1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
8 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997,
9 1998, 1999, 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
11 Permission is granted to copy, distribute and/or modify this document
12 under the terms of the GNU Free Documentation License, Version 1.2 or
13 any later version published by the Free Software Foundation; with the
14 Invariant Sections being ``GNU General Public License'' and ``Funding
15 Free Software'', the Front-Cover texts being (a) (see below), and with
16 the Back-Cover Texts being (b) (see below). A copy of the license is
17 included in the gfdl(7) man page.
19 (a) The FSF's Front-Cover Text is:
23 (b) The FSF's Back-Cover Text is:
25 You have freedom to copy and modify this GNU Manual, like GNU
26 software. Copies published by the Free Software Foundation raise
27 funds for GNU development.
29 @c Set file name and title for the man page.
31 @settitle GNU project C and C++ compiler
33 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
34 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
35 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
36 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
37 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
38 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
39 [@option{-o} @var{outfile}] @var{infile}@dots{}
41 Only the most useful options are listed here; see below for the
42 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
45 gpl(7), gfdl(7), fsf-funding(7),
46 cpp(1), gcov(1), g77(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
47 and the Info entries for @file{gcc}, @file{cpp}, @file{g77}, @file{as},
48 @file{ld}, @file{binutils} and @file{gdb}.
51 For instructions on reporting bugs, see
52 @w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug}
53 script to report bugs is recommended.
56 See the Info entry for @command{gcc}, or
57 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
58 for contributors to GCC@.
63 @chapter GCC Command Options
64 @cindex GCC command options
65 @cindex command options
66 @cindex options, GCC command
68 @c man begin DESCRIPTION
69 When you invoke GCC, it normally does preprocessing, compilation,
70 assembly and linking. The ``overall options'' allow you to stop this
71 process at an intermediate stage. For example, the @option{-c} option
72 says not to run the linker. Then the output consists of object files
73 output by the assembler.
75 Other options are passed on to one stage of processing. Some options
76 control the preprocessor and others the compiler itself. Yet other
77 options control the assembler and linker; most of these are not
78 documented here, since you rarely need to use any of them.
80 @cindex C compilation options
81 Most of the command line options that you can use with GCC are useful
82 for C programs; when an option is only useful with another language
83 (usually C++), the explanation says so explicitly. If the description
84 for a particular option does not mention a source language, you can use
85 that option with all supported languages.
87 @cindex C++ compilation options
88 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
89 options for compiling C++ programs.
91 @cindex grouping options
92 @cindex options, grouping
93 The @command{gcc} program accepts options and file names as operands. Many
94 options have multi-letter names; therefore multiple single-letter options
95 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
98 @cindex order of options
99 @cindex options, order
100 You can mix options and other arguments. For the most part, the order
101 you use doesn't matter. Order does matter when you use several options
102 of the same kind; for example, if you specify @option{-L} more than once,
103 the directories are searched in the order specified.
105 Many options have long names starting with @samp{-f} or with
106 @samp{-W}---for example, @option{-fforce-mem},
107 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
108 these have both positive and negative forms; the negative form of
109 @option{-ffoo} would be @option{-fno-foo}. This manual documents
110 only one of these two forms, whichever one is not the default.
114 @xref{Option Index}, for an index to GCC's options.
117 * Option Summary:: Brief list of all options, without explanations.
118 * Overall Options:: Controlling the kind of output:
119 an executable, object files, assembler files,
120 or preprocessed source.
121 * Invoking G++:: Compiling C++ programs.
122 * C Dialect Options:: Controlling the variant of C language compiled.
123 * C++ Dialect Options:: Variations on C++.
124 * Objective-C Dialect Options:: Variations on Objective-C.
125 * Language Independent Options:: Controlling how diagnostics should be
127 * Warning Options:: How picky should the compiler be?
128 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
129 * Optimize Options:: How much optimization?
130 * Preprocessor Options:: Controlling header files and macro definitions.
131 Also, getting dependency information for Make.
132 * Assembler Options:: Passing options to the assembler.
133 * Link Options:: Specifying libraries and so on.
134 * Directory Options:: Where to find header files and libraries.
135 Where to find the compiler executable files.
136 * Spec Files:: How to pass switches to sub-processes.
137 * Target Options:: Running a cross-compiler, or an old version of GCC.
138 * Submodel Options:: Specifying minor hardware or convention variations,
139 such as 68010 vs 68020.
140 * Code Gen Options:: Specifying conventions for function calls, data layout
142 * Environment Variables:: Env vars that affect GCC.
143 * Precompiled Headers:: Compiling a header once, and using it many times.
144 * Running Protoize:: Automatically adding or removing function prototypes.
150 @section Option Summary
152 Here is a summary of all the options, grouped by type. Explanations are
153 in the following sections.
156 @item Overall Options
157 @xref{Overall Options,,Options Controlling the Kind of Output}.
158 @gccoptlist{-c -S -E -o @var{file} -pipe -pass-exit-codes @gol
159 -x @var{language} -v -### --help --target-help --version}
161 @item C Language Options
162 @xref{C Dialect Options,,Options Controlling C Dialect}.
163 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
164 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
165 -fhosted -ffreestanding -fms-extensions @gol
166 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
167 -fallow-single-precision -fcond-mismatch @gol
168 -fsigned-bitfields -fsigned-char @gol
169 -funsigned-bitfields -funsigned-char @gol
172 @item C++ Language Options
173 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
174 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
175 -fconserve-space -fno-const-strings @gol
176 -fno-elide-constructors @gol
177 -fno-enforce-eh-specs @gol
178 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
179 -fno-implicit-templates @gol
180 -fno-implicit-inline-templates @gol
181 -fno-implement-inlines -fms-extensions @gol
182 -fno-nonansi-builtins -fno-operator-names @gol
183 -fno-optional-diags -fpermissive @gol
184 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
185 -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
186 -fno-default-inline -Wabi -Wctor-dtor-privacy @gol
187 -Wnon-virtual-dtor -Wreorder @gol
188 -Weffc++ -Wno-deprecated @gol
189 -Wno-non-template-friend -Wold-style-cast @gol
190 -Woverloaded-virtual -Wno-pmf-conversions @gol
191 -Wsign-promo -Wsynth}
193 @item Objective-C Language Options
194 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
196 -fconstant-string-class=@var{class-name} @gol
197 -fgnu-runtime -fnext-runtime @gol
198 -fno-nil-receivers @gol
199 -fobjc-exceptions @gol
200 -freplace-objc-classes @gol
203 -Wno-protocol -Wselector -Wundeclared-selector}
205 @item Language Independent Options
206 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
207 @gccoptlist{-fmessage-length=@var{n} @gol
208 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
210 @item Warning Options
211 @xref{Warning Options,,Options to Request or Suppress Warnings}.
212 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
213 -w -Wextra -Wall -Waggregate-return @gol
214 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
215 -Wconversion -Wno-deprecated-declarations @gol
216 -Wdisabled-optimization -Wno-div-by-zero -Wendif-labels @gol
217 -Werror -Werror-implicit-function-declaration @gol
218 -Wfloat-equal -Wformat -Wformat=2 @gol
219 -Wno-format-extra-args -Wformat-nonliteral @gol
220 -Wformat-security -Wformat-y2k @gol
221 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
222 -Wimport -Wno-import -Winit-self -Winline @gol
223 -Wno-invalid-offsetof -Winvalid-pch @gol
224 -Wlarger-than-@var{len} -Wlong-long @gol
225 -Wmain -Wmissing-braces @gol
226 -Wmissing-format-attribute -Wmissing-noreturn @gol
227 -Wno-multichar -Wnonnull -Wpacked -Wpadded @gol
228 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
229 -Wreturn-type -Wsequence-point -Wshadow @gol
230 -Wsign-compare -Wstrict-aliasing @gol
231 -Wswitch -Wswitch-default -Wswitch-enum @gol
232 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
233 -Wunknown-pragmas -Wunreachable-code @gol
234 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
235 -Wunused-value -Wunused-variable -Wwrite-strings}
237 @item C-only Warning Options
238 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
239 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
240 -Wstrict-prototypes -Wtraditional @gol
241 -Wdeclaration-after-statement}
243 @item Debugging Options
244 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
245 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
246 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
247 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
248 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
249 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
250 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
251 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
252 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs @gol
253 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
254 -ftest-coverage -ftime-report @gol
255 -g -g@var{level} -gcoff -gdwarf-2 @gol
256 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
257 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
258 -print-multi-directory -print-multi-lib @gol
259 -print-prog-name=@var{program} -print-search-dirs -Q @gol
262 @item Optimization Options
263 @xref{Optimize Options,,Options that Control Optimization}.
264 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
265 -falign-labels=@var{n} -falign-loops=@var{n} @gol
266 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
267 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
268 -fcaller-saves -fcprop-registers @gol
269 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
270 -fdelayed-branch -fdelete-null-pointer-checks @gol
271 -fexpensive-optimizations -ffast-math -ffloat-store @gol
272 -fforce-addr -fforce-mem -ffunction-sections @gol
273 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -floop-optimize @gol
274 -fcrossjumping -fif-conversion -fif-conversion2 @gol
275 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
276 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
277 -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
278 -fno-default-inline -fno-defer-pop @gol
279 -fno-function-cse -fno-guess-branch-probability @gol
280 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
281 -funsafe-math-optimizations -ffinite-math-only @gol
282 -fno-trapping-math -fno-zero-initialized-in-bss @gol
283 -fomit-frame-pointer -foptimize-register-move @gol
284 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
285 -fprofile-generate -fprofile-use @gol
286 -freduce-all-givs -fregmove -frename-registers @gol
287 -freorder-blocks -freorder-functions @gol
288 -frerun-cse-after-loop -frerun-loop-opt @gol
289 -frounding-math -fschedule-insns -fschedule-insns2 @gol
290 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
291 -fsched-spec-load-dangerous @gol
292 -fsched-stalled-insns=@var{n} -sched-stalled-insns-dep=@var{n} @gol
293 -fsched2-use-superblocks @gol
294 -fsched2-use-traces -fsignaling-nans @gol
295 -fsingle-precision-constant @gol
296 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
297 -funroll-all-loops -funroll-loops -fpeel-loops @gol
298 -funswitch-loops -fold-unroll-loops -fold-unroll-all-loops @gol
299 --param @var{name}=@var{value}
300 -O -O0 -O1 -O2 -O3 -Os}
302 @item Preprocessor Options
303 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
304 @gccoptlist{-A@var{question}=@var{answer} @gol
305 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
306 -C -dD -dI -dM -dN @gol
307 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
308 -idirafter @var{dir} @gol
309 -include @var{file} -imacros @var{file} @gol
310 -iprefix @var{file} -iwithprefix @var{dir} @gol
311 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
312 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
313 -P -fworking-directory -remap @gol
314 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
315 -Xpreprocessor @var{option}}
317 @item Assembler Option
318 @xref{Assembler Options,,Passing Options to the Assembler}.
319 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
322 @xref{Link Options,,Options for Linking}.
323 @gccoptlist{@var{object-file-name} -l@var{library} @gol
324 -nostartfiles -nodefaultlibs -nostdlib -pie @gol
325 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
326 -Wl,@var{option} -Xlinker @var{option} @gol
329 @item Directory Options
330 @xref{Directory Options,,Options for Directory Search}.
331 @gccoptlist{-B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
334 @c I wrote this xref this way to avoid overfull hbox. -- rms
335 @xref{Target Options}.
336 @gccoptlist{-V @var{version} -b @var{machine}}
338 @item Machine Dependent Options
339 @xref{Submodel Options,,Hardware Models and Configurations}.
341 @emph{M680x0 Options}
342 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
343 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
344 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
345 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
346 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
348 @emph{M68hc1x Options}
349 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
350 -mauto-incdec -minmax -mlong-calls -mshort @gol
351 -msoft-reg-count=@var{count}}
354 @gccoptlist{-mg -mgnu -munix}
357 @gccoptlist{-mcpu=@var{cpu-type} @gol
358 -mtune=@var{cpu-type} @gol
359 -mcmodel=@var{code-model} @gol
360 -m32 -m64 -mapp-regs -mno-app-regs @gol
361 -mfaster-structs -mno-faster-structs @gol
362 -mflat -mno-flat -mfpu -mno-fpu @gol
363 -mhard-float -msoft-float @gol
364 -mhard-quad-float -msoft-quad-float @gol
365 -mimpure-text -mno-impure-text -mlittle-endian @gol
366 -mstack-bias -mno-stack-bias @gol
367 -munaligned-doubles -mno-unaligned-doubles @gol
368 -mv8plus -mno-v8plus -mvis -mno-vis @gol
369 -mcypress -mf930 -mf934 @gol
370 -msparclite -msupersparc -mv8}
373 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
374 -mapcs-26 -mapcs-32 @gol
375 -mapcs-stack-check -mno-apcs-stack-check @gol
376 -mapcs-float -mno-apcs-float @gol
377 -mapcs-reentrant -mno-apcs-reentrant @gol
378 -msched-prolog -mno-sched-prolog @gol
379 -mlittle-endian -mbig-endian -mwords-little-endian @gol
380 -malignment-traps -mno-alignment-traps @gol
381 -msoft-float -mhard-float -mfpe @gol
382 -mthumb-interwork -mno-thumb-interwork @gol
383 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
384 -mstructure-size-boundary=@var{n} @gol
385 -mabort-on-noreturn @gol
386 -mlong-calls -mno-long-calls @gol
387 -msingle-pic-base -mno-single-pic-base @gol
388 -mpic-register=@var{reg} @gol
389 -mnop-fun-dllimport @gol
390 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
391 -mpoke-function-name @gol
393 -mtpcs-frame -mtpcs-leaf-frame @gol
394 -mcaller-super-interworking -mcallee-super-interworking}
396 @emph{MN10300 Options}
397 @gccoptlist{-mmult-bug -mno-mult-bug @gol
398 -mam33 -mno-am33 @gol
399 -mam33-2 -mno-am33-2 @gol
402 @emph{M32R/D Options}
403 @gccoptlist{-m32r2 -m32rx -m32r @gol
405 -malign-loops -mno-align-loops @gol
406 -missue-rate=@var{number} @gol
407 -mbranch-cost=@var{number} @gol
408 -mmodel=@var{code-size-model-type} @gol
409 -msdata=@var{sdata-type} @gol
410 -mno-flush-func -mflush-func=@var{name} @gol
411 -mno-flush-trap -mflush-trap=@var{number} @gol
414 @emph{RS/6000 and PowerPC Options}
415 @gccoptlist{-mcpu=@var{cpu-type} @gol
416 -mtune=@var{cpu-type} @gol
417 -mpower -mno-power -mpower2 -mno-power2 @gol
418 -mpowerpc -mpowerpc64 -mno-powerpc @gol
419 -maltivec -mno-altivec @gol
420 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
421 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
422 -mnew-mnemonics -mold-mnemonics @gol
423 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
424 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
425 -malign-power -malign-natural @gol
426 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
427 -mstring -mno-string -mupdate -mno-update @gol
428 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
429 -mstrict-align -mno-strict-align -mrelocatable @gol
430 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
431 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
432 -mdynamic-no-pic @gol
433 -mprioritize-restricted-insns=@var{priority} @gol
434 -msched-costly-dep=@var{dependence_type} @gol
435 -minsert-sched-nops=@var{scheme} @gol
436 -mcall-sysv -mcall-netbsd @gol
437 -maix-struct-return -msvr4-struct-return @gol
438 -mabi=altivec -mabi=no-altivec @gol
439 -mabi=spe -mabi=no-spe @gol
440 -misel=yes -misel=no @gol
441 -mspe=yes -mspe=no @gol
442 -mfloat-gprs=yes -mfloat-gprs=no @gol
443 -mprototype -mno-prototype @gol
444 -msim -mmvme -mads -myellowknife -memb -msdata @gol
445 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
447 @emph{Darwin Options}
448 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
449 -arch_only -bind_at_load -bundle -bundle_loader @gol
450 -client_name -compatibility_version -current_version @gol
451 -dependency-file -dylib_file -dylinker_install_name @gol
452 -dynamic -dynamiclib -exported_symbols_list @gol
453 -filelist -flat_namespace -force_cpusubtype_ALL @gol
454 -force_flat_namespace -headerpad_max_install_names @gol
455 -image_base -init -install_name -keep_private_externs @gol
456 -multi_module -multiply_defined -multiply_defined_unused @gol
457 -noall_load -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
458 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
459 -private_bundle -read_only_relocs -sectalign @gol
460 -sectobjectsymbols -whyload -seg1addr @gol
461 -sectcreate -sectobjectsymbols -sectorder @gol
462 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
463 -segprot -segs_read_only_addr -segs_read_write_addr @gol
464 -single_module -static -sub_library -sub_umbrella @gol
465 -twolevel_namespace -umbrella -undefined @gol
466 -unexported_symbols_list -weak_reference_mismatches @gol
470 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
471 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
472 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
473 -mxgot -mno-xgot -membedded-pic -mno-embedded-pic @gol
474 -mgp32 -mgp64 -mfp32 -mfp64 -mhard-float -msoft-float @gol
475 -msingle-float -mdouble-float -mint64 -mlong64 -mlong32 @gol
476 -G@var{num} -membedded-data -mno-embedded-data @gol
477 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
478 -msplit-addresses -mno-split-addresses @gol
479 -mexplicit-relocs -mno-explicit-relocs @gol
480 -mrnames -mno-rnames @gol
481 -mcheck-zero-division -mno-check-zero-division @gol
482 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
483 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
484 -mfix-sb1 -mno-fix-sb1 -mflush-func=@var{func} @gol
485 -mno-flush-func -mbranch-likely -mno-branch-likely}
487 @emph{i386 and x86-64 Options}
488 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
489 -mfpmath=@var{unit} @gol
490 -masm=@var{dialect} -mno-fancy-math-387 @gol
491 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
492 -mno-wide-multiply -mrtd -malign-double @gol
493 -mpreferred-stack-boundary=@var{num} @gol
494 -mmmx -msse -msse2 -mpni -m3dnow @gol
495 -mthreads -mno-align-stringops -minline-all-stringops @gol
496 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
497 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
498 -mno-red-zone -mno-tls-direct-seg-refs @gol
499 -mcmodel=@var{code-model} @gol
503 @gccoptlist{-march=@var{architecture-type} @gol
504 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
505 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
506 -mjump-in-delay -mlinker-opt -mlong-calls @gol
507 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
508 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
509 -mno-jump-in-delay -mno-long-load-store @gol
510 -mno-portable-runtime -mno-soft-float @gol
511 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
512 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
513 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
514 -nolibdld -static -threads}
516 @emph{Intel 960 Options}
517 @gccoptlist{-m@var{cpu-type} -masm-compat -mclean-linkage @gol
518 -mcode-align -mcomplex-addr -mleaf-procedures @gol
519 -mic-compat -mic2.0-compat -mic3.0-compat @gol
520 -mintel-asm -mno-clean-linkage -mno-code-align @gol
521 -mno-complex-addr -mno-leaf-procedures @gol
522 -mno-old-align -mno-strict-align -mno-tail-call @gol
523 -mnumerics -mold-align -msoft-float -mstrict-align @gol
526 @emph{DEC Alpha Options}
527 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
528 -mieee -mieee-with-inexact -mieee-conformant @gol
529 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
530 -mtrap-precision=@var{mode} -mbuild-constants @gol
531 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
532 -mbwx -mmax -mfix -mcix @gol
533 -mfloat-vax -mfloat-ieee @gol
534 -mexplicit-relocs -msmall-data -mlarge-data @gol
535 -msmall-text -mlarge-text @gol
536 -mmemory-latency=@var{time}}
538 @emph{DEC Alpha/VMS Options}
539 @gccoptlist{-mvms-return-codes}
541 @emph{H8/300 Options}
542 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
545 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
546 -m4-nofpu -m4-single-only -m4-single -m4 @gol
547 -m5-64media -m5-64media-nofpu @gol
548 -m5-32media -m5-32media-nofpu @gol
549 -m5-compact -m5-compact-nofpu @gol
550 -mb -ml -mdalign -mrelax @gol
551 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
552 -mieee -misize -mpadstruct -mspace @gol
553 -mprefergot -musermode}
555 @emph{System V Options}
556 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
559 @gccoptlist{-EB -EL @gol
560 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
561 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
563 @emph{TMS320C3x/C4x Options}
564 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
565 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
566 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
567 -mparallel-insns -mparallel-mpy -mpreserve-float}
570 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
571 -mprolog-function -mno-prolog-function -mspace @gol
572 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
573 -mapp-regs -mno-app-regs @gol
574 -mdisable-callt -mno-disable-callt @gol
580 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
581 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
582 -mregparam -mnoregparam -msb -mnosb @gol
583 -mbitfield -mnobitfield -mhimem -mnohimem}
586 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
587 -mcall-prologues -mno-tablejump -mtiny-stack}
590 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
591 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
592 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
593 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
594 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
597 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
598 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
599 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
600 -mno-base-addresses -msingle-exit -mno-single-exit}
603 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
604 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
605 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
606 -minline-float-divide-max-throughput @gol
607 -minline-int-divide-min-latency @gol
608 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
609 -mfixed-range=@var{register-range}}
612 @gccoptlist{-mextmem -mextmemory -monchip -mno-asm-optimize @gol
613 -masm-optimize -mbranch-cost=@var{n} -mcond-exec=@var{n}}
615 @emph{S/390 and zSeries Options}
616 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
617 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
618 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
619 -m64 -m31 -mdebug -mno-debug -mesa -mzarch -mfused-madd -mno-fused-madd}
622 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
623 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
624 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
625 -mstack-align -mdata-align -mconst-align @gol
626 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
627 -melf -maout -melinux -mlinux -sim -sim2}
629 @emph{PDP-11 Options}
630 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
631 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
632 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
633 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
634 -mbranch-expensive -mbranch-cheap @gol
635 -msplit -mno-split -munix-asm -mdec-asm}
637 @emph{Xstormy16 Options}
640 @emph{Xtensa Options}
641 @gccoptlist{-mconst16 -mno-const16 @gol
642 -mfused-madd -mno-fused-madd @gol
643 -mtext-section-literals -mno-text-section-literals @gol
644 -mtarget-align -mno-target-align @gol
645 -mlongcalls -mno-longcalls}
648 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
649 -mhard-float -msoft-float @gol
650 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
651 -mdouble -mno-double @gol
652 -mmedia -mno-media -mmuladd -mno-muladd @gol
653 -mlibrary-pic -macc-4 -macc-8 @gol
654 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
655 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
656 -mvliw-branch -mno-vliw-branch @gol
657 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
658 -mno-nested-cond-exec -mtomcat-stats @gol
661 @item Code Generation Options
662 @xref{Code Gen Options,,Options for Code Generation Conventions}.
663 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
664 -ffixed-@var{reg} -fexceptions @gol
665 -fnon-call-exceptions -funwind-tables @gol
666 -fasynchronous-unwind-tables @gol
667 -finhibit-size-directive -finstrument-functions @gol
668 -fno-common -fno-ident @gol
669 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
670 -freg-struct-return -fshared-data -fshort-enums @gol
671 -fshort-double -fshort-wchar @gol
672 -fverbose-asm -fpack-struct -fstack-check @gol
673 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
674 -fargument-alias -fargument-noalias @gol
675 -fargument-noalias-global -fleading-underscore @gol
676 -ftls-model=@var{model} @gol
677 -ftrapv -fwrapv -fbounds-check}
681 * Overall Options:: Controlling the kind of output:
682 an executable, object files, assembler files,
683 or preprocessed source.
684 * C Dialect Options:: Controlling the variant of C language compiled.
685 * C++ Dialect Options:: Variations on C++.
686 * Objective-C Dialect Options:: Variations on Objective-C.
687 * Language Independent Options:: Controlling how diagnostics should be
689 * Warning Options:: How picky should the compiler be?
690 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
691 * Optimize Options:: How much optimization?
692 * Preprocessor Options:: Controlling header files and macro definitions.
693 Also, getting dependency information for Make.
694 * Assembler Options:: Passing options to the assembler.
695 * Link Options:: Specifying libraries and so on.
696 * Directory Options:: Where to find header files and libraries.
697 Where to find the compiler executable files.
698 * Spec Files:: How to pass switches to sub-processes.
699 * Target Options:: Running a cross-compiler, or an old version of GCC.
702 @node Overall Options
703 @section Options Controlling the Kind of Output
705 Compilation can involve up to four stages: preprocessing, compilation
706 proper, assembly and linking, always in that order. GCC is capable of
707 preprocessing and compiling several files either into several
708 assembler input files, or into one assembler input file; then each
709 assembler input file produces an object file, and linking combines all
710 the object files (those newly compiled, and those specified as input)
711 into an executable file.
713 @cindex file name suffix
714 For any given input file, the file name suffix determines what kind of
719 C source code which must be preprocessed.
722 C source code which should not be preprocessed.
725 C++ source code which should not be preprocessed.
728 Objective-C source code. Note that you must link with the library
729 @file{libobjc.a} to make an Objective-C program work.
732 Objective-C source code which should not be preprocessed.
735 C or C++ header file to be turned into a precompiled header.
739 @itemx @var{file}.cxx
740 @itemx @var{file}.cpp
741 @itemx @var{file}.CPP
742 @itemx @var{file}.c++
744 C++ source code which must be preprocessed. Note that in @samp{.cxx},
745 the last two letters must both be literally @samp{x}. Likewise,
746 @samp{.C} refers to a literal capital C@.
750 C++ header file to be turned into a precompiled header.
753 @itemx @var{file}.for
754 @itemx @var{file}.FOR
755 Fortran source code which should not be preprocessed.
758 @itemx @var{file}.fpp
759 @itemx @var{file}.FPP
760 Fortran source code which must be preprocessed (with the traditional
764 Fortran source code which must be preprocessed with a RATFOR
765 preprocessor (not included with GCC)@.
767 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
768 Using and Porting GNU Fortran}, for more details of the handling of
771 @c FIXME: Descriptions of Java file types.
778 Ada source code file which contains a library unit declaration (a
779 declaration of a package, subprogram, or generic, or a generic
780 instantiation), or a library unit renaming declaration (a package,
781 generic, or subprogram renaming declaration). Such files are also
784 @itemx @var{file}.adb
785 Ada source code file containing a library unit body (a subprogram or
786 package body). Such files are also called @dfn{bodies}.
788 @c GCC also knows about some suffixes for languages not yet included:
797 Assembler code which must be preprocessed.
800 An object file to be fed straight into linking.
801 Any file name with no recognized suffix is treated this way.
805 You can specify the input language explicitly with the @option{-x} option:
808 @item -x @var{language}
809 Specify explicitly the @var{language} for the following input files
810 (rather than letting the compiler choose a default based on the file
811 name suffix). This option applies to all following input files until
812 the next @option{-x} option. Possible values for @var{language} are:
814 c c-header cpp-output
815 c++ c++-header c++-cpp-output
816 objective-c objective-c-header objc-cpp-output
817 assembler assembler-with-cpp
819 f77 f77-cpp-input ratfor
825 Turn off any specification of a language, so that subsequent files are
826 handled according to their file name suffixes (as they are if @option{-x}
827 has not been used at all).
829 @item -pass-exit-codes
830 @opindex pass-exit-codes
831 Normally the @command{gcc} program will exit with the code of 1 if any
832 phase of the compiler returns a non-success return code. If you specify
833 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
834 numerically highest error produced by any phase that returned an error
838 If you only want some of the stages of compilation, you can use
839 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
840 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
841 @command{gcc} is to stop. Note that some combinations (for example,
842 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
847 Compile or assemble the source files, but do not link. The linking
848 stage simply is not done. The ultimate output is in the form of an
849 object file for each source file.
851 By default, the object file name for a source file is made by replacing
852 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
854 Unrecognized input files, not requiring compilation or assembly, are
859 Stop after the stage of compilation proper; do not assemble. The output
860 is in the form of an assembler code file for each non-assembler input
863 By default, the assembler file name for a source file is made by
864 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
866 Input files that don't require compilation are ignored.
870 Stop after the preprocessing stage; do not run the compiler proper. The
871 output is in the form of preprocessed source code, which is sent to the
874 Input files which don't require preprocessing are ignored.
876 @cindex output file option
879 Place output in file @var{file}. This applies regardless to whatever
880 sort of output is being produced, whether it be an executable file,
881 an object file, an assembler file or preprocessed C code.
883 If you specify @option{-o} when compiling more than one input file, or
884 you are producing an executable file as output, all the source files
885 on the command line will be compiled at once.
887 If @option{-o} is not specified, the default is to put an executable file
888 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
889 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
890 all preprocessed C source on standard output.
894 Print (on standard error output) the commands executed to run the stages
895 of compilation. Also print the version number of the compiler driver
896 program and of the preprocessor and the compiler proper.
900 Like @option{-v} except the commands are not executed and all command
901 arguments are quoted. This is useful for shell scripts to capture the
902 driver-generated command lines.
906 Use pipes rather than temporary files for communication between the
907 various stages of compilation. This fails to work on some systems where
908 the assembler is unable to read from a pipe; but the GNU assembler has
913 Print (on the standard output) a description of the command line options
914 understood by @command{gcc}. If the @option{-v} option is also specified
915 then @option{--help} will also be passed on to the various processes
916 invoked by @command{gcc}, so that they can display the command line options
917 they accept. If the @option{-Wextra} option is also specified then command
918 line options which have no documentation associated with them will also
923 Print (on the standard output) a description of target specific command
924 line options for each tool.
928 Display the version number and copyrights of the invoked GCC.
932 @section Compiling C++ Programs
934 @cindex suffixes for C++ source
935 @cindex C++ source file suffixes
936 C++ source files conventionally use one of the suffixes @samp{.C},
937 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
938 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
939 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
940 files with these names and compiles them as C++ programs even if you
941 call the compiler the same way as for compiling C programs (usually
942 with the name @command{gcc}).
946 However, C++ programs often require class libraries as well as a
947 compiler that understands the C++ language---and under some
948 circumstances, you might want to compile programs or header files from
949 standard input, or otherwise without a suffix that flags them as C++
950 programs. You might also like to precompile a C header file with a
951 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
952 program that calls GCC with the default language set to C++, and
953 automatically specifies linking against the C++ library. On many
954 systems, @command{g++} is also installed with the name @command{c++}.
956 @cindex invoking @command{g++}
957 When you compile C++ programs, you may specify many of the same
958 command-line options that you use for compiling programs in any
959 language; or command-line options meaningful for C and related
960 languages; or options that are meaningful only for C++ programs.
961 @xref{C Dialect Options,,Options Controlling C Dialect}, for
962 explanations of options for languages related to C@.
963 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
964 explanations of options that are meaningful only for C++ programs.
966 @node C Dialect Options
967 @section Options Controlling C Dialect
968 @cindex dialect options
969 @cindex language dialect options
970 @cindex options, dialect
972 The following options control the dialect of C (or languages derived
973 from C, such as C++ and Objective-C) that the compiler accepts:
980 In C mode, support all ISO C90 programs. In C++ mode,
981 remove GNU extensions that conflict with ISO C++.
983 This turns off certain features of GCC that are incompatible with ISO
984 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
985 such as the @code{asm} and @code{typeof} keywords, and
986 predefined macros such as @code{unix} and @code{vax} that identify the
987 type of system you are using. It also enables the undesirable and
988 rarely used ISO trigraph feature. For the C compiler,
989 it disables recognition of C++ style @samp{//} comments as well as
990 the @code{inline} keyword.
992 The alternate keywords @code{__asm__}, @code{__extension__},
993 @code{__inline__} and @code{__typeof__} continue to work despite
994 @option{-ansi}. You would not want to use them in an ISO C program, of
995 course, but it is useful to put them in header files that might be included
996 in compilations done with @option{-ansi}. Alternate predefined macros
997 such as @code{__unix__} and @code{__vax__} are also available, with or
998 without @option{-ansi}.
1000 The @option{-ansi} option does not cause non-ISO programs to be
1001 rejected gratuitously. For that, @option{-pedantic} is required in
1002 addition to @option{-ansi}. @xref{Warning Options}.
1004 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1005 option is used. Some header files may notice this macro and refrain
1006 from declaring certain functions or defining certain macros that the
1007 ISO standard doesn't call for; this is to avoid interfering with any
1008 programs that might use these names for other things.
1010 Functions which would normally be built in but do not have semantics
1011 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1012 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1013 built-in functions provided by GCC}, for details of the functions
1018 Determine the language standard. This option is currently only
1019 supported when compiling C or C++. A value for this option must be
1020 provided; possible values are
1025 ISO C90 (same as @option{-ansi}).
1027 @item iso9899:199409
1028 ISO C90 as modified in amendment 1.
1034 ISO C99. Note that this standard is not yet fully supported; see
1035 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1036 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1039 Default, ISO C90 plus GNU extensions (including some C99 features).
1043 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1044 this will become the default. The name @samp{gnu9x} is deprecated.
1047 The 1998 ISO C++ standard plus amendments.
1050 The same as @option{-std=c++98} plus GNU extensions. This is the
1051 default for C++ code.
1054 Even when this option is not specified, you can still use some of the
1055 features of newer standards in so far as they do not conflict with
1056 previous C standards. For example, you may use @code{__restrict__} even
1057 when @option{-std=c99} is not specified.
1059 The @option{-std} options specifying some version of ISO C have the same
1060 effects as @option{-ansi}, except that features that were not in ISO C90
1061 but are in the specified version (for example, @samp{//} comments and
1062 the @code{inline} keyword in ISO C99) are not disabled.
1064 @xref{Standards,,Language Standards Supported by GCC}, for details of
1065 these standard versions.
1067 @item -aux-info @var{filename}
1069 Output to the given filename prototyped declarations for all functions
1070 declared and/or defined in a translation unit, including those in header
1071 files. This option is silently ignored in any language other than C@.
1073 Besides declarations, the file indicates, in comments, the origin of
1074 each declaration (source file and line), whether the declaration was
1075 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1076 @samp{O} for old, respectively, in the first character after the line
1077 number and the colon), and whether it came from a declaration or a
1078 definition (@samp{C} or @samp{F}, respectively, in the following
1079 character). In the case of function definitions, a K&R-style list of
1080 arguments followed by their declarations is also provided, inside
1081 comments, after the declaration.
1085 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1086 keyword, so that code can use these words as identifiers. You can use
1087 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1088 instead. @option{-ansi} implies @option{-fno-asm}.
1090 In C++, this switch only affects the @code{typeof} keyword, since
1091 @code{asm} and @code{inline} are standard keywords. You may want to
1092 use the @option{-fno-gnu-keywords} flag instead, which has the same
1093 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1094 switch only affects the @code{asm} and @code{typeof} keywords, since
1095 @code{inline} is a standard keyword in ISO C99.
1098 @itemx -fno-builtin-@var{function}
1099 @opindex fno-builtin
1100 @cindex built-in functions
1101 Don't recognize built-in functions that do not begin with
1102 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1103 functions provided by GCC}, for details of the functions affected,
1104 including those which are not built-in functions when @option{-ansi} or
1105 @option{-std} options for strict ISO C conformance are used because they
1106 do not have an ISO standard meaning.
1108 GCC normally generates special code to handle certain built-in functions
1109 more efficiently; for instance, calls to @code{alloca} may become single
1110 instructions that adjust the stack directly, and calls to @code{memcpy}
1111 may become inline copy loops. The resulting code is often both smaller
1112 and faster, but since the function calls no longer appear as such, you
1113 cannot set a breakpoint on those calls, nor can you change the behavior
1114 of the functions by linking with a different library.
1116 With the @option{-fno-builtin-@var{function}} option
1117 only the built-in function @var{function} is
1118 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1119 function is named this is not built-in in this version of GCC, this
1120 option is ignored. There is no corresponding
1121 @option{-fbuiltin-@var{function}} option; if you wish to enable
1122 built-in functions selectively when using @option{-fno-builtin} or
1123 @option{-ffreestanding}, you may define macros such as:
1126 #define abs(n) __builtin_abs ((n))
1127 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1132 @cindex hosted environment
1134 Assert that compilation takes place in a hosted environment. This implies
1135 @option{-fbuiltin}. A hosted environment is one in which the
1136 entire standard library is available, and in which @code{main} has a return
1137 type of @code{int}. Examples are nearly everything except a kernel.
1138 This is equivalent to @option{-fno-freestanding}.
1140 @item -ffreestanding
1141 @opindex ffreestanding
1142 @cindex hosted environment
1144 Assert that compilation takes place in a freestanding environment. This
1145 implies @option{-fno-builtin}. A freestanding environment
1146 is one in which the standard library may not exist, and program startup may
1147 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1148 This is equivalent to @option{-fno-hosted}.
1150 @xref{Standards,,Language Standards Supported by GCC}, for details of
1151 freestanding and hosted environments.
1153 @item -fms-extensions
1154 @opindex fms-extensions
1155 Accept some non-standard constructs used in Microsoft header files.
1159 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1160 options for strict ISO C conformance) implies @option{-trigraphs}.
1162 @item -no-integrated-cpp
1163 @opindex no-integrated-cpp
1164 Performs a compilation in two passes: preprocessing and compiling. This
1165 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1166 @option{-B} option. The user supplied compilation step can then add in
1167 an additional preprocessing step after normal preprocessing but before
1168 compiling. The default is to use the integrated cpp (internal cpp)
1170 The semantics of this option will change if "cc1", "cc1plus", and
1171 "cc1obj" are merged.
1173 @cindex traditional C language
1174 @cindex C language, traditional
1176 @itemx -traditional-cpp
1177 @opindex traditional-cpp
1178 @opindex traditional
1179 Formerly, these options caused GCC to attempt to emulate a pre-standard
1180 C compiler. They are now only supported with the @option{-E} switch.
1181 The preprocessor continues to support a pre-standard mode. See the GNU
1182 CPP manual for details.
1184 @item -fcond-mismatch
1185 @opindex fcond-mismatch
1186 Allow conditional expressions with mismatched types in the second and
1187 third arguments. The value of such an expression is void. This option
1188 is not supported for C++.
1190 @item -funsigned-char
1191 @opindex funsigned-char
1192 Let the type @code{char} be unsigned, like @code{unsigned char}.
1194 Each kind of machine has a default for what @code{char} should
1195 be. It is either like @code{unsigned char} by default or like
1196 @code{signed char} by default.
1198 Ideally, a portable program should always use @code{signed char} or
1199 @code{unsigned char} when it depends on the signedness of an object.
1200 But many programs have been written to use plain @code{char} and
1201 expect it to be signed, or expect it to be unsigned, depending on the
1202 machines they were written for. This option, and its inverse, let you
1203 make such a program work with the opposite default.
1205 The type @code{char} is always a distinct type from each of
1206 @code{signed char} or @code{unsigned char}, even though its behavior
1207 is always just like one of those two.
1210 @opindex fsigned-char
1211 Let the type @code{char} be signed, like @code{signed char}.
1213 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1214 the negative form of @option{-funsigned-char}. Likewise, the option
1215 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1217 @item -fsigned-bitfields
1218 @itemx -funsigned-bitfields
1219 @itemx -fno-signed-bitfields
1220 @itemx -fno-unsigned-bitfields
1221 @opindex fsigned-bitfields
1222 @opindex funsigned-bitfields
1223 @opindex fno-signed-bitfields
1224 @opindex fno-unsigned-bitfields
1225 These options control whether a bit-field is signed or unsigned, when the
1226 declaration does not use either @code{signed} or @code{unsigned}. By
1227 default, such a bit-field is signed, because this is consistent: the
1228 basic integer types such as @code{int} are signed types.
1230 @item -fwritable-strings
1231 @opindex fwritable-strings
1232 Store string constants in the writable data segment and don't uniquize
1233 them. This is for compatibility with old programs which assume they can
1234 write into string constants.
1236 Writing into string constants is a very bad idea; ``constants'' should
1239 This option is deprecated.
1242 @node C++ Dialect Options
1243 @section Options Controlling C++ Dialect
1245 @cindex compiler options, C++
1246 @cindex C++ options, command line
1247 @cindex options, C++
1248 This section describes the command-line options that are only meaningful
1249 for C++ programs; but you can also use most of the GNU compiler options
1250 regardless of what language your program is in. For example, you
1251 might compile a file @code{firstClass.C} like this:
1254 g++ -g -frepo -O -c firstClass.C
1258 In this example, only @option{-frepo} is an option meant
1259 only for C++ programs; you can use the other options with any
1260 language supported by GCC@.
1262 Here is a list of options that are @emph{only} for compiling C++ programs:
1266 @item -fabi-version=@var{n}
1267 @opindex fabi-version
1268 Use version @var{n} of the C++ ABI. Version 2 is the version of the
1269 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1270 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1271 the version that conforms most closely to the C++ ABI specification.
1272 Therefore, the ABI obtained using version 0 will change as ABI bugs
1275 The default is version 1.
1277 @item -fno-access-control
1278 @opindex fno-access-control
1279 Turn off all access checking. This switch is mainly useful for working
1280 around bugs in the access control code.
1284 Check that the pointer returned by @code{operator new} is non-null
1285 before attempting to modify the storage allocated. This check is
1286 normally unnecessary because the C++ standard specifies that
1287 @code{operator new} will only return @code{0} if it is declared
1288 @samp{throw()}, in which case the compiler will always check the
1289 return value even without this option. In all other cases, when
1290 @code{operator new} has a non-empty exception specification, memory
1291 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1292 @samp{new (nothrow)}.
1294 @item -fconserve-space
1295 @opindex fconserve-space
1296 Put uninitialized or runtime-initialized global variables into the
1297 common segment, as C does. This saves space in the executable at the
1298 cost of not diagnosing duplicate definitions. If you compile with this
1299 flag and your program mysteriously crashes after @code{main()} has
1300 completed, you may have an object that is being destroyed twice because
1301 two definitions were merged.
1303 This option is no longer useful on most targets, now that support has
1304 been added for putting variables into BSS without making them common.
1306 @item -fno-const-strings
1307 @opindex fno-const-strings
1308 Give string constants type @code{char *} instead of type @code{const
1309 char *}. By default, G++ uses type @code{const char *} as required by
1310 the standard. Even if you use @option{-fno-const-strings}, you cannot
1311 actually modify the value of a string constant, unless you also use
1312 @option{-fwritable-strings}.
1314 This option might be removed in a future release of G++. For maximum
1315 portability, you should structure your code so that it works with
1316 string constants that have type @code{const char *}.
1318 @item -fno-elide-constructors
1319 @opindex fno-elide-constructors
1320 The C++ standard allows an implementation to omit creating a temporary
1321 which is only used to initialize another object of the same type.
1322 Specifying this option disables that optimization, and forces G++ to
1323 call the copy constructor in all cases.
1325 @item -fno-enforce-eh-specs
1326 @opindex fno-enforce-eh-specs
1327 Don't check for violation of exception specifications at runtime. This
1328 option violates the C++ standard, but may be useful for reducing code
1329 size in production builds, much like defining @samp{NDEBUG}. The compiler
1330 will still optimize based on the exception specifications.
1333 @itemx -fno-for-scope
1335 @opindex fno-for-scope
1336 If @option{-ffor-scope} is specified, the scope of variables declared in
1337 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1338 as specified by the C++ standard.
1339 If @option{-fno-for-scope} is specified, the scope of variables declared in
1340 a @i{for-init-statement} extends to the end of the enclosing scope,
1341 as was the case in old versions of G++, and other (traditional)
1342 implementations of C++.
1344 The default if neither flag is given to follow the standard,
1345 but to allow and give a warning for old-style code that would
1346 otherwise be invalid, or have different behavior.
1348 @item -fno-gnu-keywords
1349 @opindex fno-gnu-keywords
1350 Do not recognize @code{typeof} as a keyword, so that code can use this
1351 word as an identifier. You can use the keyword @code{__typeof__} instead.
1352 @option{-ansi} implies @option{-fno-gnu-keywords}.
