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 -Wfatal-errors -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 -mgp32 -mgp64 -mfp32 -mfp64 @gol
472 -mhard-float -msoft-float -msingle-float -mdouble-float @gol
473 -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 @gol
486 -mfp-exceptions -mno-fp-exceptions}
488 @emph{i386 and x86-64 Options}
489 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
490 -mfpmath=@var{unit} @gol
491 -masm=@var{dialect} -mno-fancy-math-387 @gol
492 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
493 -mno-wide-multiply -mrtd -malign-double @gol
494 -mpreferred-stack-boundary=@var{num} @gol
495 -mmmx -msse -msse2 -msse3 -m3dnow @gol
496 -mthreads -mno-align-stringops -minline-all-stringops @gol
497 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
498 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
499 -mno-red-zone -mno-tls-direct-seg-refs @gol
500 -mcmodel=@var{code-model} @gol
504 @gccoptlist{-march=@var{architecture-type} @gol
505 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
506 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
507 -mjump-in-delay -mlinker-opt -mlong-calls @gol
508 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
509 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
510 -mno-jump-in-delay -mno-long-load-store @gol
511 -mno-portable-runtime -mno-soft-float @gol
512 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
513 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
514 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
515 -nolibdld -static -threads}
517 @emph{DEC Alpha Options}
518 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
519 -mieee -mieee-with-inexact -mieee-conformant @gol
520 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
521 -mtrap-precision=@var{mode} -mbuild-constants @gol
522 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
523 -mbwx -mmax -mfix -mcix @gol
524 -mfloat-vax -mfloat-ieee @gol
525 -mexplicit-relocs -msmall-data -mlarge-data @gol
526 -msmall-text -mlarge-text @gol
527 -mmemory-latency=@var{time}}
529 @emph{DEC Alpha/VMS Options}
530 @gccoptlist{-mvms-return-codes}
532 @emph{H8/300 Options}
533 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
536 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
537 -m4-nofpu -m4-single-only -m4-single -m4 @gol
538 -m5-64media -m5-64media-nofpu @gol
539 -m5-32media -m5-32media-nofpu @gol
540 -m5-compact -m5-compact-nofpu @gol
541 -mb -ml -mdalign -mrelax @gol
542 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
543 -mieee -misize -mpadstruct -mspace @gol
544 -mprefergot -musermode}
546 @emph{System V Options}
547 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
550 @gccoptlist{-EB -EL @gol
551 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
552 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
554 @emph{TMS320C3x/C4x Options}
555 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
556 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
557 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
558 -mparallel-insns -mparallel-mpy -mpreserve-float}
561 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
562 -mprolog-function -mno-prolog-function -mspace @gol
563 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
564 -mapp-regs -mno-app-regs @gol
565 -mdisable-callt -mno-disable-callt @gol
571 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
572 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
573 -mregparam -mnoregparam -msb -mnosb @gol
574 -mbitfield -mnobitfield -mhimem -mnohimem}
577 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
578 -mcall-prologues -mno-tablejump -mtiny-stack}
581 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
582 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
583 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
584 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
585 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
588 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
589 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
590 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
591 -mno-base-addresses -msingle-exit -mno-single-exit}
594 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
595 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
596 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
597 -minline-float-divide-max-throughput @gol
598 -minline-int-divide-min-latency @gol
599 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
600 -mfixed-range=@var{register-range}}
602 @emph{S/390 and zSeries Options}
603 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
604 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
605 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
606 -m64 -m31 -mdebug -mno-debug -mesa -mzarch -mfused-madd -mno-fused-madd}
609 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
610 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
611 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
612 -mstack-align -mdata-align -mconst-align @gol
613 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
614 -melf -maout -melinux -mlinux -sim -sim2 @gol
615 -mmul-bug-workaround -mno-mul-bug-workaround}
617 @emph{PDP-11 Options}
618 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
619 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
620 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
621 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
622 -mbranch-expensive -mbranch-cheap @gol
623 -msplit -mno-split -munix-asm -mdec-asm}
625 @emph{Xstormy16 Options}
628 @emph{Xtensa Options}
629 @gccoptlist{-mconst16 -mno-const16 @gol
630 -mfused-madd -mno-fused-madd @gol
631 -mtext-section-literals -mno-text-section-literals @gol
632 -mtarget-align -mno-target-align @gol
633 -mlongcalls -mno-longcalls}
636 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
637 -mhard-float -msoft-float @gol
638 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
639 -mdouble -mno-double @gol
640 -mmedia -mno-media -mmuladd -mno-muladd @gol
641 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic -mlinked-fp @gol
642 -mlibrary-pic -macc-4 -macc-8 @gol
643 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
644 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
645 -mvliw-branch -mno-vliw-branch @gol
646 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
647 -mno-nested-cond-exec -mtomcat-stats @gol
650 @item Code Generation Options
651 @xref{Code Gen Options,,Options for Code Generation Conventions}.
652 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
653 -ffixed-@var{reg} -fexceptions @gol
654 -fnon-call-exceptions -funwind-tables @gol
655 -fasynchronous-unwind-tables @gol
656 -finhibit-size-directive -finstrument-functions @gol
657 -fno-common -fno-ident @gol
658 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
659 -freg-struct-return -fshared-data -fshort-enums @gol
660 -fshort-double -fshort-wchar @gol
661 -fverbose-asm -fpack-struct -fstack-check @gol
662 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
663 -fargument-alias -fargument-noalias @gol
664 -fargument-noalias-global -fleading-underscore @gol
665 -ftls-model=@var{model} @gol
666 -ftrapv -fwrapv -fbounds-check}
670 * Overall Options:: Controlling the kind of output:
671 an executable, object files, assembler files,
672 or preprocessed source.
673 * C Dialect Options:: Controlling the variant of C language compiled.
674 * C++ Dialect Options:: Variations on C++.
675 * Objective-C Dialect Options:: Variations on Objective-C.
676 * Language Independent Options:: Controlling how diagnostics should be
678 * Warning Options:: How picky should the compiler be?
679 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
680 * Optimize Options:: How much optimization?
681 * Preprocessor Options:: Controlling header files and macro definitions.
682 Also, getting dependency information for Make.
683 * Assembler Options:: Passing options to the assembler.
684 * Link Options:: Specifying libraries and so on.
685 * Directory Options:: Where to find header files and libraries.
686 Where to find the compiler executable files.
687 * Spec Files:: How to pass switches to sub-processes.
688 * Target Options:: Running a cross-compiler, or an old version of GCC.
691 @node Overall Options
692 @section Options Controlling the Kind of Output
694 Compilation can involve up to four stages: preprocessing, compilation
695 proper, assembly and linking, always in that order. GCC is capable of
696 preprocessing and compiling several files either into several
697 assembler input files, or into one assembler input file; then each
698 assembler input file produces an object file, and linking combines all
699 the object files (those newly compiled, and those specified as input)
700 into an executable file.
702 @cindex file name suffix
703 For any given input file, the file name suffix determines what kind of
708 C source code which must be preprocessed.
711 C source code which should not be preprocessed.
714 C++ source code which should not be preprocessed.
717 Objective-C source code. Note that you must link with the library
718 @file{libobjc.a} to make an Objective-C program work.
721 Objective-C source code which should not be preprocessed.
724 C or C++ header file to be turned into a precompiled header.
728 @itemx @var{file}.cxx
729 @itemx @var{file}.cpp
730 @itemx @var{file}.CPP
731 @itemx @var{file}.c++
733 C++ source code which must be preprocessed. Note that in @samp{.cxx},
734 the last two letters must both be literally @samp{x}. Likewise,
735 @samp{.C} refers to a literal capital C@.
739 C++ header file to be turned into a precompiled header.
742 @itemx @var{file}.for
743 @itemx @var{file}.FOR
744 Fortran source code which should not be preprocessed.
747 @itemx @var{file}.fpp
748 @itemx @var{file}.FPP
749 Fortran source code which must be preprocessed (with the traditional
753 Fortran source code which must be preprocessed with a RATFOR
754 preprocessor (not included with GCC)@.
756 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
757 Using and Porting GNU Fortran}, for more details of the handling of
760 @c FIXME: Descriptions of Java file types.
767 Ada source code file which contains a library unit declaration (a
768 declaration of a package, subprogram, or generic, or a generic
769 instantiation), or a library unit renaming declaration (a package,
770 generic, or subprogram renaming declaration). Such files are also
773 @itemx @var{file}.adb
774 Ada source code file containing a library unit body (a subprogram or
775 package body). Such files are also called @dfn{bodies}.
777 @c GCC also knows about some suffixes for languages not yet included:
786 Assembler code which must be preprocessed.
789 An object file to be fed straight into linking.
790 Any file name with no recognized suffix is treated this way.
794 You can specify the input language explicitly with the @option{-x} option:
797 @item -x @var{language}
798 Specify explicitly the @var{language} for the following input files
799 (rather than letting the compiler choose a default based on the file
800 name suffix). This option applies to all following input files until
801 the next @option{-x} option. Possible values for @var{language} are:
803 c c-header cpp-output
804 c++ c++-header c++-cpp-output
805 objective-c objective-c-header objc-cpp-output
806 assembler assembler-with-cpp
808 f77 f77-cpp-input ratfor
814 Turn off any specification of a language, so that subsequent files are
815 handled according to their file name suffixes (as they are if @option{-x}
816 has not been used at all).
818 @item -pass-exit-codes
819 @opindex pass-exit-codes
820 Normally the @command{gcc} program will exit with the code of 1 if any
821 phase of the compiler returns a non-success return code. If you specify
822 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
823 numerically highest error produced by any phase that returned an error
827 If you only want some of the stages of compilation, you can use
828 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
829 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
830 @command{gcc} is to stop. Note that some combinations (for example,
831 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
836 Compile or assemble the source files, but do not link. The linking
837 stage simply is not done. The ultimate output is in the form of an
838 object file for each source file.
840 By default, the object file name for a source file is made by replacing
841 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
843 Unrecognized input files, not requiring compilation or assembly, are
848 Stop after the stage of compilation proper; do not assemble. The output
849 is in the form of an assembler code file for each non-assembler input
852 By default, the assembler file name for a source file is made by
853 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
855 Input files that don't require compilation are ignored.
859 Stop after the preprocessing stage; do not run the compiler proper. The
860 output is in the form of preprocessed source code, which is sent to the
863 Input files which don't require preprocessing are ignored.
865 @cindex output file option
868 Place output in file @var{file}. This applies regardless to whatever
869 sort of output is being produced, whether it be an executable file,
870 an object file, an assembler file or preprocessed C code.
872 If @option{-o} is not specified, the default is to put an executable
873 file in @file{a.out}, the object file for
874 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
875 assembler file in @file{@var{source}.s}, a precompiled header file in
876 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
881 Print (on standard error output) the commands executed to run the stages
882 of compilation. Also print the version number of the compiler driver
883 program and of the preprocessor and the compiler proper.
887 Like @option{-v} except the commands are not executed and all command
888 arguments are quoted. This is useful for shell scripts to capture the
889 driver-generated command lines.
893 Use pipes rather than temporary files for communication between the
894 various stages of compilation. This fails to work on some systems where
895 the assembler is unable to read from a pipe; but the GNU assembler has
900 If you are compiling multiple source files, this option tells the driver
901 to pass all the source files to the compiler at once (for those
902 languages for which the compiler can handle this). This will allow
903 intermodule analysis (IMA) to be performed by the compiler. Currently the only
904 language for which this is supported is C. If you pass source files for
905 multiple languages to the driver, using this option, the driver will invoke
906 the compiler(s) that support IMA once each, passing each compiler all the
907 source files appropriate for it. For those languages that do not support
908 IMA this option will be ignored, and the compiler will be invoked once for
909 each source file in that language. If you use this option in conjunction
910 with -save-temps, the compiler will generate multiple pre-processed files
911 (one for each source file), but only one (combined) .o or .s file.
915 Print (on the standard output) a description of the command line options
916 understood by @command{gcc}. If the @option{-v} option is also specified
917 then @option{--help} will also be passed on to the various processes
918 invoked by @command{gcc}, so that they can display the command line options
919 they accept. If the @option{-Wextra} option is also specified then command
920 line options which have no documentation associated with them will also
925 Print (on the standard output) a description of target specific command
926 line options for each tool.
930 Display the version number and copyrights of the invoked GCC.
934 @section Compiling C++ Programs
936 @cindex suffixes for C++ source
937 @cindex C++ source file suffixes
938 C++ source files conventionally use one of the suffixes @samp{.C},
939 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
940 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
941 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
942 files with these names and compiles them as C++ programs even if you
943 call the compiler the same way as for compiling C programs (usually
944 with the name @command{gcc}).
948 However, C++ programs often require class libraries as well as a
949 compiler that understands the C++ language---and under some
950 circumstances, you might want to compile programs or header files from
951 standard input, or otherwise without a suffix that flags them as C++
952 programs. You might also like to precompile a C header file with a
953 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
954 program that calls GCC with the default language set to C++, and
955 automatically specifies linking against the C++ library. On many
956 systems, @command{g++} is also installed with the name @command{c++}.
958 @cindex invoking @command{g++}
959 When you compile C++ programs, you may specify many of the same
960 command-line options that you use for compiling programs in any
961 language; or command-line options meaningful for C and related
962 languages; or options that are meaningful only for C++ programs.
963 @xref{C Dialect Options,,Options Controlling C Dialect}, for
964 explanations of options for languages related to C@.
965 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
966 explanations of options that are meaningful only for C++ programs.
968 @node C Dialect Options
969 @section Options Controlling C Dialect
970 @cindex dialect options
971 @cindex language dialect options
972 @cindex options, dialect
974 The following options control the dialect of C (or languages derived
975 from C, such as C++ and Objective-C) that the compiler accepts:
982 In C mode, support all ISO C90 programs. In C++ mode,
983 remove GNU extensions that conflict with ISO C++.
985 This turns off certain features of GCC that are incompatible with ISO
986 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
987 such as the @code{asm} and @code{typeof} keywords, and
988 predefined macros such as @code{unix} and @code{vax} that identify the
989 type of system you are using. It also enables the undesirable and
990 rarely used ISO trigraph feature. For the C compiler,
991 it disables recognition of C++ style @samp{//} comments as well as
992 the @code{inline} keyword.
994 The alternate keywords @code{__asm__}, @code{__extension__},
995 @code{__inline__} and @code{__typeof__} continue to work despite
996 @option{-ansi}. You would not want to use them in an ISO C program, of
997 course, but it is useful to put them in header files that might be included
998 in compilations done with @option{-ansi}. Alternate predefined macros
999 such as @code{__unix__} and @code{__vax__} are also available, with or
1000 without @option{-ansi}.
1002 The @option{-ansi} option does not cause non-ISO programs to be
1003 rejected gratuitously. For that, @option{-pedantic} is required in
1004 addition to @option{-ansi}. @xref{Warning Options}.
1006 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1007 option is used. Some header files may notice this macro and refrain
1008 from declaring certain functions or defining certain macros that the
1009 ISO standard doesn't call for; this is to avoid interfering with any
1010 programs that might use these names for other things.
1012 Functions which would normally be built in but do not have semantics
1013 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1014 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1015 built-in functions provided by GCC}, for details of the functions
1020 Determine the language standard. This option is currently only
1021 supported when compiling C or C++. A value for this option must be
1022 provided; possible values are
1027 ISO C90 (same as @option{-ansi}).
1029 @item iso9899:199409
1030 ISO C90 as modified in amendment 1.
1036 ISO C99. Note that this standard is not yet fully supported; see
1037 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1038 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1041 Default, ISO C90 plus GNU extensions (including some C99 features).
1045 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1046 this will become the default. The name @samp{gnu9x} is deprecated.
1049 The 1998 ISO C++ standard plus amendments.
1052 The same as @option{-std=c++98} plus GNU extensions. This is the
1053 default for C++ code.
1056 Even when this option is not specified, you can still use some of the
1057 features of newer standards in so far as they do not conflict with
1058 previous C standards. For example, you may use @code{__restrict__} even
1059 when @option{-std=c99} is not specified.
1061 The @option{-std} options specifying some version of ISO C have the same
1062 effects as @option{-ansi}, except that features that were not in ISO C90
1063 but are in the specified version (for example, @samp{//} comments and
1064 the @code{inline} keyword in ISO C99) are not disabled.
1066 @xref{Standards,,Language Standards Supported by GCC}, for details of
1067 these standard versions.
1069 @item -aux-info @var{filename}
1071 Output to the given filename prototyped declarations for all functions
1072 declared and/or defined in a translation unit, including those in header
1073 files. This option is silently ignored in any language other than C@.
1075 Besides declarations, the file indicates, in comments, the origin of
1076 each declaration (source file and line), whether the declaration was
1077 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1078 @samp{O} for old, respectively, in the first character after the line
1079 number and the colon), and whether it came from a declaration or a
1080 definition (@samp{C} or @samp{F}, respectively, in the following
1081 character). In the case of function definitions, a K&R-style list of
1082 arguments followed by their declarations is also provided, inside
1083 comments, after the declaration.
1087 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1088 keyword, so that code can use these words as identifiers. You can use
1089 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1090 instead. @option{-ansi} implies @option{-fno-asm}.
1092 In C++, this switch only affects the @code{typeof} keyword, since
1093 @code{asm} and @code{inline} are standard keywords. You may want to
1094 use the @option{-fno-gnu-keywords} flag instead, which has the same
1095 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1096 switch only affects the @code{asm} and @code{typeof} keywords, since
1097 @code{inline} is a standard keyword in ISO C99.
1100 @itemx -fno-builtin-@var{function}
1101 @opindex fno-builtin
1102 @cindex built-in functions
1103 Don't recognize built-in functions that do not begin with
1104 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1105 functions provided by GCC}, for details of the functions affected,
1106 including those which are not built-in functions when @option{-ansi} or
1107 @option{-std} options for strict ISO C conformance are used because they
1108 do not have an ISO standard meaning.
1110 GCC normally generates special code to handle certain built-in functions
1111 more efficiently; for instance, calls to @code{alloca} may become single
1112 instructions that adjust the stack directly, and calls to @code{memcpy}
1113 may become inline copy loops. The resulting code is often both smaller
1114 and faster, but since the function calls no longer appear as such, you
1115 cannot set a breakpoint on those calls, nor can you change the behavior
1116 of the functions by linking with a different library.
1118 With the @option{-fno-builtin-@var{function}} option
1119 only the built-in function @var{function} is
1120 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1121 function is named this is not built-in in this version of GCC, this
1122 option is ignored. There is no corresponding
1123 @option{-fbuiltin-@var{function}} option; if you wish to enable
1124 built-in functions selectively when using @option{-fno-builtin} or
1125 @option{-ffreestanding}, you may define macros such as:
1128 #define abs(n) __builtin_abs ((n))
1129 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1134 @cindex hosted environment
1136 Assert that compilation takes place in a hosted environment. This implies
1137 @option{-fbuiltin}. A hosted environment is one in which the
1138 entire standard library is available, and in which @code{main} has a return
1139 type of @code{int}. Examples are nearly everything except a kernel.
1140 This is equivalent to @option{-fno-freestanding}.
1142 @item -ffreestanding
1143 @opindex ffreestanding
1144 @cindex hosted environment
1146 Assert that compilation takes place in a freestanding environment. This
1147 implies @option{-fno-builtin}. A freestanding environment
1148 is one in which the standard library may not exist, and program startup may
1149 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1150 This is equivalent to @option{-fno-hosted}.
1152 @xref{Standards,,Language Standards Supported by GCC}, for details of
1153 freestanding and hosted environments.
1155 @item -fms-extensions
1156 @opindex fms-extensions
1157 Accept some non-standard constructs used in Microsoft header files.
1161 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1162 options for strict ISO C conformance) implies @option{-trigraphs}.
1164 @item -no-integrated-cpp
1165 @opindex no-integrated-cpp
1166 Performs a compilation in two passes: preprocessing and compiling. This
1167 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1168 @option{-B} option. The user supplied compilation step can then add in
1169 an additional preprocessing step after normal preprocessing but before
1170 compiling. The default is to use the integrated cpp (internal cpp)
1172 The semantics of this option will change if "cc1", "cc1plus", and
1173 "cc1obj" are merged.
1175 @cindex traditional C language
1176 @cindex C language, traditional
1178 @itemx -traditional-cpp
1179 @opindex traditional-cpp
1180 @opindex traditional
1181 Formerly, these options caused GCC to attempt to emulate a pre-standard
1182 C compiler. They are now only supported with the @option{-E} switch.
1183 The preprocessor continues to support a pre-standard mode. See the GNU
1184 CPP manual for details.
1186 @item -fcond-mismatch
1187 @opindex fcond-mismatch
1188 Allow conditional expressions with mismatched types in the second and
1189 third arguments. The value of such an expression is void. This option
1190 is not supported for C++.
1192 @item -funsigned-char
1193 @opindex funsigned-char
1194 Let the type @code{char} be unsigned, like @code{unsigned char}.
1196 Each kind of machine has a default for what @code{char} should
1197 be. It is either like @code{unsigned char} by default or like
1198 @code{signed char} by default.
1200 Ideally, a portable program should always use @code{signed char} or
1201 @code{unsigned char} when it depends on the signedness of an object.
1202 But many programs have been written to use plain @code{char} and
1203 expect it to be signed, or expect it to be unsigned, depending on the
1204 machines they were written for. This option, and its inverse, let you
1205 make such a program work with the opposite default.
1207 The type @code{char} is always a distinct type from each of
1208 @code{signed char} or @code{unsigned char}, even though its behavior
1209 is always just like one of those two.
1212 @opindex fsigned-char
1213 Let the type @code{char} be signed, like @code{signed char}.
1215 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1216 the negative form of @option{-funsigned-char}. Likewise, the option
1217 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1219 @item -fsigned-bitfields
1220 @itemx -funsigned-bitfields
1221 @itemx -fno-signed-bitfields
1222 @itemx -fno-unsigned-bitfields
1223 @opindex fsigned-bitfields
1224 @opindex funsigned-bitfields
1225 @opindex fno-signed-bitfields
1226 @opindex fno-unsigned-bitfields
1227 These options control whether a bit-field is signed or unsigned, when the
1228 declaration does not use either @code{signed} or @code{unsigned}. By
1229 default, such a bit-field is signed, because this is consistent: the
1230 basic integer types such as @code{int} are signed types.
1233 @node C++ Dialect Options
1234 @section Options Controlling C++ Dialect
1236 @cindex compiler options, C++
1237 @cindex C++ options, command line
1238 @cindex options, C++
1239 This section describes the command-line options that are only meaningful
1240 for C++ programs; but you can also use most of the GNU compiler options
1241 regardless of what language your program is in. For example, you
1242 might compile a file @code{firstClass.C} like this:
1245 g++ -g -frepo -O -c firstClass.C
1249 In this example, only @option{-frepo} is an option meant
1250 only for C++ programs; you can use the other options with any
1251 language supported by GCC@.
1253 Here is a list of options that are @emph{only} for compiling C++ programs:
1257 @item -fabi-version=@var{n}
1258 @opindex fabi-version
1259 Use version @var{n} of the C++ ABI. Version 2 is the version of the
1260 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1261 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1262 the version that conforms most closely to the C++ ABI specification.
1263 Therefore, the ABI obtained using version 0 will change as ABI bugs
1266 The default is version 2.
1268 @item -fno-access-control
1269 @opindex fno-access-control
1270 Turn off all access checking. This switch is mainly useful for working
1271 around bugs in the access control code.
1275 Check that the pointer returned by @code{operator new} is non-null
1276 before attempting to modify the storage allocated. This check is
1277 normally unnecessary because the C++ standard specifies that
1278 @code{operator new} will only return @code{0} if it is declared
1279 @samp{throw()}, in which case the compiler will always check the
1280 return value even without this option. In all other cases, when
1281 @code{operator new} has a non-empty exception specification, memory
1282 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1283 @samp{new (nothrow)}.
1285 @item -fconserve-space
1286 @opindex fconserve-space
1287 Put uninitialized or runtime-initialized global variables into the
1288 common segment, as C does. This saves space in the executable at the
1289 cost of not diagnosing duplicate definitions. If you compile with this
1290 flag and your program mysteriously crashes after @code{main()} has
1291 completed, you may have an object that is being destroyed twice because
1292 two definitions were merged.
1294 This option is no longer useful on most targets, now that support has
1295 been added for putting variables into BSS without making them common.
1297 @item -fno-const-strings
1298 @opindex fno-const-strings
1299 Give string constants type @code{char *} instead of type @code{const
1300 char *}. By default, G++ uses type @code{const char *} as required by
1301 the standard. Even if you use @option{-fno-const-strings}, you cannot
1302 actually modify the value of a string constant.
1304 This option might be removed in a future release of G++. For maximum
1305 portability, you should structure your code so that it works with
1306 string constants that have type @code{const char *}.
1308 @item -fno-elide-constructors
1309 @opindex fno-elide-constructors
1310 The C++ standard allows an implementation to omit creating a temporary
1311 which is only used to initialize another object of the same type.
1312 Specifying this option disables that optimization, and forces G++ to
1313 call the copy constructor in all cases.
1315 @item -fno-enforce-eh-specs
1316 @opindex fno-enforce-eh-specs
1317 Don't check for violation of exception specifications at runtime. This
1318 option violates the C++ standard, but may be useful for reducing code
1319 size in production builds, much like defining @samp{NDEBUG}. The compiler
1320 will still optimize based on the exception specifications.
1323 @itemx -fno-for-scope
1325 @opindex fno-for-scope
1326 If @option{-ffor-scope} is specified, the scope of variables declared in
1327 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1328 as specified by the C++ standard.
1329 If @option{-fno-for-scope} is specified, the scope of variables declared in
1330 a @i{for-init-statement} extends to the end of the enclosing scope,
1331 as was the case in old versions of G++, and other (traditional)
1332 implementations of C++.
1334 The default if neither flag is given to follow the standard,
1335 but to allow and give a warning for old-style code that would
1336 otherwise be invalid, or have different behavior.
1338 @item -fno-gnu-keywords
1339 @opindex fno-gnu-keywords
1340 Do not recognize @code{typeof} as a keyword, so that code can use this
1341 word as an identifier. You can use the keyword @code{__typeof__} instead.
1342 @option{-ansi} implies @option{-fno-gnu-keywords}.
1344 @item -fno-implicit-templates
1345 @opindex fno-implicit-templates
1346 Never emit code for non-inline templates which are instantiated
1347 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1348 @xref{Template Instantiation}, for more information.
1350 @item -fno-implicit-inline-templates
1351 @opindex fno-implicit-inline-templates
1352 Don't emit code for implicit instantiations of inline templates, either.
1353 The default is to handle inlines differently so that compiles with and
1354 without optimization will need the same set of explicit instantiations.
1356 @item -fno-implement-inlines
1357 @opindex fno-implement-inlines
1358 To save space, do not emit out-of-line copies of inline functions
1359 controlled by @samp{#pragma implementation}. This will cause linker
1360 errors if these functions are not inlined everywhere they are called.
1362 @item -fms-extensions
1363 @opindex fms-extensions
1364 Disable pedantic warnings about constructs used in MFC, such as implicit
1365 int and getting a pointer to member function via non-standard syntax.
1367 @item -fno-nonansi-builtins
1368 @opindex fno-nonansi-builtins
1369 Disable built-in declarations of functions that are not mandated by
1370 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1371 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1373 @item -fno-operator-names
1374 @opindex fno-operator-names
1375 Do not treat the operator name keywords @code{and}, @code{bitand},
1376 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1377 synonyms as keywords.
1379 @item -fno-optional-diags
1380 @opindex fno-optional-diags
1381 Disable diagnostics that the standard says a compiler does not need to
1382 issue. Currently, the only such diagnostic issued by G++ is the one for
1383 a name having multiple meanings within a class.
1386 @opindex fpermissive
1387 Downgrade some diagnostics about nonconformant code from errors to
1388 warnings. Thus, using @option{-fpermissive} will allow some
1389 nonconforming code to compile.
1393 Enable automatic template instantiation at link time. This option also
1394 implies @option{-fno-implicit-templates}. @xref{Template
1395 Instantiation}, for more information.
1399 Disable generation of information about every class with virtual
1400 functions for use by the C++ runtime type identification features
1401 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1402 of the language, you can save some space by using this flag. Note that
1403 exception handling uses the same information, but it will generate it as
1408 Emit statistics about front-end processing at the end of the compilation.
1409 This information is generally only useful to the G++ development team.
1411 @item -ftemplate-depth-@var{n}
1412 @opindex ftemplate-depth
1413 Set the maximum instantiation depth for template classes to @var{n}.
1414 A limit on the template instantiation depth is needed to detect
1415 endless recursions during template class instantiation. ANSI/ISO C++
1416 conforming programs must not rely on a maximum depth greater than 17.
1418 @item -fuse-cxa-atexit
1419 @opindex fuse-cxa-atexit
1420 Register destructors for objects with static storage duration with the
1421 @code{__cxa_atexit} function rather than the @code{atexit} function.
1422 This option is required for fully standards-compliant handling of static
1423 destructors, but will only work if your C library supports
1424 @code{__cxa_atexit}.
1428 Do not use weak symbol support, even if it is provided by the linker.
1429 By default, G++ will use weak symbols if they are available. This
1430 option exists only for testing, and should not be used by end-users;
1431 it will result in inferior code and has no benefits. This option may
1432 be removed in a future release of G++.
1436 Do not search for header files in the standard directories specific to
1437 C++, but do still search the other standard directories. (This option
1438 is used when building the C++ library.)
1441 In addition, these optimization, warning, and code generation options
1442 have meanings only for C++ programs:
1445 @item -fno-default-inline
1446 @opindex fno-default-inline
1447 Do not assume @samp{inline} for functions defined inside a class scope.
1448 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1449 functions will have linkage like inline functions; they just won't be
1452 @item -Wabi @r{(C++ only)}
1454 Warn when G++ generates code that is probably not compatible with the
1455 vendor-neutral C++ ABI. Although an effort has been made to warn about
1456 all such cases, there are probably some cases that are not warned about,
1457 even though G++ is generating incompatible code. There may also be
1458 cases where warnings are emitted even though the code that is generated
1461 You should rewrite your code to avoid these warnings if you are
1462 concerned about the fact that code generated by G++ may not be binary
1463 compatible with code generated by other compilers.
1465 The known incompatibilities at this point include:
1470 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1471 pack data into the same byte as a base class. For example:
1474 struct A @{ virtual void f(); int f1 : 1; @};
1475 struct B : public A @{ int f2 : 1; @};
1479 In this case, G++ will place @code{B::f2} into the same byte
1480 as@code{A::f1}; other compilers will not. You can avoid this problem
1481 by explicitly padding @code{A} so that its size is a multiple of the
1482 byte size on your platform; that will cause G++ and other compilers to
1483 layout @code{B} identically.
1486 Incorrect handling of tail-padding for virtual bases. G++ does not use
1487 tail padding when laying out virtual bases. For example:
1490 struct A @{ virtual void f(); char c1; @};
1491 struct B @{ B(); char c2; @};
1492 struct C : public A, public virtual B @{@};
1496 In this case, G++ will not place @code{B} into the tail-padding for
1497 @code{A}; other compilers will. You can avoid this problem by
1498 explicitly padding @code{A} so that its size is a multiple of its
1499 alignment (ignoring virtual base classes); that will cause G++ and other
1500 compilers to layout @code{C} identically.
1503 Incorrect handling of bit-fields with declared widths greater than that
1504 of their underlying types, when the bit-fields appear in a union. For
1508 union U @{ int i : 4096; @};
1512 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1513 union too small by the number of bits in an @code{int}.
1516 Empty classes can be placed at incorrect offsets. For example:
1526 struct C : public B, public A @{@};
1530 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1531 it should be placed at offset zero. G++ mistakenly believes that the
1532 @code{A} data member of @code{B} is already at offset zero.
1535 Names of template functions whose types involve @code{typename} or
1536 template template parameters can be mangled incorrectly.
1539 template <typename Q>
1540 void f(typename Q::X) @{@}
1542 template <template <typename> class Q>
1543 void f(typename Q<int>::X) @{@}
1547 Instantiations of these templates may be mangled incorrectly.
1551 @item -Wctor-dtor-privacy @r{(C++ only)}
1552 @opindex Wctor-dtor-privacy
1553 Warn when a class seems unusable because all the constructors or
1554 destructors in that class are private, and it has neither friends nor
1555 public static member functions.
1557 @item -Wnon-virtual-dtor @r{(C++ only)}
1558 @opindex Wnon-virtual-dtor
1559 Warn when a class appears to be polymorphic, thereby requiring a virtual
1560 destructor, yet it declares a non-virtual one.
1561 This warning is enabled by @option{-Wall}.
1563 @item -Wreorder @r{(C++ only)}
1565 @cindex reordering, warning
1566 @cindex warning for reordering of member initializers
1567 Warn when the order of member initializers given in the code does not
1568 match the order in which they must be executed. For instance:
1574 A(): j (0), i (1) @{ @}
1578 The compiler will rearrange the member initializers for @samp{i}
1579 and @samp{j} to match the declaration order of the members, emitting
1580 a warning to that effect. This warning is enabled by @option{-Wall}.
1583 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1586 @item -Weffc++ @r{(C++ only)}
1588 Warn about violations of the following style guidelines from Scott Meyers'
1589 @cite{Effective C++} book:
1593 Item 11: Define a copy constructor and an assignment operator for classes
1594 with dynamically allocated memory.
1597 Item 12: Prefer initialization to assignment in constructors.
1600 Item 14: Make destructors virtual in base classes.
1603 Item 15: Have @code{operator=} return a reference to @code{*this}.
1606 Item 23: Don't try to return a reference when you must return an object.
1610 Also warn about violations of the following style guidelines from
1611 Scott Meyers' @cite{More Effective C++} book:
1615 Item 6: Distinguish between prefix and postfix forms of increment and
1616 decrement operators.
1619 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1623 When selecting this option, be aware that the standard library
1624 headers do not obey all of these guidelines; use @samp{grep -v}
1625 to filter out those warnings.
1627 @item -Wno-deprecated @r{(C++ only)}
1628 @opindex Wno-deprecated
1629 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1631 @item -Wno-non-template-friend @r{(C++ only)}
1632 @opindex Wno-non-template-friend
1633 Disable warnings when non-templatized friend functions are declared
1634 within a template. Since the advent of explicit template specification
1635 support in G++, if the name of the friend is an unqualified-id (i.e.,
1636 @samp{friend foo(int)}), the C++ language specification demands that the
1637 friend declare or define an ordinary, nontemplate function. (Section
1638 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1639 could be interpreted as a particular specialization of a templatized
1640 function. Because this non-conforming behavior is no longer the default
1641 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1642 check existing code for potential trouble spots and is on by default.
1643 This new compiler behavior can be turned off with
1644 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1645 but disables the helpful warning.
1647 @item -Wold-style-cast @r{(C++ only)}
1648 @opindex Wold-style-cast
1649 Warn if an old-style (C-style) cast to a non-void type is used within
1650 a C++ program. The new-style casts (@samp{static_cast},
1651 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1652 unintended effects and much easier to search for.
1654 @item -Woverloaded-virtual @r{(C++ only)}
1655 @opindex Woverloaded-virtual
1656 @cindex overloaded virtual fn, warning
1657 @cindex warning for overloaded virtual fn
1658 Warn when a function declaration hides virtual functions from a
1659 base class. For example, in:
1666 struct B: public A @{
1671 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1679 will fail to compile.
1681 @item -Wno-pmf-conversions @r{(C++ only)}
1682 @opindex Wno-pmf-conversions
1683 Disable the diagnostic for converting a bound pointer to member function
1686 @item -Wsign-promo @r{(C++ only)}
1687 @opindex Wsign-promo
1688 Warn when overload resolution chooses a promotion from unsigned or
1689 enumeral type to a signed type, over a conversion to an unsigned type of
1690 the same size. Previous versions of G++ would try to preserve
1691 unsignedness, but the standard mandates the current behavior.
1693 @item -Wsynth @r{(C++ only)}
1695 @cindex warning for synthesized methods
1696 @cindex synthesized methods, warning
1697 Warn when G++'s synthesis behavior does not match that of cfront. For
1703 A& operator = (int);
1713 In this example, G++ will synthesize a default @samp{A& operator =
1714 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1717 @node Objective-C Dialect Options
1718 @section Options Controlling Objective-C Dialect
1720 @cindex compiler options, Objective-C
1721 @cindex Objective-C options, command line
1722 @cindex options, Objective-C
1723 (NOTE: This manual does not describe the Objective-C language itself. See
1724 @w{@uref{http://gcc.gnu.org/readings.html}} for references.)
1726 This section describes the command-line options that are only meaningful
1727 for Objective-C programs, but you can also use most of the GNU compiler
1728 options regardless of what language your program is in. For example,
1729 you might compile a file @code{some_class.m} like this:
1732 gcc -g -fgnu-runtime -O -c some_class.m
1736 In this example, @option{-fgnu-runtime} is an option meant only for
1737 Objective-C programs; you can use the other options with any language
1740 Here is a list of options that are @emph{only} for compiling Objective-C
1744 @item -fconstant-string-class=@var{class-name}
1745 @opindex fconstant-string-class
1746 Use @var{class-name} as the name of the class to instantiate for each
1747 literal string specified with the syntax @code{@@"@dots{}"}. The default
1748 class name is @code{NXConstantString} if the GNU runtime is being used, and
1749 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1750 @option{-fconstant-cfstrings} option, if also present, will override the
1751 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1752 to be laid out as constant CoreFoundation strings.
1755 @opindex fgnu-runtime
1756 Generate object code compatible with the standard GNU Objective-C
1757 runtime. This is the default for most types of systems.
1759 @item -fnext-runtime
1760 @opindex fnext-runtime
1761 Generate output compatible with the NeXT runtime. This is the default
1762 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1763 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1766 @item -fno-nil-receivers
1767 @opindex fno-nil-receivers
1768 Assume that all Objective-C message dispatches (e.g.,
1769 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1770 is not @code{nil}. This allows for more efficient entry points in the runtime to be
1771 used. Currently, this option is only available in conjunction with
1772 the NeXT runtime on Mac OS X 10.3 and later.
1774 @item -fobjc-exceptions
1775 @opindex fobjc-exceptions
1776 Enable syntactic support for structured exception handling in Objective-C,
1777 similar to what is offered by C++ and Java. Currently, this option is only
1778 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1786 @@catch (AnObjCClass *exc) @{
1793 @@catch (AnotherClass *exc) @{
1796 @@catch (id allOthers) @{
1806 The @code{@@throw} statement may appear anywhere in an Objective-C or
1807 Objective-C++ program; when used inside of a @code{@@catch} block, the
1808 @code{@@throw} may appear without an argument (as shown above), in which case
1809 the object caught by the @code{@@catch} will be rethrown.
1811 Note that only (pointers to) Objective-C objects may be thrown and
1812 caught using this scheme. When an object is thrown, it will be caught
1813 by the nearest @code{@@catch} clause capable of handling objects of that type,
1814 analogously to how @code{catch} blocks work in C++ and Java. A
1815 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1816 any and all Objective-C exceptions not caught by previous @code{@@catch}
1819 The @code{@@finally} clause, if present, will be executed upon exit from the
1820 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1821 regardless of whether any exceptions are thrown, caught or rethrown
1822 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1823 of the @code{finally} clause in Java.
1825 There are several caveats to using the new exception mechanism:
1829 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1830 idioms provided by the @code{NSException} class, the new
1831 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1832 systems, due to additional functionality needed in the (NeXT) Objective-C
1836 As mentioned above, the new exceptions do not support handling
1837 types other than Objective-C objects. Furthermore, when used from
1838 Objective-C++, the Objective-C exception model does not interoperate with C++
1839 exceptions at this time. This means you cannot @code{@@throw} an exception
1840 from Objective-C and @code{catch} it in C++, or vice versa
1841 (i.e., @code{throw @dots{} @@catch}).
1844 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1845 blocks for thread-safe execution:
1848 @@synchronized (ObjCClass *guard) @{
1853 Upon entering the @code{@@synchronized} block, a thread of execution shall
1854 first check whether a lock has been placed on the corresponding @code{guard}
1855 object by another thread. If it has, the current thread shall wait until
1856 the other thread relinquishes its lock. Once @code{guard} becomes available,
1857 the current thread will place its own lock on it, execute the code contained in
1858 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1859 making @code{guard} available to other threads).
1861 Unlike Java, Objective-C does not allow for entire methods to be marked
1862 @code{@@synchronized}. Note that throwing exceptions out of
1863 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1864 to be unlocked properly.
1866 @item -freplace-objc-classes
1867 @opindex freplace-objc-classes
1868 Emit a special marker instructing @command{ld(1)} not to statically link in
1869 the resulting object file, and allow @command{dyld(1)} to load it in at
1870 run time instead. This is used in conjunction with the Fix-and-Continue
1871 debugging mode, where the object file in question may be recompiled and
1872 dynamically reloaded in the course of program execution, without the need
1873 to restart the program itself. Currently, Fix-and-Continue functionality
1874 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
1879 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1880 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1881 compile time) with static class references that get initialized at load time,
1882 which improves run-time performance. Specifying the @option{-fzero-link} flag
1883 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1884 to be retained. This is useful in Zero-Link debugging mode, since it allows
1885 for individual class implementations to be modified during program execution.
1889 Dump interface declarations for all classes seen in the source file to a
1890 file named @file{@var{sourcename}.decl}.
1893 @opindex Wno-protocol
1894 If a class is declared to implement a protocol, a warning is issued for
1895 every method in the protocol that is not implemented by the class. The
1896 default behavior is to issue a warning for every method not explicitly
1897 implemented in the class, even if a method implementation is inherited
1898 from the superclass. If you use the @code{-Wno-protocol} option, then
1899 methods inherited from the superclass are considered to be implemented,
1900 and no warning is issued for them.
