1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003 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 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 -fexternal-templates @gol
178 -falt-external-templates @gol
179 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
180 -fno-implicit-templates @gol
181 -fno-implicit-inline-templates @gol
182 -fno-implement-inlines -fms-extensions @gol
183 -fno-nonansi-builtins -fno-operator-names @gol
184 -fno-optional-diags -fpermissive @gol
185 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
186 -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
187 -fno-default-inline -Wabi -Wctor-dtor-privacy @gol
188 -Wnon-virtual-dtor -Wreorder @gol
189 -Weffc++ -Wno-deprecated @gol
190 -Wno-non-template-friend -Wold-style-cast @gol
191 -Woverloaded-virtual -Wno-pmf-conversions @gol
192 -Wsign-promo -Wsynth}
194 @item Objective-C Language Options
195 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
197 -fconstant-string-class=@var{class-name} @gol
198 -fgnu-runtime -fnext-runtime @gol
199 -fno-nil-receivers @gol
200 -fobjc-exceptions @gol
201 -freplace-objc-classes @gol
204 -Wno-protocol -Wselector -Wundeclared-selector}
206 @item Language Independent Options
207 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
208 @gccoptlist{-fmessage-length=@var{n} @gol
209 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
211 @item Warning Options
212 @xref{Warning Options,,Options to Request or Suppress Warnings}.
213 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
214 -w -Wextra -Wall -Waggregate-return @gol
215 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
216 -Wconversion -Wno-deprecated-declarations @gol
217 -Wdisabled-optimization -Wno-div-by-zero -Wendif-labels @gol
218 -Werror -Werror-implicit-function-declaration @gol
219 -Wfloat-equal -Wformat -Wformat=2 @gol
220 -Wno-format-extra-args -Wformat-nonliteral @gol
221 -Wformat-security -Wformat-y2k @gol
222 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
223 -Wimport -Wno-import -Winit-self -Winline @gol
224 -Wno-invalid-offsetof -Winvalid-pch @gol
225 -Wlarger-than-@var{len} -Wlong-long @gol
226 -Wmain -Wmissing-braces @gol
227 -Wmissing-format-attribute -Wmissing-noreturn @gol
228 -Wno-multichar -Wnonnull -Wpacked -Wpadded @gol
229 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
230 -Wreturn-type -Wsequence-point -Wshadow @gol
231 -Wsign-compare -Wstrict-aliasing @gol
232 -Wswitch -Wswitch-default -Wswitch-enum @gol
233 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
234 -Wunknown-pragmas -Wunreachable-code @gol
235 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
236 -Wunused-value -Wunused-variable -Wwrite-strings}
238 @item C-only Warning Options
239 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
240 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
241 -Wstrict-prototypes -Wtraditional @gol
242 -Wdeclaration-after-statement}
244 @item Debugging Options
245 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
246 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
247 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
248 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
249 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
250 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
251 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
252 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
253 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs @gol
254 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
255 -ftest-coverage -ftime-report @gol
256 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
257 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
258 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
259 -print-multi-directory -print-multi-lib @gol
260 -print-prog-name=@var{program} -print-search-dirs -Q @gol
263 @item Optimization Options
264 @xref{Optimize Options,,Options that Control Optimization}.
265 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
266 -falign-labels=@var{n} -falign-loops=@var{n} @gol
267 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
268 -fbranch-target-load-optimize2 -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 -freduce-all-givs -fregmove -frename-registers @gol
286 -freorder-blocks -freorder-functions @gol
287 -frerun-cse-after-loop -frerun-loop-opt @gol
288 -frounding-math -fschedule-insns -fschedule-insns2 @gol
289 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
290 -fsched-spec-load-dangerous @gol
291 -fsched-stalled-insns=@var{n} -sched-stalled-insns-dep=@var{n} @gol
292 -fsched2-use-superblocks @gol
293 -fsched2-use-traces -fsignaling-nans @gol
294 -fsingle-precision-constant @gol
295 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
296 -funroll-all-loops -funroll-loops -fpeel-loops @gol
297 -funswitch-loops -fold-unroll-loops -fold-unroll-all-loops @gol
298 --param @var{name}=@var{value}
299 -O -O0 -O1 -O2 -O3 -Os}
301 @item Preprocessor Options
302 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
303 @gccoptlist{-A@var{question}=@var{answer} @gol
304 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
305 -C -dD -dI -dM -dN @gol
306 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
307 -idirafter @var{dir} @gol
308 -include @var{file} -imacros @var{file} @gol
309 -iprefix @var{file} -iwithprefix @var{dir} @gol
310 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
311 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
312 -P -fworking-directory -remap @gol
313 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
314 -Xpreprocessor @var{option}}
316 @item Assembler Option
317 @xref{Assembler Options,,Passing Options to the Assembler}.
318 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
321 @xref{Link Options,,Options for Linking}.
322 @gccoptlist{@var{object-file-name} -l@var{library} @gol
323 -nostartfiles -nodefaultlibs -nostdlib -pie @gol
324 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
325 -Wl,@var{option} -Xlinker @var{option} @gol
328 @item Directory Options
329 @xref{Directory Options,,Options for Directory Search}.
330 @gccoptlist{-B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
333 @c I wrote this xref this way to avoid overfull hbox. -- rms
334 @xref{Target Options}.
335 @gccoptlist{-V @var{version} -b @var{machine}}
337 @item Machine Dependent Options
338 @xref{Submodel Options,,Hardware Models and Configurations}.
340 @emph{M680x0 Options}
341 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
342 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
343 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
344 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
345 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
347 @emph{M68hc1x Options}
348 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
349 -mauto-incdec -minmax -mlong-calls -mshort @gol
350 -msoft-reg-count=@var{count}}
353 @gccoptlist{-mg -mgnu -munix}
356 @gccoptlist{-mcpu=@var{cpu-type} @gol
357 -mtune=@var{cpu-type} @gol
358 -mcmodel=@var{code-model} @gol
360 -mapp-regs -mbroken-saverestore -mcypress @gol
361 -mfaster-structs -mflat @gol
362 -mfpu -mhard-float -mhard-quad-float @gol
363 -mimpure-text -mlittle-endian -mlive-g0 -mno-app-regs @gol
364 -mno-faster-structs -mno-flat -mno-fpu @gol
365 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
366 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
367 -msupersparc -munaligned-doubles -mv8}
370 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
371 -mapcs-26 -mapcs-32 @gol
372 -mapcs-stack-check -mno-apcs-stack-check @gol
373 -mapcs-float -mno-apcs-float @gol
374 -mapcs-reentrant -mno-apcs-reentrant @gol
375 -msched-prolog -mno-sched-prolog @gol
376 -mlittle-endian -mbig-endian -mwords-little-endian @gol
377 -malignment-traps -mno-alignment-traps @gol
378 -msoft-float -mhard-float -mfpe @gol
379 -mthumb-interwork -mno-thumb-interwork @gol
380 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
381 -mstructure-size-boundary=@var{n} @gol
382 -mabort-on-noreturn @gol
383 -mlong-calls -mno-long-calls @gol
384 -msingle-pic-base -mno-single-pic-base @gol
385 -mpic-register=@var{reg} @gol
386 -mnop-fun-dllimport @gol
387 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
388 -mpoke-function-name @gol
390 -mtpcs-frame -mtpcs-leaf-frame @gol
391 -mcaller-super-interworking -mcallee-super-interworking}
393 @emph{MN10200 Options}
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{-m32rx -m32r -mcode-model=@var{model-type} @gol
404 -msdata=@var{sdata-type} -G @var{num}}
407 @gccoptlist{-m88000 -m88100 -m88110 -mbig-pic @gol
408 -mcheck-zero-division -mhandle-large-shift @gol
409 -midentify-revision -mno-check-zero-division @gol
410 -mno-ocs-debug-info -mno-ocs-frame-position @gol
411 -mno-optimize-arg-area -mno-serialize-volatile @gol
412 -mno-underscores -mocs-debug-info @gol
413 -mocs-frame-position -moptimize-arg-area @gol
414 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
415 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
416 -mversion-03.00 -mwarn-passed-structs}
418 @emph{RS/6000 and PowerPC Options}
419 @gccoptlist{-mcpu=@var{cpu-type} @gol
420 -mtune=@var{cpu-type} @gol
421 -mpower -mno-power -mpower2 -mno-power2 @gol
422 -mpowerpc -mpowerpc64 -mno-powerpc @gol
423 -maltivec -mno-altivec @gol
424 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
425 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
426 -mnew-mnemonics -mold-mnemonics @gol
427 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
428 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
429 -malign-power -malign-natural @gol
430 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
431 -mstring -mno-string -mupdate -mno-update @gol
432 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
433 -mstrict-align -mno-strict-align -mrelocatable @gol
434 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
435 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
436 -mdynamic-no-pic @gol
437 -mprioritize-restricted-insns=@var{priority} @gol
438 -msched-costly-dep=@var{dependence_type} @gol
439 -minsert-sched-nops=@var{scheme} @gol
440 -mcall-sysv -mcall-netbsd @gol
441 -maix-struct-return -msvr4-struct-return @gol
442 -mabi=altivec -mabi=no-altivec @gol
443 -mabi=spe -mabi=no-spe @gol
444 -misel=yes -misel=no @gol
445 -mspe=yes -mspe=no @gol
446 -mfloat-gprs=yes -mfloat-gprs=no @gol
447 -mprototype -mno-prototype @gol
448 -msim -mmvme -mads -myellowknife -memb -msdata @gol
449 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
451 @emph{Darwin Options}
452 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
453 -arch_only -bind_at_load -bundle -bundle_loader @gol
454 -client_name -compatibility_version -current_version @gol
455 -dependency-file -dylib_file -dylinker_install_name @gol
456 -dynamic -dynamiclib -exported_symbols_list @gol
457 -filelist -flat_namespace -force_cpusubtype_ALL @gol
458 -force_flat_namespace -headerpad_max_install_names @gol
459 -image_base -init -install_name -keep_private_externs @gol
460 -multi_module -multiply_defined -multiply_defined_unused @gol
461 -noall_load -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
462 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
463 -private_bundle -read_only_relocs -sectalign @gol
464 -sectobjectsymbols -whyload -seg1addr @gol
465 -sectcreate -sectobjectsymbols -sectorder @gol
466 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
467 -segprot -segs_read_only_addr -segs_read_write_addr @gol
468 -single_module -static -sub_library -sub_umbrella @gol
469 -twolevel_namespace -umbrella -undefined @gol
470 -unexported_symbols_list -weak_reference_mismatches @gol
474 @gccoptlist{-mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
475 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
476 -mminimum-fp-blocks -mnohc-struct-return}
479 @gccoptlist{-mabicalls -march=@var{cpu-type} -mtune=@var{cpu=type} @gol
480 -mcpu=@var{cpu-type} -membedded-data -muninit-const-in-rodata @gol
481 -membedded-pic -mfp32 -mfp64 -mfused-madd -mno-fused-madd @gol
482 -mgas -mgp32 -mgp64 -mhard-float -mint64 -mips1 @gol
483 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
484 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
485 -mmips-as -mmips-tfile -mno-abicalls -mxgot @gol
486 -mno-embedded-data -mno-uninit-const-in-rodata @gol
487 -mno-embedded-pic -mno-long-calls @gol
488 -mno-memcpy -mno-mips-tfile -mno-rnames @gol
489 -mrnames -msoft-float @gol
490 -m4650 -msingle-float -mmad @gol
491 -EL -EB -G @var{num} -nocpp @gol
492 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi -mabi-fake-default @gol
493 -mfix7000 -mfix-sb1 -mno-fix-sb1 @gol
494 -mno-crt0 -mflush-func=@var{func} -mno-flush-func @gol
495 -mbranch-likely -mno-branch-likely}
497 @emph{i386 and x86-64 Options}
498 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
499 -mfpmath=@var{unit} @gol
500 -masm=@var{dialect} -mno-fancy-math-387 @gol
501 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
502 -mno-wide-multiply -mrtd -malign-double @gol
503 -mpreferred-stack-boundary=@var{num} @gol
504 -mmmx -msse -msse2 -mpni -m3dnow @gol
505 -mthreads -mno-align-stringops -minline-all-stringops @gol
506 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
507 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
508 -mno-red-zone -mno-tls-direct-seg-refs @gol
509 -mcmodel=@var{code-model} @gol
513 @gccoptlist{-march=@var{architecture-type} @gol
514 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
515 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
516 -mjump-in-delay -mlinker-opt -mlong-calls @gol
517 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
518 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
519 -mno-jump-in-delay -mno-long-load-store @gol
520 -mno-portable-runtime -mno-soft-float @gol
521 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
522 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
523 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
524 -nolibdld -static -threads}
526 @emph{Intel 960 Options}
527 @gccoptlist{-m@var{cpu-type} -masm-compat -mclean-linkage @gol
528 -mcode-align -mcomplex-addr -mleaf-procedures @gol
529 -mic-compat -mic2.0-compat -mic3.0-compat @gol
530 -mintel-asm -mno-clean-linkage -mno-code-align @gol
531 -mno-complex-addr -mno-leaf-procedures @gol
532 -mno-old-align -mno-strict-align -mno-tail-call @gol
533 -mnumerics -mold-align -msoft-float -mstrict-align @gol
536 @emph{DEC Alpha Options}
537 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
538 -mieee -mieee-with-inexact -mieee-conformant @gol
539 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
540 -mtrap-precision=@var{mode} -mbuild-constants @gol
541 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
542 -mbwx -mmax -mfix -mcix @gol
543 -mfloat-vax -mfloat-ieee @gol
544 -mexplicit-relocs -msmall-data -mlarge-data @gol
545 -msmall-text -mlarge-text @gol
546 -mmemory-latency=@var{time}}
548 @emph{DEC Alpha/VMS Options}
549 @gccoptlist{-mvms-return-codes}
551 @emph{H8/300 Options}
552 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
555 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
556 -m4-nofpu -m4-single-only -m4-single -m4 @gol
557 -m5-64media -m5-64media-nofpu @gol
558 -m5-32media -m5-32media-nofpu @gol
559 -m5-compact -m5-compact-nofpu @gol
560 -mb -ml -mdalign -mrelax @gol
561 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
562 -mieee -misize -mpadstruct -mspace @gol
563 -mprefergot -musermode}
565 @emph{System V Options}
566 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
569 @gccoptlist{-EB -EL @gol
570 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
571 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
573 @emph{TMS320C3x/C4x Options}
574 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
575 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
576 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
577 -mparallel-insns -mparallel-mpy -mpreserve-float}
580 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
581 -mprolog-function -mno-prolog-function -mspace @gol
582 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
583 -mapp-regs -mno-app-regs @gol
584 -mdisable-callt -mno-disable-callt @gol
590 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
591 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
592 -mregparam -mnoregparam -msb -mnosb @gol
593 -mbitfield -mnobitfield -mhimem -mnohimem}
596 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
597 -mcall-prologues -mno-tablejump -mtiny-stack}
600 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
601 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
602 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
603 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
604 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
607 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
608 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
609 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
610 -mno-base-addresses -msingle-exit -mno-single-exit}
613 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
614 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
615 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
616 -minline-float-divide-max-throughput @gol
617 -minline-int-divide-min-latency @gol
618 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
619 -mfixed-range=@var{register-range}}
622 @gccoptlist{-mextmem -mextmemory -monchip -mno-asm-optimize @gol
623 -masm-optimize -mbranch-cost=@var{n} -mcond-exec=@var{n}}
625 @emph{S/390 and zSeries Options}
626 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
627 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
628 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
629 -m64 -m31 -mdebug -mno-debug -mesa -mzarch -mfused-madd -mno-fused-madd}
632 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
633 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
634 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
635 -mstack-align -mdata-align -mconst-align @gol
636 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
637 -melf -maout -melinux -mlinux -sim -sim2}
639 @emph{PDP-11 Options}
640 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
641 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
642 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
643 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
644 -mbranch-expensive -mbranch-cheap @gol
645 -msplit -mno-split -munix-asm -mdec-asm}
647 @emph{Xstormy16 Options}
650 @emph{Xtensa Options}
651 @gccoptlist{-mconst16 -mno-const16 @gol
652 -mfused-madd -mno-fused-madd @gol
653 -mtext-section-literals -mno-text-section-literals @gol
654 -mtarget-align -mno-target-align @gol
655 -mlongcalls -mno-longcalls}
658 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
659 -mhard-float -msoft-float @gol
660 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
661 -mdouble -mno-double @gol
662 -mmedia -mno-media -mmuladd -mno-muladd @gol
663 -mlibrary-pic -macc-4 -macc-8 @gol
664 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
665 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
666 -mvliw-branch -mno-vliw-branch @gol
667 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
668 -mno-nested-cond-exec -mtomcat-stats @gol
671 @item Code Generation Options
672 @xref{Code Gen Options,,Options for Code Generation Conventions}.
673 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
674 -ffixed-@var{reg} -fexceptions @gol
675 -fnon-call-exceptions -funwind-tables @gol
676 -fasynchronous-unwind-tables @gol
677 -finhibit-size-directive -finstrument-functions @gol
678 -fno-common -fno-ident -fno-gnu-linker @gol
679 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
680 -freg-struct-return -fshared-data -fshort-enums @gol
681 -fshort-double -fshort-wchar @gol
682 -fverbose-asm -fpack-struct -fstack-check @gol
683 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
684 -fargument-alias -fargument-noalias @gol
685 -fargument-noalias-global -fleading-underscore @gol
686 -ftls-model=@var{model} @gol
687 -ftrapv -fwrapv -fbounds-check}
691 * Overall Options:: Controlling the kind of output:
692 an executable, object files, assembler files,
693 or preprocessed source.
694 * C Dialect Options:: Controlling the variant of C language compiled.
695 * C++ Dialect Options:: Variations on C++.
696 * Objective-C Dialect Options:: Variations on Objective-C.
697 * Language Independent Options:: Controlling how diagnostics should be
699 * Warning Options:: How picky should the compiler be?
700 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
701 * Optimize Options:: How much optimization?
702 * Preprocessor Options:: Controlling header files and macro definitions.
703 Also, getting dependency information for Make.
704 * Assembler Options:: Passing options to the assembler.
705 * Link Options:: Specifying libraries and so on.
706 * Directory Options:: Where to find header files and libraries.
707 Where to find the compiler executable files.
708 * Spec Files:: How to pass switches to sub-processes.
709 * Target Options:: Running a cross-compiler, or an old version of GCC.
712 @node Overall Options
713 @section Options Controlling the Kind of Output
715 Compilation can involve up to four stages: preprocessing, compilation
716 proper, assembly and linking, always in that order. GCC is capable of
717 preprocessing and compiling several files either into several
718 assembler input files, or into one assembler input file; then each
719 assembler input file produces an object file, and linking combines all
720 the object files (those newly compiled, and those specified as input)
721 into an executable file.