1354 @item -fno-implicit-templates
1355 @opindex fno-implicit-templates
1356 Never emit code for non-inline templates which are instantiated
1357 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1358 @xref{Template Instantiation}, for more information.
1360 @item -fno-implicit-inline-templates
1361 @opindex fno-implicit-inline-templates
1362 Don't emit code for implicit instantiations of inline templates, either.
1363 The default is to handle inlines differently so that compiles with and
1364 without optimization will need the same set of explicit instantiations.
1366 @item -fno-implement-inlines
1367 @opindex fno-implement-inlines
1368 To save space, do not emit out-of-line copies of inline functions
1369 controlled by @samp{#pragma implementation}. This will cause linker
1370 errors if these functions are not inlined everywhere they are called.
1372 @item -fms-extensions
1373 @opindex fms-extensions
1374 Disable pedantic warnings about constructs used in MFC, such as implicit
1375 int and getting a pointer to member function via non-standard syntax.
1377 @item -fno-nonansi-builtins
1378 @opindex fno-nonansi-builtins
1379 Disable built-in declarations of functions that are not mandated by
1380 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1381 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1383 @item -fno-operator-names
1384 @opindex fno-operator-names
1385 Do not treat the operator name keywords @code{and}, @code{bitand},
1386 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1387 synonyms as keywords.
1389 @item -fno-optional-diags
1390 @opindex fno-optional-diags
1391 Disable diagnostics that the standard says a compiler does not need to
1392 issue. Currently, the only such diagnostic issued by G++ is the one for
1393 a name having multiple meanings within a class.
1396 @opindex fpermissive
1397 Downgrade some diagnostics about nonconformant code from errors to
1398 warnings. Thus, using @option{-fpermissive} will allow some
1399 nonconforming code to compile.
1403 Enable automatic template instantiation at link time. This option also
1404 implies @option{-fno-implicit-templates}. @xref{Template
1405 Instantiation}, for more information.
1409 Disable generation of information about every class with virtual
1410 functions for use by the C++ runtime type identification features
1411 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1412 of the language, you can save some space by using this flag. Note that
1413 exception handling uses the same information, but it will generate it as
1418 Emit statistics about front-end processing at the end of the compilation.
1419 This information is generally only useful to the G++ development team.
1421 @item -ftemplate-depth-@var{n}
1422 @opindex ftemplate-depth
1423 Set the maximum instantiation depth for template classes to @var{n}.
1424 A limit on the template instantiation depth is needed to detect
1425 endless recursions during template class instantiation. ANSI/ISO C++
1426 conforming programs must not rely on a maximum depth greater than 17.
1428 @item -fuse-cxa-atexit
1429 @opindex fuse-cxa-atexit
1430 Register destructors for objects with static storage duration with the
1431 @code{__cxa_atexit} function rather than the @code{atexit} function.
1432 This option is required for fully standards-compliant handling of static
1433 destructors, but will only work if your C library supports
1434 @code{__cxa_atexit}.
1438 Do not use weak symbol support, even if it is provided by the linker.
1439 By default, G++ will use weak symbols if they are available. This
1440 option exists only for testing, and should not be used by end-users;
1441 it will result in inferior code and has no benefits. This option may
1442 be removed in a future release of G++.
1446 Do not search for header files in the standard directories specific to
1447 C++, but do still search the other standard directories. (This option
1448 is used when building the C++ library.)
1451 In addition, these optimization, warning, and code generation options
1452 have meanings only for C++ programs:
1455 @item -fno-default-inline
1456 @opindex fno-default-inline
1457 Do not assume @samp{inline} for functions defined inside a class scope.
1458 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1459 functions will have linkage like inline functions; they just won't be
1462 @item -Wabi @r{(C++ only)}
1464 Warn when G++ generates code that is probably not compatible with the
1465 vendor-neutral C++ ABI. Although an effort has been made to warn about
1466 all such cases, there are probably some cases that are not warned about,
1467 even though G++ is generating incompatible code. There may also be
1468 cases where warnings are emitted even though the code that is generated
1471 You should rewrite your code to avoid these warnings if you are
1472 concerned about the fact that code generated by G++ may not be binary
1473 compatible with code generated by other compilers.
1475 The known incompatibilities at this point include:
1480 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1481 pack data into the same byte as a base class. For example:
1484 struct A @{ virtual void f(); int f1 : 1; @};
1485 struct B : public A @{ int f2 : 1; @};
1489 In this case, G++ will place @code{B::f2} into the same byte
1490 as@code{A::f1}; other compilers will not. You can avoid this problem
1491 by explicitly padding @code{A} so that its size is a multiple of the
1492 byte size on your platform; that will cause G++ and other compilers to
1493 layout @code{B} identically.
1496 Incorrect handling of tail-padding for virtual bases. G++ does not use
1497 tail padding when laying out virtual bases. For example:
1500 struct A @{ virtual void f(); char c1; @};
1501 struct B @{ B(); char c2; @};
1502 struct C : public A, public virtual B @{@};
1506 In this case, G++ will not place @code{B} into the tail-padding for
1507 @code{A}; other compilers will. You can avoid this problem by
1508 explicitly padding @code{A} so that its size is a multiple of its
1509 alignment (ignoring virtual base classes); that will cause G++ and other
1510 compilers to layout @code{C} identically.
1513 Incorrect handling of bit-fields with declared widths greater than that
1514 of their underlying types, when the bit-fields appear in a union. For
1518 union U @{ int i : 4096; @};
1522 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1523 union too small by the number of bits in an @code{int}.
1526 Empty classes can be placed at incorrect offsets. For example:
1536 struct C : public B, public A @{@};
1540 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1541 it should be placed at offset zero. G++ mistakenly believes that the
1542 @code{A} data member of @code{B} is already at offset zero.
1545 Names of template functions whose types involve @code{typename} or
1546 template template parameters can be mangled incorrectly.
1549 template <typename Q>
1550 void f(typename Q::X) @{@}
1552 template <template <typename> class Q>
1553 void f(typename Q<int>::X) @{@}
1557 Instantiations of these templates may be mangled incorrectly.
1561 @item -Wctor-dtor-privacy @r{(C++ only)}
1562 @opindex Wctor-dtor-privacy
1563 Warn when a class seems unusable because all the constructors or
1564 destructors in that class are private, and it has neither friends nor
1565 public static member functions.
1567 @item -Wnon-virtual-dtor @r{(C++ only)}
1568 @opindex Wnon-virtual-dtor
1569 Warn when a class appears to be polymorphic, thereby requiring a virtual
1570 destructor, yet it declares a non-virtual one.
1571 This warning is enabled by @option{-Wall}.
1573 @item -Wreorder @r{(C++ only)}
1575 @cindex reordering, warning
1576 @cindex warning for reordering of member initializers
1577 Warn when the order of member initializers given in the code does not
1578 match the order in which they must be executed. For instance:
1584 A(): j (0), i (1) @{ @}
1588 The compiler will rearrange the member initializers for @samp{i}
1589 and @samp{j} to match the declaration order of the members, emitting
1590 a warning to that effect. This warning is enabled by @option{-Wall}.
1593 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1596 @item -Weffc++ @r{(C++ only)}
1598 Warn about violations of the following style guidelines from Scott Meyers'
1599 @cite{Effective C++} book:
1603 Item 11: Define a copy constructor and an assignment operator for classes
1604 with dynamically allocated memory.
1607 Item 12: Prefer initialization to assignment in constructors.
1610 Item 14: Make destructors virtual in base classes.
1613 Item 15: Have @code{operator=} return a reference to @code{*this}.
1616 Item 23: Don't try to return a reference when you must return an object.
1620 Also warn about violations of the following style guidelines from
1621 Scott Meyers' @cite{More Effective C++} book:
1625 Item 6: Distinguish between prefix and postfix forms of increment and
1626 decrement operators.
1629 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1633 When selecting this option, be aware that the standard library
1634 headers do not obey all of these guidelines; use @samp{grep -v}
1635 to filter out those warnings.
1637 @item -Wno-deprecated @r{(C++ only)}
1638 @opindex Wno-deprecated
1639 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1641 @item -Wno-non-template-friend @r{(C++ only)}
1642 @opindex Wno-non-template-friend
1643 Disable warnings when non-templatized friend functions are declared
1644 within a template. Since the advent of explicit template specification
1645 support in G++, if the name of the friend is an unqualified-id (i.e.,
1646 @samp{friend foo(int)}), the C++ language specification demands that the
1647 friend declare or define an ordinary, nontemplate function. (Section
1648 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1649 could be interpreted as a particular specialization of a templatized
1650 function. Because this non-conforming behavior is no longer the default
1651 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1652 check existing code for potential trouble spots and is on by default.
1653 This new compiler behavior can be turned off with
1654 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1655 but disables the helpful warning.
1657 @item -Wold-style-cast @r{(C++ only)}
1658 @opindex Wold-style-cast
1659 Warn if an old-style (C-style) cast to a non-void type is used within
1660 a C++ program. The new-style casts (@samp{static_cast},
1661 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1662 unintended effects and much easier to search for.
1664 @item -Woverloaded-virtual @r{(C++ only)}
1665 @opindex Woverloaded-virtual
1666 @cindex overloaded virtual fn, warning
1667 @cindex warning for overloaded virtual fn
1668 Warn when a function declaration hides virtual functions from a
1669 base class. For example, in:
1676 struct B: public A @{
1681 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1689 will fail to compile.
1691 @item -Wno-pmf-conversions @r{(C++ only)}
1692 @opindex Wno-pmf-conversions
1693 Disable the diagnostic for converting a bound pointer to member function
1696 @item -Wsign-promo @r{(C++ only)}
1697 @opindex Wsign-promo
1698 Warn when overload resolution chooses a promotion from unsigned or
1699 enumeral type to a signed type, over a conversion to an unsigned type of
1700 the same size. Previous versions of G++ would try to preserve
1701 unsignedness, but the standard mandates the current behavior.
1703 @item -Wsynth @r{(C++ only)}
1705 @cindex warning for synthesized methods
1706 @cindex synthesized methods, warning
1707 Warn when G++'s synthesis behavior does not match that of cfront. For
1713 A& operator = (int);
1723 In this example, G++ will synthesize a default @samp{A& operator =
1724 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1727 @node Objective-C Dialect Options
1728 @section Options Controlling Objective-C Dialect
1730 @cindex compiler options, Objective-C
1731 @cindex Objective-C options, command line
1732 @cindex options, Objective-C
1733 (NOTE: This manual does not describe the Objective-C language itself. See
1734 @w{@uref{http://gcc.gnu.org/readings.html}} for references.)
1736 This section describes the command-line options that are only meaningful
1737 for Objective-C programs, but you can also use most of the GNU compiler
1738 options regardless of what language your program is in. For example,
1739 you might compile a file @code{some_class.m} like this:
1742 gcc -g -fgnu-runtime -O -c some_class.m
1746 In this example, @option{-fgnu-runtime} is an option meant only for
1747 Objective-C programs; you can use the other options with any language
1750 Here is a list of options that are @emph{only} for compiling Objective-C
1754 @item -fconstant-string-class=@var{class-name}
1755 @opindex fconstant-string-class
1756 Use @var{class-name} as the name of the class to instantiate for each
1757 literal string specified with the syntax @code{@@"@dots{}"}. The default
1758 class name is @code{NXConstantString} if the GNU runtime is being used, and
1759 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1760 @option{-fconstant-cfstrings} option, if also present, will override the
1761 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1762 to be laid out as constant CoreFoundation strings.
1765 @opindex fgnu-runtime
1766 Generate object code compatible with the standard GNU Objective-C
1767 runtime. This is the default for most types of systems.
1769 @item -fnext-runtime
1770 @opindex fnext-runtime
1771 Generate output compatible with the NeXT runtime. This is the default
1772 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1773 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1776 @item -fno-nil-receivers
1777 @opindex -fno-nil-receivers
1778 Assume that all Objective-C message dispatches (e.g.,
1779 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1780 is not @code{nil}. This allows for more efficient entry points in the runtime to be
1781 used. Currently, this option is only available in conjunction with
1782 the NeXT runtime on Mac OS X 10.3 and later.
1784 @item -fobjc-exceptions
1785 @opindex -fobjc-exceptions
1786 Enable syntactic support for structured exception handling in Objective-C,
1787 similar to what is offered by C++ and Java. Currently, this option is only
1788 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1796 @@catch (AnObjCClass *exc) @{
1803 @@catch (AnotherClass *exc) @{
1806 @@catch (id allOthers) @{
1816 The @code{@@throw} statement may appear anywhere in an Objective-C or
1817 Objective-C++ program; when used inside of a @code{@@catch} block, the
1818 @code{@@throw} may appear without an argument (as shown above), in which case
1819 the object caught by the @code{@@catch} will be rethrown.
1821 Note that only (pointers to) Objective-C objects may be thrown and
1822 caught using this scheme. When an object is thrown, it will be caught
1823 by the nearest @code{@@catch} clause capable of handling objects of that type,
1824 analogously to how @code{catch} blocks work in C++ and Java. A
1825 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1826 any and all Objective-C exceptions not caught by previous @code{@@catch}
1829 The @code{@@finally} clause, if present, will be executed upon exit from the
1830 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1831 regardless of whether any exceptions are thrown, caught or rethrown
1832 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1833 of the @code{finally} clause in Java.
1835 There are several caveats to using the new exception mechanism:
1839 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1840 idioms provided by the @code{NSException} class, the new
1841 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1842 systems, due to additional functionality needed in the (NeXT) Objective-C
1846 As mentioned above, the new exceptions do not support handling
1847 types other than Objective-C objects. Furthermore, when used from
1848 Objective-C++, the Objective-C exception model does not interoperate with C++
1849 exceptions at this time. This means you cannot @code{@@throw} an exception
1850 from Objective-C and @code{catch} it in C++, or vice versa
1851 (i.e., @code{throw @dots{} @@catch}).
1854 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1855 blocks for thread-safe execution:
1858 @@synchronized (ObjCClass *guard) @{
1863 Upon entering the @code{@@synchronized} block, a thread of execution shall
1864 first check whether a lock has been placed on the corresponding @code{guard}
1865 object by another thread. If it has, the current thread shall wait until
1866 the other thread relinquishes its lock. Once @code{guard} becomes available,
1867 the current thread will place its own lock on it, execute the code contained in
1868 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1869 making @code{guard} available to other threads).
1871 Unlike Java, Objective-C does not allow for entire methods to be marked
1872 @code{@@synchronized}. Note that throwing exceptions out of
1873 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1874 to be unlocked properly.
1876 @item -freplace-objc-classes
1877 @opindex -freplace-objc-classes
1878 Emit a special marker instructing @command{ld(1)} not to statically link in
1879 the resulting object file, and allow @command{dyld(1)} to load it in at
1880 run time instead. This is used in conjunction with the Fix-and-Continue
1881 debugging mode, where the object file in question may be recompiled and
1882 dynamically reloaded in the course of program execution, without the need
1883 to restart the program itself. Currently, Fix-and-Continue functionality
1884 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
1888 @opindex -fzero-link
1889 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1890 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1891 compile time) with static class references that get initialized at load time,
1892 which improves run-time performance. Specifying the @option{-fzero-link} flag
1893 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1894 to be retained. This is useful in Zero-Link debugging mode, since it allows
1895 for individual class implementations to be modified during program execution.
1899 Dump interface declarations for all classes seen in the source file to a
1900 file named @file{@var{sourcename}.decl}.
1903 @opindex Wno-protocol
1904 If a class is declared to implement a protocol, a warning is issued for
1905 every method in the protocol that is not implemented by the class. The
1906 default behavior is to issue a warning for every method not explicitly
1907 implemented in the class, even if a method implementation is inherited
1908 from the superclass. If you use the @code{-Wno-protocol} option, then
1909 methods inherited from the superclass are considered to be implemented,
1910 and no warning is issued for them.
1914 Warn if multiple methods of different types for the same selector are
1915 found during compilation. The check is performed on the list of methods
1916 in the final stage of compilation. Additionally, a check is performed
1917 for each selector appearing in a @code{@@selector(@dots{})}
1918 expression, and a corresponding method for that selector has been found
1919 during compilation. Because these checks scan the method table only at
1920 the end of compilation, these warnings are not produced if the final
1921 stage of compilation is not reached, for example because an error is
1922 found during compilation, or because the @code{-fsyntax-only} option is
1925 @item -Wundeclared-selector
1926 @opindex Wundeclared-selector
1927 Warn if a @code{@@selector(@dots{})} expression referring to an
1928 undeclared selector is found. A selector is considered undeclared if no
1929 method with that name has been declared before the
1930 @code{@@selector(@dots{})} expression, either explicitly in an
1931 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
1932 an @code{@@implementation} section. This option always performs its
1933 checks as soon as a @code{@@selector(@dots{})} expression is found,
1934 while @code{-Wselector} only performs its checks in the final stage of
1935 compilation. This also enforces the coding style convention
1936 that methods and selectors must be declared before being used.
1938 @item -print-objc-runtime-info
1939 @opindex -print-objc-runtime-info
1940 Generate C header describing the largest structure that is passed by
1945 @node Language Independent Options
1946 @section Options to Control Diagnostic Messages Formatting
1947 @cindex options to control diagnostics formatting
1948 @cindex diagnostic messages
1949 @cindex message formatting
1951 Traditionally, diagnostic messages have been formatted irrespective of
1952 the output device's aspect (e.g.@: its width, @dots{}). The options described
1953 below can be used to control the diagnostic messages formatting
1954 algorithm, e.g.@: how many characters per line, how often source location
1955 information should be reported. Right now, only the C++ front end can
1956 honor these options. However it is expected, in the near future, that
1957 the remaining front ends would be able to digest them correctly.
1960 @item -fmessage-length=@var{n}
1961 @opindex fmessage-length
1962 Try to format error messages so that they fit on lines of about @var{n}
1963 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1964 the front ends supported by GCC@. If @var{n} is zero, then no
1965 line-wrapping will be done; each error message will appear on a single
1968 @opindex fdiagnostics-show-location
1969 @item -fdiagnostics-show-location=once
1970 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1971 reporter to emit @emph{once} source location information; that is, in
1972 case the message is too long to fit on a single physical line and has to
1973 be wrapped, the source location won't be emitted (as prefix) again,
1974 over and over, in subsequent continuation lines. This is the default
1977 @item -fdiagnostics-show-location=every-line
1978 Only meaningful in line-wrapping mode. Instructs the diagnostic
1979 messages reporter to emit the same source location information (as
1980 prefix) for physical lines that result from the process of breaking
1981 a message which is too long to fit on a single line.
1985 @node Warning Options
1986 @section Options to Request or Suppress Warnings
1987 @cindex options to control warnings
1988 @cindex warning messages
1989 @cindex messages, warning
1990 @cindex suppressing warnings
1992 Warnings are diagnostic messages that report constructions which
1993 are not inherently erroneous but which are risky or suggest there
1994 may have been an error.
1996 You can request many specific warnings with options beginning @samp{-W},
1997 for example @option{-Wimplicit} to request warnings on implicit
1998 declarations. Each of these specific warning options also has a
1999 negative form beginning @samp{-Wno-} to turn off warnings;
2000 for example, @option{-Wno-implicit}. This manual lists only one of the
2001 two forms, whichever is not the default.
2003 The following options control the amount and kinds of warnings produced
2004 by GCC; for further, language-specific options also refer to
2005 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
2008 @cindex syntax checking
2010 @opindex fsyntax-only
2011 Check the code for syntax errors, but don't do anything beyond that.
2015 Issue all the warnings demanded by strict ISO C and ISO C++;
2016 reject all programs that use forbidden extensions, and some other
2017 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2018 version of the ISO C standard specified by any @option{-std} option used.
2020 Valid ISO C and ISO C++ programs should compile properly with or without
2021 this option (though a rare few will require @option{-ansi} or a
2022 @option{-std} option specifying the required version of ISO C)@. However,
2023 without this option, certain GNU extensions and traditional C and C++
2024 features are supported as well. With this option, they are rejected.
2026 @option{-pedantic} does not cause warning messages for use of the
2027 alternate keywords whose names begin and end with @samp{__}. Pedantic
2028 warnings are also disabled in the expression that follows
2029 @code{__extension__}. However, only system header files should use
2030 these escape routes; application programs should avoid them.
2031 @xref{Alternate Keywords}.
2033 Some users try to use @option{-pedantic} to check programs for strict ISO
2034 C conformance. They soon find that it does not do quite what they want:
2035 it finds some non-ISO practices, but not all---only those for which
2036 ISO C @emph{requires} a diagnostic, and some others for which
2037 diagnostics have been added.
2039 A feature to report any failure to conform to ISO C might be useful in
2040 some instances, but would require considerable additional work and would
2041 be quite different from @option{-pedantic}. We don't have plans to
2042 support such a feature in the near future.
2044 Where the standard specified with @option{-std} represents a GNU
2045 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2046 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2047 extended dialect is based. Warnings from @option{-pedantic} are given
2048 where they are required by the base standard. (It would not make sense
2049 for such warnings to be given only for features not in the specified GNU
2050 C dialect, since by definition the GNU dialects of C include all
2051 features the compiler supports with the given option, and there would be
2052 nothing to warn about.)
2054 @item -pedantic-errors
2055 @opindex pedantic-errors
2056 Like @option{-pedantic}, except that errors are produced rather than
2061 Inhibit all warning messages.
2065 Inhibit warning messages about the use of @samp{#import}.
2067 @item -Wchar-subscripts
2068 @opindex Wchar-subscripts
2069 Warn if an array subscript has type @code{char}. This is a common cause
2070 of error, as programmers often forget that this type is signed on some
2075 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2076 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2080 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2081 the arguments supplied have types appropriate to the format string
2082 specified, and that the conversions specified in the format string make
2083 sense. This includes standard functions, and others specified by format
2084 attributes (@pxref{Function Attributes}), in the @code{printf},
2085 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2086 not in the C standard) families.
2088 The formats are checked against the format features supported by GNU
2089 libc version 2.2. These include all ISO C90 and C99 features, as well
2090 as features from the Single Unix Specification and some BSD and GNU
2091 extensions. Other library implementations may not support all these
2092 features; GCC does not support warning about features that go beyond a
2093 particular library's limitations. However, if @option{-pedantic} is used
2094 with @option{-Wformat}, warnings will be given about format features not
2095 in the selected standard version (but not for @code{strfmon} formats,
2096 since those are not in any version of the C standard). @xref{C Dialect
2097 Options,,Options Controlling C Dialect}.
2099 Since @option{-Wformat} also checks for null format arguments for
2100 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2102 @option{-Wformat} is included in @option{-Wall}. For more control over some
2103 aspects of format checking, the options @option{-Wformat-y2k},
2104 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2105 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2106 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2109 @opindex Wformat-y2k
2110 If @option{-Wformat} is specified, also warn about @code{strftime}
2111 formats which may yield only a two-digit year.
2113 @item -Wno-format-extra-args
2114 @opindex Wno-format-extra-args
2115 If @option{-Wformat} is specified, do not warn about excess arguments to a
2116 @code{printf} or @code{scanf} format function. The C standard specifies
2117 that such arguments are ignored.
2119 Where the unused arguments lie between used arguments that are
2120 specified with @samp{$} operand number specifications, normally
2121 warnings are still given, since the implementation could not know what
2122 type to pass to @code{va_arg} to skip the unused arguments. However,
2123 in the case of @code{scanf} formats, this option will suppress the
2124 warning if the unused arguments are all pointers, since the Single
2125 Unix Specification says that such unused arguments are allowed.
2127 @item -Wno-format-zero-length
2128 @opindex Wno-format-zero-length
2129 If @option{-Wformat} is specified, do not warn about zero-length formats.
2130 The C standard specifies that zero-length formats are allowed.
2132 @item -Wformat-nonliteral
2133 @opindex Wformat-nonliteral
2134 If @option{-Wformat} is specified, also warn if the format string is not a
2135 string literal and so cannot be checked, unless the format function
2136 takes its format arguments as a @code{va_list}.
2138 @item -Wformat-security
2139 @opindex Wformat-security
2140 If @option{-Wformat} is specified, also warn about uses of format
2141 functions that represent possible security problems. At present, this
2142 warns about calls to @code{printf} and @code{scanf} functions where the
2143 format string is not a string literal and there are no format arguments,
2144 as in @code{printf (foo);}. This may be a security hole if the format
2145 string came from untrusted input and contains @samp{%n}. (This is
2146 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2147 in future warnings may be added to @option{-Wformat-security} that are not
2148 included in @option{-Wformat-nonliteral}.)
2152 Enable @option{-Wformat} plus format checks not included in
2153 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2154 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2158 Warn about passing a null pointer for arguments marked as
2159 requiring a non-null value by the @code{nonnull} function attribute.
2161 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2162 can be disabled with the @option{-Wno-nonnull} option.
2164 @item -Winit-self @r{(C, C++, and Objective-C only)}
2166 Warn about uninitialized variables which are initialized with themselves.
2167 Note this option can only be used with the @option{-Wuninitialized} option,
2168 which in turn only works with @option{-O1} and above.
2170 For example, GCC will warn about @code{i} being uninitialized in the
2171 following snippet only when @option{-Winit-self} has been specified:
2182 @item -Wimplicit-int
2183 @opindex Wimplicit-int
2184 Warn when a declaration does not specify a type.
2186 @item -Wimplicit-function-declaration
2187 @itemx -Werror-implicit-function-declaration
2188 @opindex Wimplicit-function-declaration
2189 @opindex Werror-implicit-function-declaration
2190 Give a warning (or error) whenever a function is used before being
2195 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2199 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2200 function with external linkage, returning int, taking either zero
2201 arguments, two, or three arguments of appropriate types.
2203 @item -Wmissing-braces
2204 @opindex Wmissing-braces
2205 Warn if an aggregate or union initializer is not fully bracketed. In
2206 the following example, the initializer for @samp{a} is not fully
2207 bracketed, but that for @samp{b} is fully bracketed.
2210 int a[2][2] = @{ 0, 1, 2, 3 @};
2211 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2215 @opindex Wparentheses
2216 Warn if parentheses are omitted in certain contexts, such
2217 as when there is an assignment in a context where a truth value
2218 is expected, or when operators are nested whose precedence people
2219 often get confused about.
2221 Also warn about constructions where there may be confusion to which
2222 @code{if} statement an @code{else} branch belongs. Here is an example of
2237 In C, every @code{else} branch belongs to the innermost possible @code{if}
2238 statement, which in this example is @code{if (b)}. This is often not
2239 what the programmer expected, as illustrated in the above example by
2240 indentation the programmer chose. When there is the potential for this
2241 confusion, GCC will issue a warning when this flag is specified.
2242 To eliminate the warning, add explicit braces around the innermost
2243 @code{if} statement so there is no way the @code{else} could belong to
2244 the enclosing @code{if}. The resulting code would look like this:
2260 @item -Wsequence-point
2261 @opindex Wsequence-point
2262 Warn about code that may have undefined semantics because of violations
2263 of sequence point rules in the C standard.
2265 The C standard defines the order in which expressions in a C program are
2266 evaluated in terms of @dfn{sequence points}, which represent a partial
2267 ordering between the execution of parts of the program: those executed
2268 before the sequence point, and those executed after it. These occur
2269 after the evaluation of a full expression (one which is not part of a
2270 larger expression), after the evaluation of the first operand of a
2271 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2272 function is called (but after the evaluation of its arguments and the
2273 expression denoting the called function), and in certain other places.
2274 Other than as expressed by the sequence point rules, the order of
2275 evaluation of subexpressions of an expression is not specified. All
2276 these rules describe only a partial order rather than a total order,
2277 since, for example, if two functions are called within one expression
2278 with no sequence point between them, the order in which the functions
2279 are called is not specified. However, the standards committee have
2280 ruled that function calls do not overlap.
2282 It is not specified when between sequence points modifications to the
2283 values of objects take effect. Programs whose behavior depends on this
2284 have undefined behavior; the C standard specifies that ``Between the
2285 previous and next sequence point an object shall have its stored value
2286 modified at most once by the evaluation of an expression. Furthermore,
2287 the prior value shall be read only to determine the value to be
2288 stored.''. If a program breaks these rules, the results on any
2289 particular implementation are entirely unpredictable.
2291 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2292 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2293 diagnosed by this option, and it may give an occasional false positive
2294 result, but in general it has been found fairly effective at detecting
2295 this sort of problem in programs.
2297 The present implementation of this option only works for C programs. A
2298 future implementation may also work for C++ programs.
2300 The C standard is worded confusingly, therefore there is some debate
2301 over the precise meaning of the sequence point rules in subtle cases.
2302 Links to discussions of the problem, including proposed formal
2303 definitions, may be found on our readings page, at
2304 @w{@uref{http://gcc.gnu.org/readings.html}}.
2307 @opindex Wreturn-type
2308 Warn whenever a function is defined with a return-type that defaults to
2309 @code{int}. Also warn about any @code{return} statement with no
2310 return-value in a function whose return-type is not @code{void}.
2312 For C++, a function without return type always produces a diagnostic
2313 message, even when @option{-Wno-return-type} is specified. The only
2314 exceptions are @samp{main} and functions defined in system headers.
2318 Warn whenever a @code{switch} statement has an index of enumeral type
2319 and lacks a @code{case} for one or more of the named codes of that
2320 enumeration. (The presence of a @code{default} label prevents this
2321 warning.) @code{case} labels outside the enumeration range also
2322 provoke warnings when this option is used.
2324 @item -Wswitch-default
2325 @opindex Wswitch-switch
2326 Warn whenever a @code{switch} statement does not have a @code{default}
2330 @opindex Wswitch-enum
2331 Warn whenever a @code{switch} statement has an index of enumeral type
2332 and lacks a @code{case} for one or more of the named codes of that
2333 enumeration. @code{case} labels outside the enumeration range also
2334 provoke warnings when this option is used.
2338 Warn if any trigraphs are encountered that might change the meaning of
2339 the program (trigraphs within comments are not warned about).
2341 @item -Wunused-function
2342 @opindex Wunused-function
2343 Warn whenever a static function is declared but not defined or a
2344 non\-inline static function is unused.
2346 @item -Wunused-label
2347 @opindex Wunused-label
2348 Warn whenever a label is declared but not used.
2350 To suppress this warning use the @samp{unused} attribute
2351 (@pxref{Variable Attributes}).
2353 @item -Wunused-parameter
2354 @opindex Wunused-parameter
2355 Warn whenever a function parameter is unused aside from its declaration.
2357 To suppress this warning use the @samp{unused} attribute
2358 (@pxref{Variable Attributes}).
2360 @item -Wunused-variable
2361 @opindex Wunused-variable
2362 Warn whenever a local variable or non-constant static variable is unused
2363 aside from its declaration
2365 To suppress this warning use the @samp{unused} attribute
2366 (@pxref{Variable Attributes}).
2368 @item -Wunused-value
2369 @opindex Wunused-value
2370 Warn whenever a statement computes a result that is explicitly not used.
2372 To suppress this warning cast the expression to @samp{void}.
2376 All the above @option{-Wunused} options combined.
2378 In order to get a warning about an unused function parameter, you must
2379 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2380 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2382 @item -Wuninitialized
2383 @opindex Wuninitialized
2384 Warn if an automatic variable is used without first being initialized or
2385 if a variable may be clobbered by a @code{setjmp} call.
2387 These warnings are possible only in optimizing compilation,
2388 because they require data flow information that is computed only
2389 when optimizing. If you don't specify @option{-O}, you simply won't
2392 If you want to warn about code which uses the uninitialized value of the
2393 variable in its own initializer, use the @option{-Winit-self} option.
2395 These warnings occur only for variables that are candidates for
2396 register allocation. Therefore, they do not occur for a variable that
2397 is declared @code{volatile}, or whose address is taken, or whose size
2398 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2399 structures, unions or arrays, even when they are in registers.
2401 Note that there may be no warning about a variable that is used only
2402 to compute a value that itself is never used, because such
2403 computations may be deleted by data flow analysis before the warnings
2406 These warnings are made optional because GCC is not smart
2407 enough to see all the reasons why the code might be correct
2408 despite appearing to have an error. Here is one example of how
2429 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2430 always initialized, but GCC doesn't know this. Here is
2431 another common case:
2436 if (change_y) save_y = y, y = new_y;
2438 if (change_y) y = save_y;
2443 This has no bug because @code{save_y} is used only if it is set.
2445 @cindex @code{longjmp} warnings
2446 This option also warns when a non-volatile automatic variable might be
2447 changed by a call to @code{longjmp}. These warnings as well are possible
2448 only in optimizing compilation.
2450 The compiler sees only the calls to @code{setjmp}. It cannot know
2451 where @code{longjmp} will be called; in fact, a signal handler could
2452 call it at any point in the code. As a result, you may get a warning
2453 even when there is in fact no problem because @code{longjmp} cannot
2454 in fact be called at the place which would cause a problem.
2456 Some spurious warnings can be avoided if you declare all the functions
2457 you use that never return as @code{noreturn}. @xref{Function
2460 @item -Wunknown-pragmas
2461 @opindex Wunknown-pragmas
2462 @cindex warning for unknown pragmas
2463 @cindex unknown pragmas, warning
2464 @cindex pragmas, warning of unknown
2465 Warn when a #pragma directive is encountered which is not understood by
2466 GCC@. If this command line option is used, warnings will even be issued
2467 for unknown pragmas in system header files. This is not the case if
2468 the warnings were only enabled by the @option{-Wall} command line option.
2470 @item -Wstrict-aliasing
2471 @opindex Wstrict-aliasing
2472 This option is only active when @option{-fstrict-aliasing} is active.
2473 It warns about code which might break the strict aliasing rules that the
2474 compiler is using for optimization. The warning does not catch all
2475 cases, but does attempt to catch the more common pitfalls. It is
2476 included in @option{-Wall}.
2480 All of the above @samp{-W} options combined. This enables all the
2481 warnings about constructions that some users consider questionable, and
2482 that are easy to avoid (or modify to prevent the warning), even in
2483 conjunction with macros. This also enables some language-specific
2484 warnings described in @ref{C++ Dialect Options} and
2485 @ref{Objective-C Dialect Options}.
2488 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2489 Some of them warn about constructions that users generally do not
2490 consider questionable, but which occasionally you might wish to check
2491 for; others warn about constructions that are necessary or hard to avoid
2492 in some cases, and there is no simple way to modify the code to suppress
2499 (This option used to be called @option{-W}. The older name is still
2500 supported, but the newer name is more descriptive.) Print extra warning
2501 messages for these events:
2505 A function can return either with or without a value. (Falling
2506 off the end of the function body is considered returning without
2507 a value.) For example, this function would evoke such a
2521 An expression-statement or the left-hand side of a comma expression
2522 contains no side effects.
2523 To suppress the warning, cast the unused expression to void.
2524 For example, an expression such as @samp{x[i,j]} will cause a warning,
2525 but @samp{x[(void)i,j]} will not.
2528 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2531 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2532 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2533 that of ordinary mathematical notation.
2536 Storage-class specifiers like @code{static} are not the first things in
2537 a declaration. According to the C Standard, this usage is obsolescent.
2540 The return type of a function has a type qualifier such as @code{const}.
2541 Such a type qualifier has no effect, since the value returned by a
2542 function is not an lvalue. (But don't warn about the GNU extension of
2543 @code{volatile void} return types. That extension will be warned about
2544 if @option{-pedantic} is specified.)
2547 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2551 A comparison between signed and unsigned values could produce an
2552 incorrect result when the signed value is converted to unsigned.
2553 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2556 An aggregate has an initializer which does not initialize all members.
2557 For example, the following code would cause such a warning, because
2558 @code{x.h} would be implicitly initialized to zero:
2561 struct s @{ int f, g, h; @};
2562 struct s x = @{ 3, 4 @};
2566 A function parameter is declared without a type specifier in K&R-style
2574 An empty body occurs in an @samp{if} or @samp{else} statement.
2577 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2578 @samp{>}, or @samp{>=}.
2581 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2584 Any of several floating-point events that often indicate errors, such as
2585 overflow, underflow, loss of precision, etc.
2587 @item @r{(C++ only)}
2588 An enumerator and a non-enumerator both appear in a conditional expression.
2590 @item @r{(C++ only)}
2591 A non-static reference or non-static @samp{const} member appears in a
2592 class without constructors.
2594 @item @r{(C++ only)}
2595 Ambiguous virtual bases.
2597 @item @r{(C++ only)}
2598 Subscripting an array which has been declared @samp{register}.
2600 @item @r{(C++ only)}
2601 Taking the address of a variable which has been declared @samp{register}.
2603 @item @r{(C++ only)}
2604 A base class is not initialized in a derived class' copy constructor.
2607 @item -Wno-div-by-zero
2608 @opindex Wno-div-by-zero
2609 @opindex Wdiv-by-zero
2610 Do not warn about compile-time integer division by zero. Floating point
2611 division by zero is not warned about, as it can be a legitimate way of
2612 obtaining infinities and NaNs.
2614 @item -Wsystem-headers
2615 @opindex Wsystem-headers
2616 @cindex warnings from system headers
2617 @cindex system headers, warnings from
2618 Print warning messages for constructs found in system header files.
2619 Warnings from system headers are normally suppressed, on the assumption
2620 that they usually do not indicate real problems and would only make the
2621 compiler output harder to read. Using this command line option tells
2622 GCC to emit warnings from system headers as if they occurred in user
2623 code. However, note that using @option{-Wall} in conjunction with this
2624 option will @emph{not} warn about unknown pragmas in system
2625 headers---for that, @option{-Wunknown-pragmas} must also be used.
2628 @opindex Wfloat-equal
2629 Warn if floating point values are used in equality comparisons.
2631 The idea behind this is that sometimes it is convenient (for the
2632 programmer) to consider floating-point values as approximations to
2633 infinitely precise real numbers. If you are doing this, then you need
2634 to compute (by analyzing the code, or in some other way) the maximum or
2635 likely maximum error that the computation introduces, and allow for it
2636 when performing comparisons (and when producing output, but that's a
2637 different problem). In particular, instead of testing for equality, you
2638 would check to see whether the two values have ranges that overlap; and
2639 this is done with the relational operators, so equality comparisons are
2642 @item -Wtraditional @r{(C only)}
2643 @opindex Wtraditional
2644 Warn about certain constructs that behave differently in traditional and
2645 ISO C@. Also warn about ISO C constructs that have no traditional C
2646 equivalent, and/or problematic constructs which should be avoided.
2650 Macro parameters that appear within string literals in the macro body.
2651 In traditional C macro replacement takes place within string literals,
2652 but does not in ISO C@.
2655 In traditional C, some preprocessor directives did not exist.
2656 Traditional preprocessors would only consider a line to be a directive
2657 if the @samp{#} appeared in column 1 on the line. Therefore
2658 @option{-Wtraditional} warns about directives that traditional C
2659 understands but would ignore because the @samp{#} does not appear as the
2660 first character on the line. It also suggests you hide directives like
2661 @samp{#pragma} not understood by traditional C by indenting them. Some
2662 traditional implementations would not recognize @samp{#elif}, so it
2663 suggests avoiding it altogether.
2666 A function-like macro that appears without arguments.
2669 The unary plus operator.
2672 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2673 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2674 constants.) Note, these suffixes appear in macros defined in the system
2675 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2676 Use of these macros in user code might normally lead to spurious
2677 warnings, however gcc's integrated preprocessor has enough context to
2678 avoid warning in these cases.
2681 A function declared external in one block and then used after the end of
2685 A @code{switch} statement has an operand of type @code{long}.
2688 A non-@code{static} function declaration follows a @code{static} one.
2689 This construct is not accepted by some traditional C compilers.
2692 The ISO type of an integer constant has a different width or
2693 signedness from its traditional type. This warning is only issued if
2694 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2695 typically represent bit patterns, are not warned about.
2698 Usage of ISO string concatenation is detected.
2701 Initialization of automatic aggregates.
2704 Identifier conflicts with labels. Traditional C lacks a separate
2705 namespace for labels.
2708 Initialization of unions. If the initializer is zero, the warning is
2709 omitted. This is done under the assumption that the zero initializer in
2710 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2711 initializer warnings and relies on default initialization to zero in the
2715 Conversions by prototypes between fixed/floating point values and vice
2716 versa. The absence of these prototypes when compiling with traditional
2717 C would cause serious problems. This is a subset of the possible
2718 conversion warnings, for the full set use @option{-Wconversion}.
2721 Use of ISO C style function definitions. This warning intentionally is
2722 @emph{not} issued for prototype declarations or variadic functions
2723 because these ISO C features will appear in your code when using
2724 libiberty's traditional C compatibility macros, @code{PARAMS} and
2725 @code{VPARAMS}. This warning is also bypassed for nested functions
2726 because that feature is already a gcc extension and thus not relevant to
2727 traditional C compatibility.
2730 @item -Wdeclaration-after-statement @r{(C only)}
2731 @opindex Wdeclaration-after-statement
2732 Warn when a declaration is found after a statement in a block. This
2733 construct, known from C++, was introduced with ISO C99 and is by default
2734 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2735 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2739 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2741 @item -Wendif-labels
2742 @opindex Wendif-labels
2743 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2747 Warn whenever a local variable shadows another local variable, parameter or
2748 global variable or whenever a built-in function is shadowed.
2750 @item -Wlarger-than-@var{len}
2751 @opindex Wlarger-than
2752 Warn whenever an object of larger than @var{len} bytes is defined.
2754 @item -Wpointer-arith
2755 @opindex Wpointer-arith
2756 Warn about anything that depends on the ``size of'' a function type or
2757 of @code{void}. GNU C assigns these types a size of 1, for
2758 convenience in calculations with @code{void *} pointers and pointers
2761 @item -Wbad-function-cast @r{(C only)}
2762 @opindex Wbad-function-cast
2763 Warn whenever a function call is cast to a non-matching type.
2764 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2768 Warn whenever a pointer is cast so as to remove a type qualifier from
2769 the target type. For example, warn if a @code{const char *} is cast
2770 to an ordinary @code{char *}.
2773 @opindex Wcast-align
2774 Warn whenever a pointer is cast such that the required alignment of the
2775 target is increased. For example, warn if a @code{char *} is cast to
2776 an @code{int *} on machines where integers can only be accessed at
2777 two- or four-byte boundaries.
2779 @item -Wwrite-strings
2780 @opindex Wwrite-strings
2781 When compiling C, give string constants the type @code{const
2782 char[@var{length}]} so that
2783 copying the address of one into a non-@code{const} @code{char *}
2784 pointer will get a warning; when compiling C++, warn about the
2785 deprecated conversion from string constants to @code{char *}.
2786 These warnings will help you find at
2787 compile time code that can try to write into a string constant, but
2788 only if you have been very careful about using @code{const} in
2789 declarations and prototypes. Otherwise, it will just be a nuisance;
2790 this is why we did not make @option{-Wall} request these warnings.
2793 @opindex Wconversion
2794 Warn if a prototype causes a type conversion that is different from what
2795 would happen to the same argument in the absence of a prototype. This
2796 includes conversions of fixed point to floating and vice versa, and
2797 conversions changing the width or signedness of a fixed point argument
2798 except when the same as the default promotion.
2800 Also, warn if a negative integer constant expression is implicitly
2801 converted to an unsigned type. For example, warn about the assignment
2802 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2803 casts like @code{(unsigned) -1}.
2805 @item -Wsign-compare
2806 @opindex Wsign-compare
2807 @cindex warning for comparison of signed and unsigned values
2808 @cindex comparison of signed and unsigned values, warning
2809 @cindex signed and unsigned values, comparison warning
2810 Warn when a comparison between signed and unsigned values could produce
2811 an incorrect result when the signed value is converted to unsigned.