1904 Warn if multiple methods of different types for the same selector are
1905 found during compilation. The check is performed on the list of methods
1906 in the final stage of compilation. Additionally, a check is performed
1907 for each selector appearing in a @code{@@selector(@dots{})}
1908 expression, and a corresponding method for that selector has been found
1909 during compilation. Because these checks scan the method table only at
1910 the end of compilation, these warnings are not produced if the final
1911 stage of compilation is not reached, for example because an error is
1912 found during compilation, or because the @code{-fsyntax-only} option is
1915 @item -Wundeclared-selector
1916 @opindex Wundeclared-selector
1917 Warn if a @code{@@selector(@dots{})} expression referring to an
1918 undeclared selector is found. A selector is considered undeclared if no
1919 method with that name has been declared before the
1920 @code{@@selector(@dots{})} expression, either explicitly in an
1921 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
1922 an @code{@@implementation} section. This option always performs its
1923 checks as soon as a @code{@@selector(@dots{})} expression is found,
1924 while @code{-Wselector} only performs its checks in the final stage of
1925 compilation. This also enforces the coding style convention
1926 that methods and selectors must be declared before being used.
1928 @item -print-objc-runtime-info
1929 @opindex print-objc-runtime-info
1930 Generate C header describing the largest structure that is passed by
1935 @node Language Independent Options
1936 @section Options to Control Diagnostic Messages Formatting
1937 @cindex options to control diagnostics formatting
1938 @cindex diagnostic messages
1939 @cindex message formatting
1941 Traditionally, diagnostic messages have been formatted irrespective of
1942 the output device's aspect (e.g.@: its width, @dots{}). The options described
1943 below can be used to control the diagnostic messages formatting
1944 algorithm, e.g.@: how many characters per line, how often source location
1945 information should be reported. Right now, only the C++ front end can
1946 honor these options. However it is expected, in the near future, that
1947 the remaining front ends would be able to digest them correctly.
1950 @item -fmessage-length=@var{n}
1951 @opindex fmessage-length
1952 Try to format error messages so that they fit on lines of about @var{n}
1953 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1954 the front ends supported by GCC@. If @var{n} is zero, then no
1955 line-wrapping will be done; each error message will appear on a single
1958 @opindex fdiagnostics-show-location
1959 @item -fdiagnostics-show-location=once
1960 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1961 reporter to emit @emph{once} source location information; that is, in
1962 case the message is too long to fit on a single physical line and has to
1963 be wrapped, the source location won't be emitted (as prefix) again,
1964 over and over, in subsequent continuation lines. This is the default
1967 @item -fdiagnostics-show-location=every-line
1968 Only meaningful in line-wrapping mode. Instructs the diagnostic
1969 messages reporter to emit the same source location information (as
1970 prefix) for physical lines that result from the process of breaking
1971 a message which is too long to fit on a single line.
1975 @node Warning Options
1976 @section Options to Request or Suppress Warnings
1977 @cindex options to control warnings
1978 @cindex warning messages
1979 @cindex messages, warning
1980 @cindex suppressing warnings
1982 Warnings are diagnostic messages that report constructions which
1983 are not inherently erroneous but which are risky or suggest there
1984 may have been an error.
1986 You can request many specific warnings with options beginning @samp{-W},
1987 for example @option{-Wimplicit} to request warnings on implicit
1988 declarations. Each of these specific warning options also has a
1989 negative form beginning @samp{-Wno-} to turn off warnings;
1990 for example, @option{-Wno-implicit}. This manual lists only one of the
1991 two forms, whichever is not the default.
1993 The following options control the amount and kinds of warnings produced
1994 by GCC; for further, language-specific options also refer to
1995 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
1998 @cindex syntax checking
2000 @opindex fsyntax-only
2001 Check the code for syntax errors, but don't do anything beyond that.
2005 Issue all the warnings demanded by strict ISO C and ISO C++;
2006 reject all programs that use forbidden extensions, and some other
2007 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2008 version of the ISO C standard specified by any @option{-std} option used.
2010 Valid ISO C and ISO C++ programs should compile properly with or without
2011 this option (though a rare few will require @option{-ansi} or a
2012 @option{-std} option specifying the required version of ISO C)@. However,
2013 without this option, certain GNU extensions and traditional C and C++
2014 features are supported as well. With this option, they are rejected.
2016 @option{-pedantic} does not cause warning messages for use of the
2017 alternate keywords whose names begin and end with @samp{__}. Pedantic
2018 warnings are also disabled in the expression that follows
2019 @code{__extension__}. However, only system header files should use
2020 these escape routes; application programs should avoid them.
2021 @xref{Alternate Keywords}.
2023 Some users try to use @option{-pedantic} to check programs for strict ISO
2024 C conformance. They soon find that it does not do quite what they want:
2025 it finds some non-ISO practices, but not all---only those for which
2026 ISO C @emph{requires} a diagnostic, and some others for which
2027 diagnostics have been added.
2029 A feature to report any failure to conform to ISO C might be useful in
2030 some instances, but would require considerable additional work and would
2031 be quite different from @option{-pedantic}. We don't have plans to
2032 support such a feature in the near future.
2034 Where the standard specified with @option{-std} represents a GNU
2035 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2036 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2037 extended dialect is based. Warnings from @option{-pedantic} are given
2038 where they are required by the base standard. (It would not make sense
2039 for such warnings to be given only for features not in the specified GNU
2040 C dialect, since by definition the GNU dialects of C include all
2041 features the compiler supports with the given option, and there would be
2042 nothing to warn about.)
2044 @item -pedantic-errors
2045 @opindex pedantic-errors
2046 Like @option{-pedantic}, except that errors are produced rather than
2051 Inhibit all warning messages.
2055 Inhibit warning messages about the use of @samp{#import}.
2057 @item -Wchar-subscripts
2058 @opindex Wchar-subscripts
2059 Warn if an array subscript has type @code{char}. This is a common cause
2060 of error, as programmers often forget that this type is signed on some
2065 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2066 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2068 @item -Wfatal-errors
2069 @opindex Wfatal-errors
2070 This option causes the compiler to abort compilation on the first error
2071 occurred rather than trying to keep going and printing further error
2076 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2077 the arguments supplied have types appropriate to the format string
2078 specified, and that the conversions specified in the format string make
2079 sense. This includes standard functions, and others specified by format
2080 attributes (@pxref{Function Attributes}), in the @code{printf},
2081 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2082 not in the C standard) families.
2084 The formats are checked against the format features supported by GNU
2085 libc version 2.2. These include all ISO C90 and C99 features, as well
2086 as features from the Single Unix Specification and some BSD and GNU
2087 extensions. Other library implementations may not support all these
2088 features; GCC does not support warning about features that go beyond a
2089 particular library's limitations. However, if @option{-pedantic} is used
2090 with @option{-Wformat}, warnings will be given about format features not
2091 in the selected standard version (but not for @code{strfmon} formats,
2092 since those are not in any version of the C standard). @xref{C Dialect
2093 Options,,Options Controlling C Dialect}.
2095 Since @option{-Wformat} also checks for null format arguments for
2096 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2098 @option{-Wformat} is included in @option{-Wall}. For more control over some
2099 aspects of format checking, the options @option{-Wformat-y2k},
2100 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2101 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2102 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2105 @opindex Wformat-y2k
2106 If @option{-Wformat} is specified, also warn about @code{strftime}
2107 formats which may yield only a two-digit year.
2109 @item -Wno-format-extra-args
2110 @opindex Wno-format-extra-args
2111 If @option{-Wformat} is specified, do not warn about excess arguments to a
2112 @code{printf} or @code{scanf} format function. The C standard specifies
2113 that such arguments are ignored.
2115 Where the unused arguments lie between used arguments that are
2116 specified with @samp{$} operand number specifications, normally
2117 warnings are still given, since the implementation could not know what
2118 type to pass to @code{va_arg} to skip the unused arguments. However,
2119 in the case of @code{scanf} formats, this option will suppress the
2120 warning if the unused arguments are all pointers, since the Single
2121 Unix Specification says that such unused arguments are allowed.
2123 @item -Wno-format-zero-length
2124 @opindex Wno-format-zero-length
2125 If @option{-Wformat} is specified, do not warn about zero-length formats.
2126 The C standard specifies that zero-length formats are allowed.
2128 @item -Wformat-nonliteral
2129 @opindex Wformat-nonliteral
2130 If @option{-Wformat} is specified, also warn if the format string is not a
2131 string literal and so cannot be checked, unless the format function
2132 takes its format arguments as a @code{va_list}.
2134 @item -Wformat-security
2135 @opindex Wformat-security
2136 If @option{-Wformat} is specified, also warn about uses of format
2137 functions that represent possible security problems. At present, this
2138 warns about calls to @code{printf} and @code{scanf} functions where the
2139 format string is not a string literal and there are no format arguments,
2140 as in @code{printf (foo);}. This may be a security hole if the format
2141 string came from untrusted input and contains @samp{%n}. (This is
2142 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2143 in future warnings may be added to @option{-Wformat-security} that are not
2144 included in @option{-Wformat-nonliteral}.)
2148 Enable @option{-Wformat} plus format checks not included in
2149 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2150 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2154 Warn about passing a null pointer for arguments marked as
2155 requiring a non-null value by the @code{nonnull} function attribute.
2157 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2158 can be disabled with the @option{-Wno-nonnull} option.
2160 @item -Winit-self @r{(C, C++, and Objective-C only)}
2162 Warn about uninitialized variables which are initialized with themselves.
2163 Note this option can only be used with the @option{-Wuninitialized} option,
2164 which in turn only works with @option{-O1} and above.
2166 For example, GCC will warn about @code{i} being uninitialized in the
2167 following snippet only when @option{-Winit-self} has been specified:
2178 @item -Wimplicit-int
2179 @opindex Wimplicit-int
2180 Warn when a declaration does not specify a type.
2182 @item -Wimplicit-function-declaration
2183 @itemx -Werror-implicit-function-declaration
2184 @opindex Wimplicit-function-declaration
2185 @opindex Werror-implicit-function-declaration
2186 Give a warning (or error) whenever a function is used before being
2191 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2195 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2196 function with external linkage, returning int, taking either zero
2197 arguments, two, or three arguments of appropriate types.
2199 @item -Wmissing-braces
2200 @opindex Wmissing-braces
2201 Warn if an aggregate or union initializer is not fully bracketed. In
2202 the following example, the initializer for @samp{a} is not fully
2203 bracketed, but that for @samp{b} is fully bracketed.
2206 int a[2][2] = @{ 0, 1, 2, 3 @};
2207 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2211 @opindex Wparentheses
2212 Warn if parentheses are omitted in certain contexts, such
2213 as when there is an assignment in a context where a truth value
2214 is expected, or when operators are nested whose precedence people
2215 often get confused about.
2217 Also warn about constructions where there may be confusion to which
2218 @code{if} statement an @code{else} branch belongs. Here is an example of
2233 In C, every @code{else} branch belongs to the innermost possible @code{if}
2234 statement, which in this example is @code{if (b)}. This is often not
2235 what the programmer expected, as illustrated in the above example by
2236 indentation the programmer chose. When there is the potential for this
2237 confusion, GCC will issue a warning when this flag is specified.
2238 To eliminate the warning, add explicit braces around the innermost
2239 @code{if} statement so there is no way the @code{else} could belong to
2240 the enclosing @code{if}. The resulting code would look like this:
2256 @item -Wsequence-point
2257 @opindex Wsequence-point
2258 Warn about code that may have undefined semantics because of violations
2259 of sequence point rules in the C standard.
2261 The C standard defines the order in which expressions in a C program are
2262 evaluated in terms of @dfn{sequence points}, which represent a partial
2263 ordering between the execution of parts of the program: those executed
2264 before the sequence point, and those executed after it. These occur
2265 after the evaluation of a full expression (one which is not part of a
2266 larger expression), after the evaluation of the first operand of a
2267 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2268 function is called (but after the evaluation of its arguments and the
2269 expression denoting the called function), and in certain other places.
2270 Other than as expressed by the sequence point rules, the order of
2271 evaluation of subexpressions of an expression is not specified. All
2272 these rules describe only a partial order rather than a total order,
2273 since, for example, if two functions are called within one expression
2274 with no sequence point between them, the order in which the functions
2275 are called is not specified. However, the standards committee have
2276 ruled that function calls do not overlap.
2278 It is not specified when between sequence points modifications to the
2279 values of objects take effect. Programs whose behavior depends on this
2280 have undefined behavior; the C standard specifies that ``Between the
2281 previous and next sequence point an object shall have its stored value
2282 modified at most once by the evaluation of an expression. Furthermore,
2283 the prior value shall be read only to determine the value to be
2284 stored.''. If a program breaks these rules, the results on any
2285 particular implementation are entirely unpredictable.
2287 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2288 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2289 diagnosed by this option, and it may give an occasional false positive
2290 result, but in general it has been found fairly effective at detecting
2291 this sort of problem in programs.
2293 The present implementation of this option only works for C programs. A
2294 future implementation may also work for C++ programs.
2296 The C standard is worded confusingly, therefore there is some debate
2297 over the precise meaning of the sequence point rules in subtle cases.
2298 Links to discussions of the problem, including proposed formal
2299 definitions, may be found on our readings page, at
2300 @w{@uref{http://gcc.gnu.org/readings.html}}.
2303 @opindex Wreturn-type
2304 Warn whenever a function is defined with a return-type that defaults to
2305 @code{int}. Also warn about any @code{return} statement with no
2306 return-value in a function whose return-type is not @code{void}.
2308 For C++, a function without return type always produces a diagnostic
2309 message, even when @option{-Wno-return-type} is specified. The only
2310 exceptions are @samp{main} and functions defined in system headers.
2314 Warn whenever a @code{switch} statement has an index of enumeral type
2315 and lacks a @code{case} for one or more of the named codes of that
2316 enumeration. (The presence of a @code{default} label prevents this
2317 warning.) @code{case} labels outside the enumeration range also
2318 provoke warnings when this option is used.
2320 @item -Wswitch-default
2321 @opindex Wswitch-switch
2322 Warn whenever a @code{switch} statement does not have a @code{default}
2326 @opindex Wswitch-enum
2327 Warn whenever a @code{switch} statement has an index of enumeral type
2328 and lacks a @code{case} for one or more of the named codes of that
2329 enumeration. @code{case} labels outside the enumeration range also
2330 provoke warnings when this option is used.
2334 Warn if any trigraphs are encountered that might change the meaning of
2335 the program (trigraphs within comments are not warned about).
2337 @item -Wunused-function
2338 @opindex Wunused-function
2339 Warn whenever a static function is declared but not defined or a
2340 non\-inline static function is unused.
2342 @item -Wunused-label
2343 @opindex Wunused-label
2344 Warn whenever a label is declared but not used.
2346 To suppress this warning use the @samp{unused} attribute
2347 (@pxref{Variable Attributes}).
2349 @item -Wunused-parameter
2350 @opindex Wunused-parameter
2351 Warn whenever a function parameter is unused aside from its declaration.
2353 To suppress this warning use the @samp{unused} attribute
2354 (@pxref{Variable Attributes}).
2356 @item -Wunused-variable
2357 @opindex Wunused-variable
2358 Warn whenever a local variable or non-constant static variable is unused
2359 aside from its declaration
2361 To suppress this warning use the @samp{unused} attribute
2362 (@pxref{Variable Attributes}).
2364 @item -Wunused-value
2365 @opindex Wunused-value
2366 Warn whenever a statement computes a result that is explicitly not used.
2368 To suppress this warning cast the expression to @samp{void}.
2372 All the above @option{-Wunused} options combined.
2374 In order to get a warning about an unused function parameter, you must
2375 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2376 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2378 @item -Wuninitialized
2379 @opindex Wuninitialized
2380 Warn if an automatic variable is used without first being initialized or
2381 if a variable may be clobbered by a @code{setjmp} call.
2383 These warnings are possible only in optimizing compilation,
2384 because they require data flow information that is computed only
2385 when optimizing. If you don't specify @option{-O}, you simply won't
2388 If you want to warn about code which uses the uninitialized value of the
2389 variable in its own initializer, use the @option{-Winit-self} option.
2391 These warnings occur only for variables that are candidates for
2392 register allocation. Therefore, they do not occur for a variable that
2393 is declared @code{volatile}, or whose address is taken, or whose size
2394 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2395 structures, unions or arrays, even when they are in registers.
2397 Note that there may be no warning about a variable that is used only
2398 to compute a value that itself is never used, because such
2399 computations may be deleted by data flow analysis before the warnings
2402 These warnings are made optional because GCC is not smart
2403 enough to see all the reasons why the code might be correct
2404 despite appearing to have an error. Here is one example of how
2425 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2426 always initialized, but GCC doesn't know this. Here is
2427 another common case:
2432 if (change_y) save_y = y, y = new_y;
2434 if (change_y) y = save_y;
2439 This has no bug because @code{save_y} is used only if it is set.
2441 @cindex @code{longjmp} warnings
2442 This option also warns when a non-volatile automatic variable might be
2443 changed by a call to @code{longjmp}. These warnings as well are possible
2444 only in optimizing compilation.
2446 The compiler sees only the calls to @code{setjmp}. It cannot know
2447 where @code{longjmp} will be called; in fact, a signal handler could
2448 call it at any point in the code. As a result, you may get a warning
2449 even when there is in fact no problem because @code{longjmp} cannot
2450 in fact be called at the place which would cause a problem.
2452 Some spurious warnings can be avoided if you declare all the functions
2453 you use that never return as @code{noreturn}. @xref{Function
2456 @item -Wunknown-pragmas
2457 @opindex Wunknown-pragmas
2458 @cindex warning for unknown pragmas
2459 @cindex unknown pragmas, warning
2460 @cindex pragmas, warning of unknown
2461 Warn when a #pragma directive is encountered which is not understood by
2462 GCC@. If this command line option is used, warnings will even be issued
2463 for unknown pragmas in system header files. This is not the case if
2464 the warnings were only enabled by the @option{-Wall} command line option.
2466 @item -Wstrict-aliasing
2467 @opindex Wstrict-aliasing
2468 This option is only active when @option{-fstrict-aliasing} is active.
2469 It warns about code which might break the strict aliasing rules that the
2470 compiler is using for optimization. The warning does not catch all
2471 cases, but does attempt to catch the more common pitfalls. It is
2472 included in @option{-Wall}.
2474 @item -Wstrict-aliasing=2
2475 @opindex Wstrict-aliasing=2
2476 This option is only active when @option{-fstrict-aliasing} is active.
2477 It warns about all code which might break the strict aliasing rules that the
2478 compiler is using for optimization. This warning catches all cases, but
2479 it will also give a warning for some ambiguous cases that are safe.
2483 All of the above @samp{-W} options combined. This enables all the
2484 warnings about constructions that some users consider questionable, and
2485 that are easy to avoid (or modify to prevent the warning), even in
2486 conjunction with macros. This also enables some language-specific
2487 warnings described in @ref{C++ Dialect Options} and
2488 @ref{Objective-C Dialect Options}.
2491 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2492 Some of them warn about constructions that users generally do not
2493 consider questionable, but which occasionally you might wish to check
2494 for; others warn about constructions that are necessary or hard to avoid
2495 in some cases, and there is no simple way to modify the code to suppress
2502 (This option used to be called @option{-W}. The older name is still
2503 supported, but the newer name is more descriptive.) Print extra warning
2504 messages for these events:
2508 A function can return either with or without a value. (Falling
2509 off the end of the function body is considered returning without
2510 a value.) For example, this function would evoke such a
2524 An expression-statement or the left-hand side of a comma expression
2525 contains no side effects.
2526 To suppress the warning, cast the unused expression to void.
2527 For example, an expression such as @samp{x[i,j]} will cause a warning,
2528 but @samp{x[(void)i,j]} will not.
2531 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2534 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2535 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2536 that of ordinary mathematical notation.
2539 Storage-class specifiers like @code{static} are not the first things in
2540 a declaration. According to the C Standard, this usage is obsolescent.
2543 The return type of a function has a type qualifier such as @code{const}.
2544 Such a type qualifier has no effect, since the value returned by a
2545 function is not an lvalue. (But don't warn about the GNU extension of
2546 @code{volatile void} return types. That extension will be warned about
2547 if @option{-pedantic} is specified.)
2550 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2554 A comparison between signed and unsigned values could produce an
2555 incorrect result when the signed value is converted to unsigned.
2556 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2559 An aggregate has an initializer which does not initialize all members.
2560 For example, the following code would cause such a warning, because
2561 @code{x.h} would be implicitly initialized to zero:
2564 struct s @{ int f, g, h; @};
2565 struct s x = @{ 3, 4 @};
2569 A function parameter is declared without a type specifier in K&R-style
2577 An empty body occurs in an @samp{if} or @samp{else} statement.
2580 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2581 @samp{>}, or @samp{>=}.
2584 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2587 Any of several floating-point events that often indicate errors, such as
2588 overflow, underflow, loss of precision, etc.
2590 @item @r{(C++ only)}
2591 An enumerator and a non-enumerator both appear in a conditional expression.
2593 @item @r{(C++ only)}
2594 A non-static reference or non-static @samp{const} member appears in a
2595 class without constructors.
2597 @item @r{(C++ only)}
2598 Ambiguous virtual bases.
2600 @item @r{(C++ only)}
2601 Subscripting an array which has been declared @samp{register}.
2603 @item @r{(C++ only)}
2604 Taking the address of a variable which has been declared @samp{register}.
2606 @item @r{(C++ only)}
2607 A base class is not initialized in a derived class' copy constructor.
2610 @item -Wno-div-by-zero
2611 @opindex Wno-div-by-zero
2612 @opindex Wdiv-by-zero
2613 Do not warn about compile-time integer division by zero. Floating point
2614 division by zero is not warned about, as it can be a legitimate way of
2615 obtaining infinities and NaNs.
2617 @item -Wsystem-headers
2618 @opindex Wsystem-headers
2619 @cindex warnings from system headers
2620 @cindex system headers, warnings from
2621 Print warning messages for constructs found in system header files.
2622 Warnings from system headers are normally suppressed, on the assumption
2623 that they usually do not indicate real problems and would only make the
2624 compiler output harder to read. Using this command line option tells
2625 GCC to emit warnings from system headers as if they occurred in user
2626 code. However, note that using @option{-Wall} in conjunction with this
2627 option will @emph{not} warn about unknown pragmas in system
2628 headers---for that, @option{-Wunknown-pragmas} must also be used.
2631 @opindex Wfloat-equal
2632 Warn if floating point values are used in equality comparisons.
2634 The idea behind this is that sometimes it is convenient (for the
2635 programmer) to consider floating-point values as approximations to
2636 infinitely precise real numbers. If you are doing this, then you need
2637 to compute (by analyzing the code, or in some other way) the maximum or
2638 likely maximum error that the computation introduces, and allow for it
2639 when performing comparisons (and when producing output, but that's a
2640 different problem). In particular, instead of testing for equality, you
2641 would check to see whether the two values have ranges that overlap; and
2642 this is done with the relational operators, so equality comparisons are
2645 @item -Wtraditional @r{(C only)}
2646 @opindex Wtraditional
2647 Warn about certain constructs that behave differently in traditional and
2648 ISO C@. Also warn about ISO C constructs that have no traditional C
2649 equivalent, and/or problematic constructs which should be avoided.
2653 Macro parameters that appear within string literals in the macro body.
2654 In traditional C macro replacement takes place within string literals,
2655 but does not in ISO C@.
2658 In traditional C, some preprocessor directives did not exist.
2659 Traditional preprocessors would only consider a line to be a directive
2660 if the @samp{#} appeared in column 1 on the line. Therefore
2661 @option{-Wtraditional} warns about directives that traditional C
2662 understands but would ignore because the @samp{#} does not appear as the
2663 first character on the line. It also suggests you hide directives like
2664 @samp{#pragma} not understood by traditional C by indenting them. Some
2665 traditional implementations would not recognize @samp{#elif}, so it
2666 suggests avoiding it altogether.
2669 A function-like macro that appears without arguments.
2672 The unary plus operator.
2675 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2676 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2677 constants.) Note, these suffixes appear in macros defined in the system
2678 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2679 Use of these macros in user code might normally lead to spurious
2680 warnings, however GCC's integrated preprocessor has enough context to
2681 avoid warning in these cases.
2684 A function declared external in one block and then used after the end of
2688 A @code{switch} statement has an operand of type @code{long}.
2691 A non-@code{static} function declaration follows a @code{static} one.
2692 This construct is not accepted by some traditional C compilers.
2695 The ISO type of an integer constant has a different width or
2696 signedness from its traditional type. This warning is only issued if
2697 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2698 typically represent bit patterns, are not warned about.
2701 Usage of ISO string concatenation is detected.
2704 Initialization of automatic aggregates.
2707 Identifier conflicts with labels. Traditional C lacks a separate
2708 namespace for labels.
2711 Initialization of unions. If the initializer is zero, the warning is
2712 omitted. This is done under the assumption that the zero initializer in
2713 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2714 initializer warnings and relies on default initialization to zero in the
2718 Conversions by prototypes between fixed/floating point values and vice
2719 versa. The absence of these prototypes when compiling with traditional
2720 C would cause serious problems. This is a subset of the possible
2721 conversion warnings, for the full set use @option{-Wconversion}.
2724 Use of ISO C style function definitions. This warning intentionally is
2725 @emph{not} issued for prototype declarations or variadic functions
2726 because these ISO C features will appear in your code when using
2727 libiberty's traditional C compatibility macros, @code{PARAMS} and
2728 @code{VPARAMS}. This warning is also bypassed for nested functions
2729 because that feature is already a GCC extension and thus not relevant to
2730 traditional C compatibility.
2733 @item -Wdeclaration-after-statement @r{(C only)}
2734 @opindex Wdeclaration-after-statement
2735 Warn when a declaration is found after a statement in a block. This
2736 construct, known from C++, was introduced with ISO C99 and is by default
2737 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2738 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2742 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2744 @item -Wendif-labels
2745 @opindex Wendif-labels
2746 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2750 Warn whenever a local variable shadows another local variable, parameter or
2751 global variable or whenever a built-in function is shadowed.
2753 @item -Wlarger-than-@var{len}
2754 @opindex Wlarger-than
2755 Warn whenever an object of larger than @var{len} bytes is defined.
2757 @item -Wpointer-arith
2758 @opindex Wpointer-arith
2759 Warn about anything that depends on the ``size of'' a function type or
2760 of @code{void}. GNU C assigns these types a size of 1, for
2761 convenience in calculations with @code{void *} pointers and pointers
2764 @item -Wbad-function-cast @r{(C only)}
2765 @opindex Wbad-function-cast
2766 Warn whenever a function call is cast to a non-matching type.
2767 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2771 Warn whenever a pointer is cast so as to remove a type qualifier from
2772 the target type. For example, warn if a @code{const char *} is cast
2773 to an ordinary @code{char *}.
2776 @opindex Wcast-align
2777 Warn whenever a pointer is cast such that the required alignment of the
2778 target is increased. For example, warn if a @code{char *} is cast to
2779 an @code{int *} on machines where integers can only be accessed at
2780 two- or four-byte boundaries.
2782 @item -Wwrite-strings
2783 @opindex Wwrite-strings
2784 When compiling C, give string constants the type @code{const
2785 char[@var{length}]} so that
2786 copying the address of one into a non-@code{const} @code{char *}
2787 pointer will get a warning; when compiling C++, warn about the
2788 deprecated conversion from string constants to @code{char *}.
2789 These warnings will help you find at
2790 compile time code that can try to write into a string constant, but
2791 only if you have been very careful about using @code{const} in
2792 declarations and prototypes. Otherwise, it will just be a nuisance;
2793 this is why we did not make @option{-Wall} request these warnings.
2796 @opindex Wconversion
2797 Warn if a prototype causes a type conversion that is different from what
2798 would happen to the same argument in the absence of a prototype. This
2799 includes conversions of fixed point to floating and vice versa, and
2800 conversions changing the width or signedness of a fixed point argument
2801 except when the same as the default promotion.
2803 Also, warn if a negative integer constant expression is implicitly
2804 converted to an unsigned type. For example, warn about the assignment
2805 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2806 casts like @code{(unsigned) -1}.
2808 @item -Wsign-compare
2809 @opindex Wsign-compare
2810 @cindex warning for comparison of signed and unsigned values
2811 @cindex comparison of signed and unsigned values, warning
2812 @cindex signed and unsigned values, comparison warning
2813 Warn when a comparison between signed and unsigned values could produce
2814 an incorrect result when the signed value is converted to unsigned.
2815 This warning is also enabled by @option{-Wextra}; to get the other warnings
2816 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2818 @item -Waggregate-return
2819 @opindex Waggregate-return
2820 Warn if any functions that return structures or unions are defined or
2821 called. (In languages where you can return an array, this also elicits
2824 @item -Wstrict-prototypes @r{(C only)}
2825 @opindex Wstrict-prototypes
2826 Warn if a function is declared or defined without specifying the
2827 argument types. (An old-style function definition is permitted without
2828 a warning if preceded by a declaration which specifies the argument
2831 @item -Wold-style-definition @r{(C only)}
2832 @opindex Wold-style-definition
2833 Warn if an old-style function definition is used. A warning is given
2834 even if there is a previous prototype.
2836 @item -Wmissing-prototypes @r{(C only)}
2837 @opindex Wmissing-prototypes
2838 Warn if a global function is defined without a previous prototype
2839 declaration. This warning is issued even if the definition itself
2840 provides a prototype. The aim is to detect global functions that fail
2841 to be declared in header files.
2843 @item -Wmissing-declarations @r{(C only)}
2844 @opindex Wmissing-declarations
2845 Warn if a global function is defined without a previous declaration.
2846 Do so even if the definition itself provides a prototype.
2847 Use this option to detect global functions that are not declared in
2850 @item -Wmissing-noreturn
2851 @opindex Wmissing-noreturn
2852 Warn about functions which might be candidates for attribute @code{noreturn}.
2853 Note these are only possible candidates, not absolute ones. Care should
2854 be taken to manually verify functions actually do not ever return before
2855 adding the @code{noreturn} attribute, otherwise subtle code generation
2856 bugs could be introduced. You will not get a warning for @code{main} in
2857 hosted C environments.
2859 @item -Wmissing-format-attribute
2860 @opindex Wmissing-format-attribute
2862 If @option{-Wformat} is enabled, also warn about functions which might be
2863 candidates for @code{format} attributes. Note these are only possible
2864 candidates, not absolute ones. GCC will guess that @code{format}
2865 attributes might be appropriate for any function that calls a function
2866 like @code{vprintf} or @code{vscanf}, but this might not always be the
2867 case, and some functions for which @code{format} attributes are
2868 appropriate may not be detected. This option has no effect unless
2869 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2871 @item -Wno-multichar
2872 @opindex Wno-multichar
2874 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2875 Usually they indicate a typo in the user's code, as they have
2876 implementation-defined values, and should not be used in portable code.
2878 @item -Wno-deprecated-declarations
2879 @opindex Wno-deprecated-declarations
2880 Do not warn about uses of functions, variables, and types marked as
2881 deprecated by using the @code{deprecated} attribute.
2882 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2883 @pxref{Type Attributes}.)
2887 Warn if a structure is given the packed attribute, but the packed
2888 attribute has no effect on the layout or size of the structure.
2889 Such structures may be mis-aligned for little benefit. For
2890 instance, in this code, the variable @code{f.x} in @code{struct bar}
2891 will be misaligned even though @code{struct bar} does not itself
2892 have the packed attribute:
2899 @} __attribute__((packed));
2909 Warn if padding is included in a structure, either to align an element
2910 of the structure or to align the whole structure. Sometimes when this
2911 happens it is possible to rearrange the fields of the structure to
2912 reduce the padding and so make the structure smaller.
2914 @item -Wredundant-decls
2915 @opindex Wredundant-decls
2916 Warn if anything is declared more than once in the same scope, even in
2917 cases where multiple declaration is valid and changes nothing.
2919 @item -Wnested-externs @r{(C only)}
2920 @opindex Wnested-externs
2921 Warn if an @code{extern} declaration is encountered within a function.
2923 @item -Wunreachable-code
2924 @opindex Wunreachable-code
2925 Warn if the compiler detects that code will never be executed.
2927 This option is intended to warn when the compiler detects that at
2928 least a whole line of source code will never be executed, because
2929 some condition is never satisfied or because it is after a
2930 procedure that never returns.
2932 It is possible for this option to produce a warning even though there
2933 are circumstances under which part of the affected line can be executed,
2934 so care should be taken when removing apparently-unreachable code.
2936 For instance, when a function is inlined, a warning may mean that the
2937 line is unreachable in only one inlined copy of the function.
2939 This option is not made part of @option{-Wall} because in a debugging
2940 version of a program there is often substantial code which checks
2941 correct functioning of the program and is, hopefully, unreachable
2942 because the program does work. Another common use of unreachable
2943 code is to provide behavior which is selectable at compile-time.
2947 Warn if a function can not be inlined and it was declared as inline.
2948 Even with this option, the compiler will not warn about failures to
2949 inline functions declared in system headers.
2951 The compiler uses a variety of heuristics to determine whether or not
2952 to inline a function. For example, the compiler takes into account
2953 the size of the function being inlined and the the amount of inlining
2954 that has already been done in the current function. Therefore,
2955 seemingly insignificant changes in the source program can cause the
2956 warnings produced by @option{-Winline} to appear or disappear.
2958 @item -Wno-invalid-offsetof @r{(C++ only)}
2959 @opindex Wno-invalid-offsetof
2960 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
2961 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
2962 to a non-POD type is undefined. In existing C++ implementations,
2963 however, @samp{offsetof} typically gives meaningful results even when
2964 applied to certain kinds of non-POD types. (Such as a simple
2965 @samp{struct} that fails to be a POD type only by virtue of having a
2966 constructor.) This flag is for users who are aware that they are
2967 writing nonportable code and who have deliberately chosen to ignore the
2970 The restrictions on @samp{offsetof} may be relaxed in a future version
2971 of the C++ standard.
2974 @opindex Winvalid-pch
2975 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
2976 the search path but can't be used.
2980 @opindex Wno-long-long
2981 Warn if @samp{long long} type is used. This is default. To inhibit
2982 the warning messages, use @option{-Wno-long-long}. Flags
2983 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2984 only when @option{-pedantic} flag is used.
2986 @item -Wvariadic-macros
2987 @opindex Wvariadic-macros
2988 @opindex Wno-variadic-macros
2989 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
2990 alternate syntax when in pedantic ISO C99 mode. This is default.
2991 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
2993 @item -Wdisabled-optimization
2994 @opindex Wdisabled-optimization
2995 Warn if a requested optimization pass is disabled. This warning does
2996 not generally indicate that there is anything wrong with your code; it
2997 merely indicates that GCC's optimizers were unable to handle the code
2998 effectively. Often, the problem is that your code is too big or too
2999 complex; GCC will refuse to optimize programs when the optimization
3000 itself is likely to take inordinate amounts of time.
3004 Make all warnings into errors.
3007 @node Debugging Options
3008 @section Options for Debugging Your Program or GCC
3009 @cindex options, debugging
3010 @cindex debugging information options
3012 GCC has various special options that are used for debugging
3013 either your program or GCC:
3018 Produce debugging information in the operating system's native format
3019 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
3022 On most systems that use stabs format, @option{-g} enables use of extra
3023 debugging information that only GDB can use; this extra information
3024 makes debugging work better in GDB but will probably make other debuggers
3026 refuse to read the program. If you want to control for certain whether
3027 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3028 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3030 Unlike most other C compilers, GCC allows you to use @option{-g} with
3031 @option{-O}. The shortcuts taken by optimized code may occasionally
3032 produce surprising results: some variables you declared may not exist
3033 at all; flow of control may briefly move where you did not expect it;
3034 some statements may not be executed because they compute constant
3035 results or their values were already at hand; some statements may
3036 execute in different places because they were moved out of loops.
3038 Nevertheless it proves possible to debug optimized output. This makes
3039 it reasonable to use the optimizer for programs that might have bugs.
3041 The following options are useful when GCC is generated with the
3042 capability for more than one debugging format.
3046 Produce debugging information for use by GDB@. This means to use the
3047 most expressive format available (DWARF 2, stabs, or the native format
3048 if neither of those are supported), including GDB extensions if at all
3053 Produce debugging information in stabs format (if that is supported),
3054 without GDB extensions. This is the format used by DBX on most BSD
3055 systems. On MIPS, Alpha and System V Release 4 systems this option
3056 produces stabs debugging output which is not understood by DBX or SDB@.
3057 On System V Release 4 systems this option requires the GNU assembler.
3059 @item -feliminate-unused-debug-symbols
3060 @opindex feliminate-unused-debug-symbols
3061 Produce debugging information in stabs format (if that is supported),
3062 for only symbols that are actually used.
3066 Produce debugging information in stabs format (if that is supported),
3067 using GNU extensions understood only by the GNU debugger (GDB)@. The
3068 use of these extensions is likely to make other debuggers crash or
3069 refuse to read the program.
3073 Produce debugging information in COFF format (if that is supported).
3074 This is the format used by SDB on most System V systems prior to
3079 Produce debugging information in XCOFF format (if that is supported).
3080 This is the format used by the DBX debugger on IBM RS/6000 systems.
3084 Produce debugging information in XCOFF format (if that is supported),
3085 using GNU extensions understood only by the GNU debugger (GDB)@. The
3086 use of these extensions is likely to make other debuggers crash or
3087 refuse to read the program, and may cause assemblers other than the GNU
3088 assembler (GAS) to fail with an error.
3092 Produce debugging information in DWARF version 2 format (if that is
3093 supported). This is the format used by DBX on IRIX 6.
3097 Produce debugging information in VMS debug format (if that is
3098 supported). This is the format used by DEBUG on VMS systems.
3101 @itemx -ggdb@var{level}
3102 @itemx -gstabs@var{level}
3103 @itemx -gcoff@var{level}
3104 @itemx -gxcoff@var{level}
3105 @itemx -gvms@var{level}
3106 Request debugging information and also use @var{level} to specify how
3107 much information. The default level is 2.
3109 Level 1 produces minimal information, enough for making backtraces in
3110 parts of the program that you don't plan to debug. This includes
3111 descriptions of functions and external variables, but no information
3112 about local variables and no line numbers.
3114 Level 3 includes extra information, such as all the macro definitions
3115 present in the program. Some debuggers support macro expansion when
3116 you use @option{-g3}.
3118 Note that in order to avoid confusion between DWARF1 debug level 2,
3119 and DWARF2 @option{-gdwarf-2} does not accept a concatenated debug
3120 level. Instead use an additional @option{-g@var{level}} option to
3121 change the debug level for DWARF2.
3123 @item -feliminate-dwarf2-dups
3124 @opindex feliminate-dwarf2-dups
3125 Compress DWARF2 debugging information by eliminating duplicated
3126 information about each symbol. This option only makes sense when
3127 generating DWARF2 debugging information with @option{-gdwarf-2}.
3129 @cindex @command{prof}
3132 Generate extra code to write profile information suitable for the
3133 analysis program @command{prof}. You must use this option when compiling
3134 the source files you want data about, and you must also use it when
3137 @cindex @command{gprof}
3140 Generate extra code to write profile information suitable for the
3141 analysis program @command{gprof}. You must use this option when compiling
3142 the source files you want data about, and you must also use it when
3147 Makes the compiler print out each function name as it is compiled, and
3148 print some statistics about each pass when it finishes.
3151 @opindex ftime-report
3152 Makes the compiler print some statistics about the time consumed by each
3153 pass when it finishes.
3156 @opindex fmem-report
3157 Makes the compiler print some statistics about permanent memory
3158 allocation when it finishes.
3160 @item -fprofile-arcs
3161 @opindex fprofile-arcs
3162 Add code so that program flow @dfn{arcs} are instrumented. During
3163 execution the program records how many times each branch and call is
3164 executed and how many times it is taken or returns. When the compiled
3165 program exits it saves this data to a file called
3166 @file{@var{auxname}.gcda} for each source file. The data may be used for
3167 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3168 test coverage analysis (@option{-ftest-coverage}). Each object file's
3169 @var{auxname} is generated from the name of the output file, if
3170 explicitly specified and it is not the final executable, otherwise it is
3171 the basename of the source file. In both cases any suffix is removed
3172 (e.g. @file{foo.gcda} for input file @file{dir/foo.c}, or
3173 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3178 Compile the source files with @option{-fprofile-arcs} plus optimization
3179 and code generation options. For test coverage analysis, use the
3180 additional @option{-ftest-coverage} option. You do not need to profile
3181 every source file in a program.
3184 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3185 (the latter implies the former).
3188 Run the program on a representative workload to generate the arc profile
3189 information. This may be repeated any number of times. You can run
3190 concurrent instances of your program, and provided that the file system
3191 supports locking, the data files will be correctly updated. Also
3192 @code{fork} calls are detected and correctly handled (double counting
3196 For profile-directed optimizations, compile the source files again with
3197 the same optimization and code generation options plus
3198 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3199 Control Optimization}).
3202 For test coverage analysis, use @command{gcov} to produce human readable
3203 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3204 @command{gcov} documentation for further information.
3208 With @option{-fprofile-arcs}, for each function of your program GCC
3209 creates a program flow graph, then finds a spanning tree for the graph.
3210 Only arcs that are not on the spanning tree have to be instrumented: the
3211 compiler adds code to count the number of times that these arcs are
3212 executed. When an arc is the only exit or only entrance to a block, the
3213 instrumentation code can be added to the block; otherwise, a new basic
3214 block must be created to hold the instrumentation code.
3217 @item -ftest-coverage
3218 @opindex ftest-coverage
3219 Produce a notes file that the @command{gcov} code-coverage utility
3220 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3221 show program coverage. Each source file's note file is called
3222 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3223 above for a description of @var{auxname} and instructions on how to
3224 generate test coverage data. Coverage data will match the source files
3225 more closely, if you do not optimize.
