall-saved
11059 Treat the register named @var{reg} as an allocable register saved by
11060 functions. It may be allocated even for temporaries or variables that
11061 live across a call. Functions compiled this way will save and restore
11062 the register @var{reg} if they use it.
11064 It is an error to used this flag with the frame pointer or stack pointer.
11065 Use of this flag for other registers that have fixed pervasive roles in
11066 the machine's execution model will produce disastrous results.
11068 A different sort of disaster will result from the use of this flag for
11069 a register in which function values may be returned.
11071 This flag does not have a negative form, because it specifies a
11074 @item -fpack-struct
11075 @opindex fpack-struct
11076 Pack all structure members together without holes.
11078 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11079 code that is not binary compatible with code generated without that switch.
11080 Additionally, it makes the code suboptimal.
11081 Use it to conform to a non-default application binary interface.
11083 @item -finstrument-functions
11084 @opindex finstrument-functions
11085 Generate instrumentation calls for entry and exit to functions. Just
11086 after function entry and just before function exit, the following
11087 profiling functions will be called with the address of the current
11088 function and its call site. (On some platforms,
11089 @code{__builtin_return_address} does not work beyond the current
11090 function, so the call site information may not be available to the
11091 profiling functions otherwise.)
11094 void __cyg_profile_func_enter (void *this_fn,
11096 void __cyg_profile_func_exit (void *this_fn,
11100 The first argument is the address of the start of the current function,
11101 which may be looked up exactly in the symbol table.
11103 This instrumentation is also done for functions expanded inline in other
11104 functions. The profiling calls will indicate where, conceptually, the
11105 inline function is entered and exited. This means that addressable
11106 versions of such functions must be available. If all your uses of a
11107 function are expanded inline, this may mean an additional expansion of
11108 code size. If you use @samp{extern inline} in your C code, an
11109 addressable version of such functions must be provided. (This is
11110 normally the case anyways, but if you get lucky and the optimizer always
11111 expands the functions inline, you might have gotten away without
11112 providing static copies.)
11114 A function may be given the attribute @code{no_instrument_function}, in
11115 which case this instrumentation will not be done. This can be used, for
11116 example, for the profiling functions listed above, high-priority
11117 interrupt routines, and any functions from which the profiling functions
11118 cannot safely be called (perhaps signal handlers, if the profiling
11119 routines generate output or allocate memory).
11121 @item -fstack-check
11122 @opindex fstack-check
11123 Generate code to verify that you do not go beyond the boundary of the
11124 stack. You should specify this flag if you are running in an
11125 environment with multiple threads, but only rarely need to specify it in
11126 a single-threaded environment since stack overflow is automatically
11127 detected on nearly all systems if there is only one stack.
11129 Note that this switch does not actually cause checking to be done; the
11130 operating system must do that. The switch causes generation of code
11131 to ensure that the operating system sees the stack being extended.
11133 @item -fstack-limit-register=@var{reg}
11134 @itemx -fstack-limit-symbol=@var{sym}
11135 @itemx -fno-stack-limit
11136 @opindex fstack-limit-register
11137 @opindex fstack-limit-symbol
11138 @opindex fno-stack-limit
11139 Generate code to ensure that the stack does not grow beyond a certain value,
11140 either the value of a register or the address of a symbol. If the stack
11141 would grow beyond the value, a signal is raised. For most targets,
11142 the signal is raised before the stack overruns the boundary, so
11143 it is possible to catch the signal without taking special precautions.
11145 For instance, if the stack starts at absolute address @samp{0x80000000}
11146 and grows downwards, you can use the flags
11147 @option{-fstack-limit-symbol=__stack_limit} and
11148 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11149 of 128KB@. Note that this may only work with the GNU linker.
11151 @cindex aliasing of parameters
11152 @cindex parameters, aliased
11153 @item -fargument-alias
11154 @itemx -fargument-noalias
11155 @itemx -fargument-noalias-global
11156 @opindex fargument-alias
11157 @opindex fargument-noalias
11158 @opindex fargument-noalias-global
11159 Specify the possible relationships among parameters and between
11160 parameters and global data.