2812 This warning is also enabled by @option{-Wextra}; to get the other warnings
2813 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2815 @item -Waggregate-return
2816 @opindex Waggregate-return
2817 Warn if any functions that return structures or unions are defined or
2818 called. (In languages where you can return an array, this also elicits
2821 @item -Wstrict-prototypes @r{(C only)}
2822 @opindex Wstrict-prototypes
2823 Warn if a function is declared or defined without specifying the
2824 argument types. (An old-style function definition is permitted without
2825 a warning if preceded by a declaration which specifies the argument
2828 @item -Wold-style-definition @r{(C only)}
2829 @opindex Wold-style-definition
2830 Warn if an old-style function definition is used. A warning is given
2831 even if there is a previous prototype.
2833 @item -Wmissing-prototypes @r{(C only)}
2834 @opindex Wmissing-prototypes
2835 Warn if a global function is defined without a previous prototype
2836 declaration. This warning is issued even if the definition itself
2837 provides a prototype. The aim is to detect global functions that fail
2838 to be declared in header files.
2840 @item -Wmissing-declarations @r{(C only)}
2841 @opindex Wmissing-declarations
2842 Warn if a global function is defined without a previous declaration.
2843 Do so even if the definition itself provides a prototype.
2844 Use this option to detect global functions that are not declared in
2847 @item -Wmissing-noreturn
2848 @opindex Wmissing-noreturn
2849 Warn about functions which might be candidates for attribute @code{noreturn}.
2850 Note these are only possible candidates, not absolute ones. Care should
2851 be taken to manually verify functions actually do not ever return before
2852 adding the @code{noreturn} attribute, otherwise subtle code generation
2853 bugs could be introduced. You will not get a warning for @code{main} in
2854 hosted C environments.
2856 @item -Wmissing-format-attribute
2857 @opindex Wmissing-format-attribute
2859 If @option{-Wformat} is enabled, also warn about functions which might be
2860 candidates for @code{format} attributes. Note these are only possible
2861 candidates, not absolute ones. GCC will guess that @code{format}
2862 attributes might be appropriate for any function that calls a function
2863 like @code{vprintf} or @code{vscanf}, but this might not always be the
2864 case, and some functions for which @code{format} attributes are
2865 appropriate may not be detected. This option has no effect unless
2866 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2868 @item -Wno-multichar
2869 @opindex Wno-multichar
2871 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2872 Usually they indicate a typo in the user's code, as they have
2873 implementation-defined values, and should not be used in portable code.
2875 @item -Wno-deprecated-declarations
2876 @opindex Wno-deprecated-declarations
2877 Do not warn about uses of functions, variables, and types marked as
2878 deprecated by using the @code{deprecated} attribute.
2879 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2880 @pxref{Type Attributes}.)
2884 Warn if a structure is given the packed attribute, but the packed
2885 attribute has no effect on the layout or size of the structure.
2886 Such structures may be mis-aligned for little benefit. For
2887 instance, in this code, the variable @code{f.x} in @code{struct bar}
2888 will be misaligned even though @code{struct bar} does not itself
2889 have the packed attribute:
2896 @} __attribute__((packed));
2906 Warn if padding is included in a structure, either to align an element
2907 of the structure or to align the whole structure. Sometimes when this
2908 happens it is possible to rearrange the fields of the structure to
2909 reduce the padding and so make the structure smaller.
2911 @item -Wredundant-decls
2912 @opindex Wredundant-decls
2913 Warn if anything is declared more than once in the same scope, even in
2914 cases where multiple declaration is valid and changes nothing.
2916 @item -Wnested-externs @r{(C only)}
2917 @opindex Wnested-externs
2918 Warn if an @code{extern} declaration is encountered within a function.
2920 @item -Wunreachable-code
2921 @opindex Wunreachable-code
2922 Warn if the compiler detects that code will never be executed.
2924 This option is intended to warn when the compiler detects that at
2925 least a whole line of source code will never be executed, because
2926 some condition is never satisfied or because it is after a
2927 procedure that never returns.
2929 It is possible for this option to produce a warning even though there
2930 are circumstances under which part of the affected line can be executed,
2931 so care should be taken when removing apparently-unreachable code.
2933 For instance, when a function is inlined, a warning may mean that the
2934 line is unreachable in only one inlined copy of the function.
2936 This option is not made part of @option{-Wall} because in a debugging
2937 version of a program there is often substantial code which checks
2938 correct functioning of the program and is, hopefully, unreachable
2939 because the program does work. Another common use of unreachable
2940 code is to provide behavior which is selectable at compile-time.
2944 Warn if a function can not be inlined and it was declared as inline.
2945 Even with this option, the compiler will not warn about failures to
2946 inline functions declared in system headers.
2948 The compiler uses a variety of heuristics to determine whether or not
2949 to inline a function. For example, the compiler takes into account
2950 the size of the function being inlined and the the amount of inlining
2951 that has already been done in the current function. Therefore,
2952 seemingly insignificant changes in the source program can cause the
2953 warnings produced by @option{-Winline} to appear or disappear.
2955 @item -Wno-invalid-offsetof @r{(C++ only)}
2956 @opindex Wno-invalid-offsetof
2957 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
2958 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
2959 to a non-POD type is undefined. In existing C++ implementations,
2960 however, @samp{offsetof} typically gives meaningful results even when
2961 applied to certain kinds of non-POD types. (Such as a simple
2962 @samp{struct} that fails to be a POD type only by virtue of having a
2963 constructor.) This flag is for users who are aware that they are
2964 writing nonportable code and who have deliberately chosen to ignore the
2967 The restrictions on @samp{offsetof} may be relaxed in a future version
2968 of the C++ standard.
2971 @opindex Winvalid-pch
2972 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
2973 the search path but can't be used.
2977 @opindex Wno-long-long
2978 Warn if @samp{long long} type is used. This is default. To inhibit
2979 the warning messages, use @option{-Wno-long-long}. Flags
2980 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2981 only when @option{-pedantic} flag is used.
2983 @item -Wdisabled-optimization
2984 @opindex Wdisabled-optimization
2985 Warn if a requested optimization pass is disabled. This warning does
2986 not generally indicate that there is anything wrong with your code; it
2987 merely indicates that GCC's optimizers were unable to handle the code
2988 effectively. Often, the problem is that your code is too big or too
2989 complex; GCC will refuse to optimize programs when the optimization
2990 itself is likely to take inordinate amounts of time.
2994 Make all warnings into errors.
2997 @node Debugging Options
2998 @section Options for Debugging Your Program or GCC
2999 @cindex options, debugging
3000 @cindex debugging information options
3002 GCC has various special options that are used for debugging
3003 either your program or GCC:
3008 Produce debugging information in the operating system's native format
3009 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
3012 On most systems that use stabs format, @option{-g} enables use of extra
3013 debugging information that only GDB can use; this extra information
3014 makes debugging work better in GDB but will probably make other debuggers
3016 refuse to read the program. If you want to control for certain whether
3017 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3018 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3020 Unlike most other C compilers, GCC allows you to use @option{-g} with
3021 @option{-O}. The shortcuts taken by optimized code may occasionally
3022 produce surprising results: some variables you declared may not exist
3023 at all; flow of control may briefly move where you did not expect it;
3024 some statements may not be executed because they compute constant
3025 results or their values were already at hand; some statements may
3026 execute in different places because they were moved out of loops.
3028 Nevertheless it proves possible to debug optimized output. This makes
3029 it reasonable to use the optimizer for programs that might have bugs.
3031 The following options are useful when GCC is generated with the
3032 capability for more than one debugging format.
3036 Produce debugging information for use by GDB@. This means to use the
3037 most expressive format available (DWARF 2, stabs, or the native format
3038 if neither of those are supported), including GDB extensions if at all
3043 Produce debugging information in stabs format (if that is supported),
3044 without GDB extensions. This is the format used by DBX on most BSD
3045 systems. On MIPS, Alpha and System V Release 4 systems this option
3046 produces stabs debugging output which is not understood by DBX or SDB@.
3047 On System V Release 4 systems this option requires the GNU assembler.
3049 @item -feliminate-unused-debug-symbols
3050 @opindex feliminate-unused-debug-symbols
3051 Produce debugging information in stabs format (if that is supported),
3052 for only symbols that are actually used.
3056 Produce debugging information in stabs format (if that is supported),
3057 using GNU extensions understood only by the GNU debugger (GDB)@. The
3058 use of these extensions is likely to make other debuggers crash or
3059 refuse to read the program.
3063 Produce debugging information in COFF format (if that is supported).
3064 This is the format used by SDB on most System V systems prior to
3069 Produce debugging information in XCOFF format (if that is supported).
3070 This is the format used by the DBX debugger on IBM RS/6000 systems.
3074 Produce debugging information in XCOFF format (if that is supported),
3075 using GNU extensions understood only by the GNU debugger (GDB)@. The
3076 use of these extensions is likely to make other debuggers crash or
3077 refuse to read the program, and may cause assemblers other than the GNU
3078 assembler (GAS) to fail with an error.
3082 Produce debugging information in DWARF version 2 format (if that is
3083 supported). This is the format used by DBX on IRIX 6.
3087 Produce debugging information in VMS debug format (if that is
3088 supported). This is the format used by DEBUG on VMS systems.
3091 @itemx -ggdb@var{level}
3092 @itemx -gstabs@var{level}
3093 @itemx -gcoff@var{level}
3094 @itemx -gxcoff@var{level}
3095 @itemx -gvms@var{level}
3096 Request debugging information and also use @var{level} to specify how
3097 much information. The default level is 2.
3099 Level 1 produces minimal information, enough for making backtraces in
3100 parts of the program that you don't plan to debug. This includes
3101 descriptions of functions and external variables, but no information
3102 about local variables and no line numbers.
3104 Level 3 includes extra information, such as all the macro definitions
3105 present in the program. Some debuggers support macro expansion when
3106 you use @option{-g3}.
3108 Note that in order to avoid confusion between DWARF1 debug level 2,
3109 and DWARF2 @option{-gdwarf-2} does not accept a concatenated debug
3110 level. Instead use an additional @option{-g@var{level}} option to
3111 change the debug level for DWARF2.
3113 @item -feliminate-dwarf2-dups
3114 @opindex feliminate-dwarf2-dups
3115 Compress DWARF2 debugging information by eliminating duplicated
3116 information about each symbol. This option only makes sense when
3117 generating DWARF2 debugging information with @option{-gdwarf-2}.
3119 @cindex @command{prof}
3122 Generate extra code to write profile information suitable for the
3123 analysis program @command{prof}. You must use this option when compiling
3124 the source files you want data about, and you must also use it when
3127 @cindex @command{gprof}
3130 Generate extra code to write profile information suitable for the
3131 analysis program @command{gprof}. You must use this option when compiling
3132 the source files you want data about, and you must also use it when
3137 Makes the compiler print out each function name as it is compiled, and
3138 print some statistics about each pass when it finishes.
3141 @opindex ftime-report
3142 Makes the compiler print some statistics about the time consumed by each
3143 pass when it finishes.
3146 @opindex fmem-report
3147 Makes the compiler print some statistics about permanent memory
3148 allocation when it finishes.
3150 @item -fprofile-arcs
3151 @opindex fprofile-arcs
3152 Add code so that program flow @dfn{arcs} are instrumented. During
3153 execution the program records how many times each branch and call is
3154 executed and how many times it is taken or returns. When the compiled
3155 program exits it saves this data to a file called
3156 @file{@var{auxname}.gcda} for each source file. The data may be used for
3157 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3158 test coverage analysis (@option{-ftest-coverage}). Each object file's
3159 @var{auxname} is generated from the name of the output file, if
3160 explicitly specified and it is not the final executable, otherwise it is
3161 the basename of the source file. In both cases any suffix is removed
3162 (e.g. @file{foo.gcda} for input file @file{dir/foo.c}, or
3163 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3168 Compile the source files with @option{-fprofile-arcs} plus optimization
3169 and code generation options. For test coverage analysis, use the
3170 additional @option{-ftest-coverage} option. You do not need to profile
3171 every source file in a program.
3174 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3175 (the latter implies the former).
3178 Run the program on a representative workload to generate the arc profile
3179 information. This may be repeated any number of times. You can run
3180 concurrent instances of your program, and provided that the file system
3181 supports locking, the data files will be correctly updated. Also
3182 @code{fork} calls are detected and correctly handled (double counting
3186 For profile-directed optimizations, compile the source files again with
3187 the same optimization and code generation options plus
3188 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3189 Control Optimization}).
3192 For test coverage analysis, use @command{gcov} to produce human readable
3193 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3194 @command{gcov} documentation for further information.
3198 With @option{-fprofile-arcs}, for each function of your program GCC
3199 creates a program flow graph, then finds a spanning tree for the graph.
3200 Only arcs that are not on the spanning tree have to be instrumented: the
3201 compiler adds code to count the number of times that these arcs are
3202 executed. When an arc is the only exit or only entrance to a block, the
3203 instrumentation code can be added to the block; otherwise, a new basic
3204 block must be created to hold the instrumentation code.
3207 @item -ftest-coverage
3208 @opindex ftest-coverage
3209 Produce a notes file that the @command{gcov} code-coverage utility
3210 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3211 show program coverage. Each source file's note file is called
3212 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3213 above for a description of @var{auxname} and instructions on how to
3214 generate test coverage data. Coverage data will match the source files
3215 more closely, if you do not optimize.
3217 @item -d@var{letters}
3219 Says to make debugging dumps during compilation at times specified by
3220 @var{letters}. This is used for debugging the compiler. The file names
3221 for most of the dumps are made by appending a pass number and a word to
3222 the @var{dumpname}. @var{dumpname} is generated from the name of the
3223 output file, if explicitly specified and it is not an executable,
3224 otherwise it is the basename of the source file. In both cases any
3225 suffix is removed (e.g. @file{foo.01.rtl} or @file{foo.02.sibling}).
3226 Here are the possible letters for use in @var{letters}, and their
3232 Annotate the assembler output with miscellaneous debugging information.
3235 Dump after computing branch probabilities, to @file{@var{file}.12.bp}.
3238 Dump after block reordering, to @file{@var{file}.31.bbro}.
3241 Dump after instruction combination, to the file @file{@var{file}.20.combine}.
3244 Dump after the first if conversion, to the file @file{@var{file}.14.ce1}.
3245 Also dump after the second if conversion, to the file @file{@var{file}.21.ce2}.
3248 Dump after branch target load optimization, to to @file{@var{file}.32.btl}.
3249 Also dump after delayed branch scheduling, to @file{@var{file}.36.dbr}.
3252 Dump all macro definitions, at the end of preprocessing, in addition to
3256 Dump after the third if conversion, to @file{@var{file}.30.ce3}.
3259 Dump after control and data flow analysis, to @file{@var{file}.11.cfg}.
3260 Also dump after life analysis, to @file{@var{file}.19.life}.
3263 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.07.addressof}.
3266 Dump after global register allocation, to @file{@var{file}.25.greg}.
3269 Dump after GCSE, to @file{@var{file}.08.gcse}.
3270 Also dump after jump bypassing and control flow optimizations, to
3271 @file{@var{file}.10.bypass}.
3274 Dump after finalization of EH handling code, to @file{@var{file}.03.eh}.
3277 Dump after sibling call optimizations, to @file{@var{file}.02.sibling}.
3280 Dump after the first jump optimization, to @file{@var{file}.04.jump}.
3283 Dump after conversion from registers to stack, to @file{@var{file}.34.stack}.
3286 Dump after local register allocation, to @file{@var{file}.24.lreg}.
3289 Dump after loop optimization passes, to @file{@var{file}.09.loop} and
3290 @file{@var{file}.16.loop2}.
3293 Dump after performing the machine dependent reorganization pass, to
3294 @file{@var{file}.35.mach}.
3297 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3300 Dump after the register move pass, to @file{@var{file}.22.regmove}.
3303 Dump after post-reload optimizations, to @file{@var{file}.26.postreload}.
3306 Dump after RTL generation, to @file{@var{file}.01.rtl}.
3309 Dump after the second scheduling pass, to @file{@var{file}.33.sched2}.
3312 Dump after CSE (including the jump optimization that sometimes follows
3313 CSE), to @file{@var{file}.06.cse}.
3316 Dump after the first scheduling pass, to @file{@var{file}.23.sched}.
3319 Dump after the second CSE pass (including the jump optimization that
3320 sometimes follows CSE), to @file{@var{file}.18.cse2}.
3323 Dump after running tracer, to @file{@var{file}.15.tracer}.
3326 Dump after null pointer elimination pass to @file{@var{file}.05.null}.
3329 Dump callgraph and unit-at-a-time optimization @file{@var{file}.00.unit}.
3332 Dump after the value profile transformations, to @file{@var{file}.13.vpt}.
3335 Dump after the second flow pass, to @file{@var{file}.27.flow2}.
3338 Dump after the peephole pass, to @file{@var{file}.28.peephole2}.
3341 Dump after constructing the web, to @file{@var{file}.17.web}.
3344 Produce all the dumps listed above.
3347 Produce a core dump whenever an error occurs.
3350 Print statistics on memory usage, at the end of the run, to
3354 Annotate the assembler output with a comment indicating which
3355 pattern and alternative was used. The length of each instruction is
3359 Dump the RTL in the assembler output as a comment before each instruction.
3360 Also turns on @option{-dp} annotation.
3363 For each of the other indicated dump files (except for
3364 @file{@var{file}.01.rtl}), dump a representation of the control flow graph
3365 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3368 Just generate RTL for a function instead of compiling it. Usually used
3372 Dump debugging information during parsing, to standard error.
3375 @item -fdump-unnumbered
3376 @opindex fdump-unnumbered
3377 When doing debugging dumps (see @option{-d} option above), suppress instruction
3378 numbers and line number note output. This makes it more feasible to
3379 use diff on debugging dumps for compiler invocations with different
3380 options, in particular with and without @option{-g}.
3382 @item -fdump-translation-unit @r{(C and C++ only)}
3383 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3384 @opindex fdump-translation-unit
3385 Dump a representation of the tree structure for the entire translation
3386 unit to a file. The file name is made by appending @file{.tu} to the
3387 source file name. If the @samp{-@var{options}} form is used, @var{options}
3388 controls the details of the dump as described for the
3389 @option{-fdump-tree} options.
3391 @item -fdump-class-hierarchy @r{(C++ only)}
3392 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3393 @opindex fdump-class-hierarchy
3394 Dump a representation of each class's hierarchy and virtual function
3395 table layout to a file. The file name is made by appending @file{.class}
3396 to the source file name. If the @samp{-@var{options}} form is used,
3397 @var{options} controls the details of the dump as described for the
3398 @option{-fdump-tree} options.
3400 @item -fdump-tree-@var{switch} @r{(C++ only)}
3401 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3403 Control the dumping at various stages of processing the intermediate
3404 language tree to a file. The file name is generated by appending a switch
3405 specific suffix to the source file name. If the @samp{-@var{options}}
3406 form is used, @var{options} is a list of @samp{-} separated options that
3407 control the details of the dump. Not all options are applicable to all
3408 dumps, those which are not meaningful will be ignored. The following
3409 options are available
3413 Print the address of each node. Usually this is not meaningful as it
3414 changes according to the environment and source file. Its primary use
3415 is for tying up a dump file with a debug environment.
3417 Inhibit dumping of members of a scope or body of a function merely
3418 because that scope has been reached. Only dump such items when they
3419 are directly reachable by some other path.
3421 Turn on all options.
3424 The following tree dumps are possible:
3427 Dump before any tree based optimization, to @file{@var{file}.original}.
3429 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3431 Dump after function inlining, to @file{@var{file}.inlined}.
3434 @item -frandom-seed=@var{string}
3435 @opindex frandom-string
3436 This option provides a seed that GCC uses when it would otherwise use
3437 random numbers. It is used to generate certain symbol names
3438 that have to be different in every compiled file. It is also used to
3439 place unique stamps in coverage data files and the object files that
3440 produce them. You can use the @option{-frandom-seed} option to produce
3441 reproducibly identical object files.
3443 The @var{string} should be different for every file you compile.
3445 @item -fsched-verbose=@var{n}
3446 @opindex fsched-verbose
3447 On targets that use instruction scheduling, this option controls the
3448 amount of debugging output the scheduler prints. This information is
3449 written to standard error, unless @option{-dS} or @option{-dR} is
3450 specified, in which case it is output to the usual dump
3451 listing file, @file{.sched} or @file{.sched2} respectively. However
3452 for @var{n} greater than nine, the output is always printed to standard
3455 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3456 same information as @option{-dRS}. For @var{n} greater than one, it
3457 also output basic block probabilities, detailed ready list information
3458 and unit/insn info. For @var{n} greater than two, it includes RTL
3459 at abort point, control-flow and regions info. And for @var{n} over
3460 four, @option{-fsched-verbose} also includes dependence info.
3464 Store the usual ``temporary'' intermediate files permanently; place them
3465 in the current directory and name them based on the source file. Thus,
3466 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3467 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3468 preprocessed @file{foo.i} output file even though the compiler now
3469 normally uses an integrated preprocessor.
3473 Report the CPU time taken by each subprocess in the compilation
3474 sequence. For C source files, this is the compiler proper and assembler
3475 (plus the linker if linking is done). The output looks like this:
3482 The first number on each line is the ``user time,'' that is time spent
3483 executing the program itself. The second number is ``system time,''
3484 time spent executing operating system routines on behalf of the program.
3485 Both numbers are in seconds.
3487 @item -print-file-name=@var{library}
3488 @opindex print-file-name
3489 Print the full absolute name of the library file @var{library} that
3490 would be used when linking---and don't do anything else. With this
3491 option, GCC does not compile or link anything; it just prints the
3494 @item -print-multi-directory
3495 @opindex print-multi-directory
3496 Print the directory name corresponding to the multilib selected by any
3497 other switches present in the command line. This directory is supposed
3498 to exist in @env{GCC_EXEC_PREFIX}.
3500 @item -print-multi-lib
3501 @opindex print-multi-lib
3502 Print the mapping from multilib directory names to compiler switches
3503 that enable them. The directory name is separated from the switches by
3504 @samp{;}, and each switch starts with an @samp{@@} instead of the
3505 @samp{-}, without spaces between multiple switches. This is supposed to
3506 ease shell-processing.
3508 @item -print-prog-name=@var{program}
3509 @opindex print-prog-name
3510 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3512 @item -print-libgcc-file-name
3513 @opindex print-libgcc-file-name
3514 Same as @option{-print-file-name=libgcc.a}.
3516 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3517 but you do want to link with @file{libgcc.a}. You can do
3520 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3523 @item -print-search-dirs
3524 @opindex print-search-dirs
3525 Print the name of the configured installation directory and a list of
3526 program and library directories gcc will search---and don't do anything else.
3528 This is useful when gcc prints the error message
3529 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3530 To resolve this you either need to put @file{cpp0} and the other compiler
3531 components where gcc expects to find them, or you can set the environment
3532 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3533 Don't forget the trailing '/'.
3534 @xref{Environment Variables}.
3537 @opindex dumpmachine
3538 Print the compiler's target machine (for example,
3539 @samp{i686-pc-linux-gnu})---and don't do anything else.
3542 @opindex dumpversion
3543 Print the compiler version (for example, @samp{3.0})---and don't do
3548 Print the compiler's built-in specs---and don't do anything else. (This
3549 is used when GCC itself is being built.) @xref{Spec Files}.
3551 @item -feliminate-unused-debug-types
3552 @opindex feliminate-unused-debug-types
3553 Normally, when producing DWARF2 output, GCC will emit debugging
3554 information for all types declared in a compilation
3555 unit, regardless of whether or not they are actually used
3556 in that compilation unit. Sometimes this is useful, such as
3557 if, in the debugger, you want to cast a value to a type that is
3558 not actually used in your program (but is declared). More often,
3559 however, this results in a significant amount of wasted space.
3560 With this option, GCC will avoid producing debug symbol output
3561 for types that are nowhere used in the source file being compiled.
3564 @node Optimize Options
3565 @section Options That Control Optimization
3566 @cindex optimize options
3567 @cindex options, optimization
3569 These options control various sorts of optimizations.
3571 Without any optimization option, the compiler's goal is to reduce the
3572 cost of compilation and to make debugging produce the expected
3573 results. Statements are independent: if you stop the program with a
3574 breakpoint between statements, you can then assign a new value to any
3575 variable or change the program counter to any other statement in the
3576 function and get exactly the results you would expect from the source
3579 Turning on optimization flags makes the compiler attempt to improve
3580 the performance and/or code size at the expense of compilation time
3581 and possibly the ability to debug the program.
3583 The compiler performs optimization based on the knowledge it has of
3584 the program. Using the @option{-funit-at-a-time} flag will allow the
3585 compiler to consider information gained from later functions in the
3586 file when compiling a function. Compiling multiple files at once to a
3587 single output file (and using @option{-funit-at-a-time}) will allow
3588 the compiler to use information gained from all of the files when
3589 compiling each of them.
3591 Not all optimizations are controlled directly by a flag. Only
3592 optimizations that have a flag are listed.
3599 Optimize. Optimizing compilation takes somewhat more time, and a lot
3600 more memory for a large function.
3602 With @option{-O}, the compiler tries to reduce code size and execution
3603 time, without performing any optimizations that take a great deal of
3606 @option{-O} turns on the following optimization flags:
3607 @gccoptlist{-fdefer-pop @gol
3608 -fmerge-constants @gol
3610 -floop-optimize @gol
3611 -fif-conversion @gol
3612 -fif-conversion2 @gol
3613 -fdelayed-branch @gol
3614 -fguess-branch-probability @gol
3617 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3618 where doing so does not interfere with debugging.
3622 Optimize even more. GCC performs nearly all supported optimizations
3623 that do not involve a space-speed tradeoff. The compiler does not
3624 perform loop unrolling or function inlining when you specify @option{-O2}.
3625 As compared to @option{-O}, this option increases both compilation time
3626 and the performance of the generated code.
3628 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3629 also turns on the following optimization flags:
3630 @gccoptlist{-fforce-mem @gol
3631 -foptimize-sibling-calls @gol
3632 -fstrength-reduce @gol
3633 -fcse-follow-jumps -fcse-skip-blocks @gol
3634 -frerun-cse-after-loop -frerun-loop-opt @gol
3635 -fgcse -fgcse-lm -fgcse-sm -fgcse-las @gol
3636 -fdelete-null-pointer-checks @gol
3637 -fexpensive-optimizations @gol
3639 -fschedule-insns -fschedule-insns2 @gol
3640 -fsched-interblock -fsched-spec @gol
3643 -freorder-blocks -freorder-functions @gol
3644 -fstrict-aliasing @gol
3645 -funit-at-a-time @gol
3646 -falign-functions -falign-jumps @gol
3647 -falign-loops -falign-labels @gol
3650 Please note the warning under @option{-fgcse} about
3651 invoking @option{-O2} on programs that use computed gotos.
3655 Optimize yet more. @option{-O3} turns on all optimizations specified by
3656 @option{-O2} and also turns on the @option{-finline-functions},
3657 @option{-fweb} and @option{-frename-registers} options.
3661 Do not optimize. This is the default.
3665 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3666 do not typically increase code size. It also performs further
3667 optimizations designed to reduce code size.
3669 @option{-Os} disables the following optimization flags:
3670 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3671 -falign-labels -freorder-blocks -fprefetch-loop-arrays}
3673 If you use multiple @option{-O} options, with or without level numbers,
3674 the last such option is the one that is effective.
3677 Options of the form @option{-f@var{flag}} specify machine-independent
3678 flags. Most flags have both positive and negative forms; the negative
3679 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3680 below, only one of the forms is listed---the one you typically will
3681 use. You can figure out the other form by either removing @samp{no-}
3684 The following options control specific optimizations. They are either
3685 activated by @option{-O} options or are related to ones that are. You
3686 can use the following flags in the rare cases when ``fine-tuning'' of
3687 optimizations to be performed is desired.
3690 @item -fno-default-inline
3691 @opindex fno-default-inline
3692 Do not make member functions inline by default merely because they are
3693 defined inside the class scope (C++ only). Otherwise, when you specify
3694 @w{@option{-O}}, member functions defined inside class scope are compiled
3695 inline by default; i.e., you don't need to add @samp{inline} in front of
3696 the member function name.
3698 @item -fno-defer-pop
3699 @opindex fno-defer-pop
3700 Always pop the arguments to each function call as soon as that function
3701 returns. For machines which must pop arguments after a function call,
3702 the compiler normally lets arguments accumulate on the stack for several
3703 function calls and pops them all at once.
3705 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3709 Force memory operands to be copied into registers before doing
3710 arithmetic on them. This produces better code by making all memory
3711 references potential common subexpressions. When they are not common
3712 subexpressions, instruction combination should eliminate the separate
3715 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3718 @opindex fforce-addr
3719 Force memory address constants to be copied into registers before
3720 doing arithmetic on them. This may produce better code just as
3721 @option{-fforce-mem} may.
3723 @item -fomit-frame-pointer
3724 @opindex fomit-frame-pointer
3725 Don't keep the frame pointer in a register for functions that
3726 don't need one. This avoids the instructions to save, set up and
3727 restore frame pointers; it also makes an extra register available
3728 in many functions. @strong{It also makes debugging impossible on
3731 On some machines, such as the VAX, this flag has no effect, because
3732 the standard calling sequence automatically handles the frame pointer
3733 and nothing is saved by pretending it doesn't exist. The
3734 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3735 whether a target machine supports this flag. @xref{Registers,,Register
3736 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3738 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3740 @item -foptimize-sibling-calls
3741 @opindex foptimize-sibling-calls
3742 Optimize sibling and tail recursive calls.
3744 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3748 Don't pay attention to the @code{inline} keyword. Normally this option
3749 is used to keep the compiler from expanding any functions inline.
3750 Note that if you are not optimizing, no functions can be expanded inline.
3752 @item -finline-functions
3753 @opindex finline-functions
3754 Integrate all simple functions into their callers. The compiler
3755 heuristically decides which functions are simple enough to be worth
3756 integrating in this way.
3758 If all calls to a given function are integrated, and the function is
3759 declared @code{static}, then the function is normally not output as
3760 assembler code in its own right.
3762 Enabled at level @option{-O3}.
3764 @item -finline-limit=@var{n}
3765 @opindex finline-limit
3766 By default, gcc limits the size of functions that can be inlined. This flag
3767 allows the control of this limit for functions that are explicitly marked as
3768 inline (i.e., marked with the inline keyword or defined within the class
3769 definition in c++). @var{n} is the size of functions that can be inlined in
3770 number of pseudo instructions (not counting parameter handling). The default
3771 value of @var{n} is 600.
3772 Increasing this value can result in more inlined code at
3773 the cost of compilation time and memory consumption. Decreasing usually makes
3774 the compilation faster and less code will be inlined (which presumably
3775 means slower programs). This option is particularly useful for programs that
3776 use inlining heavily such as those based on recursive templates with C++.
3778 Inlining is actually controlled by a number of parameters, which may be
3779 specified individually by using @option{--param @var{name}=@var{value}}.
3780 The @option{-finline-limit=@var{n}} option sets some of these parameters
3784 @item max-inline-insns-single
3785 is set to @var{n}/2.
3786 @item max-inline-insns-auto
3787 is set to @var{n}/2.
3788 @item min-inline-insns
3789 is set to 130 or @var{n}/4, whichever is smaller.
3790 @item max-inline-insns-rtl
3794 See below for a documentation of the individual
3795 parameters controlling inlining.
3797 @emph{Note:} pseudo instruction represents, in this particular context, an
3798 abstract measurement of function's size. In no way, it represents a count
3799 of assembly instructions and as such its exact meaning might change from one
3800 release to an another.
3802 @item -fkeep-inline-functions
3803 @opindex fkeep-inline-functions
3804 Even if all calls to a given function are integrated, and the function
3805 is declared @code{static}, nevertheless output a separate run-time
3806 callable version of the function. This switch does not affect
3807 @code{extern inline} functions.
3809 @item -fkeep-static-consts
3810 @opindex fkeep-static-consts
3811 Emit variables declared @code{static const} when optimization isn't turned
3812 on, even if the variables aren't referenced.
3814 GCC enables this option by default. If you want to force the compiler to
3815 check if the variable was referenced, regardless of whether or not
3816 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3818 @item -fmerge-constants
3819 Attempt to merge identical constants (string constants and floating point
3820 constants) across compilation units.
3822 This option is the default for optimized compilation if the assembler and
3823 linker support it. Use @option{-fno-merge-constants} to inhibit this
3826 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3828 @item -fmerge-all-constants
3829 Attempt to merge identical constants and identical variables.
3831 This option implies @option{-fmerge-constants}. In addition to
3832 @option{-fmerge-constants} this considers e.g. even constant initialized
3833 arrays or initialized constant variables with integral or floating point
3834 types. Languages like C or C++ require each non-automatic variable to
3835 have distinct location, so using this option will result in non-conforming
3840 Use a graph coloring register allocator. Currently this option is meant
3841 for testing, so we are interested to hear about miscompilations with
3844 @item -fno-branch-count-reg
3845 @opindex fno-branch-count-reg
3846 Do not use ``decrement and branch'' instructions on a count register,
3847 but instead generate a sequence of instructions that decrement a
3848 register, compare it against zero, then branch based upon the result.
3849 This option is only meaningful on architectures that support such
3850 instructions, which include x86, PowerPC, IA-64 and S/390.
3852 The default is @option{-fbranch-count-reg}, enabled when
3853 @option{-fstrength-reduce} is enabled.
3855 @item -fno-function-cse
3856 @opindex fno-function-cse
3857 Do not put function addresses in registers; make each instruction that
3858 calls a constant function contain the function's address explicitly.
3860 This option results in less efficient code, but some strange hacks
3861 that alter the assembler output may be confused by the optimizations
3862 performed when this option is not used.
3864 The default is @option{-ffunction-cse}
3866 @item -fno-zero-initialized-in-bss
3867 @opindex fno-zero-initialized-in-bss
3868 If the target supports a BSS section, GCC by default puts variables that
3869 are initialized to zero into BSS@. This can save space in the resulting
3872 This option turns off this behavior because some programs explicitly
3873 rely on variables going to the data section. E.g., so that the
3874 resulting executable can find the beginning of that section and/or make
3875 assumptions based on that.
3877 The default is @option{-fzero-initialized-in-bss}.
3879 @item -fstrength-reduce
3880 @opindex fstrength-reduce
3881 Perform the optimizations of loop strength reduction and
3882 elimination of iteration variables.
3884 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3886 @item -fthread-jumps
3887 @opindex fthread-jumps
3888 Perform optimizations where we check to see if a jump branches to a
3889 location where another comparison subsumed by the first is found. If
3890 so, the first branch is redirected to either the destination of the
3891 second branch or a point immediately following it, depending on whether
3892 the condition is known to be true or false.
3894 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3896 @item -fcse-follow-jumps
3897 @opindex fcse-follow-jumps
3898 In common subexpression elimination, scan through jump instructions
3899 when the target of the jump is not reached by any other path. For
3900 example, when CSE encounters an @code{if} statement with an
3901 @code{else} clause, CSE will follow the jump when the condition
3904 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3906 @item -fcse-skip-blocks
3907 @opindex fcse-skip-blocks
3908 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3909 follow jumps which conditionally skip over blocks. When CSE
3910 encounters a simple @code{if} statement with no else clause,
3911 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3912 body of the @code{if}.
3914 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3916 @item -frerun-cse-after-loop
3917 @opindex frerun-cse-after-loop
3918 Re-run common subexpression elimination after loop optimizations has been
3921 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3923 @item -frerun-loop-opt
3924 @opindex frerun-loop-opt
3925 Run the loop optimizer twice.
3927 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3931 Perform a global common subexpression elimination pass.
3932 This pass also performs global constant and copy propagation.
3934 @emph{Note:} When compiling a program using computed gotos, a GCC
3935 extension, you may get better runtime performance if you disable
3936 the global common subexpression elimination pass by adding
3937 @option{-fno-gcse} to the command line.
3939 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3943 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3944 attempt to move loads which are only killed by stores into themselves. This
3945 allows a loop containing a load/store sequence to be changed to a load outside
3946 the loop, and a copy/store within the loop.
3948 Enabled by default when gcse is enabled.
3952 When @option{-fgcse-sm} is enabled, a store motion pass is run after
3953 global common subexpression elimination. This pass will attempt to move
3954 stores out of loops. When used in conjunction with @option{-fgcse-lm},
3955 loops containing a load/store sequence can be changed to a load before
3956 the loop and a store after the loop.
3958 Enabled by default when gcse is enabled.
3962 When @option{-fgcse-las} is enabled, the global common subexpression
3963 elimination pass eliminates redundant loads that come after stores to the
3964 same memory location (both partial and full redundancies).
3966 Enabled by default when gcse is enabled.
3968 @item -floop-optimize
3969 @opindex floop-optimize
3970 Perform loop optimizations: move constant expressions out of loops, simplify
3971 exit test conditions and optionally do strength-reduction and loop unrolling as
3974 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3976 @item -fcrossjumping
3977 @opindex crossjumping
3978 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
3979 resulting code may or may not perform better than without cross-jumping.
3981 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3983 @item -fif-conversion
3984 @opindex if-conversion
3985 Attempt to transform conditional jumps into branch-less equivalents. This
3986 include use of conditional moves, min, max, set flags and abs instructions, and
3987 some tricks doable by standard arithmetics. The use of conditional execution
3988 on chips where it is available is controlled by @code{if-conversion2}.
3990 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3992 @item -fif-conversion2
3993 @opindex if-conversion2
3994 Use conditional execution (where available) to transform conditional jumps into
3995 branch-less equivalents.
3997 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3999 @item -fdelete-null-pointer-checks
4000 @opindex fdelete-null-pointer-checks
4001 Use global dataflow analysis to identify and eliminate useless checks
4002 for null pointers. The compiler assumes that dereferencing a null
4003 pointer would have halted the program. If a pointer is checked after
4004 it has already been dereferenced, it cannot be null.
4006 In some environments, this assumption is not true, and programs can
4007 safely dereference null pointers. Use
4008 @option{-fno-delete-null-pointer-checks} to disable this optimization
4009 for programs which depend on that behavior.
4011 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4013 @item -fexpensive-optimizations
4014 @opindex fexpensive-optimizations
4015 Perform a number of minor optimizations that are relatively expensive.
4017 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4019 @item -foptimize-register-move
4021 @opindex foptimize-register-move
4023 Attempt to reassign register numbers in move instructions and as
4024 operands of other simple instructions in order to maximize the amount of
4025 register tying. This is especially helpful on machines with two-operand
4028 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4031 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4033 @item -fdelayed-branch
4034 @opindex fdelayed-branch
4035 If supported for the target machine, attempt to reorder instructions
4036 to exploit instruction slots available after delayed branch
4039 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4041 @item -fschedule-insns
4042 @opindex fschedule-insns
4043 If supported for the target machine, attempt to reorder instructions to
4044 eliminate execution stalls due to required data being unavailable. This
4045 helps machines that have slow floating point or memory load instructions
4046 by allowing other instructions to be issued until the result of the load
4047 or floating point instruction is required.
4049 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4051 @item -fschedule-insns2
4052 @opindex fschedule-insns2
4053 Similar to @option{-fschedule-insns}, but requests an additional pass of
4054 instruction scheduling after register allocation has been done. This is
4055 especially useful on machines with a relatively small number of
4056 registers and where memory load instructions take more than one cycle.
4058 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4060 @item -fno-sched-interblock
4061 @opindex fno-sched-interblock
4062 Don't schedule instructions across basic blocks. This is normally
4063 enabled by default when scheduling before register allocation, i.e.@:
4064 with @option{-fschedule-insns} or at @option{-O2} or higher.
4066 @item -fno-sched-spec
4067 @opindex fno-sched-spec
4068 Don't allow speculative motion of non-load instructions. This is normally
4069 enabled by default when scheduling before register allocation, i.e.@:
4070 with @option{-fschedule-insns} or at @option{-O2} or higher.
4072 @item -fsched-spec-load
4073 @opindex fsched-spec-load
4074 Allow speculative motion of some load instructions. This only makes
4075 sense when scheduling before register allocation, i.e.@: with
4076 @option{-fschedule-insns} or at @option{-O2} or higher.
4078 @item -fsched-spec-load-dangerous
4079 @opindex fsched-spec-load-dangerous
4080 Allow speculative motion of more load instructions. This only makes
4081 sense when scheduling before register allocation, i.e.@: with
4082 @option{-fschedule-insns} or at @option{-O2} or higher.
4084 @item -fsched-stalled-insns=@var{n}
4085 @opindex fsched-stalled-insns
4086 Define how many insns (if any) can be moved prematurely from the queue
4087 of stalled insns into the ready list, during the second scheduling pass.
4089 @item -fsched-stalled-insns-dep=@var{n}
4090 @opindex fsched-stalled-insns-dep
4091 Define how many insn groups (cycles) will be examined for a dependency
4092 on a stalled insn that is candidate for premature removal from the queue
4093 of stalled insns. Has an effect only during the second scheduling pass,
4094 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4096 @item -fsched2-use-superblocks
4097 @opindex fsched2-use-superblocks
4098 When scheduling after register allocation, do use superblock scheduling
4099 algorithm. Superblock scheduling allows motion across basic block boundaries
4100 resulting on faster schedules. This option is experimental, as not all machine
4101 descriptions used by GCC model the CPU closely enough to avoid unreliable
4102 results from the algorithm.
4104 This only makes sense when scheduling after register allocation, i.e.@: with
4105 @option{-fschedule-insns2} or at @option{-O2} or higher.
4107 @item -fsched2-use-traces
4108 @opindex fsched2-use-traces
4109 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4110 allocation and additionally perform code duplication in order to increase the
4111 size of superblocks using tracer pass. See @option{-ftracer} for details on
4114 This mode should produce faster but significantly longer programs. Also
4115 without @code{-fbranch-probabilities} the traces constructed may not match the
4116 reality and hurt the performance. This only makes
4117 sense when scheduling after register allocation, i.e.@: with
4118 @option{-fschedule-insns2} or at @option{-O2} or higher.
4120 @item -fcaller-saves
4121 @opindex fcaller-saves
4122 Enable values to be allocated in registers that will be clobbered by
4123 function calls, by emitting extra instructions to save and restore the
4124 registers around such calls. Such allocation is done only when it
4125 seems to result in better code than would otherwise be produced.
4127 This option is always enabled by default on certain machines, usually
4128 those which have no call-preserved registers to use instead.
4130 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4132 @item -fmove-all-movables
4133 @opindex fmove-all-movables
4134 Forces all invariant computations in loops to be moved
4137 @item -freduce-all-givs
4138 @opindex freduce-all-givs
4139 Forces all general-induction variables in loops to be
4142 @emph{Note:} When compiling programs written in Fortran,
4143 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4144 by default when you use the optimizer.
4146 These options may generate better or worse code; results are highly
4147 dependent on the structure of loops within the source code.
4149 These two options are intended to be removed someday, once
4150 they have helped determine the efficacy of various
4151 approaches to improving loop optimizations.
4153 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
4154 know how use of these options affects
4155 the performance of your production code.
4156 We're very interested in code that runs @emph{slower}
4157 when these options are @emph{enabled}.
4160 @itemx -fno-peephole2
4161 @opindex fno-peephole
4162 @opindex fno-peephole2
4163 Disable any machine-specific peephole optimizations. The difference
4164 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4165 are implemented in the compiler; some targets use one, some use the
4166 other, a few use both.
4168 @option{-fpeephole} is enabled by default.
4169 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4171 @item -fno-guess-branch-probability
4172 @opindex fno-guess-branch-probability
4173 Do not guess branch probabilities using a randomized model.
4175 Sometimes gcc will opt to use a randomized model to guess branch
4176 probabilities, when none are available from either profiling feedback
4177 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4178 different runs of the compiler on the same program may produce different
4181 In a hard real-time system, people don't want different runs of the
4182 compiler to produce code that has different behavior; minimizing
4183 non-determinism is of paramount import. This switch allows users to
4184 reduce non-determinism, possibly at the expense of inferior
4187 The default is @option{-fguess-branch-probability} at levels
4188 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4190 @item -freorder-blocks
4191 @opindex freorder-blocks
4192 Reorder basic blocks in the compiled function in order to reduce number of
4193 taken branches and improve code locality.