3227 @item -d@var{letters}
3229 Says to make debugging dumps during compilation at times specified by
3230 @var{letters}. This is used for debugging the compiler. The file names
3231 for most of the dumps are made by appending a pass number and a word to
3232 the @var{dumpname}. @var{dumpname} is generated from the name of the
3233 output file, if explicitly specified and it is not an executable,
3234 otherwise it is the basename of the source file. In both cases any
3235 suffix is removed (e.g. @file{foo.01.rtl} or @file{foo.02.sibling}).
3236 Here are the possible letters for use in @var{letters}, and their
3242 Annotate the assembler output with miscellaneous debugging information.
3245 Dump after computing branch probabilities, to @file{@var{file}.12.bp}.
3248 Dump after block reordering, to @file{@var{file}.31.bbro}.
3251 Dump after instruction combination, to the file @file{@var{file}.20.combine}.
3254 Dump after the first if conversion, to the file @file{@var{file}.14.ce1}.
3255 Also dump after the second if conversion, to the file @file{@var{file}.21.ce2}.
3258 Dump after branch target load optimization, to to @file{@var{file}.32.btl}.
3259 Also dump after delayed branch scheduling, to @file{@var{file}.36.dbr}.
3262 Dump all macro definitions, at the end of preprocessing, in addition to
3266 Dump after the third if conversion, to @file{@var{file}.30.ce3}.
3269 Dump after control and data flow analysis, to @file{@var{file}.11.cfg}.
3270 Also dump after life analysis, to @file{@var{file}.19.life}.
3273 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.07.addressof}.
3276 Dump after global register allocation, to @file{@var{file}.25.greg}.
3279 Dump after GCSE, to @file{@var{file}.08.gcse}.
3280 Also dump after jump bypassing and control flow optimizations, to
3281 @file{@var{file}.10.bypass}.
3284 Dump after finalization of EH handling code, to @file{@var{file}.03.eh}.
3287 Dump after sibling call optimizations, to @file{@var{file}.02.sibling}.
3290 Dump after the first jump optimization, to @file{@var{file}.04.jump}.
3293 Dump after conversion from registers to stack, to @file{@var{file}.34.stack}.
3296 Dump after local register allocation, to @file{@var{file}.24.lreg}.
3299 Dump after loop optimization passes, to @file{@var{file}.09.loop} and
3300 @file{@var{file}.16.loop2}.
3303 Dump after performing the machine dependent reorganization pass, to
3304 @file{@var{file}.35.mach}.
3307 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3310 Dump after the register move pass, to @file{@var{file}.22.regmove}.
3313 Dump after post-reload optimizations, to @file{@var{file}.26.postreload}.
3316 Dump after RTL generation, to @file{@var{file}.01.rtl}.
3319 Dump after the second scheduling pass, to @file{@var{file}.33.sched2}.
3322 Dump after CSE (including the jump optimization that sometimes follows
3323 CSE), to @file{@var{file}.06.cse}.
3326 Dump after the first scheduling pass, to @file{@var{file}.23.sched}.
3329 Dump after the second CSE pass (including the jump optimization that
3330 sometimes follows CSE), to @file{@var{file}.18.cse2}.
3333 Dump after running tracer, to @file{@var{file}.15.tracer}.
3336 Dump after null pointer elimination pass to @file{@var{file}.05.null}.
3339 Dump callgraph and unit-at-a-time optimization @file{@var{file}.00.unit}.
3342 Dump after the value profile transformations, to @file{@var{file}.13.vpt}.
3343 Also dump after variable tracking, to @file{@var{file}.35.vartrack}.
3346 Dump after the second flow pass, to @file{@var{file}.27.flow2}.
3349 Dump after the peephole pass, to @file{@var{file}.28.peephole2}.
3352 Dump after constructing the web, to @file{@var{file}.17.web}.
3355 Produce all the dumps listed above.
3358 Produce a core dump whenever an error occurs.
3361 Print statistics on memory usage, at the end of the run, to
3365 Annotate the assembler output with a comment indicating which
3366 pattern and alternative was used. The length of each instruction is
3370 Dump the RTL in the assembler output as a comment before each instruction.
3371 Also turns on @option{-dp} annotation.
3374 For each of the other indicated dump files (except for
3375 @file{@var{file}.01.rtl}), dump a representation of the control flow graph
3376 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3379 Just generate RTL for a function instead of compiling it. Usually used
3383 Dump debugging information during parsing, to standard error.
3386 @item -fdump-unnumbered
3387 @opindex fdump-unnumbered
3388 When doing debugging dumps (see @option{-d} option above), suppress instruction
3389 numbers and line number note output. This makes it more feasible to
3390 use diff on debugging dumps for compiler invocations with different
3391 options, in particular with and without @option{-g}.
3393 @item -fdump-translation-unit @r{(C and C++ only)}
3394 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3395 @opindex fdump-translation-unit
3396 Dump a representation of the tree structure for the entire translation
3397 unit to a file. The file name is made by appending @file{.tu} to the
3398 source file name. If the @samp{-@var{options}} form is used, @var{options}
3399 controls the details of the dump as described for the
3400 @option{-fdump-tree} options.
3402 @item -fdump-class-hierarchy @r{(C++ only)}
3403 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3404 @opindex fdump-class-hierarchy
3405 Dump a representation of each class's hierarchy and virtual function
3406 table layout to a file. The file name is made by appending @file{.class}
3407 to the source file name. If the @samp{-@var{options}} form is used,
3408 @var{options} controls the details of the dump as described for the
3409 @option{-fdump-tree} options.
3411 @item -fdump-tree-@var{switch} @r{(C++ only)}
3412 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3414 Control the dumping at various stages of processing the intermediate
3415 language tree to a file. The file name is generated by appending a switch
3416 specific suffix to the source file name. If the @samp{-@var{options}}
3417 form is used, @var{options} is a list of @samp{-} separated options that
3418 control the details of the dump. Not all options are applicable to all
3419 dumps, those which are not meaningful will be ignored. The following
3420 options are available
3424 Print the address of each node. Usually this is not meaningful as it
3425 changes according to the environment and source file. Its primary use
3426 is for tying up a dump file with a debug environment.
3428 Inhibit dumping of members of a scope or body of a function merely
3429 because that scope has been reached. Only dump such items when they
3430 are directly reachable by some other path.
3432 Turn on all options.
3435 The following tree dumps are possible:
3438 Dump before any tree based optimization, to @file{@var{file}.original}.
3440 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3442 Dump after function inlining, to @file{@var{file}.inlined}.
3445 @item -frandom-seed=@var{string}
3446 @opindex frandom-string
3447 This option provides a seed that GCC uses when it would otherwise use
3448 random numbers. It is used to generate certain symbol names
3449 that have to be different in every compiled file. It is also used to
3450 place unique stamps in coverage data files and the object files that
3451 produce them. You can use the @option{-frandom-seed} option to produce
3452 reproducibly identical object files.
3454 The @var{string} should be different for every file you compile.
3456 @item -fsched-verbose=@var{n}
3457 @opindex fsched-verbose
3458 On targets that use instruction scheduling, this option controls the
3459 amount of debugging output the scheduler prints. This information is
3460 written to standard error, unless @option{-dS} or @option{-dR} is
3461 specified, in which case it is output to the usual dump
3462 listing file, @file{.sched} or @file{.sched2} respectively. However
3463 for @var{n} greater than nine, the output is always printed to standard
3466 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3467 same information as @option{-dRS}. For @var{n} greater than one, it
3468 also output basic block probabilities, detailed ready list information
3469 and unit/insn info. For @var{n} greater than two, it includes RTL
3470 at abort point, control-flow and regions info. And for @var{n} over
3471 four, @option{-fsched-verbose} also includes dependence info.
3475 Store the usual ``temporary'' intermediate files permanently; place them
3476 in the current directory and name them based on the source file. Thus,
3477 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3478 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3479 preprocessed @file{foo.i} output file even though the compiler now
3480 normally uses an integrated preprocessor.
3484 Report the CPU time taken by each subprocess in the compilation
3485 sequence. For C source files, this is the compiler proper and assembler
3486 (plus the linker if linking is done). The output looks like this:
3493 The first number on each line is the ``user time,'' that is time spent
3494 executing the program itself. The second number is ``system time,''
3495 time spent executing operating system routines on behalf of the program.
3496 Both numbers are in seconds.
3498 @item -fvar-tracking
3499 @opindex fvar-tracking
3500 Run variable tracking pass. It computes where variables are stored at each
3501 position in code. Better debugging information is then generated
3502 (if the debugging information format supports this information).
3504 It is enabled by default when compiling with optimization (@option{-Os},
3505 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
3506 the debug info format supports it.
3508 @item -print-file-name=@var{library}
3509 @opindex print-file-name
3510 Print the full absolute name of the library file @var{library} that
3511 would be used when linking---and don't do anything else. With this
3512 option, GCC does not compile or link anything; it just prints the
3515 @item -print-multi-directory
3516 @opindex print-multi-directory
3517 Print the directory name corresponding to the multilib selected by any
3518 other switches present in the command line. This directory is supposed
3519 to exist in @env{GCC_EXEC_PREFIX}.
3521 @item -print-multi-lib
3522 @opindex print-multi-lib
3523 Print the mapping from multilib directory names to compiler switches
3524 that enable them. The directory name is separated from the switches by
3525 @samp{;}, and each switch starts with an @samp{@@} instead of the
3526 @samp{-}, without spaces between multiple switches. This is supposed to
3527 ease shell-processing.
3529 @item -print-prog-name=@var{program}
3530 @opindex print-prog-name
3531 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3533 @item -print-libgcc-file-name
3534 @opindex print-libgcc-file-name
3535 Same as @option{-print-file-name=libgcc.a}.
3537 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3538 but you do want to link with @file{libgcc.a}. You can do
3541 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3544 @item -print-search-dirs
3545 @opindex print-search-dirs
3546 Print the name of the configured installation directory and a list of
3547 program and library directories @command{gcc} will search---and don't do anything else.
3549 This is useful when @command{gcc} prints the error message
3550 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3551 To resolve this you either need to put @file{cpp0} and the other compiler
3552 components where @command{gcc} expects to find them, or you can set the environment
3553 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3554 Don't forget the trailing '/'.
3555 @xref{Environment Variables}.
3558 @opindex dumpmachine
3559 Print the compiler's target machine (for example,
3560 @samp{i686-pc-linux-gnu})---and don't do anything else.
3563 @opindex dumpversion
3564 Print the compiler version (for example, @samp{3.0})---and don't do
3569 Print the compiler's built-in specs---and don't do anything else. (This
3570 is used when GCC itself is being built.) @xref{Spec Files}.
3572 @item -feliminate-unused-debug-types
3573 @opindex feliminate-unused-debug-types
3574 Normally, when producing DWARF2 output, GCC will emit debugging
3575 information for all types declared in a compilation
3576 unit, regardless of whether or not they are actually used
3577 in that compilation unit. Sometimes this is useful, such as
3578 if, in the debugger, you want to cast a value to a type that is
3579 not actually used in your program (but is declared). More often,
3580 however, this results in a significant amount of wasted space.
3581 With this option, GCC will avoid producing debug symbol output
3582 for types that are nowhere used in the source file being compiled.
3585 @node Optimize Options
3586 @section Options That Control Optimization
3587 @cindex optimize options
3588 @cindex options, optimization
3590 These options control various sorts of optimizations.
3592 Without any optimization option, the compiler's goal is to reduce the
3593 cost of compilation and to make debugging produce the expected
3594 results. Statements are independent: if you stop the program with a
3595 breakpoint between statements, you can then assign a new value to any
3596 variable or change the program counter to any other statement in the
3597 function and get exactly the results you would expect from the source
3600 Turning on optimization flags makes the compiler attempt to improve
3601 the performance and/or code size at the expense of compilation time
3602 and possibly the ability to debug the program.
3604 The compiler performs optimization based on the knowledge it has of
3605 the program. Optimization levels @option{-O2} and above, in
3606 particular, enable @emph{unit-at-a-time} mode, which allows the
3607 compiler to consider information gained from later functions in
3608 the file when compiling a function. Compiling multiple files at
3609 once to a single output file in @emph{unit-at-a-time} mode allows
3610 the compiler to use information gained from all of the files when
3611 compiling each of them.
3613 Not all optimizations are controlled directly by a flag. Only
3614 optimizations that have a flag are listed.
3621 Optimize. Optimizing compilation takes somewhat more time, and a lot
3622 more memory for a large function.
3624 With @option{-O}, the compiler tries to reduce code size and execution
3625 time, without performing any optimizations that take a great deal of
3628 @option{-O} turns on the following optimization flags:
3629 @gccoptlist{-fdefer-pop @gol
3630 -fmerge-constants @gol
3632 -floop-optimize @gol
3633 -fif-conversion @gol
3634 -fif-conversion2 @gol
3635 -fdelayed-branch @gol
3636 -fguess-branch-probability @gol
3639 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3640 where doing so does not interfere with debugging.
3644 Optimize even more. GCC performs nearly all supported optimizations
3645 that do not involve a space-speed tradeoff. The compiler does not
3646 perform loop unrolling or function inlining when you specify @option{-O2}.
3647 As compared to @option{-O}, this option increases both compilation time
3648 and the performance of the generated code.
3650 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3651 also turns on the following optimization flags:
3652 @gccoptlist{-fforce-mem @gol
3653 -foptimize-sibling-calls @gol
3654 -fstrength-reduce @gol
3655 -fcse-follow-jumps -fcse-skip-blocks @gol
3656 -frerun-cse-after-loop -frerun-loop-opt @gol
3657 -fgcse -fgcse-lm -fgcse-sm -fgcse-las @gol
3658 -fdelete-null-pointer-checks @gol
3659 -fexpensive-optimizations @gol
3661 -fschedule-insns -fschedule-insns2 @gol
3662 -fsched-interblock -fsched-spec @gol
3665 -freorder-blocks -freorder-functions @gol
3666 -fstrict-aliasing @gol
3667 -funit-at-a-time @gol
3668 -falign-functions -falign-jumps @gol
3669 -falign-loops -falign-labels @gol
3672 Please note the warning under @option{-fgcse} about
3673 invoking @option{-O2} on programs that use computed gotos.
3677 Optimize yet more. @option{-O3} turns on all optimizations specified by
3678 @option{-O2} and also turns on the @option{-finline-functions},
3679 @option{-fweb} and @option{-frename-registers} options.
3683 Do not optimize. This is the default.
3687 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3688 do not typically increase code size. It also performs further
3689 optimizations designed to reduce code size.
3691 @option{-Os} disables the following optimization flags:
3692 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3693 -falign-labels -freorder-blocks -freorder-blocks-and-partition -fprefetch-loop-arrays}
3695 If you use multiple @option{-O} options, with or without level numbers,
3696 the last such option is the one that is effective.
3699 Options of the form @option{-f@var{flag}} specify machine-independent
3700 flags. Most flags have both positive and negative forms; the negative
3701 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3702 below, only one of the forms is listed---the one you typically will
3703 use. You can figure out the other form by either removing @samp{no-}
3706 The following options control specific optimizations. They are either
3707 activated by @option{-O} options or are related to ones that are. You
3708 can use the following flags in the rare cases when ``fine-tuning'' of
3709 optimizations to be performed is desired.
3712 @item -fno-default-inline
3713 @opindex fno-default-inline
3714 Do not make member functions inline by default merely because they are
3715 defined inside the class scope (C++ only). Otherwise, when you specify
3716 @w{@option{-O}}, member functions defined inside class scope are compiled
3717 inline by default; i.e., you don't need to add @samp{inline} in front of
3718 the member function name.
3720 @item -fno-defer-pop
3721 @opindex fno-defer-pop
3722 Always pop the arguments to each function call as soon as that function
3723 returns. For machines which must pop arguments after a function call,
3724 the compiler normally lets arguments accumulate on the stack for several
3725 function calls and pops them all at once.
3727 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3731 Force memory operands to be copied into registers before doing
3732 arithmetic on them. This produces better code by making all memory
3733 references potential common subexpressions. When they are not common
3734 subexpressions, instruction combination should eliminate the separate
3737 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3740 @opindex fforce-addr
3741 Force memory address constants to be copied into registers before
3742 doing arithmetic on them. This may produce better code just as
3743 @option{-fforce-mem} may.
3745 @item -fomit-frame-pointer
3746 @opindex fomit-frame-pointer
3747 Don't keep the frame pointer in a register for functions that
3748 don't need one. This avoids the instructions to save, set up and
3749 restore frame pointers; it also makes an extra register available
3750 in many functions. @strong{It also makes debugging impossible on
3753 On some machines, such as the VAX, this flag has no effect, because
3754 the standard calling sequence automatically handles the frame pointer
3755 and nothing is saved by pretending it doesn't exist. The
3756 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3757 whether a target machine supports this flag. @xref{Registers,,Register
3758 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3760 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3762 @item -foptimize-sibling-calls
3763 @opindex foptimize-sibling-calls
3764 Optimize sibling and tail recursive calls.
3766 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3770 Don't pay attention to the @code{inline} keyword. Normally this option
3771 is used to keep the compiler from expanding any functions inline.
3772 Note that if you are not optimizing, no functions can be expanded inline.
3774 @item -finline-functions
3775 @opindex finline-functions
3776 Integrate all simple functions into their callers. The compiler
3777 heuristically decides which functions are simple enough to be worth
3778 integrating in this way.
3780 If all calls to a given function are integrated, and the function is
3781 declared @code{static}, then the function is normally not output as
3782 assembler code in its own right.
3784 Enabled at level @option{-O3}.
3786 @item -finline-limit=@var{n}
3787 @opindex finline-limit
3788 By default, GCC limits the size of functions that can be inlined. This flag
3789 allows the control of this limit for functions that are explicitly marked as
3790 inline (i.e., marked with the inline keyword or defined within the class
3791 definition in c++). @var{n} is the size of functions that can be inlined in
3792 number of pseudo instructions (not counting parameter handling). The default
3793 value of @var{n} is 600.
3794 Increasing this value can result in more inlined code at
3795 the cost of compilation time and memory consumption. Decreasing usually makes
3796 the compilation faster and less code will be inlined (which presumably
3797 means slower programs). This option is particularly useful for programs that
3798 use inlining heavily such as those based on recursive templates with C++.
3800 Inlining is actually controlled by a number of parameters, which may be
3801 specified individually by using @option{--param @var{name}=@var{value}}.
3802 The @option{-finline-limit=@var{n}} option sets some of these parameters
3806 @item max-inline-insns-single
3807 is set to @var{n}/2.
3808 @item max-inline-insns-auto
3809 is set to @var{n}/2.
3810 @item min-inline-insns
3811 is set to 130 or @var{n}/4, whichever is smaller.
3812 @item max-inline-insns-rtl
3816 See below for a documentation of the individual
3817 parameters controlling inlining.
3819 @emph{Note:} pseudo instruction represents, in this particular context, an
3820 abstract measurement of function's size. In no way, it represents a count
3821 of assembly instructions and as such its exact meaning might change from one
3822 release to an another.
3824 @item -fkeep-inline-functions
3825 @opindex fkeep-inline-functions
3826 Even if all calls to a given function are integrated, and the function
3827 is declared @code{static}, nevertheless output a separate run-time
3828 callable version of the function. This switch does not affect
3829 @code{extern inline} functions.
3831 @item -fkeep-static-consts
3832 @opindex fkeep-static-consts
3833 Emit variables declared @code{static const} when optimization isn't turned
3834 on, even if the variables aren't referenced.
3836 GCC enables this option by default. If you want to force the compiler to
3837 check if the variable was referenced, regardless of whether or not
3838 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3840 @item -fmerge-constants
3841 Attempt to merge identical constants (string constants and floating point
3842 constants) across compilation units.
3844 This option is the default for optimized compilation if the assembler and
3845 linker support it. Use @option{-fno-merge-constants} to inhibit this
3848 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3850 @item -fmerge-all-constants
3851 Attempt to merge identical constants and identical variables.
3853 This option implies @option{-fmerge-constants}. In addition to
3854 @option{-fmerge-constants} this considers e.g. even constant initialized
3855 arrays or initialized constant variables with integral or floating point
3856 types. Languages like C or C++ require each non-automatic variable to
3857 have distinct location, so using this option will result in non-conforming
3862 Use a graph coloring register allocator. Currently this option is meant
3863 only for testing. Users should not specify this option, since it is not
3864 yet ready for production use.
3866 @item -fno-branch-count-reg
3867 @opindex fno-branch-count-reg
3868 Do not use ``decrement and branch'' instructions on a count register,
3869 but instead generate a sequence of instructions that decrement a
3870 register, compare it against zero, then branch based upon the result.
3871 This option is only meaningful on architectures that support such
3872 instructions, which include x86, PowerPC, IA-64 and S/390.
3874 The default is @option{-fbranch-count-reg}, enabled when
3875 @option{-fstrength-reduce} is enabled.
3877 @item -fno-function-cse
3878 @opindex fno-function-cse
3879 Do not put function addresses in registers; make each instruction that
3880 calls a constant function contain the function's address explicitly.
3882 This option results in less efficient code, but some strange hacks
3883 that alter the assembler output may be confused by the optimizations
3884 performed when this option is not used.
3886 The default is @option{-ffunction-cse}
3888 @item -fno-zero-initialized-in-bss
3889 @opindex fno-zero-initialized-in-bss
3890 If the target supports a BSS section, GCC by default puts variables that
3891 are initialized to zero into BSS@. This can save space in the resulting
3894 This option turns off this behavior because some programs explicitly
3895 rely on variables going to the data section. E.g., so that the
3896 resulting executable can find the beginning of that section and/or make
3897 assumptions based on that.
3899 The default is @option{-fzero-initialized-in-bss}.
3901 @item -fstrength-reduce
3902 @opindex fstrength-reduce
3903 Perform the optimizations of loop strength reduction and
3904 elimination of iteration variables.
3906 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3908 @item -fthread-jumps
3909 @opindex fthread-jumps
3910 Perform optimizations where we check to see if a jump branches to a
3911 location where another comparison subsumed by the first is found. If
3912 so, the first branch is redirected to either the destination of the
3913 second branch or a point immediately following it, depending on whether
3914 the condition is known to be true or false.
3916 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3918 @item -fcse-follow-jumps
3919 @opindex fcse-follow-jumps
3920 In common subexpression elimination, scan through jump instructions
3921 when the target of the jump is not reached by any other path. For
3922 example, when CSE encounters an @code{if} statement with an
3923 @code{else} clause, CSE will follow the jump when the condition
3926 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3928 @item -fcse-skip-blocks
3929 @opindex fcse-skip-blocks
3930 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3931 follow jumps which conditionally skip over blocks. When CSE
3932 encounters a simple @code{if} statement with no else clause,
3933 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3934 body of the @code{if}.
3936 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3938 @item -frerun-cse-after-loop
3939 @opindex frerun-cse-after-loop
3940 Re-run common subexpression elimination after loop optimizations has been
3943 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3945 @item -frerun-loop-opt
3946 @opindex frerun-loop-opt
3947 Run the loop optimizer twice.
3949 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3953 Perform a global common subexpression elimination pass.
3954 This pass also performs global constant and copy propagation.
3956 @emph{Note:} When compiling a program using computed gotos, a GCC
3957 extension, you may get better runtime performance if you disable
3958 the global common subexpression elimination pass by adding
3959 @option{-fno-gcse} to the command line.
3961 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3965 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3966 attempt to move loads which are only killed by stores into themselves. This
3967 allows a loop containing a load/store sequence to be changed to a load outside
3968 the loop, and a copy/store within the loop.
3970 Enabled by default when gcse is enabled.
3974 When @option{-fgcse-sm} is enabled, a store motion pass is run after
3975 global common subexpression elimination. This pass will attempt to move
3976 stores out of loops. When used in conjunction with @option{-fgcse-lm},
3977 loops containing a load/store sequence can be changed to a load before
3978 the loop and a store after the loop.
3980 Enabled by default when gcse is enabled.
3984 When @option{-fgcse-las} is enabled, the global common subexpression
3985 elimination pass eliminates redundant loads that come after stores to the
3986 same memory location (both partial and full redundancies).
3988 Enabled by default when gcse is enabled.
3990 @item -floop-optimize
3991 @opindex floop-optimize
3992 Perform loop optimizations: move constant expressions out of loops, simplify
3993 exit test conditions and optionally do strength-reduction and loop unrolling as
3996 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3998 @item -fcrossjumping
3999 @opindex crossjumping
4000 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4001 resulting code may or may not perform better than without cross-jumping.
4003 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4005 @item -fif-conversion
4006 @opindex if-conversion
4007 Attempt to transform conditional jumps into branch-less equivalents. This
4008 include use of conditional moves, min, max, set flags and abs instructions, and
4009 some tricks doable by standard arithmetics. The use of conditional execution
4010 on chips where it is available is controlled by @code{if-conversion2}.
4012 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4014 @item -fif-conversion2
4015 @opindex if-conversion2
4016 Use conditional execution (where available) to transform conditional jumps into
4017 branch-less equivalents.
4019 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4021 @item -fdelete-null-pointer-checks
4022 @opindex fdelete-null-pointer-checks
4023 Use global dataflow analysis to identify and eliminate useless checks
4024 for null pointers. The compiler assumes that dereferencing a null
4025 pointer would have halted the program. If a pointer is checked after
4026 it has already been dereferenced, it cannot be null.
4028 In some environments, this assumption is not true, and programs can
4029 safely dereference null pointers. Use
4030 @option{-fno-delete-null-pointer-checks} to disable this optimization
4031 for programs which depend on that behavior.
4033 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4035 @item -fexpensive-optimizations
4036 @opindex fexpensive-optimizations
4037 Perform a number of minor optimizations that are relatively expensive.
4039 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4041 @item -foptimize-register-move
4043 @opindex foptimize-register-move
4045 Attempt to reassign register numbers in move instructions and as
4046 operands of other simple instructions in order to maximize the amount of
4047 register tying. This is especially helpful on machines with two-operand
4050 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4053 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4055 @item -fdelayed-branch
4056 @opindex fdelayed-branch
4057 If supported for the target machine, attempt to reorder instructions
4058 to exploit instruction slots available after delayed branch
4061 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4063 @item -fschedule-insns
4064 @opindex fschedule-insns
4065 If supported for the target machine, attempt to reorder instructions to
4066 eliminate execution stalls due to required data being unavailable. This
4067 helps machines that have slow floating point or memory load instructions
4068 by allowing other instructions to be issued until the result of the load
4069 or floating point instruction is required.
4071 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4073 @item -fschedule-insns2
4074 @opindex fschedule-insns2
4075 Similar to @option{-fschedule-insns}, but requests an additional pass of
4076 instruction scheduling after register allocation has been done. This is
4077 especially useful on machines with a relatively small number of
4078 registers and where memory load instructions take more than one cycle.
4080 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4082 @item -fno-sched-interblock
4083 @opindex fno-sched-interblock
4084 Don't schedule instructions across basic blocks. This is normally
4085 enabled by default when scheduling before register allocation, i.e.@:
4086 with @option{-fschedule-insns} or at @option{-O2} or higher.
4088 @item -fno-sched-spec
4089 @opindex fno-sched-spec
4090 Don't allow speculative motion of non-load instructions. This is normally
4091 enabled by default when scheduling before register allocation, i.e.@:
4092 with @option{-fschedule-insns} or at @option{-O2} or higher.
4094 @item -fsched-spec-load
4095 @opindex fsched-spec-load
4096 Allow speculative motion of some load instructions. This only makes
4097 sense when scheduling before register allocation, i.e.@: with
4098 @option{-fschedule-insns} or at @option{-O2} or higher.
4100 @item -fsched-spec-load-dangerous
4101 @opindex fsched-spec-load-dangerous
4102 Allow speculative motion of more load instructions. This only makes
4103 sense when scheduling before register allocation, i.e.@: with
4104 @option{-fschedule-insns} or at @option{-O2} or higher.
4106 @item -fsched-stalled-insns=@var{n}
4107 @opindex fsched-stalled-insns
4108 Define how many insns (if any) can be moved prematurely from the queue
4109 of stalled insns into the ready list, during the second scheduling pass.
4111 @item -fsched-stalled-insns-dep=@var{n}
4112 @opindex fsched-stalled-insns-dep
4113 Define how many insn groups (cycles) will be examined for a dependency
4114 on a stalled insn that is candidate for premature removal from the queue
4115 of stalled insns. Has an effect only during the second scheduling pass,
4116 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4118 @item -fsched2-use-superblocks
4119 @opindex fsched2-use-superblocks
4120 When scheduling after register allocation, do use superblock scheduling
4121 algorithm. Superblock scheduling allows motion across basic block boundaries
4122 resulting on faster schedules. This option is experimental, as not all machine
4123 descriptions used by GCC model the CPU closely enough to avoid unreliable
4124 results from the algorithm.
4126 This only makes sense when scheduling after register allocation, i.e.@: with
4127 @option{-fschedule-insns2} or at @option{-O2} or higher.
4129 @item -fsched2-use-traces
4130 @opindex fsched2-use-traces
4131 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4132 allocation and additionally perform code duplication in order to increase the
4133 size of superblocks using tracer pass. See @option{-ftracer} for details on
4136 This mode should produce faster but significantly longer programs. Also
4137 without @code{-fbranch-probabilities} the traces constructed may not match the
4138 reality and hurt the performance. This only makes
4139 sense when scheduling after register allocation, i.e.@: with
4140 @option{-fschedule-insns2} or at @option{-O2} or higher.
4142 @item -fcaller-saves
4143 @opindex fcaller-saves
4144 Enable values to be allocated in registers that will be clobbered by
4145 function calls, by emitting extra instructions to save and restore the
4146 registers around such calls. Such allocation is done only when it
4147 seems to result in better code than would otherwise be produced.
4149 This option is always enabled by default on certain machines, usually
4150 those which have no call-preserved registers to use instead.
4152 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4154 @item -fmove-all-movables
4155 @opindex fmove-all-movables
4156 Forces all invariant computations in loops to be moved
4159 @item -freduce-all-givs
4160 @opindex freduce-all-givs
4161 Forces all general-induction variables in loops to be
4164 @emph{Note:} When compiling programs written in Fortran,
4165 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4166 by default when you use the optimizer.
4168 These options may generate better or worse code; results are highly
4169 dependent on the structure of loops within the source code.
4171 These two options are intended to be removed someday, once
4172 they have helped determine the efficacy of various
4173 approaches to improving loop optimizations.
4175 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
4176 know how use of these options affects
4177 the performance of your production code.
4178 We're very interested in code that runs @emph{slower}
4179 when these options are @emph{enabled}.
4182 @itemx -fno-peephole2
4183 @opindex fno-peephole
4184 @opindex fno-peephole2
4185 Disable any machine-specific peephole optimizations. The difference
4186 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4187 are implemented in the compiler; some targets use one, some use the
4188 other, a few use both.
4190 @option{-fpeephole} is enabled by default.
4191 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4193 @item -fno-guess-branch-probability
4194 @opindex fno-guess-branch-probability
4195 Do not guess branch probabilities using a randomized model.
4197 Sometimes GCC will opt to use a randomized model to guess branch
4198 probabilities, when none are available from either profiling feedback
4199 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4200 different runs of the compiler on the same program may produce different
4203 In a hard real-time system, people don't want different runs of the
4204 compiler to produce code that has different behavior; minimizing
4205 non-determinism is of paramount import. This switch allows users to
4206 reduce non-determinism, possibly at the expense of inferior
4209 The default is @option{-fguess-branch-probability} at levels
4210 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4212 @item -freorder-blocks
4213 @opindex freorder-blocks
4214 Reorder basic blocks in the compiled function in order to reduce number of
4215 taken branches and improve code locality.
4217 Enabled at levels @option{-O2}, @option{-O3}.
4219 @item -freorder-blocks-and-partition
4220 @opindex freorder-blocks-and-partition
4221 In addition to reordering basic blocks in the compiled function, in order
4222 to reduce number of taken branches, partitions hot and cold basic blocks
4223 into separate sections of the assembly and .o files, to improve
4224 paging and cache locality performance.
4226 @item -freorder-functions
4227 @opindex freorder-functions
4228 Reorder basic blocks in the compiled function in order to reduce number of
4229 taken branches and improve code locality. This is implemented by using special
4230 subsections @code{.text.hot} for most frequently executed functions and
4231 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
4232 the linker so object file format must support named sections and linker must
4233 place them in a reasonable way.
4235 Also profile feedback must be available in to make this option effective. See
4236 @option{-fprofile-arcs} for details.
4238 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4240 @item -fstrict-aliasing
4241 @opindex fstrict-aliasing
4242 Allows the compiler to assume the strictest aliasing rules applicable to
4243 the language being compiled. For C (and C++), this activates
4244 optimizations based on the type of expressions. In particular, an
4245 object of one type is assumed never to reside at the same address as an
4246 object of a different type, unless the types are almost the same. For
4247 example, an @code{unsigned int} can alias an @code{int}, but not a
4248 @code{void*} or a @code{double}. A character type may alias any other
4251 Pay special attention to code like this:
4264 The practice of reading from a different union member than the one most
4265 recently written to (called ``type-punning'') is common. Even with
4266 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4267 is accessed through the union type. So, the code above will work as
4268 expected. However, this code might not:
4279 Every language that wishes to perform language-specific alias analysis
4280 should define a function that computes, given an @code{tree}
4281 node, an alias set for the node. Nodes in different alias sets are not
4282 allowed to alias. For an example, see the C front-end function
4283 @code{c_get_alias_set}.
4285 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4287 @item -falign-functions
4288 @itemx -falign-functions=@var{n}
4289 @opindex falign-functions
4290 Align the start of functions to the next power-of-two greater than
4291 @var{n}, skipping up to @var{n} bytes. For instance,
4292 @option{-falign-functions=32} aligns functions to the next 32-byte
4293 boundary, but @option{-falign-functions=24} would align to the next
4294 32-byte boundary only if this can be done by skipping 23 bytes or less.
4296 @option{-fno-align-functions} and @option{-falign-functions=1} are
4297 equivalent and mean that functions will not be aligned.
4299 Some assemblers only support this flag when @var{n} is a power of two;
4300 in that case, it is rounded up.
4302 If @var{n} is not specified or is zero, use a machine-dependent default.
4304 Enabled at levels @option{-O2}, @option{-O3}.
4306 @item -falign-labels
4307 @itemx -falign-labels=@var{n}
4308 @opindex falign-labels
4309 Align all branch targets to a power-of-two boundary, skipping up to
4310 @var{n} bytes like @option{-falign-functions}. This option can easily
4311 make code slower, because it must insert dummy operations for when the
4312 branch target is reached in the usual flow of the code.
4314 @option{-fno-align-labels} and @option{-falign-labels=1} are
4315 equivalent and mean that labels will not be aligned.
4317 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4318 are greater than this value, then their values are used instead.
4320 If @var{n} is not specified or is zero, use a machine-dependent default
4321 which is very likely to be @samp{1}, meaning no alignment.
4323 Enabled at levels @option{-O2}, @option{-O3}.
4326 @itemx -falign-loops=@var{n}
4327 @opindex falign-loops
4328 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4329 like @option{-falign-functions}. The hope is that the loop will be
4330 executed many times, which will make up for any execution of the dummy
4333 @option{-fno-align-loops} and @option{-falign-loops=1} are
4334 equivalent and mean that loops will not be aligned.
4336 If @var{n} is not specified or is zero, use a machine-dependent default.
4338 Enabled at levels @option{-O2}, @option{-O3}.
4341 @itemx -falign-jumps=@var{n}
4342 @opindex falign-jumps
4343 Align branch targets to a power-of-two boundary, for branch targets
4344 where the targets can only be reached by jumping, skipping up to @var{n}
4345 bytes like @option{-falign-functions}. In this case, no dummy operations
4348 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4349 equivalent and mean that loops will not be aligned.
4351 If @var{n} is not specified or is zero, use a machine-dependent default.
4353 Enabled at levels @option{-O2}, @option{-O3}.
4355 @item -funit-at-a-time
4356 @opindex funit-at-a-time
4357 Parse the whole compilation unit before starting to produce code.
4358 This allows some extra optimizations to take place but consumes
4359 more memory (in general). There are some compatibility issues
4360 with @emph{unit-at-at-time} mode:
4363 enabling @emph{unit-at-a-time} mode may change the order
4364 in which functions, variables, and top-level @code{asm} statements
4365 are emitted, and will likely break code relying on some particular
4366 ordering. The majority of such top-level @code{asm} statements,
4367 though, can be replaced by @code{section} attributes.
4370 @emph{unit-at-a-time} mode removes unreferenced static variables
4371 and functions are removed. This may result in undefined references
4372 when an @code{asm} statement refers directly to variables or functions
4373 that are otherwise unused. In that case either the variable/function
4374 shall be listed as an operand of the @code{asm} statement operand or,
4375 in the case of top-level @code{asm} statements the attribute @code{used}
4376 shall be used on the declaration.
4379 Static functions now can use non-standard passing conventions that
4380 may break @code{asm} statements calling functions directly. Again,
4381 attribute @code{used} will prevent this behavior.
4384 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
4385 but this scheme may not be supported by future releases of GCC.
4387 Enabled at levels @option{-O2}, @option{-O3}.
4391 Constructs webs as commonly used for register allocation purposes and assign
4392 each web individual pseudo register. This allows our register allocation pass
4393 to operate on pseudos directly, but also strengthens several other optimization
4394 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
4395 however, make debugging impossible, since variables will no longer stay in a
4398 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os},
4399 on targets where the default format for debugging information supports
4402 @item -fno-cprop-registers
4403 @opindex fno-cprop-registers
4404 After register allocation and post-register allocation instruction splitting,
4405 we perform a copy-propagation pass to try to reduce scheduling dependencies
4406 and occasionally eliminate the copy.
4408 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4410 @item -fprofile-generate
4411 @opindex fprofile-generate
4413 Enable options usually used for instrumenting application to produce
4414 profile useful for later recompilation with profile feedback based
4415 optimization. You must use @code{-fprofile-generate} both when
4416 compiling and when linking your program.
4418 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
4421 @opindex fprofile-use
4422 Enable profile feedback directed optimizations, and optimizations
4423 generally profitable only with profile feedback available.
4425 The following options are enabled: @code{-fbranch-probabilities},
4426 @code{-fvpt}, @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}.
4430 The following options control compiler behavior regarding floating
4431 point arithmetic. These options trade off between speed and
4432 correctness. All must be specifically enabled.
4436 @opindex ffloat-store
4437 Do not store floating point variables in registers, and inhibit other
4438 options that might change whether a floating point value is taken from a
4441 @cindex floating point precision
4442 This option prevents undesirable excess precision on machines such as
4443 the 68000 where the floating registers (of the 68881) keep more
4444 precision than a @code{double} is supposed to have. Similarly for the
4445 x86 architecture. For most programs, the excess precision does only
4446 good, but a few programs rely on the precise definition of IEEE floating
4447 point. Use @option{-ffloat-store} for such programs, after modifying
4448 them to store all pertinent intermediate computations into variables.
4452 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4453 @option{-fno-trapping-math}, @option{-ffinite-math-only},
4454 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
4456 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4458 This option should never be turned on by any @option{-O} option since
4459 it can result in incorrect output for programs which depend on
4460 an exact implementation of IEEE or ISO rules/specifications for
4463 @item -fno-math-errno
4464 @opindex fno-math-errno
4465 Do not set ERRNO after calling math functions that are executed
4466 with a single instruction, e.g., sqrt. A program that relies on
4467 IEEE exceptions for math error handling may want to use this flag
4468 for speed while maintaining IEEE arithmetic compatibility.
4470 This option should never be turned on by any @option{-O} option since
4471 it can result in incorrect output for programs which depend on
4472 an exact implementation of IEEE or ISO rules/specifications for
4475 The default is @option{-fmath-errno}.
4477 @item -funsafe-math-optimizations
4478 @opindex funsafe-math-optimizations
4479 Allow optimizations for floating-point arithmetic that (a) assume
4480 that arguments and results are valid and (b) may violate IEEE or
4481 ANSI standards. When used at link-time, it may include libraries
4482 or startup files that change the default FPU control word or other
4483 similar optimizations.
4485 This option should never be turned on by any @option{-O} option since
4486 it can result in incorrect output for programs which depend on
4487 an exact implementation of IEEE or ISO rules/specifications for
4490 The default is @option{-fno-unsafe-math-optimizations}.
4492 @item -ffinite-math-only
4493 @opindex ffinite-math-only
4494 Allow optimizations for floating-point arithmetic that assume
4495 that arguments and results are not NaNs or +-Infs.
4497 This option should never be turned on by any @option{-O} option since
4498 it can result in incorrect output for programs which depend on
4499 an exact implementation of IEEE or ISO rules/specifications.
4501 The default is @option{-fno-finite-math-only}.
4503 @item -fno-trapping-math
4504 @opindex fno-trapping-math
4505 Compile code assuming that floating-point operations cannot generate
4506 user-visible traps. These traps include division by zero, overflow,
4507 underflow, inexact result and invalid operation. This option implies
4508 @option{-fno-signaling-nans}. Setting this option may allow faster
4509 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4511 This option should never be turned on by any @option{-O} option since
4512 it can result in incorrect output for programs which depend on
4513 an exact implementation of IEEE or ISO rules/specifications for
4516 The default is @option{-ftrapping-math}.
4518 @item -frounding-math
4519 @opindex frounding-math
4520 Disable transformations and optimizations that assume default floating
4521 point rounding behavior. This is round-to-zero for all floating point
4522 to integer conversions, and round-to-nearest for all other arithmetic
4523 truncations. This option should be specified for programs that change
4524 the FP rounding mode dynamically, or that may be executed with a
4525 non-default rounding mode. This option disables constant folding of
4526 floating point expressions at compile-time (which may be affected by
4527 rounding mode) and arithmetic transformations that are unsafe in the
4528 presence of sign-dependent rounding modes.
4530 The default is @option{-fno-rounding-math}.
4532 This option is experimental and does not currently guarantee to
4533 disable all GCC optimizations that are affected by rounding mode.