723 @cindex file name suffix
724 For any given input file, the file name suffix determines what kind of
729 C source code which must be preprocessed.
732 C source code which should not be preprocessed.
735 C++ source code which should not be preprocessed.
738 Objective-C source code. Note that you must link with the library
739 @file{libobjc.a} to make an Objective-C program work.
742 Objective-C source code which should not be preprocessed.
745 C or C++ header file to be turned into a precompiled header.
749 @itemx @var{file}.cxx
750 @itemx @var{file}.cpp
751 @itemx @var{file}.CPP
752 @itemx @var{file}.c++
754 C++ source code which must be preprocessed. Note that in @samp{.cxx},
755 the last two letters must both be literally @samp{x}. Likewise,
756 @samp{.C} refers to a literal capital C@.
760 C++ header file to be turned into a precompiled header.
763 @itemx @var{file}.for
764 @itemx @var{file}.FOR
765 Fortran source code which should not be preprocessed.
768 @itemx @var{file}.fpp
769 @itemx @var{file}.FPP
770 Fortran source code which must be preprocessed (with the traditional
774 Fortran source code which must be preprocessed with a RATFOR
775 preprocessor (not included with GCC)@.
777 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
778 Using and Porting GNU Fortran}, for more details of the handling of
781 @c FIXME: Descriptions of Java file types.
788 Ada source code file which contains a library unit declaration (a
789 declaration of a package, subprogram, or generic, or a generic
790 instantiation), or a library unit renaming declaration (a package,
791 generic, or subprogram renaming declaration). Such files are also
794 @itemx @var{file}.adb
795 Ada source code file containing a library unit body (a subprogram or
796 package body). Such files are also called @dfn{bodies}.
798 @c GCC also knows about some suffixes for languages not yet included:
807 Assembler code which must be preprocessed.
810 An object file to be fed straight into linking.
811 Any file name with no recognized suffix is treated this way.
815 You can specify the input language explicitly with the @option{-x} option:
818 @item -x @var{language}
819 Specify explicitly the @var{language} for the following input files
820 (rather than letting the compiler choose a default based on the file
821 name suffix). This option applies to all following input files until
822 the next @option{-x} option. Possible values for @var{language} are:
824 c c-header cpp-output
825 c++ c++-header c++-cpp-output
826 objective-c objective-c-header objc-cpp-output
827 assembler assembler-with-cpp
829 f77 f77-cpp-input ratfor
835 Turn off any specification of a language, so that subsequent files are
836 handled according to their file name suffixes (as they are if @option{-x}
837 has not been used at all).
839 @item -pass-exit-codes
840 @opindex pass-exit-codes
841 Normally the @command{gcc} program will exit with the code of 1 if any
842 phase of the compiler returns a non-success return code. If you specify
843 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
844 numerically highest error produced by any phase that returned an error
848 If you only want some of the stages of compilation, you can use
849 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
850 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
851 @command{gcc} is to stop. Note that some combinations (for example,
852 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
857 Compile or assemble the source files, but do not link. The linking
858 stage simply is not done. The ultimate output is in the form of an
859 object file for each source file.
861 By default, the object file name for a source file is made by replacing
862 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
864 Unrecognized input files, not requiring compilation or assembly, are
869 Stop after the stage of compilation proper; do not assemble. The output
870 is in the form of an assembler code file for each non-assembler input
873 By default, the assembler file name for a source file is made by
874 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
876 Input files that don't require compilation are ignored.
880 Stop after the preprocessing stage; do not run the compiler proper. The
881 output is in the form of preprocessed source code, which is sent to the
884 Input files which don't require preprocessing are ignored.
886 @cindex output file option
889 Place output in file @var{file}. This applies regardless to whatever
890 sort of output is being produced, whether it be an executable file,
891 an object file, an assembler file or preprocessed C code.
893 If you specify @option{-o} when compiling more than one input file, or
894 you are producing an executable file as output, all the source files
895 on the command line will be compiled at once.
897 If @option{-o} is not specified, the default is to put an executable file
898 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
899 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
900 all preprocessed C source on standard output.
904 Print (on standard error output) the commands executed to run the stages
905 of compilation. Also print the version number of the compiler driver
906 program and of the preprocessor and the compiler proper.
910 Like @option{-v} except the commands are not executed and all command
911 arguments are quoted. This is useful for shell scripts to capture the
912 driver-generated command lines.
916 Use pipes rather than temporary files for communication between the
917 various stages of compilation. This fails to work on some systems where
918 the assembler is unable to read from a pipe; but the GNU assembler has
923 Print (on the standard output) a description of the command line options
924 understood by @command{gcc}. If the @option{-v} option is also specified
925 then @option{--help} will also be passed on to the various processes
926 invoked by @command{gcc}, so that they can display the command line options
927 they accept. If the @option{-Wextra} option is also specified then command
928 line options which have no documentation associated with them will also
933 Print (on the standard output) a description of target specific command
934 line options for each tool.
938 Display the version number and copyrights of the invoked GCC.
942 @section Compiling C++ Programs
944 @cindex suffixes for C++ source
945 @cindex C++ source file suffixes
946 C++ source files conventionally use one of the suffixes @samp{.C},
947 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
948 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
949 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
950 files with these names and compiles them as C++ programs even if you
951 call the compiler the same way as for compiling C programs (usually
952 with the name @command{gcc}).
956 However, C++ programs often require class libraries as well as a
957 compiler that understands the C++ language---and under some
958 circumstances, you might want to compile programs or header files from
959 standard input, or otherwise without a suffix that flags them as C++
960 programs. You might also like to precompile a C header file with a
961 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
962 program that calls GCC with the default language set to C++, and
963 automatically specifies linking against the C++ library. On many
964 systems, @command{g++} is also installed with the name @command{c++}.
966 @cindex invoking @command{g++}
967 When you compile C++ programs, you may specify many of the same
968 command-line options that you use for compiling programs in any
969 language; or command-line options meaningful for C and related
970 languages; or options that are meaningful only for C++ programs.
971 @xref{C Dialect Options,,Options Controlling C Dialect}, for
972 explanations of options for languages related to C@.
973 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
974 explanations of options that are meaningful only for C++ programs.
976 @node C Dialect Options
977 @section Options Controlling C Dialect
978 @cindex dialect options
979 @cindex language dialect options
980 @cindex options, dialect
982 The following options control the dialect of C (or languages derived
983 from C, such as C++ and Objective-C) that the compiler accepts:
990 In C mode, support all ISO C90 programs. In C++ mode,
991 remove GNU extensions that conflict with ISO C++.
993 This turns off certain features of GCC that are incompatible with ISO
994 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
995 such as the @code{asm} and @code{typeof} keywords, and
996 predefined macros such as @code{unix} and @code{vax} that identify the
997 type of system you are using. It also enables the undesirable and
998 rarely used ISO trigraph feature. For the C compiler,
999 it disables recognition of C++ style @samp{//} comments as well as
1000 the @code{inline} keyword.
1002 The alternate keywords @code{__asm__}, @code{__extension__},
1003 @code{__inline__} and @code{__typeof__} continue to work despite
1004 @option{-ansi}. You would not want to use them in an ISO C program, of
1005 course, but it is useful to put them in header files that might be included
1006 in compilations done with @option{-ansi}. Alternate predefined macros
1007 such as @code{__unix__} and @code{__vax__} are also available, with or
1008 without @option{-ansi}.
1010 The @option{-ansi} option does not cause non-ISO programs to be
1011 rejected gratuitously. For that, @option{-pedantic} is required in
1012 addition to @option{-ansi}. @xref{Warning Options}.
1014 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1015 option is used. Some header files may notice this macro and refrain
1016 from declaring certain functions or defining certain macros that the
1017 ISO standard doesn't call for; this is to avoid interfering with any
1018 programs that might use these names for other things.
1020 Functions which would normally be built in but do not have semantics
1021 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1022 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1023 built-in functions provided by GCC}, for details of the functions
1028 Determine the language standard. This option is currently only
1029 supported when compiling C or C++. A value for this option must be
1030 provided; possible values are
1035 ISO C90 (same as @option{-ansi}).
1037 @item iso9899:199409
1038 ISO C90 as modified in amendment 1.
1044 ISO C99. Note that this standard is not yet fully supported; see
1045 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1046 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1049 Default, ISO C90 plus GNU extensions (including some C99 features).
1053 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1054 this will become the default. The name @samp{gnu9x} is deprecated.
1057 The 1998 ISO C++ standard plus amendments.
1060 The same as @option{-std=c++98} plus GNU extensions. This is the
1061 default for C++ code.
1064 Even when this option is not specified, you can still use some of the
1065 features of newer standards in so far as they do not conflict with
1066 previous C standards. For example, you may use @code{__restrict__} even
1067 when @option{-std=c99} is not specified.
1069 The @option{-std} options specifying some version of ISO C have the same
1070 effects as @option{-ansi}, except that features that were not in ISO C90
1071 but are in the specified version (for example, @samp{//} comments and
1072 the @code{inline} keyword in ISO C99) are not disabled.
1074 @xref{Standards,,Language Standards Supported by GCC}, for details of
1075 these standard versions.
1077 @item -aux-info @var{filename}
1079 Output to the given filename prototyped declarations for all functions
1080 declared and/or defined in a translation unit, including those in header
1081 files. This option is silently ignored in any language other than C@.
1083 Besides declarations, the file indicates, in comments, the origin of
1084 each declaration (source file and line), whether the declaration was
1085 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1086 @samp{O} for old, respectively, in the first character after the line
1087 number and the colon), and whether it came from a declaration or a
1088 definition (@samp{C} or @samp{F}, respectively, in the following
1089 character). In the case of function definitions, a K&R-style list of
1090 arguments followed by their declarations is also provided, inside
1091 comments, after the declaration.
1095 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1096 keyword, so that code can use these words as identifiers. You can use
1097 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1098 instead. @option{-ansi} implies @option{-fno-asm}.
1100 In C++, this switch only affects the @code{typeof} keyword, since
1101 @code{asm} and @code{inline} are standard keywords. You may want to
1102 use the @option{-fno-gnu-keywords} flag instead, which has the same
1103 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1104 switch only affects the @code{asm} and @code{typeof} keywords, since
1105 @code{inline} is a standard keyword in ISO C99.
1108 @itemx -fno-builtin-@var{function}
1109 @opindex fno-builtin
1110 @cindex built-in functions
1111 Don't recognize built-in functions that do not begin with
1112 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1113 functions provided by GCC}, for details of the functions affected,
1114 including those which are not built-in functions when @option{-ansi} or
1115 @option{-std} options for strict ISO C conformance are used because they
1116 do not have an ISO standard meaning.
1118 GCC normally generates special code to handle certain built-in functions
1119 more efficiently; for instance, calls to @code{alloca} may become single
1120 instructions that adjust the stack directly, and calls to @code{memcpy}
1121 may become inline copy loops. The resulting code is often both smaller
1122 and faster, but since the function calls no longer appear as such, you
1123 cannot set a breakpoint on those calls, nor can you change the behavior
1124 of the functions by linking with a different library.
1126 With the @option{-fno-builtin-@var{function}} option
1127 only the built-in function @var{function} is
1128 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1129 function is named this is not built-in in this version of GCC, this
1130 option is ignored. There is no corresponding
1131 @option{-fbuiltin-@var{function}} option; if you wish to enable
1132 built-in functions selectively when using @option{-fno-builtin} or
1133 @option{-ffreestanding}, you may define macros such as:
1136 #define abs(n) __builtin_abs ((n))
1137 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1142 @cindex hosted environment
1144 Assert that compilation takes place in a hosted environment. This implies
1145 @option{-fbuiltin}. A hosted environment is one in which the
1146 entire standard library is available, and in which @code{main} has a return
1147 type of @code{int}. Examples are nearly everything except a kernel.
1148 This is equivalent to @option{-fno-freestanding}.
1150 @item -ffreestanding
1151 @opindex ffreestanding
1152 @cindex hosted environment
1154 Assert that compilation takes place in a freestanding environment. This
1155 implies @option{-fno-builtin}. A freestanding environment
1156 is one in which the standard library may not exist, and program startup may
1157 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1158 This is equivalent to @option{-fno-hosted}.
1160 @xref{Standards,,Language Standards Supported by GCC}, for details of
1161 freestanding and hosted environments.
1163 @item -fms-extensions
1164 @opindex fms-extensions
1165 Accept some non-standard constructs used in Microsoft header files.
1169 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1170 options for strict ISO C conformance) implies @option{-trigraphs}.
1172 @item -no-integrated-cpp
1173 @opindex no-integrated-cpp
1174 Performs a compilation in two passes: preprocessing and compiling. This
1175 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1176 @option{-B} option. The user supplied compilation step can then add in
1177 an additional preprocessing step after normal preprocessing but before
1178 compiling. The default is to use the integrated cpp (internal cpp)
1180 The semantics of this option will change if "cc1", "cc1plus", and
1181 "cc1obj" are merged.
1183 @cindex traditional C language
1184 @cindex C language, traditional
1186 @itemx -traditional-cpp
1187 @opindex traditional-cpp
1188 @opindex traditional
1189 Formerly, these options caused GCC to attempt to emulate a pre-standard
1190 C compiler. They are now only supported with the @option{-E} switch.
1191 The preprocessor continues to support a pre-standard mode. See the GNU
1192 CPP manual for details.
1194 @item -fcond-mismatch
1195 @opindex fcond-mismatch
1196 Allow conditional expressions with mismatched types in the second and
1197 third arguments. The value of such an expression is void. This option
1198 is not supported for C++.
1200 @item -funsigned-char
1201 @opindex funsigned-char
1202 Let the type @code{char} be unsigned, like @code{unsigned char}.
1204 Each kind of machine has a default for what @code{char} should
1205 be. It is either like @code{unsigned char} by default or like
1206 @code{signed char} by default.
1208 Ideally, a portable program should always use @code{signed char} or
1209 @code{unsigned char} when it depends on the signedness of an object.
1210 But many programs have been written to use plain @code{char} and
1211 expect it to be signed, or expect it to be unsigned, depending on the
1212 machines they were written for. This option, and its inverse, let you
1213 make such a program work with the opposite default.
1215 The type @code{char} is always a distinct type from each of
1216 @code{signed char} or @code{unsigned char}, even though its behavior
1217 is always just like one of those two.
1220 @opindex fsigned-char
1221 Let the type @code{char} be signed, like @code{signed char}.
1223 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1224 the negative form of @option{-funsigned-char}. Likewise, the option
1225 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1227 @item -fsigned-bitfields
1228 @itemx -funsigned-bitfields
1229 @itemx -fno-signed-bitfields
1230 @itemx -fno-unsigned-bitfields
1231 @opindex fsigned-bitfields
1232 @opindex funsigned-bitfields
1233 @opindex fno-signed-bitfields
1234 @opindex fno-unsigned-bitfields
1235 These options control whether a bit-field is signed or unsigned, when the
1236 declaration does not use either @code{signed} or @code{unsigned}. By
1237 default, such a bit-field is signed, because this is consistent: the
1238 basic integer types such as @code{int} are signed types.
1240 @item -fwritable-strings
1241 @opindex fwritable-strings
1242 Store string constants in the writable data segment and don't uniquize
1243 them. This is for compatibility with old programs which assume they can
1244 write into string constants.
1246 Writing into string constants is a very bad idea; ``constants'' should
1250 @node C++ Dialect Options
1251 @section Options Controlling C++ Dialect
1253 @cindex compiler options, C++
1254 @cindex C++ options, command line
1255 @cindex options, C++
1256 This section describes the command-line options that are only meaningful
1257 for C++ programs; but you can also use most of the GNU compiler options
1258 regardless of what language your program is in. For example, you
1259 might compile a file @code{firstClass.C} like this:
1262 g++ -g -frepo -O -c firstClass.C
1266 In this example, only @option{-frepo} is an option meant
1267 only for C++ programs; you can use the other options with any
1268 language supported by GCC@.
1270 Here is a list of options that are @emph{only} for compiling C++ programs:
1274 @item -fabi-version=@var{n}
1275 @opindex fabi-version
1276 Use version @var{n} of the C++ ABI. Version 1 is the version of the C++
1277 ABI that first appeared in G++ 3.2. Version 0 will always be the
1278 version that conforms most closely to the C++ ABI specification.
1279 Therefore, the ABI obtained using version 0 will change as ABI bugs are
1282 The default is version 1.
1284 @item -fno-access-control
1285 @opindex fno-access-control
1286 Turn off all access checking. This switch is mainly useful for working
1287 around bugs in the access control code.
1291 Check that the pointer returned by @code{operator new} is non-null
1292 before attempting to modify the storage allocated. This check is
1293 normally unnecessary because the C++ standard specifies that
1294 @code{operator new} will only return @code{0} if it is declared
1295 @samp{throw()}, in which case the compiler will always check the
1296 return value even without this option. In all other cases, when
1297 @code{operator new} has a non-empty exception specification, memory
1298 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1299 @samp{new (nothrow)}.
1301 @item -fconserve-space
1302 @opindex fconserve-space
1303 Put uninitialized or runtime-initialized global variables into the
1304 common segment, as C does. This saves space in the executable at the
1305 cost of not diagnosing duplicate definitions. If you compile with this
1306 flag and your program mysteriously crashes after @code{main()} has
1307 completed, you may have an object that is being destroyed twice because
1308 two definitions were merged.
1310 This option is no longer useful on most targets, now that support has
1311 been added for putting variables into BSS without making them common.
1313 @item -fno-const-strings
1314 @opindex fno-const-strings
1315 Give string constants type @code{char *} instead of type @code{const
1316 char *}. By default, G++ uses type @code{const char *} as required by
1317 the standard. Even if you use @option{-fno-const-strings}, you cannot
1318 actually modify the value of a string constant, unless you also use
1319 @option{-fwritable-strings}.
1321 This option might be removed in a future release of G++. For maximum
1322 portability, you should structure your code so that it works with
1323 string constants that have type @code{const char *}.
1325 @item -fno-elide-constructors
1326 @opindex fno-elide-constructors
1327 The C++ standard allows an implementation to omit creating a temporary
1328 which is only used to initialize another object of the same type.
1329 Specifying this option disables that optimization, and forces G++ to
1330 call the copy constructor in all cases.
1332 @item -fno-enforce-eh-specs
1333 @opindex fno-enforce-eh-specs
1334 Don't check for violation of exception specifications at runtime. This
1335 option violates the C++ standard, but may be useful for reducing code
1336 size in production builds, much like defining @samp{NDEBUG}. The compiler
1337 will still optimize based on the exception specifications.