4195 Enabled at levels @option{-O2}, @option{-O3}.
4197 @item -freorder-functions
4198 @opindex freorder-functions
4199 Reorder basic blocks in the compiled function in order to reduce number of
4200 taken branches and improve code locality. This is implemented by using special
4201 subsections @code{text.hot} for most frequently executed functions and
4202 @code{text.unlikely} for unlikely executed functions. Reordering is done by
4203 the linker so object file format must support named sections and linker must
4204 place them in a reasonable way.
4206 Also profile feedback must be available in to make this option effective. See
4207 @option{-fprofile-arcs} for details.
4209 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4211 @item -fstrict-aliasing
4212 @opindex fstrict-aliasing
4213 Allows the compiler to assume the strictest aliasing rules applicable to
4214 the language being compiled. For C (and C++), this activates
4215 optimizations based on the type of expressions. In particular, an
4216 object of one type is assumed never to reside at the same address as an
4217 object of a different type, unless the types are almost the same. For
4218 example, an @code{unsigned int} can alias an @code{int}, but not a
4219 @code{void*} or a @code{double}. A character type may alias any other
4222 Pay special attention to code like this:
4235 The practice of reading from a different union member than the one most
4236 recently written to (called ``type-punning'') is common. Even with
4237 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4238 is accessed through the union type. So, the code above will work as
4239 expected. However, this code might not:
4250 Every language that wishes to perform language-specific alias analysis
4251 should define a function that computes, given an @code{tree}
4252 node, an alias set for the node. Nodes in different alias sets are not
4253 allowed to alias. For an example, see the C front-end function
4254 @code{c_get_alias_set}.
4256 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4258 @item -falign-functions
4259 @itemx -falign-functions=@var{n}
4260 @opindex falign-functions
4261 Align the start of functions to the next power-of-two greater than
4262 @var{n}, skipping up to @var{n} bytes. For instance,
4263 @option{-falign-functions=32} aligns functions to the next 32-byte
4264 boundary, but @option{-falign-functions=24} would align to the next
4265 32-byte boundary only if this can be done by skipping 23 bytes or less.
4267 @option{-fno-align-functions} and @option{-falign-functions=1} are
4268 equivalent and mean that functions will not be aligned.
4270 Some assemblers only support this flag when @var{n} is a power of two;
4271 in that case, it is rounded up.
4273 If @var{n} is not specified or is zero, use a machine-dependent default.
4275 Enabled at levels @option{-O2}, @option{-O3}.
4277 @item -falign-labels
4278 @itemx -falign-labels=@var{n}
4279 @opindex falign-labels
4280 Align all branch targets to a power-of-two boundary, skipping up to
4281 @var{n} bytes like @option{-falign-functions}. This option can easily
4282 make code slower, because it must insert dummy operations for when the
4283 branch target is reached in the usual flow of the code.
4285 @option{-fno-align-labels} and @option{-falign-labels=1} are
4286 equivalent and mean that labels will not be aligned.
4288 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4289 are greater than this value, then their values are used instead.
4291 If @var{n} is not specified or is zero, use a machine-dependent default
4292 which is very likely to be @samp{1}, meaning no alignment.
4294 Enabled at levels @option{-O2}, @option{-O3}.
4297 @itemx -falign-loops=@var{n}
4298 @opindex falign-loops
4299 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4300 like @option{-falign-functions}. The hope is that the loop will be
4301 executed many times, which will make up for any execution of the dummy
4304 @option{-fno-align-loops} and @option{-falign-loops=1} are
4305 equivalent and mean that loops will not be aligned.
4307 If @var{n} is not specified or is zero, use a machine-dependent default.
4309 Enabled at levels @option{-O2}, @option{-O3}.
4312 @itemx -falign-jumps=@var{n}
4313 @opindex falign-jumps
4314 Align branch targets to a power-of-two boundary, for branch targets
4315 where the targets can only be reached by jumping, skipping up to @var{n}
4316 bytes like @option{-falign-functions}. In this case, no dummy operations
4319 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4320 equivalent and mean that loops will not be aligned.
4322 If @var{n} is not specified or is zero, use a machine-dependent default.
4324 Enabled at levels @option{-O2}, @option{-O3}.
4326 @item -frename-registers
4327 @opindex frename-registers
4328 Attempt to avoid false dependencies in scheduled code by making use
4329 of registers left over after register allocation. This optimization
4330 will most benefit processors with lots of registers. It can, however,
4331 make debugging impossible, since variables will no longer stay in
4332 a ``home register''.
4336 Constructs webs as commonly used for register allocation purposes and assign
4337 each web individual pseudo register. This allows our register allocation pass
4338 to operate on pseudos directly, but also strengthens several other optimization
4339 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
4340 however, make debugging impossible, since variables will no longer stay in a
4343 Enabled at levels @option{-O3}.
4345 @item -fno-cprop-registers
4346 @opindex fno-cprop-registers
4347 After register allocation and post-register allocation instruction splitting,
4348 we perform a copy-propagation pass to try to reduce scheduling dependencies
4349 and occasionally eliminate the copy.
4351 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4353 @item -fprofile-generate
4354 @opindex fprofile-generate
4355 Enable options usually used for instrumenting application to produce profile usefull
4356 for later recompilation profile feedback based optimization.
4358 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}
4361 @opindex fprofile-use
4362 Enable profile feedback directed optimizations, and optimizations
4363 generally profitable only with profile feedback available.
4365 The following options are enabled: @code{-fbranch-probabilities},
4366 @code{-fvpt}, @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}.
4370 The following options control compiler behavior regarding floating
4371 point arithmetic. These options trade off between speed and
4372 correctness. All must be specifically enabled.
4376 @opindex ffloat-store
4377 Do not store floating point variables in registers, and inhibit other
4378 options that might change whether a floating point value is taken from a
4381 @cindex floating point precision
4382 This option prevents undesirable excess precision on machines such as
4383 the 68000 where the floating registers (of the 68881) keep more
4384 precision than a @code{double} is supposed to have. Similarly for the
4385 x86 architecture. For most programs, the excess precision does only
4386 good, but a few programs rely on the precise definition of IEEE floating
4387 point. Use @option{-ffloat-store} for such programs, after modifying
4388 them to store all pertinent intermediate computations into variables.
4392 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4393 @option{-fno-trapping-math}, @option{-ffinite-math-only},
4394 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
4396 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4398 This option should never be turned on by any @option{-O} option since
4399 it can result in incorrect output for programs which depend on
4400 an exact implementation of IEEE or ISO rules/specifications for
4403 @item -fno-math-errno
4404 @opindex fno-math-errno
4405 Do not set ERRNO after calling math functions that are executed
4406 with a single instruction, e.g., sqrt. A program that relies on
4407 IEEE exceptions for math error handling may want to use this flag
4408 for speed while maintaining IEEE arithmetic compatibility.
4410 This option should never be turned on by any @option{-O} option since
4411 it can result in incorrect output for programs which depend on
4412 an exact implementation of IEEE or ISO rules/specifications for
4415 The default is @option{-fmath-errno}.
4417 @item -funsafe-math-optimizations
4418 @opindex funsafe-math-optimizations
4419 Allow optimizations for floating-point arithmetic that (a) assume
4420 that arguments and results are valid and (b) may violate IEEE or
4421 ANSI standards. When used at link-time, it may include libraries
4422 or startup files that change the default FPU control word or other
4423 similar optimizations.
4425 This option should never be turned on by any @option{-O} option since
4426 it can result in incorrect output for programs which depend on
4427 an exact implementation of IEEE or ISO rules/specifications for
4430 The default is @option{-fno-unsafe-math-optimizations}.
4432 @item -ffinite-math-only
4433 @opindex ffinite-math-only
4434 Allow optimizations for floating-point arithmetic that assume
4435 that arguments and results are not NaNs or +-Infs.
4437 This option should never be turned on by any @option{-O} option since
4438 it can result in incorrect output for programs which depend on
4439 an exact implementation of IEEE or ISO rules/specifications.
4441 The default is @option{-fno-finite-math-only}.
4443 @item -fno-trapping-math
4444 @opindex fno-trapping-math
4445 Compile code assuming that floating-point operations cannot generate
4446 user-visible traps. These traps include division by zero, overflow,
4447 underflow, inexact result and invalid operation. This option implies
4448 @option{-fno-signaling-nans}. Setting this option may allow faster
4449 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4451 This option should never be turned on by any @option{-O} option since
4452 it can result in incorrect output for programs which depend on
4453 an exact implementation of IEEE or ISO rules/specifications for
4456 The default is @option{-ftrapping-math}.
4458 @item -frounding-math
4459 @opindex frounding-math
4460 Disable transformations and optimizations that assume default floating
4461 point rounding behavior. This is round-to-zero for all floating point
4462 to integer conversions, and round-to-nearest for all other arithmetic
4463 truncations. This option should be specified for programs that change
4464 the FP rounding mode dynamically, or that may be executed with a
4465 non-default rounding mode. This option disables constant folding of
4466 floating point expressions at compile-time (which may be affected by
4467 rounding mode) and arithmetic transformations that are unsafe in the
4468 presence of sign-dependent rounding modes.
4470 The default is @option{-fno-rounding-math}.
4472 This option is experimental and does not currently guarantee to
4473 disable all GCC optimizations that are affected by rounding mode.
4474 Future versions of gcc may provide finer control of this setting
4475 using C99's @code{FENV_ACCESS} pragma. This command line option
4476 will be used to specify the default state for @code{FENV_ACCESS}.
4478 @item -fsignaling-nans
4479 @opindex fsignaling-nans
4480 Compile code assuming that IEEE signaling NaNs may generate user-visible
4481 traps during floating-point operations. Setting this option disables
4482 optimizations that may change the number of exceptions visible with
4483 signaling NaNs. This option implies @option{-ftrapping-math}.
4485 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4488 The default is @option{-fno-signaling-nans}.
4490 This option is experimental and does not currently guarantee to
4491 disable all GCC optimizations that affect signaling NaN behavior.
4493 @item -fsingle-precision-constant
4494 @opindex fsingle-precision-constant
4495 Treat floating point constant as single precision constant instead of
4496 implicitly converting it to double precision constant.
4501 The following options control optimizations that may improve
4502 performance, but are not enabled by any @option{-O} options. This
4503 section includes experimental options that may produce broken code.
4506 @item -fbranch-probabilities
4507 @opindex fbranch-probabilities
4508 After running a program compiled with @option{-fprofile-arcs}
4509 (@pxref{Debugging Options,, Options for Debugging Your Program or
4510 @command{gcc}}), you can compile it a second time using
4511 @option{-fbranch-probabilities}, to improve optimizations based on
4512 the number of times each branch was taken. When the program
4513 compiled with @option{-fprofile-arcs} exits it saves arc execution
4514 counts to a file called @file{@var{sourcename}.gcda} for each source
4515 file The information in this data file is very dependent on the
4516 structure of the generated code, so you must use the same source code
4517 and the same optimization options for both compilations.
4519 With @option{-fbranch-probabilities}, GCC puts a
4520 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4521 These can be used to improve optimization. Currently, they are only
4522 used in one place: in @file{reorg.c}, instead of guessing which path a
4523 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4524 exactly determine which path is taken more often.
4526 @item -fprofile-values
4527 @opindex fprofile-values
4528 If combined with @option{-fprofile-arcs}, it adds code so that some
4529 data about values of expressions in the program is gathered.
4531 With @option{-fbranch-probabilities}, it reads back the data gathered
4532 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
4533 notes to instructions for their later usage in optimizations.
4537 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
4538 a code to gather information about values of expressions.
4540 With @option{-fbranch-probabilities}, it reads back the data gathered
4541 and actually performs the optimizations based on them.
4542 Currently the optimizations include specialization of division operation
4543 using the knowledge about the value of the denominator.
4547 Use a graph coloring register allocator. Currently this option is meant
4548 for testing, so we are interested to hear about miscompilations with
4553 Perform tail duplication to enlarge superblock size. This transformation
4554 simplifies the control flow of the function allowing other optimizations to do
4557 @item -funit-at-a-time
4558 @opindex funit-at-a-time
4559 Parse the whole compilation unit before starting to produce code.
4560 This allows some extra optimizations to take place but consumes more
4563 @item -funroll-loops
4564 @opindex funroll-loops
4565 Unroll loops whose number of iterations can be determined at compile time or
4566 upon entry to the loop. @option{-funroll-loops} implies
4567 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4568 (i.e. complete removal of loops with small constant number of iterations).
4569 This option makes code larger, and may or may not make it run faster.
4571 @item -funroll-all-loops
4572 @opindex funroll-all-loops
4573 Unroll all loops, even if their number of iterations is uncertain when
4574 the loop is entered. This usually makes programs run more slowly.
4575 @option{-funroll-all-loops} implies the same options as
4576 @option{-funroll-loops}.
4579 @opindex fpeel-loops
4580 Peels the loops for that there is enough information that they do not
4581 roll much (from profile feedback). It also turns on complete loop peeling
4582 (i.e. complete removal of loops with small constant number of iterations).
4584 @item -funswitch-loops
4585 @opindex funswitch-loops
4586 Move branches with loop invariant conditions out of the loop, with duplicates
4587 of the loop on both branches (modified according to result of the condition).
4589 @item -fold-unroll-loops
4590 @opindex fold-unroll-loops
4591 Unroll loops whose number of iterations can be determined at compile
4592 time or upon entry to the loop, using the old loop unroller whose loop
4593 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4594 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4595 option makes code larger, and may or may not make it run faster.
4597 @item -fold-unroll-all-loops
4598 @opindex fold-unroll-all-loops
4599 Unroll all loops, even if their number of iterations is uncertain when
4600 the loop is entered. This is done using the old loop unroller whose loop
4601 recognition is based on notes from frontend. This usually makes programs run more slowly.
4602 @option{-fold-unroll-all-loops} implies the same options as
4603 @option{-fold-unroll-loops}.
4605 @item -funswitch-loops
4606 @opindex funswitch-loops
4607 Move branches with loop invariant conditions out of the loop, with duplicates
4608 of the loop on both branches (modified according to result of the condition).
4610 @item -funswitch-loops
4611 @opindex funswitch-loops
4612 Move branches with loop invariant conditions out of the loop, with duplicates
4613 of the loop on both branches (modified according to result of the condition).
4615 @item -fprefetch-loop-arrays
4616 @opindex fprefetch-loop-arrays
4617 If supported by the target machine, generate instructions to prefetch
4618 memory to improve the performance of loops that access large arrays.
4620 Disabled at level @option{-Os}.
4622 @item -ffunction-sections
4623 @itemx -fdata-sections
4624 @opindex ffunction-sections
4625 @opindex fdata-sections
4626 Place each function or data item into its own section in the output
4627 file if the target supports arbitrary sections. The name of the
4628 function or the name of the data item determines the section's name
4631 Use these options on systems where the linker can perform optimizations
4632 to improve locality of reference in the instruction space. Most systems
4633 using the ELF object format and SPARC processors running Solaris 2 have
4634 linkers with such optimizations. AIX may have these optimizations in
4637 Only use these options when there are significant benefits from doing
4638 so. When you specify these options, the assembler and linker will
4639 create larger object and executable files and will also be slower.
4640 You will not be able to use @code{gprof} on all systems if you
4641 specify this option and you may have problems with debugging if
4642 you specify both this option and @option{-g}.
4644 @item -fbranch-target-load-optimize
4645 @opindex fbranch-target-load-optimize
4646 Perform branch target register load optimization before prologue / epilogue
4648 The use of target registers can typically be exposed only during reload,
4649 thus hoisting loads out of loops and doing inter-block scheduling needs
4650 a separate optimization pass.
4652 @item -fbranch-target-load-optimize2
4653 @opindex fbranch-target-load-optimize2
4654 Perform branch target register load optimization after prologue / epilogue
4657 @item -fbtr-bb-exclusive
4658 @opindex fbtr-bb-exclusive
4659 WHen performing branch target register load optimization, don't reuse
4660 branch target registers in within any basic block.
4662 @item --param @var{name}=@var{value}
4664 In some places, GCC uses various constants to control the amount of
4665 optimization that is done. For example, GCC will not inline functions
4666 that contain more that a certain number of instructions. You can
4667 control some of these constants on the command-line using the
4668 @option{--param} option.
4670 The names of specific parameters, and the meaning of the values, are
4671 tied to the internals of the compiler, and are subject to change
4672 without notice in future releases.
4674 In each case, the @var{value} is an integer. The allowable choices for
4675 @var{name} are given in the following table:
4678 @item max-crossjump-edges
4679 The maximum number of incoming edges to consider for crossjumping.
4680 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
4681 the number of edges incoming to each block. Increasing values mean
4682 more aggressive optimization, making the compile time increase with
4683 probably small improvement in executable size.
4685 @item max-delay-slot-insn-search
4686 The maximum number of instructions to consider when looking for an
4687 instruction to fill a delay slot. If more than this arbitrary number of
4688 instructions is searched, the time savings from filling the delay slot
4689 will be minimal so stop searching. Increasing values mean more
4690 aggressive optimization, making the compile time increase with probably
4691 small improvement in executable run time.
4693 @item max-delay-slot-live-search
4694 When trying to fill delay slots, the maximum number of instructions to
4695 consider when searching for a block with valid live register
4696 information. Increasing this arbitrarily chosen value means more
4697 aggressive optimization, increasing the compile time. This parameter
4698 should be removed when the delay slot code is rewritten to maintain the
4701 @item max-gcse-memory
4702 The approximate maximum amount of memory that will be allocated in
4703 order to perform the global common subexpression elimination
4704 optimization. If more memory than specified is required, the
4705 optimization will not be done.
4707 @item max-gcse-passes
4708 The maximum number of passes of GCSE to run.
4710 @item max-pending-list-length
4711 The maximum number of pending dependencies scheduling will allow
4712 before flushing the current state and starting over. Large functions
4713 with few branches or calls can create excessively large lists which
4714 needlessly consume memory and resources.
4716 @item max-inline-insns-single
4717 Several parameters control the tree inliner used in gcc.
4718 This number sets the maximum number of instructions (counted in gcc's
4719 internal representation) in a single function that the tree inliner
4720 will consider for inlining. This only affects functions declared
4721 inline and methods implemented in a class declaration (C++).
4722 The default value is 500.
4724 @item max-inline-insns-auto
4725 When you use @option{-finline-functions} (included in @option{-O3}),
4726 a lot of functions that would otherwise not be considered for inlining
4727 by the compiler will be investigated. To those functions, a different
4728 (more restrictive) limit compared to functions declared inline can
4730 The default value is 120.
4732 @item large-function-insns
4733 The limit specifying really large functions. For functions greater than this
4734 limit inlining is constrained by @option{--param large-function-growth}.
4735 This parameter is useful primarily to avoid extreme compilation time caused by non-linear
4736 algorithms used by the backend.
4737 This parameter is ignored when @option{-funit-at-a-time} is not used.
4738 The default value is 3000.
4740 @item large-function-growth
4741 Specifies maximal growth of large function caused by inlining in percents.
4742 This parameter is ignored when @option{-funit-at-a-time} is not used.
4743 The default value is 200.
4745 @item inline-unit-growth
4746 Specifies maximal overall growth of the compilation unit caused by inlining.
4747 This parameter is ignored when @option{-funit-at-a-time} is not used.
4748 The default value is 150.
4750 @item max-inline-insns-rtl
4751 For languages that use the RTL inliner (this happens at a later stage
4752 than tree inlining), you can set the maximum allowable size (counted
4753 in RTL instructions) for the RTL inliner with this parameter.
4754 The default value is 600.
4756 @item max-unrolled-insns
4757 The maximum number of instructions that a loop should have if that loop
4758 is unrolled, and if the loop is unrolled, it determines how many times
4759 the loop code is unrolled.
4761 @item max-average-unrolled-insns
4762 The maximum number of instructions biased by probabilities of their execution
4763 that a loop should have if that loop is unrolled, and if the loop is unrolled,
4764 it determines how many times the loop code is unrolled.
4766 @item max-unroll-times
4767 The maximum number of unrollings of a single loop.
4769 @item max-peeled-insns
4770 The maximum number of instructions that a loop should have if that loop
4771 is peeled, and if the loop is peeled, it determines how many times
4772 the loop code is peeled.
4774 @item max-peel-times
4775 The maximum number of peelings of a single loop.
4777 @item max-completely-peeled-insns
4778 The maximum number of insns of a completely peeled loop.
4780 @item max-completely-peel-times
4781 The maximum number of iterations of a loop to be suitable for complete peeling.
4783 @item max-unswitch-insns
4784 The maximum number of insns of an unswitched loop.
4786 @item max-unswitch-level
4787 The maximum number of branches unswitched in a single loop.
4789 @item hot-bb-count-fraction
4790 Select fraction of the maximal count of repetitions of basic block in program
4791 given basic block needs to have to be considered hot.
4793 @item hot-bb-frequency-fraction
4794 Select fraction of the maximal frequency of executions of basic block in
4795 function given basic block needs to have to be considered hot
4797 @item tracer-dynamic-coverage
4798 @itemx tracer-dynamic-coverage-feedback
4800 This value is used to limit superblock formation once the given percentage of
4801 executed instructions is covered. This limits unnecessary code size
4804 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4805 feedback is available. The real profiles (as opposed to statically estimated
4806 ones) are much less balanced allowing the threshold to be larger value.
4808 @item tracer-max-code-growth
4809 Stop tail duplication once code growth has reached given percentage. This is
4810 rather hokey argument, as most of the duplicates will be eliminated later in
4811 cross jumping, so it may be set to much higher values than is the desired code
4814 @item tracer-min-branch-ratio
4816 Stop reverse growth when the reverse probability of best edge is less than this
4817 threshold (in percent).
4819 @item tracer-min-branch-ratio
4820 @itemx tracer-min-branch-ratio-feedback
4822 Stop forward growth if the best edge do have probability lower than this
4825 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
4826 compilation for profile feedback and one for compilation without. The value
4827 for compilation with profile feedback needs to be more conservative (higher) in
4828 order to make tracer effective.
4830 @item max-cse-path-length
4832 Maximum number of basic blocks on path that cse considers.
4834 @item ggc-min-expand
4836 GCC uses a garbage collector to manage its own memory allocation. This
4837 parameter specifies the minimum percentage by which the garbage
4838 collector's heap should be allowed to expand between collections.
4839 Tuning this may improve compilation speed; it has no effect on code
4842 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
4843 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
4844 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
4845 GCC is not able to calculate RAM on a particular platform, the lower
4846 bound of 30% is used. Setting this parameter and
4847 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4848 every opportunity. This is extremely slow, but can be useful for
4851 @item ggc-min-heapsize
4853 Minimum size of the garbage collector's heap before it begins bothering
4854 to collect garbage. The first collection occurs after the heap expands
4855 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4856 tuning this may improve compilation speed, and has no effect on code
4859 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
4860 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
4861 available, the notion of "RAM" is the smallest of actual RAM,
4862 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
4863 RAM on a particular platform, the lower bound is used. Setting this
4864 parameter very large effectively disables garbage collection. Setting
4865 this parameter and @option{ggc-min-expand} to zero causes a full
4866 collection to occur at every opportunity.
4868 @item max-reload-search-insns
4869 The maximum number of instruction reload should look backward for equivalent
4870 register. Increasing values mean more aggressive optimization, making the
4871 compile time increase with probably slightly better performance. The default
4874 @item max-cselib-memory-location
4875 The maximum number of memory locations cselib should take into acount.
4876 Increasing values mean more aggressive optimization, making the compile time
4877 increase with probably slightly better performance. The default value is 500.
4879 @item reorder-blocks-duplicate
4880 @itemx reorder-blocks-duplicate-feedback
4882 Used by basic block reordering pass to decide whether to use unconditional
4883 branch or duplicate the code on its destination. Code is duplicated when its
4884 estimated size is smaller than this value multiplied by the estimated size of
4885 unconditional jump in the hot spots of the program.
4887 The @option{reorder-block-duplicate-feedback} is used only when profile
4888 feedback is available and may be set to higher values than
4889 @option{reorder-block-duplicate} since information about the hot spots is more
4894 @node Preprocessor Options
4895 @section Options Controlling the Preprocessor
4896 @cindex preprocessor options
4897 @cindex options, preprocessor
4899 These options control the C preprocessor, which is run on each C source
4900 file before actual compilation.
4902 If you use the @option{-E} option, nothing is done except preprocessing.
4903 Some of these options make sense only together with @option{-E} because
4904 they cause the preprocessor output to be unsuitable for actual
4909 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4910 and pass @var{option} directly through to the preprocessor. If
4911 @var{option} contains commas, it is split into multiple options at the
4912 commas. However, many options are modified, translated or interpreted
4913 by the compiler driver before being passed to the preprocessor, and
4914 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4915 interface is undocumented and subject to change, so whenever possible
4916 you should avoid using @option{-Wp} and let the driver handle the
4919 @item -Xpreprocessor @var{option}
4920 @opindex preprocessor
4921 Pass @var{option} as an option to the preprocessor. You can use this to
4922 supply system-specific preprocessor options which GCC does not know how to
4925 If you want to pass an option that takes an argument, you must use
4926 @option{-Xpreprocessor} twice, once for the option and once for the argument.
4929 @include cppopts.texi
4931 @node Assembler Options
4932 @section Passing Options to the Assembler
4934 @c prevent bad page break with this line
4935 You can pass options to the assembler.
4938 @item -Wa,@var{option}
4940 Pass @var{option} as an option to the assembler. If @var{option}
4941 contains commas, it is split into multiple options at the commas.
4943 @item -Xassembler @var{option}
4945 Pass @var{option} as an option to the assembler. You can use this to
4946 supply system-specific assembler options which GCC does not know how to
4949 If you want to pass an option that takes an argument, you must use
4950 @option{-Xassembler} twice, once for the option and once for the argument.
4955 @section Options for Linking
4956 @cindex link options
4957 @cindex options, linking
4959 These options come into play when the compiler links object files into
4960 an executable output file. They are meaningless if the compiler is
4961 not doing a link step.
4965 @item @var{object-file-name}
4966 A file name that does not end in a special recognized suffix is
4967 considered to name an object file or library. (Object files are
4968 distinguished from libraries by the linker according to the file
4969 contents.) If linking is done, these object files are used as input
4978 If any of these options is used, then the linker is not run, and
4979 object file names should not be used as arguments. @xref{Overall
4983 @item -l@var{library}
4984 @itemx -l @var{library}
4986 Search the library named @var{library} when linking. (The second
4987 alternative with the library as a separate argument is only for
4988 POSIX compliance and is not recommended.)
4990 It makes a difference where in the command you write this option; the
4991 linker searches and processes libraries and object files in the order they
4992 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4993 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4994 to functions in @samp{z}, those functions may not be loaded.
4996 The linker searches a standard list of directories for the library,
4997 which is actually a file named @file{lib@var{library}.a}. The linker
4998 then uses this file as if it had been specified precisely by name.
5000 The directories searched include several standard system directories
5001 plus any that you specify with @option{-L}.
5003 Normally the files found this way are library files---archive files
5004 whose members are object files. The linker handles an archive file by
5005 scanning through it for members which define symbols that have so far
5006 been referenced but not defined. But if the file that is found is an
5007 ordinary object file, it is linked in the usual fashion. The only
5008 difference between using an @option{-l} option and specifying a file name
5009 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
5010 and searches several directories.
5014 You need this special case of the @option{-l} option in order to
5015 link an Objective-C program.
5018 @opindex nostartfiles
5019 Do not use the standard system startup files when linking.
5020 The standard system libraries are used normally, unless @option{-nostdlib}
5021 or @option{-nodefaultlibs} is used.
5023 @item -nodefaultlibs
5024 @opindex nodefaultlibs
5025 Do not use the standard system libraries when linking.
5026 Only the libraries you specify will be passed to the linker.
5027 The standard startup files are used normally, unless @option{-nostartfiles}
5028 is used. The compiler may generate calls to memcmp, memset, and memcpy
5029 for System V (and ISO C) environments or to bcopy and bzero for
5030 BSD environments. These entries are usually resolved by entries in
5031 libc. These entry points should be supplied through some other
5032 mechanism when this option is specified.
5036 Do not use the standard system startup files or libraries when linking.
5037 No startup files and only the libraries you specify will be passed to
5038 the linker. The compiler may generate calls to memcmp, memset, and memcpy
5039 for System V (and ISO C) environments or to bcopy and bzero for
5040 BSD environments. These entries are usually resolved by entries in
5041 libc. These entry points should be supplied through some other
5042 mechanism when this option is specified.
5044 @cindex @option{-lgcc}, use with @option{-nostdlib}
5045 @cindex @option{-nostdlib} and unresolved references
5046 @cindex unresolved references and @option{-nostdlib}
5047 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5048 @cindex @option{-nodefaultlibs} and unresolved references
5049 @cindex unresolved references and @option{-nodefaultlibs}
5050 One of the standard libraries bypassed by @option{-nostdlib} and
5051 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5052 that GCC uses to overcome shortcomings of particular machines, or special
5053 needs for some languages.
5054 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5055 Collection (GCC) Internals},
5056 for more discussion of @file{libgcc.a}.)
5057 In most cases, you need @file{libgcc.a} even when you want to avoid
5058 other standard libraries. In other words, when you specify @option{-nostdlib}
5059 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5060 This ensures that you have no unresolved references to internal GCC
5061 library subroutines. (For example, @samp{__main}, used to ensure C++
5062 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5063 GNU Compiler Collection (GCC) Internals}.)
5067 Produce a position independent executable on targets which support it.
5068 For predictable results, you must also specify the same set of options
5069 that were used to generate code (@option{-fpie}, @option{-fPIE},
5070 or model suboptions) when you specify this option.
5074 Remove all symbol table and relocation information from the executable.
5078 On systems that support dynamic linking, this prevents linking with the shared
5079 libraries. On other systems, this option has no effect.
5083 Produce a shared object which can then be linked with other objects to
5084 form an executable. Not all systems support this option. For predictable
5085 results, you must also specify the same set of options that were used to
5086 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5087 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5088 needs to build supplementary stub code for constructors to work. On
5089 multi-libbed systems, @samp{gcc -shared} must select the correct support
5090 libraries to link against. Failing to supply the correct flags may lead
5091 to subtle defects. Supplying them in cases where they are not necessary
5094 @item -shared-libgcc
5095 @itemx -static-libgcc
5096 @opindex shared-libgcc
5097 @opindex static-libgcc
5098 On systems that provide @file{libgcc} as a shared library, these options
5099 force the use of either the shared or static version respectively.
5100 If no shared version of @file{libgcc} was built when the compiler was
5101 configured, these options have no effect.
5103 There are several situations in which an application should use the
5104 shared @file{libgcc} instead of the static version. The most common
5105 of these is when the application wishes to throw and catch exceptions
5106 across different shared libraries. In that case, each of the libraries
5107 as well as the application itself should use the shared @file{libgcc}.
5109 Therefore, the G++ and GCJ drivers automatically add
5110 @option{-shared-libgcc} whenever you build a shared library or a main
5111 executable, because C++ and Java programs typically use exceptions, so
5112 this is the right thing to do.
5114 If, instead, you use the GCC driver to create shared libraries, you may
5115 find that they will not always be linked with the shared @file{libgcc}.
5116 If GCC finds, at its configuration time, that you have a GNU linker that
5117 does not support option @option{--eh-frame-hdr}, it will link the shared
5118 version of @file{libgcc} into shared libraries by default. Otherwise,
5119 it will take advantage of the linker and optimize away the linking with
5120 the shared version of @file{libgcc}, linking with the static version of
5121 libgcc by default. This allows exceptions to propagate through such
5122 shared libraries, without incurring relocation costs at library load
5125 However, if a library or main executable is supposed to throw or catch
5126 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5127 for the languages used in the program, or using the option
5128 @option{-shared-libgcc}, such that it is linked with the shared
5133 Bind references to global symbols when building a shared object. Warn
5134 about any unresolved references (unless overridden by the link editor
5135 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5138 @item -Xlinker @var{option}
5140 Pass @var{option} as an option to the linker. You can use this to
5141 supply system-specific linker options which GCC does not know how to
5144 If you want to pass an option that takes an argument, you must use
5145 @option{-Xlinker} twice, once for the option and once for the argument.
5146 For example, to pass @option{-assert definitions}, you must write
5147 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5148 @option{-Xlinker "-assert definitions"}, because this passes the entire
5149 string as a single argument, which is not what the linker expects.
5151 @item -Wl,@var{option}
5153 Pass @var{option} as an option to the linker. If @var{option} contains
5154 commas, it is split into multiple options at the commas.
5156 @item -u @var{symbol}
5158 Pretend the symbol @var{symbol} is undefined, to force linking of
5159 library modules to define it. You can use @option{-u} multiple times with
5160 different symbols to force loading of additional library modules.
5163 @node Directory Options
5164 @section Options for Directory Search
5165 @cindex directory options
5166 @cindex options, directory search
5169 These options specify directories to search for header files, for
5170 libraries and for parts of the compiler:
5175 Add the directory @var{dir} to the head of the list of directories to be
5176 searched for header files. This can be used to override a system header
5177 file, substituting your own version, since these directories are
5178 searched before the system header file directories. However, you should
5179 not use this option to add directories that contain vendor-supplied
5180 system header files (use @option{-isystem} for that). If you use more than
5181 one @option{-I} option, the directories are scanned in left-to-right
5182 order; the standard system directories come after.
5184 If a standard system include directory, or a directory specified with
5185 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5186 option will be ignored. The directory will still be searched but as a
5187 system directory at its normal position in the system include chain.
5188 This is to ensure that GCC's procedure to fix buggy system headers and
5189 the ordering for the include_next directive are not inadvertently changed.
5190 If you really need to change the search order for system directories,
5191 use the @option{-nostdinc} and/or @option{-isystem} options.
5195 Any directories you specify with @option{-I} options before the @option{-I-}
5196 option are searched only for the case of @samp{#include "@var{file}"};
5197 they are not searched for @samp{#include <@var{file}>}.
5199 If additional directories are specified with @option{-I} options after
5200 the @option{-I-}, these directories are searched for all @samp{#include}
5201 directives. (Ordinarily @emph{all} @option{-I} directories are used
5204 In addition, the @option{-I-} option inhibits the use of the current
5205 directory (where the current input file came from) as the first search
5206 directory for @samp{#include "@var{file}"}. There is no way to
5207 override this effect of @option{-I-}. With @option{-I.} you can specify
5208 searching the directory which was current when the compiler was
5209 invoked. That is not exactly the same as what the preprocessor does
5210 by default, but it is often satisfactory.
5212 @option{-I-} does not inhibit the use of the standard system directories
5213 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5218 Add directory @var{dir} to the list of directories to be searched
5221 @item -B@var{prefix}
5223 This option specifies where to find the executables, libraries,
5224 include files, and data files of the compiler itself.
5226 The compiler driver program runs one or more of the subprograms
5227 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5228 @var{prefix} as a prefix for each program it tries to run, both with and
5229 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5231 For each subprogram to be run, the compiler driver first tries the
5232 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5233 was not specified, the driver tries two standard prefixes, which are
5234 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5235 those results in a file name that is found, the unmodified program
5236 name is searched for using the directories specified in your
5237 @env{PATH} environment variable.
5239 The compiler will check to see if the path provided by the @option{-B}
5240 refers to a directory, and if necessary it will add a directory
5241 separator character at the end of the path.
5243 @option{-B} prefixes that effectively specify directory names also apply
5244 to libraries in the linker, because the compiler translates these
5245 options into @option{-L} options for the linker. They also apply to
5246 includes files in the preprocessor, because the compiler translates these
5247 options into @option{-isystem} options for the preprocessor. In this case,
5248 the compiler appends @samp{include} to the prefix.
5250 The run-time support file @file{libgcc.a} can also be searched for using
5251 the @option{-B} prefix, if needed. If it is not found there, the two
5252 standard prefixes above are tried, and that is all. The file is left
5253 out of the link if it is not found by those means.
5255 Another way to specify a prefix much like the @option{-B} prefix is to use
5256 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5259 As a special kludge, if the path provided by @option{-B} is
5260 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5261 9, then it will be replaced by @file{[dir/]include}. This is to help
5262 with boot-strapping the compiler.
5264 @item -specs=@var{file}
5266 Process @var{file} after the compiler reads in the standard @file{specs}
5267 file, in order to override the defaults that the @file{gcc} driver
5268 program uses when determining what switches to pass to @file{cc1},
5269 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5270 @option{-specs=@var{file}} can be specified on the command line, and they
5271 are processed in order, from left to right.
5277 @section Specifying subprocesses and the switches to pass to them
5280 @command{gcc} is a driver program. It performs its job by invoking a
5281 sequence of other programs to do the work of compiling, assembling and
5282 linking. GCC interprets its command-line parameters and uses these to
5283 deduce which programs it should invoke, and which command-line options
5284 it ought to place on their command lines. This behavior is controlled
5285 by @dfn{spec strings}. In most cases there is one spec string for each
5286 program that GCC can invoke, but a few programs have multiple spec
5287 strings to control their behavior. The spec strings built into GCC can
5288 be overridden by using the @option{-specs=} command-line switch to specify
5291 @dfn{Spec files} are plaintext files that are used to construct spec
5292 strings. They consist of a sequence of directives separated by blank
5293 lines. The type of directive is determined by the first non-whitespace
5294 character on the line and it can be one of the following:
5297 @item %@var{command}
5298 Issues a @var{command} to the spec file processor. The commands that can
5302 @item %include <@var{file}>
5304 Search for @var{file} and insert its text at the current point in the
5307 @item %include_noerr <@var{file}>
5308 @cindex %include_noerr
5309 Just like @samp{%include}, but do not generate an error message if the include
5310 file cannot be found.
5312 @item %rename @var{old_name} @var{new_name}
5314 Rename the spec string @var{old_name} to @var{new_name}.
5318 @item *[@var{spec_name}]:
5319 This tells the compiler to create, override or delete the named spec
5320 string. All lines after this directive up to the next directive or
5321 blank line are considered to be the text for the spec string. If this
5322 results in an empty string then the spec will be deleted. (Or, if the
5323 spec did not exist, then nothing will happened.) Otherwise, if the spec
5324 does not currently exist a new spec will be created. If the spec does
5325 exist then its contents will be overridden by the text of this
5326 directive, unless the first character of that text is the @samp{+}
5327 character, in which case the text will be appended to the spec.
5329 @item [@var{suffix}]:
5330 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5331 and up to the next directive or blank line are considered to make up the
5332 spec string for the indicated suffix. When the compiler encounters an
5333 input file with the named suffix, it will processes the spec string in
5334 order to work out how to compile that file. For example:
5341 This says that any input file whose name ends in @samp{.ZZ} should be
5342 passed to the program @samp{z-compile}, which should be invoked with the
5343 command-line switch @option{-input} and with the result of performing the
5344 @samp{%i} substitution. (See below.)
5346 As an alternative to providing a spec string, the text that follows a
5347 suffix directive can be one of the following:
5350 @item @@@var{language}
5351 This says that the suffix is an alias for a known @var{language}. This is
5352 similar to using the @option{-x} command-line switch to GCC to specify a
5353 language explicitly. For example:
5360 Says that .ZZ files are, in fact, C++ source files.
5363 This causes an error messages saying:
5366 @var{name} compiler not installed on this system.
5370 GCC already has an extensive list of suffixes built into it.
5371 This directive will add an entry to the end of the list of suffixes, but
5372 since the list is searched from the end backwards, it is effectively
5373 possible to override earlier entries using this technique.
5377 GCC has the following spec strings built into it. Spec files can
5378 override these strings or create their own. Note that individual
5379 targets can also add their own spec strings to this list.
5382 asm Options to pass to the assembler
5383 asm_final Options to pass to the assembler post-processor
5384 cpp Options to pass to the C preprocessor
5385 cc1 Options to pass to the C compiler
5386 cc1plus Options to pass to the C++ compiler
5387 endfile Object files to include at the end of the link
5388 link Options to pass to the linker
5389 lib Libraries to include on the command line to the linker
5390 libgcc Decides which GCC support library to pass to the linker
5391 linker Sets the name of the linker
5392 predefines Defines to be passed to the C preprocessor
5393 signed_char Defines to pass to CPP to say whether @code{char} is signed
5395 startfile Object files to include at the start of the link
5398 Here is a small example of a spec file:
5404 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5407 This example renames the spec called @samp{lib} to @samp{old_lib} and
5408 then overrides the previous definition of @samp{lib} with a new one.
5409 The new definition adds in some extra command-line options before
5410 including the text of the old definition.
5412 @dfn{Spec strings} are a list of command-line options to be passed to their
5413 corresponding program. In addition, the spec strings can contain
5414 @samp{%}-prefixed sequences to substitute variable text or to
5415 conditionally insert text into the command line. Using these constructs
5416 it is possible to generate quite complex command lines.
5418 Here is a table of all defined @samp{%}-sequences for spec
5419 strings. Note that spaces are not generated automatically around the
5420 results of expanding these sequences. Therefore you can concatenate them
5421 together or combine them with constant text in a single argument.
5425 Substitute one @samp{%} into the program name or argument.
5428 Substitute the name of the input file being processed.
5431 Substitute the basename of the input file being processed.
5432 This is the substring up to (and not including) the last period
5433 and not including the directory.
5436 This is the same as @samp{%b}, but include the file suffix (text after
5440 Marks the argument containing or following the @samp{%d} as a
5441 temporary file name, so that that file will be deleted if GCC exits
5442 successfully. Unlike @samp{%g}, this contributes no text to the
5445 @item %g@var{suffix}
5446 Substitute a file name that has suffix @var{suffix} and is chosen
5447 once per compilation, and mark the argument in the same way as
5448 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5449 name is now chosen in a way that is hard to predict even when previously
5450 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5451 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5452 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5453 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5454 was simply substituted with a file name chosen once per compilation,
5455 without regard to any appended suffix (which was therefore treated
5456 just like ordinary text), making such attacks more likely to succeed.
5458 @item %u@var{suffix}
5459 Like @samp{%g}, but generates a new temporary file name even if
5460 @samp{%u@var{suffix}} was already seen.
5462 @item %U@var{suffix}
5463 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5464 new one if there is no such last file name. In the absence of any
5465 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5466 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5467 would involve the generation of two distinct file names, one
5468 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5469 simply substituted with a file name chosen for the previous @samp{%u},
5470 without regard to any appended suffix.
5472 @item %j@var{suffix}
5473 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5474 writable, and if save-temps is off; otherwise, substitute the name
5475 of a temporary file, just like @samp{%u}. This temporary file is not
5476 meant for communication between processes, but rather as a junk
5479 @item %|@var{suffix}
5480 @itemx %m@var{suffix}
5481 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5482 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5483 all. These are the two most common ways to instruct a program that it
5484 should read from standard input or write to standard output. If you
5485 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5486 construct: see for example @file{f/lang-specs.h}.
5488 @item %.@var{SUFFIX}
5489 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5490 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5491 terminated by the next space or %.
5494 Marks the argument containing or following the @samp{%w} as the
5495 designated output file of this compilation. This puts the argument
5496 into the sequence of arguments that @samp{%o} will substitute later.
5499 Substitutes the names of all the output files, with spaces
5500 automatically placed around them. You should write spaces
5501 around the @samp{%o} as well or the results are undefined.
5502 @samp{%o} is for use in the specs for running the linker.
5503 Input files whose names have no recognized suffix are not compiled
5504 at all, but they are included among the output files, so they will
5508 Substitutes the suffix for object files. Note that this is
5509 handled specially when it immediately follows @samp{%g, %u, or %U},
5510 because of the need for those to form complete file names. The
5511 handling is such that @samp{%O} is treated exactly as if it had already
5512 been substituted, except that @samp{%g, %u, and %U} do not currently
5513 support additional @var{suffix} characters following @samp{%O} as they would
5514 following, for example, @samp{.o}.