4534 Future versions of GCC may provide finer control of this setting
4535 using C99's @code{FENV_ACCESS} pragma. This command line option
4536 will be used to specify the default state for @code{FENV_ACCESS}.
4538 @item -fsignaling-nans
4539 @opindex fsignaling-nans
4540 Compile code assuming that IEEE signaling NaNs may generate user-visible
4541 traps during floating-point operations. Setting this option disables
4542 optimizations that may change the number of exceptions visible with
4543 signaling NaNs. This option implies @option{-ftrapping-math}.
4545 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4548 The default is @option{-fno-signaling-nans}.
4550 This option is experimental and does not currently guarantee to
4551 disable all GCC optimizations that affect signaling NaN behavior.
4553 @item -fsingle-precision-constant
4554 @opindex fsingle-precision-constant
4555 Treat floating point constant as single precision constant instead of
4556 implicitly converting it to double precision constant.
4561 The following options control optimizations that may improve
4562 performance, but are not enabled by any @option{-O} options. This
4563 section includes experimental options that may produce broken code.
4566 @item -fbranch-probabilities
4567 @opindex fbranch-probabilities
4568 After running a program compiled with @option{-fprofile-arcs}
4569 (@pxref{Debugging Options,, Options for Debugging Your Program or
4570 @command{gcc}}), you can compile it a second time using
4571 @option{-fbranch-probabilities}, to improve optimizations based on
4572 the number of times each branch was taken. When the program
4573 compiled with @option{-fprofile-arcs} exits it saves arc execution
4574 counts to a file called @file{@var{sourcename}.gcda} for each source
4575 file The information in this data file is very dependent on the
4576 structure of the generated code, so you must use the same source code
4577 and the same optimization options for both compilations.
4579 With @option{-fbranch-probabilities}, GCC puts a
4580 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4581 These can be used to improve optimization. Currently, they are only
4582 used in one place: in @file{reorg.c}, instead of guessing which path a
4583 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4584 exactly determine which path is taken more often.
4586 @item -fprofile-values
4587 @opindex fprofile-values
4588 If combined with @option{-fprofile-arcs}, it adds code so that some
4589 data about values of expressions in the program is gathered.
4591 With @option{-fbranch-probabilities}, it reads back the data gathered
4592 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
4593 notes to instructions for their later usage in optimizations.
4595 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
4599 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
4600 a code to gather information about values of expressions.
4602 With @option{-fbranch-probabilities}, it reads back the data gathered
4603 and actually performs the optimizations based on them.
4604 Currently the optimizations include specialization of division operation
4605 using the knowledge about the value of the denominator.
4607 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
4609 @item -frename-registers
4610 @opindex frename-registers
4611 Attempt to avoid false dependencies in scheduled code by making use
4612 of registers left over after register allocation. This optimization
4613 will most benefit processors with lots of registers. Depending on the
4614 debug information format adopted by the target, however, it can
4615 make debugging impossible, since variables will no longer stay in
4616 a ``home register''.
4618 Not enabled by default at any level because it has known bugs.
4622 Use a graph coloring register allocator. Currently this option is meant
4623 for testing, so we are interested to hear about miscompilations with
4628 Perform tail duplication to enlarge superblock size. This transformation
4629 simplifies the control flow of the function allowing other optimizations to do
4632 Enabled with @option{-fprofile-use}.
4634 @item -funroll-loops
4635 @opindex funroll-loops
4636 Unroll loops whose number of iterations can be determined at compile time or
4637 upon entry to the loop. @option{-funroll-loops} implies
4638 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4639 (i.e. complete removal of loops with small constant number of iterations).
4640 This option makes code larger, and may or may not make it run faster.
4642 Enabled with @option{-fprofile-use}.
4644 @item -funroll-all-loops
4645 @opindex funroll-all-loops
4646 Unroll all loops, even if their number of iterations is uncertain when
4647 the loop is entered. This usually makes programs run more slowly.
4648 @option{-funroll-all-loops} implies the same options as
4649 @option{-funroll-loops}.
4652 @opindex fpeel-loops
4653 Peels the loops for that there is enough information that they do not
4654 roll much (from profile feedback). It also turns on complete loop peeling
4655 (i.e. complete removal of loops with small constant number of iterations).
4657 Enabled with @option{-fprofile-use}.
4659 @item -funswitch-loops
4660 @opindex funswitch-loops
4661 Move branches with loop invariant conditions out of the loop, with duplicates
4662 of the loop on both branches (modified according to result of the condition).
4664 @item -fold-unroll-loops
4665 @opindex fold-unroll-loops
4666 Unroll loops whose number of iterations can be determined at compile
4667 time or upon entry to the loop, using the old loop unroller whose loop
4668 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4669 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4670 option makes code larger, and may or may not make it run faster.
4672 @item -fold-unroll-all-loops
4673 @opindex fold-unroll-all-loops
4674 Unroll all loops, even if their number of iterations is uncertain when
4675 the loop is entered. This is done using the old loop unroller whose loop
4676 recognition is based on notes from frontend. This usually makes programs run more slowly.
4677 @option{-fold-unroll-all-loops} implies the same options as
4678 @option{-fold-unroll-loops}.
4680 @item -fprefetch-loop-arrays
4681 @opindex fprefetch-loop-arrays
4682 If supported by the target machine, generate instructions to prefetch
4683 memory to improve the performance of loops that access large arrays.
4685 Disabled at level @option{-Os}.
4687 @item -ffunction-sections
4688 @itemx -fdata-sections
4689 @opindex ffunction-sections
4690 @opindex fdata-sections
4691 Place each function or data item into its own section in the output
4692 file if the target supports arbitrary sections. The name of the
4693 function or the name of the data item determines the section's name
4696 Use these options on systems where the linker can perform optimizations
4697 to improve locality of reference in the instruction space. Most systems
4698 using the ELF object format and SPARC processors running Solaris 2 have
4699 linkers with such optimizations. AIX may have these optimizations in
4702 Only use these options when there are significant benefits from doing
4703 so. When you specify these options, the assembler and linker will
4704 create larger object and executable files and will also be slower.
4705 You will not be able to use @code{gprof} on all systems if you
4706 specify this option and you may have problems with debugging if
4707 you specify both this option and @option{-g}.
4709 @item -fbranch-target-load-optimize
4710 @opindex fbranch-target-load-optimize
4711 Perform branch target register load optimization before prologue / epilogue
4713 The use of target registers can typically be exposed only during reload,
4714 thus hoisting loads out of loops and doing inter-block scheduling needs
4715 a separate optimization pass.
4717 @item -fbranch-target-load-optimize2
4718 @opindex fbranch-target-load-optimize2
4719 Perform branch target register load optimization after prologue / epilogue
4722 @item -fbtr-bb-exclusive
4723 @opindex fbtr-bb-exclusive
4724 When performing branch target register load optimization, don't reuse
4725 branch target registers in within any basic block.
4727 @item --param @var{name}=@var{value}
4729 In some places, GCC uses various constants to control the amount of
4730 optimization that is done. For example, GCC will not inline functions
4731 that contain more that a certain number of instructions. You can
4732 control some of these constants on the command-line using the
4733 @option{--param} option.
4735 The names of specific parameters, and the meaning of the values, are
4736 tied to the internals of the compiler, and are subject to change
4737 without notice in future releases.
4739 In each case, the @var{value} is an integer. The allowable choices for
4740 @var{name} are given in the following table:
4743 @item max-crossjump-edges
4744 The maximum number of incoming edges to consider for crossjumping.
4745 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
4746 the number of edges incoming to each block. Increasing values mean
4747 more aggressive optimization, making the compile time increase with
4748 probably small improvement in executable size.
4750 @item max-delay-slot-insn-search
4751 The maximum number of instructions to consider when looking for an
4752 instruction to fill a delay slot. If more than this arbitrary number of
4753 instructions is searched, the time savings from filling the delay slot
4754 will be minimal so stop searching. Increasing values mean more
4755 aggressive optimization, making the compile time increase with probably
4756 small improvement in executable run time.
4758 @item max-delay-slot-live-search
4759 When trying to fill delay slots, the maximum number of instructions to
4760 consider when searching for a block with valid live register
4761 information. Increasing this arbitrarily chosen value means more
4762 aggressive optimization, increasing the compile time. This parameter
4763 should be removed when the delay slot code is rewritten to maintain the
4766 @item max-gcse-memory
4767 The approximate maximum amount of memory that will be allocated in
4768 order to perform the global common subexpression elimination
4769 optimization. If more memory than specified is required, the
4770 optimization will not be done.
4772 @item max-gcse-passes
4773 The maximum number of passes of GCSE to run. The default is 1.
4775 @item max-pending-list-length
4776 The maximum number of pending dependencies scheduling will allow
4777 before flushing the current state and starting over. Large functions
4778 with few branches or calls can create excessively large lists which
4779 needlessly consume memory and resources.
4781 @item max-inline-insns-single
4782 Several parameters control the tree inliner used in gcc.
4783 This number sets the maximum number of instructions (counted in GCC's
4784 internal representation) in a single function that the tree inliner
4785 will consider for inlining. This only affects functions declared
4786 inline and methods implemented in a class declaration (C++).
4787 The default value is 500.
4789 @item max-inline-insns-auto
4790 When you use @option{-finline-functions} (included in @option{-O3}),
4791 a lot of functions that would otherwise not be considered for inlining
4792 by the compiler will be investigated. To those functions, a different
4793 (more restrictive) limit compared to functions declared inline can
4795 The default value is 120.
4797 @item large-function-insns
4798 The limit specifying really large functions. For functions greater than this
4799 limit inlining is constrained by @option{--param large-function-growth}.
4800 This parameter is useful primarily to avoid extreme compilation time caused by non-linear
4801 algorithms used by the backend.
4802 This parameter is ignored when @option{-funit-at-a-time} is not used.
4803 The default value is 3000.
4805 @item large-function-growth
4806 Specifies maximal growth of large function caused by inlining in percents.
4807 This parameter is ignored when @option{-funit-at-a-time} is not used.
4808 The default value is 200.
4810 @item inline-unit-growth
4811 Specifies maximal overall growth of the compilation unit caused by inlining.
4812 This parameter is ignored when @option{-funit-at-a-time} is not used.
4813 The default value is 150.
4815 @item max-inline-insns-rtl
4816 For languages that use the RTL inliner (this happens at a later stage
4817 than tree inlining), you can set the maximum allowable size (counted
4818 in RTL instructions) for the RTL inliner with this parameter.
4819 The default value is 600.
4821 @item max-unrolled-insns
4822 The maximum number of instructions that a loop should have if that loop
4823 is unrolled, and if the loop is unrolled, it determines how many times
4824 the loop code is unrolled.
4826 @item max-average-unrolled-insns
4827 The maximum number of instructions biased by probabilities of their execution
4828 that a loop should have if that loop is unrolled, and if the loop is unrolled,
4829 it determines how many times the loop code is unrolled.
4831 @item max-unroll-times
4832 The maximum number of unrollings of a single loop.
4834 @item max-peeled-insns
4835 The maximum number of instructions that a loop should have if that loop
4836 is peeled, and if the loop is peeled, it determines how many times
4837 the loop code is peeled.
4839 @item max-peel-times
4840 The maximum number of peelings of a single loop.
4842 @item max-completely-peeled-insns
4843 The maximum number of insns of a completely peeled loop.
4845 @item max-completely-peel-times
4846 The maximum number of iterations of a loop to be suitable for complete peeling.
4848 @item max-unswitch-insns
4849 The maximum number of insns of an unswitched loop.
4851 @item max-unswitch-level
4852 The maximum number of branches unswitched in a single loop.
4854 @item hot-bb-count-fraction
4855 Select fraction of the maximal count of repetitions of basic block in program
4856 given basic block needs to have to be considered hot.
4858 @item hot-bb-frequency-fraction
4859 Select fraction of the maximal frequency of executions of basic block in
4860 function given basic block needs to have to be considered hot
4862 @item tracer-dynamic-coverage
4863 @itemx tracer-dynamic-coverage-feedback
4865 This value is used to limit superblock formation once the given percentage of
4866 executed instructions is covered. This limits unnecessary code size
4869 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4870 feedback is available. The real profiles (as opposed to statically estimated
4871 ones) are much less balanced allowing the threshold to be larger value.
4873 @item tracer-max-code-growth
4874 Stop tail duplication once code growth has reached given percentage. This is
4875 rather hokey argument, as most of the duplicates will be eliminated later in
4876 cross jumping, so it may be set to much higher values than is the desired code
4879 @item tracer-min-branch-ratio
4881 Stop reverse growth when the reverse probability of best edge is less than this
4882 threshold (in percent).
4884 @item tracer-min-branch-ratio
4885 @itemx tracer-min-branch-ratio-feedback
4887 Stop forward growth if the best edge do have probability lower than this
4890 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
4891 compilation for profile feedback and one for compilation without. The value
4892 for compilation with profile feedback needs to be more conservative (higher) in
4893 order to make tracer effective.
4895 @item max-cse-path-length
4897 Maximum number of basic blocks on path that cse considers. The default is 10.
4899 @item ggc-min-expand
4901 GCC uses a garbage collector to manage its own memory allocation. This
4902 parameter specifies the minimum percentage by which the garbage
4903 collector's heap should be allowed to expand between collections.
4904 Tuning this may improve compilation speed; it has no effect on code
4907 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
4908 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
4909 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
4910 GCC is not able to calculate RAM on a particular platform, the lower
4911 bound of 30% is used. Setting this parameter and
4912 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4913 every opportunity. This is extremely slow, but can be useful for
4916 @item ggc-min-heapsize
4918 Minimum size of the garbage collector's heap before it begins bothering
4919 to collect garbage. The first collection occurs after the heap expands
4920 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4921 tuning this may improve compilation speed, and has no effect on code
4924 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
4925 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
4926 available, the notion of "RAM" is the smallest of actual RAM,
4927 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
4928 RAM on a particular platform, the lower bound is used. Setting this
4929 parameter very large effectively disables garbage collection. Setting
4930 this parameter and @option{ggc-min-expand} to zero causes a full
4931 collection to occur at every opportunity.
4933 @item max-reload-search-insns
4934 The maximum number of instruction reload should look backward for equivalent
4935 register. Increasing values mean more aggressive optimization, making the
4936 compile time increase with probably slightly better performance. The default
4939 @item max-cselib-memory-location
4940 The maximum number of memory locations cselib should take into acount.
4941 Increasing values mean more aggressive optimization, making the compile time
4942 increase with probably slightly better performance. The default value is 500.
4944 @item reorder-blocks-duplicate
4945 @itemx reorder-blocks-duplicate-feedback
4947 Used by basic block reordering pass to decide whether to use unconditional
4948 branch or duplicate the code on its destination. Code is duplicated when its
4949 estimated size is smaller than this value multiplied by the estimated size of
4950 unconditional jump in the hot spots of the program.
4952 The @option{reorder-block-duplicate-feedback} is used only when profile
4953 feedback is available and may be set to higher values than
4954 @option{reorder-block-duplicate} since information about the hot spots is more
4957 @item max-sched-region-blocks
4958 The maximum number of blocks in a region to be considered for
4959 interblock scheduling. The default value is 10.
4961 @item max-sched-region-insns
4962 The maximum number of insns in a region to be considered for
4963 interblock scheduling. The default value is 100.
4967 @node Preprocessor Options
4968 @section Options Controlling the Preprocessor
4969 @cindex preprocessor options
4970 @cindex options, preprocessor
4972 These options control the C preprocessor, which is run on each C source
4973 file before actual compilation.
4975 If you use the @option{-E} option, nothing is done except preprocessing.
4976 Some of these options make sense only together with @option{-E} because
4977 they cause the preprocessor output to be unsuitable for actual
4982 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4983 and pass @var{option} directly through to the preprocessor. If
4984 @var{option} contains commas, it is split into multiple options at the
4985 commas. However, many options are modified, translated or interpreted
4986 by the compiler driver before being passed to the preprocessor, and
4987 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4988 interface is undocumented and subject to change, so whenever possible
4989 you should avoid using @option{-Wp} and let the driver handle the
4992 @item -Xpreprocessor @var{option}
4993 @opindex preprocessor
4994 Pass @var{option} as an option to the preprocessor. You can use this to
4995 supply system-specific preprocessor options which GCC does not know how to
4998 If you want to pass an option that takes an argument, you must use
4999 @option{-Xpreprocessor} twice, once for the option and once for the argument.
5002 @include cppopts.texi
5004 @node Assembler Options
5005 @section Passing Options to the Assembler
5007 @c prevent bad page break with this line
5008 You can pass options to the assembler.
5011 @item -Wa,@var{option}
5013 Pass @var{option} as an option to the assembler. If @var{option}
5014 contains commas, it is split into multiple options at the commas.
5016 @item -Xassembler @var{option}
5018 Pass @var{option} as an option to the assembler. You can use this to
5019 supply system-specific assembler options which GCC does not know how to
5022 If you want to pass an option that takes an argument, you must use
5023 @option{-Xassembler} twice, once for the option and once for the argument.
5028 @section Options for Linking
5029 @cindex link options
5030 @cindex options, linking
5032 These options come into play when the compiler links object files into
5033 an executable output file. They are meaningless if the compiler is
5034 not doing a link step.
5038 @item @var{object-file-name}
5039 A file name that does not end in a special recognized suffix is
5040 considered to name an object file or library. (Object files are
5041 distinguished from libraries by the linker according to the file
5042 contents.) If linking is done, these object files are used as input
5051 If any of these options is used, then the linker is not run, and
5052 object file names should not be used as arguments. @xref{Overall
5056 @item -l@var{library}
5057 @itemx -l @var{library}
5059 Search the library named @var{library} when linking. (The second
5060 alternative with the library as a separate argument is only for
5061 POSIX compliance and is not recommended.)
5063 It makes a difference where in the command you write this option; the
5064 linker searches and processes libraries and object files in the order they
5065 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
5066 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
5067 to functions in @samp{z}, those functions may not be loaded.
5069 The linker searches a standard list of directories for the library,
5070 which is actually a file named @file{lib@var{library}.a}. The linker
5071 then uses this file as if it had been specified precisely by name.
5073 The directories searched include several standard system directories
5074 plus any that you specify with @option{-L}.
5076 Normally the files found this way are library files---archive files
5077 whose members are object files. The linker handles an archive file by
5078 scanning through it for members which define symbols that have so far
5079 been referenced but not defined. But if the file that is found is an
5080 ordinary object file, it is linked in the usual fashion. The only
5081 difference between using an @option{-l} option and specifying a file name
5082 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
5083 and searches several directories.
5087 You need this special case of the @option{-l} option in order to
5088 link an Objective-C program.
5091 @opindex nostartfiles
5092 Do not use the standard system startup files when linking.
5093 The standard system libraries are used normally, unless @option{-nostdlib}
5094 or @option{-nodefaultlibs} is used.
5096 @item -nodefaultlibs
5097 @opindex nodefaultlibs
5098 Do not use the standard system libraries when linking.
5099 Only the libraries you specify will be passed to the linker.
5100 The standard startup files are used normally, unless @option{-nostartfiles}
5101 is used. The compiler may generate calls to memcmp, memset, and memcpy
5102 for System V (and ISO C) environments or to bcopy and bzero for
5103 BSD environments. These entries are usually resolved by entries in
5104 libc. These entry points should be supplied through some other
5105 mechanism when this option is specified.
5109 Do not use the standard system startup files or libraries when linking.
5110 No startup files and only the libraries you specify will be passed to
5111 the linker. The compiler may generate calls to memcmp, memset, and memcpy
5112 for System V (and ISO C) environments or to bcopy and bzero for
5113 BSD environments. These entries are usually resolved by entries in
5114 libc. These entry points should be supplied through some other
5115 mechanism when this option is specified.
5117 @cindex @option{-lgcc}, use with @option{-nostdlib}
5118 @cindex @option{-nostdlib} and unresolved references
5119 @cindex unresolved references and @option{-nostdlib}
5120 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5121 @cindex @option{-nodefaultlibs} and unresolved references
5122 @cindex unresolved references and @option{-nodefaultlibs}
5123 One of the standard libraries bypassed by @option{-nostdlib} and
5124 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5125 that GCC uses to overcome shortcomings of particular machines, or special
5126 needs for some languages.
5127 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5128 Collection (GCC) Internals},
5129 for more discussion of @file{libgcc.a}.)
5130 In most cases, you need @file{libgcc.a} even when you want to avoid
5131 other standard libraries. In other words, when you specify @option{-nostdlib}
5132 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5133 This ensures that you have no unresolved references to internal GCC
5134 library subroutines. (For example, @samp{__main}, used to ensure C++
5135 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5136 GNU Compiler Collection (GCC) Internals}.)
5140 Produce a position independent executable on targets which support it.
5141 For predictable results, you must also specify the same set of options
5142 that were used to generate code (@option{-fpie}, @option{-fPIE},
5143 or model suboptions) when you specify this option.
5147 Remove all symbol table and relocation information from the executable.
5151 On systems that support dynamic linking, this prevents linking with the shared
5152 libraries. On other systems, this option has no effect.
5156 Produce a shared object which can then be linked with other objects to
5157 form an executable. Not all systems support this option. For predictable
5158 results, you must also specify the same set of options that were used to
5159 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5160 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5161 needs to build supplementary stub code for constructors to work. On
5162 multi-libbed systems, @samp{gcc -shared} must select the correct support
5163 libraries to link against. Failing to supply the correct flags may lead
5164 to subtle defects. Supplying them in cases where they are not necessary
5167 @item -shared-libgcc
5168 @itemx -static-libgcc
5169 @opindex shared-libgcc
5170 @opindex static-libgcc
5171 On systems that provide @file{libgcc} as a shared library, these options
5172 force the use of either the shared or static version respectively.
5173 If no shared version of @file{libgcc} was built when the compiler was
5174 configured, these options have no effect.
5176 There are several situations in which an application should use the
5177 shared @file{libgcc} instead of the static version. The most common
5178 of these is when the application wishes to throw and catch exceptions
5179 across different shared libraries. In that case, each of the libraries
5180 as well as the application itself should use the shared @file{libgcc}.
5182 Therefore, the G++ and GCJ drivers automatically add
5183 @option{-shared-libgcc} whenever you build a shared library or a main
5184 executable, because C++ and Java programs typically use exceptions, so
5185 this is the right thing to do.
5187 If, instead, you use the GCC driver to create shared libraries, you may
5188 find that they will not always be linked with the shared @file{libgcc}.
5189 If GCC finds, at its configuration time, that you have a GNU linker that
5190 does not support option @option{--eh-frame-hdr}, it will link the shared
5191 version of @file{libgcc} into shared libraries by default. Otherwise,
5192 it will take advantage of the linker and optimize away the linking with
5193 the shared version of @file{libgcc}, linking with the static version of
5194 libgcc by default. This allows exceptions to propagate through such
5195 shared libraries, without incurring relocation costs at library load
5198 However, if a library or main executable is supposed to throw or catch
5199 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5200 for the languages used in the program, or using the option
5201 @option{-shared-libgcc}, such that it is linked with the shared
5206 Bind references to global symbols when building a shared object. Warn
5207 about any unresolved references (unless overridden by the link editor
5208 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5211 @item -Xlinker @var{option}
5213 Pass @var{option} as an option to the linker. You can use this to
5214 supply system-specific linker options which GCC does not know how to
5217 If you want to pass an option that takes an argument, you must use
5218 @option{-Xlinker} twice, once for the option and once for the argument.
5219 For example, to pass @option{-assert definitions}, you must write
5220 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5221 @option{-Xlinker "-assert definitions"}, because this passes the entire
5222 string as a single argument, which is not what the linker expects.
5224 @item -Wl,@var{option}
5226 Pass @var{option} as an option to the linker. If @var{option} contains
5227 commas, it is split into multiple options at the commas.
5229 @item -u @var{symbol}
5231 Pretend the symbol @var{symbol} is undefined, to force linking of
5232 library modules to define it. You can use @option{-u} multiple times with
5233 different symbols to force loading of additional library modules.
5236 @node Directory Options
5237 @section Options for Directory Search
5238 @cindex directory options
5239 @cindex options, directory search
5242 These options specify directories to search for header files, for
5243 libraries and for parts of the compiler:
5248 Add the directory @var{dir} to the head of the list of directories to be
5249 searched for header files. This can be used to override a system header
5250 file, substituting your own version, since these directories are
5251 searched before the system header file directories. However, you should
5252 not use this option to add directories that contain vendor-supplied
5253 system header files (use @option{-isystem} for that). If you use more than
5254 one @option{-I} option, the directories are scanned in left-to-right
5255 order; the standard system directories come after.
5257 If a standard system include directory, or a directory specified with
5258 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5259 option will be ignored. The directory will still be searched but as a
5260 system directory at its normal position in the system include chain.
5261 This is to ensure that GCC's procedure to fix buggy system headers and
5262 the ordering for the include_next directive are not inadvertently changed.
5263 If you really need to change the search order for system directories,
5264 use the @option{-nostdinc} and/or @option{-isystem} options.
5268 Any directories you specify with @option{-I} options before the @option{-I-}
5269 option are searched only for the case of @samp{#include "@var{file}"};
5270 they are not searched for @samp{#include <@var{file}>}.
5272 If additional directories are specified with @option{-I} options after
5273 the @option{-I-}, these directories are searched for all @samp{#include}
5274 directives. (Ordinarily @emph{all} @option{-I} directories are used
5277 In addition, the @option{-I-} option inhibits the use of the current
5278 directory (where the current input file came from) as the first search
5279 directory for @samp{#include "@var{file}"}. There is no way to
5280 override this effect of @option{-I-}. With @option{-I.} you can specify
5281 searching the directory which was current when the compiler was
5282 invoked. That is not exactly the same as what the preprocessor does
5283 by default, but it is often satisfactory.
5285 @option{-I-} does not inhibit the use of the standard system directories
5286 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5291 Add directory @var{dir} to the list of directories to be searched
5294 @item -B@var{prefix}
5296 This option specifies where to find the executables, libraries,
5297 include files, and data files of the compiler itself.
5299 The compiler driver program runs one or more of the subprograms
5300 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5301 @var{prefix} as a prefix for each program it tries to run, both with and
5302 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5304 For each subprogram to be run, the compiler driver first tries the
5305 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5306 was not specified, the driver tries two standard prefixes, which are
5307 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5308 those results in a file name that is found, the unmodified program
5309 name is searched for using the directories specified in your
5310 @env{PATH} environment variable.
5312 The compiler will check to see if the path provided by the @option{-B}
5313 refers to a directory, and if necessary it will add a directory
5314 separator character at the end of the path.
5316 @option{-B} prefixes that effectively specify directory names also apply
5317 to libraries in the linker, because the compiler translates these
5318 options into @option{-L} options for the linker. They also apply to
5319 includes files in the preprocessor, because the compiler translates these
5320 options into @option{-isystem} options for the preprocessor. In this case,
5321 the compiler appends @samp{include} to the prefix.
5323 The run-time support file @file{libgcc.a} can also be searched for using
5324 the @option{-B} prefix, if needed. If it is not found there, the two
5325 standard prefixes above are tried, and that is all. The file is left
5326 out of the link if it is not found by those means.
5328 Another way to specify a prefix much like the @option{-B} prefix is to use
5329 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5332 As a special kludge, if the path provided by @option{-B} is
5333 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5334 9, then it will be replaced by @file{[dir/]include}. This is to help
5335 with boot-strapping the compiler.
5337 @item -specs=@var{file}
5339 Process @var{file} after the compiler reads in the standard @file{specs}
5340 file, in order to override the defaults that the @file{gcc} driver
5341 program uses when determining what switches to pass to @file{cc1},
5342 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5343 @option{-specs=@var{file}} can be specified on the command line, and they
5344 are processed in order, from left to right.
5350 @section Specifying subprocesses and the switches to pass to them
5353 @command{gcc} is a driver program. It performs its job by invoking a
5354 sequence of other programs to do the work of compiling, assembling and
5355 linking. GCC interprets its command-line parameters and uses these to
5356 deduce which programs it should invoke, and which command-line options
5357 it ought to place on their command lines. This behavior is controlled
5358 by @dfn{spec strings}. In most cases there is one spec string for each
5359 program that GCC can invoke, but a few programs have multiple spec
5360 strings to control their behavior. The spec strings built into GCC can
5361 be overridden by using the @option{-specs=} command-line switch to specify
5364 @dfn{Spec files} are plaintext files that are used to construct spec
5365 strings. They consist of a sequence of directives separated by blank
5366 lines. The type of directive is determined by the first non-whitespace
5367 character on the line and it can be one of the following:
5370 @item %@var{command}
5371 Issues a @var{command} to the spec file processor. The commands that can
5375 @item %include <@var{file}>
5377 Search for @var{file} and insert its text at the current point in the
5380 @item %include_noerr <@var{file}>
5381 @cindex %include_noerr
5382 Just like @samp{%include}, but do not generate an error message if the include
5383 file cannot be found.
5385 @item %rename @var{old_name} @var{new_name}
5387 Rename the spec string @var{old_name} to @var{new_name}.
5391 @item *[@var{spec_name}]:
5392 This tells the compiler to create, override or delete the named spec
5393 string. All lines after this directive up to the next directive or
5394 blank line are considered to be the text for the spec string. If this
5395 results in an empty string then the spec will be deleted. (Or, if the
5396 spec did not exist, then nothing will happened.) Otherwise, if the spec
5397 does not currently exist a new spec will be created. If the spec does
5398 exist then its contents will be overridden by the text of this
5399 directive, unless the first character of that text is the @samp{+}
5400 character, in which case the text will be appended to the spec.
5402 @item [@var{suffix}]:
5403 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5404 and up to the next directive or blank line are considered to make up the
5405 spec string for the indicated suffix. When the compiler encounters an
5406 input file with the named suffix, it will processes the spec string in
5407 order to work out how to compile that file. For example:
5414 This says that any input file whose name ends in @samp{.ZZ} should be
5415 passed to the program @samp{z-compile}, which should be invoked with the
5416 command-line switch @option{-input} and with the result of performing the
5417 @samp{%i} substitution. (See below.)
5419 As an alternative to providing a spec string, the text that follows a
5420 suffix directive can be one of the following:
5423 @item @@@var{language}
5424 This says that the suffix is an alias for a known @var{language}. This is
5425 similar to using the @option{-x} command-line switch to GCC to specify a
5426 language explicitly. For example:
5433 Says that .ZZ files are, in fact, C++ source files.
5436 This causes an error messages saying:
5439 @var{name} compiler not installed on this system.
5443 GCC already has an extensive list of suffixes built into it.
5444 This directive will add an entry to the end of the list of suffixes, but
5445 since the list is searched from the end backwards, it is effectively
5446 possible to override earlier entries using this technique.
5450 GCC has the following spec strings built into it. Spec files can
5451 override these strings or create their own. Note that individual
5452 targets can also add their own spec strings to this list.
5455 asm Options to pass to the assembler
5456 asm_final Options to pass to the assembler post-processor
5457 cpp Options to pass to the C preprocessor
5458 cc1 Options to pass to the C compiler
5459 cc1plus Options to pass to the C++ compiler
5460 endfile Object files to include at the end of the link
5461 link Options to pass to the linker
5462 lib Libraries to include on the command line to the linker
5463 libgcc Decides which GCC support library to pass to the linker
5464 linker Sets the name of the linker
5465 predefines Defines to be passed to the C preprocessor
5466 signed_char Defines to pass to CPP to say whether @code{char} is signed
5468 startfile Object files to include at the start of the link
5471 Here is a small example of a spec file:
5477 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5480 This example renames the spec called @samp{lib} to @samp{old_lib} and
5481 then overrides the previous definition of @samp{lib} with a new one.
5482 The new definition adds in some extra command-line options before
5483 including the text of the old definition.
5485 @dfn{Spec strings} are a list of command-line options to be passed to their
5486 corresponding program. In addition, the spec strings can contain
5487 @samp{%}-prefixed sequences to substitute variable text or to
5488 conditionally insert text into the command line. Using these constructs
5489 it is possible to generate quite complex command lines.
5491 Here is a table of all defined @samp{%}-sequences for spec
5492 strings. Note that spaces are not generated automatically around the
5493 results of expanding these sequences. Therefore you can concatenate them
5494 together or combine them with constant text in a single argument.
5498 Substitute one @samp{%} into the program name or argument.
5501 Substitute the name of the input file being processed.
5504 Substitute the basename of the input file being processed.
5505 This is the substring up to (and not including) the last period
5506 and not including the directory.
5509 This is the same as @samp{%b}, but include the file suffix (text after
5513 Marks the argument containing or following the @samp{%d} as a
5514 temporary file name, so that that file will be deleted if GCC exits
5515 successfully. Unlike @samp{%g}, this contributes no text to the
5518 @item %g@var{suffix}
5519 Substitute a file name that has suffix @var{suffix} and is chosen
5520 once per compilation, and mark the argument in the same way as
5521 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5522 name is now chosen in a way that is hard to predict even when previously
5523 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5524 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5525 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5526 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5527 was simply substituted with a file name chosen once per compilation,
5528 without regard to any appended suffix (which was therefore treated
5529 just like ordinary text), making such attacks more likely to succeed.
5531 @item %u@var{suffix}
5532 Like @samp{%g}, but generates a new temporary file name even if
5533 @samp{%u@var{suffix}} was already seen.
5535 @item %U@var{suffix}
5536 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5537 new one if there is no such last file name. In the absence of any
5538 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5539 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5540 would involve the generation of two distinct file names, one
5541 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5542 simply substituted with a file name chosen for the previous @samp{%u},
5543 without regard to any appended suffix.
5545 @item %j@var{suffix}
5546 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5547 writable, and if save-temps is off; otherwise, substitute the name
5548 of a temporary file, just like @samp{%u}. This temporary file is not
5549 meant for communication between processes, but rather as a junk
5552 @item %|@var{suffix}
5553 @itemx %m@var{suffix}
5554 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5555 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5556 all. These are the two most common ways to instruct a program that it
5557 should read from standard input or write to standard output. If you
5558 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5559 construct: see for example @file{f/lang-specs.h}.
5561 @item %.@var{SUFFIX}
5562 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5563 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5564 terminated by the next space or %.
5567 Marks the argument containing or following the @samp{%w} as the
5568 designated output file of this compilation. This puts the argument
5569 into the sequence of arguments that @samp{%o} will substitute later.
5572 Substitutes the names of all the output files, with spaces
5573 automatically placed around them. You should write spaces
5574 around the @samp{%o} as well or the results are undefined.
5575 @samp{%o} is for use in the specs for running the linker.
5576 Input files whose names have no recognized suffix are not compiled
5577 at all, but they are included among the output files, so they will
5581 Substitutes the suffix for object files. Note that this is
5582 handled specially when it immediately follows @samp{%g, %u, or %U},
5583 because of the need for those to form complete file names. The
5584 handling is such that @samp{%O} is treated exactly as if it had already
5585 been substituted, except that @samp{%g, %u, and %U} do not currently
5586 support additional @var{suffix} characters following @samp{%O} as they would
5587 following, for example, @samp{.o}.
5590 Substitutes the standard macro predefinitions for the
5591 current target machine. Use this when running @code{cpp}.
5594 Like @samp{%p}, but puts @samp{__} before and after the name of each
5595 predefined macro, except for macros that start with @samp{__} or with
5596 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5600 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5601 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5602 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5606 Current argument is the name of a library or startup file of some sort.
5607 Search for that file in a standard list of directories and substitute
5608 the full name found.
5611 Print @var{str} as an error message. @var{str} is terminated by a newline.
5612 Use this when inconsistent options are detected.
5615 Substitute the contents of spec string @var{name} at this point.
5618 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5620 @item %x@{@var{option}@}
5621 Accumulate an option for @samp{%X}.
5624 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5628 Output the accumulated assembler options specified by @option{-Wa}.
5631 Output the accumulated preprocessor options specified by @option{-Wp}.
5634 Process the @code{asm} spec. This is used to compute the
5635 switches to be passed to the assembler.
5638 Process the @code{asm_final} spec. This is a spec string for
5639 passing switches to an assembler post-processor, if such a program is
5643 Process the @code{link} spec. This is the spec for computing the
5644 command line passed to the linker. Typically it will make use of the
5645 @samp{%L %G %S %D and %E} sequences.
5648 Dump out a @option{-L} option for each directory that GCC believes might
5649 contain startup files. If the target supports multilibs then the
5650 current multilib directory will be prepended to each of these paths.
5653 Output the multilib directory with directory separators replaced with
5654 @samp{_}. If multilib directories are not set, or the multilib directory is
5655 @file{.} then this option emits nothing.
5658 Process the @code{lib} spec. This is a spec string for deciding which
5659 libraries should be included on the command line to the linker.
5662 Process the @code{libgcc} spec. This is a spec string for deciding
5663 which GCC support library should be included on the command line to the linker.
5666 Process the @code{startfile} spec. This is a spec for deciding which
5667 object files should be the first ones passed to the linker. Typically
5668 this might be a file named @file{crt0.o}.
5671 Process the @code{endfile} spec. This is a spec string that specifies
5672 the last object files that will be passed to the linker.
5675 Process the @code{cpp} spec. This is used to construct the arguments
5676 to be passed to the C preprocessor.
5679 Process the @code{signed_char} spec. This is intended to be used
5680 to tell cpp whether a char is signed. It typically has the definition:
5682 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
5686 Process the @code{cc1} spec. This is used to construct the options to be
5687 passed to the actual C compiler (@samp{cc1}).
5690 Process the @code{cc1plus} spec. This is used to construct the options to be
5691 passed to the actual C++ compiler (@samp{cc1plus}).
5694 Substitute the variable part of a matched option. See below.
5695 Note that each comma in the substituted string is replaced by
5699 Remove all occurrences of @code{-S} from the command line. Note---this
5700 command is position dependent. @samp{%} commands in the spec string
5701 before this one will see @code{-S}, @samp{%} commands in the spec string
5702 after this one will not.
5704 @item %:@var{function}(@var{args})
5705 Call the named function @var{function}, passing it @var{args}.
5706 @var{args} is first processed as a nested spec string, then split
5707 into an argument vector in the usual fashion. The function returns
5708 a string which is processed as if it had appeared literally as part
5709 of the current spec.
5711 The following built-in spec functions are provided:
5714 @item @code{if-exists}
5715 The @code{if-exists} spec function takes one argument, an absolute
5716 pathname to a file. If the file exists, @code{if-exists} returns the
5717 pathname. Here is a small example of its usage:
5721 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5724 @item @code{if-exists-else}
5725 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5726 spec function, except that it takes two arguments. The first argument is
5727 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5728 returns the pathname. If it does not exist, it returns the second argument.
5729 This way, @code{if-exists-else} can be used to select one file or another,
5730 based on the existence of the first. Here is a small example of its usage:
5734 crt0%O%s %:if-exists(crti%O%s) \
5735 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5740 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5741 If that switch was not specified, this substitutes nothing. Note that
5742 the leading dash is omitted when specifying this option, and it is
5743 automatically inserted if the substitution is performed. Thus the spec
5744 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5745 and would output the command line option @option{-foo}.
5747 @item %W@{@code{S}@}
5748 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5751 @item %@{@code{S}*@}
5752 Substitutes all the switches specified to GCC whose names start
5753 with @code{-S}, but which also take an argument. This is used for
5754 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5755 GCC considers @option{-o foo} as being
5756 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5757 text, including the space. Thus two arguments would be generated.
5759 @item %@{@code{S}*&@code{T}*@}
5760 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5761 (the order of @code{S} and @code{T} in the spec is not significant).
5762 There can be any number of ampersand-separated variables; for each the
5763 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5765 @item %@{@code{S}:@code{X}@}
5766 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
5768 @item %@{!@code{S}:@code{X}@}
5769 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
5771 @item %@{@code{S}*:@code{X}@}
5772 Substitutes @code{X} if one or more switches whose names start with
5773 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
5774 once, no matter how many such switches appeared. However, if @code{%*}
5775 appears somewhere in @code{X}, then @code{X} will be substituted once
5776 for each matching switch, with the @code{%*} replaced by the part of
5777 that switch that matched the @code{*}.
5779 @item %@{.@code{S}:@code{X}@}
5780 Substitutes @code{X}, if processing a file with suffix @code{S}.
5782 @item %@{!.@code{S}:@code{X}@}
5783 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
5785 @item %@{@code{S}|@code{P}:@code{X}@}
5786 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
5787 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
5788 although they have a stronger binding than the @samp{|}. If @code{%*}
5789 appears in @code{X}, all of the alternatives must be starred, and only
5790 the first matching alternative is substituted.
5792 For example, a spec string like this:
5795 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5798 will output the following command-line options from the following input
5799 command-line options:
5804 -d fred.c -foo -baz -boggle
5805 -d jim.d -bar -baz -boggle
5808 @item %@{S:X; T:Y; :D@}
5810 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
5811 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
5812 be as many clauses as you need. This may be combined with @code{.},
5813 @code{!}, @code{|}, and @code{*} as needed.
5818 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
5819 construct may contain other nested @samp{%} constructs or spaces, or
5820 even newlines. They are processed as usual, as described above.
5821 Trailing white space in @code{X} is ignored. White space may also
5822 appear anywhere on the left side of the colon in these constructs,
5823 except between @code{.} or @code{*} and the corresponding word.
5825 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
5826 handled specifically in these constructs. If another value of
5827 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5828 @option{-W} switch is found later in the command line, the earlier
5829 switch value is ignored, except with @{@code{S}*@} where @code{S} is
5830 just one letter, which passes all matching options.
5832 The character @samp{|} at the beginning of the predicate text is used to
5833 indicate that a command should be piped to the following command, but
5834 only if @option{-pipe} is specified.
5836 It is built into GCC which switches take arguments and which do not.
5837 (You might think it would be useful to generalize this to allow each
5838 compiler's spec to say which switches take arguments. But this cannot
5839 be done in a consistent fashion. GCC cannot even decide which input
5840 files have been specified without knowing which switches take arguments,
5841 and it must know which input files to compile in order to tell which
5844 GCC also knows implicitly that arguments starting in @option{-l} are to be
5845 treated as compiler output files, and passed to the linker in their
5846 proper position among the other output files.