1339 @item -fexternal-templates
1340 @opindex fexternal-templates
1342 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1343 template instantiation; template instances are emitted or not according
1344 to the location of the template definition. @xref{Template
1345 Instantiation}, for more information.
1347 This option is deprecated.
1349 @item -falt-external-templates
1350 @opindex falt-external-templates
1351 Similar to @option{-fexternal-templates}, but template instances are
1352 emitted or not according to the place where they are first instantiated.
1353 @xref{Template Instantiation}, for more information.
1355 This option is deprecated.
1358 @itemx -fno-for-scope
1360 @opindex fno-for-scope
1361 If @option{-ffor-scope} is specified, the scope of variables declared in
1362 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1363 as specified by the C++ standard.
1364 If @option{-fno-for-scope} is specified, the scope of variables declared in
1365 a @i{for-init-statement} extends to the end of the enclosing scope,
1366 as was the case in old versions of G++, and other (traditional)
1367 implementations of C++.
1369 The default if neither flag is given to follow the standard,
1370 but to allow and give a warning for old-style code that would
1371 otherwise be invalid, or have different behavior.
1373 @item -fno-gnu-keywords
1374 @opindex fno-gnu-keywords
1375 Do not recognize @code{typeof} as a keyword, so that code can use this
1376 word as an identifier. You can use the keyword @code{__typeof__} instead.
1377 @option{-ansi} implies @option{-fno-gnu-keywords}.
1379 @item -fno-implicit-templates
1380 @opindex fno-implicit-templates
1381 Never emit code for non-inline templates which are instantiated
1382 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1383 @xref{Template Instantiation}, for more information.
1385 @item -fno-implicit-inline-templates
1386 @opindex fno-implicit-inline-templates
1387 Don't emit code for implicit instantiations of inline templates, either.
1388 The default is to handle inlines differently so that compiles with and
1389 without optimization will need the same set of explicit instantiations.
1391 @item -fno-implement-inlines
1392 @opindex fno-implement-inlines
1393 To save space, do not emit out-of-line copies of inline functions
1394 controlled by @samp{#pragma implementation}. This will cause linker
1395 errors if these functions are not inlined everywhere they are called.
1397 @item -fms-extensions
1398 @opindex fms-extensions
1399 Disable pedantic warnings about constructs used in MFC, such as implicit
1400 int and getting a pointer to member function via non-standard syntax.
1402 @item -fno-nonansi-builtins
1403 @opindex fno-nonansi-builtins
1404 Disable built-in declarations of functions that are not mandated by
1405 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1406 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1408 @item -fno-operator-names
1409 @opindex fno-operator-names
1410 Do not treat the operator name keywords @code{and}, @code{bitand},
1411 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1412 synonyms as keywords.
1414 @item -fno-optional-diags
1415 @opindex fno-optional-diags
1416 Disable diagnostics that the standard says a compiler does not need to
1417 issue. Currently, the only such diagnostic issued by G++ is the one for
1418 a name having multiple meanings within a class.
1421 @opindex fpermissive
1422 Downgrade some diagnostics about nonconformant code from errors to
1423 warnings. Thus, using @option{-fpermissive} will allow some
1424 nonconforming code to compile.
1428 Enable automatic template instantiation at link time. This option also
1429 implies @option{-fno-implicit-templates}. @xref{Template
1430 Instantiation}, for more information.
1434 Disable generation of information about every class with virtual
1435 functions for use by the C++ runtime type identification features
1436 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1437 of the language, you can save some space by using this flag. Note that
1438 exception handling uses the same information, but it will generate it as
1443 Emit statistics about front-end processing at the end of the compilation.
1444 This information is generally only useful to the G++ development team.
1446 @item -ftemplate-depth-@var{n}
1447 @opindex ftemplate-depth
1448 Set the maximum instantiation depth for template classes to @var{n}.
1449 A limit on the template instantiation depth is needed to detect
1450 endless recursions during template class instantiation. ANSI/ISO C++
1451 conforming programs must not rely on a maximum depth greater than 17.
1453 @item -fuse-cxa-atexit
1454 @opindex fuse-cxa-atexit
1455 Register destructors for objects with static storage duration with the
1456 @code{__cxa_atexit} function rather than the @code{atexit} function.
1457 This option is required for fully standards-compliant handling of static
1458 destructors, but will only work if your C library supports
1459 @code{__cxa_atexit}.
1463 Do not use weak symbol support, even if it is provided by the linker.
1464 By default, G++ will use weak symbols if they are available. This
1465 option exists only for testing, and should not be used by end-users;
1466 it will result in inferior code and has no benefits. This option may
1467 be removed in a future release of G++.
1471 Do not search for header files in the standard directories specific to
1472 C++, but do still search the other standard directories. (This option
1473 is used when building the C++ library.)
1476 In addition, these optimization, warning, and code generation options
1477 have meanings only for C++ programs:
1480 @item -fno-default-inline
1481 @opindex fno-default-inline
1482 Do not assume @samp{inline} for functions defined inside a class scope.
1483 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1484 functions will have linkage like inline functions; they just won't be
1487 @item -Wabi @r{(C++ only)}
1489 Warn when G++ generates code that is probably not compatible with the
1490 vendor-neutral C++ ABI. Although an effort has been made to warn about
1491 all such cases, there are probably some cases that are not warned about,
1492 even though G++ is generating incompatible code. There may also be
1493 cases where warnings are emitted even though the code that is generated
1496 You should rewrite your code to avoid these warnings if you are
1497 concerned about the fact that code generated by G++ may not be binary
1498 compatible with code generated by other compilers.
1500 The known incompatibilities at this point include:
1505 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1506 pack data into the same byte as a base class. For example:
1509 struct A @{ virtual void f(); int f1 : 1; @};
1510 struct B : public A @{ int f2 : 1; @};
1514 In this case, G++ will place @code{B::f2} into the same byte
1515 as@code{A::f1}; other compilers will not. You can avoid this problem
1516 by explicitly padding @code{A} so that its size is a multiple of the
1517 byte size on your platform; that will cause G++ and other compilers to
1518 layout @code{B} identically.
1521 Incorrect handling of tail-padding for virtual bases. G++ does not use
1522 tail padding when laying out virtual bases. For example:
1525 struct A @{ virtual void f(); char c1; @};
1526 struct B @{ B(); char c2; @};
1527 struct C : public A, public virtual B @{@};
1531 In this case, G++ will not place @code{B} into the tail-padding for
1532 @code{A}; other compilers will. You can avoid this problem by
1533 explicitly padding @code{A} so that its size is a multiple of its
1534 alignment (ignoring virtual base classes); that will cause G++ and other
1535 compilers to layout @code{C} identically.
1538 Incorrect handling of bit-fields with declared widths greater than that
1539 of their underlying types, when the bit-fields appear in a union. For
1543 union U @{ int i : 4096; @};
1547 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1548 union too small by the number of bits in an @code{int}.
1551 Empty classes can be placed at incorrect offsets. For example:
1561 struct C : public B, public A @{@};
1565 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1566 it should be placed at offset zero. G++ mistakenly believes that the
1567 @code{A} data member of @code{B} is already at offset zero.
1570 Names of template functions whose types involve @code{typename} or
1571 template template parameters can be mangled incorrectly.
1574 template <typename Q>
1575 void f(typename Q::X) @{@}
1577 template <template <typename> class Q>
1578 void f(typename Q<int>::X) @{@}
1582 Instantiations of these templates may be mangled incorrectly.
1586 @item -Wctor-dtor-privacy @r{(C++ only)}
1587 @opindex Wctor-dtor-privacy
1588 Warn when a class seems unusable because all the constructors or
1589 destructors in that class are private, and it has neither friends nor
1590 public static member functions.
1592 @item -Wnon-virtual-dtor @r{(C++ only)}
1593 @opindex Wnon-virtual-dtor
1594 Warn when a class appears to be polymorphic, thereby requiring a virtual
1595 destructor, yet it declares a non-virtual one.
1596 This warning is enabled by @option{-Wall}.
1598 @item -Wreorder @r{(C++ only)}
1600 @cindex reordering, warning
1601 @cindex warning for reordering of member initializers
1602 Warn when the order of member initializers given in the code does not
1603 match the order in which they must be executed. For instance:
1609 A(): j (0), i (1) @{ @}
1613 The compiler will rearrange the member initializers for @samp{i}
1614 and @samp{j} to match the declaration order of the members, emitting
1615 a warning to that effect. This warning is enabled by @option{-Wall}.
1618 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1621 @item -Weffc++ @r{(C++ only)}
1623 Warn about violations of the following style guidelines from Scott Meyers'
1624 @cite{Effective C++} book:
1628 Item 11: Define a copy constructor and an assignment operator for classes
1629 with dynamically allocated memory.
1632 Item 12: Prefer initialization to assignment in constructors.
1635 Item 14: Make destructors virtual in base classes.
1638 Item 15: Have @code{operator=} return a reference to @code{*this}.
1641 Item 23: Don't try to return a reference when you must return an object.
1645 Also warn about violations of the following style guidelines from
1646 Scott Meyers' @cite{More Effective C++} book:
1650 Item 6: Distinguish between prefix and postfix forms of increment and
1651 decrement operators.
1654 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1658 When selecting this option, be aware that the standard library
1659 headers do not obey all of these guidelines; use @samp{grep -v}
1660 to filter out those warnings.
1662 @item -Wno-deprecated @r{(C++ only)}
1663 @opindex Wno-deprecated
1664 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1666 @item -Wno-non-template-friend @r{(C++ only)}
1667 @opindex Wno-non-template-friend
1668 Disable warnings when non-templatized friend functions are declared
1669 within a template. Since the advent of explicit template specification
1670 support in G++, if the name of the friend is an unqualified-id (i.e.,
1671 @samp{friend foo(int)}), the C++ language specification demands that the
1672 friend declare or define an ordinary, nontemplate function. (Section
1673 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1674 could be interpreted as a particular specialization of a templatized
1675 function. Because this non-conforming behavior is no longer the default
1676 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1677 check existing code for potential trouble spots and is on by default.
1678 This new compiler behavior can be turned off with
1679 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1680 but disables the helpful warning.
1682 @item -Wold-style-cast @r{(C++ only)}
1683 @opindex Wold-style-cast
1684 Warn if an old-style (C-style) cast to a non-void type is used within
1685 a C++ program. The new-style casts (@samp{static_cast},
1686 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1687 unintended effects and much easier to search for.
1689 @item -Woverloaded-virtual @r{(C++ only)}
1690 @opindex Woverloaded-virtual
1691 @cindex overloaded virtual fn, warning
1692 @cindex warning for overloaded virtual fn
1693 Warn when a function declaration hides virtual functions from a
1694 base class. For example, in:
1701 struct B: public A @{
1706 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1714 will fail to compile.
1716 @item -Wno-pmf-conversions @r{(C++ only)}
1717 @opindex Wno-pmf-conversions
1718 Disable the diagnostic for converting a bound pointer to member function
1721 @item -Wsign-promo @r{(C++ only)}
1722 @opindex Wsign-promo
1723 Warn when overload resolution chooses a promotion from unsigned or
1724 enumeral type to a signed type, over a conversion to an unsigned type of
1725 the same size. Previous versions of G++ would try to preserve
1726 unsignedness, but the standard mandates the current behavior.
1728 @item -Wsynth @r{(C++ only)}
1730 @cindex warning for synthesized methods
1731 @cindex synthesized methods, warning
1732 Warn when G++'s synthesis behavior does not match that of cfront. For
1738 A& operator = (int);
1748 In this example, G++ will synthesize a default @samp{A& operator =
1749 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1752 @node Objective-C Dialect Options
1753 @section Options Controlling Objective-C Dialect
1755 @cindex compiler options, Objective-C
1756 @cindex Objective-C options, command line
1757 @cindex options, Objective-C
1758 (NOTE: This manual does not describe the Objective-C language itself. See
1759 @w{@uref{http://gcc.gnu.org/readings.html}} for references.)
1761 This section describes the command-line options that are only meaningful
1762 for Objective-C programs, but you can also use most of the GNU compiler
1763 options regardless of what language your program is in. For example,
1764 you might compile a file @code{some_class.m} like this:
1767 gcc -g -fgnu-runtime -O -c some_class.m
1771 In this example, @option{-fgnu-runtime} is an option meant only for
1772 Objective-C programs; you can use the other options with any language
1775 Here is a list of options that are @emph{only} for compiling Objective-C
1779 @item -fconstant-string-class=@var{class-name}
1780 @opindex fconstant-string-class
1781 Use @var{class-name} as the name of the class to instantiate for each
1782 literal string specified with the syntax @code{@@"@dots{}"}. The default
1783 class name is @code{NXConstantString} if the GNU runtime is being used, and
1784 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1785 @option{-fconstant-cfstrings} option, if also present, will override the
1786 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1787 to be laid out as constant CoreFoundation strings.
1790 @opindex fgnu-runtime
1791 Generate object code compatible with the standard GNU Objective-C
1792 runtime. This is the default for most types of systems.
1794 @item -fnext-runtime
1795 @opindex fnext-runtime
1796 Generate output compatible with the NeXT runtime. This is the default
1797 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1798 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1801 @item -fno-nil-receivers
1802 @opindex -fno-nil-receivers
1803 Assume that all Objective-C message dispatches (e.g.,
1804 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1805 is not @code{nil}. This allows for more efficient entry points in the runtime to be
1806 used. Currently, this option is only available in conjunction with
1807 the NeXT runtime on Mac OS X 10.3 and later.
1809 @item -fobjc-exceptions
1810 @opindex -fobjc-exceptions
1811 Enable syntactic support for structured exception handling in Objective-C,
1812 similar to what is offered by C++ and Java. Currently, this option is only
1813 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1821 @@catch (AnObjCClass *exc) @{
1828 @@catch (AnotherClass *exc) @{
1831 @@catch (id allOthers) @{
1841 The @code{@@throw} statement may appear anywhere in an Objective-C or
1842 Objective-C++ program; when used inside of a @code{@@catch} block, the
1843 @code{@@throw} may appear without an argument (as shown above), in which case
1844 the object caught by the @code{@@catch} will be rethrown.
1846 Note that only (pointers to) Objective-C objects may be thrown and
1847 caught using this scheme. When an object is thrown, it will be caught
1848 by the nearest @code{@@catch} clause capable of handling objects of that type,
1849 analogously to how @code{catch} blocks work in C++ and Java. A
1850 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1851 any and all Objective-C exceptions not caught by previous @code{@@catch}
1854 The @code{@@finally} clause, if present, will be executed upon exit from the
1855 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1856 regardless of whether any exceptions are thrown, caught or rethrown
1857 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1858 of the @code{finally} clause in Java.
1860 There are several caveats to using the new exception mechanism:
1864 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1865 idioms provided by the @code{NSException} class, the new
1866 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1867 systems, due to additional functionality needed in the (NeXT) Objective-C
1871 As mentioned above, the new exceptions do not support handling
1872 types other than Objective-C objects. Furthermore, when used from
1873 Objective-C++, the Objective-C exception model does not interoperate with C++
1874 exceptions at this time. This means you cannot @code{@@throw} an exception
1875 from Objective-C and @code{catch} it in C++, or vice versa
1876 (i.e., @code{throw @dots{} @@catch}).
1879 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1880 blocks for thread-safe execution:
1883 @@synchronized (ObjCClass *guard) @{
1888 Upon entering the @code{@@synchronized} block, a thread of execution shall
1889 first check whether a lock has been placed on the corresponding @code{guard}
1890 object by another thread. If it has, the current thread shall wait until
1891 the other thread relinquishes its lock. Once @code{guard} becomes available,
1892 the current thread will place its own lock on it, execute the code contained in
1893 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1894 making @code{guard} available to other threads).
1896 Unlike Java, Objective-C does not allow for entire methods to be marked
1897 @code{@@synchronized}. Note that throwing exceptions out of
1898 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1899 to be unlocked properly.
1901 @item -freplace-objc-classes
1902 @opindex -freplace-objc-classes
1903 Emit a special marker instructing @command{ld(1)} not to statically link in
1904 the resulting object file, and allow @command{dyld(1)} to load it in at
1905 run time instead. This is used in conjunction with the Fix-and-Continue
1906 debugging mode, where the object file in question may be recompiled and
1907 dynamically reloaded in the course of program execution, without the need
1908 to restart the program itself. Currently, Fix-and-Continue functionality
1909 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
1913 @opindex -fzero-link
1914 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1915 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1916 compile time) with static class references that get initialized at load time,
1917 which improves run-time performance. Specifying the @option{-fzero-link} flag
1918 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1919 to be retained. This is useful in Zero-Link debugging mode, since it allows
1920 for individual class implementations to be modified during program execution.
1924 Dump interface declarations for all classes seen in the source file to a
1925 file named @file{@var{sourcename}.decl}.
1928 @opindex Wno-protocol
1929 If a class is declared to implement a protocol, a warning is issued for
1930 every method in the protocol that is not implemented by the class. The
1931 default behavior is to issue a warning for every method not explicitly
1932 implemented in the class, even if a method implementation is inherited
1933 from the superclass. If you use the @code{-Wno-protocol} option, then
1934 methods inherited from the superclass are considered to be implemented,
1935 and no warning is issued for them.
1939 Warn if multiple methods of different types for the same selector are
1940 found during compilation. The check is performed on the list of methods
1941 in the final stage of compilation. Additionally, a check is performed
1942 for each selector appearing in a @code{@@selector(@dots{})}
1943 expression, and a corresponding method for that selector has been found
1944 during compilation. Because these checks scan the method table only at
1945 the end of compilation, these warnings are not produced if the final
1946 stage of compilation is not reached, for example because an error is
1947 found during compilation, or because the @code{-fsyntax-only} option is
1950 @item -Wundeclared-selector
1951 @opindex Wundeclared-selector
1952 Warn if a @code{@@selector(@dots{})} expression referring to an
1953 undeclared selector is found. A selector is considered undeclared if no
1954 method with that name has been declared before the
1955 @code{@@selector(@dots{})} expression, either explicitly in an
1956 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
1957 an @code{@@implementation} section. This option always performs its
1958 checks as soon as a @code{@@selector(@dots{})} expression is found,
1959 while @code{-Wselector} only performs its checks in the final stage of
1960 compilation. This also enforces the coding style convention
1961 that methods and selectors must be declared before being used.
1963 @item -print-objc-runtime-info
1964 @opindex -print-objc-runtime-info
1965 Generate C header describing the largest structure that is passed by
1970 @node Language Independent Options
1971 @section Options to Control Diagnostic Messages Formatting
1972 @cindex options to control diagnostics formatting
1973 @cindex diagnostic messages
1974 @cindex message formatting
1976 Traditionally, diagnostic messages have been formatted irrespective of
1977 the output device's aspect (e.g.@: its width, @dots{}). The options described
1978 below can be used to control the diagnostic messages formatting
1979 algorithm, e.g.@: how many characters per line, how often source location
1980 information should be reported. Right now, only the C++ front end can
1981 honor these options. However it is expected, in the near future, that
1982 the remaining front ends would be able to digest them correctly.