5517 Substitutes the standard macro predefinitions for the
5518 current target machine. Use this when running @code{cpp}.
5521 Like @samp{%p}, but puts @samp{__} before and after the name of each
5522 predefined macro, except for macros that start with @samp{__} or with
5523 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5527 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5528 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5529 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5533 Current argument is the name of a library or startup file of some sort.
5534 Search for that file in a standard list of directories and substitute
5535 the full name found.
5538 Print @var{str} as an error message. @var{str} is terminated by a newline.
5539 Use this when inconsistent options are detected.
5542 Substitute the contents of spec string @var{name} at this point.
5545 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5547 @item %x@{@var{option}@}
5548 Accumulate an option for @samp{%X}.
5551 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5555 Output the accumulated assembler options specified by @option{-Wa}.
5558 Output the accumulated preprocessor options specified by @option{-Wp}.
5561 Process the @code{asm} spec. This is used to compute the
5562 switches to be passed to the assembler.
5565 Process the @code{asm_final} spec. This is a spec string for
5566 passing switches to an assembler post-processor, if such a program is
5570 Process the @code{link} spec. This is the spec for computing the
5571 command line passed to the linker. Typically it will make use of the
5572 @samp{%L %G %S %D and %E} sequences.
5575 Dump out a @option{-L} option for each directory that GCC believes might
5576 contain startup files. If the target supports multilibs then the
5577 current multilib directory will be prepended to each of these paths.
5580 Output the multilib directory with directory separators replaced with
5581 @samp{_}. If multilib directories are not set, or the multilib directory is
5582 @file{.} then this option emits nothing.
5585 Process the @code{lib} spec. This is a spec string for deciding which
5586 libraries should be included on the command line to the linker.
5589 Process the @code{libgcc} spec. This is a spec string for deciding
5590 which GCC support library should be included on the command line to the linker.
5593 Process the @code{startfile} spec. This is a spec for deciding which
5594 object files should be the first ones passed to the linker. Typically
5595 this might be a file named @file{crt0.o}.
5598 Process the @code{endfile} spec. This is a spec string that specifies
5599 the last object files that will be passed to the linker.
5602 Process the @code{cpp} spec. This is used to construct the arguments
5603 to be passed to the C preprocessor.
5606 Process the @code{signed_char} spec. This is intended to be used
5607 to tell cpp whether a char is signed. It typically has the definition:
5609 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
5613 Process the @code{cc1} spec. This is used to construct the options to be
5614 passed to the actual C compiler (@samp{cc1}).
5617 Process the @code{cc1plus} spec. This is used to construct the options to be
5618 passed to the actual C++ compiler (@samp{cc1plus}).
5621 Substitute the variable part of a matched option. See below.
5622 Note that each comma in the substituted string is replaced by
5626 Remove all occurrences of @code{-S} from the command line. Note---this
5627 command is position dependent. @samp{%} commands in the spec string
5628 before this one will see @code{-S}, @samp{%} commands in the spec string
5629 after this one will not.
5631 @item %:@var{function}(@var{args})
5632 Call the named function @var{function}, passing it @var{args}.
5633 @var{args} is first processed as a nested spec string, then split
5634 into an argument vector in the usual fashion. The function returns
5635 a string which is processed as if it had appeared literally as part
5636 of the current spec.
5638 The following built-in spec functions are provided:
5641 @item @code{if-exists}
5642 The @code{if-exists} spec function takes one argument, an absolute
5643 pathname to a file. If the file exists, @code{if-exists} returns the
5644 pathname. Here is a small example of its usage:
5648 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5651 @item @code{if-exists-else}
5652 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5653 spec function, except that it takes two arguments. The first argument is
5654 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5655 returns the pathname. If it does not exist, it returns the second argument.
5656 This way, @code{if-exists-else} can be used to select one file or another,
5657 based on the existence of the first. Here is a small example of its usage:
5661 crt0%O%s %:if-exists(crti%O%s) \
5662 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5667 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5668 If that switch was not specified, this substitutes nothing. Note that
5669 the leading dash is omitted when specifying this option, and it is
5670 automatically inserted if the substitution is performed. Thus the spec
5671 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5672 and would output the command line option @option{-foo}.
5674 @item %W@{@code{S}@}
5675 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5678 @item %@{@code{S}*@}
5679 Substitutes all the switches specified to GCC whose names start
5680 with @code{-S}, but which also take an argument. This is used for
5681 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5682 GCC considers @option{-o foo} as being
5683 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5684 text, including the space. Thus two arguments would be generated.
5686 @item %@{@code{S}*&@code{T}*@}
5687 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5688 (the order of @code{S} and @code{T} in the spec is not significant).
5689 There can be any number of ampersand-separated variables; for each the
5690 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5692 @item %@{@code{S}:@code{X}@}
5693 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
5695 @item %@{!@code{S}:@code{X}@}
5696 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
5698 @item %@{@code{S}*:@code{X}@}
5699 Substitutes @code{X} if one or more switches whose names start with
5700 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
5701 once, no matter how many such switches appeared. However, if @code{%*}
5702 appears somewhere in @code{X}, then @code{X} will be substituted once
5703 for each matching switch, with the @code{%*} replaced by the part of
5704 that switch that matched the @code{*}.
5706 @item %@{.@code{S}:@code{X}@}
5707 Substitutes @code{X}, if processing a file with suffix @code{S}.
5709 @item %@{!.@code{S}:@code{X}@}
5710 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
5712 @item %@{@code{S}|@code{P}:@code{X}@}
5713 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
5714 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
5715 although they have a stronger binding than the @samp{|}. If @code{%*}
5716 appears in @code{X}, all of the alternatives must be starred, and only
5717 the first matching alternative is substituted.
5719 For example, a spec string like this:
5722 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5725 will output the following command-line options from the following input
5726 command-line options:
5731 -d fred.c -foo -baz -boggle
5732 -d jim.d -bar -baz -boggle
5735 @item %@{S:X; T:Y; :D@}
5737 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
5738 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
5739 be as many clauses as you need. This may be combined with @code{.},
5740 @code{!}, @code{|}, and @code{*} as needed.
5745 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
5746 construct may contain other nested @samp{%} constructs or spaces, or
5747 even newlines. They are processed as usual, as described above.
5748 Trailing white space in @code{X} is ignored. White space may also
5749 appear anywhere on the left side of the colon in these constructs,
5750 except between @code{.} or @code{*} and the corresponding word.
5752 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
5753 handled specifically in these constructs. If another value of
5754 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5755 @option{-W} switch is found later in the command line, the earlier
5756 switch value is ignored, except with @{@code{S}*@} where @code{S} is
5757 just one letter, which passes all matching options.
5759 The character @samp{|} at the beginning of the predicate text is used to
5760 indicate that a command should be piped to the following command, but
5761 only if @option{-pipe} is specified.
5763 It is built into GCC which switches take arguments and which do not.
5764 (You might think it would be useful to generalize this to allow each
5765 compiler's spec to say which switches take arguments. But this cannot
5766 be done in a consistent fashion. GCC cannot even decide which input
5767 files have been specified without knowing which switches take arguments,
5768 and it must know which input files to compile in order to tell which
5771 GCC also knows implicitly that arguments starting in @option{-l} are to be
5772 treated as compiler output files, and passed to the linker in their
5773 proper position among the other output files.
5775 @c man begin OPTIONS
5777 @node Target Options
5778 @section Specifying Target Machine and Compiler Version
5779 @cindex target options
5780 @cindex cross compiling
5781 @cindex specifying machine version
5782 @cindex specifying compiler version and target machine
5783 @cindex compiler version, specifying
5784 @cindex target machine, specifying
5786 The usual way to run GCC is to run the executable called @file{gcc}, or
5787 @file{<machine>-gcc} when cross-compiling, or
5788 @file{<machine>-gcc-<version>} to run a version other than the one that
5789 was installed last. Sometimes this is inconvenient, so GCC provides
5790 options that will switch to another cross-compiler or version.
5793 @item -b @var{machine}
5795 The argument @var{machine} specifies the target machine for compilation.
5797 The value to use for @var{machine} is the same as was specified as the
5798 machine type when configuring GCC as a cross-compiler. For
5799 example, if a cross-compiler was configured with @samp{configure
5800 i386v}, meaning to compile for an 80386 running System V, then you
5801 would specify @option{-b i386v} to run that cross compiler.
5803 @item -V @var{version}
5805 The argument @var{version} specifies which version of GCC to run.
5806 This is useful when multiple versions are installed. For example,
5807 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5810 The @option{-V} and @option{-b} options work by running the
5811 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5812 use them if you can just run that directly.
5814 @node Submodel Options
5815 @section Hardware Models and Configurations
5816 @cindex submodel options
5817 @cindex specifying hardware config
5818 @cindex hardware models and configurations, specifying
5819 @cindex machine dependent options
5821 Earlier we discussed the standard option @option{-b} which chooses among
5822 different installed compilers for completely different target
5823 machines, such as VAX vs.@: 68000 vs.@: 80386.
5825 In addition, each of these target machine types can have its own
5826 special options, starting with @samp{-m}, to choose among various
5827 hardware models or configurations---for example, 68010 vs 68020,
5828 floating coprocessor or none. A single installed version of the
5829 compiler can compile for any model or configuration, according to the
5832 Some configurations of the compiler also support additional special
5833 options, usually for compatibility with other compilers on the same
5836 These options are defined by the macro @code{TARGET_SWITCHES} in the
5837 machine description. The default for the options is also defined by
5838 that macro, which enables you to change the defaults.
5848 * RS/6000 and PowerPC Options::
5851 * i386 and x86-64 Options::
5853 * Intel 960 Options::
5854 * DEC Alpha Options::
5855 * DEC Alpha/VMS Options::
5858 * System V Options::
5859 * TMS320C3x/C4x Options::
5867 * S/390 and zSeries Options::
5871 * Xstormy16 Options::
5876 @node M680x0 Options
5877 @subsection M680x0 Options
5878 @cindex M680x0 options
5880 These are the @samp{-m} options defined for the 68000 series. The default
5881 values for these options depends on which style of 68000 was selected when
5882 the compiler was configured; the defaults for the most common choices are
5890 Generate output for a 68000. This is the default
5891 when the compiler is configured for 68000-based systems.
5893 Use this option for microcontrollers with a 68000 or EC000 core,
5894 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5900 Generate output for a 68020. This is the default
5901 when the compiler is configured for 68020-based systems.
5905 Generate output containing 68881 instructions for floating point.
5906 This is the default for most 68020 systems unless @option{--nfp} was
5907 specified when the compiler was configured.
5911 Generate output for a 68030. This is the default when the compiler is
5912 configured for 68030-based systems.
5916 Generate output for a 68040. This is the default when the compiler is
5917 configured for 68040-based systems.
5919 This option inhibits the use of 68881/68882 instructions that have to be
5920 emulated by software on the 68040. Use this option if your 68040 does not
5921 have code to emulate those instructions.
5925 Generate output for a 68060. This is the default when the compiler is
5926 configured for 68060-based systems.
5928 This option inhibits the use of 68020 and 68881/68882 instructions that
5929 have to be emulated by software on the 68060. Use this option if your 68060
5930 does not have code to emulate those instructions.
5934 Generate output for a CPU32. This is the default
5935 when the compiler is configured for CPU32-based systems.
5937 Use this option for microcontrollers with a
5938 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5939 68336, 68340, 68341, 68349 and 68360.
5943 Generate output for a 520X ``coldfire'' family cpu. This is the default
5944 when the compiler is configured for 520X-based systems.
5946 Use this option for microcontroller with a 5200 core, including
5947 the MCF5202, MCF5203, MCF5204 and MCF5202.
5952 Generate output for a 68040, without using any of the new instructions.
5953 This results in code which can run relatively efficiently on either a
5954 68020/68881 or a 68030 or a 68040. The generated code does use the
5955 68881 instructions that are emulated on the 68040.
5959 Generate output for a 68060, without using any of the new instructions.
5960 This results in code which can run relatively efficiently on either a
5961 68020/68881 or a 68030 or a 68040. The generated code does use the
5962 68881 instructions that are emulated on the 68060.
5965 @opindex msoft-float
5966 Generate output containing library calls for floating point.
5967 @strong{Warning:} the requisite libraries are not available for all m68k
5968 targets. Normally the facilities of the machine's usual C compiler are
5969 used, but this can't be done directly in cross-compilation. You must
5970 make your own arrangements to provide suitable library functions for
5971 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5972 @samp{m68k-*-coff} do provide software floating point support.
5976 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5979 @opindex mnobitfield
5980 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5981 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5985 Do use the bit-field instructions. The @option{-m68020} option implies
5986 @option{-mbitfield}. This is the default if you use a configuration
5987 designed for a 68020.
5991 Use a different function-calling convention, in which functions
5992 that take a fixed number of arguments return with the @code{rtd}
5993 instruction, which pops their arguments while returning. This
5994 saves one instruction in the caller since there is no need to pop
5995 the arguments there.
5997 This calling convention is incompatible with the one normally
5998 used on Unix, so you cannot use it if you need to call libraries
5999 compiled with the Unix compiler.
6001 Also, you must provide function prototypes for all functions that
6002 take variable numbers of arguments (including @code{printf});
6003 otherwise incorrect code will be generated for calls to those
6006 In addition, seriously incorrect code will result if you call a
6007 function with too many arguments. (Normally, extra arguments are
6008 harmlessly ignored.)
6010 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
6011 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
6014 @itemx -mno-align-int
6016 @opindex mno-align-int
6017 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
6018 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
6019 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
6020 Aligning variables on 32-bit boundaries produces code that runs somewhat
6021 faster on processors with 32-bit busses at the expense of more memory.
6023 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
6024 align structures containing the above types differently than
6025 most published application binary interface specifications for the m68k.
6029 Use the pc-relative addressing mode of the 68000 directly, instead of
6030 using a global offset table. At present, this option implies @option{-fpic},
6031 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
6032 not presently supported with @option{-mpcrel}, though this could be supported for
6033 68020 and higher processors.
6035 @item -mno-strict-align
6036 @itemx -mstrict-align
6037 @opindex mno-strict-align
6038 @opindex mstrict-align
6039 Do not (do) assume that unaligned memory references will be handled by
6043 Generate code that allows the data segment to be located in a different
6044 area of memory from the text segment. This allows for execute in place in
6045 an environment without virtual memory management. This option implies -fPIC.
6048 Generate code that assumes that the data segment follows the text segment.
6049 This is the default.
6051 @item -mid-shared-library
6052 Generate code that supports shared libraries via the library ID method.
6053 This allows for execute in place and shared libraries in an environment
6054 without virtual memory management. This option implies -fPIC.
6056 @item -mno-id-shared-library
6057 Generate code that doesn't assume ID based shared libraries are being used.
6058 This is the default.
6060 @item -mshared-library-id=n
6061 Specified the identification number of the ID based shared library being
6062 compiled. Specifying a value of 0 will generate more compact code, specifying
6063 other values will force the allocation of that number to the current
6064 library but is no more space or time efficient than omitting this option.
6068 @node M68hc1x Options
6069 @subsection M68hc1x Options
6070 @cindex M68hc1x options
6072 These are the @samp{-m} options defined for the 68hc11 and 68hc12
6073 microcontrollers. The default values for these options depends on
6074 which style of microcontroller was selected when the compiler was configured;
6075 the defaults for the most common choices are given below.
6082 Generate output for a 68HC11. This is the default
6083 when the compiler is configured for 68HC11-based systems.
6089 Generate output for a 68HC12. This is the default
6090 when the compiler is configured for 68HC12-based systems.
6096 Generate output for a 68HCS12.
6099 @opindex mauto-incdec
6100 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
6107 Enable the use of 68HC12 min and max instructions.
6110 @itemx -mno-long-calls
6111 @opindex mlong-calls
6112 @opindex mno-long-calls
6113 Treat all calls as being far away (near). If calls are assumed to be
6114 far away, the compiler will use the @code{call} instruction to
6115 call a function and the @code{rtc} instruction for returning.
6119 Consider type @code{int} to be 16 bits wide, like @code{short int}.
6121 @item -msoft-reg-count=@var{count}
6122 @opindex msoft-reg-count
6123 Specify the number of pseudo-soft registers which are used for the
6124 code generation. The maximum number is 32. Using more pseudo-soft
6125 register may or may not result in better code depending on the program.
6126 The default is 4 for 68HC11 and 2 for 68HC12.
6131 @subsection VAX Options
6134 These @samp{-m} options are defined for the VAX:
6139 Do not output certain jump instructions (@code{aobleq} and so on)
6140 that the Unix assembler for the VAX cannot handle across long
6145 Do output those jump instructions, on the assumption that you
6146 will assemble with the GNU assembler.
6150 Output code for g-format floating point numbers instead of d-format.
6154 @subsection SPARC Options
6155 @cindex SPARC options
6157 These @samp{-m} options are supported on the SPARC:
6162 @opindex mno-app-regs
6164 Specify @option{-mapp-regs} to generate output using the global registers
6165 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
6168 To be fully SVR4 ABI compliant at the cost of some performance loss,
6169 specify @option{-mno-app-regs}. You should compile libraries and system
6170 software with this option.
6175 @opindex mhard-float
6176 Generate output containing floating point instructions. This is the
6182 @opindex msoft-float
6183 Generate output containing library calls for floating point.
6184 @strong{Warning:} the requisite libraries are not available for all SPARC
6185 targets. Normally the facilities of the machine's usual C compiler are
6186 used, but this cannot be done directly in cross-compilation. You must make
6187 your own arrangements to provide suitable library functions for
6188 cross-compilation. The embedded targets @samp{sparc-*-aout} and
6189 @samp{sparclite-*-*} do provide software floating point support.
6191 @option{-msoft-float} changes the calling convention in the output file;
6192 therefore, it is only useful if you compile @emph{all} of a program with
6193 this option. In particular, you need to compile @file{libgcc.a}, the
6194 library that comes with GCC, with @option{-msoft-float} in order for
6197 @item -mhard-quad-float
6198 @opindex mhard-quad-float
6199 Generate output containing quad-word (long double) floating point
6202 @item -msoft-quad-float
6203 @opindex msoft-quad-float
6204 Generate output containing library calls for quad-word (long double)
6205 floating point instructions. The functions called are those specified
6206 in the SPARC ABI@. This is the default.
6208 As of this writing, there are no SPARC implementations that have hardware
6209 support for the quad-word floating point instructions. They all invoke
6210 a trap handler for one of these instructions, and then the trap handler
6211 emulates the effect of the instruction. Because of the trap handler overhead,
6212 this is much slower than calling the ABI library routines. Thus the
6213 @option{-msoft-quad-float} option is the default.
6219 With @option{-mflat}, the compiler does not generate save/restore instructions
6220 and will use a ``flat'' or single register window calling convention.
6221 This model uses %i7 as the frame pointer and is compatible with the normal
6222 register window model. Code from either may be intermixed.
6223 The local registers and the input registers (0--5) are still treated as
6224 ``call saved'' registers and will be saved on the stack as necessary.
6226 With @option{-mno-flat} (the default), the compiler emits save/restore
6227 instructions (except for leaf functions) and is the normal mode of operation.
6229 @item -mno-unaligned-doubles
6230 @itemx -munaligned-doubles
6231 @opindex mno-unaligned-doubles
6232 @opindex munaligned-doubles
6233 Assume that doubles have 8 byte alignment. This is the default.
6235 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
6236 alignment only if they are contained in another type, or if they have an
6237 absolute address. Otherwise, it assumes they have 4 byte alignment.
6238 Specifying this option avoids some rare compatibility problems with code
6239 generated by other compilers. It is not the default because it results
6240 in a performance loss, especially for floating point code.
6242 @item -mno-faster-structs
6243 @itemx -mfaster-structs
6244 @opindex mno-faster-structs
6245 @opindex mfaster-structs
6246 With @option{-mfaster-structs}, the compiler assumes that structures
6247 should have 8 byte alignment. This enables the use of pairs of
6248 @code{ldd} and @code{std} instructions for copies in structure
6249 assignment, in place of twice as many @code{ld} and @code{st} pairs.
6250 However, the use of this changed alignment directly violates the SPARC
6251 ABI@. Thus, it's intended only for use on targets where the developer
6252 acknowledges that their resulting code will not be directly in line with
6253 the rules of the ABI@.
6256 @opindex mimpure-text
6257 @option{-mimpure-text}, used in addition to @option{-shared}, tells
6258 the compiler to not pass @option{-z text} to the linker when linking a
6259 shared object. Using this option, you can link position-dependent
6260 code into a shared object.
6262 @option{-mimpure-text} suppresses the ``relocations remain against
6263 allocatable but non-writable sections'' linker error message.
6264 However, the necessary relocations will trigger copy-on-write, and the
6265 shared object is not actually shared across processes. Instead of
6266 using @option{-mimpure-text}, you should compile all source code with
6267 @option{-fpic} or @option{-fPIC}.
6269 This option is only available on SunOS and Solaris.
6275 These two options select variations on the SPARC architecture.
6277 By default (unless specifically configured for the Fujitsu SPARClite),
6278 GCC generates code for the V7 variant of the SPARC architecture.
6280 @option{-mv8} will give you SPARC-V8 code. The only difference from V7
6281 code is that the compiler emits the integer multiply and integer
6282 divide instructions which exist in SPARC-V8 but not in SPARC-V7.
6284 @option{-msparclite} will give you SPARClite code. This adds the integer
6285 multiply, integer divide step and scan (@code{ffs}) instructions which
6286 exist in SPARClite but not in SPARC-V7.
6288 These options are deprecated and will be deleted in a future GCC release.
6289 They have been replaced with @option{-mcpu=xxx}.
6296 @opindex msupersparc
6299 These four options select the processor for which the code is optimized.
6301 With @option{-mcypress} (the default), the compiler optimizes code for the
6302 Cypress CY7C602 chip, as used in the SPARCStation/SPARCServer 3xx series.
6303 This is also appropriate for the older SPARCStation 1, 2, IPX etc.
6305 With @option{-msupersparc} the compiler optimizes code for the SuperSPARC cpu, as
6306 used in the SPARCStation 10, 1000 and 2000 series. This flag also enables use
6307 of the SPARC-V8 instruction set.
6309 With @option{-mf930} the compiler optimizes code for the Fujitsu MB86930 chip,
6310 which is the original SPARClite, with no FPU. This flag also enables use
6311 of the SPARClite instruction set and disables the FPU support.
6313 With @option{-mf934} the compiler optimizes code for the Fujitsu MB86934 chip,
6314 which is the more recent SPARClite with FPU. This flag also enables use
6315 of the SPARClite instruction set and enables the FPU support.
6317 These options are deprecated and will be deleted in a future GCC release.
6318 They have been replaced with @option{-mcpu=xxx}.
6320 @item -mcpu=@var{cpu_type}
6322 Set the instruction set, register set, and instruction scheduling parameters
6323 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
6324 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
6325 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
6326 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
6329 Default instruction scheduling parameters are used for values that select
6330 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
6331 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
6333 Here is a list of each supported architecture and their supported
6338 v8: supersparc, hypersparc
6339 sparclite: f930, f934, sparclite86x
6341 v9: ultrasparc, ultrasparc3
6344 @item -mtune=@var{cpu_type}
6346 Set the instruction scheduling parameters for machine type
6347 @var{cpu_type}, but do not set the instruction set or register set that the
6348 option @option{-mcpu=@var{cpu_type}} would.
6350 The same values for @option{-mcpu=@var{cpu_type}} can be used for
6351 @option{-mtune=@var{cpu_type}}, but the only useful values are those
6352 that select a particular cpu implementation. Those are @samp{cypress},
6353 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
6354 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
6360 @opindex -mno-v8plus
6361 @option{-mv8plus} will give you SPARC-V8+ code. The difference from V8 is
6362 that the global and out registers are considered 64-bit wide. This is the
6363 default on Solaris in 32-bit mode for all SPARC-V9 processors.
6369 @option{-mvis} will give you code that takes advantage of the UltraSPARC
6370 Visual Instruction Set extensions. The default is @option{-mno-vis}.
6373 These @samp{-m} options are supported in addition to the above
6374 on SPARC-V9 processors in 64-bit environments:
6377 @item -mlittle-endian
6378 @opindex mlittle-endian
6379 Generate code for a processor running in little-endian mode. It is only
6380 available for a few configurations and most notably not on Solaris.
6386 Generate code for a 32-bit or 64-bit environment.
6387 The 32-bit environment sets int, long and pointer to 32 bits.
6388 The 64-bit environment sets int to 32 bits and long and pointer
6391 @item -mcmodel=medlow
6392 @opindex mcmodel=medlow
6393 Generate code for the Medium/Low code model: the program must be linked
6394 in the low 32 bits of the address space. Pointers are 64 bits.
6395 Programs can be statically or dynamically linked.
6397 @item -mcmodel=medmid
6398 @opindex mcmodel=medmid
6399 Generate code for the Medium/Middle code model: the program must be linked
6400 in the low 44 bits of the address space, the text segment must be less than
6401 2G bytes, and data segment must be within 2G of the text segment.
6402 Pointers are 64 bits.
6404 @item -mcmodel=medany
6405 @opindex mcmodel=medany
6406 Generate code for the Medium/Anywhere code model: the program may be linked
6407 anywhere in the address space, the text segment must be less than
6408 2G bytes, and data segment must be within 2G of the text segment.
6409 Pointers are 64 bits.
6411 @item -mcmodel=embmedany
6412 @opindex mcmodel=embmedany
6413 Generate code for the Medium/Anywhere code model for embedded systems:
6414 assume a 32-bit text and a 32-bit data segment, both starting anywhere
6415 (determined at link time). Register %g4 points to the base of the
6416 data segment. Pointers are still 64 bits.
6417 Programs are statically linked, PIC is not supported.
6420 @itemx -mno-stack-bias
6421 @opindex mstack-bias
6422 @opindex mno-stack-bias
6423 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
6424 frame pointer if present, are offset by @minus{}2047 which must be added back
6425 when making stack frame references. This is the default in 64-bit mode.
6426 Otherwise, assume no such offset is present.
6430 @subsection ARM Options
6433 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6438 @opindex mapcs-frame
6439 Generate a stack frame that is compliant with the ARM Procedure Call
6440 Standard for all functions, even if this is not strictly necessary for
6441 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6442 with this option will cause the stack frames not to be generated for
6443 leaf functions. The default is @option{-mno-apcs-frame}.
6447 This is a synonym for @option{-mapcs-frame}.
6451 Generate code for a processor running with a 26-bit program counter,
6452 and conforming to the function calling standards for the APCS 26-bit
6453 option. This option replaces the @option{-m2} and @option{-m3} options
6454 of previous releases of the compiler.
6458 Generate code for a processor running with a 32-bit program counter,
6459 and conforming to the function calling standards for the APCS 32-bit
6460 option. This option replaces the @option{-m6} option of previous releases
6464 @c not currently implemented
6465 @item -mapcs-stack-check
6466 @opindex mapcs-stack-check
6467 Generate code to check the amount of stack space available upon entry to
6468 every function (that actually uses some stack space). If there is
6469 insufficient space available then either the function
6470 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6471 called, depending upon the amount of stack space required. The run time
6472 system is required to provide these functions. The default is
6473 @option{-mno-apcs-stack-check}, since this produces smaller code.
6475 @c not currently implemented
6477 @opindex mapcs-float
6478 Pass floating point arguments using the float point registers. This is
6479 one of the variants of the APCS@. This option is recommended if the
6480 target hardware has a floating point unit or if a lot of floating point
6481 arithmetic is going to be performed by the code. The default is
6482 @option{-mno-apcs-float}, since integer only code is slightly increased in
6483 size if @option{-mapcs-float} is used.
6485 @c not currently implemented
6486 @item -mapcs-reentrant
6487 @opindex mapcs-reentrant
6488 Generate reentrant, position independent code. The default is
6489 @option{-mno-apcs-reentrant}.
6492 @item -mthumb-interwork
6493 @opindex mthumb-interwork
6494 Generate code which supports calling between the ARM and Thumb
6495 instruction sets. Without this option the two instruction sets cannot
6496 be reliably used inside one program. The default is
6497 @option{-mno-thumb-interwork}, since slightly larger code is generated
6498 when @option{-mthumb-interwork} is specified.
6500 @item -mno-sched-prolog
6501 @opindex mno-sched-prolog
6502 Prevent the reordering of instructions in the function prolog, or the
6503 merging of those instruction with the instructions in the function's
6504 body. This means that all functions will start with a recognizable set
6505 of instructions (or in fact one of a choice from a small set of
6506 different function prologues), and this information can be used to
6507 locate the start if functions inside an executable piece of code. The
6508 default is @option{-msched-prolog}.
6511 @opindex mhard-float
6512 Generate output containing floating point instructions. This is the
6516 @opindex msoft-float
6517 Generate output containing library calls for floating point.
6518 @strong{Warning:} the requisite libraries are not available for all ARM
6519 targets. Normally the facilities of the machine's usual C compiler are
6520 used, but this cannot be done directly in cross-compilation. You must make
6521 your own arrangements to provide suitable library functions for
6524 @option{-msoft-float} changes the calling convention in the output file;
6525 therefore, it is only useful if you compile @emph{all} of a program with
6526 this option. In particular, you need to compile @file{libgcc.a}, the
6527 library that comes with GCC, with @option{-msoft-float} in order for
6530 @item -mlittle-endian
6531 @opindex mlittle-endian
6532 Generate code for a processor running in little-endian mode. This is
6533 the default for all standard configurations.
6536 @opindex mbig-endian
6537 Generate code for a processor running in big-endian mode; the default is
6538 to compile code for a little-endian processor.
6540 @item -mwords-little-endian
6541 @opindex mwords-little-endian
6542 This option only applies when generating code for big-endian processors.
6543 Generate code for a little-endian word order but a big-endian byte
6544 order. That is, a byte order of the form @samp{32107654}. Note: this
6545 option should only be used if you require compatibility with code for
6546 big-endian ARM processors generated by versions of the compiler prior to
6549 @item -malignment-traps
6550 @opindex malignment-traps
6551 Generate code that will not trap if the MMU has alignment traps enabled.
6552 On ARM architectures prior to ARMv4, there were no instructions to
6553 access half-word objects stored in memory. However, when reading from
6554 memory a feature of the ARM architecture allows a word load to be used,
6555 even if the address is unaligned, and the processor core will rotate the
6556 data as it is being loaded. This option tells the compiler that such
6557 misaligned accesses will cause a MMU trap and that it should instead
6558 synthesize the access as a series of byte accesses. The compiler can
6559 still use word accesses to load half-word data if it knows that the
6560 address is aligned to a word boundary.
6562 This option is ignored when compiling for ARM architecture 4 or later,
6563 since these processors have instructions to directly access half-word
6566 @item -mno-alignment-traps
6567 @opindex mno-alignment-traps
6568 Generate code that assumes that the MMU will not trap unaligned
6569 accesses. This produces better code when the target instruction set
6570 does not have half-word memory operations (i.e.@: implementations prior to
6573 Note that you cannot use this option to access unaligned word objects,
6574 since the processor will only fetch one 32-bit aligned object from
6577 The default setting for most targets is @option{-mno-alignment-traps}, since
6578 this produces better code when there are no half-word memory
6579 instructions available.
6581 @item -mshort-load-bytes
6582 @itemx -mno-short-load-words
6583 @opindex mshort-load-bytes
6584 @opindex mno-short-load-words
6585 These are deprecated aliases for @option{-malignment-traps}.
6587 @item -mno-short-load-bytes
6588 @itemx -mshort-load-words
6589 @opindex mno-short-load-bytes
6590 @opindex mshort-load-words
6591 This are deprecated aliases for @option{-mno-alignment-traps}.
6593 @item -mcpu=@var{name}
6595 This specifies the name of the target ARM processor. GCC uses this name
6596 to determine what kind of instructions it can emit when generating
6597 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6598 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6599 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6600 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6601 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6602 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6603 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6604 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6605 @samp{arm920t}, @samp{arm926ejs}, @samp{arm940t}, @samp{arm9tdmi},
6606 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ejs},
6607 @samp{arm1136js}, @samp{arm1136jfs} ,@samp{xscale}, @samp{iwmmxt},
6610 @itemx -mtune=@var{name}
6612 This option is very similar to the @option{-mcpu=} option, except that
6613 instead of specifying the actual target processor type, and hence
6614 restricting which instructions can be used, it specifies that GCC should
6615 tune the performance of the code as if the target were of the type
6616 specified in this option, but still choosing the instructions that it
6617 will generate based on the cpu specified by a @option{-mcpu=} option.
6618 For some ARM implementations better performance can be obtained by using
6621 @item -march=@var{name}
6623 This specifies the name of the target ARM architecture. GCC uses this
6624 name to determine what kind of instructions it can emit when generating
6625 assembly code. This option can be used in conjunction with or instead
6626 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6627 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6628 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6j},
6629 @samp{iwmmxt}, @samp{ep9312}.
6631 @item -mfpe=@var{number}
6632 @itemx -mfp=@var{number}
6635 This specifies the version of the floating point emulation available on
6636 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6637 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6639 @item -mstructure-size-boundary=@var{n}
6640 @opindex mstructure-size-boundary
6641 The size of all structures and unions will be rounded up to a multiple
6642 of the number of bits set by this option. Permissible values are 8 and
6643 32. The default value varies for different toolchains. For the COFF
6644 targeted toolchain the default value is 8. Specifying the larger number
6645 can produce faster, more efficient code, but can also increase the size
6646 of the program. The two values are potentially incompatible. Code
6647 compiled with one value cannot necessarily expect to work with code or
6648 libraries compiled with the other value, if they exchange information
6649 using structures or unions.
6651 @item -mabort-on-noreturn
6652 @opindex mabort-on-noreturn
6653 Generate a call to the function @code{abort} at the end of a
6654 @code{noreturn} function. It will be executed if the function tries to
6658 @itemx -mno-long-calls
6659 @opindex mlong-calls
6660 @opindex mno-long-calls
6661 Tells the compiler to perform function calls by first loading the
6662 address of the function into a register and then performing a subroutine
6663 call on this register. This switch is needed if the target function
6664 will lie outside of the 64 megabyte addressing range of the offset based
6665 version of subroutine call instruction.
6667 Even if this switch is enabled, not all function calls will be turned
6668 into long calls. The heuristic is that static functions, functions
6669 which have the @samp{short-call} attribute, functions that are inside
6670 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6671 definitions have already been compiled within the current compilation
6672 unit, will not be turned into long calls. The exception to this rule is
6673 that weak function definitions, functions with the @samp{long-call}
6674 attribute or the @samp{section} attribute, and functions that are within
6675 the scope of a @samp{#pragma long_calls} directive, will always be
6676 turned into long calls.
6678 This feature is not enabled by default. Specifying
6679 @option{-mno-long-calls} will restore the default behavior, as will
6680 placing the function calls within the scope of a @samp{#pragma
6681 long_calls_off} directive. Note these switches have no effect on how
6682 the compiler generates code to handle function calls via function
6685 @item -mnop-fun-dllimport
6686 @opindex mnop-fun-dllimport
6687 Disable support for the @code{dllimport} attribute.
6689 @item -msingle-pic-base
6690 @opindex msingle-pic-base
6691 Treat the register used for PIC addressing as read-only, rather than
6692 loading it in the prologue for each function. The run-time system is
6693 responsible for initializing this register with an appropriate value
6694 before execution begins.
6696 @item -mpic-register=@var{reg}
6697 @opindex mpic-register
6698 Specify the register to be used for PIC addressing. The default is R10
6699 unless stack-checking is enabled, when R9 is used.
6701 @item -mcirrus-fix-invalid-insns
6702 @opindex mcirrus-fix-invalid-insns
6703 @opindex mno-cirrus-fix-invalid-insns
6704 Insert NOPs into the instruction stream to in order to work around
6705 problems with invalid Maverick instruction combinations. This option
6706 is only valid if the @option{-mcpu=ep9312} option has been used to
6707 enable generation of instructions for the Cirrus Maverick floating
6708 point co-processor. This option is not enabled by default, since the
6709 problem is only present in older Maverick implementations. The default
6710 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6713 @item -mpoke-function-name
6714 @opindex mpoke-function-name
6715 Write the name of each function into the text section, directly
6716 preceding the function prologue. The generated code is similar to this:
6720 .ascii "arm_poke_function_name", 0
6723 .word 0xff000000 + (t1 - t0)
6724 arm_poke_function_name
6726 stmfd sp!, @{fp, ip, lr, pc@}
6730 When performing a stack backtrace, code can inspect the value of
6731 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6732 location @code{pc - 12} and the top 8 bits are set, then we know that
6733 there is a function name embedded immediately preceding this location
6734 and has length @code{((pc[-3]) & 0xff000000)}.
6738 Generate code for the 16-bit Thumb instruction set. The default is to
6739 use the 32-bit ARM instruction set.
6742 @opindex mtpcs-frame
6743 Generate a stack frame that is compliant with the Thumb Procedure Call
6744 Standard for all non-leaf functions. (A leaf function is one that does
6745 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6747 @item -mtpcs-leaf-frame
6748 @opindex mtpcs-leaf-frame
6749 Generate a stack frame that is compliant with the Thumb Procedure Call
6750 Standard for all leaf functions. (A leaf function is one that does
6751 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6753 @item -mcallee-super-interworking
6754 @opindex mcallee-super-interworking
6755 Gives all externally visible functions in the file being compiled an ARM
6756 instruction set header which switches to Thumb mode before executing the
6757 rest of the function. This allows these functions to be called from
6758 non-interworking code.
6760 @item -mcaller-super-interworking
6761 @opindex mcaller-super-interworking
6762 Allows calls via function pointers (including virtual functions) to
6763 execute correctly regardless of whether the target code has been
6764 compiled for interworking or not. There is a small overhead in the cost
6765 of executing a function pointer if this option is enabled.
6769 @node MN10300 Options
6770 @subsection MN10300 Options
6771 @cindex MN10300 options
6773 These @option{-m} options are defined for Matsushita MN10300 architectures:
6778 Generate code to avoid bugs in the multiply instructions for the MN10300
6779 processors. This is the default.
6782 @opindex mno-mult-bug
6783 Do not generate code to avoid bugs in the multiply instructions for the
6788 Generate code which uses features specific to the AM33 processor.
6792 Do not generate code which uses features specific to the AM33 processor. This
6797 Do not link in the C run-time initialization object file.
6801 Indicate to the linker that it should perform a relaxation optimization pass
6802 to shorten branches, calls and absolute memory addresses. This option only
6803 has an effect when used on the command line for the final link step.
6805 This option makes symbolic debugging impossible.
6809 @node M32R/D Options
6810 @subsection M32R/D Options
6811 @cindex M32R/D options
6813 These @option{-m} options are defined for Renesas M32R/D architectures:
6818 Generate code for the M32R/2@.
6822 Generate code for the M32R/X@.
6826 Generate code for the M32R@. This is the default.
6829 @opindex mmodel=small
6830 Assume all objects live in the lower 16MB of memory (so that their addresses
6831 can be loaded with the @code{ld24} instruction), and assume all subroutines
6832 are reachable with the @code{bl} instruction.
6833 This is the default.
6835 The addressability of a particular object can be set with the
6836 @code{model} attribute.
6838 @item -mmodel=medium
6839 @opindex mmodel=medium
6840 Assume objects may be anywhere in the 32-bit address space (the compiler
6841 will generate @code{seth/add3} instructions to load their addresses), and
6842 assume all subroutines are reachable with the @code{bl} instruction.
6845 @opindex mmodel=large
6846 Assume objects may be anywhere in the 32-bit address space (the compiler
6847 will generate @code{seth/add3} instructions to load their addresses), and
6848 assume subroutines may not be reachable with the @code{bl} instruction
6849 (the compiler will generate the much slower @code{seth/add3/jl}
6850 instruction sequence).
6853 @opindex msdata=none
6854 Disable use of the small data area. Variables will be put into
6855 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6856 @code{section} attribute has been specified).
6857 This is the default.
6859 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6860 Objects may be explicitly put in the small data area with the
6861 @code{section} attribute using one of these sections.
6864 @opindex msdata=sdata
6865 Put small global and static data in the small data area, but do not
6866 generate special code to reference them.
6870 Put small global and static data in the small data area, and generate
6871 special instructions to reference them.
6875 @cindex smaller data references
6876 Put global and static objects less than or equal to @var{num} bytes
6877 into the small data or bss sections instead of the normal data or bss
6878 sections. The default value of @var{num} is 8.
6879 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6880 for this option to have any effect.
6882 All modules should be compiled with the same @option{-G @var{num}} value.
6883 Compiling with different values of @var{num} may or may not work; if it
6884 doesn't the linker will give an error message---incorrect code will not be
6889 Makes the M32R specific code in the compiler display some statistics
6890 that might help in debugging programs.
6893 @opindex malign-loops
6894 Align all loops to a 32-byte boundary.
6896 @item -mno-align-loops
6897 @opindex mno-align-loops
6898 Do not enforce a 32-byte alignment for loops. This is the default.
6900 @item -missue-rate=@var{number}
6901 @opindex missue-rate=@var{number}
6902 Issue @var{number} instructions per cycle. @var{number} can only be 1
6905 @item -mbranch-cost=@var{number}
6906 @opindex mbranch-cost=@var{number}
6907 @var{number} can only be 1 or 2. If it is 1 then branches will be
6908 preferred over conditional code, if it is 2, then the opposite will
6911 @item -mflush-trap=@var{number}
6912 @opindex mflush-trap=@var{number}
6913 Specifies the trap number to use to flush the cache. The default is
6914 12. Valid numbers are between 0 and 15 inclusive.
6916 @item -mno-flush-trap
6917 @opindex mno-flush-trap
6918 Specifies that the cache cannot be flushed by using a trap.
6920 @item -mflush-func=@var{name}
6921 @opindex mflush-func=@var{name}
6922 Specifies the name of the operating system function to call to flush
6923 the cache. The default is @emph{_flush_cache}, but a function call
6924 will only be used if a trap is not available.
6926 @item -mno-flush-func
6927 @opindex mno-flush-func
6928 Indicates that there is no OS function for flushing the cache.
6932 @node RS/6000 and PowerPC Options
6933 @subsection IBM RS/6000 and PowerPC Options
6934 @cindex RS/6000 and PowerPC Options
6935 @cindex IBM RS/6000 and PowerPC Options
6937 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6945 @itemx -mpowerpc-gpopt
6946 @itemx -mno-powerpc-gpopt
6947 @itemx -mpowerpc-gfxopt
6948 @itemx -mno-powerpc-gfxopt
6950 @itemx -mno-powerpc64
6956 @opindex mno-powerpc
6957 @opindex mpowerpc-gpopt
6958 @opindex mno-powerpc-gpopt
6959 @opindex mpowerpc-gfxopt
6960 @opindex mno-powerpc-gfxopt
6962 @opindex mno-powerpc64
6963 GCC supports two related instruction set architectures for the
6964 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6965 instructions supported by the @samp{rios} chip set used in the original
6966 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6967 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6968 the IBM 4xx microprocessors.
6970 Neither architecture is a subset of the other. However there is a
6971 large common subset of instructions supported by both. An MQ
6972 register is included in processors supporting the POWER architecture.
6974 You use these options to specify which instructions are available on the
6975 processor you are using. The default value of these options is
6976 determined when configuring GCC@. Specifying the
6977 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6978 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6979 rather than the options listed above.
6981 The @option{-mpower} option allows GCC to generate instructions that
6982 are found only in the POWER architecture and to use the MQ register.
6983 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
6984 to generate instructions that are present in the POWER2 architecture but
6985 not the original POWER architecture.
6987 The @option{-mpowerpc} option allows GCC to generate instructions that
6988 are found only in the 32-bit subset of the PowerPC architecture.
6989 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
6990 GCC to use the optional PowerPC architecture instructions in the
6991 General Purpose group, including floating-point square root. Specifying
6992 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
6993 use the optional PowerPC architecture instructions in the Graphics
6994 group, including floating-point select.