5848 @c man begin OPTIONS
5850 @node Target Options
5851 @section Specifying Target Machine and Compiler Version
5852 @cindex target options
5853 @cindex cross compiling
5854 @cindex specifying machine version
5855 @cindex specifying compiler version and target machine
5856 @cindex compiler version, specifying
5857 @cindex target machine, specifying
5859 The usual way to run GCC is to run the executable called @file{gcc}, or
5860 @file{<machine>-gcc} when cross-compiling, or
5861 @file{<machine>-gcc-<version>} to run a version other than the one that
5862 was installed last. Sometimes this is inconvenient, so GCC provides
5863 options that will switch to another cross-compiler or version.
5866 @item -b @var{machine}
5868 The argument @var{machine} specifies the target machine for compilation.
5870 The value to use for @var{machine} is the same as was specified as the
5871 machine type when configuring GCC as a cross-compiler. For
5872 example, if a cross-compiler was configured with @samp{configure
5873 i386v}, meaning to compile for an 80386 running System V, then you
5874 would specify @option{-b i386v} to run that cross compiler.
5876 @item -V @var{version}
5878 The argument @var{version} specifies which version of GCC to run.
5879 This is useful when multiple versions are installed. For example,
5880 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5883 The @option{-V} and @option{-b} options work by running the
5884 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5885 use them if you can just run that directly.
5887 @node Submodel Options
5888 @section Hardware Models and Configurations
5889 @cindex submodel options
5890 @cindex specifying hardware config
5891 @cindex hardware models and configurations, specifying
5892 @cindex machine dependent options
5894 Earlier we discussed the standard option @option{-b} which chooses among
5895 different installed compilers for completely different target
5896 machines, such as VAX vs.@: 68000 vs.@: 80386.
5898 In addition, each of these target machine types can have its own
5899 special options, starting with @samp{-m}, to choose among various
5900 hardware models or configurations---for example, 68010 vs 68020,
5901 floating coprocessor or none. A single installed version of the
5902 compiler can compile for any model or configuration, according to the
5905 Some configurations of the compiler also support additional special
5906 options, usually for compatibility with other compilers on the same
5909 These options are defined by the macro @code{TARGET_SWITCHES} in the
5910 machine description. The default for the options is also defined by
5911 that macro, which enables you to change the defaults.
5921 * RS/6000 and PowerPC Options::
5924 * i386 and x86-64 Options::
5926 * DEC Alpha Options::
5927 * DEC Alpha/VMS Options::
5930 * System V Options::
5931 * TMS320C3x/C4x Options::
5938 * S/390 and zSeries Options::
5942 * Xstormy16 Options::
5947 @node M680x0 Options
5948 @subsection M680x0 Options
5949 @cindex M680x0 options
5951 These are the @samp{-m} options defined for the 68000 series. The default
5952 values for these options depends on which style of 68000 was selected when
5953 the compiler was configured; the defaults for the most common choices are
5961 Generate output for a 68000. This is the default
5962 when the compiler is configured for 68000-based systems.
5964 Use this option for microcontrollers with a 68000 or EC000 core,
5965 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5971 Generate output for a 68020. This is the default
5972 when the compiler is configured for 68020-based systems.
5976 Generate output containing 68881 instructions for floating point.
5977 This is the default for most 68020 systems unless @option{--nfp} was
5978 specified when the compiler was configured.
5982 Generate output for a 68030. This is the default when the compiler is
5983 configured for 68030-based systems.
5987 Generate output for a 68040. This is the default when the compiler is
5988 configured for 68040-based systems.
5990 This option inhibits the use of 68881/68882 instructions that have to be
5991 emulated by software on the 68040. Use this option if your 68040 does not
5992 have code to emulate those instructions.
5996 Generate output for a 68060. This is the default when the compiler is
5997 configured for 68060-based systems.
5999 This option inhibits the use of 68020 and 68881/68882 instructions that
6000 have to be emulated by software on the 68060. Use this option if your 68060
6001 does not have code to emulate those instructions.
6005 Generate output for a CPU32. This is the default
6006 when the compiler is configured for CPU32-based systems.
6008 Use this option for microcontrollers with a
6009 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
6010 68336, 68340, 68341, 68349 and 68360.
6014 Generate output for a 520X ``coldfire'' family cpu. This is the default
6015 when the compiler is configured for 520X-based systems.
6017 Use this option for microcontroller with a 5200 core, including
6018 the MCF5202, MCF5203, MCF5204 and MCF5202.
6023 Generate output for a 68040, without using any of the new instructions.
6024 This results in code which can run relatively efficiently on either a
6025 68020/68881 or a 68030 or a 68040. The generated code does use the
6026 68881 instructions that are emulated on the 68040.
6030 Generate output for a 68060, without using any of the new instructions.
6031 This results in code which can run relatively efficiently on either a
6032 68020/68881 or a 68030 or a 68040. The generated code does use the
6033 68881 instructions that are emulated on the 68060.
6036 @opindex msoft-float
6037 Generate output containing library calls for floating point.
6038 @strong{Warning:} the requisite libraries are not available for all m68k
6039 targets. Normally the facilities of the machine's usual C compiler are
6040 used, but this can't be done directly in cross-compilation. You must
6041 make your own arrangements to provide suitable library functions for
6042 cross-compilation. The embedded targets @samp{m68k-*-aout} and
6043 @samp{m68k-*-coff} do provide software floating point support.
6047 Consider type @code{int} to be 16 bits wide, like @code{short int}.
6050 @opindex mnobitfield
6051 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
6052 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
6056 Do use the bit-field instructions. The @option{-m68020} option implies
6057 @option{-mbitfield}. This is the default if you use a configuration
6058 designed for a 68020.
6062 Use a different function-calling convention, in which functions
6063 that take a fixed number of arguments return with the @code{rtd}
6064 instruction, which pops their arguments while returning. This
6065 saves one instruction in the caller since there is no need to pop
6066 the arguments there.
6068 This calling convention is incompatible with the one normally
6069 used on Unix, so you cannot use it if you need to call libraries
6070 compiled with the Unix compiler.
6072 Also, you must provide function prototypes for all functions that
6073 take variable numbers of arguments (including @code{printf});
6074 otherwise incorrect code will be generated for calls to those
6077 In addition, seriously incorrect code will result if you call a
6078 function with too many arguments. (Normally, extra arguments are
6079 harmlessly ignored.)
6081 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
6082 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
6085 @itemx -mno-align-int
6087 @opindex mno-align-int
6088 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
6089 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
6090 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
6091 Aligning variables on 32-bit boundaries produces code that runs somewhat
6092 faster on processors with 32-bit busses at the expense of more memory.
6094 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
6095 align structures containing the above types differently than
6096 most published application binary interface specifications for the m68k.
6100 Use the pc-relative addressing mode of the 68000 directly, instead of
6101 using a global offset table. At present, this option implies @option{-fpic},
6102 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
6103 not presently supported with @option{-mpcrel}, though this could be supported for
6104 68020 and higher processors.
6106 @item -mno-strict-align
6107 @itemx -mstrict-align
6108 @opindex mno-strict-align
6109 @opindex mstrict-align
6110 Do not (do) assume that unaligned memory references will be handled by
6114 Generate code that allows the data segment to be located in a different
6115 area of memory from the text segment. This allows for execute in place in
6116 an environment without virtual memory management. This option implies -fPIC.
6119 Generate code that assumes that the data segment follows the text segment.
6120 This is the default.
6122 @item -mid-shared-library
6123 Generate code that supports shared libraries via the library ID method.
6124 This allows for execute in place and shared libraries in an environment
6125 without virtual memory management. This option implies -fPIC.
6127 @item -mno-id-shared-library
6128 Generate code that doesn't assume ID based shared libraries are being used.
6129 This is the default.
6131 @item -mshared-library-id=n
6132 Specified the identification number of the ID based shared library being
6133 compiled. Specifying a value of 0 will generate more compact code, specifying
6134 other values will force the allocation of that number to the current
6135 library but is no more space or time efficient than omitting this option.
6139 @node M68hc1x Options
6140 @subsection M68hc1x Options
6141 @cindex M68hc1x options
6143 These are the @samp{-m} options defined for the 68hc11 and 68hc12
6144 microcontrollers. The default values for these options depends on
6145 which style of microcontroller was selected when the compiler was configured;
6146 the defaults for the most common choices are given below.
6153 Generate output for a 68HC11. This is the default
6154 when the compiler is configured for 68HC11-based systems.
6160 Generate output for a 68HC12. This is the default
6161 when the compiler is configured for 68HC12-based systems.
6167 Generate output for a 68HCS12.
6170 @opindex mauto-incdec
6171 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
6178 Enable the use of 68HC12 min and max instructions.
6181 @itemx -mno-long-calls
6182 @opindex mlong-calls
6183 @opindex mno-long-calls
6184 Treat all calls as being far away (near). If calls are assumed to be
6185 far away, the compiler will use the @code{call} instruction to
6186 call a function and the @code{rtc} instruction for returning.
6190 Consider type @code{int} to be 16 bits wide, like @code{short int}.
6192 @item -msoft-reg-count=@var{count}
6193 @opindex msoft-reg-count
6194 Specify the number of pseudo-soft registers which are used for the
6195 code generation. The maximum number is 32. Using more pseudo-soft
6196 register may or may not result in better code depending on the program.
6197 The default is 4 for 68HC11 and 2 for 68HC12.
6202 @subsection VAX Options
6205 These @samp{-m} options are defined for the VAX:
6210 Do not output certain jump instructions (@code{aobleq} and so on)
6211 that the Unix assembler for the VAX cannot handle across long
6216 Do output those jump instructions, on the assumption that you
6217 will assemble with the GNU assembler.
6221 Output code for g-format floating point numbers instead of d-format.
6225 @subsection SPARC Options
6226 @cindex SPARC options
6228 These @samp{-m} options are supported on the SPARC:
6233 @opindex mno-app-regs
6235 Specify @option{-mapp-regs} to generate output using the global registers
6236 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
6239 To be fully SVR4 ABI compliant at the cost of some performance loss,
6240 specify @option{-mno-app-regs}. You should compile libraries and system
6241 software with this option.
6246 @opindex mhard-float
6247 Generate output containing floating point instructions. This is the
6253 @opindex msoft-float
6254 Generate output containing library calls for floating point.
6255 @strong{Warning:} the requisite libraries are not available for all SPARC
6256 targets. Normally the facilities of the machine's usual C compiler are
6257 used, but this cannot be done directly in cross-compilation. You must make
6258 your own arrangements to provide suitable library functions for
6259 cross-compilation. The embedded targets @samp{sparc-*-aout} and
6260 @samp{sparclite-*-*} do provide software floating point support.
6262 @option{-msoft-float} changes the calling convention in the output file;
6263 therefore, it is only useful if you compile @emph{all} of a program with
6264 this option. In particular, you need to compile @file{libgcc.a}, the
6265 library that comes with GCC, with @option{-msoft-float} in order for
6268 @item -mhard-quad-float
6269 @opindex mhard-quad-float
6270 Generate output containing quad-word (long double) floating point
6273 @item -msoft-quad-float
6274 @opindex msoft-quad-float
6275 Generate output containing library calls for quad-word (long double)
6276 floating point instructions. The functions called are those specified
6277 in the SPARC ABI@. This is the default.
6279 As of this writing, there are no SPARC implementations that have hardware
6280 support for the quad-word floating point instructions. They all invoke
6281 a trap handler for one of these instructions, and then the trap handler
6282 emulates the effect of the instruction. Because of the trap handler overhead,
6283 this is much slower than calling the ABI library routines. Thus the
6284 @option{-msoft-quad-float} option is the default.
6286 @item -mno-unaligned-doubles
6287 @itemx -munaligned-doubles
6288 @opindex mno-unaligned-doubles
6289 @opindex munaligned-doubles
6290 Assume that doubles have 8 byte alignment. This is the default.
6292 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
6293 alignment only if they are contained in another type, or if they have an
6294 absolute address. Otherwise, it assumes they have 4 byte alignment.
6295 Specifying this option avoids some rare compatibility problems with code
6296 generated by other compilers. It is not the default because it results
6297 in a performance loss, especially for floating point code.
6299 @item -mno-faster-structs
6300 @itemx -mfaster-structs
6301 @opindex mno-faster-structs
6302 @opindex mfaster-structs
6303 With @option{-mfaster-structs}, the compiler assumes that structures
6304 should have 8 byte alignment. This enables the use of pairs of
6305 @code{ldd} and @code{std} instructions for copies in structure
6306 assignment, in place of twice as many @code{ld} and @code{st} pairs.
6307 However, the use of this changed alignment directly violates the SPARC
6308 ABI@. Thus, it's intended only for use on targets where the developer
6309 acknowledges that their resulting code will not be directly in line with
6310 the rules of the ABI@.
6313 @opindex mimpure-text
6314 @option{-mimpure-text}, used in addition to @option{-shared}, tells
6315 the compiler to not pass @option{-z text} to the linker when linking a
6316 shared object. Using this option, you can link position-dependent
6317 code into a shared object.
6319 @option{-mimpure-text} suppresses the ``relocations remain against
6320 allocatable but non-writable sections'' linker error message.
6321 However, the necessary relocations will trigger copy-on-write, and the
6322 shared object is not actually shared across processes. Instead of
6323 using @option{-mimpure-text}, you should compile all source code with
6324 @option{-fpic} or @option{-fPIC}.
6326 This option is only available on SunOS and Solaris.
6328 @item -mcpu=@var{cpu_type}
6330 Set the instruction set, register set, and instruction scheduling parameters
6331 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
6332 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
6333 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
6334 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
6337 Default instruction scheduling parameters are used for values that select
6338 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
6339 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
6341 Here is a list of each supported architecture and their supported
6346 v8: supersparc, hypersparc
6347 sparclite: f930, f934, sparclite86x
6349 v9: ultrasparc, ultrasparc3
6352 By default (unless configured otherwise), GCC generates code for the V7
6353 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
6354 additionally optimizes it for the Cypress CY7C602 chip, as used in the
6355 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
6356 SPARCStation 1, 2, IPX etc.
6358 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
6359 architecture. The only difference from V7 code is that the compiler emits
6360 the integer multiply and integer divide instructions which exist in SPARC-V8
6361 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
6362 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
6365 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
6366 the SPARC architecture. This adds the integer multiply, integer divide step
6367 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
6368 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
6369 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU. With
6370 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
6371 MB86934 chip, which is the more recent SPARClite with FPU.
6373 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
6374 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
6375 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
6376 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
6377 optimizes it for the TEMIC SPARClet chip.
6379 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
6380 architecture. This adds 64-bit integer and floating-point move instructions,
6381 3 additional floating-point condition code registers and conditional move
6382 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
6383 optimizes it for the Sun UltraSPARC I/II chips. With
6384 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
6385 Sun UltraSPARC III chip.
6387 @item -mtune=@var{cpu_type}
6389 Set the instruction scheduling parameters for machine type
6390 @var{cpu_type}, but do not set the instruction set or register set that the
6391 option @option{-mcpu=@var{cpu_type}} would.
6393 The same values for @option{-mcpu=@var{cpu_type}} can be used for
6394 @option{-mtune=@var{cpu_type}}, but the only useful values are those
6395 that select a particular cpu implementation. Those are @samp{cypress},
6396 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
6397 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
6404 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI. The
6405 difference from the V8 ABI is that the global and out registers are
6406 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
6407 mode for all SPARC-V9 processors.
6413 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
6414 Visual Instruction Set extensions. The default is @option{-mno-vis}.
6417 These @samp{-m} options are supported in addition to the above
6418 on SPARC-V9 processors in 64-bit environments:
6421 @item -mlittle-endian
6422 @opindex mlittle-endian
6423 Generate code for a processor running in little-endian mode. It is only
6424 available for a few configurations and most notably not on Solaris.
6430 Generate code for a 32-bit or 64-bit environment.
6431 The 32-bit environment sets int, long and pointer to 32 bits.
6432 The 64-bit environment sets int to 32 bits and long and pointer
6435 @item -mcmodel=medlow
6436 @opindex mcmodel=medlow
6437 Generate code for the Medium/Low code model: 64-bit addresses, programs
6438 must be linked in the low 32 bits of memory. Programs can be statically
6439 or dynamically linked.
6441 @item -mcmodel=medmid
6442 @opindex mcmodel=medmid
6443 Generate code for the Medium/Middle code model: 64-bit addresses, programs
6444 must be linked in the low 44 bits of memory, the text and data segments must
6445 be less than 2GB in size and the data segment must be located within 2GB of
6448 @item -mcmodel=medany
6449 @opindex mcmodel=medany
6450 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
6451 may be linked anywhere in memory, the text and data segments must be less
6452 than 2GB in size and the data segment must be located within 2GB of the
6455 @item -mcmodel=embmedany
6456 @opindex mcmodel=embmedany
6457 Generate code for the Medium/Anywhere code model for embedded systems:
6458 64-bit addresses, the text and data segments must be less than 2GB in
6459 size, both starting anywhere in memory (determined at link time). The
6460 global register %g4 points to the base of the data segment. Programs
6461 are statically linked and PIC is not supported.
6464 @itemx -mno-stack-bias
6465 @opindex mstack-bias
6466 @opindex mno-stack-bias
6467 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
6468 frame pointer if present, are offset by @minus{}2047 which must be added back
6469 when making stack frame references. This is the default in 64-bit mode.
6470 Otherwise, assume no such offset is present.
6474 @subsection ARM Options
6477 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6481 @item -mabi=@var{name}
6483 Generate code for the specified ABI. Permissible values are: @samp{apcs-gnu},
6484 @samp{atpcs}, @samp{aapcs} and @samp{iwmmxt}.
6487 @opindex mapcs-frame
6488 Generate a stack frame that is compliant with the ARM Procedure Call
6489 Standard for all functions, even if this is not strictly necessary for
6490 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6491 with this option will cause the stack frames not to be generated for
6492 leaf functions. The default is @option{-mno-apcs-frame}.
6496 This is a synonym for @option{-mapcs-frame}.
6500 Generate code for a processor running with a 26-bit program counter,
6501 and conforming to the function calling standards for the APCS 26-bit
6502 option. This option replaces the @option{-m2} and @option{-m3} options
6503 of previous releases of the compiler.
6507 Generate code for a processor running with a 32-bit program counter,
6508 and conforming to the function calling standards for the APCS 32-bit
6509 option. This option replaces the @option{-m6} option of previous releases
6513 @c not currently implemented
6514 @item -mapcs-stack-check
6515 @opindex mapcs-stack-check
6516 Generate code to check the amount of stack space available upon entry to
6517 every function (that actually uses some stack space). If there is
6518 insufficient space available then either the function
6519 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6520 called, depending upon the amount of stack space required. The run time
6521 system is required to provide these functions. The default is
6522 @option{-mno-apcs-stack-check}, since this produces smaller code.
6524 @c not currently implemented
6526 @opindex mapcs-float
6527 Pass floating point arguments using the float point registers. This is
6528 one of the variants of the APCS@. This option is recommended if the
6529 target hardware has a floating point unit or if a lot of floating point
6530 arithmetic is going to be performed by the code. The default is
6531 @option{-mno-apcs-float}, since integer only code is slightly increased in
6532 size if @option{-mapcs-float} is used.
6534 @c not currently implemented
6535 @item -mapcs-reentrant
6536 @opindex mapcs-reentrant
6537 Generate reentrant, position independent code. The default is
6538 @option{-mno-apcs-reentrant}.
6541 @item -mthumb-interwork
6542 @opindex mthumb-interwork
6543 Generate code which supports calling between the ARM and Thumb
6544 instruction sets. Without this option the two instruction sets cannot
6545 be reliably used inside one program. The default is
6546 @option{-mno-thumb-interwork}, since slightly larger code is generated
6547 when @option{-mthumb-interwork} is specified.
6549 @item -mno-sched-prolog
6550 @opindex mno-sched-prolog
6551 Prevent the reordering of instructions in the function prolog, or the
6552 merging of those instruction with the instructions in the function's
6553 body. This means that all functions will start with a recognizable set
6554 of instructions (or in fact one of a choice from a small set of
6555 different function prologues), and this information can be used to
6556 locate the start if functions inside an executable piece of code. The
6557 default is @option{-msched-prolog}.
6560 @opindex mhard-float
6561 Generate output containing floating point instructions. This is the
6565 @opindex msoft-float
6566 Generate output containing library calls for floating point.
6567 @strong{Warning:} the requisite libraries are not available for all ARM
6568 targets. Normally the facilities of the machine's usual C compiler are
6569 used, but this cannot be done directly in cross-compilation. You must make
6570 your own arrangements to provide suitable library functions for
6573 @option{-msoft-float} changes the calling convention in the output file;
6574 therefore, it is only useful if you compile @emph{all} of a program with
6575 this option. In particular, you need to compile @file{libgcc.a}, the
6576 library that comes with GCC, with @option{-msoft-float} in order for
6579 @item -mfloat-abi=@var{name}
6581 Specifies which ABI to use for floating point values. Permissible values
6582 are: @samp{soft}, @samp{softfp} and @samp{hard}.
6584 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
6585 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
6586 of floating point instructions, but still uses the soft-float calling
6589 @item -mlittle-endian
6590 @opindex mlittle-endian
6591 Generate code for a processor running in little-endian mode. This is
6592 the default for all standard configurations.
6595 @opindex mbig-endian
6596 Generate code for a processor running in big-endian mode; the default is
6597 to compile code for a little-endian processor.
6599 @item -mwords-little-endian
6600 @opindex mwords-little-endian
6601 This option only applies when generating code for big-endian processors.
6602 Generate code for a little-endian word order but a big-endian byte
6603 order. That is, a byte order of the form @samp{32107654}. Note: this
6604 option should only be used if you require compatibility with code for
6605 big-endian ARM processors generated by versions of the compiler prior to
6608 @item -malignment-traps
6609 @opindex malignment-traps
6610 Generate code that will not trap if the MMU has alignment traps enabled.
6611 On ARM architectures prior to ARMv4, there were no instructions to
6612 access half-word objects stored in memory. However, when reading from
6613 memory a feature of the ARM architecture allows a word load to be used,
6614 even if the address is unaligned, and the processor core will rotate the
6615 data as it is being loaded. This option tells the compiler that such
6616 misaligned accesses will cause a MMU trap and that it should instead
6617 synthesize the access as a series of byte accesses. The compiler can
6618 still use word accesses to load half-word data if it knows that the
6619 address is aligned to a word boundary.
6621 This option is ignored when compiling for ARM architecture 4 or later,
6622 since these processors have instructions to directly access half-word
6625 @item -mno-alignment-traps
6626 @opindex mno-alignment-traps
6627 Generate code that assumes that the MMU will not trap unaligned
6628 accesses. This produces better code when the target instruction set
6629 does not have half-word memory operations (i.e.@: implementations prior to
6632 Note that you cannot use this option to access unaligned word objects,
6633 since the processor will only fetch one 32-bit aligned object from
6636 The default setting for most targets is @option{-mno-alignment-traps}, since
6637 this produces better code when there are no half-word memory
6638 instructions available.
6640 @item -mshort-load-bytes
6641 @itemx -mno-short-load-words
6642 @opindex mshort-load-bytes
6643 @opindex mno-short-load-words
6644 These are deprecated aliases for @option{-malignment-traps}.
6646 @item -mno-short-load-bytes
6647 @itemx -mshort-load-words
6648 @opindex mno-short-load-bytes
6649 @opindex mshort-load-words
6650 This are deprecated aliases for @option{-mno-alignment-traps}.
6652 @item -mcpu=@var{name}
6654 This specifies the name of the target ARM processor. GCC uses this name
6655 to determine what kind of instructions it can emit when generating
6656 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6657 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6658 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6659 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6660 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6661 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6662 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6663 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6664 @samp{arm920t}, @samp{arm926ejs}, @samp{arm940t}, @samp{arm9tdmi},
6665 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ejs},
6666 @samp{arm1136js}, @samp{arm1136jfs} ,@samp{xscale}, @samp{iwmmxt},
6669 @itemx -mtune=@var{name}
6671 This option is very similar to the @option{-mcpu=} option, except that
6672 instead of specifying the actual target processor type, and hence
6673 restricting which instructions can be used, it specifies that GCC should
6674 tune the performance of the code as if the target were of the type
6675 specified in this option, but still choosing the instructions that it
6676 will generate based on the cpu specified by a @option{-mcpu=} option.
6677 For some ARM implementations better performance can be obtained by using
6680 @item -march=@var{name}
6682 This specifies the name of the target ARM architecture. GCC uses this
6683 name to determine what kind of instructions it can emit when generating
6684 assembly code. This option can be used in conjunction with or instead
6685 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6686 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6687 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
6688 @samp{iwmmxt}, @samp{ep9312}.
6690 @item -mfpu=@var{name}
6691 @itemx -mfpe=@var{number}
6692 @itemx -mfp=@var{number}
6696 This specifies what floating point hardware (or hardware emulation) is
6697 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
6698 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
6699 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
6700 with older versions of GCC@.
6702 If @option{-msoft-float} is specified this specifies the format of
6703 floating point values.
6705 @item -mstructure-size-boundary=@var{n}
6706 @opindex mstructure-size-boundary
6707 The size of all structures and unions will be rounded up to a multiple
6708 of the number of bits set by this option. Permissible values are 8, 32
6709 and 64. The default value varies for different toolchains. For the COFF
6710 targeted toolchain the default value is 8. A value of 64 is only allowed
6711 if the underlying ABI supports it.
6713 Specifying the larger number can produce faster, more efficient code, but
6714 can also increase the size of the program. Different values are potentially
6715 incompatible. Code compiled with one value cannot necessarily expect to
6716 work with code or libraries compiled with another value, if they exchange
6717 information using structures or unions.
6719 @item -mabort-on-noreturn
6720 @opindex mabort-on-noreturn
6721 Generate a call to the function @code{abort} at the end of a
6722 @code{noreturn} function. It will be executed if the function tries to
6726 @itemx -mno-long-calls
6727 @opindex mlong-calls
6728 @opindex mno-long-calls
6729 Tells the compiler to perform function calls by first loading the
6730 address of the function into a register and then performing a subroutine
6731 call on this register. This switch is needed if the target function
6732 will lie outside of the 64 megabyte addressing range of the offset based
6733 version of subroutine call instruction.
6735 Even if this switch is enabled, not all function calls will be turned
6736 into long calls. The heuristic is that static functions, functions
6737 which have the @samp{short-call} attribute, functions that are inside
6738 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6739 definitions have already been compiled within the current compilation
6740 unit, will not be turned into long calls. The exception to this rule is
6741 that weak function definitions, functions with the @samp{long-call}
6742 attribute or the @samp{section} attribute, and functions that are within
6743 the scope of a @samp{#pragma long_calls} directive, will always be
6744 turned into long calls.
6746 This feature is not enabled by default. Specifying
6747 @option{-mno-long-calls} will restore the default behavior, as will
6748 placing the function calls within the scope of a @samp{#pragma
6749 long_calls_off} directive. Note these switches have no effect on how
6750 the compiler generates code to handle function calls via function
6753 @item -mnop-fun-dllimport
6754 @opindex mnop-fun-dllimport
6755 Disable support for the @code{dllimport} attribute.
6757 @item -msingle-pic-base
6758 @opindex msingle-pic-base
6759 Treat the register used for PIC addressing as read-only, rather than
6760 loading it in the prologue for each function. The run-time system is
6761 responsible for initializing this register with an appropriate value
6762 before execution begins.
6764 @item -mpic-register=@var{reg}
6765 @opindex mpic-register
6766 Specify the register to be used for PIC addressing. The default is R10
6767 unless stack-checking is enabled, when R9 is used.
6769 @item -mcirrus-fix-invalid-insns
6770 @opindex mcirrus-fix-invalid-insns
6771 @opindex mno-cirrus-fix-invalid-insns
6772 Insert NOPs into the instruction stream to in order to work around
6773 problems with invalid Maverick instruction combinations. This option
6774 is only valid if the @option{-mcpu=ep9312} option has been used to
6775 enable generation of instructions for the Cirrus Maverick floating
6776 point co-processor. This option is not enabled by default, since the
6777 problem is only present in older Maverick implementations. The default
6778 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6781 @item -mpoke-function-name
6782 @opindex mpoke-function-name
6783 Write the name of each function into the text section, directly
6784 preceding the function prologue. The generated code is similar to this:
6788 .ascii "arm_poke_function_name", 0
6791 .word 0xff000000 + (t1 - t0)
6792 arm_poke_function_name
6794 stmfd sp!, @{fp, ip, lr, pc@}
6798 When performing a stack backtrace, code can inspect the value of
6799 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6800 location @code{pc - 12} and the top 8 bits are set, then we know that
6801 there is a function name embedded immediately preceding this location
6802 and has length @code{((pc[-3]) & 0xff000000)}.
6806 Generate code for the 16-bit Thumb instruction set. The default is to
6807 use the 32-bit ARM instruction set.
6810 @opindex mtpcs-frame
6811 Generate a stack frame that is compliant with the Thumb Procedure Call
6812 Standard for all non-leaf functions. (A leaf function is one that does
6813 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6815 @item -mtpcs-leaf-frame
6816 @opindex mtpcs-leaf-frame
6817 Generate a stack frame that is compliant with the Thumb Procedure Call
6818 Standard for all leaf functions. (A leaf function is one that does
6819 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6821 @item -mcallee-super-interworking
6822 @opindex mcallee-super-interworking
6823 Gives all externally visible functions in the file being compiled an ARM
6824 instruction set header which switches to Thumb mode before executing the
6825 rest of the function. This allows these functions to be called from
6826 non-interworking code.
6828 @item -mcaller-super-interworking
6829 @opindex mcaller-super-interworking
6830 Allows calls via function pointers (including virtual functions) to
6831 execute correctly regardless of whether the target code has been
6832 compiled for interworking or not. There is a small overhead in the cost
6833 of executing a function pointer if this option is enabled.
6837 @node MN10300 Options
6838 @subsection MN10300 Options
6839 @cindex MN10300 options
6841 These @option{-m} options are defined for Matsushita MN10300 architectures:
6846 Generate code to avoid bugs in the multiply instructions for the MN10300
6847 processors. This is the default.
6850 @opindex mno-mult-bug
6851 Do not generate code to avoid bugs in the multiply instructions for the
6856 Generate code which uses features specific to the AM33 processor.
6860 Do not generate code which uses features specific to the AM33 processor. This
6865 Do not link in the C run-time initialization object file.
6869 Indicate to the linker that it should perform a relaxation optimization pass
6870 to shorten branches, calls and absolute memory addresses. This option only
6871 has an effect when used on the command line for the final link step.
6873 This option makes symbolic debugging impossible.
6877 @node M32R/D Options
6878 @subsection M32R/D Options
6879 @cindex M32R/D options
6881 These @option{-m} options are defined for Renesas M32R/D architectures:
6886 Generate code for the M32R/2@.
6890 Generate code for the M32R/X@.
6894 Generate code for the M32R@. This is the default.
6897 @opindex mmodel=small
6898 Assume all objects live in the lower 16MB of memory (so that their addresses
6899 can be loaded with the @code{ld24} instruction), and assume all subroutines
6900 are reachable with the @code{bl} instruction.
6901 This is the default.
6903 The addressability of a particular object can be set with the
6904 @code{model} attribute.
6906 @item -mmodel=medium
6907 @opindex mmodel=medium
6908 Assume objects may be anywhere in the 32-bit address space (the compiler
6909 will generate @code{seth/add3} instructions to load their addresses), and
6910 assume all subroutines are reachable with the @code{bl} instruction.
6913 @opindex mmodel=large
6914 Assume objects may be anywhere in the 32-bit address space (the compiler
6915 will generate @code{seth/add3} instructions to load their addresses), and
6916 assume subroutines may not be reachable with the @code{bl} instruction
6917 (the compiler will generate the much slower @code{seth/add3/jl}
6918 instruction sequence).
6921 @opindex msdata=none
6922 Disable use of the small data area. Variables will be put into
6923 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6924 @code{section} attribute has been specified).
6925 This is the default.
6927 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6928 Objects may be explicitly put in the small data area with the
6929 @code{section} attribute using one of these sections.
6932 @opindex msdata=sdata
6933 Put small global and static data in the small data area, but do not
6934 generate special code to reference them.
6938 Put small global and static data in the small data area, and generate
6939 special instructions to reference them.
6943 @cindex smaller data references
6944 Put global and static objects less than or equal to @var{num} bytes
6945 into the small data or bss sections instead of the normal data or bss
6946 sections. The default value of @var{num} is 8.
6947 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6948 for this option to have any effect.
6950 All modules should be compiled with the same @option{-G @var{num}} value.
6951 Compiling with different values of @var{num} may or may not work; if it
6952 doesn't the linker will give an error message---incorrect code will not be
6957 Makes the M32R specific code in the compiler display some statistics
6958 that might help in debugging programs.
6961 @opindex malign-loops
6962 Align all loops to a 32-byte boundary.
6964 @item -mno-align-loops
6965 @opindex mno-align-loops
6966 Do not enforce a 32-byte alignment for loops. This is the default.
6968 @item -missue-rate=@var{number}
6969 @opindex missue-rate=@var{number}
6970 Issue @var{number} instructions per cycle. @var{number} can only be 1
6973 @item -mbranch-cost=@var{number}
6974 @opindex mbranch-cost=@var{number}
6975 @var{number} can only be 1 or 2. If it is 1 then branches will be
6976 preferred over conditional code, if it is 2, then the opposite will
6979 @item -mflush-trap=@var{number}
6980 @opindex mflush-trap=@var{number}
6981 Specifies the trap number to use to flush the cache. The default is
6982 12. Valid numbers are between 0 and 15 inclusive.
6984 @item -mno-flush-trap
6985 @opindex mno-flush-trap
6986 Specifies that the cache cannot be flushed by using a trap.
6988 @item -mflush-func=@var{name}
6989 @opindex mflush-func=@var{name}
6990 Specifies the name of the operating system function to call to flush
6991 the cache. The default is @emph{_flush_cache}, but a function call
6992 will only be used if a trap is not available.
6994 @item -mno-flush-func
6995 @opindex mno-flush-func
6996 Indicates that there is no OS function for flushing the cache.
7000 @node RS/6000 and PowerPC Options
7001 @subsection IBM RS/6000 and PowerPC Options
7002 @cindex RS/6000 and PowerPC Options
7003 @cindex IBM RS/6000 and PowerPC Options
7005 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
7013 @itemx -mpowerpc-gpopt
7014 @itemx -mno-powerpc-gpopt
7015 @itemx -mpowerpc-gfxopt
7016 @itemx -mno-powerpc-gfxopt
7018 @itemx -mno-powerpc64
7024 @opindex mno-powerpc
7025 @opindex mpowerpc-gpopt
7026 @opindex mno-powerpc-gpopt
7027 @opindex mpowerpc-gfxopt
7028 @opindex mno-powerpc-gfxopt
7030 @opindex mno-powerpc64
7031 GCC supports two related instruction set architectures for the
7032 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
7033 instructions supported by the @samp{rios} chip set used in the original
7034 RS/6000 systems and the @dfn{PowerPC} instruction set is the
7035 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
7036 the IBM 4xx microprocessors.
7038 Neither architecture is a subset of the other. However there is a
7039 large common subset of instructions supported by both. An MQ
7040 register is included in processors supporting the POWER architecture.
7042 You use these options to specify which instructions are available on the
7043 processor you are using. The default value of these options is
7044 determined when configuring GCC@. Specifying the
7045 @option{-mcpu=@var{cpu_type}} overrides the specification of these
7046 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
7047 rather than the options listed above.
7049 The @option{-mpower} option allows GCC to generate instructions that
7050 are found only in the POWER architecture and to use the MQ register.
7051 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
7052 to generate instructions that are present in the POWER2 architecture but
7053 not the original POWER architecture.
7055 The @option{-mpowerpc} option allows GCC to generate instructions that
7056 are found only in the 32-bit subset of the PowerPC architecture.
7057 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
7058 GCC to use the optional PowerPC architecture instructions in the
7059 General Purpose group, including floating-point square root. Specifying
7060 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
7061 use the optional PowerPC architecture instructions in the Graphics
7062 group, including floating-point select.
7064 The @option{-mpowerpc64} option allows GCC to generate the additional
7065 64-bit instructions that are found in the full PowerPC64 architecture
7066 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
7067 @option{-mno-powerpc64}.
7069 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
7070 will use only the instructions in the common subset of both
7071 architectures plus some special AIX common-mode calls, and will not use
7072 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
7073 permits GCC to use any instruction from either architecture and to
7074 allow use of the MQ register; specify this for the Motorola MPC601.
7076 @item -mnew-mnemonics
7077 @itemx -mold-mnemonics
7078 @opindex mnew-mnemonics
7079 @opindex mold-mnemonics
7080 Select which mnemonics to use in the generated assembler code. With
7081 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
7082 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
7083 assembler mnemonics defined for the POWER architecture. Instructions
7084 defined in only one architecture have only one mnemonic; GCC uses that
7085 mnemonic irrespective of which of these options is specified.
7087 GCC defaults to the mnemonics appropriate for the architecture in
7088 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
7089 value of these option. Unless you are building a cross-compiler, you
7090 should normally not specify either @option{-mnew-mnemonics} or
7091 @option{-mold-mnemonics}, but should instead accept the default.
7093 @item -mcpu=@var{cpu_type}
7095 Set architecture type, register usage, choice of mnemonics, and
7096 instruction scheduling parameters for machine type @var{cpu_type}.
7097 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
7098 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
7099 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
7100 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
7101 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
7102 @samp{860}, @samp{970}, @samp{common}, @samp{ec603e}, @samp{G3},
7103 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
7104 @samp{power4}, @samp{power5}, @samp{powerpc}, @samp{powerpc64},
7105 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64a}.
7107 @option{-mcpu=common} selects a completely generic processor. Code
7108 generated under this option will run on any POWER or PowerPC processor.
7109 GCC will use only the instructions in the common subset of both
7110 architectures, and will not use the MQ register. GCC assumes a generic
7111 processor model for scheduling purposes.
7113 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
7114 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
7115 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
7116 types, with an appropriate, generic processor model assumed for
7117 scheduling purposes.
7119 The other options specify a specific processor. Code generated under
7120 those options will run best on that processor, and may not run at all on
7123 The @option{-mcpu} options automatically enable or disable the
7124 following options: @option{-maltivec}, @option{-mhard-float},
7125 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
7126 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
7127 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
7128 @option{-mstring}. The particular options set for any particular CPU
7129 will vary between compiler versions, depending on what setting seems
7130 to produce optimal code for that CPU; it doesn't necessarily reflect
7131 the actual hardware's capabilities. If you wish to set an individual
7132 option to a particular value, you may specify it after the
7133 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
7135 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
7136 not enabled or disabled by the @option{-mcpu} option at present, since
7137 AIX does not have full support for these options. You may still
7138 enable or disable them individually if you're sure it'll work in your
7141 @item -mtune=@var{cpu_type}
7143 Set the instruction scheduling parameters for machine type
7144 @var{cpu_type}, but do not set the architecture type, register usage, or
7145 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
7146 values for @var{cpu_type} are used for @option{-mtune} as for
7147 @option{-mcpu}. If both are specified, the code generated will use the
7148 architecture, registers, and mnemonics set by @option{-mcpu}, but the
7149 scheduling parameters set by @option{-mtune}.
7154 @opindex mno-altivec
7155 These switches enable or disable the use of built-in functions that
7156 allow access to the AltiVec instruction set. You may also need to set
7157 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
7162 Extend the current ABI with SPE ABI extensions. This does not change
7163 the default ABI, instead it adds the SPE ABI extensions to the current
7167 @opindex mabi=no-spe
7168 Disable Booke SPE ABI extensions for the current ABI.
7170 @item -misel=@var{yes/no}
7173 This switch enables or disables the generation of ISEL instructions.
7175 @item -mspe=@var{yes/no}
7178 This switch enables or disables the generation of SPE simd
7181 @item -mfloat-gprs=@var{yes/no}
7183 @opindex mfloat-gprs
7184 This switch enables or disables the generation of floating point
7185 operations on the general purpose registers for architectures that
7186 support it. This option is currently only available on the MPC8540.
7189 @itemx -mno-fp-in-toc
7190 @itemx -mno-sum-in-toc
7191 @itemx -mminimal-toc
7193 @opindex mno-fp-in-toc
7194 @opindex mno-sum-in-toc
7195 @opindex mminimal-toc
7196 Modify generation of the TOC (Table Of Contents), which is created for
7197 every executable file. The @option{-mfull-toc} option is selected by
7198 default. In that case, GCC will allocate at least one TOC entry for
7199 each unique non-automatic variable reference in your program. GCC
7200 will also place floating-point constants in the TOC@. However, only
7201 16,384 entries are available in the TOC@.
7203 If you receive a linker error message that saying you have overflowed
7204 the available TOC space, you can reduce the amount of TOC space used
7205 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
7206 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
7207 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
7208 generate code to calculate the sum of an address and a constant at
7209 run-time instead of putting that sum into the TOC@. You may specify one
7210 or both of these options. Each causes GCC to produce very slightly
7211 slower and larger code at the expense of conserving TOC space.
7213 If you still run out of space in the TOC even when you specify both of
7214 these options, specify @option{-mminimal-toc} instead. This option causes
7215 GCC to make only one TOC entry for every file. When you specify this
7216 option, GCC will produce code that is slower and larger but which
7217 uses extremely little TOC space. You may wish to use this option
7218 only on files that contain less frequently executed code.
7224 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
7225 @code{long} type, and the infrastructure needed to support them.