1985 @item -fmessage-length=@var{n}
1986 @opindex fmessage-length
1987 Try to format error messages so that they fit on lines of about @var{n}
1988 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1989 the front ends supported by GCC@. If @var{n} is zero, then no
1990 line-wrapping will be done; each error message will appear on a single
1993 @opindex fdiagnostics-show-location
1994 @item -fdiagnostics-show-location=once
1995 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1996 reporter to emit @emph{once} source location information; that is, in
1997 case the message is too long to fit on a single physical line and has to
1998 be wrapped, the source location won't be emitted (as prefix) again,
1999 over and over, in subsequent continuation lines. This is the default
2002 @item -fdiagnostics-show-location=every-line
2003 Only meaningful in line-wrapping mode. Instructs the diagnostic
2004 messages reporter to emit the same source location information (as
2005 prefix) for physical lines that result from the process of breaking
2006 a message which is too long to fit on a single line.
2010 @node Warning Options
2011 @section Options to Request or Suppress Warnings
2012 @cindex options to control warnings
2013 @cindex warning messages
2014 @cindex messages, warning
2015 @cindex suppressing warnings
2017 Warnings are diagnostic messages that report constructions which
2018 are not inherently erroneous but which are risky or suggest there
2019 may have been an error.
2021 You can request many specific warnings with options beginning @samp{-W},
2022 for example @option{-Wimplicit} to request warnings on implicit
2023 declarations. Each of these specific warning options also has a
2024 negative form beginning @samp{-Wno-} to turn off warnings;
2025 for example, @option{-Wno-implicit}. This manual lists only one of the
2026 two forms, whichever is not the default.
2028 The following options control the amount and kinds of warnings produced
2029 by GCC; for further, language-specific options also refer to
2030 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
2033 @cindex syntax checking
2035 @opindex fsyntax-only
2036 Check the code for syntax errors, but don't do anything beyond that.
2040 Issue all the warnings demanded by strict ISO C and ISO C++;
2041 reject all programs that use forbidden extensions, and some other
2042 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2043 version of the ISO C standard specified by any @option{-std} option used.
2045 Valid ISO C and ISO C++ programs should compile properly with or without
2046 this option (though a rare few will require @option{-ansi} or a
2047 @option{-std} option specifying the required version of ISO C)@. However,
2048 without this option, certain GNU extensions and traditional C and C++
2049 features are supported as well. With this option, they are rejected.
2051 @option{-pedantic} does not cause warning messages for use of the
2052 alternate keywords whose names begin and end with @samp{__}. Pedantic
2053 warnings are also disabled in the expression that follows
2054 @code{__extension__}. However, only system header files should use
2055 these escape routes; application programs should avoid them.
2056 @xref{Alternate Keywords}.
2058 Some users try to use @option{-pedantic} to check programs for strict ISO
2059 C conformance. They soon find that it does not do quite what they want:
2060 it finds some non-ISO practices, but not all---only those for which
2061 ISO C @emph{requires} a diagnostic, and some others for which
2062 diagnostics have been added.
2064 A feature to report any failure to conform to ISO C might be useful in
2065 some instances, but would require considerable additional work and would
2066 be quite different from @option{-pedantic}. We don't have plans to
2067 support such a feature in the near future.
2069 Where the standard specified with @option{-std} represents a GNU
2070 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2071 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2072 extended dialect is based. Warnings from @option{-pedantic} are given
2073 where they are required by the base standard. (It would not make sense
2074 for such warnings to be given only for features not in the specified GNU
2075 C dialect, since by definition the GNU dialects of C include all
2076 features the compiler supports with the given option, and there would be
2077 nothing to warn about.)
2079 @item -pedantic-errors
2080 @opindex pedantic-errors
2081 Like @option{-pedantic}, except that errors are produced rather than
2086 Inhibit all warning messages.
2090 Inhibit warning messages about the use of @samp{#import}.
2092 @item -Wchar-subscripts
2093 @opindex Wchar-subscripts
2094 Warn if an array subscript has type @code{char}. This is a common cause
2095 of error, as programmers often forget that this type is signed on some
2100 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2101 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2105 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2106 the arguments supplied have types appropriate to the format string
2107 specified, and that the conversions specified in the format string make
2108 sense. This includes standard functions, and others specified by format
2109 attributes (@pxref{Function Attributes}), in the @code{printf},
2110 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2111 not in the C standard) families.
2113 The formats are checked against the format features supported by GNU
2114 libc version 2.2. These include all ISO C90 and C99 features, as well
2115 as features from the Single Unix Specification and some BSD and GNU
2116 extensions. Other library implementations may not support all these
2117 features; GCC does not support warning about features that go beyond a
2118 particular library's limitations. However, if @option{-pedantic} is used
2119 with @option{-Wformat}, warnings will be given about format features not
2120 in the selected standard version (but not for @code{strfmon} formats,
2121 since those are not in any version of the C standard). @xref{C Dialect
2122 Options,,Options Controlling C Dialect}.
2124 Since @option{-Wformat} also checks for null format arguments for
2125 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2127 @option{-Wformat} is included in @option{-Wall}. For more control over some
2128 aspects of format checking, the options @option{-Wformat-y2k},
2129 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2130 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2131 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2134 @opindex Wformat-y2k
2135 If @option{-Wformat} is specified, also warn about @code{strftime}
2136 formats which may yield only a two-digit year.
2138 @item -Wno-format-extra-args
2139 @opindex Wno-format-extra-args
2140 If @option{-Wformat} is specified, do not warn about excess arguments to a
2141 @code{printf} or @code{scanf} format function. The C standard specifies
2142 that such arguments are ignored.
2144 Where the unused arguments lie between used arguments that are
2145 specified with @samp{$} operand number specifications, normally
2146 warnings are still given, since the implementation could not know what
2147 type to pass to @code{va_arg} to skip the unused arguments. However,
2148 in the case of @code{scanf} formats, this option will suppress the
2149 warning if the unused arguments are all pointers, since the Single
2150 Unix Specification says that such unused arguments are allowed.
2152 @item -Wno-format-zero-length
2153 @opindex Wno-format-zero-length
2154 If @option{-Wformat} is specified, do not warn about zero-length formats.
2155 The C standard specifies that zero-length formats are allowed.
2157 @item -Wformat-nonliteral
2158 @opindex Wformat-nonliteral
2159 If @option{-Wformat} is specified, also warn if the format string is not a
2160 string literal and so cannot be checked, unless the format function
2161 takes its format arguments as a @code{va_list}.
2163 @item -Wformat-security
2164 @opindex Wformat-security
2165 If @option{-Wformat} is specified, also warn about uses of format
2166 functions that represent possible security problems. At present, this
2167 warns about calls to @code{printf} and @code{scanf} functions where the
2168 format string is not a string literal and there are no format arguments,
2169 as in @code{printf (foo);}. This may be a security hole if the format
2170 string came from untrusted input and contains @samp{%n}. (This is
2171 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2172 in future warnings may be added to @option{-Wformat-security} that are not
2173 included in @option{-Wformat-nonliteral}.)
2177 Enable @option{-Wformat} plus format checks not included in
2178 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2179 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2183 Warn about passing a null pointer for arguments marked as
2184 requiring a non-null value by the @code{nonnull} function attribute.
2186 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2187 can be disabled with the @option{-Wno-nonnull} option.
2189 @item -Winit-self @r{(C, C++, and Objective-C only)}
2191 Warn about uninitialized variables which are initialized with themselves.
2192 Note this option can only be used with the @option{-Wuninitialized} option,
2193 which in turn only works with @option{-O1} and above.
2195 For example, GCC will warn about @code{i} being uninitialized in the
2196 following snippet only when @option{-Winit-self} has been specified:
2207 @item -Wimplicit-int
2208 @opindex Wimplicit-int
2209 Warn when a declaration does not specify a type.
2211 @item -Wimplicit-function-declaration
2212 @itemx -Werror-implicit-function-declaration
2213 @opindex Wimplicit-function-declaration
2214 @opindex Werror-implicit-function-declaration
2215 Give a warning (or error) whenever a function is used before being
2220 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2224 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2225 function with external linkage, returning int, taking either zero
2226 arguments, two, or three arguments of appropriate types.
2228 @item -Wmissing-braces
2229 @opindex Wmissing-braces
2230 Warn if an aggregate or union initializer is not fully bracketed. In
2231 the following example, the initializer for @samp{a} is not fully
2232 bracketed, but that for @samp{b} is fully bracketed.
2235 int a[2][2] = @{ 0, 1, 2, 3 @};
2236 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2240 @opindex Wparentheses
2241 Warn if parentheses are omitted in certain contexts, such
2242 as when there is an assignment in a context where a truth value
2243 is expected, or when operators are nested whose precedence people
2244 often get confused about.
2246 Also warn about constructions where there may be confusion to which
2247 @code{if} statement an @code{else} branch belongs. Here is an example of
2262 In C, every @code{else} branch belongs to the innermost possible @code{if}
2263 statement, which in this example is @code{if (b)}. This is often not
2264 what the programmer expected, as illustrated in the above example by
2265 indentation the programmer chose. When there is the potential for this
2266 confusion, GCC will issue a warning when this flag is specified.
2267 To eliminate the warning, add explicit braces around the innermost
2268 @code{if} statement so there is no way the @code{else} could belong to
2269 the enclosing @code{if}. The resulting code would look like this:
2285 @item -Wsequence-point
2286 @opindex Wsequence-point
2287 Warn about code that may have undefined semantics because of violations
2288 of sequence point rules in the C standard.
2290 The C standard defines the order in which expressions in a C program are
2291 evaluated in terms of @dfn{sequence points}, which represent a partial
2292 ordering between the execution of parts of the program: those executed
2293 before the sequence point, and those executed after it. These occur
2294 after the evaluation of a full expression (one which is not part of a
2295 larger expression), after the evaluation of the first operand of a
2296 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2297 function is called (but after the evaluation of its arguments and the
2298 expression denoting the called function), and in certain other places.
2299 Other than as expressed by the sequence point rules, the order of
2300 evaluation of subexpressions of an expression is not specified. All
2301 these rules describe only a partial order rather than a total order,
2302 since, for example, if two functions are called within one expression
2303 with no sequence point between them, the order in which the functions
2304 are called is not specified. However, the standards committee have
2305 ruled that function calls do not overlap.
2307 It is not specified when between sequence points modifications to the
2308 values of objects take effect. Programs whose behavior depends on this
2309 have undefined behavior; the C standard specifies that ``Between the
2310 previous and next sequence point an object shall have its stored value
2311 modified at most once by the evaluation of an expression. Furthermore,
2312 the prior value shall be read only to determine the value to be
2313 stored.''. If a program breaks these rules, the results on any
2314 particular implementation are entirely unpredictable.
2316 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2317 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2318 diagnosed by this option, and it may give an occasional false positive
2319 result, but in general it has been found fairly effective at detecting
2320 this sort of problem in programs.
2322 The present implementation of this option only works for C programs. A
2323 future implementation may also work for C++ programs.
2325 The C standard is worded confusingly, therefore there is some debate
2326 over the precise meaning of the sequence point rules in subtle cases.
2327 Links to discussions of the problem, including proposed formal
2328 definitions, may be found on our readings page, at
2329 @w{@uref{http://gcc.gnu.org/readings.html}}.
2332 @opindex Wreturn-type
2333 Warn whenever a function is defined with a return-type that defaults to
2334 @code{int}. Also warn about any @code{return} statement with no
2335 return-value in a function whose return-type is not @code{void}.
2337 For C++, a function without return type always produces a diagnostic
2338 message, even when @option{-Wno-return-type} is specified. The only
2339 exceptions are @samp{main} and functions defined in system headers.
2343 Warn whenever a @code{switch} statement has an index of enumeral type
2344 and lacks a @code{case} for one or more of the named codes of that
2345 enumeration. (The presence of a @code{default} label prevents this
2346 warning.) @code{case} labels outside the enumeration range also
2347 provoke warnings when this option is used.
2349 @item -Wswitch-default
2350 @opindex Wswitch-switch
2351 Warn whenever a @code{switch} statement does not have a @code{default}
2355 @opindex Wswitch-enum
2356 Warn whenever a @code{switch} statement has an index of enumeral type
2357 and lacks a @code{case} for one or more of the named codes of that
2358 enumeration. @code{case} labels outside the enumeration range also
2359 provoke warnings when this option is used.
2363 Warn if any trigraphs are encountered that might change the meaning of
2364 the program (trigraphs within comments are not warned about).
2366 @item -Wunused-function
2367 @opindex Wunused-function
2368 Warn whenever a static function is declared but not defined or a
2369 non\-inline static function is unused.
2371 @item -Wunused-label
2372 @opindex Wunused-label
2373 Warn whenever a label is declared but not used.
2375 To suppress this warning use the @samp{unused} attribute
2376 (@pxref{Variable Attributes}).
2378 @item -Wunused-parameter
2379 @opindex Wunused-parameter
2380 Warn whenever a function parameter is unused aside from its declaration.
2382 To suppress this warning use the @samp{unused} attribute
2383 (@pxref{Variable Attributes}).
2385 @item -Wunused-variable
2386 @opindex Wunused-variable
2387 Warn whenever a local variable or non-constant static variable is unused
2388 aside from its declaration
2390 To suppress this warning use the @samp{unused} attribute
2391 (@pxref{Variable Attributes}).
2393 @item -Wunused-value
2394 @opindex Wunused-value
2395 Warn whenever a statement computes a result that is explicitly not used.
2397 To suppress this warning cast the expression to @samp{void}.
2401 All the above @option{-Wunused} options combined.
2403 In order to get a warning about an unused function parameter, you must
2404 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2405 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2407 @item -Wuninitialized
2408 @opindex Wuninitialized
2409 Warn if an automatic variable is used without first being initialized or
2410 if a variable may be clobbered by a @code{setjmp} call.
2412 These warnings are possible only in optimizing compilation,
2413 because they require data flow information that is computed only
2414 when optimizing. If you don't specify @option{-O}, you simply won't
2417 If you want to warn about code which uses the uninitialized value of the
2418 variable in its own initializer, use the @option{-Winit-self} option.
2420 These warnings occur only for variables that are candidates for
2421 register allocation. Therefore, they do not occur for a variable that
2422 is declared @code{volatile}, or whose address is taken, or whose size
2423 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2424 structures, unions or arrays, even when they are in registers.
2426 Note that there may be no warning about a variable that is used only
2427 to compute a value that itself is never used, because such
2428 computations may be deleted by data flow analysis before the warnings
2431 These warnings are made optional because GCC is not smart
2432 enough to see all the reasons why the code might be correct
2433 despite appearing to have an error. Here is one example of how
2454 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2455 always initialized, but GCC doesn't know this. Here is
2456 another common case:
2461 if (change_y) save_y = y, y = new_y;
2463 if (change_y) y = save_y;
2468 This has no bug because @code{save_y} is used only if it is set.
2470 @cindex @code{longjmp} warnings
2471 This option also warns when a non-volatile automatic variable might be
2472 changed by a call to @code{longjmp}. These warnings as well are possible
2473 only in optimizing compilation.
2475 The compiler sees only the calls to @code{setjmp}. It cannot know
2476 where @code{longjmp} will be called; in fact, a signal handler could
2477 call it at any point in the code. As a result, you may get a warning
2478 even when there is in fact no problem because @code{longjmp} cannot
2479 in fact be called at the place which would cause a problem.
2481 Some spurious warnings can be avoided if you declare all the functions
2482 you use that never return as @code{noreturn}. @xref{Function
2485 @item -Wunknown-pragmas
2486 @opindex Wunknown-pragmas
2487 @cindex warning for unknown pragmas
2488 @cindex unknown pragmas, warning
2489 @cindex pragmas, warning of unknown
2490 Warn when a #pragma directive is encountered which is not understood by
2491 GCC@. If this command line option is used, warnings will even be issued
2492 for unknown pragmas in system header files. This is not the case if
2493 the warnings were only enabled by the @option{-Wall} command line option.
2495 @item -Wstrict-aliasing
2496 @opindex Wstrict-aliasing
2497 This option is only active when @option{-fstrict-aliasing} is active.
2498 It warns about code which might break the strict aliasing rules that the
2499 compiler is using for optimization. The warning does not catch all
2500 cases, but does attempt to catch the more common pitfalls. It is
2501 included in @option{-Wall}.
2505 All of the above @samp{-W} options combined. This enables all the
2506 warnings about constructions that some users consider questionable, and
2507 that are easy to avoid (or modify to prevent the warning), even in
2508 conjunction with macros. This also enables some language-specific
2509 warnings described in @ref{C++ Dialect Options} and
2510 @ref{Objective-C Dialect Options}.
2513 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2514 Some of them warn about constructions that users generally do not
2515 consider questionable, but which occasionally you might wish to check
2516 for; others warn about constructions that are necessary or hard to avoid
2517 in some cases, and there is no simple way to modify the code to suppress
2524 (This option used to be called @option{-W}. The older name is still
2525 supported, but the newer name is more descriptive.) Print extra warning
2526 messages for these events:
2530 A function can return either with or without a value. (Falling
2531 off the end of the function body is considered returning without
2532 a value.) For example, this function would evoke such a
2546 An expression-statement or the left-hand side of a comma expression
2547 contains no side effects.
2548 To suppress the warning, cast the unused expression to void.
2549 For example, an expression such as @samp{x[i,j]} will cause a warning,
2550 but @samp{x[(void)i,j]} will not.
2553 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2556 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2557 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2558 that of ordinary mathematical notation.
2561 Storage-class specifiers like @code{static} are not the first things in
2562 a declaration. According to the C Standard, this usage is obsolescent.
2565 The return type of a function has a type qualifier such as @code{const}.
2566 Such a type qualifier has no effect, since the value returned by a
2567 function is not an lvalue. (But don't warn about the GNU extension of
2568 @code{volatile void} return types. That extension will be warned about
2569 if @option{-pedantic} is specified.)
2572 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2576 A comparison between signed and unsigned values could produce an
2577 incorrect result when the signed value is converted to unsigned.
2578 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2581 An aggregate has an initializer which does not initialize all members.
2582 For example, the following code would cause such a warning, because
2583 @code{x.h} would be implicitly initialized to zero:
2586 struct s @{ int f, g, h; @};
2587 struct s x = @{ 3, 4 @};
2591 A function parameter is declared without a type specifier in K&R-style
2599 An empty body occurs in an @samp{if} or @samp{else} statement.
2602 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2603 @samp{>}, or @samp{>=}.