6996 The @option{-mpowerpc64} option allows GCC to generate the additional
6997 64-bit instructions that are found in the full PowerPC64 architecture
6998 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6999 @option{-mno-powerpc64}.
7001 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
7002 will use only the instructions in the common subset of both
7003 architectures plus some special AIX common-mode calls, and will not use
7004 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
7005 permits GCC to use any instruction from either architecture and to
7006 allow use of the MQ register; specify this for the Motorola MPC601.
7008 @item -mnew-mnemonics
7009 @itemx -mold-mnemonics
7010 @opindex mnew-mnemonics
7011 @opindex mold-mnemonics
7012 Select which mnemonics to use in the generated assembler code. With
7013 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
7014 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
7015 assembler mnemonics defined for the POWER architecture. Instructions
7016 defined in only one architecture have only one mnemonic; GCC uses that
7017 mnemonic irrespective of which of these options is specified.
7019 GCC defaults to the mnemonics appropriate for the architecture in
7020 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
7021 value of these option. Unless you are building a cross-compiler, you
7022 should normally not specify either @option{-mnew-mnemonics} or
7023 @option{-mold-mnemonics}, but should instead accept the default.
7025 @item -mcpu=@var{cpu_type}
7027 Set architecture type, register usage, choice of mnemonics, and
7028 instruction scheduling parameters for machine type @var{cpu_type}.
7029 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
7030 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
7031 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
7032 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
7033 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
7034 @samp{860}, @samp{970}, @samp{common}, @samp{ec603e}, @samp{G3},
7035 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
7036 @samp{power4}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
7037 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64a}.
7039 @option{-mcpu=common} selects a completely generic processor. Code
7040 generated under this option will run on any POWER or PowerPC processor.
7041 GCC will use only the instructions in the common subset of both
7042 architectures, and will not use the MQ register. GCC assumes a generic
7043 processor model for scheduling purposes.
7045 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
7046 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
7047 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
7048 types, with an appropriate, generic processor model assumed for
7049 scheduling purposes.
7051 The other options specify a specific processor. Code generated under
7052 those options will run best on that processor, and may not run at all on
7055 The @option{-mcpu} options automatically enable or disable the
7056 following options: @option{-maltivec}, @option{-mhard-float},
7057 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
7058 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
7059 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
7060 @option{-mstring}. The particular options set for any particular CPU
7061 will vary between compiler versions, depending on what setting seems
7062 to produce optimal code for that CPU; it doesn't necessarily reflect
7063 the actual hardware's capabilities. If you wish to set an individual
7064 option to a particular value, you may specify it after the
7065 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
7067 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
7068 not enabled or disabled by the @option{-mcpu} option at present, since
7069 AIX does not have full support for these options. You may still
7070 enable or disable them individually if you're sure it'll work in your
7073 @item -mtune=@var{cpu_type}
7075 Set the instruction scheduling parameters for machine type
7076 @var{cpu_type}, but do not set the architecture type, register usage, or
7077 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
7078 values for @var{cpu_type} are used for @option{-mtune} as for
7079 @option{-mcpu}. If both are specified, the code generated will use the
7080 architecture, registers, and mnemonics set by @option{-mcpu}, but the
7081 scheduling parameters set by @option{-mtune}.
7086 @opindex mno-altivec
7087 These switches enable or disable the use of built-in functions that
7088 allow access to the AltiVec instruction set. You may also need to set
7089 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
7094 Extend the current ABI with SPE ABI extensions. This does not change
7095 the default ABI, instead it adds the SPE ABI extensions to the current
7099 @opindex mabi=no-spe
7100 Disable Booke SPE ABI extensions for the current ABI.
7102 @item -misel=@var{yes/no}
7105 This switch enables or disables the generation of ISEL instructions.
7107 @item -mspe=@var{yes/no}
7110 This switch enables or disables the generation of SPE simd
7113 @item -mfloat-gprs=@var{yes/no}
7115 @opindex mfloat-gprs
7116 This switch enables or disables the generation of floating point
7117 operations on the general purpose registers for architectures that
7118 support it. This option is currently only available on the MPC8540.
7121 @itemx -mno-fp-in-toc
7122 @itemx -mno-sum-in-toc
7123 @itemx -mminimal-toc
7125 @opindex mno-fp-in-toc
7126 @opindex mno-sum-in-toc
7127 @opindex mminimal-toc
7128 Modify generation of the TOC (Table Of Contents), which is created for
7129 every executable file. The @option{-mfull-toc} option is selected by
7130 default. In that case, GCC will allocate at least one TOC entry for
7131 each unique non-automatic variable reference in your program. GCC
7132 will also place floating-point constants in the TOC@. However, only
7133 16,384 entries are available in the TOC@.
7135 If you receive a linker error message that saying you have overflowed
7136 the available TOC space, you can reduce the amount of TOC space used
7137 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
7138 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
7139 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
7140 generate code to calculate the sum of an address and a constant at
7141 run-time instead of putting that sum into the TOC@. You may specify one
7142 or both of these options. Each causes GCC to produce very slightly
7143 slower and larger code at the expense of conserving TOC space.
7145 If you still run out of space in the TOC even when you specify both of
7146 these options, specify @option{-mminimal-toc} instead. This option causes
7147 GCC to make only one TOC entry for every file. When you specify this
7148 option, GCC will produce code that is slower and larger but which
7149 uses extremely little TOC space. You may wish to use this option
7150 only on files that contain less frequently executed code.
7156 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
7157 @code{long} type, and the infrastructure needed to support them.
7158 Specifying @option{-maix64} implies @option{-mpowerpc64} and
7159 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
7160 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
7165 @opindex mno-xl-call
7166 On AIX, pass floating-point arguments to prototyped functions beyond the
7167 register save area (RSA) on the stack in addition to argument FPRs. The
7168 AIX calling convention was extended but not initially documented to
7169 handle an obscure K&R C case of calling a function that takes the
7170 address of its arguments with fewer arguments than declared. AIX XL
7171 compilers access floating point arguments which do not fit in the
7172 RSA from the stack when a subroutine is compiled without
7173 optimization. Because always storing floating-point arguments on the
7174 stack is inefficient and rarely needed, this option is not enabled by
7175 default and only is necessary when calling subroutines compiled by AIX
7176 XL compilers without optimization.
7180 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
7181 application written to use message passing with special startup code to
7182 enable the application to run. The system must have PE installed in the
7183 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
7184 must be overridden with the @option{-specs=} option to specify the
7185 appropriate directory location. The Parallel Environment does not
7186 support threads, so the @option{-mpe} option and the @option{-pthread}
7187 option are incompatible.
7189 @item -malign-natural
7190 @itemx -malign-power
7191 @opindex malign-natural
7192 @opindex malign-power
7193 On AIX, Darwin, and 64-bit PowerPC GNU/Linux, the option
7194 @option{-malign-natural} overrides the ABI-defined alignment of larger
7195 types, such as floating-point doubles, on their natural size-based boundary.
7196 The option @option{-malign-power} instructs GCC to follow the ABI-specified
7197 alignment rules. GCC defaults to the standard alignment defined in the ABI.
7201 @opindex msoft-float
7202 @opindex mhard-float
7203 Generate code that does not use (uses) the floating-point register set.
7204 Software floating point emulation is provided if you use the
7205 @option{-msoft-float} option, and pass the option to GCC when linking.
7208 @itemx -mno-multiple
7210 @opindex mno-multiple
7211 Generate code that uses (does not use) the load multiple word
7212 instructions and the store multiple word instructions. These
7213 instructions are generated by default on POWER systems, and not
7214 generated on PowerPC systems. Do not use @option{-mmultiple} on little
7215 endian PowerPC systems, since those instructions do not work when the
7216 processor is in little endian mode. The exceptions are PPC740 and
7217 PPC750 which permit the instructions usage in little endian mode.
7223 Generate code that uses (does not use) the load string instructions
7224 and the store string word instructions to save multiple registers and
7225 do small block moves. These instructions are generated by default on
7226 POWER systems, and not generated on PowerPC systems. Do not use
7227 @option{-mstring} on little endian PowerPC systems, since those
7228 instructions do not work when the processor is in little endian mode.
7229 The exceptions are PPC740 and PPC750 which permit the instructions
7230 usage in little endian mode.
7236 Generate code that uses (does not use) the load or store instructions
7237 that update the base register to the address of the calculated memory
7238 location. These instructions are generated by default. If you use
7239 @option{-mno-update}, there is a small window between the time that the
7240 stack pointer is updated and the address of the previous frame is
7241 stored, which means code that walks the stack frame across interrupts or
7242 signals may get corrupted data.
7245 @itemx -mno-fused-madd
7246 @opindex mfused-madd
7247 @opindex mno-fused-madd
7248 Generate code that uses (does not use) the floating point multiply and
7249 accumulate instructions. These instructions are generated by default if
7250 hardware floating is used.
7252 @item -mno-bit-align
7254 @opindex mno-bit-align
7256 On System V.4 and embedded PowerPC systems do not (do) force structures
7257 and unions that contain bit-fields to be aligned to the base type of the
7260 For example, by default a structure containing nothing but 8
7261 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7262 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7263 the structure would be aligned to a 1 byte boundary and be one byte in
7266 @item -mno-strict-align
7267 @itemx -mstrict-align
7268 @opindex mno-strict-align
7269 @opindex mstrict-align
7270 On System V.4 and embedded PowerPC systems do not (do) assume that
7271 unaligned memory references will be handled by the system.
7274 @itemx -mno-relocatable
7275 @opindex mrelocatable
7276 @opindex mno-relocatable
7277 On embedded PowerPC systems generate code that allows (does not allow)
7278 the program to be relocated to a different address at runtime. If you
7279 use @option{-mrelocatable} on any module, all objects linked together must
7280 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7282 @item -mrelocatable-lib
7283 @itemx -mno-relocatable-lib
7284 @opindex mrelocatable-lib
7285 @opindex mno-relocatable-lib
7286 On embedded PowerPC systems generate code that allows (does not allow)
7287 the program to be relocated to a different address at runtime. Modules
7288 compiled with @option{-mrelocatable-lib} can be linked with either modules
7289 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7290 with modules compiled with the @option{-mrelocatable} options.
7296 On System V.4 and embedded PowerPC systems do not (do) assume that
7297 register 2 contains a pointer to a global area pointing to the addresses
7298 used in the program.
7301 @itemx -mlittle-endian
7303 @opindex mlittle-endian
7304 On System V.4 and embedded PowerPC systems compile code for the
7305 processor in little endian mode. The @option{-mlittle-endian} option is
7306 the same as @option{-mlittle}.
7311 @opindex mbig-endian
7312 On System V.4 and embedded PowerPC systems compile code for the
7313 processor in big endian mode. The @option{-mbig-endian} option is
7314 the same as @option{-mbig}.
7316 @item -mdynamic-no-pic
7317 @opindex mdynamic-no-pic
7318 On Darwin and Mac OS X systems, compile code so that it is not
7319 relocatable, but that its external references are relocatable. The
7320 resulting code is suitable for applications, but not shared
7323 @item -mprioritize-restricted-insns=@var{priority}
7324 @opindex mprioritize-restricted-insns
7325 This option controls the priority that is assigned to
7326 dispatch-slot restricted instructions during the second scheduling
7327 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
7328 @var{no/highest/second-highest} priority to dispatch slot restricted
7331 @item -msched-costly-dep=@var{dependence_type}
7332 @opindex msched-costly-dep
7333 This option controls which dependences are considered costly
7334 by the target during instruction scheduling. The argument
7335 @var{dependence_type} takes one of the following values:
7336 @var{no}: no dependence is costly,
7337 @var{all}: all dependences are costly,
7338 @var{true_store_to_load}: a true dependence from store to load is costly,
7339 @var{store_to_load}: any dependence from store to load is costly,
7340 @var{number}: any dependence which latency >= @var{number} is costly.
7342 @item -minsert-sched-nops=@var{scheme}
7343 @opindex minsert-sched-nops
7344 This option controls which nop insertion scheme will be used during
7345 the second scheduling pass. The argument @var{scheme} takes one of the
7347 @var{no}: Don't insert nops.
7348 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
7349 according to the scheduler's grouping.
7350 @var{regroup_exact}: Insert nops to force costly dependent insns into
7351 separate groups. Insert exactly as many nops as needed to force an insn
7352 to a new group, according to the estimated processor grouping.
7353 @var{number}: Insert nops to force costly dependent insns into
7354 separate groups. Insert @var{number} nops to force an insn to a new group.
7358 On System V.4 and embedded PowerPC systems compile code using calling
7359 conventions that adheres to the March 1995 draft of the System V
7360 Application Binary Interface, PowerPC processor supplement. This is the
7361 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7363 @item -mcall-sysv-eabi
7364 @opindex mcall-sysv-eabi
7365 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7367 @item -mcall-sysv-noeabi
7368 @opindex mcall-sysv-noeabi
7369 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7371 @item -mcall-solaris
7372 @opindex mcall-solaris
7373 On System V.4 and embedded PowerPC systems compile code for the Solaris
7377 @opindex mcall-linux
7378 On System V.4 and embedded PowerPC systems compile code for the
7379 Linux-based GNU system.
7383 On System V.4 and embedded PowerPC systems compile code for the
7384 Hurd-based GNU system.
7387 @opindex mcall-netbsd
7388 On System V.4 and embedded PowerPC systems compile code for the
7389 NetBSD operating system.
7391 @item -maix-struct-return
7392 @opindex maix-struct-return
7393 Return all structures in memory (as specified by the AIX ABI)@.
7395 @item -msvr4-struct-return
7396 @opindex msvr4-struct-return
7397 Return structures smaller than 8 bytes in registers (as specified by the
7401 @opindex mabi=altivec
7402 Extend the current ABI with AltiVec ABI extensions. This does not
7403 change the default ABI, instead it adds the AltiVec ABI extensions to
7406 @item -mabi=no-altivec
7407 @opindex mabi=no-altivec
7408 Disable AltiVec ABI extensions for the current ABI.
7411 @itemx -mno-prototype
7413 @opindex mno-prototype
7414 On System V.4 and embedded PowerPC systems assume that all calls to
7415 variable argument functions are properly prototyped. Otherwise, the
7416 compiler must insert an instruction before every non prototyped call to
7417 set or clear bit 6 of the condition code register (@var{CR}) to
7418 indicate whether floating point values were passed in the floating point
7419 registers in case the function takes a variable arguments. With
7420 @option{-mprototype}, only calls to prototyped variable argument functions
7421 will set or clear the bit.
7425 On embedded PowerPC systems, assume that the startup module is called
7426 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7427 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7432 On embedded PowerPC systems, assume that the startup module is called
7433 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7438 On embedded PowerPC systems, assume that the startup module is called
7439 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7443 @opindex myellowknife
7444 On embedded PowerPC systems, assume that the startup module is called
7445 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7450 On System V.4 and embedded PowerPC systems, specify that you are
7451 compiling for a VxWorks system.
7455 Specify that you are compiling for the WindISS simulation environment.
7459 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7460 header to indicate that @samp{eabi} extended relocations are used.
7466 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7467 Embedded Applications Binary Interface (eabi) which is a set of
7468 modifications to the System V.4 specifications. Selecting @option{-meabi}
7469 means that the stack is aligned to an 8 byte boundary, a function
7470 @code{__eabi} is called to from @code{main} to set up the eabi
7471 environment, and the @option{-msdata} option can use both @code{r2} and
7472 @code{r13} to point to two separate small data areas. Selecting
7473 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7474 do not call an initialization function from @code{main}, and the
7475 @option{-msdata} option will only use @code{r13} to point to a single
7476 small data area. The @option{-meabi} option is on by default if you
7477 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7480 @opindex msdata=eabi
7481 On System V.4 and embedded PowerPC systems, put small initialized
7482 @code{const} global and static data in the @samp{.sdata2} section, which
7483 is pointed to by register @code{r2}. Put small initialized
7484 non-@code{const} global and static data in the @samp{.sdata} section,
7485 which is pointed to by register @code{r13}. Put small uninitialized
7486 global and static data in the @samp{.sbss} section, which is adjacent to
7487 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7488 incompatible with the @option{-mrelocatable} option. The
7489 @option{-msdata=eabi} option also sets the @option{-memb} option.
7492 @opindex msdata=sysv
7493 On System V.4 and embedded PowerPC systems, put small global and static
7494 data in the @samp{.sdata} section, which is pointed to by register
7495 @code{r13}. Put small uninitialized global and static data in the
7496 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7497 The @option{-msdata=sysv} option is incompatible with the
7498 @option{-mrelocatable} option.
7500 @item -msdata=default
7502 @opindex msdata=default
7504 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7505 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7506 same as @option{-msdata=sysv}.
7509 @opindex msdata-data
7510 On System V.4 and embedded PowerPC systems, put small global and static
7511 data in the @samp{.sdata} section. Put small uninitialized global and
7512 static data in the @samp{.sbss} section. Do not use register @code{r13}
7513 to address small data however. This is the default behavior unless
7514 other @option{-msdata} options are used.
7518 @opindex msdata=none
7520 On embedded PowerPC systems, put all initialized global and static data
7521 in the @samp{.data} section, and all uninitialized data in the
7522 @samp{.bss} section.
7526 @cindex smaller data references (PowerPC)
7527 @cindex .sdata/.sdata2 references (PowerPC)
7528 On embedded PowerPC systems, put global and static items less than or
7529 equal to @var{num} bytes into the small data or bss sections instead of
7530 the normal data or bss section. By default, @var{num} is 8. The
7531 @option{-G @var{num}} switch is also passed to the linker.
7532 All modules should be compiled with the same @option{-G @var{num}} value.
7535 @itemx -mno-regnames
7537 @opindex mno-regnames
7538 On System V.4 and embedded PowerPC systems do (do not) emit register
7539 names in the assembly language output using symbolic forms.
7542 @itemx -mno-longcall
7544 @opindex mno-longcall
7545 Default to making all function calls via pointers, so that functions
7546 which reside further than 64 megabytes (67,108,864 bytes) from the
7547 current location can be called. This setting can be overridden by the
7548 @code{shortcall} function attribute, or by @code{#pragma longcall(0)}.
7550 Some linkers are capable of detecting out-of-range calls and generating
7551 glue code on the fly. On these systems, long calls are unnecessary and
7552 generate slower code. As of this writing, the AIX linker can do this,
7553 as can the GNU linker for PowerPC/64. It is planned to add this feature
7554 to the GNU linker for 32-bit PowerPC systems as well.
7556 On Mach-O (Darwin) systems, this option directs the compiler emit to
7557 the glue for every direct call, and the Darwin linker decides whether
7558 to use or discard it.
7560 In the future, we may cause GCC to ignore all longcall specifications
7561 when the linker is known to generate glue.
7565 Adds support for multithreading with the @dfn{pthreads} library.
7566 This option sets flags for both the preprocessor and linker.
7570 @node Darwin Options
7571 @subsection Darwin Options
7572 @cindex Darwin options
7574 These options are defined for all architectures running the Darwin operating
7575 system. They are useful for compatibility with other Mac OS compilers.
7580 Loads all members of static archive libraries.
7581 See man ld(1) for more information.
7583 @item -arch_errors_fatal
7584 @opindex arch_errors_fatal
7585 Cause the errors having to do with files that have the wrong architecture
7589 @opindex bind_at_load
7590 Causes the output file to be marked such that the dynamic linker will
7591 bind all undefined references when the file is loaded or launched.
7595 Produce a Mach-o bundle format file.
7596 See man ld(1) for more information.
7598 @item -bundle_loader @var{executable}
7599 @opindex bundle_loader
7600 This specifies the @var{executable} that will be loading the build
7601 output file being linked. See man ld(1) for more information.
7603 @item -allowable_client @var{client_name}
7607 @itemx -compatibility_version
7608 @itemx -current_version
7609 @itemx -dependency-file
7611 @itemx -dylinker_install_name
7614 @itemx -exported_symbols_list
7616 @itemx -flat_namespace
7617 @itemx -force_cpusubtype_ALL
7618 @itemx -force_flat_namespace
7619 @itemx -headerpad_max_install_names
7622 @itemx -install_name
7623 @itemx -keep_private_externs
7624 @itemx -multi_module
7625 @itemx -multiply_defined
7626 @itemx -multiply_defined_unused
7628 @itemx -nofixprebinding
7631 @itemx -noseglinkedit
7632 @itemx -pagezero_size
7634 @itemx -prebind_all_twolevel_modules
7635 @itemx -private_bundle
7636 @itemx -read_only_relocs
7638 @itemx -sectobjectsymbols
7642 @itemx -sectobjectsymbols
7644 @itemx -seg_addr_table
7645 @itemx -seg_addr_table_filename
7648 @itemx -segs_read_only_addr
7649 @itemx -segs_read_write_addr
7650 @itemx -single_module
7653 @itemx -sub_umbrella
7654 @itemx -twolevel_namespace
7657 @itemx -unexported_symbols_list
7658 @itemx -weak_reference_mismatches
7661 @opindex allowable_client
7663 @opindex client_name
7664 @opindex compatibility_version
7665 @opindex current_version
7666 @opindex dependency-file
7668 @opindex dylinker_install_name
7671 @opindex exported_symbols_list
7673 @opindex flat_namespace
7674 @opindex force_cpusubtype_ALL
7675 @opindex force_flat_namespace
7676 @opindex headerpad_max_install_names
7679 @opindex install_name
7680 @opindex keep_private_externs
7681 @opindex multi_module
7682 @opindex multiply_defined
7683 @opindex multiply_defined_unused
7685 @opindex nofixprebinding
7686 @opindex nomultidefs
7688 @opindex noseglinkedit
7689 @opindex pagezero_size
7691 @opindex prebind_all_twolevel_modules
7692 @opindex private_bundle
7693 @opindex read_only_relocs
7695 @opindex sectobjectsymbols
7699 @opindex sectobjectsymbols
7701 @opindex seg_addr_table
7702 @opindex seg_addr_table_filename
7703 @opindex seglinkedit
7705 @opindex segs_read_only_addr
7706 @opindex segs_read_write_addr
7707 @opindex single_module
7709 @opindex sub_library
7710 @opindex sub_umbrella
7711 @opindex twolevel_namespace
7714 @opindex unexported_symbols_list
7715 @opindex weak_reference_mismatches
7716 @opindex whatsloaded
7718 These options are available for Darwin linker. Darwin linker man page
7719 describes them in detail.
7724 @subsection MIPS Options
7725 @cindex MIPS options
7731 Generate big-endian code.
7735 Generate little-endian code. This is the default for @samp{mips*el-*-*}
7738 @item -march=@var{arch}
7740 Generate code that will run on @var{arch}, which can be the name of a
7741 generic MIPS ISA, or the name of a particular processor.
7743 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
7744 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
7745 The processor names are:
7746 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
7748 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
7749 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
7753 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4300},
7754 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
7755 The special value @samp{from-abi} selects the
7756 most compatible architecture for the selected ABI (that is,
7757 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
7759 In processor names, a final @samp{000} can be abbreviated as @samp{k}
7760 (for example, @samp{-march=r2k}). Prefixes are optional, and
7761 @samp{vr} may be written @samp{r}.
7763 GCC defines two macros based on the value of this option. The first
7764 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
7765 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
7766 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
7767 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
7768 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
7770 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
7771 above. In other words, it will have the full prefix and will not
7772 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
7773 the macro names the resolved architecture (either @samp{"mips1"} or
7774 @samp{"mips3"}). It names the default architecture when no
7775 @option{-march} option is given.
7777 @item -mtune=@var{arch}
7779 Optimize for @var{arch}. Among other things, this option controls
7780 the way instructions are scheduled, and the perceived cost of arithmetic
7781 operations. The list of @var{arch} values is the same as for
7784 When this option is not used, GCC will optimize for the processor
7785 specified by @option{-march}. By using @option{-march} and
7786 @option{-mtune} together, it is possible to generate code that will
7787 run on a family of processors, but optimize the code for one
7788 particular member of that family.
7790 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
7791 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
7792 @samp{-march} ones described above.
7796 Equivalent to @samp{-march=mips1}.
7800 Equivalent to @samp{-march=mips2}.
7804 Equivalent to @samp{-march=mips3}.
7808 Equivalent to @samp{-march=mips4}.
7812 Equivalent to @samp{-march=mips32}.
7816 Equivalent to @samp{-march=mips32r2}.
7820 Equivalent to @samp{-march=mips64}.
7826 Use (do not use) the MIPS16 ISA.
7838 Generate code for the given ABI@.
7840 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
7841 generates 64-bit code when you select a 64-bit architecture, but you
7842 can use @option{-mgp32} to get 32-bit code instead.
7845 @itemx -mno-abicalls
7847 @opindex mno-abicalls
7848 Generate (do not generate) SVR4-style position-independent code.
7849 @option{-mabicalls} is the default for SVR4-based systems.
7855 Lift (do not lift) the usual restrictions on the size of the global
7858 GCC normally uses a single instruction to load values from the GOT.
7859 While this is relatively efficient, it will only work if the GOT
7860 is smaller than about 64k. Anything larger will cause the linker
7861 to report an error such as:
7863 @cindex relocation truncated to fit (MIPS)
7865 relocation truncated to fit: R_MIPS_GOT16 foobar
7868 If this happens, you should recompile your code with @option{-mxgot}.
7869 It should then work with very large GOTs, although it will also be
7870 less efficient, since it will take three instructions to fetch the
7871 value of a global symbol.
7873 Note that some linkers can create multiple GOTs. If you have such a
7874 linker, you should only need to use @option{-mxgot} when a single object
7875 file accesses more than 64k's worth of GOT entries. Very few do.
7877 These options have no effect unless GCC is generating position
7880 @item -membedded-pic
7881 @itemx -mno-embedded-pic
7882 @opindex membedded-pic
7883 @opindex mno-embedded-pic
7884 Generate (do not generate) position-independent code suitable for some
7885 embedded systems. All calls are made using PC relative addresses, and
7886 all data is addressed using the $gp register. No more than 65536
7887 bytes of global data may be used. This requires GNU as and GNU ld,
7888 which do most of the work.
7892 Assume that general-purpose registers are 32 bits wide.
7896 Assume that general-purpose registers are 64 bits wide.
7900 Assume that floating-point registers are 32 bits wide.
7904 Assume that floating-point registers are 64 bits wide.
7907 @opindex mhard-float
7908 Use floating-point coprocessor instructions.
7911 @opindex msoft-float
7912 Do not use floating-point coprocessor instructions. Implement
7913 floating-point calculations using library calls instead.
7915 @item -msingle-float
7916 @opindex msingle-float
7917 Assume that the floating-point coprocessor only supports single-precision
7920 @itemx -mdouble-float
7921 @opindex mdouble-float
7922 Assume that the floating-point coprocessor supports double-precision
7923 operations. This is the default.
7927 Force @code{int} and @code{long} types to be 64 bits wide. See
7928 @option{-mlong32} for an explanation of the default and the way
7929 that the pointer size is determined.
7933 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
7934 an explanation of the default and the way that the pointer size is
7939 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
7941 The default size of @code{int}s, @code{long}s and pointers depends on
7942 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
7943 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
7944 32-bit @code{long}s. Pointers are the same size as @code{long}s,
7945 or the same size as integer registers, whichever is smaller.
7949 @cindex smaller data references (MIPS)
7950 @cindex gp-relative references (MIPS)
7951 Put global and static items less than or equal to @var{num} bytes into
7952 the small data or bss section instead of the normal data or bss section.
7953 This allows the data to be accessed using a single instruction.
7955 All modules should be compiled with the same @option{-G @var{num}}
7958 @item -membedded-data
7959 @itemx -mno-embedded-data
7960 @opindex membedded-data
7961 @opindex mno-embedded-data
7962 Allocate variables to the read-only data section first if possible, then
7963 next in the small data section if possible, otherwise in data. This gives
7964 slightly slower code than the default, but reduces the amount of RAM required
7965 when executing, and thus may be preferred for some embedded systems.
7967 @item -muninit-const-in-rodata
7968 @itemx -mno-uninit-const-in-rodata
7969 @opindex muninit-const-in-rodata
7970 @opindex mno-uninit-const-in-rodata
7971 Put uninitialized @code{const} variables in the read-only data section.
7972 This option is only meaningful in conjunction with @option{-membedded-data}.
7974 @item -msplit-addresses
7975 @itemx -mno-split-addresses
7976 @opindex msplit-addresses
7977 @opindex mno-split-addresses
7978 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
7979 relocation operators. This option has been superceded by
7980 @option{-mexplicit-relocs} but is retained for backwards compatibility.
7982 @item -mexplicit-relocs
7983 @itemx -mno-explicit-relocs
7984 @opindex mexplicit-relocs
7985 @opindex mno-explicit-relocs
7986 Use (do not use) assembler relocation operators when dealing with symbolic
7987 addresses. The alternative, selected by @option{-mno-explicit-relocs},
7988 is to use assembler macros instead.
7990 @option{-mexplicit-relocs} is usually the default if GCC was
7991 configured to use an assembler that supports relocation operators.
7992 However, there are two exceptions:
7996 GCC is not yet able to generate explicit relocations for the combination
7997 of @option{-mabi=64} and @option{-mno-abicalls}. This will be addressed
7998 in a future release.
8001 The combination of @option{-mabicalls} and @option{-fno-unit-at-a-time}
8002 implies @option{-mno-explicit-relocs} unless explicitly overridden.
8003 This is because, when generating abicalls, the choice of relocation
8004 depends on whether a symbol is local or global. In some rare cases,
8005 GCC will not be able to decide this until the whole compilation unit
8013 Generate (do not generate) code that refers to registers using their
8014 software names. The default is @option{-mno-rnames}, which tells GCC
8015 to use hardware names like @samp{$4} instead of software names like
8016 @samp{a0}. The only assembler known to support @option{-rnames} is
8017 the Algorithmics assembler.
8019 @item -mcheck-zero-division
8020 @itemx -mno-check-zero-division
8021 @opindex mcheck-zero-division
8022 @opindex mno-check-zero-division
8023 Trap (do not trap) on integer division by zero. The default is
8024 @option{-mcheck-zero-division}.
8030 Force (do not force) the use of @code{memcpy()} for non-trivial block
8031 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
8032 most constant-sized copies.
8035 @itemx -mno-long-calls
8036 @opindex mlong-calls
8037 @opindex mno-long-calls
8038 Disable (do not disable) use of the @code{jal} instruction. Calling
8039 functions using @code{jal} is more efficient but requires the caller
8040 and callee to be in the same 256 megabyte segment.
8042 This option has no effect on abicalls code. The default is
8043 @option{-mno-long-calls}.
8049 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
8050 instructions, as provided by the R4650 ISA.
8053 @itemx -mno-fused-madd
8054 @opindex mfused-madd
8055 @opindex mno-fused-madd
8056 Enable (disable) use of the floating point multiply-accumulate
8057 instructions, when they are available. The default is
8058 @option{-mfused-madd}.
8060 When multiply-accumulate instructions are used, the intermediate
8061 product is calculated to infinite precision and is not subject to
8062 the FCSR Flush to Zero bit. This may be undesirable in some
8067 Tell the MIPS assembler to not run its preprocessor over user
8068 assembler files (with a @samp{.s} suffix) when assembling them.
8073 Work around certain SB-1 CPU core errata.
8074 (This flag currently works around the SB-1 revision 2
8075 ``F1'' and ``F2'' floating point errata.)
8077 @item -mflush-func=@var{func}
8078 @itemx -mno-flush-func
8079 @opindex mflush-func
8080 Specifies the function to call to flush the I and D caches, or to not
8081 call any such function. If called, the function must take the same
8082 arguments as the common @code{_flush_func()}, that is, the address of the
8083 memory range for which the cache is being flushed, the size of the
8084 memory range, and the number 3 (to flush both caches). The default
8085 depends on the target gcc was configured for, but commonly is either
8086 @samp{_flush_func} or @samp{__cpu_flush}.
8088 @item -mbranch-likely
8089 @itemx -mno-branch-likely
8090 @opindex mbranch-likely
8091 @opindex mno-branch-likely
8092 Enable or disable use of Branch Likely instructions, regardless of the
8093 default for the selected architecture. By default, Branch Likely
8094 instructions may be generated if they are supported by the selected
8095 architecture. An exception is for the MIPS32 and MIPS64 architectures
8096 and processors which implement those architectures; for those, Branch
8097 Likely instructions will not be generated by default because the MIPS32
8098 and MIPS64 architectures specifically deprecate their use.
8101 @node i386 and x86-64 Options
8102 @subsection Intel 386 and AMD x86-64 Options
8103 @cindex i386 Options
8104 @cindex x86-64 Options
8105 @cindex Intel 386 Options
8106 @cindex AMD x86-64 Options
8108 These @samp{-m} options are defined for the i386 and x86-64 family of
8112 @item -mtune=@var{cpu-type}
8114 Tune to @var{cpu-type} everything applicable about the generated code, except
8115 for the ABI and the set of available instructions. The choices for
8119 Original Intel's i386 CPU.
8121 Intel's i486 CPU. (No scheduling is implemented for this chip.)
8123 Intel Pentium CPU with no MMX support.
8125 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
8126 @item i686, pentiumpro
8127 Intel PentiumPro CPU.
8129 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
8131 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
8134 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
8136 AMD K6 CPU with MMX instruction set support.
8138 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
8139 @item athlon, athlon-tbird
8140 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
8142 @item athlon-4, athlon-xp, athlon-mp
8143 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
8144 instruction set support.
8145 @item k8, opteron, athlon64, athlon-fx
8146 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
8147 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
8149 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
8152 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
8153 instruction set support.
8155 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
8156 implemented for this chip.)
8158 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
8159 implemented for this chip.)
8162 While picking a specific @var{cpu-type} will schedule things appropriately
8163 for that particular chip, the compiler will not generate any code that
8164 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8167 @item -march=@var{cpu-type}
8169 Generate instructions for the machine type @var{cpu-type}. The choices
8170 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8171 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8173 @item -mcpu=@var{cpu-type}
8175 A deprecated synonym for @option{-mtune}.
8184 @opindex mpentiumpro
8185 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8186 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8187 These synonyms are deprecated.
8189 @item -mfpmath=@var{unit}
8191 generate floating point arithmetics for selected unit @var{unit}. the choices
8196 Use the standard 387 floating point coprocessor present majority of chips and
8197 emulated otherwise. Code compiled with this option will run almost everywhere.
8198 The temporary results are computed in 80bit precision instead of precision
8199 specified by the type resulting in slightly different results compared to most
8200 of other chips. See @option{-ffloat-store} for more detailed description.
8202 This is the default choice for i386 compiler.
8205 Use scalar floating point instructions present in the SSE instruction set.
8206 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8207 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8208 instruction set supports only single precision arithmetics, thus the double and
8209 extended precision arithmetics is still done using 387. Later version, present
8210 only in Pentium4 and the future AMD x86-64 chips supports double precision
8213 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8214 @option{-msse2} switches to enable SSE extensions and make this option
8215 effective. For x86-64 compiler, these extensions are enabled by default.
8217 The resulting code should be considerably faster in majority of cases and avoid
8218 the numerical instability problems of 387 code, but may break some existing
8219 code that expects temporaries to be 80bit.
8221 This is the default choice for x86-64 compiler.
8224 Use all SSE extensions enabled by @option{-msse2} as well as the new
8225 SSE extensions in Prescott New Instructions. @option{-mpni} also
8226 enables 2 builtin functions, @code{__builtin_ia32_monitor} and
8227 @code{__builtin_ia32_mwait}, for new instructions @code{monitor} and
8231 Attempt to utilize both instruction sets at once. This effectively double the
8232 amount of available registers and on chips with separate execution units for
8233 387 and SSE the execution resources too. Use this option with care, as it is
8234 still experimental, because gcc register allocator does not model separate
8235 functional units well resulting in instable performance.
8238 @item -masm=@var{dialect}
8239 @opindex masm=@var{dialect}
8240 Output asm instructions using selected @var{dialect}. Supported choices are
8241 @samp{intel} or @samp{att} (the default one).
8246 @opindex mno-ieee-fp
8247 Control whether or not the compiler uses IEEE floating point
8248 comparisons. These handle correctly the case where the result of a
8249 comparison is unordered.
8252 @opindex msoft-float
8253 Generate output containing library calls for floating point.
8254 @strong{Warning:} the requisite libraries are not part of GCC@.
8255 Normally the facilities of the machine's usual C compiler are used, but
8256 this can't be done directly in cross-compilation. You must make your
8257 own arrangements to provide suitable library functions for
8260 On machines where a function returns floating point results in the 80387
8261 register stack, some floating point opcodes may be emitted even if
8262 @option{-msoft-float} is used.
8264 @item -mno-fp-ret-in-387
8265 @opindex mno-fp-ret-in-387
8266 Do not use the FPU registers for return values of functions.
8268 The usual calling convention has functions return values of types
8269 @code{float} and @code{double} in an FPU register, even if there
8270 is no FPU@. The idea is that the operating system should emulate
8273 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8274 in ordinary CPU registers instead.
8276 @item -mno-fancy-math-387
8277 @opindex mno-fancy-math-387
8278 Some 387 emulators do not support the @code{sin}, @code{cos} and
8279 @code{sqrt} instructions for the 387. Specify this option to avoid
8280 generating those instructions. This option is the default on FreeBSD,
8281 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8282 indicates that the target cpu will always have an FPU and so the
8283 instruction will not need emulation. As of revision 2.6.1, these
8284 instructions are not generated unless you also use the
8285 @option{-funsafe-math-optimizations} switch.
8287 @item -malign-double
8288 @itemx -mno-align-double
8289 @opindex malign-double
8290 @opindex mno-align-double
8291 Control whether GCC aligns @code{double}, @code{long double}, and
8292 @code{long long} variables on a two word boundary or a one word
8293 boundary. Aligning @code{double} variables on a two word boundary will
8294 produce code that runs somewhat faster on a @samp{Pentium} at the
8295 expense of more memory.
8297 @strong{Warning:} if you use the @option{-malign-double} switch,
8298 structures containing the above types will be aligned differently than
8299 the published application binary interface specifications for the 386
8300 and will not be binary compatible with structures in code compiled
8301 without that switch.
8303 @item -m96bit-long-double
8304 @itemx -m128bit-long-double
8305 @opindex m96bit-long-double
8306 @opindex m128bit-long-double
8307 These switches control the size of @code{long double} type. The i386
8308 application binary interface specifies the size to be 96 bits,
8309 so @option{-m96bit-long-double} is the default in 32 bit mode.
8311 Modern architectures (Pentium and newer) would prefer @code{long double}
8312 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8313 conforming to the ABI, this would not be possible. So specifying a
8314 @option{-m128bit-long-double} will align @code{long double}
8315 to a 16 byte boundary by padding the @code{long double} with an additional
8318 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8319 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8321 Notice that neither of these options enable any extra precision over the x87
8322 standard of 80 bits for a @code{long double}.
8324 @strong{Warning:} if you override the default value for your target ABI, the
8325 structures and arrays containing @code{long double} will change their size as
8326 well as function calling convention for function taking @code{long double}
8327 will be modified. Hence they will not be binary compatible with arrays or
8328 structures in code compiled without that switch.
8332 @itemx -mno-svr3-shlib
8333 @opindex msvr3-shlib
8334 @opindex mno-svr3-shlib
8335 Control whether GCC places uninitialized local variables into the
8336 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8337 into @code{bss}. These options are meaningful only on System V Release 3.
8341 Use a different function-calling convention, in which functions that
8342 take a fixed number of arguments return with the @code{ret} @var{num}
8343 instruction, which pops their arguments while returning. This saves one
8344 instruction in the caller since there is no need to pop the arguments
8347 You can specify that an individual function is called with this calling
8348 sequence with the function attribute @samp{stdcall}. You can also
8349 override the @option{-mrtd} option by using the function attribute
8350 @samp{cdecl}. @xref{Function Attributes}.
8352 @strong{Warning:} this calling convention is incompatible with the one
8353 normally used on Unix, so you cannot use it if you need to call
8354 libraries compiled with the Unix compiler.
8356 Also, you must provide function prototypes for all functions that
8357 take variable numbers of arguments (including @code{printf});
8358 otherwise incorrect code will be generated for calls to those
8361 In addition, seriously incorrect code will result if you call a
8362 function with too many arguments. (Normally, extra arguments are
8363 harmlessly ignored.)
8365 @item -mregparm=@var{num}
8367 Control how many registers are used to pass integer arguments. By
8368 default, no registers are used to pass arguments, and at most 3
8369 registers can be used. You can control this behavior for a specific
8370 function by using the function attribute @samp{regparm}.
8371 @xref{Function Attributes}.
8373 @strong{Warning:} if you use this switch, and
8374 @var{num} is nonzero, then you must build all modules with the same
8375 value, including any libraries. This includes the system libraries and
8378 @item -mpreferred-stack-boundary=@var{num}
8379 @opindex mpreferred-stack-boundary
8380 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8381 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8382 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8383 size (@option{-Os}), in which case the default is the minimum correct
8384 alignment (4 bytes for x86, and 8 bytes for x86-64).
8386 On Pentium and PentiumPro, @code{double} and @code{long double} values
8387 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8388 suffer significant run time performance penalties. On Pentium III, the
8389 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8390 penalties if it is not 16 byte aligned.
8392 To ensure proper alignment of this values on the stack, the stack boundary
8393 must be as aligned as that required by any value stored on the stack.
8394 Further, every function must be generated such that it keeps the stack
8395 aligned. Thus calling a function compiled with a higher preferred
8396 stack boundary from a function compiled with a lower preferred stack
8397 boundary will most likely misalign the stack. It is recommended that
8398 libraries that use callbacks always use the default setting.
8400 This extra alignment does consume extra stack space, and generally
8401 increases code size. Code that is sensitive to stack space usage, such
8402 as embedded systems and operating system kernels, may want to reduce the
8403 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8421 These switches enable or disable the use of built-in functions that allow
8422 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
8424 @xref{X86 Built-in Functions}, for details of the functions enabled
8425 and disabled by these switches.
8427 To have SSE/SSE2 instructions generated automatically from floating-point
8428 code, see @option{-mfpmath=sse}.
8431 @itemx -mno-push-args
8433 @opindex mno-push-args
8434 Use PUSH operations to store outgoing parameters. This method is shorter
8435 and usually equally fast as method using SUB/MOV operations and is enabled
8436 by default. In some cases disabling it may improve performance because of
8437 improved scheduling and reduced dependencies.
8439 @item -maccumulate-outgoing-args
8440 @opindex maccumulate-outgoing-args
8441 If enabled, the maximum amount of space required for outgoing arguments will be
8442 computed in the function prologue. This is faster on most modern CPUs
8443 because of reduced dependencies, improved scheduling and reduced stack usage
8444 when preferred stack boundary is not equal to 2. The drawback is a notable
8445 increase in code size. This switch implies @option{-mno-push-args}.
8449 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8450 on thread-safe exception handling must compile and link all code with the
8451 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8452 @option{-D_MT}; when linking, it links in a special thread helper library
8453 @option{-lmingwthrd} which cleans up per thread exception handling data.
8455 @item -mno-align-stringops
8456 @opindex mno-align-stringops
8457 Do not align destination of inlined string operations. This switch reduces
8458 code size and improves performance in case the destination is already aligned,
8459 but gcc don't know about it.
8461 @item -minline-all-stringops
8462 @opindex minline-all-stringops
8463 By default GCC inlines string operations only when destination is known to be
8464 aligned at least to 4 byte boundary. This enables more inlining, increase code
8465 size, but may improve performance of code that depends on fast memcpy, strlen
8466 and memset for short lengths.
8468 @item -momit-leaf-frame-pointer
8469 @opindex momit-leaf-frame-pointer
8470 Don't keep the frame pointer in a register for leaf functions. This
8471 avoids the instructions to save, set up and restore frame pointers and
8472 makes an extra register available in leaf functions. The option
8473 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8474 which might make debugging harder.