7226 Specifying @option{-maix64} implies @option{-mpowerpc64} and
7227 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
7228 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
7233 @opindex mno-xl-call
7234 On AIX, pass floating-point arguments to prototyped functions beyond the
7235 register save area (RSA) on the stack in addition to argument FPRs. The
7236 AIX calling convention was extended but not initially documented to
7237 handle an obscure K&R C case of calling a function that takes the
7238 address of its arguments with fewer arguments than declared. AIX XL
7239 compilers access floating point arguments which do not fit in the
7240 RSA from the stack when a subroutine is compiled without
7241 optimization. Because always storing floating-point arguments on the
7242 stack is inefficient and rarely needed, this option is not enabled by
7243 default and only is necessary when calling subroutines compiled by AIX
7244 XL compilers without optimization.
7248 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
7249 application written to use message passing with special startup code to
7250 enable the application to run. The system must have PE installed in the
7251 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
7252 must be overridden with the @option{-specs=} option to specify the
7253 appropriate directory location. The Parallel Environment does not
7254 support threads, so the @option{-mpe} option and the @option{-pthread}
7255 option are incompatible.
7257 @item -malign-natural
7258 @itemx -malign-power
7259 @opindex malign-natural
7260 @opindex malign-power
7261 On AIX, Darwin, and 64-bit PowerPC GNU/Linux, the option
7262 @option{-malign-natural} overrides the ABI-defined alignment of larger
7263 types, such as floating-point doubles, on their natural size-based boundary.
7264 The option @option{-malign-power} instructs GCC to follow the ABI-specified
7265 alignment rules. GCC defaults to the standard alignment defined in the ABI.
7269 @opindex msoft-float
7270 @opindex mhard-float
7271 Generate code that does not use (uses) the floating-point register set.
7272 Software floating point emulation is provided if you use the
7273 @option{-msoft-float} option, and pass the option to GCC when linking.
7276 @itemx -mno-multiple
7278 @opindex mno-multiple
7279 Generate code that uses (does not use) the load multiple word
7280 instructions and the store multiple word instructions. These
7281 instructions are generated by default on POWER systems, and not
7282 generated on PowerPC systems. Do not use @option{-mmultiple} on little
7283 endian PowerPC systems, since those instructions do not work when the
7284 processor is in little endian mode. The exceptions are PPC740 and
7285 PPC750 which permit the instructions usage in little endian mode.
7291 Generate code that uses (does not use) the load string instructions
7292 and the store string word instructions to save multiple registers and
7293 do small block moves. These instructions are generated by default on
7294 POWER systems, and not generated on PowerPC systems. Do not use
7295 @option{-mstring} on little endian PowerPC systems, since those
7296 instructions do not work when the processor is in little endian mode.
7297 The exceptions are PPC740 and PPC750 which permit the instructions
7298 usage in little endian mode.
7304 Generate code that uses (does not use) the load or store instructions
7305 that update the base register to the address of the calculated memory
7306 location. These instructions are generated by default. If you use
7307 @option{-mno-update}, there is a small window between the time that the
7308 stack pointer is updated and the address of the previous frame is
7309 stored, which means code that walks the stack frame across interrupts or
7310 signals may get corrupted data.
7313 @itemx -mno-fused-madd
7314 @opindex mfused-madd
7315 @opindex mno-fused-madd
7316 Generate code that uses (does not use) the floating point multiply and
7317 accumulate instructions. These instructions are generated by default if
7318 hardware floating is used.
7320 @item -mno-bit-align
7322 @opindex mno-bit-align
7324 On System V.4 and embedded PowerPC systems do not (do) force structures
7325 and unions that contain bit-fields to be aligned to the base type of the
7328 For example, by default a structure containing nothing but 8
7329 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7330 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7331 the structure would be aligned to a 1 byte boundary and be one byte in
7334 @item -mno-strict-align
7335 @itemx -mstrict-align
7336 @opindex mno-strict-align
7337 @opindex mstrict-align
7338 On System V.4 and embedded PowerPC systems do not (do) assume that
7339 unaligned memory references will be handled by the system.
7342 @itemx -mno-relocatable
7343 @opindex mrelocatable
7344 @opindex mno-relocatable
7345 On embedded PowerPC systems generate code that allows (does not allow)
7346 the program to be relocated to a different address at runtime. If you
7347 use @option{-mrelocatable} on any module, all objects linked together must
7348 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7350 @item -mrelocatable-lib
7351 @itemx -mno-relocatable-lib
7352 @opindex mrelocatable-lib
7353 @opindex mno-relocatable-lib
7354 On embedded PowerPC systems generate code that allows (does not allow)
7355 the program to be relocated to a different address at runtime. Modules
7356 compiled with @option{-mrelocatable-lib} can be linked with either modules
7357 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7358 with modules compiled with the @option{-mrelocatable} options.
7364 On System V.4 and embedded PowerPC systems do not (do) assume that
7365 register 2 contains a pointer to a global area pointing to the addresses
7366 used in the program.
7369 @itemx -mlittle-endian
7371 @opindex mlittle-endian
7372 On System V.4 and embedded PowerPC systems compile code for the
7373 processor in little endian mode. The @option{-mlittle-endian} option is
7374 the same as @option{-mlittle}.
7379 @opindex mbig-endian
7380 On System V.4 and embedded PowerPC systems compile code for the
7381 processor in big endian mode. The @option{-mbig-endian} option is
7382 the same as @option{-mbig}.
7384 @item -mdynamic-no-pic
7385 @opindex mdynamic-no-pic
7386 On Darwin and Mac OS X systems, compile code so that it is not
7387 relocatable, but that its external references are relocatable. The
7388 resulting code is suitable for applications, but not shared
7391 @item -mprioritize-restricted-insns=@var{priority}
7392 @opindex mprioritize-restricted-insns
7393 This option controls the priority that is assigned to
7394 dispatch-slot restricted instructions during the second scheduling
7395 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
7396 @var{no/highest/second-highest} priority to dispatch slot restricted
7399 @item -msched-costly-dep=@var{dependence_type}
7400 @opindex msched-costly-dep
7401 This option controls which dependences are considered costly
7402 by the target during instruction scheduling. The argument
7403 @var{dependence_type} takes one of the following values:
7404 @var{no}: no dependence is costly,
7405 @var{all}: all dependences are costly,
7406 @var{true_store_to_load}: a true dependence from store to load is costly,
7407 @var{store_to_load}: any dependence from store to load is costly,
7408 @var{number}: any dependence which latency >= @var{number} is costly.
7410 @item -minsert-sched-nops=@var{scheme}
7411 @opindex minsert-sched-nops
7412 This option controls which nop insertion scheme will be used during
7413 the second scheduling pass. The argument @var{scheme} takes one of the
7415 @var{no}: Don't insert nops.
7416 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
7417 according to the scheduler's grouping.
7418 @var{regroup_exact}: Insert nops to force costly dependent insns into
7419 separate groups. Insert exactly as many nops as needed to force an insn
7420 to a new group, according to the estimated processor grouping.
7421 @var{number}: Insert nops to force costly dependent insns into
7422 separate groups. Insert @var{number} nops to force an insn to a new group.
7426 On System V.4 and embedded PowerPC systems compile code using calling
7427 conventions that adheres to the March 1995 draft of the System V
7428 Application Binary Interface, PowerPC processor supplement. This is the
7429 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7431 @item -mcall-sysv-eabi
7432 @opindex mcall-sysv-eabi
7433 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7435 @item -mcall-sysv-noeabi
7436 @opindex mcall-sysv-noeabi
7437 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7439 @item -mcall-solaris
7440 @opindex mcall-solaris
7441 On System V.4 and embedded PowerPC systems compile code for the Solaris
7445 @opindex mcall-linux
7446 On System V.4 and embedded PowerPC systems compile code for the
7447 Linux-based GNU system.
7451 On System V.4 and embedded PowerPC systems compile code for the
7452 Hurd-based GNU system.
7455 @opindex mcall-netbsd
7456 On System V.4 and embedded PowerPC systems compile code for the
7457 NetBSD operating system.
7459 @item -maix-struct-return
7460 @opindex maix-struct-return
7461 Return all structures in memory (as specified by the AIX ABI)@.
7463 @item -msvr4-struct-return
7464 @opindex msvr4-struct-return
7465 Return structures smaller than 8 bytes in registers (as specified by the
7469 @opindex mabi=altivec
7470 Extend the current ABI with AltiVec ABI extensions. This does not
7471 change the default ABI, instead it adds the AltiVec ABI extensions to
7474 @item -mabi=no-altivec
7475 @opindex mabi=no-altivec
7476 Disable AltiVec ABI extensions for the current ABI.
7479 @itemx -mno-prototype
7481 @opindex mno-prototype
7482 On System V.4 and embedded PowerPC systems assume that all calls to
7483 variable argument functions are properly prototyped. Otherwise, the
7484 compiler must insert an instruction before every non prototyped call to
7485 set or clear bit 6 of the condition code register (@var{CR}) to
7486 indicate whether floating point values were passed in the floating point
7487 registers in case the function takes a variable arguments. With
7488 @option{-mprototype}, only calls to prototyped variable argument functions
7489 will set or clear the bit.
7493 On embedded PowerPC systems, assume that the startup module is called
7494 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7495 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7500 On embedded PowerPC systems, assume that the startup module is called
7501 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7506 On embedded PowerPC systems, assume that the startup module is called
7507 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7511 @opindex myellowknife
7512 On embedded PowerPC systems, assume that the startup module is called
7513 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7518 On System V.4 and embedded PowerPC systems, specify that you are
7519 compiling for a VxWorks system.
7523 Specify that you are compiling for the WindISS simulation environment.
7527 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7528 header to indicate that @samp{eabi} extended relocations are used.
7534 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7535 Embedded Applications Binary Interface (eabi) which is a set of
7536 modifications to the System V.4 specifications. Selecting @option{-meabi}
7537 means that the stack is aligned to an 8 byte boundary, a function
7538 @code{__eabi} is called to from @code{main} to set up the eabi
7539 environment, and the @option{-msdata} option can use both @code{r2} and
7540 @code{r13} to point to two separate small data areas. Selecting
7541 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7542 do not call an initialization function from @code{main}, and the
7543 @option{-msdata} option will only use @code{r13} to point to a single
7544 small data area. The @option{-meabi} option is on by default if you
7545 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7548 @opindex msdata=eabi
7549 On System V.4 and embedded PowerPC systems, put small initialized
7550 @code{const} global and static data in the @samp{.sdata2} section, which
7551 is pointed to by register @code{r2}. Put small initialized
7552 non-@code{const} global and static data in the @samp{.sdata} section,
7553 which is pointed to by register @code{r13}. Put small uninitialized
7554 global and static data in the @samp{.sbss} section, which is adjacent to
7555 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7556 incompatible with the @option{-mrelocatable} option. The
7557 @option{-msdata=eabi} option also sets the @option{-memb} option.
7560 @opindex msdata=sysv
7561 On System V.4 and embedded PowerPC systems, put small global and static
7562 data in the @samp{.sdata} section, which is pointed to by register
7563 @code{r13}. Put small uninitialized global and static data in the
7564 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7565 The @option{-msdata=sysv} option is incompatible with the
7566 @option{-mrelocatable} option.
7568 @item -msdata=default
7570 @opindex msdata=default
7572 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7573 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7574 same as @option{-msdata=sysv}.
7577 @opindex msdata-data
7578 On System V.4 and embedded PowerPC systems, put small global and static
7579 data in the @samp{.sdata} section. Put small uninitialized global and
7580 static data in the @samp{.sbss} section. Do not use register @code{r13}
7581 to address small data however. This is the default behavior unless
7582 other @option{-msdata} options are used.
7586 @opindex msdata=none
7588 On embedded PowerPC systems, put all initialized global and static data
7589 in the @samp{.data} section, and all uninitialized data in the
7590 @samp{.bss} section.
7594 @cindex smaller data references (PowerPC)
7595 @cindex .sdata/.sdata2 references (PowerPC)
7596 On embedded PowerPC systems, put global and static items less than or
7597 equal to @var{num} bytes into the small data or bss sections instead of
7598 the normal data or bss section. By default, @var{num} is 8. The
7599 @option{-G @var{num}} switch is also passed to the linker.
7600 All modules should be compiled with the same @option{-G @var{num}} value.
7603 @itemx -mno-regnames
7605 @opindex mno-regnames
7606 On System V.4 and embedded PowerPC systems do (do not) emit register
7607 names in the assembly language output using symbolic forms.
7610 @itemx -mno-longcall
7612 @opindex mno-longcall
7613 Default to making all function calls indirectly, using a register, so
7614 that functions which reside further than 32 megabytes (33,554,432
7615 bytes) from the current location can be called. This setting can be
7616 overridden by the @code{shortcall} function attribute, or by
7617 @code{#pragma longcall(0)}.
7619 Some linkers are capable of detecting out-of-range calls and generating
7620 glue code on the fly. On these systems, long calls are unnecessary and
7621 generate slower code. As of this writing, the AIX linker can do this,
7622 as can the GNU linker for PowerPC/64. It is planned to add this feature
7623 to the GNU linker for 32-bit PowerPC systems as well.
7625 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
7626 callee, L42'', plus a ``branch island'' (glue code). The two target
7627 addresses represent the callee and the ``branch island.'' The
7628 Darwin/PPC linker will prefer the first address and generate a ``bl
7629 callee'' if the PPC ``bl'' instruction will reach the callee directly;
7630 otherwise, the linker will generate ``bl L42'' to call the ``branch
7631 island.'' The ``branch island'' is appended to the body of the
7632 calling function; it computes the full 32-bit address of the callee
7635 On Mach-O (Darwin) systems, this option directs the compiler emit to
7636 the glue for every direct call, and the Darwin linker decides whether
7637 to use or discard it.
7639 In the future, we may cause GCC to ignore all longcall specifications
7640 when the linker is known to generate glue.
7644 Adds support for multithreading with the @dfn{pthreads} library.
7645 This option sets flags for both the preprocessor and linker.
7649 @node Darwin Options
7650 @subsection Darwin Options
7651 @cindex Darwin options
7653 These options are defined for all architectures running the Darwin operating
7654 system. They are useful for compatibility with other Mac OS compilers.
7659 Loads all members of static archive libraries.
7660 See man ld(1) for more information.
7662 @item -arch_errors_fatal
7663 @opindex arch_errors_fatal
7664 Cause the errors having to do with files that have the wrong architecture
7668 @opindex bind_at_load
7669 Causes the output file to be marked such that the dynamic linker will
7670 bind all undefined references when the file is loaded or launched.
7674 Produce a Mach-o bundle format file.
7675 See man ld(1) for more information.
7677 @item -bundle_loader @var{executable}
7678 @opindex bundle_loader
7679 This specifies the @var{executable} that will be loading the build
7680 output file being linked. See man ld(1) for more information.
7682 @item -allowable_client @var{client_name}
7686 @itemx -compatibility_version
7687 @itemx -current_version
7688 @itemx -dependency-file
7690 @itemx -dylinker_install_name
7693 @itemx -exported_symbols_list
7695 @itemx -flat_namespace
7696 @itemx -force_cpusubtype_ALL
7697 @itemx -force_flat_namespace
7698 @itemx -headerpad_max_install_names
7701 @itemx -install_name
7702 @itemx -keep_private_externs
7703 @itemx -multi_module
7704 @itemx -multiply_defined
7705 @itemx -multiply_defined_unused
7707 @itemx -nofixprebinding
7710 @itemx -noseglinkedit
7711 @itemx -pagezero_size
7713 @itemx -prebind_all_twolevel_modules
7714 @itemx -private_bundle
7715 @itemx -read_only_relocs
7717 @itemx -sectobjectsymbols
7721 @itemx -sectobjectsymbols
7723 @itemx -seg_addr_table
7724 @itemx -seg_addr_table_filename
7727 @itemx -segs_read_only_addr
7728 @itemx -segs_read_write_addr
7729 @itemx -single_module
7732 @itemx -sub_umbrella
7733 @itemx -twolevel_namespace
7736 @itemx -unexported_symbols_list
7737 @itemx -weak_reference_mismatches
7740 @opindex allowable_client
7742 @opindex client_name
7743 @opindex compatibility_version
7744 @opindex current_version
7745 @opindex dependency-file
7747 @opindex dylinker_install_name
7750 @opindex exported_symbols_list
7752 @opindex flat_namespace
7753 @opindex force_cpusubtype_ALL
7754 @opindex force_flat_namespace
7755 @opindex headerpad_max_install_names
7758 @opindex install_name
7759 @opindex keep_private_externs
7760 @opindex multi_module
7761 @opindex multiply_defined
7762 @opindex multiply_defined_unused
7764 @opindex nofixprebinding
7765 @opindex nomultidefs
7767 @opindex noseglinkedit
7768 @opindex pagezero_size
7770 @opindex prebind_all_twolevel_modules
7771 @opindex private_bundle
7772 @opindex read_only_relocs
7774 @opindex sectobjectsymbols
7778 @opindex sectobjectsymbols
7780 @opindex seg_addr_table
7781 @opindex seg_addr_table_filename
7782 @opindex seglinkedit
7784 @opindex segs_read_only_addr
7785 @opindex segs_read_write_addr
7786 @opindex single_module
7788 @opindex sub_library
7789 @opindex sub_umbrella
7790 @opindex twolevel_namespace
7793 @opindex unexported_symbols_list
7794 @opindex weak_reference_mismatches
7795 @opindex whatsloaded
7797 These options are available for Darwin linker. Darwin linker man page
7798 describes them in detail.
7803 @subsection MIPS Options
7804 @cindex MIPS options
7810 Generate big-endian code.
7814 Generate little-endian code. This is the default for @samp{mips*el-*-*}
7817 @item -march=@var{arch}
7819 Generate code that will run on @var{arch}, which can be the name of a
7820 generic MIPS ISA, or the name of a particular processor.
7822 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
7823 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
7824 The processor names are:
7825 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
7827 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
7828 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
7832 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
7833 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
7834 The special value @samp{from-abi} selects the
7835 most compatible architecture for the selected ABI (that is,
7836 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
7838 In processor names, a final @samp{000} can be abbreviated as @samp{k}
7839 (for example, @samp{-march=r2k}). Prefixes are optional, and
7840 @samp{vr} may be written @samp{r}.
7842 GCC defines two macros based on the value of this option. The first
7843 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
7844 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
7845 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
7846 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
7847 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
7849 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
7850 above. In other words, it will have the full prefix and will not
7851 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
7852 the macro names the resolved architecture (either @samp{"mips1"} or
7853 @samp{"mips3"}). It names the default architecture when no
7854 @option{-march} option is given.
7856 @item -mtune=@var{arch}
7858 Optimize for @var{arch}. Among other things, this option controls
7859 the way instructions are scheduled, and the perceived cost of arithmetic
7860 operations. The list of @var{arch} values is the same as for
7863 When this option is not used, GCC will optimize for the processor
7864 specified by @option{-march}. By using @option{-march} and
7865 @option{-mtune} together, it is possible to generate code that will
7866 run on a family of processors, but optimize the code for one
7867 particular member of that family.
7869 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
7870 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
7871 @samp{-march} ones described above.
7875 Equivalent to @samp{-march=mips1}.
7879 Equivalent to @samp{-march=mips2}.
7883 Equivalent to @samp{-march=mips3}.
7887 Equivalent to @samp{-march=mips4}.
7891 Equivalent to @samp{-march=mips32}.
7895 Equivalent to @samp{-march=mips32r2}.
7899 Equivalent to @samp{-march=mips64}.
7905 Use (do not use) the MIPS16 ISA.
7917 Generate code for the given ABI@.
7919 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
7920 generates 64-bit code when you select a 64-bit architecture, but you
7921 can use @option{-mgp32} to get 32-bit code instead.
7923 For information about the O64 ABI, see
7924 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
7927 @itemx -mno-abicalls
7929 @opindex mno-abicalls
7930 Generate (do not generate) SVR4-style position-independent code.
7931 @option{-mabicalls} is the default for SVR4-based systems.
7937 Lift (do not lift) the usual restrictions on the size of the global
7940 GCC normally uses a single instruction to load values from the GOT.
7941 While this is relatively efficient, it will only work if the GOT
7942 is smaller than about 64k. Anything larger will cause the linker
7943 to report an error such as:
7945 @cindex relocation truncated to fit (MIPS)
7947 relocation truncated to fit: R_MIPS_GOT16 foobar
7950 If this happens, you should recompile your code with @option{-mxgot}.
7951 It should then work with very large GOTs, although it will also be
7952 less efficient, since it will take three instructions to fetch the
7953 value of a global symbol.
7955 Note that some linkers can create multiple GOTs. If you have such a
7956 linker, you should only need to use @option{-mxgot} when a single object
7957 file accesses more than 64k's worth of GOT entries. Very few do.
7959 These options have no effect unless GCC is generating position
7964 Assume that general-purpose registers are 32 bits wide.
7968 Assume that general-purpose registers are 64 bits wide.
7972 Assume that floating-point registers are 32 bits wide.
7976 Assume that floating-point registers are 64 bits wide.
7979 @opindex mhard-float
7980 Use floating-point coprocessor instructions.
7983 @opindex msoft-float
7984 Do not use floating-point coprocessor instructions. Implement
7985 floating-point calculations using library calls instead.
7987 @item -msingle-float
7988 @opindex msingle-float
7989 Assume that the floating-point coprocessor only supports single-precision
7992 @itemx -mdouble-float
7993 @opindex mdouble-float
7994 Assume that the floating-point coprocessor supports double-precision
7995 operations. This is the default.
7999 Force @code{int} and @code{long} types to be 64 bits wide. See
8000 @option{-mlong32} for an explanation of the default and the way
8001 that the pointer size is determined.
8005 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
8006 an explanation of the default and the way that the pointer size is
8011 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
8013 The default size of @code{int}s, @code{long}s and pointers depends on
8014 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
8015 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
8016 32-bit @code{long}s. Pointers are the same size as @code{long}s,
8017 or the same size as integer registers, whichever is smaller.
8021 @cindex smaller data references (MIPS)
8022 @cindex gp-relative references (MIPS)
8023 Put global and static items less than or equal to @var{num} bytes into
8024 the small data or bss section instead of the normal data or bss section.
8025 This allows the data to be accessed using a single instruction.
8027 All modules should be compiled with the same @option{-G @var{num}}
8030 @item -membedded-data
8031 @itemx -mno-embedded-data
8032 @opindex membedded-data
8033 @opindex mno-embedded-data
8034 Allocate variables to the read-only data section first if possible, then
8035 next in the small data section if possible, otherwise in data. This gives
8036 slightly slower code than the default, but reduces the amount of RAM required
8037 when executing, and thus may be preferred for some embedded systems.
8039 @item -muninit-const-in-rodata
8040 @itemx -mno-uninit-const-in-rodata
8041 @opindex muninit-const-in-rodata
8042 @opindex mno-uninit-const-in-rodata
8043 Put uninitialized @code{const} variables in the read-only data section.
8044 This option is only meaningful in conjunction with @option{-membedded-data}.
8046 @item -msplit-addresses
8047 @itemx -mno-split-addresses
8048 @opindex msplit-addresses
8049 @opindex mno-split-addresses
8050 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
8051 relocation operators. This option has been superceded by
8052 @option{-mexplicit-relocs} but is retained for backwards compatibility.
8054 @item -mexplicit-relocs
8055 @itemx -mno-explicit-relocs
8056 @opindex mexplicit-relocs
8057 @opindex mno-explicit-relocs
8058 Use (do not use) assembler relocation operators when dealing with symbolic
8059 addresses. The alternative, selected by @option{-mno-explicit-relocs},
8060 is to use assembler macros instead.
8062 @option{-mexplicit-relocs} is usually the default if GCC was configured
8063 to use an assembler that supports relocation operators. However, the
8064 combination of @option{-mabicalls} and @option{-fno-unit-at-a-time}
8065 implies @option{-mno-explicit-relocs} unless explicitly overridden.
8066 This is because, when generating abicalls, the choice of relocation
8067 depends on whether a symbol is local or global. In some rare cases,
8068 GCC will not be able to decide this until the whole compilation unit
8075 Generate (do not generate) code that refers to registers using their
8076 software names. The default is @option{-mno-rnames}, which tells GCC
8077 to use hardware names like @samp{$4} instead of software names like
8078 @samp{a0}. The only assembler known to support @option{-rnames} is
8079 the Algorithmics assembler.
8081 @item -mcheck-zero-division
8082 @itemx -mno-check-zero-division
8083 @opindex mcheck-zero-division
8084 @opindex mno-check-zero-division
8085 Trap (do not trap) on integer division by zero. The default is
8086 @option{-mcheck-zero-division}.
8092 Force (do not force) the use of @code{memcpy()} for non-trivial block
8093 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
8094 most constant-sized copies.
8097 @itemx -mno-long-calls
8098 @opindex mlong-calls
8099 @opindex mno-long-calls
8100 Disable (do not disable) use of the @code{jal} instruction. Calling
8101 functions using @code{jal} is more efficient but requires the caller
8102 and callee to be in the same 256 megabyte segment.
8104 This option has no effect on abicalls code. The default is
8105 @option{-mno-long-calls}.
8111 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
8112 instructions, as provided by the R4650 ISA.
8115 @itemx -mno-fused-madd
8116 @opindex mfused-madd
8117 @opindex mno-fused-madd
8118 Enable (disable) use of the floating point multiply-accumulate
8119 instructions, when they are available. The default is
8120 @option{-mfused-madd}.
8122 When multiply-accumulate instructions are used, the intermediate
8123 product is calculated to infinite precision and is not subject to
8124 the FCSR Flush to Zero bit. This may be undesirable in some
8129 Tell the MIPS assembler to not run its preprocessor over user
8130 assembler files (with a @samp{.s} suffix) when assembling them.
8133 @itemx -mno-fix-r4000
8135 @opindex mno-fix-r4000
8136 Work around certain R4000 CPU errata:
8139 A double-word or a variable shift may give an incorrect result if executed
8140 immediately after starting an integer division.
8142 A double-word or a variable shift may give an incorrect result if executed
8143 while an integer multiplication is in progress.
8145 An integer division may give an incorrect result if started in a delay slot
8146 of a taken branch or a jump.
8150 @itemx -mno-fix-r4400
8152 @opindex mno-fix-r4400
8153 Work around certain R4400 CPU errata:
8156 A double-word or a variable shift may give an incorrect result if executed
8157 immediately after starting an integer division.
8161 @itemx -mno-fix-vr4120
8162 @opindex mfix-vr4120
8163 Work around certain VR4120 errata:
8166 @code{dmultu} does not always produce the correct result.
8168 @code{div} and @code{ddiv} do not always produce the correct result if one
8169 of the operands is negative.
8171 The workarounds for the division errata rely on special functions in
8172 @file{libgcc.a}. At present, these functions are only provided by
8173 the @code{mips64vr*-elf} configurations.
8175 Other VR4120 errata require a nop to be inserted between certain pairs of
8176 instructions. These errata are handled by the assembler, not by GCC itself.
8181 Work around certain SB-1 CPU core errata.
8182 (This flag currently works around the SB-1 revision 2
8183 ``F1'' and ``F2'' floating point errata.)
8185 @item -mflush-func=@var{func}
8186 @itemx -mno-flush-func
8187 @opindex mflush-func
8188 Specifies the function to call to flush the I and D caches, or to not
8189 call any such function. If called, the function must take the same
8190 arguments as the common @code{_flush_func()}, that is, the address of the
8191 memory range for which the cache is being flushed, the size of the
8192 memory range, and the number 3 (to flush both caches). The default
8193 depends on the target GCC was configured for, but commonly is either
8194 @samp{_flush_func} or @samp{__cpu_flush}.
8196 @item -mbranch-likely
8197 @itemx -mno-branch-likely
8198 @opindex mbranch-likely
8199 @opindex mno-branch-likely
8200 Enable or disable use of Branch Likely instructions, regardless of the
8201 default for the selected architecture. By default, Branch Likely
8202 instructions may be generated if they are supported by the selected
8203 architecture. An exception is for the MIPS32 and MIPS64 architectures
8204 and processors which implement those architectures; for those, Branch
8205 Likely instructions will not be generated by default because the MIPS32
8206 and MIPS64 architectures specifically deprecate their use.
8208 @item -mfp-exceptions
8209 @itemx -mno-fp-exceptions
8210 @opindex mfp-exceptions
8211 Specifies whether FP exceptions are enabled. This affects how we schedule
8212 FP instructions for some processors. The default is that FP exceptions are
8215 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
8216 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
8220 @node i386 and x86-64 Options
8221 @subsection Intel 386 and AMD x86-64 Options
8222 @cindex i386 Options
8223 @cindex x86-64 Options
8224 @cindex Intel 386 Options
8225 @cindex AMD x86-64 Options
8227 These @samp{-m} options are defined for the i386 and x86-64 family of
8231 @item -mtune=@var{cpu-type}
8233 Tune to @var{cpu-type} everything applicable about the generated code, except
8234 for the ABI and the set of available instructions. The choices for
8238 Original Intel's i386 CPU.
8240 Intel's i486 CPU. (No scheduling is implemented for this chip.)
8242 Intel Pentium CPU with no MMX support.
8244 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
8245 @item i686, pentiumpro
8246 Intel PentiumPro CPU.
8248 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
8249 @item pentium3, pentium3m
8250 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
8253 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
8254 support. Used by Centrino notebooks.
8255 @item pentium4, pentium4m
8256 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
8258 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
8261 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
8262 SSE2 and SSE3 instruction set support.
8264 AMD K6 CPU with MMX instruction set support.
8266 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
8267 @item athlon, athlon-tbird
8268 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
8270 @item athlon-4, athlon-xp, athlon-mp
8271 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
8272 instruction set support.
8273 @item k8, opteron, athlon64, athlon-fx
8274 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
8275 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
8277 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
8280 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
8281 instruction set support.
8283 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
8284 implemented for this chip.)
8286 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
8287 implemented for this chip.)
8290 While picking a specific @var{cpu-type} will schedule things appropriately
8291 for that particular chip, the compiler will not generate any code that
8292 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8295 @item -march=@var{cpu-type}
8297 Generate instructions for the machine type @var{cpu-type}. The choices
8298 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8299 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8301 @item -mcpu=@var{cpu-type}
8303 A deprecated synonym for @option{-mtune}.
8312 @opindex mpentiumpro
8313 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8314 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8315 These synonyms are deprecated.
8317 @item -mfpmath=@var{unit}
8319 Generate floating point arithmetics for selected unit @var{unit}. The choices
8324 Use the standard 387 floating point coprocessor present majority of chips and
8325 emulated otherwise. Code compiled with this option will run almost everywhere.
8326 The temporary results are computed in 80bit precision instead of precision
8327 specified by the type resulting in slightly different results compared to most
8328 of other chips. See @option{-ffloat-store} for more detailed description.
8330 This is the default choice for i386 compiler.
8333 Use scalar floating point instructions present in the SSE instruction set.
8334 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8335 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8336 instruction set supports only single precision arithmetics, thus the double and
8337 extended precision arithmetics is still done using 387. Later version, present
8338 only in Pentium4 and the future AMD x86-64 chips supports double precision
8341 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8342 @option{-msse2} switches to enable SSE extensions and make this option
8343 effective. For x86-64 compiler, these extensions are enabled by default.
8345 The resulting code should be considerably faster in the majority of cases and avoid
8346 the numerical instability problems of 387 code, but may break some existing
8347 code that expects temporaries to be 80bit.
8349 This is the default choice for the x86-64 compiler.
8352 Attempt to utilize both instruction sets at once. This effectively double the
8353 amount of available registers and on chips with separate execution units for
8354 387 and SSE the execution resources too. Use this option with care, as it is
8355 still experimental, because the GCC register allocator does not model separate
8356 functional units well resulting in instable performance.
8359 @item -masm=@var{dialect}
8360 @opindex masm=@var{dialect}
8361 Output asm instructions using selected @var{dialect}. Supported choices are
8362 @samp{intel} or @samp{att} (the default one).
8367 @opindex mno-ieee-fp
8368 Control whether or not the compiler uses IEEE floating point
8369 comparisons. These handle correctly the case where the result of a
8370 comparison is unordered.
8373 @opindex msoft-float
8374 Generate output containing library calls for floating point.
8375 @strong{Warning:} the requisite libraries are not part of GCC@.
8376 Normally the facilities of the machine's usual C compiler are used, but
8377 this can't be done directly in cross-compilation. You must make your
8378 own arrangements to provide suitable library functions for
8381 On machines where a function returns floating point results in the 80387
8382 register stack, some floating point opcodes may be emitted even if
8383 @option{-msoft-float} is used.
8385 @item -mno-fp-ret-in-387
8386 @opindex mno-fp-ret-in-387
8387 Do not use the FPU registers for return values of functions.
8389 The usual calling convention has functions return values of types
8390 @code{float} and @code{double} in an FPU register, even if there
8391 is no FPU@. The idea is that the operating system should emulate
8394 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8395 in ordinary CPU registers instead.
8397 @item -mno-fancy-math-387
8398 @opindex mno-fancy-math-387
8399 Some 387 emulators do not support the @code{sin}, @code{cos} and
8400 @code{sqrt} instructions for the 387. Specify this option to avoid
8401 generating those instructions. This option is the default on FreeBSD,
8402 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8403 indicates that the target cpu will always have an FPU and so the
8404 instruction will not need emulation. As of revision 2.6.1, these
8405 instructions are not generated unless you also use the
8406 @option{-funsafe-math-optimizations} switch.
8408 @item -malign-double
8409 @itemx -mno-align-double
8410 @opindex malign-double
8411 @opindex mno-align-double
8412 Control whether GCC aligns @code{double}, @code{long double}, and
8413 @code{long long} variables on a two word boundary or a one word
8414 boundary. Aligning @code{double} variables on a two word boundary will
8415 produce code that runs somewhat faster on a @samp{Pentium} at the
8416 expense of more memory.
8418 @strong{Warning:} if you use the @option{-malign-double} switch,
8419 structures containing the above types will be aligned differently than
8420 the published application binary interface specifications for the 386
8421 and will not be binary compatible with structures in code compiled
8422 without that switch.
8424 @item -m96bit-long-double
8425 @itemx -m128bit-long-double
8426 @opindex m96bit-long-double
8427 @opindex m128bit-long-double
8428 These switches control the size of @code{long double} type. The i386
8429 application binary interface specifies the size to be 96 bits,
8430 so @option{-m96bit-long-double} is the default in 32 bit mode.
8432 Modern architectures (Pentium and newer) would prefer @code{long double}
8433 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8434 conforming to the ABI, this would not be possible. So specifying a
8435 @option{-m128bit-long-double} will align @code{long double}
8436 to a 16 byte boundary by padding the @code{long double} with an additional
8439 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8440 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8442 Notice that neither of these options enable any extra precision over the x87
8443 standard of 80 bits for a @code{long double}.
8445 @strong{Warning:} if you override the default value for your target ABI, the
8446 structures and arrays containing @code{long double} variables will change
8447 their size as well as function calling convention for function taking
8448 @code{long double} will be modified. Hence they will not be binary
8449 compatible with arrays or structures in code compiled without that switch.
8453 @itemx -mno-svr3-shlib
8454 @opindex msvr3-shlib
8455 @opindex mno-svr3-shlib
8456 Control whether GCC places uninitialized local variables into the
8457 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8458 into @code{bss}. These options are meaningful only on System V Release 3.
8462 Use a different function-calling convention, in which functions that
8463 take a fixed number of arguments return with the @code{ret} @var{num}
8464 instruction, which pops their arguments while returning. This saves one
8465 instruction in the caller since there is no need to pop the arguments
8468 You can specify that an individual function is called with this calling
8469 sequence with the function attribute @samp{stdcall}. You can also
8470 override the @option{-mrtd} option by using the function attribute
8471 @samp{cdecl}. @xref{Function Attributes}.
8473 @strong{Warning:} this calling convention is incompatible with the one
8474 normally used on Unix, so you cannot use it if you need to call
8475 libraries compiled with the Unix compiler.
8477 Also, you must provide function prototypes for all functions that
8478 take variable numbers of arguments (including @code{printf});
8479 otherwise incorrect code will be generated for calls to those
8482 In addition, seriously incorrect code will result if you call a
8483 function with too many arguments. (Normally, extra arguments are
8484 harmlessly ignored.)
8486 @item -mregparm=@var{num}
8488 Control how many registers are used to pass integer arguments. By
8489 default, no registers are used to pass arguments, and at most 3
8490 registers can be used. You can control this behavior for a specific
8491 function by using the function attribute @samp{regparm}.
8492 @xref{Function Attributes}.
8494 @strong{Warning:} if you use this switch, and
8495 @var{num} is nonzero, then you must build all modules with the same
8496 value, including any libraries. This includes the system libraries and
8499 @item -mpreferred-stack-boundary=@var{num}
8500 @opindex mpreferred-stack-boundary
8501 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8502 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8503 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8504 size (@option{-Os}), in which case the default is the minimum correct
8505 alignment (4 bytes for x86, and 8 bytes for x86-64).
8507 On Pentium and PentiumPro, @code{double} and @code{long double} values
8508 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8509 suffer significant run time performance penalties. On Pentium III, the
8510 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8511 penalties if it is not 16 byte aligned.
8513 To ensure proper alignment of this values on the stack, the stack boundary
8514 must be as aligned as that required by any value stored on the stack.
8515 Further, every function must be generated such that it keeps the stack
8516 aligned. Thus calling a function compiled with a higher preferred
8517 stack boundary from a function compiled with a lower preferred stack
8518 boundary will most likely misalign the stack. It is recommended that
8519 libraries that use callbacks always use the default setting.
8521 This extra alignment does consume extra stack space, and generally
8522 increases code size. Code that is sensitive to stack space usage, such
8523 as embedded systems and operating system kernels, may want to reduce the
8524 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8542 These switches enable or disable the use of built-in functions that allow
8543 direct access to the MMX, SSE, SSE2, SSE3 and 3Dnow extensions of the
8546 @xref{X86 Built-in Functions}, for details of the functions enabled
8547 and disabled by these switches.
8549 To have SSE/SSE2 instructions generated automatically from floating-point
8550 code, see @option{-mfpmath=sse}.
8553 @itemx -mno-push-args
8555 @opindex mno-push-args
8556 Use PUSH operations to store outgoing parameters. This method is shorter
8557 and usually equally fast as method using SUB/MOV operations and is enabled
8558 by default. In some cases disabling it may improve performance because of
8559 improved scheduling and reduced dependencies.
8561 @item -maccumulate-outgoing-args
8562 @opindex maccumulate-outgoing-args
8563 If enabled, the maximum amount of space required for outgoing arguments will be
8564 computed in the function prologue. This is faster on most modern CPUs
8565 because of reduced dependencies, improved scheduling and reduced stack usage
8566 when preferred stack boundary is not equal to 2. The drawback is a notable
8567 increase in code size. This switch implies @option{-mno-push-args}.
8571 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8572 on thread-safe exception handling must compile and link all code with the
8573 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8574 @option{-D_MT}; when linking, it links in a special thread helper library
8575 @option{-lmingwthrd} which cleans up per thread exception handling data.
8577 @item -mno-align-stringops
8578 @opindex mno-align-stringops
8579 Do not align destination of inlined string operations. This switch reduces
8580 code size and improves performance in case the destination is already aligned,
8581 but GCC doesn't know about it.
8583 @item -minline-all-stringops
8584 @opindex minline-all-stringops
8585 By default GCC inlines string operations only when destination is known to be
8586 aligned at least to 4 byte boundary. This enables more inlining, increase code
8587 size, but may improve performance of code that depends on fast memcpy, strlen
8588 and memset for short lengths.
8590 @item -momit-leaf-frame-pointer
8591 @opindex momit-leaf-frame-pointer
8592 Don't keep the frame pointer in a register for leaf functions. This
8593 avoids the instructions to save, set up and restore frame pointers and
8594 makes an extra register available in leaf functions. The option
8595 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8596 which might make debugging harder.
8598 @item -mtls-direct-seg-refs
8599 @itemx -mno-tls-direct-seg-refs
8600 @opindex mtls-direct-seg-refs
8601 Controls whether TLS variables may be accessed with offsets from the
8602 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8603 or whether the thread base pointer must be added. Whether or not this
8604 is legal depends on the operating system, and whether it maps the
8605 segment to cover the entire TLS area.
8607 For systems that use GNU libc, the default is on.
8610 These @samp{-m} switches are supported in addition to the above
8611 on AMD x86-64 processors in 64-bit environments.
8618 Generate code for a 32-bit or 64-bit environment.
8619 The 32-bit environment sets int, long and pointer to 32 bits and
8620 generates code that runs on any i386 system.
8621 The 64-bit environment sets int to 32 bits and long and pointer
8622 to 64 bits and generates code for AMD's x86-64 architecture.
8625 @opindex no-red-zone
8626 Do not use a so called red zone for x86-64 code. The red zone is mandated
8627 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8628 stack pointer that will not be modified by signal or interrupt handlers
8629 and therefore can be used for temporary data without adjusting the stack
8630 pointer. The flag @option{-mno-red-zone} disables this red zone.
8632 @item -mcmodel=small
8633 @opindex mcmodel=small
8634 Generate code for the small code model: the program and its symbols must
8635 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8636 Programs can be statically or dynamically linked. This is the default
8639 @item -mcmodel=kernel
8640 @opindex mcmodel=kernel
8641 Generate code for the kernel code model. The kernel runs in the
8642 negative 2 GB of the address space.
8643 This model has to be used for Linux kernel code.
8645 @item -mcmodel=medium
8646 @opindex mcmodel=medium
8647 Generate code for the medium model: The program is linked in the lower 2
8648 GB of the address space but symbols can be located anywhere in the
8649 address space. Programs can be statically or dynamically linked, but
8650 building of shared libraries are not supported with the medium model.