11162 @option{-fargument-alias} specifies that arguments (parameters) may
11163 alias each other and may alias global storage.@*
11164 @option{-fargument-noalias} specifies that arguments do not alias
11165 each other, but may alias global storage.@*
11166 @option{-fargument-noalias-global} specifies that arguments do not
11167 alias each other and do not alias global storage.
11169 Each language will automatically use whatever option is required by
11170 the language standard. You should not need to use these options yourself.
11172 @item -fleading-underscore
11173 @opindex fleading-underscore
11174 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11175 change the way C symbols are represented in the object file. One use
11176 is to help link with legacy assembly code.
11178 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11179 generate code that is not binary compatible with code generated without that
11180 switch. Use it to conform to a non-default application binary interface.
11181 Not all targets provide complete support for this switch.
11183 @item -ftls-model=@var{model}
11184 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11185 The @var{model} argument should be one of @code{global-dynamic},
11186 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11188 The default without @option{-fpic} is @code{initial-exec}; with
11189 @option{-fpic} the default is @code{global-dynamic}.
11194 @node Environment Variables
11195 @section Environment Variables Affecting GCC
11196 @cindex environment variables
11198 @c man begin ENVIRONMENT
11199 This section describes several environment variables that affect how GCC
11200 operates. Some of them work by specifying directories or prefixes to use
11201 when searching for various kinds of files. Some are used to specify other
11202 aspects of the compilation environment.
11204 Note that you can also specify places to search using options such as
11205 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11206 take precedence over places specified using environment variables, which
11207 in turn take precedence over those specified by the configuration of GCC@.
11208 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11209 GNU Compiler Collection (GCC) Internals}.
11214 @c @itemx LC_COLLATE
11216 @c @itemx LC_MONETARY
11217 @c @itemx LC_NUMERIC
11222 @c @findex LC_COLLATE
11223 @findex LC_MESSAGES
11224 @c @findex LC_MONETARY
11225 @c @findex LC_NUMERIC
11229 These environment variables control the way that GCC uses
11230 localization information that allow GCC to work with different
11231 national conventions. GCC inspects the locale categories
11232 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11233 so. These locale categories can be set to any value supported by your
11234 installation. A typical value is @samp{en_UK} for English in the United
11237 The @env{LC_CTYPE} environment variable specifies character
11238 classification. GCC uses it to determine the character boundaries in
11239 a string; this is needed for some multibyte encodings that contain quote
11240 and escape characters that would otherwise be interpreted as a string
11243 The @env{LC_MESSAGES} environment variable specifies the language to
11244 use in diagnostic messages.
11246 If the @env{LC_ALL} environment variable is set, it overrides the value
11247 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11248 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11249 environment variable. If none of these variables are set, GCC
11250 defaults to traditional C English behavior.
11254 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11255 files. GCC uses temporary files to hold the output of one stage of
11256 compilation which is to be used as input to the next stage: for example,
11257 the output of the preprocessor, which is the input to the compiler
11260 @item GCC_EXEC_PREFIX
11261 @findex GCC_EXEC_PREFIX
11262 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11263 names of the subprograms executed by the compiler. No slash is added
11264 when this prefix is combined with the name of a subprogram, but you can
11265 specify a prefix that ends with a slash if you wish.
11267 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11268 an appropriate prefix to use based on the pathname it was invoked with.
11270 If GCC cannot find the subprogram using the specified prefix, it
11271 tries looking in the usual places for the subprogram.
11273 The default value of @env{GCC_EXEC_PREFIX} is
11274 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
11275 of @code{prefix} when you ran the @file{configure} script.