2606 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2609 Any of several floating-point events that often indicate errors, such as
2610 overflow, underflow, loss of precision, etc.
2612 @item @r{(C++ only)}
2613 An enumerator and a non-enumerator both appear in a conditional expression.
2615 @item @r{(C++ only)}
2616 A non-static reference or non-static @samp{const} member appears in a
2617 class without constructors.
2619 @item @r{(C++ only)}
2620 Ambiguous virtual bases.
2622 @item @r{(C++ only)}
2623 Subscripting an array which has been declared @samp{register}.
2625 @item @r{(C++ only)}
2626 Taking the address of a variable which has been declared @samp{register}.
2628 @item @r{(C++ only)}
2629 A base class is not initialized in a derived class' copy constructor.
2632 @item -Wno-div-by-zero
2633 @opindex Wno-div-by-zero
2634 @opindex Wdiv-by-zero
2635 Do not warn about compile-time integer division by zero. Floating point
2636 division by zero is not warned about, as it can be a legitimate way of
2637 obtaining infinities and NaNs.
2639 @item -Wsystem-headers
2640 @opindex Wsystem-headers
2641 @cindex warnings from system headers
2642 @cindex system headers, warnings from
2643 Print warning messages for constructs found in system header files.
2644 Warnings from system headers are normally suppressed, on the assumption
2645 that they usually do not indicate real problems and would only make the
2646 compiler output harder to read. Using this command line option tells
2647 GCC to emit warnings from system headers as if they occurred in user
2648 code. However, note that using @option{-Wall} in conjunction with this
2649 option will @emph{not} warn about unknown pragmas in system
2650 headers---for that, @option{-Wunknown-pragmas} must also be used.
2653 @opindex Wfloat-equal
2654 Warn if floating point values are used in equality comparisons.
2656 The idea behind this is that sometimes it is convenient (for the
2657 programmer) to consider floating-point values as approximations to
2658 infinitely precise real numbers. If you are doing this, then you need
2659 to compute (by analyzing the code, or in some other way) the maximum or
2660 likely maximum error that the computation introduces, and allow for it
2661 when performing comparisons (and when producing output, but that's a
2662 different problem). In particular, instead of testing for equality, you
2663 would check to see whether the two values have ranges that overlap; and
2664 this is done with the relational operators, so equality comparisons are
2667 @item -Wtraditional @r{(C only)}
2668 @opindex Wtraditional
2669 Warn about certain constructs that behave differently in traditional and
2670 ISO C@. Also warn about ISO C constructs that have no traditional C
2671 equivalent, and/or problematic constructs which should be avoided.
2675 Macro parameters that appear within string literals in the macro body.
2676 In traditional C macro replacement takes place within string literals,
2677 but does not in ISO C@.
2680 In traditional C, some preprocessor directives did not exist.
2681 Traditional preprocessors would only consider a line to be a directive
2682 if the @samp{#} appeared in column 1 on the line. Therefore
2683 @option{-Wtraditional} warns about directives that traditional C
2684 understands but would ignore because the @samp{#} does not appear as the
2685 first character on the line. It also suggests you hide directives like
2686 @samp{#pragma} not understood by traditional C by indenting them. Some
2687 traditional implementations would not recognize @samp{#elif}, so it
2688 suggests avoiding it altogether.
2691 A function-like macro that appears without arguments.
2694 The unary plus operator.
2697 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2698 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2699 constants.) Note, these suffixes appear in macros defined in the system
2700 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2701 Use of these macros in user code might normally lead to spurious
2702 warnings, however gcc's integrated preprocessor has enough context to
2703 avoid warning in these cases.
2706 A function declared external in one block and then used after the end of
2710 A @code{switch} statement has an operand of type @code{long}.
2713 A non-@code{static} function declaration follows a @code{static} one.
2714 This construct is not accepted by some traditional C compilers.
2717 The ISO type of an integer constant has a different width or
2718 signedness from its traditional type. This warning is only issued if
2719 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2720 typically represent bit patterns, are not warned about.
2723 Usage of ISO string concatenation is detected.
2726 Initialization of automatic aggregates.
2729 Identifier conflicts with labels. Traditional C lacks a separate
2730 namespace for labels.
2733 Initialization of unions. If the initializer is zero, the warning is
2734 omitted. This is done under the assumption that the zero initializer in
2735 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2736 initializer warnings and relies on default initialization to zero in the
2740 Conversions by prototypes between fixed/floating point values and vice
2741 versa. The absence of these prototypes when compiling with traditional
2742 C would cause serious problems. This is a subset of the possible
2743 conversion warnings, for the full set use @option{-Wconversion}.
2746 Use of ISO C style function definitions. This warning intentionally is
2747 @emph{not} issued for prototype declarations or variadic functions
2748 because these ISO C features will appear in your code when using
2749 libiberty's traditional C compatibility macros, @code{PARAMS} and
2750 @code{VPARAMS}. This warning is also bypassed for nested functions
2751 because that feature is already a gcc extension and thus not relevant to
2752 traditional C compatibility.
2755 @item -Wdeclaration-after-statement @r{(C only)}
2756 @opindex Wdeclaration-after-statement
2757 Warn when a declaration is found after a statement in a block. This
2758 construct, known from C++, was introduced with ISO C99 and is by default
2759 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2760 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2764 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2766 @item -Wendif-labels
2767 @opindex Wendif-labels
2768 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2772 Warn whenever a local variable shadows another local variable, parameter or
2773 global variable or whenever a built-in function is shadowed.
2775 @item -Wlarger-than-@var{len}
2776 @opindex Wlarger-than
2777 Warn whenever an object of larger than @var{len} bytes is defined.
2779 @item -Wpointer-arith
2780 @opindex Wpointer-arith
2781 Warn about anything that depends on the ``size of'' a function type or
2782 of @code{void}. GNU C assigns these types a size of 1, for
2783 convenience in calculations with @code{void *} pointers and pointers
2786 @item -Wbad-function-cast @r{(C only)}
2787 @opindex Wbad-function-cast
2788 Warn whenever a function call is cast to a non-matching type.
2789 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2793 Warn whenever a pointer is cast so as to remove a type qualifier from
2794 the target type. For example, warn if a @code{const char *} is cast
2795 to an ordinary @code{char *}.
2798 @opindex Wcast-align
2799 Warn whenever a pointer is cast such that the required alignment of the
2800 target is increased. For example, warn if a @code{char *} is cast to
2801 an @code{int *} on machines where integers can only be accessed at
2802 two- or four-byte boundaries.
2804 @item -Wwrite-strings
2805 @opindex Wwrite-strings
2806 When compiling C, give string constants the type @code{const
2807 char[@var{length}]} so that
2808 copying the address of one into a non-@code{const} @code{char *}
2809 pointer will get a warning; when compiling C++, warn about the
2810 deprecated conversion from string constants to @code{char *}.
2811 These warnings will help you find at
2812 compile time code that can try to write into a string constant, but
2813 only if you have been very careful about using @code{const} in
2814 declarations and prototypes. Otherwise, it will just be a nuisance;
2815 this is why we did not make @option{-Wall} request these warnings.
2818 @opindex Wconversion
2819 Warn if a prototype causes a type conversion that is different from what
2820 would happen to the same argument in the absence of a prototype. This
2821 includes conversions of fixed point to floating and vice versa, and
2822 conversions changing the width or signedness of a fixed point argument
2823 except when the same as the default promotion.
2825 Also, warn if a negative integer constant expression is implicitly
2826 converted to an unsigned type. For example, warn about the assignment
2827 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2828 casts like @code{(unsigned) -1}.
2830 @item -Wsign-compare
2831 @opindex Wsign-compare
2832 @cindex warning for comparison of signed and unsigned values
2833 @cindex comparison of signed and unsigned values, warning
2834 @cindex signed and unsigned values, comparison warning
2835 Warn when a comparison between signed and unsigned values could produce
2836 an incorrect result when the signed value is converted to unsigned.
2837 This warning is also enabled by @option{-Wextra}; to get the other warnings
2838 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2840 @item -Waggregate-return
2841 @opindex Waggregate-return
2842 Warn if any functions that return structures or unions are defined or
2843 called. (In languages where you can return an array, this also elicits
2846 @item -Wstrict-prototypes @r{(C only)}
2847 @opindex Wstrict-prototypes
2848 Warn if a function is declared or defined without specifying the
2849 argument types. (An old-style function definition is permitted without
2850 a warning if preceded by a declaration which specifies the argument
2853 @item -Wold-style-definition @r{(C only)}
2854 @opindex Wold-style-definition
2855 Warn if an old-style function definition is used. A warning is given
2856 even if there is a previous prototype.
2858 @item -Wmissing-prototypes @r{(C only)}
2859 @opindex Wmissing-prototypes
2860 Warn if a global function is defined without a previous prototype
2861 declaration. This warning is issued even if the definition itself
2862 provides a prototype. The aim is to detect global functions that fail
2863 to be declared in header files.
2865 @item -Wmissing-declarations @r{(C only)}
2866 @opindex Wmissing-declarations
2867 Warn if a global function is defined without a previous declaration.
2868 Do so even if the definition itself provides a prototype.
2869 Use this option to detect global functions that are not declared in
2872 @item -Wmissing-noreturn
2873 @opindex Wmissing-noreturn
2874 Warn about functions which might be candidates for attribute @code{noreturn}.
2875 Note these are only possible candidates, not absolute ones. Care should
2876 be taken to manually verify functions actually do not ever return before
2877 adding the @code{noreturn} attribute, otherwise subtle code generation
2878 bugs could be introduced. You will not get a warning for @code{main} in
2879 hosted C environments.
2881 @item -Wmissing-format-attribute
2882 @opindex Wmissing-format-attribute
2884 If @option{-Wformat} is enabled, also warn about functions which might be
2885 candidates for @code{format} attributes. Note these are only possible
2886 candidates, not absolute ones. GCC will guess that @code{format}
2887 attributes might be appropriate for any function that calls a function
2888 like @code{vprintf} or @code{vscanf}, but this might not always be the
2889 case, and some functions for which @code{format} attributes are
2890 appropriate may not be detected. This option has no effect unless
2891 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2893 @item -Wno-multichar
2894 @opindex Wno-multichar
2896 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2897 Usually they indicate a typo in the user's code, as they have
2898 implementation-defined values, and should not be used in portable code.
2900 @item -Wno-deprecated-declarations
2901 @opindex Wno-deprecated-declarations
2902 Do not warn about uses of functions, variables, and types marked as
2903 deprecated by using the @code{deprecated} attribute.
2904 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2905 @pxref{Type Attributes}.)
2909 Warn if a structure is given the packed attribute, but the packed
2910 attribute has no effect on the layout or size of the structure.
2911 Such structures may be mis-aligned for little benefit. For
2912 instance, in this code, the variable @code{f.x} in @code{struct bar}
2913 will be misaligned even though @code{struct bar} does not itself
2914 have the packed attribute:
2921 @} __attribute__((packed));
2931 Warn if padding is included in a structure, either to align an element
2932 of the structure or to align the whole structure. Sometimes when this
2933 happens it is possible to rearrange the fields of the structure to
2934 reduce the padding and so make the structure smaller.
2936 @item -Wredundant-decls
2937 @opindex Wredundant-decls
2938 Warn if anything is declared more than once in the same scope, even in
2939 cases where multiple declaration is valid and changes nothing.
2941 @item -Wnested-externs @r{(C only)}
2942 @opindex Wnested-externs
2943 Warn if an @code{extern} declaration is encountered within a function.
2945 @item -Wunreachable-code
2946 @opindex Wunreachable-code
2947 Warn if the compiler detects that code will never be executed.
2949 This option is intended to warn when the compiler detects that at
2950 least a whole line of source code will never be executed, because
2951 some condition is never satisfied or because it is after a
2952 procedure that never returns.
2954 It is possible for this option to produce a warning even though there
2955 are circumstances under which part of the affected line can be executed,
2956 so care should be taken when removing apparently-unreachable code.
2958 For instance, when a function is inlined, a warning may mean that the
2959 line is unreachable in only one inlined copy of the function.
2961 This option is not made part of @option{-Wall} because in a debugging
2962 version of a program there is often substantial code which checks
2963 correct functioning of the program and is, hopefully, unreachable
2964 because the program does work. Another common use of unreachable
2965 code is to provide behavior which is selectable at compile-time.
2969 Warn if a function can not be inlined and it was declared as inline.
2970 Even with this option, the compiler will not warn about failures to
2971 inline functions declared in system headers.
2973 The compiler uses a variety of heuristics to determine whether or not
2974 to inline a function. For example, the compiler takes into account
2975 the size of the function being inlined and the the amount of inlining
2976 that has already been done in the current function. Therefore,
2977 seemingly insignificant changes in the source program can cause the
2978 warnings produced by @option{-Winline} to appear or disappear.
2980 @item -Wno-invalid-offsetof @r{(C++ only)}
2981 @opindex Wno-invalid-offsetof
2982 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
2983 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
2984 to a non-POD type is undefined. In existing C++ implementations,
2985 however, @samp{offsetof} typically gives meaningful results even when
2986 applied to certain kinds of non-POD types. (Such as a simple
2987 @samp{struct} that fails to be a POD type only by virtue of having a
2988 constructor.) This flag is for users who are aware that they are
2989 writing nonportable code and who have deliberately chosen to ignore the
2992 The restrictions on @samp{offsetof} may be relaxed in a future version
2993 of the C++ standard.
2996 @opindex Winvalid-pch
2997 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
2998 the search path but can't be used.
3002 @opindex Wno-long-long
3003 Warn if @samp{long long} type is used. This is default. To inhibit
3004 the warning messages, use @option{-Wno-long-long}. Flags
3005 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3006 only when @option{-pedantic} flag is used.
3008 @item -Wdisabled-optimization
3009 @opindex Wdisabled-optimization
3010 Warn if a requested optimization pass is disabled. This warning does
3011 not generally indicate that there is anything wrong with your code; it
3012 merely indicates that GCC's optimizers were unable to handle the code
3013 effectively. Often, the problem is that your code is too big or too
3014 complex; GCC will refuse to optimize programs when the optimization
3015 itself is likely to take inordinate amounts of time.
3019 Make all warnings into errors.
3022 @node Debugging Options
3023 @section Options for Debugging Your Program or GCC
3024 @cindex options, debugging
3025 @cindex debugging information options
3027 GCC has various special options that are used for debugging
3028 either your program or GCC:
3033 Produce debugging information in the operating system's native format
3034 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
3037 On most systems that use stabs format, @option{-g} enables use of extra
3038 debugging information that only GDB can use; this extra information
3039 makes debugging work better in GDB but will probably make other debuggers
3041 refuse to read the program. If you want to control for certain whether
3042 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3043 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, @option{-gdwarf-1},
3044 or @option{-gvms} (see below).
3046 Unlike most other C compilers, GCC allows you to use @option{-g} with
3047 @option{-O}. The shortcuts taken by optimized code may occasionally
3048 produce surprising results: some variables you declared may not exist
3049 at all; flow of control may briefly move where you did not expect it;
3050 some statements may not be executed because they compute constant
3051 results or their values were already at hand; some statements may
3052 execute in different places because they were moved out of loops.
3054 Nevertheless it proves possible to debug optimized output. This makes
3055 it reasonable to use the optimizer for programs that might have bugs.
3057 The following options are useful when GCC is generated with the
3058 capability for more than one debugging format.
3062 Produce debugging information for use by GDB@. This means to use the
3063 most expressive format available (DWARF 2, stabs, or the native format
3064 if neither of those are supported), including GDB extensions if at all
3069 Produce debugging information in stabs format (if that is supported),
3070 without GDB extensions. This is the format used by DBX on most BSD
3071 systems. On MIPS, Alpha and System V Release 4 systems this option
3072 produces stabs debugging output which is not understood by DBX or SDB@.
3073 On System V Release 4 systems this option requires the GNU assembler.
3075 @item -feliminate-unused-debug-symbols
3076 @opindex feliminate-unused-debug-symbols
3077 Produce debugging information in stabs format (if that is supported),
3078 for only symbols that are actually used.
3082 Produce debugging information in stabs format (if that is supported),
3083 using GNU extensions understood only by the GNU debugger (GDB)@. The
3084 use of these extensions is likely to make other debuggers crash or
3085 refuse to read the program.
3089 Produce debugging information in COFF format (if that is supported).
3090 This is the format used by SDB on most System V systems prior to
3095 Produce debugging information in XCOFF format (if that is supported).
3096 This is the format used by the DBX debugger on IBM RS/6000 systems.
3100 Produce debugging information in XCOFF format (if that is supported),
3101 using GNU extensions understood only by the GNU debugger (GDB)@. The
3102 use of these extensions is likely to make other debuggers crash or
3103 refuse to read the program, and may cause assemblers other than the GNU
3104 assembler (GAS) to fail with an error.
3108 Produce debugging information in DWARF version 1 format (if that is
3109 supported). This is the format used by SDB on most System V Release 4
3112 This option is deprecated.
3116 Produce debugging information in DWARF version 1 format (if that is
3117 supported), using GNU extensions understood only by the GNU debugger
3118 (GDB)@. The use of these extensions is likely to make other debuggers
3119 crash or refuse to read the program.
3121 This option is deprecated.
3125 Produce debugging information in DWARF version 2 format (if that is
3126 supported). This is the format used by DBX on IRIX 6.
3130 Produce debugging information in VMS debug format (if that is
3131 supported). This is the format used by DEBUG on VMS systems.
3134 @itemx -ggdb@var{level}
3135 @itemx -gstabs@var{level}
3136 @itemx -gcoff@var{level}
3137 @itemx -gxcoff@var{level}
3138 @itemx -gvms@var{level}
3139 Request debugging information and also use @var{level} to specify how
3140 much information. The default level is 2.
3142 Level 1 produces minimal information, enough for making backtraces in
3143 parts of the program that you don't plan to debug. This includes
3144 descriptions of functions and external variables, but no information
3145 about local variables and no line numbers.
3147 Level 3 includes extra information, such as all the macro definitions
3148 present in the program. Some debuggers support macro expansion when
3149 you use @option{-g3}.
3151 Note that in order to avoid confusion between DWARF1 debug level 2,
3152 and DWARF2, neither @option{-gdwarf} nor @option{-gdwarf-2} accept
3153 a concatenated debug level. Instead use an additional @option{-g@var{level}}
3154 option to change the debug level for DWARF1 or DWARF2.
3156 @item -feliminate-dwarf2-dups
3157 @opindex feliminate-dwarf2-dups
3158 Compress DWARF2 debugging information by eliminating duplicated
3159 information about each symbol. This option only makes sense when
3160 generating DWARF2 debugging information with @option{-gdwarf-2}.