8652 @item -mcmodel=large
8653 @opindex mcmodel=large
8654 Generate code for the large model: This model makes no assumptions
8655 about addresses and sizes of sections. Currently GCC does not implement
8660 @subsection HPPA Options
8661 @cindex HPPA Options
8663 These @samp{-m} options are defined for the HPPA family of computers:
8666 @item -march=@var{architecture-type}
8668 Generate code for the specified architecture. The choices for
8669 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8670 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8671 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8672 architecture option for your machine. Code compiled for lower numbered
8673 architectures will run on higher numbered architectures, but not the
8676 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8677 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8681 @itemx -mpa-risc-1-1
8682 @itemx -mpa-risc-2-0
8683 @opindex mpa-risc-1-0
8684 @opindex mpa-risc-1-1
8685 @opindex mpa-risc-2-0
8686 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8689 @opindex mbig-switch
8690 Generate code suitable for big switch tables. Use this option only if
8691 the assembler/linker complain about out of range branches within a switch
8694 @item -mjump-in-delay
8695 @opindex mjump-in-delay
8696 Fill delay slots of function calls with unconditional jump instructions
8697 by modifying the return pointer for the function call to be the target
8698 of the conditional jump.
8700 @item -mdisable-fpregs
8701 @opindex mdisable-fpregs
8702 Prevent floating point registers from being used in any manner. This is
8703 necessary for compiling kernels which perform lazy context switching of
8704 floating point registers. If you use this option and attempt to perform
8705 floating point operations, the compiler will abort.
8707 @item -mdisable-indexing
8708 @opindex mdisable-indexing
8709 Prevent the compiler from using indexing address modes. This avoids some
8710 rather obscure problems when compiling MIG generated code under MACH@.
8712 @item -mno-space-regs
8713 @opindex mno-space-regs
8714 Generate code that assumes the target has no space registers. This allows
8715 GCC to generate faster indirect calls and use unscaled index address modes.
8717 Such code is suitable for level 0 PA systems and kernels.
8719 @item -mfast-indirect-calls
8720 @opindex mfast-indirect-calls
8721 Generate code that assumes calls never cross space boundaries. This
8722 allows GCC to emit code which performs faster indirect calls.
8724 This option will not work in the presence of shared libraries or nested
8727 @item -mlong-load-store
8728 @opindex mlong-load-store
8729 Generate 3-instruction load and store sequences as sometimes required by
8730 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8733 @item -mportable-runtime
8734 @opindex mportable-runtime
8735 Use the portable calling conventions proposed by HP for ELF systems.
8739 Enable the use of assembler directives only GAS understands.
8741 @item -mschedule=@var{cpu-type}
8743 Schedule code according to the constraints for the machine type
8744 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8745 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8746 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8747 proper scheduling option for your machine. The default scheduling is
8751 @opindex mlinker-opt
8752 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8753 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8754 linkers in which they give bogus error messages when linking some programs.
8757 @opindex msoft-float
8758 Generate output containing library calls for floating point.
8759 @strong{Warning:} the requisite libraries are not available for all HPPA
8760 targets. Normally the facilities of the machine's usual C compiler are
8761 used, but this cannot be done directly in cross-compilation. You must make
8762 your own arrangements to provide suitable library functions for
8763 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8764 does provide software floating point support.
8766 @option{-msoft-float} changes the calling convention in the output file;
8767 therefore, it is only useful if you compile @emph{all} of a program with
8768 this option. In particular, you need to compile @file{libgcc.a}, the
8769 library that comes with GCC, with @option{-msoft-float} in order for
8774 Generate the predefine, @code{_SIO}, for server IO. The default is
8775 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8776 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8777 options are available under HP-UX and HI-UX.
8781 Use GNU ld specific options. This passes @option{-shared} to ld when
8782 building a shared library. It is the default when GCC is configured,
8783 explicitly or implicitly, with the GNU linker. This option does not
8784 have any affect on which ld is called, it only changes what parameters
8785 are passed to that ld. The ld that is called is determined by the
8786 @option{--with-ld} configure option, GCC's program search path, and
8787 finally by the user's @env{PATH}. The linker used by GCC can be printed
8788 using @samp{which `gcc -print-prog-name=ld`}.
8792 Use HP ld specific options. This passes @option{-b} to ld when building
8793 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8794 links. It is the default when GCC is configured, explicitly or
8795 implicitly, with the HP linker. This option does not have any affect on
8796 which ld is called, it only changes what parameters are passed to that
8797 ld. The ld that is called is determined by the @option{--with-ld}
8798 configure option, GCC's program search path, and finally by the user's
8799 @env{PATH}. The linker used by GCC can be printed using @samp{which
8800 `gcc -print-prog-name=ld`}.
8805 Select the FDPIC ABI, that uses function descriptors to represent
8806 pointers to functions. Without any PIC/PIE-related options, it
8807 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
8808 assumes GOT entries and small data are within a 12-bit range from the
8809 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
8810 are computed with 32 bits.
8813 @opindex minline-plt
8815 Enable inlining of PLT entries in function calls to functions that are
8816 not known to bind locally. It has no effect without @option{-mfdpic}.
8817 It's enabled by default if optimizing for speed and compiling for
8818 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
8819 optimization option such as @option{-O3} or above is present in the
8825 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
8826 that is known to be in read-only sections. It's enabled by default,
8827 except for @option{-fpic} or @option{-fpie}: even though it may help
8828 make the global offset table smaller, it trades 1 instruction for 4.
8829 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
8830 one of which may be shared by multiple symbols, and it avoids the need
8831 for a GOT entry for the referenced symbol, so it's more likely to be a
8832 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
8834 @item -multilib-library-pic
8835 @opindex multilib-library-pic
8837 Link with the (library, not FD) pic libraries. It's implied by
8838 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
8839 @option{-fpic} without @option{-mfdpic}. You should never have to use
8845 Follow the EABI requirement of always creating a frame pointer whenever
8846 a stack frame is allocated. This option is enabled by default and can
8847 be disabled with @option{-mno-linked-fp}.
8850 @opindex mno-long-calls
8851 Generate code that uses long call sequences. This ensures that a call
8852 is always able to reach linker generated stubs. The default is to generate
8853 long calls only when the distance from the call site to the beginning
8854 of the function or translation unit, as the case may be, exceeds a
8855 predefined limit set by the branch type being used. The limits for
8856 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8857 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8860 Distances are measured from the beginning of functions when using the
8861 @option{-ffunction-sections} option, or when using the @option{-mgas}
8862 and @option{-mno-portable-runtime} options together under HP-UX with
8865 It is normally not desirable to use this option as it will degrade
8866 performance. However, it may be useful in large applications,
8867 particularly when partial linking is used to build the application.
8869 The types of long calls used depends on the capabilities of the
8870 assembler and linker, and the type of code being generated. The
8871 impact on systems that support long absolute calls, and long pic
8872 symbol-difference or pc-relative calls should be relatively small.
8873 However, an indirect call is used on 32-bit ELF systems in pic code
8874 and it is quite long.
8878 Suppress the generation of link options to search libdld.sl when the
8879 @option{-static} option is specified on HP-UX 10 and later.
8883 The HP-UX implementation of setlocale in libc has a dependency on
8884 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8885 when the @option{-static} option is specified, special link options
8886 are needed to resolve this dependency.
8888 On HP-UX 10 and later, the GCC driver adds the necessary options to
8889 link with libdld.sl when the @option{-static} option is specified.
8890 This causes the resulting binary to be dynamic. On the 64-bit port,
8891 the linkers generate dynamic binaries by default in any case. The
8892 @option{-nolibdld} option can be used to prevent the GCC driver from
8893 adding these link options.
8897 Add support for multithreading with the @dfn{dce thread} library
8898 under HP-UX. This option sets flags for both the preprocessor and
8902 @node DEC Alpha Options
8903 @subsection DEC Alpha Options
8905 These @samp{-m} options are defined for the DEC Alpha implementations:
8908 @item -mno-soft-float
8910 @opindex mno-soft-float
8911 @opindex msoft-float
8912 Use (do not use) the hardware floating-point instructions for
8913 floating-point operations. When @option{-msoft-float} is specified,
8914 functions in @file{libgcc.a} will be used to perform floating-point
8915 operations. Unless they are replaced by routines that emulate the
8916 floating-point operations, or compiled in such a way as to call such
8917 emulations routines, these routines will issue floating-point
8918 operations. If you are compiling for an Alpha without floating-point
8919 operations, you must ensure that the library is built so as not to call
8922 Note that Alpha implementations without floating-point operations are
8923 required to have floating-point registers.
8928 @opindex mno-fp-regs
8929 Generate code that uses (does not use) the floating-point register set.
8930 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8931 register set is not used, floating point operands are passed in integer
8932 registers as if they were integers and floating-point results are passed
8933 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8934 so any function with a floating-point argument or return value called by code
8935 compiled with @option{-mno-fp-regs} must also be compiled with that
8938 A typical use of this option is building a kernel that does not use,
8939 and hence need not save and restore, any floating-point registers.
8943 The Alpha architecture implements floating-point hardware optimized for
8944 maximum performance. It is mostly compliant with the IEEE floating
8945 point standard. However, for full compliance, software assistance is
8946 required. This option generates code fully IEEE compliant code
8947 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8948 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8949 defined during compilation. The resulting code is less efficient but is
8950 able to correctly support denormalized numbers and exceptional IEEE
8951 values such as not-a-number and plus/minus infinity. Other Alpha
8952 compilers call this option @option{-ieee_with_no_inexact}.
8954 @item -mieee-with-inexact
8955 @opindex mieee-with-inexact
8956 This is like @option{-mieee} except the generated code also maintains
8957 the IEEE @var{inexact-flag}. Turning on this option causes the
8958 generated code to implement fully-compliant IEEE math. In addition to
8959 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8960 macro. On some Alpha implementations the resulting code may execute
8961 significantly slower than the code generated by default. Since there is
8962 very little code that depends on the @var{inexact-flag}, you should
8963 normally not specify this option. Other Alpha compilers call this
8964 option @option{-ieee_with_inexact}.
8966 @item -mfp-trap-mode=@var{trap-mode}
8967 @opindex mfp-trap-mode
8968 This option controls what floating-point related traps are enabled.
8969 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8970 The trap mode can be set to one of four values:
8974 This is the default (normal) setting. The only traps that are enabled
8975 are the ones that cannot be disabled in software (e.g., division by zero
8979 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8983 Like @samp{su}, but the instructions are marked to be safe for software
8984 completion (see Alpha architecture manual for details).
8987 Like @samp{su}, but inexact traps are enabled as well.
8990 @item -mfp-rounding-mode=@var{rounding-mode}
8991 @opindex mfp-rounding-mode
8992 Selects the IEEE rounding mode. Other Alpha compilers call this option
8993 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8998 Normal IEEE rounding mode. Floating point numbers are rounded towards
8999 the nearest machine number or towards the even machine number in case
9003 Round towards minus infinity.
9006 Chopped rounding mode. Floating point numbers are rounded towards zero.
9009 Dynamic rounding mode. A field in the floating point control register
9010 (@var{fpcr}, see Alpha architecture reference manual) controls the
9011 rounding mode in effect. The C library initializes this register for
9012 rounding towards plus infinity. Thus, unless your program modifies the
9013 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9016 @item -mtrap-precision=@var{trap-precision}
9017 @opindex mtrap-precision
9018 In the Alpha architecture, floating point traps are imprecise. This
9019 means without software assistance it is impossible to recover from a
9020 floating trap and program execution normally needs to be terminated.
9021 GCC can generate code that can assist operating system trap handlers
9022 in determining the exact location that caused a floating point trap.
9023 Depending on the requirements of an application, different levels of
9024 precisions can be selected:
9028 Program precision. This option is the default and means a trap handler
9029 can only identify which program caused a floating point exception.
9032 Function precision. The trap handler can determine the function that
9033 caused a floating point exception.
9036 Instruction precision. The trap handler can determine the exact
9037 instruction that caused a floating point exception.
9040 Other Alpha compilers provide the equivalent options called
9041 @option{-scope_safe} and @option{-resumption_safe}.
9043 @item -mieee-conformant
9044 @opindex mieee-conformant
9045 This option marks the generated code as IEEE conformant. You must not
9046 use this option unless you also specify @option{-mtrap-precision=i} and either
9047 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9048 is to emit the line @samp{.eflag 48} in the function prologue of the
9049 generated assembly file. Under DEC Unix, this has the effect that
9050 IEEE-conformant math library routines will be linked in.
9052 @item -mbuild-constants
9053 @opindex mbuild-constants
9054 Normally GCC examines a 32- or 64-bit integer constant to
9055 see if it can construct it from smaller constants in two or three
9056 instructions. If it cannot, it will output the constant as a literal and
9057 generate code to load it from the data segment at runtime.
9059 Use this option to require GCC to construct @emph{all} integer constants
9060 using code, even if it takes more instructions (the maximum is six).
9062 You would typically use this option to build a shared library dynamic
9063 loader. Itself a shared library, it must relocate itself in memory
9064 before it can find the variables and constants in its own data segment.
9070 Select whether to generate code to be assembled by the vendor-supplied
9071 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9089 Indicate whether GCC should generate code to use the optional BWX,
9090 CIX, FIX and MAX instruction sets. The default is to use the instruction
9091 sets supported by the CPU type specified via @option{-mcpu=} option or that
9092 of the CPU on which GCC was built if none was specified.
9097 @opindex mfloat-ieee
9098 Generate code that uses (does not use) VAX F and G floating point
9099 arithmetic instead of IEEE single and double precision.
9101 @item -mexplicit-relocs
9102 @itemx -mno-explicit-relocs
9103 @opindex mexplicit-relocs
9104 @opindex mno-explicit-relocs
9105 Older Alpha assemblers provided no way to generate symbol relocations
9106 except via assembler macros. Use of these macros does not allow
9107 optimal instruction scheduling. GNU binutils as of version 2.12
9108 supports a new syntax that allows the compiler to explicitly mark
9109 which relocations should apply to which instructions. This option
9110 is mostly useful for debugging, as GCC detects the capabilities of
9111 the assembler when it is built and sets the default accordingly.
9115 @opindex msmall-data
9116 @opindex mlarge-data
9117 When @option{-mexplicit-relocs} is in effect, static data is
9118 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9119 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9120 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9121 16-bit relocations off of the @code{$gp} register. This limits the
9122 size of the small data area to 64KB, but allows the variables to be
9123 directly accessed via a single instruction.
9125 The default is @option{-mlarge-data}. With this option the data area
9126 is limited to just below 2GB. Programs that require more than 2GB of
9127 data must use @code{malloc} or @code{mmap} to allocate the data in the
9128 heap instead of in the program's data segment.
9130 When generating code for shared libraries, @option{-fpic} implies
9131 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9135 @opindex msmall-text
9136 @opindex mlarge-text
9137 When @option{-msmall-text} is used, the compiler assumes that the
9138 code of the entire program (or shared library) fits in 4MB, and is
9139 thus reachable with a branch instruction. When @option{-msmall-data}
9140 is used, the compiler can assume that all local symbols share the
9141 same @code{$gp} value, and thus reduce the number of instructions
9142 required for a function call from 4 to 1.
9144 The default is @option{-mlarge-text}.
9146 @item -mcpu=@var{cpu_type}
9148 Set the instruction set and instruction scheduling parameters for
9149 machine type @var{cpu_type}. You can specify either the @samp{EV}
9150 style name or the corresponding chip number. GCC supports scheduling
9151 parameters for the EV4, EV5 and EV6 family of processors and will
9152 choose the default values for the instruction set from the processor
9153 you specify. If you do not specify a processor type, GCC will default
9154 to the processor on which the compiler was built.
9156 Supported values for @var{cpu_type} are
9162 Schedules as an EV4 and has no instruction set extensions.
9166 Schedules as an EV5 and has no instruction set extensions.
9170 Schedules as an EV5 and supports the BWX extension.
9175 Schedules as an EV5 and supports the BWX and MAX extensions.
9179 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9183 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9186 @item -mtune=@var{cpu_type}
9188 Set only the instruction scheduling parameters for machine type
9189 @var{cpu_type}. The instruction set is not changed.
9191 @item -mmemory-latency=@var{time}
9192 @opindex mmemory-latency
9193 Sets the latency the scheduler should assume for typical memory
9194 references as seen by the application. This number is highly
9195 dependent on the memory access patterns used by the application
9196 and the size of the external cache on the machine.
9198 Valid options for @var{time} are
9202 A decimal number representing clock cycles.
9208 The compiler contains estimates of the number of clock cycles for
9209 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9210 (also called Dcache, Scache, and Bcache), as well as to main memory.
9211 Note that L3 is only valid for EV5.
9216 @node DEC Alpha/VMS Options
9217 @subsection DEC Alpha/VMS Options
9219 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9222 @item -mvms-return-codes
9223 @opindex mvms-return-codes
9224 Return VMS condition codes from main. The default is to return POSIX
9225 style condition (e.g.@ error) codes.
9228 @node H8/300 Options
9229 @subsection H8/300 Options
9231 These @samp{-m} options are defined for the H8/300 implementations:
9236 Shorten some address references at link time, when possible; uses the
9237 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9238 ld, Using ld}, for a fuller description.
9242 Generate code for the H8/300H@.
9246 Generate code for the H8S@.
9250 Generate code for the H8S and H8/300H in the normal mode. This switch
9251 must be used either with -mh or -ms.
9255 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9259 Make @code{int} data 32 bits by default.
9263 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9264 The default for the H8/300H and H8S is to align longs and floats on 4
9266 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9267 This option has no effect on the H8/300.
9271 @subsection SH Options
9273 These @samp{-m} options are defined for the SH implementations:
9278 Generate code for the SH1.
9282 Generate code for the SH2.
9285 Generate code for the SH2e.
9289 Generate code for the SH3.
9293 Generate code for the SH3e.
9297 Generate code for the SH4 without a floating-point unit.
9299 @item -m4-single-only
9300 @opindex m4-single-only
9301 Generate code for the SH4 with a floating-point unit that only
9302 supports single-precision arithmetic.
9306 Generate code for the SH4 assuming the floating-point unit is in
9307 single-precision mode by default.
9311 Generate code for the SH4.
9315 Compile code for the processor in big endian mode.
9319 Compile code for the processor in little endian mode.
9323 Align doubles at 64-bit boundaries. Note that this changes the calling
9324 conventions, and thus some functions from the standard C library will
9325 not work unless you recompile it first with @option{-mdalign}.
9329 Shorten some address references at link time, when possible; uses the
9330 linker option @option{-relax}.
9334 Use 32-bit offsets in @code{switch} tables. The default is to use
9339 Enable the use of the instruction @code{fmovd}.
9343 Comply with the calling conventions defined by Renesas.
9347 Mark the @code{MAC} register as call-clobbered, even if
9348 @option{-mhitachi} is given.
9352 Increase IEEE-compliance of floating-point code.
9356 Dump instruction size and location in the assembly code.
9360 This option is deprecated. It pads structures to multiple of 4 bytes,
9361 which is incompatible with the SH ABI@.
9365 Optimize for space instead of speed. Implied by @option{-Os}.
9369 When generating position-independent code, emit function calls using
9370 the Global Offset Table instead of the Procedure Linkage Table.
9374 Generate a library function call to invalidate instruction cache
9375 entries, after fixing up a trampoline. This library function call
9376 doesn't assume it can write to the whole memory address space. This
9377 is the default when the target is @code{sh-*-linux*}.
9380 @node System V Options
9381 @subsection Options for System V
9383 These additional options are available on System V Release 4 for
9384 compatibility with other compilers on those systems:
9389 Create a shared object.
9390 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9394 Identify the versions of each tool used by the compiler, in a
9395 @code{.ident} assembler directive in the output.
9399 Refrain from adding @code{.ident} directives to the output file (this is
9402 @item -YP,@var{dirs}
9404 Search the directories @var{dirs}, and no others, for libraries
9405 specified with @option{-l}.
9409 Look in the directory @var{dir} to find the M4 preprocessor.
9410 The assembler uses this option.
9411 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9412 @c the generic assembler that comes with Solaris takes just -Ym.
9415 @node TMS320C3x/C4x Options
9416 @subsection TMS320C3x/C4x Options
9417 @cindex TMS320C3x/C4x Options
9419 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9423 @item -mcpu=@var{cpu_type}
9425 Set the instruction set, register set, and instruction scheduling
9426 parameters for machine type @var{cpu_type}. Supported values for
9427 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9428 @samp{c44}. The default is @samp{c40} to generate code for the
9433 @itemx -msmall-memory
9435 @opindex mbig-memory
9437 @opindex msmall-memory
9439 Generates code for the big or small memory model. The small memory
9440 model assumed that all data fits into one 64K word page. At run-time
9441 the data page (DP) register must be set to point to the 64K page
9442 containing the .bss and .data program sections. The big memory model is
9443 the default and requires reloading of the DP register for every direct
9450 Allow (disallow) allocation of general integer operands into the block
9457 Enable (disable) generation of code using decrement and branch,
9458 DBcond(D), instructions. This is enabled by default for the C4x. To be
9459 on the safe side, this is disabled for the C3x, since the maximum
9460 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9461 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9462 that it can utilize the decrement and branch instruction, but will give
9463 up if there is more than one memory reference in the loop. Thus a loop
9464 where the loop counter is decremented can generate slightly more
9465 efficient code, in cases where the RPTB instruction cannot be utilized.
9467 @item -mdp-isr-reload
9469 @opindex mdp-isr-reload
9471 Force the DP register to be saved on entry to an interrupt service
9472 routine (ISR), reloaded to point to the data section, and restored on
9473 exit from the ISR@. This should not be required unless someone has
9474 violated the small memory model by modifying the DP register, say within
9481 For the C3x use the 24-bit MPYI instruction for integer multiplies
9482 instead of a library call to guarantee 32-bit results. Note that if one
9483 of the operands is a constant, then the multiplication will be performed
9484 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9485 then squaring operations are performed inline instead of a library call.
9488 @itemx -mno-fast-fix
9490 @opindex mno-fast-fix
9491 The C3x/C4x FIX instruction to convert a floating point value to an
9492 integer value chooses the nearest integer less than or equal to the
9493 floating point value rather than to the nearest integer. Thus if the
9494 floating point number is negative, the result will be incorrectly
9495 truncated an additional code is necessary to detect and correct this
9496 case. This option can be used to disable generation of the additional
9497 code required to correct the result.
9503 Enable (disable) generation of repeat block sequences using the RPTB
9504 instruction for zero overhead looping. The RPTB construct is only used
9505 for innermost loops that do not call functions or jump across the loop
9506 boundaries. There is no advantage having nested RPTB loops due to the
9507 overhead required to save and restore the RC, RS, and RE registers.
9508 This is enabled by default with @option{-O2}.
9510 @item -mrpts=@var{count}
9514 Enable (disable) the use of the single instruction repeat instruction
9515 RPTS@. If a repeat block contains a single instruction, and the loop
9516 count can be guaranteed to be less than the value @var{count}, GCC will
9517 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9518 then a RPTS will be emitted even if the loop count cannot be determined
9519 at compile time. Note that the repeated instruction following RPTS does
9520 not have to be reloaded from memory each iteration, thus freeing up the
9521 CPU buses for operands. However, since interrupts are blocked by this
9522 instruction, it is disabled by default.
9524 @item -mloop-unsigned
9525 @itemx -mno-loop-unsigned
9526 @opindex mloop-unsigned
9527 @opindex mno-loop-unsigned
9528 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9529 is @math{2^{31} + 1} since these instructions test if the iteration count is
9530 negative to terminate the loop. If the iteration count is unsigned
9531 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9532 exceeded. This switch allows an unsigned iteration count.
9536 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9537 with. This also enforces compatibility with the API employed by the TI
9538 C3x C compiler. For example, long doubles are passed as structures
9539 rather than in floating point registers.
9545 Generate code that uses registers (stack) for passing arguments to functions.
9546 By default, arguments are passed in registers where possible rather
9547 than by pushing arguments on to the stack.
9549 @item -mparallel-insns
9550 @itemx -mno-parallel-insns
9551 @opindex mparallel-insns
9552 @opindex mno-parallel-insns
9553 Allow the generation of parallel instructions. This is enabled by
9554 default with @option{-O2}.
9556 @item -mparallel-mpy
9557 @itemx -mno-parallel-mpy
9558 @opindex mparallel-mpy
9559 @opindex mno-parallel-mpy
9560 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9561 provided @option{-mparallel-insns} is also specified. These instructions have
9562 tight register constraints which can pessimize the code generation
9568 @subsection V850 Options
9569 @cindex V850 Options
9571 These @samp{-m} options are defined for V850 implementations:
9575 @itemx -mno-long-calls
9576 @opindex mlong-calls
9577 @opindex mno-long-calls
9578 Treat all calls as being far away (near). If calls are assumed to be
9579 far away, the compiler will always load the functions address up into a
9580 register, and call indirect through the pointer.
9586 Do not optimize (do optimize) basic blocks that use the same index
9587 pointer 4 or more times to copy pointer into the @code{ep} register, and
9588 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9589 option is on by default if you optimize.
9591 @item -mno-prolog-function
9592 @itemx -mprolog-function
9593 @opindex mno-prolog-function
9594 @opindex mprolog-function
9595 Do not use (do use) external functions to save and restore registers
9596 at the prologue and epilogue of a function. The external functions
9597 are slower, but use less code space if more than one function saves
9598 the same number of registers. The @option{-mprolog-function} option
9599 is on by default if you optimize.
9603 Try to make the code as small as possible. At present, this just turns
9604 on the @option{-mep} and @option{-mprolog-function} options.
9608 Put static or global variables whose size is @var{n} bytes or less into
9609 the tiny data area that register @code{ep} points to. The tiny data
9610 area can hold up to 256 bytes in total (128 bytes for byte references).
9614 Put static or global variables whose size is @var{n} bytes or less into
9615 the small data area that register @code{gp} points to. The small data
9616 area can hold up to 64 kilobytes.
9620 Put static or global variables whose size is @var{n} bytes or less into
9621 the first 32 kilobytes of memory.
9625 Specify that the target processor is the V850.
9628 @opindex mbig-switch
9629 Generate code suitable for big switch tables. Use this option only if
9630 the assembler/linker complain about out of range branches within a switch
9635 This option will cause r2 and r5 to be used in the code generated by
9636 the compiler. This setting is the default.
9639 @opindex mno-app-regs
9640 This option will cause r2 and r5 to be treated as fixed registers.
9644 Specify that the target processor is the V850E1. The preprocessor
9645 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
9646 this option is used.
9650 Specify that the target processor is the V850E. The preprocessor
9651 constant @samp{__v850e__} will be defined if this option is used.
9653 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
9654 are defined then a default target processor will be chosen and the
9655 relevant @samp{__v850*__} preprocessor constant will be defined.
9657 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9658 defined, regardless of which processor variant is the target.
9660 @item -mdisable-callt
9661 @opindex mdisable-callt
9662 This option will suppress generation of the CALLT instruction for the
9663 v850e and v850e1 flavors of the v850 architecture. The default is
9664 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9669 @subsection ARC Options
9672 These options are defined for ARC implementations:
9677 Compile code for little endian mode. This is the default.
9681 Compile code for big endian mode.
9684 @opindex mmangle-cpu
9685 Prepend the name of the cpu to all public symbol names.
9686 In multiple-processor systems, there are many ARC variants with different
9687 instruction and register set characteristics. This flag prevents code
9688 compiled for one cpu to be linked with code compiled for another.
9689 No facility exists for handling variants that are ``almost identical''.
9690 This is an all or nothing option.
9692 @item -mcpu=@var{cpu}
9694 Compile code for ARC variant @var{cpu}.
9695 Which variants are supported depend on the configuration.
9696 All variants support @option{-mcpu=base}, this is the default.
9698 @item -mtext=@var{text-section}
9699 @itemx -mdata=@var{data-section}
9700 @itemx -mrodata=@var{readonly-data-section}
9704 Put functions, data, and readonly data in @var{text-section},
9705 @var{data-section}, and @var{readonly-data-section} respectively
9706 by default. This can be overridden with the @code{section} attribute.
9707 @xref{Variable Attributes}.
9712 @subsection NS32K Options
9713 @cindex NS32K options
9715 These are the @samp{-m} options defined for the 32000 series. The default
9716 values for these options depends on which style of 32000 was selected when
9717 the compiler was configured; the defaults for the most common choices are
9725 Generate output for a 32032. This is the default
9726 when the compiler is configured for 32032 and 32016 based systems.
9732 Generate output for a 32332. This is the default
9733 when the compiler is configured for 32332-based systems.
9739 Generate output for a 32532. This is the default
9740 when the compiler is configured for 32532-based systems.
9744 Generate output containing 32081 instructions for floating point.
9745 This is the default for all systems.
9749 Generate output containing 32381 instructions for floating point. This
9750 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9751 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9755 Try and generate multiply-add floating point instructions @code{polyF}
9756 and @code{dotF}. This option is only available if the @option{-m32381}
9757 option is in effect. Using these instructions requires changes to
9758 register allocation which generally has a negative impact on
9759 performance. This option should only be enabled when compiling code
9760 particularly likely to make heavy use of multiply-add instructions.
9763 @opindex mnomulti-add
9764 Do not try and generate multiply-add floating point instructions
9765 @code{polyF} and @code{dotF}. This is the default on all platforms.
9768 @opindex msoft-float
9769 Generate output containing library calls for floating point.
9770 @strong{Warning:} the requisite libraries may not be available.
9772 @item -mieee-compare
9773 @itemx -mno-ieee-compare
9774 @opindex mieee-compare
9775 @opindex mno-ieee-compare
9776 Control whether or not the compiler uses IEEE floating point
9777 comparisons. These handle correctly the case where the result of a
9778 comparison is unordered.
9779 @strong{Warning:} the requisite kernel support may not be available.
9782 @opindex mnobitfield
9783 Do not use the bit-field instructions. On some machines it is faster to
9784 use shifting and masking operations. This is the default for the pc532.
9788 Do use the bit-field instructions. This is the default for all platforms
9793 Use a different function-calling convention, in which functions
9794 that take a fixed number of arguments return pop their
9795 arguments on return with the @code{ret} instruction.
9797 This calling convention is incompatible with the one normally
9798 used on Unix, so you cannot use it if you need to call libraries
9799 compiled with the Unix compiler.
9801 Also, you must provide function prototypes for all functions that
9802 take variable numbers of arguments (including @code{printf});
9803 otherwise incorrect code will be generated for calls to those
9806 In addition, seriously incorrect code will result if you call a
9807 function with too many arguments. (Normally, extra arguments are
9808 harmlessly ignored.)
9810 This option takes its name from the 680x0 @code{rtd} instruction.
9815 Use a different function-calling convention where the first two arguments
9816 are passed in registers.
9818 This calling convention is incompatible with the one normally
9819 used on Unix, so you cannot use it if you need to call libraries
9820 compiled with the Unix compiler.
9823 @opindex mnoregparam
9824 Do not pass any arguments in registers. This is the default for all
9829 It is OK to use the sb as an index register which is always loaded with
9830 zero. This is the default for the pc532-netbsd target.
9834 The sb register is not available for use or has not been initialized to
9835 zero by the run time system. This is the default for all targets except
9836 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9837 @option{-fpic} is set.
9841 Many ns32000 series addressing modes use displacements of up to 512MB@.
9842 If an address is above 512MB then displacements from zero can not be used.
9843 This option causes code to be generated which can be loaded above 512MB@.
9844 This may be useful for operating systems or ROM code.
9848 Assume code will be loaded in the first 512MB of virtual address space.
9849 This is the default for all platforms.
9855 @subsection AVR Options
9858 These options are defined for AVR implementations:
9861 @item -mmcu=@var{mcu}
9863 Specify ATMEL AVR instruction set or MCU type.
9865 Instruction set avr1 is for the minimal AVR core, not supported by the C
9866 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9867 attiny11, attiny12, attiny15, attiny28).
9869 Instruction set avr2 (default) is for the classic AVR core with up to
9870 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9871 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9872 at90c8534, at90s8535).
9874 Instruction set avr3 is for the classic AVR core with up to 128K program
9875 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9877 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9878 memory space (MCU types: atmega8, atmega83, atmega85).
9880 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9881 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9882 atmega64, atmega128, at43usb355, at94k).
9886 Output instruction sizes to the asm file.
9888 @item -minit-stack=@var{N}
9889 @opindex minit-stack
9890 Specify the initial stack address, which may be a symbol or numeric value,
9891 @samp{__stack} is the default.
9893 @item -mno-interrupts
9894 @opindex mno-interrupts
9895 Generated code is not compatible with hardware interrupts.
9896 Code size will be smaller.
9898 @item -mcall-prologues
9899 @opindex mcall-prologues
9900 Functions prologues/epilogues expanded as call to appropriate
9901 subroutines. Code size will be smaller.
9903 @item -mno-tablejump
9904 @opindex mno-tablejump
9905 Do not generate tablejump insns which sometimes increase code size.
9908 @opindex mtiny-stack
9909 Change only the low 8 bits of the stack pointer.
9913 @subsection MCore Options
9914 @cindex MCore options
9916 These are the @samp{-m} options defined for the Motorola M*Core
9924 @opindex mno-hardlit
9925 Inline constants into the code stream if it can be done in two
9926 instructions or less.
9932 Use the divide instruction. (Enabled by default).
9934 @item -mrelax-immediate
9935 @itemx -mno-relax-immediate
9936 @opindex mrelax-immediate
9937 @opindex mno-relax-immediate
9938 Allow arbitrary sized immediates in bit operations.
9940 @item -mwide-bitfields
9941 @itemx -mno-wide-bitfields
9942 @opindex mwide-bitfields
9943 @opindex mno-wide-bitfields
9944 Always treat bit-fields as int-sized.
9946 @item -m4byte-functions
9947 @itemx -mno-4byte-functions
9948 @opindex m4byte-functions
9949 @opindex mno-4byte-functions
9950 Force all functions to be aligned to a four byte boundary.
9952 @item -mcallgraph-data
9953 @itemx -mno-callgraph-data
9954 @opindex mcallgraph-data
9955 @opindex mno-callgraph-data
9956 Emit callgraph information.
9959 @itemx -mno-slow-bytes
9960 @opindex mslow-bytes
9961 @opindex mno-slow-bytes
9962 Prefer word access when reading byte quantities.
9964 @item -mlittle-endian
9966 @opindex mlittle-endian
9967 @opindex mbig-endian
9968 Generate code for a little endian target.
9974 Generate code for the 210 processor.
9978 @subsection IA-64 Options
9979 @cindex IA-64 Options
9981 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9985 @opindex mbig-endian
9986 Generate code for a big endian target. This is the default for HP-UX@.
9988 @item -mlittle-endian
9989 @opindex mlittle-endian
9990 Generate code for a little endian target. This is the default for AIX5
9997 Generate (or don't) code for the GNU assembler. This is the default.
9998 @c Also, this is the default if the configure option @option{--with-gnu-as}
10004 @opindex mno-gnu-ld
10005 Generate (or don't) code for the GNU linker. This is the default.
10006 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10011 Generate code that does not use a global pointer register. The result
10012 is not position independent code, and violates the IA-64 ABI@.
10014 @item -mvolatile-asm-stop
10015 @itemx -mno-volatile-asm-stop
10016 @opindex mvolatile-asm-stop
10017 @opindex mno-volatile-asm-stop
10018 Generate (or don't) a stop bit immediately before and after volatile asm
10023 Generate code that works around Itanium B step errata.
10025 @item -mregister-names
10026 @itemx -mno-register-names
10027 @opindex mregister-names
10028 @opindex mno-register-names
10029 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10030 the stacked registers. This may make assembler output more readable.
10036 Disable (or enable) optimizations that use the small data section. This may
10037 be useful for working around optimizer bugs.
10039 @item -mconstant-gp
10040 @opindex mconstant-gp
10041 Generate code that uses a single constant global pointer value. This is
10042 useful when compiling kernel code.
10046 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10047 This is useful when compiling firmware code.
10049 @item -minline-float-divide-min-latency
10050 @opindex minline-float-divide-min-latency
10051 Generate code for inline divides of floating point values
10052 using the minimum latency algorithm.
10054 @item -minline-float-divide-max-throughput
10055 @opindex minline-float-divide-max-throughput
10056 Generate code for inline divides of floating point values
10057 using the maximum throughput algorithm.
10059 @item -minline-int-divide-min-latency
10060 @opindex minline-int-divide-min-latency
10061 Generate code for inline divides of integer values
10062 using the minimum latency algorithm.
10064 @item -minline-int-divide-max-throughput
10065 @opindex minline-int-divide-max-throughput
10066 Generate code for inline divides of integer values
10067 using the maximum throughput algorithm.
10069 @item -mno-dwarf2-asm
10070 @itemx -mdwarf2-asm
10071 @opindex mno-dwarf2-asm
10072 @opindex mdwarf2-asm
10073 Don't (or do) generate assembler code for the DWARF2 line number debugging
10074 info. This may be useful when not using the GNU assembler.
10076 @item -mfixed-range=@var{register-range}
10077 @opindex mfixed-range
10078 Generate code treating the given register range as fixed registers.
10079 A fixed register is one that the register allocator can not use. This is
10080 useful when compiling kernel code. A register range is specified as
10081 two registers separated by a dash. Multiple register ranges can be
10082 specified separated by a comma.
10084 @item -mearly-stop-bits
10085 @itemx -mno-early-stop-bits
10086 @opindex mearly-stop-bits
10087 @opindex mno-early-stop-bits
10088 Allow stop bits to be placed earlier than immediately preceding the
10089 instruction that triggered the stop bit. This can improve instruction
10090 scheduling, but does not always do so.
10093 @node S/390 and zSeries Options
10094 @subsection S/390 and zSeries Options
10095 @cindex S/390 and zSeries Options
10097 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10101 @itemx -msoft-float
10102 @opindex mhard-float
10103 @opindex msoft-float
10104 Use (do not use) the hardware floating-point instructions and registers
10105 for floating-point operations. When @option{-msoft-float} is specified,
10106 functions in @file{libgcc.a} will be used to perform floating-point
10107 operations. When @option{-mhard-float} is specified, the compiler
10108 generates IEEE floating-point instructions. This is the default.
10111 @itemx -mno-backchain
10112 @opindex mbackchain
10113 @opindex mno-backchain
10114 Generate (or do not generate) code which maintains an explicit
10115 backchain within the stack frame that points to the caller's frame.
10116 This may be needed to allow debugging using tools that do not understand
10117 DWARF-2 call frame information. The default is not to generate the
10121 @itemx -mno-small-exec
10122 @opindex msmall-exec
10123 @opindex mno-small-exec
10124 Generate (or do not generate) code using the @code{bras} instruction
10125 to do subroutine calls.
10126 This only works reliably if the total executable size does not
10127 exceed 64k. The default is to use the @code{basr} instruction instead,
10128 which does not have this limitation.
10134 When @option{-m31} is specified, generate code compliant to the
10135 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
10136 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
10137 particular to generate 64-bit instructions. For the @samp{s390}
10138 targets, the default is @option{-m31}, while the @samp{s390x}
10139 targets default to @option{-m64}.
10145 When @option{-mzarch} is specified, generate code using the
10146 instructions available on z/Architecture.
10147 When @option{-mesa} is specified, generate code using the
10148 instructions available on ESA/390. Note that @option{-mesa} is
10149 not possible with @option{-m64}.
10150 When generating code compliant to the GNU/Linux for S/390 ABI,
10151 the default is @option{-mesa}. When generating code compliant
10152 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
10158 Generate (or do not generate) code using the @code{mvcle} instruction
10159 to perform block moves. When @option{-mno-mvcle} is specified,
10160 use a @code{mvc} loop instead. This is the default.
10166 Print (or do not print) additional debug information when compiling.
10167 The default is to not print debug information.
10169 @item -march=@var{cpu-type}
10171 Generate code that will run on @var{cpu-type}, which is the name of a system
10172 representing a certain processor type. Possible values for
10173 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10174 When generating code using the instructions available on z/Architecture,
10175 the default is @option{-march=z900}. Otherwise, the default is
10176 @option{-march=g5}.
10178 @item -mtune=@var{cpu-type}
10180 Tune to @var{cpu-type} everything applicable about the generated code,
10181 except for the ABI and the set of available instructions.
10182 The list of @var{cpu-type} values is the same as for @option{-march}.
10183 The default is the value used for @option{-march}.
10186 @itemx -mno-fused-madd
10187 @opindex mfused-madd
10188 @opindex mno-fused-madd
10189 Generate code that uses (does not use) the floating point multiply and
10190 accumulate instructions. These instructions are generated by default if
10191 hardware floating point is used.
10195 @subsection CRIS Options
10196 @cindex CRIS Options
10198 These options are defined specifically for the CRIS ports.
10201 @item -march=@var{architecture-type}
10202 @itemx -mcpu=@var{architecture-type}
10205 Generate code for the specified architecture. The choices for
10206 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10207 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
10208 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10211 @item -mtune=@var{architecture-type}
10213 Tune to @var{architecture-type} everything applicable about the generated
10214 code, except for the ABI and the set of available instructions. The
10215 choices for @var{architecture-type} are the same as for
10216 @option{-march=@var{architecture-type}}.
10218 @item -mmax-stack-frame=@var{n}
10219 @opindex mmax-stack-frame
10220 Warn when the stack frame of a function exceeds @var{n} bytes.
10222 @item -melinux-stacksize=@var{n}
10223 @opindex melinux-stacksize
10224 Only available with the @samp{cris-axis-aout} target. Arranges for
10225 indications in the program to the kernel loader that the stack of the
10226 program should be set to @var{n} bytes.
10232 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10233 @option{-march=v3} and @option{-march=v8} respectively.
10235 @item -mmul-bug-workaround
10236 @itemx -mno-mul-bug-workaround
10237 @opindex mmul-bug-workaround
10238 @opindex mno-mul-bug-workaround
10239 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
10240 models where it applies. This option is active by default.
10244 Enable CRIS-specific verbose debug-related information in the assembly
10245 code. This option also has the effect to turn off the @samp{#NO_APP}
10246 formatted-code indicator to the assembler at the beginning of the
10251 Do not use condition-code results from previous instruction; always emit
10252 compare and test instructions before use of condition codes.