11277 Other prefixes specified with @option{-B} take precedence over this prefix.
11279 This prefix is also used for finding files such as @file{crt0.o} that are
11282 In addition, the prefix is used in an unusual way in finding the
11283 directories to search for header files. For each of the standard
11284 directories whose name normally begins with @samp{/usr/local/lib/gcc}
11285 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11286 replacing that beginning with the specified prefix to produce an
11287 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11288 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11289 These alternate directories are searched first; the standard directories
11292 @item COMPILER_PATH
11293 @findex COMPILER_PATH
11294 The value of @env{COMPILER_PATH} is a colon-separated list of
11295 directories, much like @env{PATH}. GCC tries the directories thus
11296 specified when searching for subprograms, if it can't find the
11297 subprograms using @env{GCC_EXEC_PREFIX}.
11300 @findex LIBRARY_PATH
11301 The value of @env{LIBRARY_PATH} is a colon-separated list of
11302 directories, much like @env{PATH}. When configured as a native compiler,
11303 GCC tries the directories thus specified when searching for special
11304 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11305 using GCC also uses these directories when searching for ordinary
11306 libraries for the @option{-l} option (but directories specified with
11307 @option{-L} come first).
11311 @cindex locale definition
11312 This variable is used to pass locale information to the compiler. One way in
11313 which this information is used is to determine the character set to be used
11314 when character literals, string literals and comments are parsed in C and C++.
11315 When the compiler is configured to allow multibyte characters,
11316 the following values for @env{LANG} are recognized:
11320 Recognize JIS characters.
11322 Recognize SJIS characters.
11324 Recognize EUCJP characters.
11327 If @env{LANG} is not defined, or if it has some other value, then the
11328 compiler will use mblen and mbtowc as defined by the default locale to
11329 recognize and translate multibyte characters.
11333 Some additional environments variables affect the behavior of the
11336 @include cppenv.texi
11340 @node Precompiled Headers
11341 @section Using Precompiled Headers
11342 @cindex precompiled headers
11343 @cindex speed of compilation
11345 Often large projects have many header files that are included in every
11346 source file. The time the compiler takes to process these header files
11347 over and over again can account for nearly all of the time required to
11348 build the project. To make builds faster, GCC allows users to
11349 `precompile' a header file; then, if builds can use the precompiled
11350 header file they will be much faster.
11352 @strong{Caution:} There are a few known situations where GCC will
11353 crash when trying to use a precompiled header. If you have trouble
11354 with a precompiled header, you should remove the precompiled header
11355 and compile without it. In addition, please use GCC's on-line
11356 defect-tracking system to report any problems you encounter with
11357 precompiled headers. @xref{Bugs}.
11359 To create a precompiled header file, simply compile it as you would any
11360 other file, if necessary using the @option{-x} option to make the driver
11361 treat it as a C or C++ header file. You will probably want to use a
11362 tool like @command{make} to keep the precompiled header up-to-date when
11363 the headers it contains change.
11365 A precompiled header file will be searched for when @code{#include} is
11366 seen in the compilation. As it searches for the included file
11367 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
11368 compiler looks for a precompiled header in each directory just before it
11369 looks for the include file in that directory. The name searched for is
11370 the name specified in the @code{#include} with @samp{.gch} appended. If
11371 the precompiled header file can't be used, it is ignored.
11373 For instance, if you have @code{#include "all.h"}, and you have
11374 @file{all.h.gch} in the same directory as @file{all.h}, then the
11375 precompiled header file will be used if possible, and the original
11376 header will be used otherwise.
11378 Alternatively, you might decide to put the precompiled header file in a
11379 directory and use @option{-I} to ensure that directory is searched
11380 before (or instead of) the directory containing the original header.
11381 Then, if you want to check that the precompiled header file is always
11382 used, you can put a file of the same name as the original header in this
11383 directory containing an @code{#error} command.
11385 This also works with @option{-include}. So yet another way to use
11386 precompiled headers, good for projects not designed with precompiled
11387 header files in mind, is to simply take most of the header files used by
11388 a project, include them from another header file, precompile that header
11389 file, and @option{-include} the precompiled header. If the header files
11390 have guards against multiple inclusion, they will be skipped because
11391 they've already been included (in the precompiled header).