3162 @cindex @command{prof}
3165 Generate extra code to write profile information suitable for the
3166 analysis program @command{prof}. You must use this option when compiling
3167 the source files you want data about, and you must also use it when
3170 @cindex @command{gprof}
3173 Generate extra code to write profile information suitable for the
3174 analysis program @command{gprof}. You must use this option when compiling
3175 the source files you want data about, and you must also use it when
3180 Makes the compiler print out each function name as it is compiled, and
3181 print some statistics about each pass when it finishes.
3184 @opindex ftime-report
3185 Makes the compiler print some statistics about the time consumed by each
3186 pass when it finishes.
3189 @opindex fmem-report
3190 Makes the compiler print some statistics about permanent memory
3191 allocation when it finishes.
3193 @item -fprofile-arcs
3194 @opindex fprofile-arcs
3195 Add code so that program flow @dfn{arcs} are instrumented. During
3196 execution the program records how many times each branch and call is
3197 executed and how many times it is taken or returns. When the compiled
3198 program exits it saves this data to a file called
3199 @file{@var{auxname}.gcda} for each source file. The data may be used for
3200 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3201 test coverage analysis (@option{-ftest-coverage}). Each object file's
3202 @var{auxname} is generated from the name of the output file, if
3203 explicitly specified and it is not the final executable, otherwise it is
3204 the basename of the source file. In both cases any suffix is removed
3205 (e.g. @file{foo.gcda} for input file @file{dir/foo.c}, or
3206 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3211 Compile the source files with @option{-fprofile-arcs} plus optimization
3212 and code generation options. For test coverage analysis, use the
3213 additional @option{-ftest-coverage} option. You do not need to profile
3214 every source file in a program.
3217 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3218 (the latter implies the former).
3221 Run the program on a representative workload to generate the arc profile
3222 information. This may be repeated any number of times. You can run
3223 concurrent instances of your program, and provided that the file system
3224 supports locking, the data files will be correctly updated. Also
3225 @code{fork} calls are detected and correctly handled (double counting
3229 For profile-directed optimizations, compile the source files again with
3230 the same optimization and code generation options plus
3231 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3232 Control Optimization}).
3235 For test coverage analysis, use @command{gcov} to produce human readable
3236 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3237 @command{gcov} documentation for further information.
3241 With @option{-fprofile-arcs}, for each function of your program GCC
3242 creates a program flow graph, then finds a spanning tree for the graph.
3243 Only arcs that are not on the spanning tree have to be instrumented: the
3244 compiler adds code to count the number of times that these arcs are
3245 executed. When an arc is the only exit or only entrance to a block, the
3246 instrumentation code can be added to the block; otherwise, a new basic
3247 block must be created to hold the instrumentation code.
3250 @item -ftest-coverage
3251 @opindex ftest-coverage
3252 Produce a notes file that the @command{gcov} code-coverage utility
3253 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3254 show program coverage. Each source file's note file is called
3255 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3256 above for a description of @var{auxname} and instructions on how to
3257 generate test coverage data. Coverage data will match the source files
3258 more closely, if you do not optimize.
3260 @item -d@var{letters}
3262 Says to make debugging dumps during compilation at times specified by
3263 @var{letters}. This is used for debugging the compiler. The file names
3264 for most of the dumps are made by appending a pass number and a word to
3265 the @var{dumpname}. @var{dumpname} is generated from the name of the
3266 output file, if explicitly specified and it is not an executable,
3267 otherwise it is the basename of the source file. In both cases any
3268 suffix is removed (e.g. @file{foo.01.rtl} or @file{foo.02.sibling}).
3269 Here are the possible letters for use in @var{letters}, and their
3275 Annotate the assembler output with miscellaneous debugging information.
3278 Dump after computing branch probabilities, to @file{@var{file}.12.bp}.
3281 Dump after block reordering, to @file{@var{file}.30.bbro}.
3284 Dump after instruction combination, to the file @file{@var{file}.20.combine}.
3287 Dump after the first if conversion, to the file @file{@var{file}.14.ce1}.
3288 Also dump after the second if conversion, to the file @file{@var{file}.21.ce2}.
3291 Dump after branch target load optimization, to to @file{@var{file}.32.btl}.
3292 Also dump after delayed branch scheduling, to @file{@var{file}.36.dbr}.
3295 Dump all macro definitions, at the end of preprocessing, in addition to
3299 Dump after the second if conversion, to @file{@var{file}.31.ce3}.
3302 Dump after control and data flow analysis, to @file{@var{file}.11.cfg}.
3303 Also dump after life analysis, to @file{@var{file}.19.life}.
3306 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.07.addressof}.
3309 Dump after global register allocation, to @file{@var{file}.25.greg}.
3312 Dump after GCSE, to @file{@var{file}.08.gcse}.
3313 Also dump after jump bypassing and control flow optimizations, to
3314 @file{@var{file}.10.bypass}.
3317 Dump after finalization of EH handling code, to @file{@var{file}.03.eh}.
3320 Dump after sibling call optimizations, to @file{@var{file}.02.sibling}.
3323 Dump after the first jump optimization, to @file{@var{file}.04.jump}.
3326 Dump after conversion from registers to stack, to @file{@var{file}.34.stack}.
3329 Dump after local register allocation, to @file{@var{file}.24.lreg}.
3332 Dump after loop optimization passes, to @file{@var{file}.09.loop} and
3333 @file{@var{file}.16.loop2}.
3336 Dump after performing the machine dependent reorganization pass, to
3337 @file{@var{file}.35.mach}.
3340 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3343 Dump after the register move pass, to @file{@var{file}.22.regmove}.
3346 Dump after post-reload optimizations, to @file{@var{file}.26.postreload}.
3349 Dump after RTL generation, to @file{@var{file}.01.rtl}.
3352 Dump after the second scheduling pass, to @file{@var{file}.33.sched2}.
3355 Dump after CSE (including the jump optimization that sometimes follows
3356 CSE), to @file{@var{file}.06.cse}.
3359 Dump after the first scheduling pass, to @file{@var{file}.23.sched}.
3362 Dump after the second CSE pass (including the jump optimization that
3363 sometimes follows CSE), to @file{@var{file}.18.cse2}.
3366 Dump after running tracer, to @file{@var{file}.15.tracer}.
3369 Dump after null pointer elimination pass to @file{@var{file}.05.null}.
3372 Dump callgraph and unit-at-a-time optimization @file{@var{file}.00.unit}.
3375 Dump after the value profile transformations, to @file{@var{file}.13.vpt}.
3378 Dump after the second flow pass, to @file{@var{file}.27.flow2}.
3381 Dump after the peephole pass, to @file{@var{file}.28.peephole2}.
3384 Dump after constructing the web, to @file{@var{file}.17.web}.
3387 Produce all the dumps listed above.
3390 Produce a core dump whenever an error occurs.
3393 Print statistics on memory usage, at the end of the run, to
3397 Annotate the assembler output with a comment indicating which
3398 pattern and alternative was used. The length of each instruction is
3402 Dump the RTL in the assembler output as a comment before each instruction.
3403 Also turns on @option{-dp} annotation.
3406 For each of the other indicated dump files (except for
3407 @file{@var{file}.01.rtl}), dump a representation of the control flow graph
3408 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3411 Just generate RTL for a function instead of compiling it. Usually used
3415 Dump debugging information during parsing, to standard error.
3418 @item -fdump-unnumbered
3419 @opindex fdump-unnumbered
3420 When doing debugging dumps (see @option{-d} option above), suppress instruction
3421 numbers and line number note output. This makes it more feasible to
3422 use diff on debugging dumps for compiler invocations with different
3423 options, in particular with and without @option{-g}.
3425 @item -fdump-translation-unit @r{(C and C++ only)}
3426 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3427 @opindex fdump-translation-unit
3428 Dump a representation of the tree structure for the entire translation
3429 unit to a file. The file name is made by appending @file{.tu} to the
3430 source file name. If the @samp{-@var{options}} form is used, @var{options}
3431 controls the details of the dump as described for the
3432 @option{-fdump-tree} options.
3434 @item -fdump-class-hierarchy @r{(C++ only)}
3435 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3436 @opindex fdump-class-hierarchy
3437 Dump a representation of each class's hierarchy and virtual function
3438 table layout to a file. The file name is made by appending @file{.class}
3439 to the source file name. If the @samp{-@var{options}} form is used,
3440 @var{options} controls the details of the dump as described for the
3441 @option{-fdump-tree} options.
3443 @item -fdump-tree-@var{switch} @r{(C++ only)}
3444 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3446 Control the dumping at various stages of processing the intermediate
3447 language tree to a file. The file name is generated by appending a switch
3448 specific suffix to the source file name. If the @samp{-@var{options}}
3449 form is used, @var{options} is a list of @samp{-} separated options that
3450 control the details of the dump. Not all options are applicable to all
3451 dumps, those which are not meaningful will be ignored. The following
3452 options are available
3456 Print the address of each node. Usually this is not meaningful as it
3457 changes according to the environment and source file. Its primary use
3458 is for tying up a dump file with a debug environment.
3460 Inhibit dumping of members of a scope or body of a function merely
3461 because that scope has been reached. Only dump such items when they
3462 are directly reachable by some other path.
3464 Turn on all options.
3467 The following tree dumps are possible:
3470 Dump before any tree based optimization, to @file{@var{file}.original}.
3472 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3474 Dump after function inlining, to @file{@var{file}.inlined}.
3477 @item -frandom-seed=@var{string}
3478 @opindex frandom-string
3479 This option provides a seed that GCC uses when it would otherwise use
3480 random numbers. It is used to generate certain symbol names
3481 that have to be different in every compiled file. It is also used to
3482 place unique stamps in coverage data files and the object files that
3483 produce them. You can use the @option{-frandom-seed} option to produce
3484 reproducibly identical object files.
3486 The @var{string} should be different for every file you compile.
3488 @item -fsched-verbose=@var{n}
3489 @opindex fsched-verbose
3490 On targets that use instruction scheduling, this option controls the
3491 amount of debugging output the scheduler prints. This information is
3492 written to standard error, unless @option{-dS} or @option{-dR} is
3493 specified, in which case it is output to the usual dump
3494 listing file, @file{.sched} or @file{.sched2} respectively. However
3495 for @var{n} greater than nine, the output is always printed to standard
3498 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3499 same information as @option{-dRS}. For @var{n} greater than one, it
3500 also output basic block probabilities, detailed ready list information
3501 and unit/insn info. For @var{n} greater than two, it includes RTL
3502 at abort point, control-flow and regions info. And for @var{n} over
3503 four, @option{-fsched-verbose} also includes dependence info.
3507 Store the usual ``temporary'' intermediate files permanently; place them
3508 in the current directory and name them based on the source file. Thus,
3509 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3510 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3511 preprocessed @file{foo.i} output file even though the compiler now
3512 normally uses an integrated preprocessor.
3516 Report the CPU time taken by each subprocess in the compilation
3517 sequence. For C source files, this is the compiler proper and assembler
3518 (plus the linker if linking is done). The output looks like this:
3525 The first number on each line is the ``user time,'' that is time spent
3526 executing the program itself. The second number is ``system time,''
3527 time spent executing operating system routines on behalf of the program.
3528 Both numbers are in seconds.
3530 @item -print-file-name=@var{library}
3531 @opindex print-file-name
3532 Print the full absolute name of the library file @var{library} that
3533 would be used when linking---and don't do anything else. With this
3534 option, GCC does not compile or link anything; it just prints the
3537 @item -print-multi-directory
3538 @opindex print-multi-directory
3539 Print the directory name corresponding to the multilib selected by any
3540 other switches present in the command line. This directory is supposed
3541 to exist in @env{GCC_EXEC_PREFIX}.
3543 @item -print-multi-lib
3544 @opindex print-multi-lib
3545 Print the mapping from multilib directory names to compiler switches
3546 that enable them. The directory name is separated from the switches by
3547 @samp{;}, and each switch starts with an @samp{@@} instead of the
3548 @samp{-}, without spaces between multiple switches. This is supposed to
3549 ease shell-processing.
3551 @item -print-prog-name=@var{program}
3552 @opindex print-prog-name
3553 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3555 @item -print-libgcc-file-name
3556 @opindex print-libgcc-file-name
3557 Same as @option{-print-file-name=libgcc.a}.
3559 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3560 but you do want to link with @file{libgcc.a}. You can do
3563 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3566 @item -print-search-dirs
3567 @opindex print-search-dirs
3568 Print the name of the configured installation directory and a list of
3569 program and library directories gcc will search---and don't do anything else.
3571 This is useful when gcc prints the error message
3572 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3573 To resolve this you either need to put @file{cpp0} and the other compiler
3574 components where gcc expects to find them, or you can set the environment
3575 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3576 Don't forget the trailing '/'.
3577 @xref{Environment Variables}.
3580 @opindex dumpmachine
3581 Print the compiler's target machine (for example,
3582 @samp{i686-pc-linux-gnu})---and don't do anything else.
3585 @opindex dumpversion
3586 Print the compiler version (for example, @samp{3.0})---and don't do
3591 Print the compiler's built-in specs---and don't do anything else. (This
3592 is used when GCC itself is being built.) @xref{Spec Files}.
3594 @item -feliminate-unused-debug-types
3595 @opindex feliminate-unused-debug-types
3596 Normally, when producing DWARF2 output, GCC will emit debugging
3597 information for all types declared in a compilation
3598 unit, regardless of whether or not they are actually used
3599 in that compilation unit. Sometimes this is useful, such as
3600 if, in the debugger, you want to cast a value to a type that is
3601 not actually used in your program (but is declared). More often,
3602 however, this results in a significant amount of wasted space.
3603 With this option, GCC will avoid producing debug symbol output
3604 for types that are nowhere used in the source file being compiled.
3607 @node Optimize Options
3608 @section Options That Control Optimization
3609 @cindex optimize options
3610 @cindex options, optimization
3612 These options control various sorts of optimizations.
3614 Without any optimization option, the compiler's goal is to reduce the
3615 cost of compilation and to make debugging produce the expected
3616 results. Statements are independent: if you stop the program with a
3617 breakpoint between statements, you can then assign a new value to any
3618 variable or change the program counter to any other statement in the
3619 function and get exactly the results you would expect from the source
3622 Turning on optimization flags makes the compiler attempt to improve
3623 the performance and/or code size at the expense of compilation time
3624 and possibly the ability to debug the program.
3626 The compiler performs optimisation based on the knowledge it has of
3627 the program. Using the @option{-funit-at-a-time} flag will allow the
3628 compiler to consider information gained from later functions in the
3629 file when compiling a function. Compiling multiple files at once to a
3630 single output file (and using @option{-funit-at-a-time}) will allow
3631 the compiler to use information gained from all of the files when
3632 compiling each of them.
3634 Not all optimizations are controlled directly by a flag. Only
3635 optimizations that have a flag are listed.
3642 Optimize. Optimizing compilation takes somewhat more time, and a lot
3643 more memory for a large function.
3645 With @option{-O}, the compiler tries to reduce code size and execution
3646 time, without performing any optimizations that take a great deal of
3649 @option{-O} turns on the following optimization flags:
3650 @gccoptlist{-fdefer-pop @gol
3651 -fmerge-constants @gol
3653 -floop-optimize @gol
3655 -fif-conversion @gol
3656 -fif-conversion2 @gol
3657 -fdelayed-branch @gol
3658 -fguess-branch-probability @gol
3661 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3662 where doing so does not interfere with debugging.
3666 Optimize even more. GCC performs nearly all supported optimizations
3667 that do not involve a space-speed tradeoff. The compiler does not
3668 perform loop unrolling or function inlining when you specify @option{-O2}.
3669 As compared to @option{-O}, this option increases both compilation time
3670 and the performance of the generated code.
3672 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3673 also turns on the following optimization flags:
3674 @gccoptlist{-fforce-mem @gol
3675 -foptimize-sibling-calls @gol
3676 -fstrength-reduce @gol
3677 -fcse-follow-jumps -fcse-skip-blocks @gol
3678 -frerun-cse-after-loop -frerun-loop-opt @gol
3679 -fgcse -fgcse-lm -fgcse-sm -fgcse-las @gol
3680 -fdelete-null-pointer-checks @gol
3681 -fexpensive-optimizations @gol
3683 -fschedule-insns -fschedule-insns2 @gol
3684 -fsched-interblock -fsched-spec @gol
3687 -freorder-blocks -freorder-functions @gol
3688 -fstrict-aliasing @gol
3689 -funit-at-a-time -fweb @gol
3690 -falign-functions -falign-jumps @gol
3691 -falign-loops -falign-labels}
3693 Please note the warning under @option{-fgcse} about
3694 invoking @option{-O2} on programs that use computed gotos.
3698 Optimize yet more. @option{-O3} turns on all optimizations specified by
3699 @option{-O2} and also turns on the @option{-finline-functions},
3700 @option{-fweb} and @option{-frename-registers} options.
3704 Do not optimize. This is the default.
3708 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3709 do not typically increase code size. It also performs further
3710 optimizations designed to reduce code size.
3712 @option{-Os} disables the following optimization flags:
3713 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3714 -falign-labels -freorder-blocks -fprefetch-loop-arrays}
3716 If you use multiple @option{-O} options, with or without level numbers,
3717 the last such option is the one that is effective.
3720 Options of the form @option{-f@var{flag}} specify machine-independent
3721 flags. Most flags have both positive and negative forms; the negative
3722 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3723 below, only one of the forms is listed---the one you typically will
3724 use. You can figure out the other form by either removing @samp{no-}
3727 The following options control specific optimizations. They are either
3728 activated by @option{-O} options or are related to ones that are. You
3729 can use the following flags in the rare cases when ``fine-tuning'' of
3730 optimizations to be performed is desired.
3733 @item -fno-default-inline
3734 @opindex fno-default-inline
3735 Do not make member functions inline by default merely because they are
3736 defined inside the class scope (C++ only). Otherwise, when you specify
3737 @w{@option{-O}}, member functions defined inside class scope are compiled
3738 inline by default; i.e., you don't need to add @samp{inline} in front of
3739 the member function name.
3741 @item -fno-defer-pop
3742 @opindex fno-defer-pop
3743 Always pop the arguments to each function call as soon as that function
3744 returns. For machines which must pop arguments after a function call,
3745 the compiler normally lets arguments accumulate on the stack for several
3746 function calls and pops them all at once.
3748 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3752 Force memory operands to be copied into registers before doing
3753 arithmetic on them. This produces better code by making all memory
3754 references potential common subexpressions. When they are not common
3755 subexpressions, instruction combination should eliminate the separate
3758 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3761 @opindex fforce-addr
3762 Force memory address constants to be copied into registers before
3763 doing arithmetic on them. This may produce better code just as
3764 @option{-fforce-mem} may.