10254 @item -mno-side-effects
10255 @opindex mno-side-effects
10256 Do not emit instructions with side-effects in addressing modes other than
10259 @item -mstack-align
10260 @itemx -mno-stack-align
10261 @itemx -mdata-align
10262 @itemx -mno-data-align
10263 @itemx -mconst-align
10264 @itemx -mno-const-align
10265 @opindex mstack-align
10266 @opindex mno-stack-align
10267 @opindex mdata-align
10268 @opindex mno-data-align
10269 @opindex mconst-align
10270 @opindex mno-const-align
10271 These options (no-options) arranges (eliminate arrangements) for the
10272 stack-frame, individual data and constants to be aligned for the maximum
10273 single data access size for the chosen CPU model. The default is to
10274 arrange for 32-bit alignment. ABI details such as structure layout are
10275 not affected by these options.
10283 Similar to the stack- data- and const-align options above, these options
10284 arrange for stack-frame, writable data and constants to all be 32-bit,
10285 16-bit or 8-bit aligned. The default is 32-bit alignment.
10287 @item -mno-prologue-epilogue
10288 @itemx -mprologue-epilogue
10289 @opindex mno-prologue-epilogue
10290 @opindex mprologue-epilogue
10291 With @option{-mno-prologue-epilogue}, the normal function prologue and
10292 epilogue that sets up the stack-frame are omitted and no return
10293 instructions or return sequences are generated in the code. Use this
10294 option only together with visual inspection of the compiled code: no
10295 warnings or errors are generated when call-saved registers must be saved,
10296 or storage for local variable needs to be allocated.
10300 @opindex mno-gotplt
10302 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10303 instruction sequences that load addresses for functions from the PLT part
10304 of the GOT rather than (traditional on other architectures) calls to the
10305 PLT. The default is @option{-mgotplt}.
10309 Legacy no-op option only recognized with the cris-axis-aout target.
10313 Legacy no-op option only recognized with the cris-axis-elf and
10314 cris-axis-linux-gnu targets.
10318 Only recognized with the cris-axis-aout target, where it selects a
10319 GNU/linux-like multilib, include files and instruction set for
10320 @option{-march=v8}.
10324 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10328 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
10329 to link with input-output functions from a simulator library. Code,
10330 initialized data and zero-initialized data are allocated consecutively.
10334 Like @option{-sim}, but pass linker options to locate initialized data at
10335 0x40000000 and zero-initialized data at 0x80000000.
10339 @subsection MMIX Options
10340 @cindex MMIX Options
10342 These options are defined for the MMIX:
10346 @itemx -mno-libfuncs
10348 @opindex mno-libfuncs
10349 Specify that intrinsic library functions are being compiled, passing all
10350 values in registers, no matter the size.
10353 @itemx -mno-epsilon
10355 @opindex mno-epsilon
10356 Generate floating-point comparison instructions that compare with respect
10357 to the @code{rE} epsilon register.
10359 @item -mabi=mmixware
10361 @opindex mabi-mmixware
10363 Generate code that passes function parameters and return values that (in
10364 the called function) are seen as registers @code{$0} and up, as opposed to
10365 the GNU ABI which uses global registers @code{$231} and up.
10367 @item -mzero-extend
10368 @itemx -mno-zero-extend
10369 @opindex mzero-extend
10370 @opindex mno-zero-extend
10371 When reading data from memory in sizes shorter than 64 bits, use (do not
10372 use) zero-extending load instructions by default, rather than
10373 sign-extending ones.
10376 @itemx -mno-knuthdiv
10378 @opindex mno-knuthdiv
10379 Make the result of a division yielding a remainder have the same sign as
10380 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10381 remainder follows the sign of the dividend. Both methods are
10382 arithmetically valid, the latter being almost exclusively used.
10384 @item -mtoplevel-symbols
10385 @itemx -mno-toplevel-symbols
10386 @opindex mtoplevel-symbols
10387 @opindex mno-toplevel-symbols
10388 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10389 code can be used with the @code{PREFIX} assembly directive.
10393 Generate an executable in the ELF format, rather than the default
10394 @samp{mmo} format used by the @command{mmix} simulator.
10396 @item -mbranch-predict
10397 @itemx -mno-branch-predict
10398 @opindex mbranch-predict
10399 @opindex mno-branch-predict
10400 Use (do not use) the probable-branch instructions, when static branch
10401 prediction indicates a probable branch.
10403 @item -mbase-addresses
10404 @itemx -mno-base-addresses
10405 @opindex mbase-addresses
10406 @opindex mno-base-addresses
10407 Generate (do not generate) code that uses @emph{base addresses}. Using a
10408 base address automatically generates a request (handled by the assembler
10409 and the linker) for a constant to be set up in a global register. The
10410 register is used for one or more base address requests within the range 0
10411 to 255 from the value held in the register. The generally leads to short
10412 and fast code, but the number of different data items that can be
10413 addressed is limited. This means that a program that uses lots of static
10414 data may require @option{-mno-base-addresses}.
10416 @item -msingle-exit
10417 @itemx -mno-single-exit
10418 @opindex msingle-exit
10419 @opindex mno-single-exit
10420 Force (do not force) generated code to have a single exit point in each
10424 @node PDP-11 Options
10425 @subsection PDP-11 Options
10426 @cindex PDP-11 Options
10428 These options are defined for the PDP-11:
10433 Use hardware FPP floating point. This is the default. (FIS floating
10434 point on the PDP-11/40 is not supported.)
10437 @opindex msoft-float
10438 Do not use hardware floating point.
10442 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10446 Return floating-point results in memory. This is the default.
10450 Generate code for a PDP-11/40.
10454 Generate code for a PDP-11/45. This is the default.
10458 Generate code for a PDP-11/10.
10460 @item -mbcopy-builtin
10461 @opindex bcopy-builtin
10462 Use inline @code{movstrhi} patterns for copying memory. This is the
10467 Do not use inline @code{movstrhi} patterns for copying memory.
10473 Use 16-bit @code{int}. This is the default.
10479 Use 32-bit @code{int}.
10482 @itemx -mno-float32
10484 @opindex mno-float32
10485 Use 64-bit @code{float}. This is the default.
10488 @itemx -mno-float64
10490 @opindex mno-float64
10491 Use 32-bit @code{float}.
10495 Use @code{abshi2} pattern. This is the default.
10499 Do not use @code{abshi2} pattern.
10501 @item -mbranch-expensive
10502 @opindex mbranch-expensive
10503 Pretend that branches are expensive. This is for experimenting with
10504 code generation only.
10506 @item -mbranch-cheap
10507 @opindex mbranch-cheap
10508 Do not pretend that branches are expensive. This is the default.
10512 Generate code for a system with split I&D.
10516 Generate code for a system without split I&D. This is the default.
10520 Use Unix assembler syntax. This is the default when configured for
10521 @samp{pdp11-*-bsd}.
10525 Use DEC assembler syntax. This is the default when configured for any
10526 PDP-11 target other than @samp{pdp11-*-bsd}.
10529 @node Xstormy16 Options
10530 @subsection Xstormy16 Options
10531 @cindex Xstormy16 Options
10533 These options are defined for Xstormy16:
10538 Choose startup files and linker script suitable for the simulator.
10542 @subsection FRV Options
10543 @cindex FRV Options
10549 Only use the first 32 general purpose registers.
10554 Use all 64 general purpose registers.
10559 Use only the first 32 floating point registers.
10564 Use all 64 floating point registers
10567 @opindex mhard-float
10569 Use hardware instructions for floating point operations.
10572 @opindex msoft-float
10574 Use library routines for floating point operations.
10579 Dynamically allocate condition code registers.
10584 Do not try to dynamically allocate condition code registers, only
10585 use @code{icc0} and @code{fcc0}.
10590 Change ABI to use double word insns.
10595 Do not use double word instructions.
10600 Use floating point double instructions.
10603 @opindex mno-double
10605 Do not use floating point double instructions.
10610 Use media instructions.
10615 Do not use media instructions.
10620 Use multiply and add/subtract instructions.
10623 @opindex mno-muladd
10625 Do not use multiply and add/subtract instructions.
10627 @item -mlibrary-pic
10628 @opindex mlibrary-pic
10630 Generate position-independent EABI code.
10635 Use only the first four media accumulator registers.
10640 Use all eight media accumulator registers.
10645 Pack VLIW instructions.
10650 Do not pack VLIW instructions.
10653 @opindex mno-eflags
10655 Do not mark ABI switches in e_flags.
10658 @opindex mcond-move
10660 Enable the use of conditional-move instructions (default).
10662 This switch is mainly for debugging the compiler and will likely be removed
10663 in a future version.
10665 @item -mno-cond-move
10666 @opindex mno-cond-move
10668 Disable the use of conditional-move instructions.
10670 This switch is mainly for debugging the compiler and will likely be removed
10671 in a future version.
10676 Enable the use of conditional set instructions (default).
10678 This switch is mainly for debugging the compiler and will likely be removed
10679 in a future version.
10684 Disable the use of conditional set instructions.
10686 This switch is mainly for debugging the compiler and will likely be removed
10687 in a future version.
10690 @opindex mcond-exec
10692 Enable the use of conditional execution (default).
10694 This switch is mainly for debugging the compiler and will likely be removed
10695 in a future version.
10697 @item -mno-cond-exec
10698 @opindex mno-cond-exec
10700 Disable the use of conditional execution.
10702 This switch is mainly for debugging the compiler and will likely be removed
10703 in a future version.
10705 @item -mvliw-branch
10706 @opindex mvliw-branch
10708 Run a pass to pack branches into VLIW instructions (default).
10710 This switch is mainly for debugging the compiler and will likely be removed
10711 in a future version.
10713 @item -mno-vliw-branch
10714 @opindex mno-vliw-branch
10716 Do not run a pass to pack branches into VLIW instructions.
10718 This switch is mainly for debugging the compiler and will likely be removed
10719 in a future version.
10721 @item -mmulti-cond-exec
10722 @opindex mmulti-cond-exec
10724 Enable optimization of @code{&&} and @code{||} in conditional execution
10727 This switch is mainly for debugging the compiler and will likely be removed
10728 in a future version.
10730 @item -mno-multi-cond-exec
10731 @opindex mno-multi-cond-exec
10733 Disable optimization of @code{&&} and @code{||} in conditional execution.
10735 This switch is mainly for debugging the compiler and will likely be removed
10736 in a future version.
10738 @item -mnested-cond-exec
10739 @opindex mnested-cond-exec
10741 Enable nested conditional execution optimizations (default).
10743 This switch is mainly for debugging the compiler and will likely be removed
10744 in a future version.
10746 @item -mno-nested-cond-exec
10747 @opindex mno-nested-cond-exec
10749 Disable nested conditional execution optimizations.
10751 This switch is mainly for debugging the compiler and will likely be removed
10752 in a future version.
10754 @item -mtomcat-stats
10755 @opindex mtomcat-stats
10757 Cause gas to print out tomcat statistics.
10759 @item -mcpu=@var{cpu}
10762 Select the processor type for which to generate code. Possible values are
10763 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
10768 @node Xtensa Options
10769 @subsection Xtensa Options
10770 @cindex Xtensa Options
10772 These options are supported for Xtensa targets:
10776 @itemx -mno-const16
10778 @opindex mno-const16
10779 Enable or disable use of @code{CONST16} instructions for loading
10780 constant values. The @code{CONST16} instruction is currently not a
10781 standard option from Tensilica. When enabled, @code{CONST16}
10782 instructions are always used in place of the standard @code{L32R}
10783 instructions. The use of @code{CONST16} is enabled by default only if
10784 the @code{L32R} instruction is not available.
10787 @itemx -mno-fused-madd
10788 @opindex mfused-madd
10789 @opindex mno-fused-madd
10790 Enable or disable use of fused multiply/add and multiply/subtract
10791 instructions in the floating-point option. This has no effect if the
10792 floating-point option is not also enabled. Disabling fused multiply/add
10793 and multiply/subtract instructions forces the compiler to use separate
10794 instructions for the multiply and add/subtract operations. This may be
10795 desirable in some cases where strict IEEE 754-compliant results are
10796 required: the fused multiply add/subtract instructions do not round the
10797 intermediate result, thereby producing results with @emph{more} bits of
10798 precision than specified by the IEEE standard. Disabling fused multiply
10799 add/subtract instructions also ensures that the program output is not
10800 sensitive to the compiler's ability to combine multiply and add/subtract
10803 @item -mtext-section-literals
10804 @itemx -mno-text-section-literals
10805 @opindex mtext-section-literals
10806 @opindex mno-text-section-literals
10807 Control the treatment of literal pools. The default is
10808 @option{-mno-text-section-literals}, which places literals in a separate
10809 section in the output file. This allows the literal pool to be placed
10810 in a data RAM/ROM, and it also allows the linker to combine literal
10811 pools from separate object files to remove redundant literals and
10812 improve code size. With @option{-mtext-section-literals}, the literals
10813 are interspersed in the text section in order to keep them as close as
10814 possible to their references. This may be necessary for large assembly
10817 @item -mtarget-align
10818 @itemx -mno-target-align
10819 @opindex mtarget-align
10820 @opindex mno-target-align
10821 When this option is enabled, GCC instructs the assembler to
10822 automatically align instructions to reduce branch penalties at the
10823 expense of some code density. The assembler attempts to widen density
10824 instructions to align branch targets and the instructions following call
10825 instructions. If there are not enough preceding safe density
10826 instructions to align a target, no widening will be performed. The
10827 default is @option{-mtarget-align}. These options do not affect the
10828 treatment of auto-aligned instructions like @code{LOOP}, which the
10829 assembler will always align, either by widening density instructions or
10830 by inserting no-op instructions.
10833 @itemx -mno-longcalls
10834 @opindex mlongcalls
10835 @opindex mno-longcalls
10836 When this option is enabled, GCC instructs the assembler to translate
10837 direct calls to indirect calls unless it can determine that the target
10838 of a direct call is in the range allowed by the call instruction. This
10839 translation typically occurs for calls to functions in other source
10840 files. Specifically, the assembler translates a direct @code{CALL}
10841 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
10842 The default is @option{-mno-longcalls}. This option should be used in
10843 programs where the call target can potentially be out of range. This
10844 option is implemented in the assembler, not the compiler, so the
10845 assembly code generated by GCC will still show direct call
10846 instructions---look at the disassembled object code to see the actual
10847 instructions. Note that the assembler will use an indirect call for
10848 every cross-file call, not just those that really will be out of range.
10851 @node Code Gen Options
10852 @section Options for Code Generation Conventions
10853 @cindex code generation conventions
10854 @cindex options, code generation
10855 @cindex run-time options
10857 These machine-independent options control the interface conventions
10858 used in code generation.
10860 Most of them have both positive and negative forms; the negative form
10861 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
10862 one of the forms is listed---the one which is not the default. You
10863 can figure out the other form by either removing @samp{no-} or adding
10867 @item -fbounds-check
10868 @opindex fbounds-check
10869 For front-ends that support it, generate additional code to check that
10870 indices used to access arrays are within the declared range. This is
10871 currently only supported by the Java and Fortran 77 front-ends, where
10872 this option defaults to true and false respectively.
10876 This option generates traps for signed overflow on addition, subtraction,
10877 multiplication operations.
10881 This option instructs the compiler to assume that signed arithmetic
10882 overflow of addition, subtraction and multiplication wraps around
10883 using twos-complement representation. This flag enables some optimizations
10884 and disables other. This option is enabled by default for the Java
10885 front-end, as required by the Java language specification.
10888 @opindex fexceptions
10889 Enable exception handling. Generates extra code needed to propagate
10890 exceptions. For some targets, this implies GCC will generate frame
10891 unwind information for all functions, which can produce significant data
10892 size overhead, although it does not affect execution. If you do not
10893 specify this option, GCC will enable it by default for languages like
10894 C++ which normally require exception handling, and disable it for
10895 languages like C that do not normally require it. However, you may need
10896 to enable this option when compiling C code that needs to interoperate
10897 properly with exception handlers written in C++. You may also wish to
10898 disable this option if you are compiling older C++ programs that don't
10899 use exception handling.
10901 @item -fnon-call-exceptions
10902 @opindex fnon-call-exceptions
10903 Generate code that allows trapping instructions to throw exceptions.
10904 Note that this requires platform-specific runtime support that does
10905 not exist everywhere. Moreover, it only allows @emph{trapping}
10906 instructions to throw exceptions, i.e.@: memory references or floating
10907 point instructions. It does not allow exceptions to be thrown from
10908 arbitrary signal handlers such as @code{SIGALRM}.
10910 @item -funwind-tables
10911 @opindex funwind-tables
10912 Similar to @option{-fexceptions}, except that it will just generate any needed
10913 static data, but will not affect the generated code in any other way.
10914 You will normally not enable this option; instead, a language processor
10915 that needs this handling would enable it on your behalf.
10917 @item -fasynchronous-unwind-tables
10918 @opindex funwind-tables
10919 Generate unwind table in dwarf2 format, if supported by target machine. The
10920 table is exact at each instruction boundary, so it can be used for stack
10921 unwinding from asynchronous events (such as debugger or garbage collector).
10923 @item -fpcc-struct-return
10924 @opindex fpcc-struct-return
10925 Return ``short'' @code{struct} and @code{union} values in memory like
10926 longer ones, rather than in registers. This convention is less
10927 efficient, but it has the advantage of allowing intercallability between
10928 GCC-compiled files and files compiled with other compilers, particularly
10929 the Portable C Compiler (pcc).
10931 The precise convention for returning structures in memory depends
10932 on the target configuration macros.
10934 Short structures and unions are those whose size and alignment match
10935 that of some integer type.
10937 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
10938 switch is not binary compatible with code compiled with the
10939 @option{-freg-struct-return} switch.
10940 Use it to conform to a non-default application binary interface.
10942 @item -freg-struct-return
10943 @opindex freg-struct-return
10944 Return @code{struct} and @code{union} values in registers when possible.
10945 This is more efficient for small structures than
10946 @option{-fpcc-struct-return}.
10948 If you specify neither @option{-fpcc-struct-return} nor
10949 @option{-freg-struct-return}, GCC defaults to whichever convention is
10950 standard for the target. If there is no standard convention, GCC
10951 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
10952 the principal compiler. In those cases, we can choose the standard, and
10953 we chose the more efficient register return alternative.
10955 @strong{Warning:} code compiled with the @option{-freg-struct-return}
10956 switch is not binary compatible with code compiled with the
10957 @option{-fpcc-struct-return} switch.
10958 Use it to conform to a non-default application binary interface.
10960 @item -fshort-enums
10961 @opindex fshort-enums
10962 Allocate to an @code{enum} type only as many bytes as it needs for the
10963 declared range of possible values. Specifically, the @code{enum} type
10964 will be equivalent to the smallest integer type which has enough room.
10966 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
10967 code that is not binary compatible with code generated without that switch.
10968 Use it to conform to a non-default application binary interface.
10970 @item -fshort-double
10971 @opindex fshort-double
10972 Use the same size for @code{double} as for @code{float}.
10974 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
10975 code that is not binary compatible with code generated without that switch.
10976 Use it to conform to a non-default application binary interface.
10978 @item -fshort-wchar
10979 @opindex fshort-wchar
10980 Override the underlying type for @samp{wchar_t} to be @samp{short
10981 unsigned int} instead of the default for the target. This option is
10982 useful for building programs to run under WINE@.
10984 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
10985 code that is not binary compatible with code generated without that switch.
10986 Use it to conform to a non-default application binary interface.
10988 @item -fshared-data
10989 @opindex fshared-data
10990 Requests that the data and non-@code{const} variables of this
10991 compilation be shared data rather than private data. The distinction
10992 makes sense only on certain operating systems, where shared data is
10993 shared between processes running the same program, while private data
10994 exists in one copy per process.
10997 @opindex fno-common
10998 In C, allocate even uninitialized global variables in the data section of the
10999 object file, rather than generating them as common blocks. This has the
11000 effect that if the same variable is declared (without @code{extern}) in
11001 two different compilations, you will get an error when you link them.
11002 The only reason this might be useful is if you wish to verify that the
11003 program will work on other systems which always work this way.
11007 Ignore the @samp{#ident} directive.
11009 @item -finhibit-size-directive
11010 @opindex finhibit-size-directive
11011 Don't output a @code{.size} assembler directive, or anything else that
11012 would cause trouble if the function is split in the middle, and the
11013 two halves are placed at locations far apart in memory. This option is
11014 used when compiling @file{crtstuff.c}; you should not need to use it
11017 @item -fverbose-asm
11018 @opindex fverbose-asm
11019 Put extra commentary information in the generated assembly code to
11020 make it more readable. This option is generally only of use to those
11021 who actually need to read the generated assembly code (perhaps while
11022 debugging the compiler itself).
11024 @option{-fno-verbose-asm}, the default, causes the
11025 extra information to be omitted and is useful when comparing two assembler
11030 @cindex global offset table
11032 Generate position-independent code (PIC) suitable for use in a shared
11033 library, if supported for the target machine. Such code accesses all
11034 constant addresses through a global offset table (GOT)@. The dynamic
11035 loader resolves the GOT entries when the program starts (the dynamic
11036 loader is not part of GCC; it is part of the operating system). If
11037 the GOT size for the linked executable exceeds a machine-specific
11038 maximum size, you get an error message from the linker indicating that
11039 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11040 instead. (These maximums are 8k on the SPARC and 32k
11041 on the m68k and RS/6000. The 386 has no such limit.)
11043 Position-independent code requires special support, and therefore works
11044 only on certain machines. For the 386, GCC supports PIC for System V
11045 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11046 position-independent.
11050 If supported for the target machine, emit position-independent code,
11051 suitable for dynamic linking and avoiding any limit on the size of the
11052 global offset table. This option makes a difference on the m68k
11055 Position-independent code requires special support, and therefore works
11056 only on certain machines.
11062 These options are similar to @option{-fpic} and @option{-fPIC}, but
11063 generated position independent code can be only linked into executables.
11064 Usually these options are used when @option{-pie} GCC option will be
11065 used during linking.
11067 @item -ffixed-@var{reg}
11069 Treat the register named @var{reg} as a fixed register; generated code
11070 should never refer to it (except perhaps as a stack pointer, frame
11071 pointer or in some other fixed role).
11073 @var{reg} must be the name of a register. The register names accepted
11074 are machine-specific and are defined in the @code{REGISTER_NAMES}
11075 macro in the machine description macro file.
11077 This flag does not have a negative form, because it specifies a
11080 @item -fcall-used-@var{reg}
11081 @opindex fcall-used
11082 Treat the register named @var{reg} as an allocable register that is
11083 clobbered by function calls. It may be allocated for temporaries or
11084 variables that do not live across a call. Functions compiled this way
11085 will not save and restore the register @var{reg}.
11087 It is an error to used this flag with the frame pointer or stack pointer.
11088 Use of this flag for other registers that have fixed pervasive roles in
11089 the machine's execution model will produce disastrous results.
11091 This flag does not have a negative form, because it specifies a
11094 @item -fcall-saved-@var{reg}
11095 @opindex fcall-saved
11096 Treat the register named @var{reg} as an allocable register saved by
11097 functions. It may be allocated even for temporaries or variables that
11098 live across a call. Functions compiled this way will save and restore
11099 the register @var{reg} if they use it.
11101 It is an error to used this flag with the frame pointer or stack pointer.
11102 Use of this flag for other registers that have fixed pervasive roles in
11103 the machine's execution model will produce disastrous results.
11105 A different sort of disaster will result from the use of this flag for
11106 a register in which function values may be returned.
11108 This flag does not have a negative form, because it specifies a
11111 @item -fpack-struct
11112 @opindex fpack-struct
11113 Pack all structure members together without holes.
11115 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11116 code that is not binary compatible with code generated without that switch.
11117 Additionally, it makes the code suboptimal.
11118 Use it to conform to a non-default application binary interface.
11120 @item -finstrument-functions
11121 @opindex finstrument-functions
11122 Generate instrumentation calls for entry and exit to functions. Just
11123 after function entry and just before function exit, the following
11124 profiling functions will be called with the address of the current
11125 function and its call site. (On some platforms,
11126 @code{__builtin_return_address} does not work beyond the current
11127 function, so the call site information may not be available to the
11128 profiling functions otherwise.)
11131 void __cyg_profile_func_enter (void *this_fn,
11133 void __cyg_profile_func_exit (void *this_fn,
11137 The first argument is the address of the start of the current function,
11138 which may be looked up exactly in the symbol table.
11140 This instrumentation is also done for functions expanded inline in other
11141 functions. The profiling calls will indicate where, conceptually, the
11142 inline function is entered and exited. This means that addressable
11143 versions of such functions must be available. If all your uses of a
11144 function are expanded inline, this may mean an additional expansion of
11145 code size. If you use @samp{extern inline} in your C code, an
11146 addressable version of such functions must be provided. (This is
11147 normally the case anyways, but if you get lucky and the optimizer always
11148 expands the functions inline, you might have gotten away without
11149 providing static copies.)
11151 A function may be given the attribute @code{no_instrument_function}, in
11152 which case this instrumentation will not be done. This can be used, for
11153 example, for the profiling functions listed above, high-priority
11154 interrupt routines, and any functions from which the profiling functions
11155 cannot safely be called (perhaps signal handlers, if the profiling
11156 routines generate output or allocate memory).
11158 @item -fstack-check
11159 @opindex fstack-check
11160 Generate code to verify that you do not go beyond the boundary of the
11161 stack. You should specify this flag if you are running in an
11162 environment with multiple threads, but only rarely need to specify it in
11163 a single-threaded environment since stack overflow is automatically
11164 detected on nearly all systems if there is only one stack.
11166 Note that this switch does not actually cause checking to be done; the
11167 operating system must do that. The switch causes generation of code
11168 to ensure that the operating system sees the stack being extended.
11170 @item -fstack-limit-register=@var{reg}
11171 @itemx -fstack-limit-symbol=@var{sym}
11172 @itemx -fno-stack-limit
11173 @opindex fstack-limit-register
11174 @opindex fstack-limit-symbol
11175 @opindex fno-stack-limit
11176 Generate code to ensure that the stack does not grow beyond a certain value,
11177 either the value of a register or the address of a symbol. If the stack
11178 would grow beyond the value, a signal is raised. For most targets,
11179 the signal is raised before the stack overruns the boundary, so
11180 it is possible to catch the signal without taking special precautions.
11182 For instance, if the stack starts at absolute address @samp{0x80000000}
11183 and grows downwards, you can use the flags
11184 @option{-fstack-limit-symbol=__stack_limit} and
11185 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11186 of 128KB@. Note that this may only work with the GNU linker.
11188 @cindex aliasing of parameters
11189 @cindex parameters, aliased
11190 @item -fargument-alias
11191 @itemx -fargument-noalias
11192 @itemx -fargument-noalias-global
11193 @opindex fargument-alias
11194 @opindex fargument-noalias
11195 @opindex fargument-noalias-global
11196 Specify the possible relationships among parameters and between
11197 parameters and global data.
11199 @option{-fargument-alias} specifies that arguments (parameters) may
11200 alias each other and may alias global storage.@*
11201 @option{-fargument-noalias} specifies that arguments do not alias
11202 each other, but may alias global storage.@*
11203 @option{-fargument-noalias-global} specifies that arguments do not
11204 alias each other and do not alias global storage.
11206 Each language will automatically use whatever option is required by
11207 the language standard. You should not need to use these options yourself.
11209 @item -fleading-underscore
11210 @opindex fleading-underscore
11211 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11212 change the way C symbols are represented in the object file. One use
11213 is to help link with legacy assembly code.
11215 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11216 generate code that is not binary compatible with code generated without that
11217 switch. Use it to conform to a non-default application binary interface.
11218 Not all targets provide complete support for this switch.
11220 @item -ftls-model=@var{model}
11221 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11222 The @var{model} argument should be one of @code{global-dynamic},
11223 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11225 The default without @option{-fpic} is @code{initial-exec}; with
11226 @option{-fpic} the default is @code{global-dynamic}.
11231 @node Environment Variables
11232 @section Environment Variables Affecting GCC
11233 @cindex environment variables
11235 @c man begin ENVIRONMENT
11236 This section describes several environment variables that affect how GCC
11237 operates. Some of them work by specifying directories or prefixes to use
11238 when searching for various kinds of files. Some are used to specify other
11239 aspects of the compilation environment.
11241 Note that you can also specify places to search using options such as
11242 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11243 take precedence over places specified using environment variables, which
11244 in turn take precedence over those specified by the configuration of GCC@.
11245 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11246 GNU Compiler Collection (GCC) Internals}.
11251 @c @itemx LC_COLLATE
11253 @c @itemx LC_MONETARY
11254 @c @itemx LC_NUMERIC
11259 @c @findex LC_COLLATE
11260 @findex LC_MESSAGES
11261 @c @findex LC_MONETARY
11262 @c @findex LC_NUMERIC
11266 These environment variables control the way that GCC uses
11267 localization information that allow GCC to work with different
11268 national conventions. GCC inspects the locale categories
11269 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11270 so. These locale categories can be set to any value supported by your
11271 installation. A typical value is @samp{en_UK} for English in the United
11274 The @env{LC_CTYPE} environment variable specifies character
11275 classification. GCC uses it to determine the character boundaries in
11276 a string; this is needed for some multibyte encodings that contain quote
11277 and escape characters that would otherwise be interpreted as a string
11280 The @env{LC_MESSAGES} environment variable specifies the language to
11281 use in diagnostic messages.
11283 If the @env{LC_ALL} environment variable is set, it overrides the value
11284 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11285 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11286 environment variable. If none of these variables are set, GCC
11287 defaults to traditional C English behavior.
11291 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11292 files. GCC uses temporary files to hold the output of one stage of
11293 compilation which is to be used as input to the next stage: for example,
11294 the output of the preprocessor, which is the input to the compiler
11297 @item GCC_EXEC_PREFIX
11298 @findex GCC_EXEC_PREFIX
11299 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11300 names of the subprograms executed by the compiler. No slash is added
11301 when this prefix is combined with the name of a subprogram, but you can
11302 specify a prefix that ends with a slash if you wish.
11304 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11305 an appropriate prefix to use based on the pathname it was invoked with.
11307 If GCC cannot find the subprogram using the specified prefix, it
11308 tries looking in the usual places for the subprogram.
11310 The default value of @env{GCC_EXEC_PREFIX} is
11311 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
11312 of @code{prefix} when you ran the @file{configure} script.
11314 Other prefixes specified with @option{-B} take precedence over this prefix.
11316 This prefix is also used for finding files such as @file{crt0.o} that are
11319 In addition, the prefix is used in an unusual way in finding the
11320 directories to search for header files. For each of the standard
11321 directories whose name normally begins with @samp{/usr/local/lib/gcc}
11322 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11323 replacing that beginning with the specified prefix to produce an
11324 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11325 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11326 These alternate directories are searched first; the standard directories
11329 @item COMPILER_PATH
11330 @findex COMPILER_PATH
11331 The value of @env{COMPILER_PATH} is a colon-separated list of
11332 directories, much like @env{PATH}. GCC tries the directories thus
11333 specified when searching for subprograms, if it can't find the
11334 subprograms using @env{GCC_EXEC_PREFIX}.
11337 @findex LIBRARY_PATH
11338 The value of @env{LIBRARY_PATH} is a colon-separated list of
11339 directories, much like @env{PATH}. When configured as a native compiler,
11340 GCC tries the directories thus specified when searching for special
11341 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11342 using GCC also uses these directories when searching for ordinary
11343 libraries for the @option{-l} option (but directories specified with
11344 @option{-L} come first).
11348 @cindex locale definition
11349 This variable is used to pass locale information to the compiler. One way in
11350 which this information is used is to determine the character set to be used
11351 when character literals, string literals and comments are parsed in C and C++.
11352 When the compiler is configured to allow multibyte characters,
11353 the following values for @env{LANG} are recognized:
11357 Recognize JIS characters.
11359 Recognize SJIS characters.
11361 Recognize EUCJP characters.
11364 If @env{LANG} is not defined, or if it has some other value, then the
11365 compiler will use mblen and mbtowc as defined by the default locale to
11366 recognize and translate multibyte characters.
11370 Some additional environments variables affect the behavior of the
11373 @include cppenv.texi
11377 @node Precompiled Headers
11378 @section Using Precompiled Headers
11379 @cindex precompiled headers
11380 @cindex speed of compilation
11382 Often large projects have many header files that are included in every
11383 source file. The time the compiler takes to process these header files
11384 over and over again can account for nearly all of the time required to
11385 build the project. To make builds faster, GCC allows users to
11386 `precompile' a header file; then, if builds can use the precompiled
11387 header file they will be much faster.
11389 @strong{Caution:} There are a few known situations where GCC will
11390 crash when trying to use a precompiled header. If you have trouble
11391 with a precompiled header, you should remove the precompiled header
11392 and compile without it. In addition, please use GCC's on-line
11393 defect-tracking system to report any problems you encounter with
11394 precompiled headers. @xref{Bugs}.
11396 To create a precompiled header file, simply compile it as you would any
11397 other file, if necessary using the @option{-x} option to make the driver
11398 treat it as a C or C++ header file. You will probably want to use a
11399 tool like @command{make} to keep the precompiled header up-to-date when
11400 the headers it contains change.
11402 A precompiled header file will be searched for when @code{#include} is
11403 seen in the compilation. As it searches for the included file
11404 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
11405 compiler looks for a precompiled header in each directory just before it
11406 looks for the include file in that directory. The name searched for is
11407 the name specified in the @code{#include} with @samp{.gch} appended. If
11408 the precompiled header file can't be used, it is ignored.
11410 For instance, if you have @code{#include "all.h"}, and you have
11411 @file{all.h.gch} in the same directory as @file{all.h}, then the
11412 precompiled header file will be used if possible, and the original
11413 header will be used otherwise.
11415 Alternatively, you might decide to put the precompiled header file in a
11416 directory and use @option{-I} to ensure that directory is searched
11417 before (or instead of) the directory containing the original header.
11418 Then, if you want to check that the precompiled header file is always
11419 used, you can put a file of the same name as the original header in this
11420 directory containing an @code{#error} command.
11422 This also works with @option{-include}. So yet another way to use
11423 precompiled headers, good for projects not designed with precompiled
11424 header files in mind, is to simply take most of the header files used by
11425 a project, include them from another header file, precompile that header
11426 file, and @option{-include} the precompiled header. If the header files
11427 have guards against multiple inclusion, they will be skipped because
11428 they've already been included (in the precompiled header).
11430 If you need to precompile the same header file for different
11431 languages, targets, or compiler options, you can instead make a
11432 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11433 header in the directory, perhaps using @option{-o}. It doesn't matter
11434 what you call the files in the directory, every precompiled header in
11435 the directory will be considered. The first precompiled header
11436 encountered in the directory that is valid for this compilation will
11437 be used; they're searched in no particular order.
11439 There are many other possibilities, limited only by your imagination,
11440 good sense, and the constraints of your build system.
11442 A precompiled header file can be used only when these conditions apply:
11446 Only one precompiled header can be used in a particular compilation.
11449 A precompiled header can't be used once the first C token is seen. You
11450 can have preprocessor directives before a precompiled header; you can
11451 even include a precompiled header from inside another header, so long as
11452 there are no C tokens before the @code{#include}.
11455 The precompiled header file must be produced for the same language as
11456 the current compilation. You can't use a C precompiled header for a C++
11460 The precompiled header file must be produced by the same compiler
11461 version and configuration as the current compilation is using.
11462 The easiest way to guarantee this is to use the same compiler binary
11463 for creating and using precompiled headers.
11466 Any macros defined before the precompiled header is included must
11467 either be defined in the same way as when the precompiled header was
11468 generated, or must not affect the precompiled header, which usually
11469 means that the they don't appear in the precompiled header at all.
11471 The @option{-D} option is one way to define a macro before a
11472 precompiled header is included; using a @code{#define} can also do it.
11473 There are also some options that define macros implicitly, like
11474 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
11477 @item If debugging information is output when using the precompiled
11478 header, using @option{-g} or similar, the same kind of debugging information
11479 must have been output when building the precompiled header. However,
11480 a precompiled header built using @option{-g} can be used in a compilation
11481 when no debugging information is being output.
11483 @item The same @option{-m} options must generally be used when building
11484 and using the precompiled header. @xref{Submodel Options},
11485 for any cases where this rule is relaxed.
11487 @item Each of the following options must be the same when building and using
11488 the precompiled header:
11490 @gccoptlist{-fexceptions -funit-at-a-time}
11493 Some other command-line options starting with @option{-f},
11494 @option{-p}, or @option{-O} must be defined in the same way as when
11495 the precompiled header was generated. At present, it's not clear
11496 which options are safe to change and which are not; the safest choice
11497 is to use exactly the same options when generating and using the
11498 precompiled header. The following are known to be safe:
11500 @gccoptlist{-pedantic-errors}
11504 For all of these except the last, the compiler will automatically
11505 ignore the precompiled header if the conditions aren't met. If you
11506 find an option combination that doesn't work and doesn't cause the
11507 precompiled header to be ignored, please consider filing a bug report,
11510 @node Running Protoize
11511 @section Running Protoize
11513 The program @code{protoize} is an optional part of GCC@. You can use
11514 it to add prototypes to a program, thus converting the program to ISO
11515 C in one respect. The companion program @code{unprotoize} does the
11516 reverse: it removes argument types from any prototypes that are found.
11518 When you run these programs, you must specify a set of source files as
11519 command line arguments. The conversion programs start out by compiling
11520 these files to see what functions they define. The information gathered
11521 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11523 After scanning comes actual conversion. The specified files are all
11524 eligible to be converted; any files they include (whether sources or
11525 just headers) are eligible as well.
11527 But not all the eligible files are converted. By default,
11528 @code{protoize} and @code{unprotoize} convert only source and header
11529 files in the current directory. You can specify additional directories
11530 whose files should be converted with the @option{-d @var{directory}}
11531 option. You can also specify particular files to exclude with the
11532 @option{-x @var{file}} option. A file is converted if it is eligible, its
11533 directory name matches one of the specified directory names, and its
11534 name within the directory has not been excluded.
11536 Basic conversion with @code{protoize} consists of rewriting most
11537 function definitions and function declarations to specify the types of
11538 the arguments. The only ones not rewritten are those for varargs
11541 @code{protoize} optionally inserts prototype declarations at the
11542 beginning of the source file, to make them available for any calls that
11543 precede the function's definition. Or it can insert prototype
11544 declarations with block scope in the blocks where undeclared functions
11547 Basic conversion with @code{unprotoize} consists of rewriting most
11548 function declarations to remove any argument types, and rewriting
11549 function definitions to the old-style pre-ISO form.
11551 Both conversion programs print a warning for any function declaration or
11552 definition that they can't convert. You can suppress these warnings
11555 The output from @code{protoize} or @code{unprotoize} replaces the
11556 original source file. The original file is renamed to a name ending
11557 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11558 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11559 for DOS) file already exists, then the source file is simply discarded.
11561 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11562 scan the program and collect information about the functions it uses.
11563 So neither of these programs will work until GCC is installed.
11565 Here is a table of the options you can use with @code{protoize} and
11566 @code{unprotoize}. Each option works with both programs unless
11570 @item -B @var{directory}
11571 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11572 usual directory (normally @file{/usr/local/lib}). This file contains
11573 prototype information about standard system functions. This option
11574 applies only to @code{protoize}.
11576 @item -c @var{compilation-options}
11577 Use @var{compilation-options} as the options when running @command{gcc} to
11578 produce the @samp{.X} files. The special option @option{-aux-info} is
11579 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11581 Note that the compilation options must be given as a single argument to
11582 @code{protoize} or @code{unprotoize}. If you want to specify several
11583 @command{gcc} options, you must quote the entire set of compilation options
11584 to make them a single word in the shell.
11586 There are certain @command{gcc} arguments that you cannot use, because they
11587 would produce the wrong kind of output. These include @option{-g},
11588 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11589 the @var{compilation-options}, they are ignored.
11592 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11593 systems) instead of @samp{.c}. This is convenient if you are converting
11594 a C program to C++. This option applies only to @code{protoize}.
11597 Add explicit global declarations. This means inserting explicit
11598 declarations at the beginning of each source file for each function
11599 that is called in the file and was not declared. These declarations
11600 precede the first function definition that contains a call to an
11601 undeclared function. This option applies only to @code{protoize}.
11603 @item -i @var{string}
11604 Indent old-style parameter declarations with the string @var{string}.
11605 This option applies only to @code{protoize}.
11607 @code{unprotoize} converts prototyped function definitions to old-style
11608 function definitions, where the arguments are declared between the
11609 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11610 uses five spaces as the indentation. If you want to indent with just
11611 one space instead, use @option{-i " "}.
11614 Keep the @samp{.X} files. Normally, they are deleted after conversion
11618 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11619 a prototype declaration for each function in each block which calls the
11620 function without any declaration. This option applies only to
11624 Make no real changes. This mode just prints information about the conversions
11625 that would have been done without @option{-n}.
11628 Make no @samp{.save} files. The original files are simply deleted.
11629 Use this option with caution.
11631 @item -p @var{program}
11632 Use the program @var{program} as the compiler. Normally, the name
11633 @file{gcc} is used.
11636 Work quietly. Most warnings are suppressed.
11639 Print the version number, just like @option{-v} for @command{gcc}.
11642 If you need special compiler options to compile one of your program's
11643 source files, then you should generate that file's @samp{.X} file
11644 specially, by running @command{gcc} on that source file with the
11645 appropriate options and the option @option{-aux-info}. Then run
11646 @code{protoize} on the entire set of files. @code{protoize} will use
11647 the existing @samp{.X} file because it is newer than the source file.
11651 gcc -Dfoo=bar file1.c -aux-info file1.X
11656 You need to include the special files along with the rest in the
11657 @code{protoize} command, even though their @samp{.X} files already
11658 exist, because otherwise they won't get converted.
11660 @xref{Protoize Caveats}, for more information on how to use
11661 @code{protoize} successfully.