11393 If you need to precompile the same header file for different
11394 languages, targets, or compiler options, you can instead make a
11395 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11396 header in the directory, perhaps using @option{-o}. It doesn't matter
11397 what you call the files in the directory, every precompiled header in
11398 the directory will be considered. The first precompiled header
11399 encountered in the directory that is valid for this compilation will
11400 be used; they're searched in no particular order.
11402 There are many other possibilities, limited only by your imagination,
11403 good sense, and the constraints of your build system.
11405 A precompiled header file can be used only when these conditions apply:
11409 Only one precompiled header can be used in a particular compilation.
11412 A precompiled header can't be used once the first C token is seen. You
11413 can have preprocessor directives before a precompiled header; you can
11414 even include a precompiled header from inside another header, so long as
11415 there are no C tokens before the @code{#include}.
11418 The precompiled header file must be produced for the same language as
11419 the current compilation. You can't use a C precompiled header for a C++
11423 The precompiled header file must be produced by the same compiler
11424 version and configuration as the current compilation is using.
11425 The easiest way to guarantee this is to use the same compiler binary
11426 for creating and using precompiled headers.
11429 Any macros defined before the precompiled header is included must
11430 either be defined in the same way as when the precompiled header was
11431 generated, or must not affect the precompiled header, which usually
11432 means that the they don't appear in the precompiled header at all.
11434 The @option{-D} option is one way to define a macro before a
11435 precompiled header is included; using a @code{#define} can also do it.
11436 There are also some options that define macros implicitly, like
11437 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
11440 @item If debugging information is output when using the precompiled
11441 header, using @option{-g} or similar, the same kind of debugging information
11442 must have been output when building the precompiled header. However,
11443 a precompiled header built using @option{-g} can be used in a compilation
11444 when no debugging information is being output.
11446 @item The same @option{-m} options must generally be used when building
11447 and using the precompiled header. @xref{Submodel Options},
11448 for any cases where this rule is relaxed.
11450 @item Each of the following options must be the same when building and using
11451 the precompiled header:
11453 @gccoptlist{-fexceptions -funit-at-a-time}
11456 Some other command-line options starting with @option{-f},
11457 @option{-p}, or @option{-O} must be defined in the same way as when
11458 the precompiled header was generated. At present, it's not clear
11459 which options are safe to change and which are not; the safest choice
11460 is to use exactly the same options when generating and using the
11461 precompiled header. The following are known to be safe:
11463 @gccoptlist{-pedantic-errors}
11467 For all of these except the last, the compiler will automatically
11468 ignore the precompiled header if the conditions aren't met. If you
11469 find an option combination that doesn't work and doesn't cause the
11470 precompiled header to be ignored, please consider filing a bug report,
11473 @node Running Protoize
11474 @section Running Protoize
11476 The program @code{protoize} is an optional part of GCC@. You can use
11477 it to add prototypes to a program, thus converting the program to ISO
11478 C in one respect. The companion program @code{unprotoize} does the
11479 reverse: it removes argument types from any prototypes that are found.
11481 When you run these programs, you must specify a set of source files as
11482 command line arguments. The conversion programs start out by compiling
11483 these files to see what functions they define. The information gathered
11484 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11486 After scanning comes actual conversion. The specified files are all
11487 eligible to be converted; any files they include (whether sources or
11488 just headers) are eligible as well.
11490 But not all the eligible files are converted. By default,
11491 @code{protoize} and @code{unprotoize} convert only source and header
11492 files in the current directory. You can specify additional directories
11493 whose files should be converted with the @option{-d @var{directory}}
11494 option. You can also specify particular files to exclude with the
11495 @option{-x @var{file}} option. A file is converted if it is eligible, its
11496 directory name matches one of the specified directory names, and its
11497 name within the directory has not been excluded.