3766 @item -fomit-frame-pointer
3767 @opindex fomit-frame-pointer
3768 Don't keep the frame pointer in a register for functions that
3769 don't need one. This avoids the instructions to save, set up and
3770 restore frame pointers; it also makes an extra register available
3771 in many functions. @strong{It also makes debugging impossible on
3774 On some machines, such as the VAX, this flag has no effect, because
3775 the standard calling sequence automatically handles the frame pointer
3776 and nothing is saved by pretending it doesn't exist. The
3777 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3778 whether a target machine supports this flag. @xref{Registers,,Register
3779 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3781 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3783 @item -foptimize-sibling-calls
3784 @opindex foptimize-sibling-calls
3785 Optimize sibling and tail recursive calls.
3787 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3791 Don't pay attention to the @code{inline} keyword. Normally this option
3792 is used to keep the compiler from expanding any functions inline.
3793 Note that if you are not optimizing, no functions can be expanded inline.
3795 @item -finline-functions
3796 @opindex finline-functions
3797 Integrate all simple functions into their callers. The compiler
3798 heuristically decides which functions are simple enough to be worth
3799 integrating in this way.
3801 If all calls to a given function are integrated, and the function is
3802 declared @code{static}, then the function is normally not output as
3803 assembler code in its own right.
3805 Enabled at level @option{-O3}.
3807 @item -finline-limit=@var{n}
3808 @opindex finline-limit
3809 By default, gcc limits the size of functions that can be inlined. This flag
3810 allows the control of this limit for functions that are explicitly marked as
3811 inline (i.e., marked with the inline keyword or defined within the class
3812 definition in c++). @var{n} is the size of functions that can be inlined in
3813 number of pseudo instructions (not counting parameter handling). The default
3814 value of @var{n} is 600.
3815 Increasing this value can result in more inlined code at
3816 the cost of compilation time and memory consumption. Decreasing usually makes
3817 the compilation faster and less code will be inlined (which presumably
3818 means slower programs). This option is particularly useful for programs that
3819 use inlining heavily such as those based on recursive templates with C++.
3821 Inlining is actually controlled by a number of parameters, which may be
3822 specified individually by using @option{--param @var{name}=@var{value}}.
3823 The @option{-finline-limit=@var{n}} option sets some of these parameters
3827 @item max-inline-insns-single
3828 is set to @var{n}/2.
3829 @item max-inline-insns-auto
3830 is set to @var{n}/2.
3831 @item min-inline-insns
3832 is set to 130 or @var{n}/4, whichever is smaller.
3833 @item max-inline-insns-rtl
3837 See below for a documentation of the individual
3838 parameters controlling inlining.
3840 @emph{Note:} pseudo instruction represents, in this particular context, an
3841 abstract measurement of function's size. In no way, it represents a count
3842 of assembly instructions and as such its exact meaning might change from one
3843 release to an another.
3845 @item -fkeep-inline-functions
3846 @opindex fkeep-inline-functions
3847 Even if all calls to a given function are integrated, and the function
3848 is declared @code{static}, nevertheless output a separate run-time
3849 callable version of the function. This switch does not affect
3850 @code{extern inline} functions.
3852 @item -fkeep-static-consts
3853 @opindex fkeep-static-consts
3854 Emit variables declared @code{static const} when optimization isn't turned
3855 on, even if the variables aren't referenced.
3857 GCC enables this option by default. If you want to force the compiler to
3858 check if the variable was referenced, regardless of whether or not
3859 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3861 @item -fmerge-constants
3862 Attempt to merge identical constants (string constants and floating point
3863 constants) across compilation units.
3865 This option is the default for optimized compilation if the assembler and
3866 linker support it. Use @option{-fno-merge-constants} to inhibit this
3869 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3871 @item -fmerge-all-constants
3872 Attempt to merge identical constants and identical variables.
3874 This option implies @option{-fmerge-constants}. In addition to
3875 @option{-fmerge-constants} this considers e.g. even constant initialized
3876 arrays or initialized constant variables with integral or floating point
3877 types. Languages like C or C++ require each non-automatic variable to
3878 have distinct location, so using this option will result in non-conforming
3883 Use a graph coloring register allocator. Currently this option is meant
3884 for testing, so we are interested to hear about miscompilations with
3887 @item -fno-branch-count-reg
3888 @opindex fno-branch-count-reg
3889 Do not use ``decrement and branch'' instructions on a count register,
3890 but instead generate a sequence of instructions that decrement a
3891 register, compare it against zero, then branch based upon the result.
3892 This option is only meaningful on architectures that support such
3893 instructions, which include x86, PowerPC, IA-64 and S/390.
3895 The default is @option{-fbranch-count-reg}, enabled when
3896 @option{-fstrength-reduce} is enabled.
3898 @item -fno-function-cse
3899 @opindex fno-function-cse
3900 Do not put function addresses in registers; make each instruction that
3901 calls a constant function contain the function's address explicitly.
3903 This option results in less efficient code, but some strange hacks
3904 that alter the assembler output may be confused by the optimizations
3905 performed when this option is not used.
3907 The default is @option{-ffunction-cse}
3909 @item -fno-zero-initialized-in-bss
3910 @opindex fno-zero-initialized-in-bss
3911 If the target supports a BSS section, GCC by default puts variables that
3912 are initialized to zero into BSS@. This can save space in the resulting
3915 This option turns off this behavior because some programs explicitly
3916 rely on variables going to the data section. E.g., so that the
3917 resulting executable can find the beginning of that section and/or make
3918 assumptions based on that.
3920 The default is @option{-fzero-initialized-in-bss}.
3922 @item -fstrength-reduce
3923 @opindex fstrength-reduce
3924 Perform the optimizations of loop strength reduction and
3925 elimination of iteration variables.
3927 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3929 @item -fthread-jumps
3930 @opindex fthread-jumps
3931 Perform optimizations where we check to see if a jump branches to a
3932 location where another comparison subsumed by the first is found. If
3933 so, the first branch is redirected to either the destination of the
3934 second branch or a point immediately following it, depending on whether
3935 the condition is known to be true or false.
3937 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3939 @item -fcse-follow-jumps
3940 @opindex fcse-follow-jumps
3941 In common subexpression elimination, scan through jump instructions
3942 when the target of the jump is not reached by any other path. For
3943 example, when CSE encounters an @code{if} statement with an
3944 @code{else} clause, CSE will follow the jump when the condition
3947 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3949 @item -fcse-skip-blocks
3950 @opindex fcse-skip-blocks
3951 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3952 follow jumps which conditionally skip over blocks. When CSE
3953 encounters a simple @code{if} statement with no else clause,
3954 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3955 body of the @code{if}.
3957 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3959 @item -frerun-cse-after-loop
3960 @opindex frerun-cse-after-loop
3961 Re-run common subexpression elimination after loop optimizations has been
3964 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3966 @item -frerun-loop-opt
3967 @opindex frerun-loop-opt
3968 Run the loop optimizer twice.
3970 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3974 Perform a global common subexpression elimination pass.
3975 This pass also performs global constant and copy propagation.
3977 @emph{Note:} When compiling a program using computed gotos, a GCC
3978 extension, you may get better runtime performance if you disable
3979 the global common subexpression elimination pass by adding
3980 @option{-fno-gcse} to the command line.
3982 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3986 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3987 attempt to move loads which are only killed by stores into themselves. This
3988 allows a loop containing a load/store sequence to be changed to a load outside
3989 the loop, and a copy/store within the loop.
3991 Enabled by default when gcse is enabled.
3995 When @option{-fgcse-sm} is enabled, a store motion pass is run after
3996 global common subexpression elimination. This pass will attempt to move
3997 stores out of loops. When used in conjunction with @option{-fgcse-lm},
3998 loops containing a load/store sequence can be changed to a load before
3999 the loop and a store after the loop.
4001 Enabled by default when gcse is enabled.
4005 When @option{-fgcse-las} is enabled, the global common subexpression
4006 elimination pass eliminates redundant loads that come after stores to the
4007 same memory location (both partial and full redundancies).
4009 Enabled by default when gcse is enabled.
4011 @item -floop-optimize
4012 @opindex floop-optimize
4013 Perform loop optimizations: move constant expressions out of loops, simplify
4014 exit test conditions and optionally do strength-reduction and loop unrolling as
4017 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4019 @item -fcrossjumping
4020 @opindex crossjumping
4021 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4022 resulting code may or may not perform better than without cross-jumping.
4024 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4026 @item -fif-conversion
4027 @opindex if-conversion
4028 Attempt to transform conditional jumps into branch-less equivalents. This
4029 include use of conditional moves, min, max, set flags and abs instructions, and
4030 some tricks doable by standard arithmetics. The use of conditional execution
4031 on chips where it is available is controlled by @code{if-conversion2}.
4033 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4035 @item -fif-conversion2
4036 @opindex if-conversion2
4037 Use conditional execution (where available) to transform conditional jumps into
4038 branch-less equivalents.
4040 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4042 @item -fdelete-null-pointer-checks
4043 @opindex fdelete-null-pointer-checks
4044 Use global dataflow analysis to identify and eliminate useless checks
4045 for null pointers. The compiler assumes that dereferencing a null
4046 pointer would have halted the program. If a pointer is checked after
4047 it has already been dereferenced, it cannot be null.
4049 In some environments, this assumption is not true, and programs can
4050 safely dereference null pointers. Use
4051 @option{-fno-delete-null-pointer-checks} to disable this optimization
4052 for programs which depend on that behavior.
4054 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4056 @item -fexpensive-optimizations
4057 @opindex fexpensive-optimizations
4058 Perform a number of minor optimizations that are relatively expensive.
4060 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4062 @item -foptimize-register-move
4064 @opindex foptimize-register-move
4066 Attempt to reassign register numbers in move instructions and as
4067 operands of other simple instructions in order to maximize the amount of
4068 register tying. This is especially helpful on machines with two-operand
4071 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4074 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4076 @item -fdelayed-branch
4077 @opindex fdelayed-branch
4078 If supported for the target machine, attempt to reorder instructions
4079 to exploit instruction slots available after delayed branch
4082 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4084 @item -fschedule-insns
4085 @opindex fschedule-insns
4086 If supported for the target machine, attempt to reorder instructions to
4087 eliminate execution stalls due to required data being unavailable. This
4088 helps machines that have slow floating point or memory load instructions
4089 by allowing other instructions to be issued until the result of the load
4090 or floating point instruction is required.
4092 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4094 @item -fschedule-insns2
4095 @opindex fschedule-insns2
4096 Similar to @option{-fschedule-insns}, but requests an additional pass of
4097 instruction scheduling after register allocation has been done. This is
4098 especially useful on machines with a relatively small number of
4099 registers and where memory load instructions take more than one cycle.
4101 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4103 @item -fno-sched-interblock
4104 @opindex fno-sched-interblock
4105 Don't schedule instructions across basic blocks. This is normally
4106 enabled by default when scheduling before register allocation, i.e.@:
4107 with @option{-fschedule-insns} or at @option{-O2} or higher.
4109 @item -fno-sched-spec
4110 @opindex fno-sched-spec
4111 Don't allow speculative motion of non-load instructions. This is normally
4112 enabled by default when scheduling before register allocation, i.e.@:
4113 with @option{-fschedule-insns} or at @option{-O2} or higher.
4115 @item -fsched-spec-load
4116 @opindex fsched-spec-load
4117 Allow speculative motion of some load instructions. This only makes
4118 sense when scheduling before register allocation, i.e.@: with
4119 @option{-fschedule-insns} or at @option{-O2} or higher.
4121 @item -fsched-spec-load-dangerous
4122 @opindex fsched-spec-load-dangerous
4123 Allow speculative motion of more load instructions. This only makes
4124 sense when scheduling before register allocation, i.e.@: with
4125 @option{-fschedule-insns} or at @option{-O2} or higher.
4127 @item -fsched-stalled-insns=@var{n}
4128 @opindex fsched-stalled-insns
4129 Define how many insns (if any) can be moved prematurely from the queue
4130 of stalled insns into the ready list, during the second scheduling pass.
4132 @item -fsched-stalled-insns-dep=@var{n}
4133 @opindex fsched-stalled-insns-dep
4134 Define how many insn groups (cycles) will be examined for a dependency
4135 on a stalled insn that is candidate for premature removal from the queue
4136 of stalled insns. Has an effect only during the second scheduling pass,
4137 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4139 @item -fsched2-use-superblocks
4140 @opindex fsched2-use-superblocks
4141 When scheduling after register allocation, do use superblock scheduling
4142 algorithm. Superblock scheduling allows motion across basic block boundaries
4143 resulting on faster schedules. This option is experimental, as not all machine
4144 descriptions used by GCC model the CPU closely enough to avoid unreliable
4145 results from the algorithm.
4147 This only makes sense when scheduling after register allocation, i.e.@: with
4148 @option{-fschedule-insns2} or at @option{-O2} or higher.
4150 @item -fsched2-use-traces
4151 @opindex fsched2-use-traces
4152 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4153 allocation and additionally perform code duplication in order to increase the
4154 size of superblocks using tracer pass. See @option{-ftracer} for details on
4157 This mode should produce faster but significantly longer programs. Also
4158 without @code{-fbranch-probabilities} the traces constructed may not match the
4159 reality and hurt the performance. This only makes
4160 sense when scheduling after register allocation, i.e.@: with
4161 @option{-fschedule-insns2} or at @option{-O2} or higher.
4163 @item -fcaller-saves
4164 @opindex fcaller-saves
4165 Enable values to be allocated in registers that will be clobbered by
4166 function calls, by emitting extra instructions to save and restore the
4167 registers around such calls. Such allocation is done only when it
4168 seems to result in better code than would otherwise be produced.
4170 This option is always enabled by default on certain machines, usually
4171 those which have no call-preserved registers to use instead.
4173 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4175 @item -fmove-all-movables
4176 @opindex fmove-all-movables
4177 Forces all invariant computations in loops to be moved
4180 @item -freduce-all-givs
4181 @opindex freduce-all-givs
4182 Forces all general-induction variables in loops to be
4185 @emph{Note:} When compiling programs written in Fortran,
4186 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4187 by default when you use the optimizer.
4189 These options may generate better or worse code; results are highly
4190 dependent on the structure of loops within the source code.
4192 These two options are intended to be removed someday, once
4193 they have helped determine the efficacy of various
4194 approaches to improving loop optimizations.
4196 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
4197 know how use of these options affects
4198 the performance of your production code.
4199 We're very interested in code that runs @emph{slower}
4200 when these options are @emph{enabled}.
4203 @itemx -fno-peephole2
4204 @opindex fno-peephole
4205 @opindex fno-peephole2
4206 Disable any machine-specific peephole optimizations. The difference
4207 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4208 are implemented in the compiler; some targets use one, some use the
4209 other, a few use both.
4211 @option{-fpeephole} is enabled by default.
4212 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4214 @item -fno-guess-branch-probability
4215 @opindex fno-guess-branch-probability
4216 Do not guess branch probabilities using a randomized model.
4218 Sometimes gcc will opt to use a randomized model to guess branch
4219 probabilities, when none are available from either profiling feedback
4220 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4221 different runs of the compiler on the same program may produce different
4224 In a hard real-time system, people don't want different runs of the
4225 compiler to produce code that has different behavior; minimizing
4226 non-determinism is of paramount import. This switch allows users to
4227 reduce non-determinism, possibly at the expense of inferior
4230 The default is @option{-fguess-branch-probability} at levels
4231 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4233 @item -freorder-blocks
4234 @opindex freorder-blocks
4235 Reorder basic blocks in the compiled function in order to reduce number of
4236 taken branches and improve code locality.
4238 Enabled at levels @option{-O2}, @option{-O3}.
4240 @item -freorder-functions
4241 @opindex freorder-functions
4242 Reorder basic blocks in the compiled function in order to reduce number of
4243 taken branches and improve code locality. This is implemented by using special
4244 subsections @code{text.hot} for most frequently executed functions and
4245 @code{text.unlikely} for unlikely executed functions. Reordering is done by
4246 the linker so object file format must support named sections and linker must
4247 place them in a reasonable way.
4249 Also profile feedback must be available in to make this option effective. See
4250 @option{-fprofile-arcs} for details.
4252 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4254 @item -fstrict-aliasing
4255 @opindex fstrict-aliasing
4256 Allows the compiler to assume the strictest aliasing rules applicable to
4257 the language being compiled. For C (and C++), this activates
4258 optimizations based on the type of expressions. In particular, an
4259 object of one type is assumed never to reside at the same address as an
4260 object of a different type, unless the types are almost the same. For
4261 example, an @code{unsigned int} can alias an @code{int}, but not a
4262 @code{void*} or a @code{double}. A character type may alias any other
4265 Pay special attention to code like this:
4278 The practice of reading from a different union member than the one most
4279 recently written to (called ``type-punning'') is common. Even with
4280 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4281 is accessed through the union type. So, the code above will work as
4282 expected. However, this code might not:
4293 Every language that wishes to perform language-specific alias analysis
4294 should define a function that computes, given an @code{tree}
4295 node, an alias set for the node. Nodes in different alias sets are not
4296 allowed to alias. For an example, see the C front-end function
4297 @code{c_get_alias_set}.
4299 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4301 @item -falign-functions
4302 @itemx -falign-functions=@var{n}
4303 @opindex falign-functions
4304 Align the start of functions to the next power-of-two greater than
4305 @var{n}, skipping up to @var{n} bytes. For instance,
4306 @option{-falign-functions=32} aligns functions to the next 32-byte
4307 boundary, but @option{-falign-functions=24} would align to the next
4308 32-byte boundary only if this can be done by skipping 23 bytes or less.
4310 @option{-fno-align-functions} and @option{-falign-functions=1} are
4311 equivalent and mean that functions will not be aligned.
4313 Some assemblers only support this flag when @var{n} is a power of two;
4314 in that case, it is rounded up.
4316 If @var{n} is not specified or is zero, use a machine-dependent default.
4318 Enabled at levels @option{-O2}, @option{-O3}.
4320 @item -falign-labels
4321 @itemx -falign-labels=@var{n}
4322 @opindex falign-labels
4323 Align all branch targets to a power-of-two boundary, skipping up to
4324 @var{n} bytes like @option{-falign-functions}. This option can easily
4325 make code slower, because it must insert dummy operations for when the
4326 branch target is reached in the usual flow of the code.
4328 @option{-fno-align-labels} and @option{-falign-labels=1} are
4329 equivalent and mean that labels will not be aligned.
4331 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4332 are greater than this value, then their values are used instead.
4334 If @var{n} is not specified or is zero, use a machine-dependent default
4335 which is very likely to be @samp{1}, meaning no alignment.