8476 @item -mtls-direct-seg-refs
8477 @itemx -mno-tls-direct-seg-refs
8478 @opindex mtls-direct-seg-refs
8479 Controls whether TLS variables may be accessed with offsets from the
8480 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8481 or whether the thread base pointer must be added. Whether or not this
8482 is legal depends on the operating system, and whether it maps the
8483 segment to cover the entire TLS area.
8485 For systems that use GNU libc, the default is on.
8488 These @samp{-m} switches are supported in addition to the above
8489 on AMD x86-64 processors in 64-bit environments.
8496 Generate code for a 32-bit or 64-bit environment.
8497 The 32-bit environment sets int, long and pointer to 32 bits and
8498 generates code that runs on any i386 system.
8499 The 64-bit environment sets int to 32 bits and long and pointer
8500 to 64 bits and generates code for AMD's x86-64 architecture.
8503 @opindex no-red-zone
8504 Do not use a so called red zone for x86-64 code. The red zone is mandated
8505 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8506 stack pointer that will not be modified by signal or interrupt handlers
8507 and therefore can be used for temporary data without adjusting the stack
8508 pointer. The flag @option{-mno-red-zone} disables this red zone.
8510 @item -mcmodel=small
8511 @opindex mcmodel=small
8512 Generate code for the small code model: the program and its symbols must
8513 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8514 Programs can be statically or dynamically linked. This is the default
8517 @item -mcmodel=kernel
8518 @opindex mcmodel=kernel
8519 Generate code for the kernel code model. The kernel runs in the
8520 negative 2 GB of the address space.
8521 This model has to be used for Linux kernel code.
8523 @item -mcmodel=medium
8524 @opindex mcmodel=medium
8525 Generate code for the medium model: The program is linked in the lower 2
8526 GB of the address space but symbols can be located anywhere in the
8527 address space. Programs can be statically or dynamically linked, but
8528 building of shared libraries are not supported with the medium model.
8530 @item -mcmodel=large
8531 @opindex mcmodel=large
8532 Generate code for the large model: This model makes no assumptions
8533 about addresses and sizes of sections. Currently GCC does not implement
8538 @subsection HPPA Options
8539 @cindex HPPA Options
8541 These @samp{-m} options are defined for the HPPA family of computers:
8544 @item -march=@var{architecture-type}
8546 Generate code for the specified architecture. The choices for
8547 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8548 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8549 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8550 architecture option for your machine. Code compiled for lower numbered
8551 architectures will run on higher numbered architectures, but not the
8554 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8555 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8559 @itemx -mpa-risc-1-1
8560 @itemx -mpa-risc-2-0
8561 @opindex mpa-risc-1-0
8562 @opindex mpa-risc-1-1
8563 @opindex mpa-risc-2-0
8564 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8567 @opindex mbig-switch
8568 Generate code suitable for big switch tables. Use this option only if
8569 the assembler/linker complain about out of range branches within a switch
8572 @item -mjump-in-delay
8573 @opindex mjump-in-delay
8574 Fill delay slots of function calls with unconditional jump instructions
8575 by modifying the return pointer for the function call to be the target
8576 of the conditional jump.
8578 @item -mdisable-fpregs
8579 @opindex mdisable-fpregs
8580 Prevent floating point registers from being used in any manner. This is
8581 necessary for compiling kernels which perform lazy context switching of
8582 floating point registers. If you use this option and attempt to perform
8583 floating point operations, the compiler will abort.
8585 @item -mdisable-indexing
8586 @opindex mdisable-indexing
8587 Prevent the compiler from using indexing address modes. This avoids some
8588 rather obscure problems when compiling MIG generated code under MACH@.
8590 @item -mno-space-regs
8591 @opindex mno-space-regs
8592 Generate code that assumes the target has no space registers. This allows
8593 GCC to generate faster indirect calls and use unscaled index address modes.
8595 Such code is suitable for level 0 PA systems and kernels.
8597 @item -mfast-indirect-calls
8598 @opindex mfast-indirect-calls
8599 Generate code that assumes calls never cross space boundaries. This
8600 allows GCC to emit code which performs faster indirect calls.
8602 This option will not work in the presence of shared libraries or nested
8605 @item -mlong-load-store
8606 @opindex mlong-load-store
8607 Generate 3-instruction load and store sequences as sometimes required by
8608 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8611 @item -mportable-runtime
8612 @opindex mportable-runtime
8613 Use the portable calling conventions proposed by HP for ELF systems.
8617 Enable the use of assembler directives only GAS understands.
8619 @item -mschedule=@var{cpu-type}
8621 Schedule code according to the constraints for the machine type
8622 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8623 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8624 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8625 proper scheduling option for your machine. The default scheduling is
8629 @opindex mlinker-opt
8630 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8631 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8632 linkers in which they give bogus error messages when linking some programs.
8635 @opindex msoft-float
8636 Generate output containing library calls for floating point.
8637 @strong{Warning:} the requisite libraries are not available for all HPPA
8638 targets. Normally the facilities of the machine's usual C compiler are
8639 used, but this cannot be done directly in cross-compilation. You must make
8640 your own arrangements to provide suitable library functions for
8641 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8642 does provide software floating point support.
8644 @option{-msoft-float} changes the calling convention in the output file;
8645 therefore, it is only useful if you compile @emph{all} of a program with
8646 this option. In particular, you need to compile @file{libgcc.a}, the
8647 library that comes with GCC, with @option{-msoft-float} in order for
8652 Generate the predefine, @code{_SIO}, for server IO. The default is
8653 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8654 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8655 options are available under HP-UX and HI-UX.
8659 Use GNU ld specific options. This passes @option{-shared} to ld when
8660 building a shared library. It is the default when GCC is configured,
8661 explicitly or implicitly, with the GNU linker. This option does not
8662 have any affect on which ld is called, it only changes what parameters
8663 are passed to that ld. The ld that is called is determined by the
8664 @option{--with-ld} configure option, gcc's program search path, and
8665 finally by the user's @env{PATH}. The linker used by GCC can be printed
8666 using @samp{which `gcc -print-prog-name=ld`}.
8670 Use HP ld specific options. This passes @option{-b} to ld when building
8671 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8672 links. It is the default when GCC is configured, explicitly or
8673 implicitly, with the HP linker. This option does not have any affect on
8674 which ld is called, it only changes what parameters are passed to that
8675 ld. The ld that is called is determined by the @option{--with-ld}
8676 configure option, gcc's program search path, and finally by the user's
8677 @env{PATH}. The linker used by GCC can be printed using @samp{which
8678 `gcc -print-prog-name=ld`}.
8681 @opindex mno-long-calls
8682 Generate code that uses long call sequences. This ensures that a call
8683 is always able to reach linker generated stubs. The default is to generate
8684 long calls only when the distance from the call site to the beginning
8685 of the function or translation unit, as the case may be, exceeds a
8686 predefined limit set by the branch type being used. The limits for
8687 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8688 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8691 Distances are measured from the beginning of functions when using the
8692 @option{-ffunction-sections} option, or when using the @option{-mgas}
8693 and @option{-mno-portable-runtime} options together under HP-UX with
8696 It is normally not desirable to use this option as it will degrade
8697 performance. However, it may be useful in large applications,
8698 particularly when partial linking is used to build the application.
8700 The types of long calls used depends on the capabilities of the
8701 assembler and linker, and the type of code being generated. The
8702 impact on systems that support long absolute calls, and long pic
8703 symbol-difference or pc-relative calls should be relatively small.
8704 However, an indirect call is used on 32-bit ELF systems in pic code
8705 and it is quite long.
8709 Suppress the generation of link options to search libdld.sl when the
8710 @option{-static} option is specified on HP-UX 10 and later.
8714 The HP-UX implementation of setlocale in libc has a dependency on
8715 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8716 when the @option{-static} option is specified, special link options
8717 are needed to resolve this dependency.
8719 On HP-UX 10 and later, the GCC driver adds the necessary options to
8720 link with libdld.sl when the @option{-static} option is specified.
8721 This causes the resulting binary to be dynamic. On the 64-bit port,
8722 the linkers generate dynamic binaries by default in any case. The
8723 @option{-nolibdld} option can be used to prevent the GCC driver from
8724 adding these link options.
8728 Add support for multithreading with the @dfn{dce thread} library
8729 under HP-UX. This option sets flags for both the preprocessor and
8733 @node Intel 960 Options
8734 @subsection Intel 960 Options
8736 These @samp{-m} options are defined for the Intel 960 implementations:
8739 @item -m@var{cpu-type}
8747 Assume the defaults for the machine type @var{cpu-type} for some of
8748 the other options, including instruction scheduling, floating point
8749 support, and addressing modes. The choices for @var{cpu-type} are
8750 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
8751 @samp{sa}, and @samp{sb}.
8758 @opindex msoft-float
8759 The @option{-mnumerics} option indicates that the processor does support
8760 floating-point instructions. The @option{-msoft-float} option indicates
8761 that floating-point support should not be assumed.
8763 @item -mleaf-procedures
8764 @itemx -mno-leaf-procedures
8765 @opindex mleaf-procedures
8766 @opindex mno-leaf-procedures
8767 Do (or do not) attempt to alter leaf procedures to be callable with the
8768 @code{bal} instruction as well as @code{call}. This will result in more
8769 efficient code for explicit calls when the @code{bal} instruction can be
8770 substituted by the assembler or linker, but less efficient code in other
8771 cases, such as calls via function pointers, or using a linker that doesn't
8772 support this optimization.
8775 @itemx -mno-tail-call
8777 @opindex mno-tail-call
8778 Do (or do not) make additional attempts (beyond those of the
8779 machine-independent portions of the compiler) to optimize tail-recursive
8780 calls into branches. You may not want to do this because the detection of
8781 cases where this is not valid is not totally complete. The default is
8782 @option{-mno-tail-call}.
8784 @item -mcomplex-addr
8785 @itemx -mno-complex-addr
8786 @opindex mcomplex-addr
8787 @opindex mno-complex-addr
8788 Assume (or do not assume) that the use of a complex addressing mode is a
8789 win on this implementation of the i960. Complex addressing modes may not
8790 be worthwhile on the K-series, but they definitely are on the C-series.
8791 The default is currently @option{-mcomplex-addr} for all processors except
8795 @itemx -mno-code-align
8796 @opindex mcode-align
8797 @opindex mno-code-align
8798 Align code to 8-byte boundaries for faster fetching (or don't bother).
8799 Currently turned on by default for C-series implementations only.
8802 @item -mclean-linkage
8803 @itemx -mno-clean-linkage
8804 @opindex mclean-linkage
8805 @opindex mno-clean-linkage
8806 These options are not fully implemented.
8810 @itemx -mic2.0-compat
8811 @itemx -mic3.0-compat
8813 @opindex mic2.0-compat
8814 @opindex mic3.0-compat
8815 Enable compatibility with iC960 v2.0 or v3.0.
8819 @opindex masm-compat
8821 Enable compatibility with the iC960 assembler.
8823 @item -mstrict-align
8824 @itemx -mno-strict-align
8825 @opindex mstrict-align
8826 @opindex mno-strict-align
8827 Do not permit (do permit) unaligned accesses.
8831 Enable structure-alignment compatibility with Intel's gcc release version
8832 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
8834 @item -mlong-double-64
8835 @opindex mlong-double-64
8836 Implement type @samp{long double} as 64-bit floating point numbers.
8837 Without the option @samp{long double} is implemented by 80-bit
8838 floating point numbers. The only reason we have it because there is
8839 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
8840 is only useful for people using soft-float targets. Otherwise, we
8841 should recommend against use of it.
8845 @node DEC Alpha Options
8846 @subsection DEC Alpha Options
8848 These @samp{-m} options are defined for the DEC Alpha implementations:
8851 @item -mno-soft-float
8853 @opindex mno-soft-float
8854 @opindex msoft-float
8855 Use (do not use) the hardware floating-point instructions for
8856 floating-point operations. When @option{-msoft-float} is specified,
8857 functions in @file{libgcc.a} will be used to perform floating-point
8858 operations. Unless they are replaced by routines that emulate the
8859 floating-point operations, or compiled in such a way as to call such
8860 emulations routines, these routines will issue floating-point
8861 operations. If you are compiling for an Alpha without floating-point
8862 operations, you must ensure that the library is built so as not to call
8865 Note that Alpha implementations without floating-point operations are
8866 required to have floating-point registers.
8871 @opindex mno-fp-regs
8872 Generate code that uses (does not use) the floating-point register set.
8873 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8874 register set is not used, floating point operands are passed in integer
8875 registers as if they were integers and floating-point results are passed
8876 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8877 so any function with a floating-point argument or return value called by code
8878 compiled with @option{-mno-fp-regs} must also be compiled with that
8881 A typical use of this option is building a kernel that does not use,
8882 and hence need not save and restore, any floating-point registers.
8886 The Alpha architecture implements floating-point hardware optimized for
8887 maximum performance. It is mostly compliant with the IEEE floating
8888 point standard. However, for full compliance, software assistance is
8889 required. This option generates code fully IEEE compliant code
8890 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8891 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8892 defined during compilation. The resulting code is less efficient but is
8893 able to correctly support denormalized numbers and exceptional IEEE
8894 values such as not-a-number and plus/minus infinity. Other Alpha
8895 compilers call this option @option{-ieee_with_no_inexact}.
8897 @item -mieee-with-inexact
8898 @opindex mieee-with-inexact
8899 This is like @option{-mieee} except the generated code also maintains
8900 the IEEE @var{inexact-flag}. Turning on this option causes the
8901 generated code to implement fully-compliant IEEE math. In addition to
8902 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8903 macro. On some Alpha implementations the resulting code may execute
8904 significantly slower than the code generated by default. Since there is
8905 very little code that depends on the @var{inexact-flag}, you should
8906 normally not specify this option. Other Alpha compilers call this
8907 option @option{-ieee_with_inexact}.
8909 @item -mfp-trap-mode=@var{trap-mode}
8910 @opindex mfp-trap-mode
8911 This option controls what floating-point related traps are enabled.
8912 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8913 The trap mode can be set to one of four values:
8917 This is the default (normal) setting. The only traps that are enabled
8918 are the ones that cannot be disabled in software (e.g., division by zero
8922 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8926 Like @samp{su}, but the instructions are marked to be safe for software
8927 completion (see Alpha architecture manual for details).
8930 Like @samp{su}, but inexact traps are enabled as well.
8933 @item -mfp-rounding-mode=@var{rounding-mode}
8934 @opindex mfp-rounding-mode
8935 Selects the IEEE rounding mode. Other Alpha compilers call this option
8936 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8941 Normal IEEE rounding mode. Floating point numbers are rounded towards
8942 the nearest machine number or towards the even machine number in case
8946 Round towards minus infinity.
8949 Chopped rounding mode. Floating point numbers are rounded towards zero.
8952 Dynamic rounding mode. A field in the floating point control register
8953 (@var{fpcr}, see Alpha architecture reference manual) controls the
8954 rounding mode in effect. The C library initializes this register for
8955 rounding towards plus infinity. Thus, unless your program modifies the
8956 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8959 @item -mtrap-precision=@var{trap-precision}
8960 @opindex mtrap-precision
8961 In the Alpha architecture, floating point traps are imprecise. This
8962 means without software assistance it is impossible to recover from a
8963 floating trap and program execution normally needs to be terminated.
8964 GCC can generate code that can assist operating system trap handlers
8965 in determining the exact location that caused a floating point trap.
8966 Depending on the requirements of an application, different levels of
8967 precisions can be selected:
8971 Program precision. This option is the default and means a trap handler
8972 can only identify which program caused a floating point exception.
8975 Function precision. The trap handler can determine the function that
8976 caused a floating point exception.
8979 Instruction precision. The trap handler can determine the exact
8980 instruction that caused a floating point exception.
8983 Other Alpha compilers provide the equivalent options called
8984 @option{-scope_safe} and @option{-resumption_safe}.
8986 @item -mieee-conformant
8987 @opindex mieee-conformant
8988 This option marks the generated code as IEEE conformant. You must not
8989 use this option unless you also specify @option{-mtrap-precision=i} and either
8990 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8991 is to emit the line @samp{.eflag 48} in the function prologue of the
8992 generated assembly file. Under DEC Unix, this has the effect that
8993 IEEE-conformant math library routines will be linked in.
8995 @item -mbuild-constants
8996 @opindex mbuild-constants
8997 Normally GCC examines a 32- or 64-bit integer constant to
8998 see if it can construct it from smaller constants in two or three
8999 instructions. If it cannot, it will output the constant as a literal and
9000 generate code to load it from the data segment at runtime.
9002 Use this option to require GCC to construct @emph{all} integer constants
9003 using code, even if it takes more instructions (the maximum is six).
9005 You would typically use this option to build a shared library dynamic
9006 loader. Itself a shared library, it must relocate itself in memory
9007 before it can find the variables and constants in its own data segment.
9013 Select whether to generate code to be assembled by the vendor-supplied
9014 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9032 Indicate whether GCC should generate code to use the optional BWX,
9033 CIX, FIX and MAX instruction sets. The default is to use the instruction
9034 sets supported by the CPU type specified via @option{-mcpu=} option or that
9035 of the CPU on which GCC was built if none was specified.
9040 @opindex mfloat-ieee
9041 Generate code that uses (does not use) VAX F and G floating point
9042 arithmetic instead of IEEE single and double precision.
9044 @item -mexplicit-relocs
9045 @itemx -mno-explicit-relocs
9046 @opindex mexplicit-relocs
9047 @opindex mno-explicit-relocs
9048 Older Alpha assemblers provided no way to generate symbol relocations
9049 except via assembler macros. Use of these macros does not allow
9050 optimal instruction scheduling. GNU binutils as of version 2.12
9051 supports a new syntax that allows the compiler to explicitly mark
9052 which relocations should apply to which instructions. This option
9053 is mostly useful for debugging, as GCC detects the capabilities of
9054 the assembler when it is built and sets the default accordingly.
9058 @opindex msmall-data
9059 @opindex mlarge-data
9060 When @option{-mexplicit-relocs} is in effect, static data is
9061 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9062 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9063 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9064 16-bit relocations off of the @code{$gp} register. This limits the
9065 size of the small data area to 64KB, but allows the variables to be
9066 directly accessed via a single instruction.
9068 The default is @option{-mlarge-data}. With this option the data area
9069 is limited to just below 2GB. Programs that require more than 2GB of
9070 data must use @code{malloc} or @code{mmap} to allocate the data in the
9071 heap instead of in the program's data segment.
9073 When generating code for shared libraries, @option{-fpic} implies
9074 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9078 @opindex msmall-text
9079 @opindex mlarge-text
9080 When @option{-msmall-text} is used, the compiler assumes that the
9081 code of the entire program (or shared library) fits in 4MB, and is
9082 thus reachable with a branch instruction. When @option{-msmall-data}
9083 is used, the compiler can assume that all local symbols share the
9084 same @code{$gp} value, and thus reduce the number of instructions
9085 required for a function call from 4 to 1.
9087 The default is @option{-mlarge-text}.
9089 @item -mcpu=@var{cpu_type}
9091 Set the instruction set and instruction scheduling parameters for
9092 machine type @var{cpu_type}. You can specify either the @samp{EV}
9093 style name or the corresponding chip number. GCC supports scheduling
9094 parameters for the EV4, EV5 and EV6 family of processors and will
9095 choose the default values for the instruction set from the processor
9096 you specify. If you do not specify a processor type, GCC will default
9097 to the processor on which the compiler was built.
9099 Supported values for @var{cpu_type} are
9105 Schedules as an EV4 and has no instruction set extensions.
9109 Schedules as an EV5 and has no instruction set extensions.
9113 Schedules as an EV5 and supports the BWX extension.
9118 Schedules as an EV5 and supports the BWX and MAX extensions.
9122 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9126 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9129 @item -mtune=@var{cpu_type}
9131 Set only the instruction scheduling parameters for machine type
9132 @var{cpu_type}. The instruction set is not changed.
9134 @item -mmemory-latency=@var{time}
9135 @opindex mmemory-latency
9136 Sets the latency the scheduler should assume for typical memory
9137 references as seen by the application. This number is highly
9138 dependent on the memory access patterns used by the application
9139 and the size of the external cache on the machine.
9141 Valid options for @var{time} are
9145 A decimal number representing clock cycles.
9151 The compiler contains estimates of the number of clock cycles for
9152 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9153 (also called Dcache, Scache, and Bcache), as well as to main memory.
9154 Note that L3 is only valid for EV5.
9159 @node DEC Alpha/VMS Options
9160 @subsection DEC Alpha/VMS Options
9162 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9165 @item -mvms-return-codes
9166 @opindex mvms-return-codes
9167 Return VMS condition codes from main. The default is to return POSIX
9168 style condition (e.g.@ error) codes.
9171 @node H8/300 Options
9172 @subsection H8/300 Options
9174 These @samp{-m} options are defined for the H8/300 implementations:
9179 Shorten some address references at link time, when possible; uses the
9180 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9181 ld, Using ld}, for a fuller description.
9185 Generate code for the H8/300H@.
9189 Generate code for the H8S@.
9193 Generate code for the H8S and H8/300H in the normal mode. This switch
9194 must be used either with -mh or -ms.
9198 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9202 Make @code{int} data 32 bits by default.
9206 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9207 The default for the H8/300H and H8S is to align longs and floats on 4
9209 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9210 This option has no effect on the H8/300.
9214 @subsection SH Options
9216 These @samp{-m} options are defined for the SH implementations:
9221 Generate code for the SH1.
9225 Generate code for the SH2.
9228 Generate code for the SH2e.
9232 Generate code for the SH3.
9236 Generate code for the SH3e.
9240 Generate code for the SH4 without a floating-point unit.
9242 @item -m4-single-only
9243 @opindex m4-single-only
9244 Generate code for the SH4 with a floating-point unit that only
9245 supports single-precision arithmetic.
9249 Generate code for the SH4 assuming the floating-point unit is in
9250 single-precision mode by default.
9254 Generate code for the SH4.
9258 Compile code for the processor in big endian mode.
9262 Compile code for the processor in little endian mode.
9266 Align doubles at 64-bit boundaries. Note that this changes the calling
9267 conventions, and thus some functions from the standard C library will
9268 not work unless you recompile it first with @option{-mdalign}.
9272 Shorten some address references at link time, when possible; uses the
9273 linker option @option{-relax}.
9277 Use 32-bit offsets in @code{switch} tables. The default is to use
9282 Enable the use of the instruction @code{fmovd}.
9286 Comply with the calling conventions defined by Renesas.
9290 Mark the @code{MAC} register as call-clobbered, even if
9291 @option{-mhitachi} is given.
9295 Increase IEEE-compliance of floating-point code.
9299 Dump instruction size and location in the assembly code.
9303 This option is deprecated. It pads structures to multiple of 4 bytes,
9304 which is incompatible with the SH ABI@.
9308 Optimize for space instead of speed. Implied by @option{-Os}.
9312 When generating position-independent code, emit function calls using
9313 the Global Offset Table instead of the Procedure Linkage Table.
9317 Generate a library function call to invalidate instruction cache
9318 entries, after fixing up a trampoline. This library function call
9319 doesn't assume it can write to the whole memory address space. This
9320 is the default when the target is @code{sh-*-linux*}.
9323 @node System V Options
9324 @subsection Options for System V
9326 These additional options are available on System V Release 4 for
9327 compatibility with other compilers on those systems:
9332 Create a shared object.
9333 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9337 Identify the versions of each tool used by the compiler, in a
9338 @code{.ident} assembler directive in the output.
9342 Refrain from adding @code{.ident} directives to the output file (this is
9345 @item -YP,@var{dirs}
9347 Search the directories @var{dirs}, and no others, for libraries
9348 specified with @option{-l}.
9352 Look in the directory @var{dir} to find the M4 preprocessor.
9353 The assembler uses this option.
9354 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9355 @c the generic assembler that comes with Solaris takes just -Ym.
9358 @node TMS320C3x/C4x Options
9359 @subsection TMS320C3x/C4x Options
9360 @cindex TMS320C3x/C4x Options
9362 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9366 @item -mcpu=@var{cpu_type}
9368 Set the instruction set, register set, and instruction scheduling
9369 parameters for machine type @var{cpu_type}. Supported values for
9370 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9371 @samp{c44}. The default is @samp{c40} to generate code for the
9376 @itemx -msmall-memory
9378 @opindex mbig-memory
9380 @opindex msmall-memory
9382 Generates code for the big or small memory model. The small memory
9383 model assumed that all data fits into one 64K word page. At run-time
9384 the data page (DP) register must be set to point to the 64K page
9385 containing the .bss and .data program sections. The big memory model is
9386 the default and requires reloading of the DP register for every direct
9393 Allow (disallow) allocation of general integer operands into the block
9400 Enable (disable) generation of code using decrement and branch,
9401 DBcond(D), instructions. This is enabled by default for the C4x. To be
9402 on the safe side, this is disabled for the C3x, since the maximum
9403 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9404 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9405 that it can utilize the decrement and branch instruction, but will give
9406 up if there is more than one memory reference in the loop. Thus a loop
9407 where the loop counter is decremented can generate slightly more
9408 efficient code, in cases where the RPTB instruction cannot be utilized.
9410 @item -mdp-isr-reload
9412 @opindex mdp-isr-reload
9414 Force the DP register to be saved on entry to an interrupt service
9415 routine (ISR), reloaded to point to the data section, and restored on
9416 exit from the ISR@. This should not be required unless someone has
9417 violated the small memory model by modifying the DP register, say within
9424 For the C3x use the 24-bit MPYI instruction for integer multiplies
9425 instead of a library call to guarantee 32-bit results. Note that if one
9426 of the operands is a constant, then the multiplication will be performed
9427 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9428 then squaring operations are performed inline instead of a library call.
9431 @itemx -mno-fast-fix
9433 @opindex mno-fast-fix
9434 The C3x/C4x FIX instruction to convert a floating point value to an
9435 integer value chooses the nearest integer less than or equal to the
9436 floating point value rather than to the nearest integer. Thus if the
9437 floating point number is negative, the result will be incorrectly
9438 truncated an additional code is necessary to detect and correct this
9439 case. This option can be used to disable generation of the additional
9440 code required to correct the result.
9446 Enable (disable) generation of repeat block sequences using the RPTB
9447 instruction for zero overhead looping. The RPTB construct is only used
9448 for innermost loops that do not call functions or jump across the loop
9449 boundaries. There is no advantage having nested RPTB loops due to the
9450 overhead required to save and restore the RC, RS, and RE registers.
9451 This is enabled by default with @option{-O2}.
9453 @item -mrpts=@var{count}
9457 Enable (disable) the use of the single instruction repeat instruction
9458 RPTS@. If a repeat block contains a single instruction, and the loop
9459 count can be guaranteed to be less than the value @var{count}, GCC will
9460 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9461 then a RPTS will be emitted even if the loop count cannot be determined
9462 at compile time. Note that the repeated instruction following RPTS does
9463 not have to be reloaded from memory each iteration, thus freeing up the
9464 CPU buses for operands. However, since interrupts are blocked by this
9465 instruction, it is disabled by default.
9467 @item -mloop-unsigned
9468 @itemx -mno-loop-unsigned
9469 @opindex mloop-unsigned
9470 @opindex mno-loop-unsigned
9471 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9472 is @math{2^{31} + 1} since these instructions test if the iteration count is
9473 negative to terminate the loop. If the iteration count is unsigned
9474 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9475 exceeded. This switch allows an unsigned iteration count.
9479 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9480 with. This also enforces compatibility with the API employed by the TI
9481 C3x C compiler. For example, long doubles are passed as structures
9482 rather than in floating point registers.
9488 Generate code that uses registers (stack) for passing arguments to functions.
9489 By default, arguments are passed in registers where possible rather
9490 than by pushing arguments on to the stack.
9492 @item -mparallel-insns
9493 @itemx -mno-parallel-insns
9494 @opindex mparallel-insns
9495 @opindex mno-parallel-insns
9496 Allow the generation of parallel instructions. This is enabled by
9497 default with @option{-O2}.
9499 @item -mparallel-mpy
9500 @itemx -mno-parallel-mpy
9501 @opindex mparallel-mpy
9502 @opindex mno-parallel-mpy
9503 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9504 provided @option{-mparallel-insns} is also specified. These instructions have
9505 tight register constraints which can pessimize the code generation
9511 @subsection V850 Options
9512 @cindex V850 Options
9514 These @samp{-m} options are defined for V850 implementations:
9518 @itemx -mno-long-calls
9519 @opindex mlong-calls
9520 @opindex mno-long-calls
9521 Treat all calls as being far away (near). If calls are assumed to be
9522 far away, the compiler will always load the functions address up into a
9523 register, and call indirect through the pointer.
9529 Do not optimize (do optimize) basic blocks that use the same index
9530 pointer 4 or more times to copy pointer into the @code{ep} register, and
9531 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9532 option is on by default if you optimize.
9534 @item -mno-prolog-function
9535 @itemx -mprolog-function
9536 @opindex mno-prolog-function
9537 @opindex mprolog-function
9538 Do not use (do use) external functions to save and restore registers
9539 at the prologue and epilogue of a function. The external functions
9540 are slower, but use less code space if more than one function saves
9541 the same number of registers. The @option{-mprolog-function} option
9542 is on by default if you optimize.
9546 Try to make the code as small as possible. At present, this just turns
9547 on the @option{-mep} and @option{-mprolog-function} options.
9551 Put static or global variables whose size is @var{n} bytes or less into
9552 the tiny data area that register @code{ep} points to. The tiny data
9553 area can hold up to 256 bytes in total (128 bytes for byte references).
9557 Put static or global variables whose size is @var{n} bytes or less into
9558 the small data area that register @code{gp} points to. The small data
9559 area can hold up to 64 kilobytes.
9563 Put static or global variables whose size is @var{n} bytes or less into
9564 the first 32 kilobytes of memory.
9568 Specify that the target processor is the V850.
9571 @opindex mbig-switch
9572 Generate code suitable for big switch tables. Use this option only if
9573 the assembler/linker complain about out of range branches within a switch
9578 This option will cause r2 and r5 to be used in the code generated by
9579 the compiler. This setting is the default.
9582 @opindex mno-app-regs
9583 This option will cause r2 and r5 to be treated as fixed registers.
9587 Specify that the target processor is the V850E1. The preprocessor
9588 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
9589 this option is used.
9593 Specify that the target processor is the V850E. The preprocessor
9594 constant @samp{__v850e__} will be defined if this option is used.
9596 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
9597 are defined then a default target processor will be chosen and the
9598 relevant @samp{__v850*__} preprocessor constant will be defined.
9600 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9601 defined, regardless of which processor variant is the target.
9603 @item -mdisable-callt
9604 @opindex mdisable-callt
9605 This option will suppress generation of the CALLT instruction for the
9606 v850e and v850e1 flavors of the v850 architecture. The default is
9607 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9612 @subsection ARC Options
9615 These options are defined for ARC implementations:
9620 Compile code for little endian mode. This is the default.
9624 Compile code for big endian mode.
9627 @opindex mmangle-cpu
9628 Prepend the name of the cpu to all public symbol names.
9629 In multiple-processor systems, there are many ARC variants with different
9630 instruction and register set characteristics. This flag prevents code
9631 compiled for one cpu to be linked with code compiled for another.
9632 No facility exists for handling variants that are ``almost identical''.
9633 This is an all or nothing option.
9635 @item -mcpu=@var{cpu}
9637 Compile code for ARC variant @var{cpu}.
9638 Which variants are supported depend on the configuration.
9639 All variants support @option{-mcpu=base}, this is the default.
9641 @item -mtext=@var{text-section}
9642 @itemx -mdata=@var{data-section}
9643 @itemx -mrodata=@var{readonly-data-section}
9647 Put functions, data, and readonly data in @var{text-section},
9648 @var{data-section}, and @var{readonly-data-section} respectively
9649 by default. This can be overridden with the @code{section} attribute.
9650 @xref{Variable Attributes}.
9655 @subsection NS32K Options
9656 @cindex NS32K options
9658 These are the @samp{-m} options defined for the 32000 series. The default
9659 values for these options depends on which style of 32000 was selected when
9660 the compiler was configured; the defaults for the most common choices are
9668 Generate output for a 32032. This is the default
9669 when the compiler is configured for 32032 and 32016 based systems.
9675 Generate output for a 32332. This is the default
9676 when the compiler is configured for 32332-based systems.
9682 Generate output for a 32532. This is the default
9683 when the compiler is configured for 32532-based systems.
9687 Generate output containing 32081 instructions for floating point.
9688 This is the default for all systems.
9692 Generate output containing 32381 instructions for floating point. This
9693 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9694 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9698 Try and generate multiply-add floating point instructions @code{polyF}
9699 and @code{dotF}. This option is only available if the @option{-m32381}
9700 option is in effect. Using these instructions requires changes to
9701 register allocation which generally has a negative impact on
9702 performance. This option should only be enabled when compiling code
9703 particularly likely to make heavy use of multiply-add instructions.
9706 @opindex mnomulti-add
9707 Do not try and generate multiply-add floating point instructions
9708 @code{polyF} and @code{dotF}. This is the default on all platforms.
9711 @opindex msoft-float
9712 Generate output containing library calls for floating point.
9713 @strong{Warning:} the requisite libraries may not be available.
9715 @item -mieee-compare
9716 @itemx -mno-ieee-compare
9717 @opindex mieee-compare
9718 @opindex mno-ieee-compare
9719 Control whether or not the compiler uses IEEE floating point
9720 comparisons. These handle correctly the case where the result of a
9721 comparison is unordered.
9722 @strong{Warning:} the requisite kernel support may not be available.
9725 @opindex mnobitfield
9726 Do not use the bit-field instructions. On some machines it is faster to
9727 use shifting and masking operations. This is the default for the pc532.
9731 Do use the bit-field instructions. This is the default for all platforms
9736 Use a different function-calling convention, in which functions
9737 that take a fixed number of arguments return pop their
9738 arguments on return with the @code{ret} instruction.
9740 This calling convention is incompatible with the one normally
9741 used on Unix, so you cannot use it if you need to call libraries
9742 compiled with the Unix compiler.
9744 Also, you must provide function prototypes for all functions that
9745 take variable numbers of arguments (including @code{printf});
9746 otherwise incorrect code will be generated for calls to those
9749 In addition, seriously incorrect code will result if you call a
9750 function with too many arguments. (Normally, extra arguments are
9751 harmlessly ignored.)
9753 This option takes its name from the 680x0 @code{rtd} instruction.
9758 Use a different function-calling convention where the first two arguments
9759 are passed in registers.
9761 This calling convention is incompatible with the one normally
9762 used on Unix, so you cannot use it if you need to call libraries
9763 compiled with the Unix compiler.
9766 @opindex mnoregparam
9767 Do not pass any arguments in registers. This is the default for all
9772 It is OK to use the sb as an index register which is always loaded with
9773 zero. This is the default for the pc532-netbsd target.
9777 The sb register is not available for use or has not been initialized to
9778 zero by the run time system. This is the default for all targets except
9779 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9780 @option{-fpic} is set.
9784 Many ns32000 series addressing modes use displacements of up to 512MB@.
9785 If an address is above 512MB then displacements from zero can not be used.
9786 This option causes code to be generated which can be loaded above 512MB@.
9787 This may be useful for operating systems or ROM code.
9791 Assume code will be loaded in the first 512MB of virtual address space.
9792 This is the default for all platforms.
9798 @subsection AVR Options
9801 These options are defined for AVR implementations:
9804 @item -mmcu=@var{mcu}
9806 Specify ATMEL AVR instruction set or MCU type.
9808 Instruction set avr1 is for the minimal AVR core, not supported by the C
9809 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9810 attiny11, attiny12, attiny15, attiny28).
9812 Instruction set avr2 (default) is for the classic AVR core with up to
9813 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9814 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9815 at90c8534, at90s8535).
9817 Instruction set avr3 is for the classic AVR core with up to 128K program
9818 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9820 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9821 memory space (MCU types: atmega8, atmega83, atmega85).
9823 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9824 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9825 atmega64, atmega128, at43usb355, at94k).
9829 Output instruction sizes to the asm file.
9831 @item -minit-stack=@var{N}
9832 @opindex minit-stack
9833 Specify the initial stack address, which may be a symbol or numeric value,
9834 @samp{__stack} is the default.
9836 @item -mno-interrupts
9837 @opindex mno-interrupts
9838 Generated code is not compatible with hardware interrupts.
9839 Code size will be smaller.
9841 @item -mcall-prologues
9842 @opindex mcall-prologues
9843 Functions prologues/epilogues expanded as call to appropriate
9844 subroutines. Code size will be smaller.
9846 @item -mno-tablejump
9847 @opindex mno-tablejump
9848 Do not generate tablejump insns which sometimes increase code size.
9851 @opindex mtiny-stack
9852 Change only the low 8 bits of the stack pointer.
9856 @subsection MCore Options
9857 @cindex MCore options
9859 These are the @samp{-m} options defined for the Motorola M*Core
9867 @opindex mno-hardlit
9868 Inline constants into the code stream if it can be done in two
9869 instructions or less.
9875 Use the divide instruction. (Enabled by default).
9877 @item -mrelax-immediate
9878 @itemx -mno-relax-immediate
9879 @opindex mrelax-immediate
9880 @opindex mno-relax-immediate
9881 Allow arbitrary sized immediates in bit operations.
9883 @item -mwide-bitfields
9884 @itemx -mno-wide-bitfields
9885 @opindex mwide-bitfields
9886 @opindex mno-wide-bitfields
9887 Always treat bit-fields as int-sized.
9889 @item -m4byte-functions
9890 @itemx -mno-4byte-functions
9891 @opindex m4byte-functions
9892 @opindex mno-4byte-functions
9893 Force all functions to be aligned to a four byte boundary.
9895 @item -mcallgraph-data
9896 @itemx -mno-callgraph-data
9897 @opindex mcallgraph-data
9898 @opindex mno-callgraph-data
9899 Emit callgraph information.
9902 @itemx -mno-slow-bytes
9903 @opindex mslow-bytes
9904 @opindex mno-slow-bytes
9905 Prefer word access when reading byte quantities.
9907 @item -mlittle-endian
9909 @opindex mlittle-endian
9910 @opindex mbig-endian
9911 Generate code for a little endian target.
9917 Generate code for the 210 processor.
9921 @subsection IA-64 Options
9922 @cindex IA-64 Options
9924 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9928 @opindex mbig-endian
9929 Generate code for a big endian target. This is the default for HP-UX@.
9931 @item -mlittle-endian
9932 @opindex mlittle-endian
9933 Generate code for a little endian target. This is the default for AIX5
9940 Generate (or don't) code for the GNU assembler. This is the default.
9941 @c Also, this is the default if the configure option @option{--with-gnu-as}
9948 Generate (or don't) code for the GNU linker. This is the default.
9949 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9954 Generate code that does not use a global pointer register. The result
9955 is not position independent code, and violates the IA-64 ABI@.
9957 @item -mvolatile-asm-stop
9958 @itemx -mno-volatile-asm-stop
9959 @opindex mvolatile-asm-stop
9960 @opindex mno-volatile-asm-stop
9961 Generate (or don't) a stop bit immediately before and after volatile asm
9966 Generate code that works around Itanium B step errata.
9968 @item -mregister-names
9969 @itemx -mno-register-names
9970 @opindex mregister-names
9971 @opindex mno-register-names
9972 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9973 the stacked registers. This may make assembler output more readable.
9979 Disable (or enable) optimizations that use the small data section. This may
9980 be useful for working around optimizer bugs.
9983 @opindex mconstant-gp
9984 Generate code that uses a single constant global pointer value. This is
9985 useful when compiling kernel code.
9989 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9990 This is useful when compiling firmware code.
9992 @item -minline-float-divide-min-latency
9993 @opindex minline-float-divide-min-latency
9994 Generate code for inline divides of floating point values
9995 using the minimum latency algorithm.
9997 @item -minline-float-divide-max-throughput
9998 @opindex minline-float-divide-max-throughput
9999 Generate code for inline divides of floating point values
10000 using the maximum throughput algorithm.
10002 @item -minline-int-divide-min-latency
10003 @opindex minline-int-divide-min-latency
10004 Generate code for inline divides of integer values
10005 using the minimum latency algorithm.
10007 @item -minline-int-divide-max-throughput
10008 @opindex minline-int-divide-max-throughput
10009 Generate code for inline divides of integer values
10010 using the maximum throughput algorithm.
10012 @item -mno-dwarf2-asm
10013 @itemx -mdwarf2-asm
10014 @opindex mno-dwarf2-asm
10015 @opindex mdwarf2-asm
10016 Don't (or do) generate assembler code for the DWARF2 line number debugging
10017 info. This may be useful when not using the GNU assembler.
10019 @item -mfixed-range=@var{register-range}
10020 @opindex mfixed-range
10021 Generate code treating the given register range as fixed registers.
10022 A fixed register is one that the register allocator can not use. This is
10023 useful when compiling kernel code. A register range is specified as
10024 two registers separated by a dash. Multiple register ranges can be
10025 specified separated by a comma.
10027 @item -mearly-stop-bits
10028 @itemx -mno-early-stop-bits
10029 @opindex mearly-stop-bits
10030 @opindex mno-early-stop-bits
10031 Allow stop bits to be placed earlier than immediately preceding the
10032 instruction that triggered the stop bit. This can improve instruction
10033 scheduling, but does not always do so.
10037 @subsection D30V Options
10038 @cindex D30V Options
10040 These @samp{-m} options are defined for D30V implementations:
10045 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
10046 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
10047 memory, which starts at location @code{0x80000000}.
10050 @opindex mextmemory
10051 Same as the @option{-mextmem} switch.
10055 Link the @samp{.text} section into onchip text memory, which starts at
10056 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
10057 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
10058 into onchip data memory, which starts at location @code{0x20000000}.
10060 @item -mno-asm-optimize
10061 @itemx -masm-optimize
10062 @opindex mno-asm-optimize
10063 @opindex masm-optimize
10064 Disable (enable) passing @option{-O} to the assembler when optimizing.
10065 The assembler uses the @option{-O} option to automatically parallelize
10066 adjacent short instructions where possible.
10068 @item -mbranch-cost=@var{n}
10069 @opindex mbranch-cost
10070 Increase the internal costs of branches to @var{n}. Higher costs means
10071 that the compiler will issue more instructions to avoid doing a branch.
10074 @item -mcond-exec=@var{n}
10075 @opindex mcond-exec
10076 Specify the maximum number of conditionally executed instructions that
10077 replace a branch. The default is 4.
10080 @node S/390 and zSeries Options
10081 @subsection S/390 and zSeries Options
10082 @cindex S/390 and zSeries Options
10084 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10088 @itemx -msoft-float
10089 @opindex mhard-float
10090 @opindex msoft-float
10091 Use (do not use) the hardware floating-point instructions and registers
10092 for floating-point operations. When @option{-msoft-float} is specified,
10093 functions in @file{libgcc.a} will be used to perform floating-point
10094 operations. When @option{-mhard-float} is specified, the compiler
10095 generates IEEE floating-point instructions. This is the default.
10098 @itemx -mno-backchain
10099 @opindex mbackchain
10100 @opindex mno-backchain
10101 Generate (or do not generate) code which maintains an explicit
10102 backchain within the stack frame that points to the caller's frame.
10103 This may be needed to allow debugging using tools that do not understand
10104 DWARF-2 call frame information. The default is not to generate the
10108 @itemx -mno-small-exec
10109 @opindex msmall-exec
10110 @opindex mno-small-exec
10111 Generate (or do not generate) code using the @code{bras} instruction
10112 to do subroutine calls.
10113 This only works reliably if the total executable size does not
10114 exceed 64k. The default is to use the @code{basr} instruction instead,
10115 which does not have this limitation.