11499 Basic conversion with @code{protoize} consists of rewriting most
11500 function definitions and function declarations to specify the types of
11501 the arguments. The only ones not rewritten are those for varargs
11504 @code{protoize} optionally inserts prototype declarations at the
11505 beginning of the source file, to make them available for any calls that
11506 precede the function's definition. Or it can insert prototype
11507 declarations with block scope in the blocks where undeclared functions
11510 Basic conversion with @code{unprotoize} consists of rewriting most
11511 function declarations to remove any argument types, and rewriting
11512 function definitions to the old-style pre-ISO form.
11514 Both conversion programs print a warning for any function declaration or
11515 definition that they can't convert. You can suppress these warnings
11518 The output from @code{protoize} or @code{unprotoize} replaces the
11519 original source file. The original file is renamed to a name ending
11520 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11521 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11522 for DOS) file already exists, then the source file is simply discarded.
11524 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11525 scan the program and collect information about the functions it uses.
11526 So neither of these programs will work until GCC is installed.
11528 Here is a table of the options you can use with @code{protoize} and
11529 @code{unprotoize}. Each option works with both programs unless
11533 @item -B @var{directory}
11534 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11535 usual directory (normally @file{/usr/local/lib}). This file contains
11536 prototype information about standard system functions. This option
11537 applies only to @code{protoize}.
11539 @item -c @var{compilation-options}
11540 Use @var{compilation-options} as the options when running @command{gcc} to
11541 produce the @samp{.X} files. The special option @option{-aux-info} is
11542 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11544 Note that the compilation options must be given as a single argument to
11545 @code{protoize} or @code{unprotoize}. If you want to specify several
11546 @command{gcc} options, you must quote the entire set of compilation options
11547 to make them a single word in the shell.
11549 There are certain @command{gcc} arguments that you cannot use, because they
11550 would produce the wrong kind of output. These include @option{-g},
11551 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11552 the @var{compilation-options}, they are ignored.
11555 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11556 systems) instead of @samp{.c}. This is convenient if you are converting
11557 a C program to C++. This option applies only to @code{protoize}.
11560 Add explicit global declarations. This means inserting explicit
11561 declarations at the beginning of each source file for each function
11562 that is called in the file and was not declared. These declarations
11563 precede the first function definition that contains a call to an
11564 undeclared function. This option applies only to @code{protoize}.
11566 @item -i @var{string}
11567 Indent old-style parameter declarations with the string @var{string}.
11568 This option applies only to @code{protoize}.
11570 @code{unprotoize} converts prototyped function definitions to old-style
11571 function definitions, where the arguments are declared between the
11572 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11573 uses five spaces as the indentation. If you want to indent with just
11574 one space instead, use @option{-i " "}.
11577 Keep the @samp{.X} files. Normally, they are deleted after conversion
11581 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11582 a prototype declaration for each function in each block which calls the
11583 function without any declaration. This option applies only to
11587 Make no real changes. This mode just prints information about the conversions
11588 that would have been done without @option{-n}.
11591 Make no @samp{.save} files. The original files are simply deleted.
11592 Use this option with caution.
11594 @item -p @var{program}
11595 Use the program @var{program} as the compiler. Normally, the name
11596 @file{gcc} is used.
11599 Work quietly. Most warnings are suppressed.
11602 Print the version number, just like @option{-v} for @command{gcc}.
11605 If you need special compiler options to compile one of your program's
11606 source files, then you should generate that file's @samp{.X} file
11607 specially, by running @command{gcc} on that source file with the
11608 appropriate options and the option @option{-aux-info}. Then run
11609 @code{protoize} on the entire set of files. @code{protoize} will use
11610 the existing @samp{.X} file because it is newer than the source file.
11614 gcc -Dfoo=bar file1.c -aux-info file1.X
11619 You need to include the special files along with the rest in the
11620 @code{protoize} command, even though their @samp{.X} files already
11621 exist, because otherwise they won't get converted.
11623 @xref{Protoize Caveats}, for more information on how to use
11624 @code{protoize} successfully.