10121 When @option{-m31} is specified, generate code compliant to the
10122 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
10123 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
10124 particular to generate 64-bit instructions. For the @samp{s390}
10125 targets, the default is @option{-m31}, while the @samp{s390x}
10126 targets default to @option{-m64}.
10132 When @option{-mzarch} is specified, generate code using the
10133 instructions available on z/Architecture.
10134 When @option{-mesa} is specified, generate code using the
10135 instructions available on ESA/390. Note that @option{-mesa} is
10136 not possible with @option{-m64}.
10137 When generating code compliant to the GNU/Linux for S/390 ABI,
10138 the default is @option{-mesa}. When generating code compliant
10139 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
10145 Generate (or do not generate) code using the @code{mvcle} instruction
10146 to perform block moves. When @option{-mno-mvcle} is specified,
10147 use a @code{mvc} loop instead. This is the default.
10153 Print (or do not print) additional debug information when compiling.
10154 The default is to not print debug information.
10156 @item -march=@var{cpu-type}
10158 Generate code that will run on @var{cpu-type}, which is the name of a system
10159 representing a certain processor type. Possible values for
10160 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10161 When generating code using the instructions available on z/Architecture,
10162 the default is @option{-march=z900}. Otherwise, the default is
10163 @option{-march=g5}.
10165 @item -mtune=@var{cpu-type}
10167 Tune to @var{cpu-type} everything applicable about the generated code,
10168 except for the ABI and the set of available instructions.
10169 The list of @var{cpu-type} values is the same as for @option{-march}.
10170 The default is the value used for @option{-march}.
10173 @itemx -mno-fused-madd
10174 @opindex mfused-madd
10175 @opindex mno-fused-madd
10176 Generate code that uses (does not use) the floating point multiply and
10177 accumulate instructions. These instructions are generated by default if
10178 hardware floating point is used.
10182 @subsection CRIS Options
10183 @cindex CRIS Options
10185 These options are defined specifically for the CRIS ports.
10188 @item -march=@var{architecture-type}
10189 @itemx -mcpu=@var{architecture-type}
10192 Generate code for the specified architecture. The choices for
10193 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10194 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
10195 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10198 @item -mtune=@var{architecture-type}
10200 Tune to @var{architecture-type} everything applicable about the generated
10201 code, except for the ABI and the set of available instructions. The
10202 choices for @var{architecture-type} are the same as for
10203 @option{-march=@var{architecture-type}}.
10205 @item -mmax-stack-frame=@var{n}
10206 @opindex mmax-stack-frame
10207 Warn when the stack frame of a function exceeds @var{n} bytes.
10209 @item -melinux-stacksize=@var{n}
10210 @opindex melinux-stacksize
10211 Only available with the @samp{cris-axis-aout} target. Arranges for
10212 indications in the program to the kernel loader that the stack of the
10213 program should be set to @var{n} bytes.
10219 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10220 @option{-march=v3} and @option{-march=v8} respectively.
10224 Enable CRIS-specific verbose debug-related information in the assembly
10225 code. This option also has the effect to turn off the @samp{#NO_APP}
10226 formatted-code indicator to the assembler at the beginning of the
10231 Do not use condition-code results from previous instruction; always emit
10232 compare and test instructions before use of condition codes.
10234 @item -mno-side-effects
10235 @opindex mno-side-effects
10236 Do not emit instructions with side-effects in addressing modes other than
10239 @item -mstack-align
10240 @itemx -mno-stack-align
10241 @itemx -mdata-align
10242 @itemx -mno-data-align
10243 @itemx -mconst-align
10244 @itemx -mno-const-align
10245 @opindex mstack-align
10246 @opindex mno-stack-align
10247 @opindex mdata-align
10248 @opindex mno-data-align
10249 @opindex mconst-align
10250 @opindex mno-const-align
10251 These options (no-options) arranges (eliminate arrangements) for the
10252 stack-frame, individual data and constants to be aligned for the maximum
10253 single data access size for the chosen CPU model. The default is to
10254 arrange for 32-bit alignment. ABI details such as structure layout are
10255 not affected by these options.
10263 Similar to the stack- data- and const-align options above, these options
10264 arrange for stack-frame, writable data and constants to all be 32-bit,
10265 16-bit or 8-bit aligned. The default is 32-bit alignment.
10267 @item -mno-prologue-epilogue
10268 @itemx -mprologue-epilogue
10269 @opindex mno-prologue-epilogue
10270 @opindex mprologue-epilogue
10271 With @option{-mno-prologue-epilogue}, the normal function prologue and
10272 epilogue that sets up the stack-frame are omitted and no return
10273 instructions or return sequences are generated in the code. Use this
10274 option only together with visual inspection of the compiled code: no
10275 warnings or errors are generated when call-saved registers must be saved,
10276 or storage for local variable needs to be allocated.
10280 @opindex mno-gotplt
10282 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10283 instruction sequences that load addresses for functions from the PLT part
10284 of the GOT rather than (traditional on other architectures) calls to the
10285 PLT. The default is @option{-mgotplt}.
10289 Legacy no-op option only recognized with the cris-axis-aout target.
10293 Legacy no-op option only recognized with the cris-axis-elf and
10294 cris-axis-linux-gnu targets.
10298 Only recognized with the cris-axis-aout target, where it selects a
10299 GNU/linux-like multilib, include files and instruction set for
10300 @option{-march=v8}.
10304 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10308 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
10309 to link with input-output functions from a simulator library. Code,
10310 initialized data and zero-initialized data are allocated consecutively.
10314 Like @option{-sim}, but pass linker options to locate initialized data at
10315 0x40000000 and zero-initialized data at 0x80000000.
10319 @subsection MMIX Options
10320 @cindex MMIX Options
10322 These options are defined for the MMIX:
10326 @itemx -mno-libfuncs
10328 @opindex mno-libfuncs
10329 Specify that intrinsic library functions are being compiled, passing all
10330 values in registers, no matter the size.
10333 @itemx -mno-epsilon
10335 @opindex mno-epsilon
10336 Generate floating-point comparison instructions that compare with respect
10337 to the @code{rE} epsilon register.
10339 @item -mabi=mmixware
10341 @opindex mabi-mmixware
10343 Generate code that passes function parameters and return values that (in
10344 the called function) are seen as registers @code{$0} and up, as opposed to
10345 the GNU ABI which uses global registers @code{$231} and up.
10347 @item -mzero-extend
10348 @itemx -mno-zero-extend
10349 @opindex mzero-extend
10350 @opindex mno-zero-extend
10351 When reading data from memory in sizes shorter than 64 bits, use (do not
10352 use) zero-extending load instructions by default, rather than
10353 sign-extending ones.
10356 @itemx -mno-knuthdiv
10358 @opindex mno-knuthdiv
10359 Make the result of a division yielding a remainder have the same sign as
10360 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10361 remainder follows the sign of the dividend. Both methods are
10362 arithmetically valid, the latter being almost exclusively used.
10364 @item -mtoplevel-symbols
10365 @itemx -mno-toplevel-symbols
10366 @opindex mtoplevel-symbols
10367 @opindex mno-toplevel-symbols
10368 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10369 code can be used with the @code{PREFIX} assembly directive.
10373 Generate an executable in the ELF format, rather than the default
10374 @samp{mmo} format used by the @command{mmix} simulator.
10376 @item -mbranch-predict
10377 @itemx -mno-branch-predict
10378 @opindex mbranch-predict
10379 @opindex mno-branch-predict
10380 Use (do not use) the probable-branch instructions, when static branch
10381 prediction indicates a probable branch.
10383 @item -mbase-addresses
10384 @itemx -mno-base-addresses
10385 @opindex mbase-addresses
10386 @opindex mno-base-addresses
10387 Generate (do not generate) code that uses @emph{base addresses}. Using a
10388 base address automatically generates a request (handled by the assembler
10389 and the linker) for a constant to be set up in a global register. The
10390 register is used for one or more base address requests within the range 0
10391 to 255 from the value held in the register. The generally leads to short
10392 and fast code, but the number of different data items that can be
10393 addressed is limited. This means that a program that uses lots of static
10394 data may require @option{-mno-base-addresses}.
10396 @item -msingle-exit
10397 @itemx -mno-single-exit
10398 @opindex msingle-exit
10399 @opindex mno-single-exit
10400 Force (do not force) generated code to have a single exit point in each
10404 @node PDP-11 Options
10405 @subsection PDP-11 Options
10406 @cindex PDP-11 Options
10408 These options are defined for the PDP-11:
10413 Use hardware FPP floating point. This is the default. (FIS floating
10414 point on the PDP-11/40 is not supported.)
10417 @opindex msoft-float
10418 Do not use hardware floating point.
10422 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10426 Return floating-point results in memory. This is the default.
10430 Generate code for a PDP-11/40.
10434 Generate code for a PDP-11/45. This is the default.
10438 Generate code for a PDP-11/10.
10440 @item -mbcopy-builtin
10441 @opindex bcopy-builtin
10442 Use inline @code{movstrhi} patterns for copying memory. This is the
10447 Do not use inline @code{movstrhi} patterns for copying memory.
10453 Use 16-bit @code{int}. This is the default.
10459 Use 32-bit @code{int}.
10462 @itemx -mno-float32
10464 @opindex mno-float32
10465 Use 64-bit @code{float}. This is the default.
10468 @itemx -mno-float64
10470 @opindex mno-float64
10471 Use 32-bit @code{float}.
10475 Use @code{abshi2} pattern. This is the default.
10479 Do not use @code{abshi2} pattern.
10481 @item -mbranch-expensive
10482 @opindex mbranch-expensive
10483 Pretend that branches are expensive. This is for experimenting with
10484 code generation only.
10486 @item -mbranch-cheap
10487 @opindex mbranch-cheap
10488 Do not pretend that branches are expensive. This is the default.
10492 Generate code for a system with split I&D.
10496 Generate code for a system without split I&D. This is the default.
10500 Use Unix assembler syntax. This is the default when configured for
10501 @samp{pdp11-*-bsd}.
10505 Use DEC assembler syntax. This is the default when configured for any
10506 PDP-11 target other than @samp{pdp11-*-bsd}.
10509 @node Xstormy16 Options
10510 @subsection Xstormy16 Options
10511 @cindex Xstormy16 Options
10513 These options are defined for Xstormy16:
10518 Choose startup files and linker script suitable for the simulator.
10522 @subsection FRV Options
10523 @cindex FRV Options
10529 Only use the first 32 general purpose registers.
10534 Use all 64 general purpose registers.
10539 Use only the first 32 floating point registers.
10544 Use all 64 floating point registers
10547 @opindex mhard-float
10549 Use hardware instructions for floating point operations.
10552 @opindex msoft-float
10554 Use library routines for floating point operations.
10559 Dynamically allocate condition code registers.
10564 Do not try to dynamically allocate condition code registers, only
10565 use @code{icc0} and @code{fcc0}.
10570 Change ABI to use double word insns.
10575 Do not use double word instructions.
10580 Use floating point double instructions.
10583 @opindex mno-double
10585 Do not use floating point double instructions.
10590 Use media instructions.
10595 Do not use media instructions.
10600 Use multiply and add/subtract instructions.
10603 @opindex mno-muladd
10605 Do not use multiply and add/subtract instructions.
10607 @item -mlibrary-pic
10608 @opindex mlibrary-pic
10610 Enable PIC support for building libraries
10615 Use only the first four media accumulator registers.
10620 Use all eight media accumulator registers.
10625 Pack VLIW instructions.
10630 Do not pack VLIW instructions.
10633 @opindex mno-eflags
10635 Do not mark ABI switches in e_flags.
10638 @opindex mcond-move
10640 Enable the use of conditional-move instructions (default).
10642 This switch is mainly for debugging the compiler and will likely be removed
10643 in a future version.
10645 @item -mno-cond-move
10646 @opindex mno-cond-move
10648 Disable the use of conditional-move instructions.
10650 This switch is mainly for debugging the compiler and will likely be removed
10651 in a future version.
10656 Enable the use of conditional set instructions (default).
10658 This switch is mainly for debugging the compiler and will likely be removed
10659 in a future version.
10664 Disable the use of conditional set instructions.
10666 This switch is mainly for debugging the compiler and will likely be removed
10667 in a future version.
10670 @opindex mcond-exec
10672 Enable the use of conditional execution (default).
10674 This switch is mainly for debugging the compiler and will likely be removed
10675 in a future version.
10677 @item -mno-cond-exec
10678 @opindex mno-cond-exec
10680 Disable the use of conditional execution.
10682 This switch is mainly for debugging the compiler and will likely be removed
10683 in a future version.
10685 @item -mvliw-branch
10686 @opindex mvliw-branch
10688 Run a pass to pack branches into VLIW instructions (default).
10690 This switch is mainly for debugging the compiler and will likely be removed
10691 in a future version.
10693 @item -mno-vliw-branch
10694 @opindex mno-vliw-branch
10696 Do not run a pass to pack branches into VLIW instructions.
10698 This switch is mainly for debugging the compiler and will likely be removed
10699 in a future version.
10701 @item -mmulti-cond-exec
10702 @opindex mmulti-cond-exec
10704 Enable optimization of @code{&&} and @code{||} in conditional execution
10707 This switch is mainly for debugging the compiler and will likely be removed
10708 in a future version.
10710 @item -mno-multi-cond-exec
10711 @opindex mno-multi-cond-exec
10713 Disable optimization of @code{&&} and @code{||} in conditional execution.
10715 This switch is mainly for debugging the compiler and will likely be removed
10716 in a future version.
10718 @item -mnested-cond-exec
10719 @opindex mnested-cond-exec
10721 Enable nested conditional execution optimizations (default).
10723 This switch is mainly for debugging the compiler and will likely be removed
10724 in a future version.
10726 @item -mno-nested-cond-exec
10727 @opindex mno-nested-cond-exec
10729 Disable nested conditional execution optimizations.
10731 This switch is mainly for debugging the compiler and will likely be removed
10732 in a future version.
10734 @item -mtomcat-stats
10735 @opindex mtomcat-stats
10737 Cause gas to print out tomcat statistics.
10739 @item -mcpu=@var{cpu}
10742 Select the processor type for which to generate code. Possible values are
10743 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
10748 @node Xtensa Options
10749 @subsection Xtensa Options
10750 @cindex Xtensa Options
10752 These options are supported for Xtensa targets:
10756 @itemx -mno-const16
10758 @opindex mno-const16
10759 Enable or disable use of @code{CONST16} instructions for loading
10760 constant values. The @code{CONST16} instruction is currently not a
10761 standard option from Tensilica. When enabled, @code{CONST16}
10762 instructions are always used in place of the standard @code{L32R}
10763 instructions. The use of @code{CONST16} is enabled by default only if
10764 the @code{L32R} instruction is not available.
10767 @itemx -mno-fused-madd
10768 @opindex mfused-madd
10769 @opindex mno-fused-madd
10770 Enable or disable use of fused multiply/add and multiply/subtract
10771 instructions in the floating-point option. This has no effect if the
10772 floating-point option is not also enabled. Disabling fused multiply/add
10773 and multiply/subtract instructions forces the compiler to use separate
10774 instructions for the multiply and add/subtract operations. This may be
10775 desirable in some cases where strict IEEE 754-compliant results are
10776 required: the fused multiply add/subtract instructions do not round the
10777 intermediate result, thereby producing results with @emph{more} bits of
10778 precision than specified by the IEEE standard. Disabling fused multiply
10779 add/subtract instructions also ensures that the program output is not
10780 sensitive to the compiler's ability to combine multiply and add/subtract
10783 @item -mtext-section-literals
10784 @itemx -mno-text-section-literals
10785 @opindex mtext-section-literals
10786 @opindex mno-text-section-literals
10787 Control the treatment of literal pools. The default is
10788 @option{-mno-text-section-literals}, which places literals in a separate
10789 section in the output file. This allows the literal pool to be placed
10790 in a data RAM/ROM, and it also allows the linker to combine literal
10791 pools from separate object files to remove redundant literals and
10792 improve code size. With @option{-mtext-section-literals}, the literals
10793 are interspersed in the text section in order to keep them as close as
10794 possible to their references. This may be necessary for large assembly
10797 @item -mtarget-align
10798 @itemx -mno-target-align
10799 @opindex mtarget-align
10800 @opindex mno-target-align
10801 When this option is enabled, GCC instructs the assembler to
10802 automatically align instructions to reduce branch penalties at the
10803 expense of some code density. The assembler attempts to widen density
10804 instructions to align branch targets and the instructions following call
10805 instructions. If there are not enough preceding safe density
10806 instructions to align a target, no widening will be performed. The
10807 default is @option{-mtarget-align}. These options do not affect the
10808 treatment of auto-aligned instructions like @code{LOOP}, which the
10809 assembler will always align, either by widening density instructions or
10810 by inserting no-op instructions.
10813 @itemx -mno-longcalls
10814 @opindex mlongcalls
10815 @opindex mno-longcalls
10816 When this option is enabled, GCC instructs the assembler to translate
10817 direct calls to indirect calls unless it can determine that the target
10818 of a direct call is in the range allowed by the call instruction. This
10819 translation typically occurs for calls to functions in other source
10820 files. Specifically, the assembler translates a direct @code{CALL}
10821 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
10822 The default is @option{-mno-longcalls}. This option should be used in
10823 programs where the call target can potentially be out of range. This
10824 option is implemented in the assembler, not the compiler, so the
10825 assembly code generated by GCC will still show direct call
10826 instructions---look at the disassembled object code to see the actual
10827 instructions. Note that the assembler will use an indirect call for
10828 every cross-file call, not just those that really will be out of range.
10831 @node Code Gen Options
10832 @section Options for Code Generation Conventions
10833 @cindex code generation conventions
10834 @cindex options, code generation
10835 @cindex run-time options
10837 These machine-independent options control the interface conventions
10838 used in code generation.
10840 Most of them have both positive and negative forms; the negative form
10841 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
10842 one of the forms is listed---the one which is not the default. You
10843 can figure out the other form by either removing @samp{no-} or adding
10847 @item -fbounds-check
10848 @opindex fbounds-check
10849 For front-ends that support it, generate additional code to check that
10850 indices used to access arrays are within the declared range. This is
10851 currently only supported by the Java and Fortran 77 front-ends, where
10852 this option defaults to true and false respectively.
10856 This option generates traps for signed overflow on addition, subtraction,
10857 multiplication operations.
10861 This option instructs the compiler to assume that signed arithmetic
10862 overflow of addition, subtraction and multiplication wraps around
10863 using twos-complement representation. This flag enables some optimizations
10864 and disables other. This option is enabled by default for the Java
10865 front-end, as required by the Java language specification.
10868 @opindex fexceptions
10869 Enable exception handling. Generates extra code needed to propagate
10870 exceptions. For some targets, this implies GCC will generate frame
10871 unwind information for all functions, which can produce significant data
10872 size overhead, although it does not affect execution. If you do not
10873 specify this option, GCC will enable it by default for languages like
10874 C++ which normally require exception handling, and disable it for
10875 languages like C that do not normally require it. However, you may need
10876 to enable this option when compiling C code that needs to interoperate
10877 properly with exception handlers written in C++. You may also wish to
10878 disable this option if you are compiling older C++ programs that don't
10879 use exception handling.
10881 @item -fnon-call-exceptions
10882 @opindex fnon-call-exceptions
10883 Generate code that allows trapping instructions to throw exceptions.
10884 Note that this requires platform-specific runtime support that does
10885 not exist everywhere. Moreover, it only allows @emph{trapping}
10886 instructions to throw exceptions, i.e.@: memory references or floating
10887 point instructions. It does not allow exceptions to be thrown from
10888 arbitrary signal handlers such as @code{SIGALRM}.
10890 @item -funwind-tables
10891 @opindex funwind-tables
10892 Similar to @option{-fexceptions}, except that it will just generate any needed
10893 static data, but will not affect the generated code in any other way.
10894 You will normally not enable this option; instead, a language processor
10895 that needs this handling would enable it on your behalf.
10897 @item -fasynchronous-unwind-tables
10898 @opindex funwind-tables
10899 Generate unwind table in dwarf2 format, if supported by target machine. The
10900 table is exact at each instruction boundary, so it can be used for stack
10901 unwinding from asynchronous events (such as debugger or garbage collector).
10903 @item -fpcc-struct-return
10904 @opindex fpcc-struct-return
10905 Return ``short'' @code{struct} and @code{union} values in memory like
10906 longer ones, rather than in registers. This convention is less
10907 efficient, but it has the advantage of allowing intercallability between
10908 GCC-compiled files and files compiled with other compilers, particularly
10909 the Portable C Compiler (pcc).
10911 The precise convention for returning structures in memory depends
10912 on the target configuration macros.
10914 Short structures and unions are those whose size and alignment match
10915 that of some integer type.
10917 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
10918 switch is not binary compatible with code compiled with the
10919 @option{-freg-struct-return} switch.
10920 Use it to conform to a non-default application binary interface.
10922 @item -freg-struct-return
10923 @opindex freg-struct-return
10924 Return @code{struct} and @code{union} values in registers when possible.
10925 This is more efficient for small structures than
10926 @option{-fpcc-struct-return}.
10928 If you specify neither @option{-fpcc-struct-return} nor
10929 @option{-freg-struct-return}, GCC defaults to whichever convention is
10930 standard for the target. If there is no standard convention, GCC
10931 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
10932 the principal compiler. In those cases, we can choose the standard, and
10933 we chose the more efficient register return alternative.
10935 @strong{Warning:} code compiled with the @option{-freg-struct-return}
10936 switch is not binary compatible with code compiled with the
10937 @option{-fpcc-struct-return} switch.
10938 Use it to conform to a non-default application binary interface.
10940 @item -fshort-enums
10941 @opindex fshort-enums
10942 Allocate to an @code{enum} type only as many bytes as it needs for the
10943 declared range of possible values. Specifically, the @code{enum} type
10944 will be equivalent to the smallest integer type which has enough room.
10946 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
10947 code that is not binary compatible with code generated without that switch.
10948 Use it to conform to a non-default application binary interface.
10950 @item -fshort-double
10951 @opindex fshort-double
10952 Use the same size for @code{double} as for @code{float}.
10954 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
10955 code that is not binary compatible with code generated without that switch.
10956 Use it to conform to a non-default application binary interface.
10958 @item -fshort-wchar
10959 @opindex fshort-wchar
10960 Override the underlying type for @samp{wchar_t} to be @samp{short
10961 unsigned int} instead of the default for the target. This option is
10962 useful for building programs to run under WINE@.
10964 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
10965 code that is not binary compatible with code generated without that switch.
10966 Use it to conform to a non-default application binary interface.
10968 @item -fshared-data
10969 @opindex fshared-data
10970 Requests that the data and non-@code{const} variables of this
10971 compilation be shared data rather than private data. The distinction
10972 makes sense only on certain operating systems, where shared data is
10973 shared between processes running the same program, while private data
10974 exists in one copy per process.
10977 @opindex fno-common
10978 In C, allocate even uninitialized global variables in the data section of the
10979 object file, rather than generating them as common blocks. This has the
10980 effect that if the same variable is declared (without @code{extern}) in
10981 two different compilations, you will get an error when you link them.
10982 The only reason this might be useful is if you wish to verify that the
10983 program will work on other systems which always work this way.
10987 Ignore the @samp{#ident} directive.
10989 @item -finhibit-size-directive
10990 @opindex finhibit-size-directive
10991 Don't output a @code{.size} assembler directive, or anything else that
10992 would cause trouble if the function is split in the middle, and the
10993 two halves are placed at locations far apart in memory. This option is
10994 used when compiling @file{crtstuff.c}; you should not need to use it
10997 @item -fverbose-asm
10998 @opindex fverbose-asm
10999 Put extra commentary information in the generated assembly code to
11000 make it more readable. This option is generally only of use to those
11001 who actually need to read the generated assembly code (perhaps while
11002 debugging the compiler itself).
11004 @option{-fno-verbose-asm}, the default, causes the
11005 extra information to be omitted and is useful when comparing two assembler
11010 @cindex global offset table
11012 Generate position-independent code (PIC) suitable for use in a shared
11013 library, if supported for the target machine. Such code accesses all
11014 constant addresses through a global offset table (GOT)@. The dynamic
11015 loader resolves the GOT entries when the program starts (the dynamic
11016 loader is not part of GCC; it is part of the operating system). If
11017 the GOT size for the linked executable exceeds a machine-specific
11018 maximum size, you get an error message from the linker indicating that
11019 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11020 instead. (These maximums are 8k on the SPARC and 32k
11021 on the m68k and RS/6000. The 386 has no such limit.)
11023 Position-independent code requires special support, and therefore works
11024 only on certain machines. For the 386, GCC supports PIC for System V
11025 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11026 position-independent.
11030 If supported for the target machine, emit position-independent code,
11031 suitable for dynamic linking and avoiding any limit on the size of the
11032 global offset table. This option makes a difference on the m68k
11035 Position-independent code requires special support, and therefore works
11036 only on certain machines.
11042 These options are similar to @option{-fpic} and @option{-fPIC}, but
11043 generated position independent code can be only linked into executables.
11044 Usually these options are used when @option{-pie} GCC option will be
11045 used during linking.
11047 @item -ffixed-@var{reg}
11049 Treat the register named @var{reg} as a fixed register; generated code
11050 should never refer to it (except perhaps as a stack pointer, frame
11051 pointer or in some other fixed role).
11053 @var{reg} must be the name of a register. The register names accepted
11054 are machine-specific and are defined in the @code{REGISTER_NAMES}
11055 macro in the machine description macro file.
11057 This flag does not have a negative form, because it specifies a
11060 @item -fcall-used-@var{reg}
11061 @opindex fcall-used
11062 Treat the register named @var{reg} as an allocable register that is
11063 clobbered by function calls. It may be allocated for temporaries or
11064 variables that do not live across a call. Functions compiled this way
11065 will not save and restore the register @var{reg}.
11067 It is an error to used this flag with the frame pointer or stack pointer.
11068 Use of this flag for other registers that have fixed pervasive roles in
11069 the machine's execution model will produce disastrous results.
11071 This flag does not have a negative form, because it specifies a
11074 @item -fcall-saved-@var{reg}
11075 @opindex fcall-saved
11076 Treat the register named @var{reg} as an allocable register saved by
11077 functions. It may be allocated even for temporaries or variables that
11078 live across a call. Functions compiled this way will save and restore
11079 the register @var{reg} if they use it.
11081 It is an error to used this flag with the frame pointer or stack pointer.
11082 Use of this flag for other registers that have fixed pervasive roles in
11083 the machine's execution model will produce disastrous results.
11085 A different sort of disaster will result from the use of this flag for
11086 a register in which function values may be returned.
11088 This flag does not have a negative form, because it specifies a
11091 @item -fpack-struct
11092 @opindex fpack-struct
11093 Pack all structure members together without holes.
11095 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11096 code that is not binary compatible with code generated without that switch.
11097 Additionally, it makes the code suboptimal.
11098 Use it to conform to a non-default application binary interface.
11100 @item -finstrument-functions
11101 @opindex finstrument-functions
11102 Generate instrumentation calls for entry and exit to functions. Just
11103 after function entry and just before function exit, the following
11104 profiling functions will be called with the address of the current
11105 function and its call site. (On some platforms,
11106 @code{__builtin_return_address} does not work beyond the current
11107 function, so the call site information may not be available to the
11108 profiling functions otherwise.)
11111 void __cyg_profile_func_enter (void *this_fn,
11113 void __cyg_profile_func_exit (void *this_fn,
11117 The first argument is the address of the start of the current function,
11118 which may be looked up exactly in the symbol table.
11120 This instrumentation is also done for functions expanded inline in other
11121 functions. The profiling calls will indicate where, conceptually, the
11122 inline function is entered and exited. This means that addressable
11123 versions of such functions must be available. If all your uses of a
11124 function are expanded inline, this may mean an additional expansion of
11125 code size. If you use @samp{extern inline} in your C code, an
11126 addressable version of such functions must be provided. (This is
11127 normally the case anyways, but if you get lucky and the optimizer always
11128 expands the functions inline, you might have gotten away without
11129 providing static copies.)
11131 A function may be given the attribute @code{no_instrument_function}, in
11132 which case this instrumentation will not be done. This can be used, for
11133 example, for the profiling functions listed above, high-priority
11134 interrupt routines, and any functions from which the profiling functions
11135 cannot safely be called (perhaps signal handlers, if the profiling
11136 routines generate output or allocate memory).
11138 @item -fstack-check
11139 @opindex fstack-check
11140 Generate code to verify that you do not go beyond the boundary of the
11141 stack. You should specify this flag if you are running in an
11142 environment with multiple threads, but only rarely need to specify it in
11143 a single-threaded environment since stack overflow is automatically
11144 detected on nearly all systems if there is only one stack.
11146 Note that this switch does not actually cause checking to be done; the
11147 operating system must do that. The switch causes generation of code
11148 to ensure that the operating system sees the stack being extended.
11150 @item -fstack-limit-register=@var{reg}
11151 @itemx -fstack-limit-symbol=@var{sym}
11152 @itemx -fno-stack-limit
11153 @opindex fstack-limit-register
11154 @opindex fstack-limit-symbol
11155 @opindex fno-stack-limit
11156 Generate code to ensure that the stack does not grow beyond a certain value,
11157 either the value of a register or the address of a symbol. If the stack
11158 would grow beyond the value, a signal is raised. For most targets,
11159 the signal is raised before the stack overruns the boundary, so
11160 it is possible to catch the signal without taking special precautions.
11162 For instance, if the stack starts at absolute address @samp{0x80000000}
11163 and grows downwards, you can use the flags
11164 @option{-fstack-limit-symbol=__stack_limit} and
11165 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11166 of 128KB@. Note that this may only work with the GNU linker.
11168 @cindex aliasing of parameters
11169 @cindex parameters, aliased
11170 @item -fargument-alias
11171 @itemx -fargument-noalias
11172 @itemx -fargument-noalias-global
11173 @opindex fargument-alias
11174 @opindex fargument-noalias
11175 @opindex fargument-noalias-global
11176 Specify the possible relationships among parameters and between
11177 parameters and global data.
11179 @option{-fargument-alias} specifies that arguments (parameters) may
11180 alias each other and may alias global storage.@*
11181 @option{-fargument-noalias} specifies that arguments do not alias
11182 each other, but may alias global storage.@*
11183 @option{-fargument-noalias-global} specifies that arguments do not
11184 alias each other and do not alias global storage.
11186 Each language will automatically use whatever option is required by
11187 the language standard. You should not need to use these options yourself.
11189 @item -fleading-underscore
11190 @opindex fleading-underscore
11191 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11192 change the way C symbols are represented in the object file. One use
11193 is to help link with legacy assembly code.
11195 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11196 generate code that is not binary compatible with code generated without that
11197 switch. Use it to conform to a non-default application binary interface.
11198 Not all targets provide complete support for this switch.
11200 @item -ftls-model=@var{model}
11201 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11202 The @var{model} argument should be one of @code{global-dynamic},
11203 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11205 The default without @option{-fpic} is @code{initial-exec}; with
11206 @option{-fpic} the default is @code{global-dynamic}.
11211 @node Environment Variables
11212 @section Environment Variables Affecting GCC
11213 @cindex environment variables
11215 @c man begin ENVIRONMENT
11216 This section describes several environment variables that affect how GCC
11217 operates. Some of them work by specifying directories or prefixes to use
11218 when searching for various kinds of files. Some are used to specify other
11219 aspects of the compilation environment.
11221 Note that you can also specify places to search using options such as
11222 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11223 take precedence over places specified using environment variables, which
11224 in turn take precedence over those specified by the configuration of GCC@.
11225 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11226 GNU Compiler Collection (GCC) Internals}.
11231 @c @itemx LC_COLLATE
11233 @c @itemx LC_MONETARY
11234 @c @itemx LC_NUMERIC
11239 @c @findex LC_COLLATE
11240 @findex LC_MESSAGES
11241 @c @findex LC_MONETARY
11242 @c @findex LC_NUMERIC
11246 These environment variables control the way that GCC uses
11247 localization information that allow GCC to work with different
11248 national conventions. GCC inspects the locale categories
11249 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11250 so. These locale categories can be set to any value supported by your
11251 installation. A typical value is @samp{en_UK} for English in the United
11254 The @env{LC_CTYPE} environment variable specifies character
11255 classification. GCC uses it to determine the character boundaries in
11256 a string; this is needed for some multibyte encodings that contain quote
11257 and escape characters that would otherwise be interpreted as a string
11260 The @env{LC_MESSAGES} environment variable specifies the language to
11261 use in diagnostic messages.
11263 If the @env{LC_ALL} environment variable is set, it overrides the value
11264 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11265 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11266 environment variable. If none of these variables are set, GCC
11267 defaults to traditional C English behavior.
11271 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11272 files. GCC uses temporary files to hold the output of one stage of
11273 compilation which is to be used as input to the next stage: for example,
11274 the output of the preprocessor, which is the input to the compiler
11277 @item GCC_EXEC_PREFIX
11278 @findex GCC_EXEC_PREFIX
11279 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11280 names of the subprograms executed by the compiler. No slash is added
11281 when this prefix is combined with the name of a subprogram, but you can
11282 specify a prefix that ends with a slash if you wish.
11284 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11285 an appropriate prefix to use based on the pathname it was invoked with.
11287 If GCC cannot find the subprogram using the specified prefix, it
11288 tries looking in the usual places for the subprogram.
11290 The default value of @env{GCC_EXEC_PREFIX} is
11291 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
11292 of @code{prefix} when you ran the @file{configure} script.
11294 Other prefixes specified with @option{-B} take precedence over this prefix.
11296 This prefix is also used for finding files such as @file{crt0.o} that are
11299 In addition, the prefix is used in an unusual way in finding the
11300 directories to search for header files. For each of the standard
11301 directories whose name normally begins with @samp{/usr/local/lib/gcc}
11302 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11303 replacing that beginning with the specified prefix to produce an
11304 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11305 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11306 These alternate directories are searched first; the standard directories
11309 @item COMPILER_PATH
11310 @findex COMPILER_PATH
11311 The value of @env{COMPILER_PATH} is a colon-separated list of
11312 directories, much like @env{PATH}. GCC tries the directories thus
11313 specified when searching for subprograms, if it can't find the
11314 subprograms using @env{GCC_EXEC_PREFIX}.
11317 @findex LIBRARY_PATH
11318 The value of @env{LIBRARY_PATH} is a colon-separated list of
11319 directories, much like @env{PATH}. When configured as a native compiler,
11320 GCC tries the directories thus specified when searching for special
11321 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11322 using GCC also uses these directories when searching for ordinary
11323 libraries for the @option{-l} option (but directories specified with
11324 @option{-L} come first).
11328 @cindex locale definition
11329 This variable is used to pass locale information to the compiler. One way in
11330 which this information is used is to determine the character set to be used
11331 when character literals, string literals and comments are parsed in C and C++.
11332 When the compiler is configured to allow multibyte characters,
11333 the following values for @env{LANG} are recognized:
11337 Recognize JIS characters.
11339 Recognize SJIS characters.
11341 Recognize EUCJP characters.
11344 If @env{LANG} is not defined, or if it has some other value, then the
11345 compiler will use mblen and mbtowc as defined by the default locale to
11346 recognize and translate multibyte characters.
11350 Some additional environments variables affect the behavior of the
11353 @include cppenv.texi
11357 @node Precompiled Headers
11358 @section Using Precompiled Headers
11359 @cindex precompiled headers
11360 @cindex speed of compilation
11362 Often large projects have many header files that are included in every
11363 source file. The time the compiler takes to process these header files
11364 over and over again can account for nearly all of the time required to
11365 build the project. To make builds faster, GCC allows users to
11366 `precompile' a header file; then, if builds can use the precompiled
11367 header file they will be much faster.
11369 To create a precompiled header file, simply compile it as you would any
11370 other file, if necessary using the @option{-x} option to make the driver
11371 treat it as a C or C++ header file. You will probably want to use a
11372 tool like @command{make} to keep the precompiled header up-to-date when
11373 the headers it contains change.
11375 A precompiled header file will be searched for when @code{#include} is
11376 seen in the compilation. As it searches for the included file
11377 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
11378 compiler looks for a precompiled header in each directory just before it
11379 looks for the include file in that directory. The name searched for is
11380 the name specified in the @code{#include} with @samp{.gch} appended. If
11381 the precompiled header file can't be used, it is ignored.
11383 For instance, if you have @code{#include "all.h"}, and you have
11384 @file{all.h.gch} in the same directory as @file{all.h}, then the
11385 precompiled header file will be used if possible, and the original
11386 header will be used otherwise.
11388 Alternatively, you might decide to put the precompiled header file in a
11389 directory and use @option{-I} to ensure that directory is searched
11390 before (or instead of) the directory containing the original header.
11391 Then, if you want to check that the precompiled header file is always
11392 used, you can put a file of the same name as the original header in this
11393 directory containing an @code{#error} command.
11395 This also works with @option{-include}. So yet another way to use
11396 precompiled headers, good for projects not designed with precompiled
11397 header files in mind, is to simply take most of the header files used by
11398 a project, include them from another header file, precompile that header
11399 file, and @option{-include} the precompiled header. If the header files
11400 have guards against multiple inclusion, they will be skipped because
11401 they've already been included (in the precompiled header).
11403 If you need to precompile the same header file for different
11404 languages, targets, or compiler options, you can instead make a
11405 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11406 header in the directory. (It doesn't matter what you call the files
11407 in the directory, every precompiled header in the directory will be
11408 considered.) The first precompiled header encountered in the
11409 directory that is valid for this compilation will be used; they're
11410 searched in no particular order.
11412 There are many other possibilities, limited only by your imagination,
11413 good sense, and the constraints of your build system.
11415 A precompiled header file can be used only when these conditions apply:
11419 Only one precompiled header can be used in a particular compilation.
11421 A precompiled header can't be used once the first C token is seen. You
11422 can have preprocessor directives before a precompiled header; you can
11423 even include a precompiled header from inside another header, so long as
11424 there are no C tokens before the @code{#include}.
11426 The precompiled header file must be produced for the same language as
11427 the current compilation. You can't use a C precompiled header for a C++
11430 The precompiled header file must be produced by the same compiler
11431 version and configuration as the current compilation is using.
11432 The easiest way to guarantee this is to use the same compiler binary
11433 for creating and using precompiled headers.
11435 Any macros defined before the precompiled header (including with
11436 @option{-D}) must either be defined in the same way as when the
11437 precompiled header was generated, or must not affect the precompiled
11438 header, which usually means that the they don't appear in the
11439 precompiled header at all.
11441 Certain command-line options must be defined in the same way as when the
11442 precompiled header was generated. At present, it's not clear which
11443 options are safe to change and which are not; the safest choice is to
11444 use exactly the same options when generating and using the precompiled
11448 For all of these but the last, the compiler will automatically ignore
11449 the precompiled header if the conditions aren't met. For the last item,
11450 some option changes will cause the precompiled header to be rejected,
11451 but not all incompatible option combinations have yet been found. If
11452 you find a new incompatible combination, please consider filing a bug
11453 report, see @ref{Bugs}.
11455 @node Running Protoize
11456 @section Running Protoize
11458 The program @code{protoize} is an optional part of GCC@. You can use
11459 it to add prototypes to a program, thus converting the program to ISO
11460 C in one respect. The companion program @code{unprotoize} does the
11461 reverse: it removes argument types from any prototypes that are found.
11463 When you run these programs, you must specify a set of source files as
11464 command line arguments. The conversion programs start out by compiling
11465 these files to see what functions they define. The information gathered
11466 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11468 After scanning comes actual conversion. The specified files are all
11469 eligible to be converted; any files they include (whether sources or
11470 just headers) are eligible as well.
11472 But not all the eligible files are converted. By default,
11473 @code{protoize} and @code{unprotoize} convert only source and header
11474 files in the current directory. You can specify additional directories
11475 whose files should be converted with the @option{-d @var{directory}}
11476 option. You can also specify particular files to exclude with the
11477 @option{-x @var{file}} option. A file is converted if it is eligible, its
11478 directory name matches one of the specified directory names, and its
11479 name within the directory has not been excluded.
11481 Basic conversion with @code{protoize} consists of rewriting most
11482 function definitions and function declarations to specify the types of
11483 the arguments. The only ones not rewritten are those for varargs
11486 @code{protoize} optionally inserts prototype declarations at the
11487 beginning of the source file, to make them available for any calls that
11488 precede the function's definition. Or it can insert prototype
11489 declarations with block scope in the blocks where undeclared functions
11492 Basic conversion with @code{unprotoize} consists of rewriting most
11493 function declarations to remove any argument types, and rewriting
11494 function definitions to the old-style pre-ISO form.
11496 Both conversion programs print a warning for any function declaration or
11497 definition that they can't convert. You can suppress these warnings
11500 The output from @code{protoize} or @code{unprotoize} replaces the
11501 original source file. The original file is renamed to a name ending
11502 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11503 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11504 for DOS) file already exists, then the source file is simply discarded.
11506 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11507 scan the program and collect information about the functions it uses.
11508 So neither of these programs will work until GCC is installed.
11510 Here is a table of the options you can use with @code{protoize} and
11511 @code{unprotoize}. Each option works with both programs unless
11515 @item -B @var{directory}
11516 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11517 usual directory (normally @file{/usr/local/lib}). This file contains
11518 prototype information about standard system functions. This option
11519 applies only to @code{protoize}.
11521 @item -c @var{compilation-options}
11522 Use @var{compilation-options} as the options when running @command{gcc} to
11523 produce the @samp{.X} files. The special option @option{-aux-info} is
11524 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11526 Note that the compilation options must be given as a single argument to
11527 @code{protoize} or @code{unprotoize}. If you want to specify several
11528 @command{gcc} options, you must quote the entire set of compilation options
11529 to make them a single word in the shell.
11531 There are certain @command{gcc} arguments that you cannot use, because they
11532 would produce the wrong kind of output. These include @option{-g},
11533 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11534 the @var{compilation-options}, they are ignored.
11537 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11538 systems) instead of @samp{.c}. This is convenient if you are converting
11539 a C program to C++. This option applies only to @code{protoize}.
11542 Add explicit global declarations. This means inserting explicit
11543 declarations at the beginning of each source file for each function
11544 that is called in the file and was not declared. These declarations
11545 precede the first function definition that contains a call to an
11546 undeclared function. This option applies only to @code{protoize}.
11548 @item -i @var{string}
11549 Indent old-style parameter declarations with the string @var{string}.
11550 This option applies only to @code{protoize}.
11552 @code{unprotoize} converts prototyped function definitions to old-style
11553 function definitions, where the arguments are declared between the
11554 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11555 uses five spaces as the indentation. If you want to indent with just
11556 one space instead, use @option{-i " "}.
11559 Keep the @samp{.X} files. Normally, they are deleted after conversion
11563 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11564 a prototype declaration for each function in each block which calls the
11565 function without any declaration. This option applies only to
11569 Make no real changes. This mode just prints information about the conversions
11570 that would have been done without @option{-n}.
11573 Make no @samp{.save} files. The original files are simply deleted.
11574 Use this option with caution.
11576 @item -p @var{program}
11577 Use the program @var{program} as the compiler. Normally, the name
11578 @file{gcc} is used.
11581 Work quietly. Most warnings are suppressed.
11584 Print the version number, just like @option{-v} for @command{gcc}.
11587 If you need special compiler options to compile one of your program's
11588 source files, then you should generate that file's @samp{.X} file
11589 specially, by running @command{gcc} on that source file with the
11590 appropriate options and the option @option{-aux-info}. Then run
11591 @code{protoize} on the entire set of files. @code{protoize} will use
11592 the existing @samp{.X} file because it is newer than the source file.
11596 gcc -Dfoo=bar file1.c -aux-info file1.X
11601 You need to include the special files along with the rest in the
11602 @code{protoize} command, even though their @samp{.X} files already
11603 exist, because otherwise they won't get converted.
11605 @xref{Protoize Caveats}, for more information on how to use
11606 @code{protoize} successfully.