4337 Enabled at levels @option{-O2}, @option{-O3}.
4340 @itemx -falign-loops=@var{n}
4341 @opindex falign-loops
4342 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4343 like @option{-falign-functions}. The hope is that the loop will be
4344 executed many times, which will make up for any execution of the dummy
4347 @option{-fno-align-loops} and @option{-falign-loops=1} are
4348 equivalent and mean that loops will not be aligned.
4350 If @var{n} is not specified or is zero, use a machine-dependent default.
4352 Enabled at levels @option{-O2}, @option{-O3}.
4355 @itemx -falign-jumps=@var{n}
4356 @opindex falign-jumps
4357 Align branch targets to a power-of-two boundary, for branch targets
4358 where the targets can only be reached by jumping, skipping up to @var{n}
4359 bytes like @option{-falign-functions}. In this case, no dummy operations
4362 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4363 equivalent and mean that loops will not be aligned.
4365 If @var{n} is not specified or is zero, use a machine-dependent default.
4367 Enabled at levels @option{-O2}, @option{-O3}.
4369 @item -frename-registers
4370 @opindex frename-registers
4371 Attempt to avoid false dependencies in scheduled code by making use
4372 of registers left over after register allocation. This optimization
4373 will most benefit processors with lots of registers. It can, however,
4374 make debugging impossible, since variables will no longer stay in
4375 a ``home register''.
4379 Constructs webs as commonly used for register allocation purposes and assign
4380 each web individual pseudo register. This allows our register allocation pass
4381 to operate on pseudos directly, but also strengthens several other optimization
4382 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
4383 however, make debugging impossible, since variables will no longer stay in a
4386 Enabled at levels @option{-O3}.
4388 @item -fno-cprop-registers
4389 @opindex fno-cprop-registers
4390 After register allocation and post-register allocation instruction splitting,
4391 we perform a copy-propagation pass to try to reduce scheduling dependencies
4392 and occasionally eliminate the copy.
4394 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4398 The following options control compiler behavior regarding floating
4399 point arithmetic. These options trade off between speed and
4400 correctness. All must be specifically enabled.
4404 @opindex ffloat-store
4405 Do not store floating point variables in registers, and inhibit other
4406 options that might change whether a floating point value is taken from a
4409 @cindex floating point precision
4410 This option prevents undesirable excess precision on machines such as
4411 the 68000 where the floating registers (of the 68881) keep more
4412 precision than a @code{double} is supposed to have. Similarly for the
4413 x86 architecture. For most programs, the excess precision does only
4414 good, but a few programs rely on the precise definition of IEEE floating
4415 point. Use @option{-ffloat-store} for such programs, after modifying
4416 them to store all pertinent intermediate computations into variables.
4420 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4421 @option{-fno-trapping-math}, @option{-ffinite-math-only},
4422 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
4424 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4426 This option should never be turned on by any @option{-O} option since
4427 it can result in incorrect output for programs which depend on
4428 an exact implementation of IEEE or ISO rules/specifications for
4431 @item -fno-math-errno
4432 @opindex fno-math-errno
4433 Do not set ERRNO after calling math functions that are executed
4434 with a single instruction, e.g., sqrt. A program that relies on
4435 IEEE exceptions for math error handling may want to use this flag
4436 for speed while maintaining IEEE arithmetic compatibility.
4438 This option should never be turned on by any @option{-O} option since
4439 it can result in incorrect output for programs which depend on
4440 an exact implementation of IEEE or ISO rules/specifications for
4443 The default is @option{-fmath-errno}.
4445 @item -funsafe-math-optimizations
4446 @opindex funsafe-math-optimizations
4447 Allow optimizations for floating-point arithmetic that (a) assume
4448 that arguments and results are valid and (b) may violate IEEE or
4449 ANSI standards. When used at link-time, it may include libraries
4450 or startup files that change the default FPU control word or other
4451 similar optimizations.
4453 This option should never be turned on by any @option{-O} option since
4454 it can result in incorrect output for programs which depend on
4455 an exact implementation of IEEE or ISO rules/specifications for
4458 The default is @option{-fno-unsafe-math-optimizations}.
4460 @item -ffinite-math-only
4461 @opindex ffinite-math-only
4462 Allow optimizations for floating-point arithmetic that assume
4463 that arguments and results are not NaNs or +-Infs.
4465 This option should never be turned on by any @option{-O} option since
4466 it can result in incorrect output for programs which depend on
4467 an exact implementation of IEEE or ISO rules/specifications.
4469 The default is @option{-fno-finite-math-only}.
4471 @item -fno-trapping-math
4472 @opindex fno-trapping-math
4473 Compile code assuming that floating-point operations cannot generate
4474 user-visible traps. These traps include division by zero, overflow,
4475 underflow, inexact result and invalid operation. This option implies
4476 @option{-fno-signaling-nans}. Setting this option may allow faster
4477 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4479 This option should never be turned on by any @option{-O} option since
4480 it can result in incorrect output for programs which depend on
4481 an exact implementation of IEEE or ISO rules/specifications for
4484 The default is @option{-ftrapping-math}.
4486 @item -frounding-math
4487 @opindex frounding-math
4488 Disable transformations and optimizations that assume default floating
4489 point rounding behavior. This is round-to-zero for all floating point
4490 to integer conversions, and round-to-nearest for all other arithmetic
4491 truncations. This option should be specified for programs that change
4492 the FP rounding mode dynamically, or that may be executed with a
4493 non-default rounding mode. This option disables constant folding of
4494 floating point expressions at compile-time (which may be affected by
4495 rounding mode) and arithmetic transformations that are unsafe in the
4496 presence of sign-dependent rounding modes.
4498 The default is @option{-fno-rounding-math}.
4500 This option is experimental and does not currently guarantee to
4501 disable all GCC optimizations that are affected by rounding mode.
4502 Future versions of gcc may provide finer control of this setting
4503 using C99's @code{FENV_ACCESS} pragma. This command line option
4504 will be used to specify the default state for @code{FENV_ACCESS}.
4506 @item -fsignaling-nans
4507 @opindex fsignaling-nans
4508 Compile code assuming that IEEE signaling NaNs may generate user-visible
4509 traps during floating-point operations. Setting this option disables
4510 optimizations that may change the number of exceptions visible with
4511 signaling NaNs. This option implies @option{-ftrapping-math}.
4513 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4516 The default is @option{-fno-signaling-nans}.
4518 This option is experimental and does not currently guarantee to
4519 disable all GCC optimizations that affect signaling NaN behavior.
4521 @item -fsingle-precision-constant
4522 @opindex fsingle-precision-constant
4523 Treat floating point constant as single precision constant instead of
4524 implicitly converting it to double precision constant.
4529 The following options control optimizations that may improve
4530 performance, but are not enabled by any @option{-O} options. This
4531 section includes experimental options that may produce broken code.
4534 @item -fbranch-probabilities
4535 @opindex fbranch-probabilities
4536 After running a program compiled with @option{-fprofile-arcs}
4537 (@pxref{Debugging Options,, Options for Debugging Your Program or
4538 @command{gcc}}), you can compile it a second time using
4539 @option{-fbranch-probabilities}, to improve optimizations based on
4540 the number of times each branch was taken. When the program
4541 compiled with @option{-fprofile-arcs} exits it saves arc execution
4542 counts to a file called @file{@var{sourcename}.gcda} for each source
4543 file The information in this data file is very dependent on the
4544 structure of the generated code, so you must use the same source code
4545 and the same optimization options for both compilations.
4547 With @option{-fbranch-probabilities}, GCC puts a
4548 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4549 These can be used to improve optimization. Currently, they are only
4550 used in one place: in @file{reorg.c}, instead of guessing which path a
4551 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4552 exactly determine which path is taken more often.
4554 @item -fprofile-values
4555 @opindex fprofile-values
4556 If combined with @option{-fprofile-arcs}, it adds code so that some
4557 data about values of expressions in the program is gathered.
4559 With @option{-fbranch-probabilities}, it reads back the data gathered
4560 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
4561 notes to instructions for their later usage in optimizations.
4565 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
4566 a code to gather information about values of expressions.
4568 With @option{-fbranch-probabilities}, it reads back the data gathered
4569 and actually performs the optimizations based on them.
4570 Currently the optimizations include specialization of division operation
4571 using the knowledge about the value of the denominator.
4575 Use a graph coloring register allocator. Currently this option is meant
4576 for testing, so we are interested to hear about miscompilations with
4581 Perform tail duplication to enlarge superblock size. This transformation
4582 simplifies the control flow of the function allowing other optimizations to do
4585 @item -funit-at-a-time
4586 @opindex funit-at-a-time
4587 Parse the whole compilation unit before starting to produce code.
4588 This allows some extra optimizations to take place but consumes more
4591 @item -funroll-loops
4592 @opindex funroll-loops
4593 Unroll loops whose number of iterations can be determined at compile time or
4594 upon entry to the loop. @option{-funroll-loops} implies
4595 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4596 (i.e. complete removal of loops with small constant number of iterations).
4597 This option makes code larger, and may or may not make it run faster.
4599 @item -funroll-all-loops
4600 @opindex funroll-all-loops
4601 Unroll all loops, even if their number of iterations is uncertain when
4602 the loop is entered. This usually makes programs run more slowly.
4603 @option{-funroll-all-loops} implies the same options as
4604 @option{-funroll-loops}.
4607 @opindex fpeel-loops
4608 Peels the loops for that there is enough information that they do not
4609 roll much (from profile feedback). It also turns on complete loop peeling
4610 (i.e. complete removal of loops with small constant number of iterations).
4612 @item -funswitch-loops
4613 @opindex funswitch-loops
4614 Move branches with loop invariant conditions out of the loop, with duplicates
4615 of the loop on both branches (modified according to result of the condition).
4617 @item -fold-unroll-loops
4618 @opindex fold-unroll-loops
4619 Unroll loops whose number of iterations can be determined at compile
4620 time or upon entry to the loop, using the old loop unroller whose loop
4621 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4622 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4623 option makes code larger, and may or may not make it run faster.
4625 @item -fold-unroll-all-loops
4626 @opindex fold-unroll-all-loops
4627 Unroll all loops, even if their number of iterations is uncertain when
4628 the loop is entered. This is done using the old loop unroller whose loop
4629 recognition is based on notes from frontend. This usually makes programs run more slowly.
4630 @option{-fold-unroll-all-loops} implies the same options as
4631 @option{-fold-unroll-loops}.
4633 @item -funswitch-loops
4634 @opindex funswitch-loops
4635 Move branches with loop invariant conditions out of the loop, with duplicates
4636 of the loop on both branches (modified according to result of the condition).
4638 @item -funswitch-loops
4639 @opindex funswitch-loops
4640 Move branches with loop invariant conditions out of the loop, with duplicates
4641 of the loop on both branches (modified according to result of the condition).
4643 @item -fprefetch-loop-arrays
4644 @opindex fprefetch-loop-arrays
4645 If supported by the target machine, generate instructions to prefetch
4646 memory to improve the performance of loops that access large arrays.
4648 Disabled at level @option{-Os}.
4650 @item -ffunction-sections
4651 @itemx -fdata-sections
4652 @opindex ffunction-sections
4653 @opindex fdata-sections
4654 Place each function or data item into its own section in the output
4655 file if the target supports arbitrary sections. The name of the
4656 function or the name of the data item determines the section's name
4659 Use these options on systems where the linker can perform optimizations
4660 to improve locality of reference in the instruction space. Most systems
4661 using the ELF object format and SPARC processors running Solaris 2 have
4662 linkers with such optimizations. AIX may have these optimizations in
4665 Only use these options when there are significant benefits from doing
4666 so. When you specify these options, the assembler and linker will
4667 create larger object and executable files and will also be slower.
4668 You will not be able to use @code{gprof} on all systems if you
4669 specify this option and you may have problems with debugging if
4670 you specify both this option and @option{-g}.
4672 @item -fbranch-target-load-optimize
4673 @opindex fbranch-target-load-optimize
4674 Perform branch target register load optimization before prologue / epilogue
4676 The use of target registers can typically be exposed only during reload,
4677 thus hoisting loads out of loops and doing inter-block scheduling needs
4678 a separate optimization pass.
4680 @item -fbranch-target-load-optimize2
4681 @opindex fbranch-target-load-optimize2
4682 Perform branch target register load optimization after prologue / epilogue
4685 @item --param @var{name}=@var{value}
4687 In some places, GCC uses various constants to control the amount of
4688 optimization that is done. For example, GCC will not inline functions
4689 that contain more that a certain number of instructions. You can
4690 control some of these constants on the command-line using the
4691 @option{--param} option.
4693 In each case, the @var{value} is an integer. The allowable choices for
4694 @var{name} are given in the following table:
4697 @item max-crossjump-edges
4698 The maximum number of incoming edges to consider for crossjumping.
4699 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
4700 the number of edges incoming to each block. Increasing values mean
4701 more aggressive optimization, making the compile time increase with
4702 probably small improvement in executable size.
4704 @item max-delay-slot-insn-search
4705 The maximum number of instructions to consider when looking for an
4706 instruction to fill a delay slot. If more than this arbitrary number of
4707 instructions is searched, the time savings from filling the delay slot
4708 will be minimal so stop searching. Increasing values mean more
4709 aggressive optimization, making the compile time increase with probably
4710 small improvement in executable run time.
4712 @item max-delay-slot-live-search
4713 When trying to fill delay slots, the maximum number of instructions to
4714 consider when searching for a block with valid live register
4715 information. Increasing this arbitrarily chosen value means more
4716 aggressive optimization, increasing the compile time. This parameter
4717 should be removed when the delay slot code is rewritten to maintain the
4720 @item max-gcse-memory
4721 The approximate maximum amount of memory that will be allocated in
4722 order to perform the global common subexpression elimination
4723 optimization. If more memory than specified is required, the
4724 optimization will not be done.
4726 @item max-gcse-passes
4727 The maximum number of passes of GCSE to run.
4729 @item max-pending-list-length
4730 The maximum number of pending dependencies scheduling will allow
4731 before flushing the current state and starting over. Large functions
4732 with few branches or calls can create excessively large lists which
4733 needlessly consume memory and resources.
4735 @item max-inline-insns-single
4736 Several parameters control the tree inliner used in gcc.
4737 This number sets the maximum number of instructions (counted in gcc's
4738 internal representation) in a single function that the tree inliner
4739 will consider for inlining. This only affects functions declared
4740 inline and methods implemented in a class declaration (C++).
4741 The default value is 500.
4743 @item max-inline-insns-auto
4744 When you use @option{-finline-functions} (included in @option{-O3}),
4745 a lot of functions that would otherwise not be considered for inlining
4746 by the compiler will be investigated. To those functions, a different
4747 (more restrictive) limit compared to functions declared inline can
4749 The default value is 150.
4751 @item large-function-insns
4752 The limit specifying really large functions. For functions greater than this
4753 limit inlining is constrained by @option{--param large-function-growth}.
4754 This parameter is useful primarily to avoid extreme compilation time caused by non-linear
4755 algorithms used by the backend.
4756 This parameter is ignored when @option{-funit-at-a-time} is not used.
4757 The default value is 30000.
4759 @item large-function-growth
4760 Specifies maximal growth of large functtion caused by inlining in percents.
4761 This parameter is ignored when @option{-funit-at-a-time} is not used.
4762 The default value is 200.
4764 @item inline-unit-growth
4765 Specifies maximal overall growth of the compilation unit caused by inlining.
4766 This parameter is ignored when @option{-funit-at-a-time} is not used.
4767 The default value is 150.
4769 @item max-inline-insns-rtl
4770 For languages that use the RTL inliner (this happens at a later stage
4771 than tree inlining), you can set the maximum allowable size (counted
4772 in RTL instructions) for the RTL inliner with this parameter.
4773 The default value is 600.
4775 @item max-unrolled-insns
4776 The maximum number of instructions that a loop should have if that loop
4777 is unrolled, and if the loop is unrolled, it determines how many times
4778 the loop code is unrolled.
4780 @item max-average-unrolled-insns
4781 The maximum number of instructions biased by probabilities of their execution
4782 that a loop should have if that loop is unrolled, and if the loop is unrolled,
4783 it determines how many times the loop code is unrolled.
4785 @item max-unroll-times
4786 The maximum number of unrollings of a single loop.
4788 @item max-peeled-insns
4789 The maximum number of instructions that a loop should have if that loop
4790 is peeled, and if the loop is peeled, it determines how many times
4791 the loop code is peeled.
4793 @item max-peel-times
4794 The maximum number of peelings of a single loop.
4796 @item max-completely-peeled-insns
4797 The maximum number of insns of a completely peeled loop.
4799 @item max-completely-peel-times
4800 The maximum number of iterations of a loop to be suitable for complete peeling.
4802 @item max-unswitch-insns
4803 The maximum number of insns of an unswitched loop.
4805 @item max-unswitch-level
4806 The maximum number of branches unswitched in a single loop.
4808 @item hot-bb-count-fraction
4809 Select fraction of the maximal count of repetitions of basic block in program
4810 given basic block needs to have to be considered hot.
4812 @item hot-bb-frequency-fraction
4813 Select fraction of the maximal frequency of executions of basic block in
4814 function given basic block needs to have to be considered hot
4816 @item tracer-dynamic-coverage
4817 @itemx tracer-dynamic-coverage-feedback
4819 This value is used to limit superblock formation once the given percentage of
4820 executed instructions is covered. This limits unnecessary code size
4823 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4824 feedback is available. The real profiles (as opposed to statically estimated
4825 ones) are much less balanced allowing the threshold to be larger value.
4827 @item tracer-max-code-growth
4828 Stop tail duplication once code growth has reached given percentage. This is
4829 rather hokey argument, as most of the duplicates will be eliminated later in
4830 cross jumping, so it may be set to much higher values than is the desired code
4833 @item tracer-min-branch-ratio
4835 Stop reverse growth when the reverse probability of best edge is less than this
4836 threshold (in percent).
4838 @item tracer-min-branch-ratio
4839 @itemx tracer-min-branch-ratio-feedback
4841 Stop forward growth if the best edge do have probability lower than this
4844 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
4845 compilation for profile feedback and one for compilation without. The value
4846 for compilation with profile feedback needs to be more conservative (higher) in
4847 order to make tracer effective.
4849 @item max-cse-path-length
4851 Maximum number of basic blocks on path that cse considers.
4853 @item ggc-min-expand
4855 GCC uses a garbage collector to manage its own memory allocation. This
4856 parameter specifies the minimum percentage by which the garbage
4857 collector's heap should be allowed to expand between collections.
4858 Tuning this may improve compilation speed; it has no effect on code
4861 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
4862 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
4863 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
4864 GCC is not able to calculate RAM on a particular platform, the lower
4865 bound of 30% is used. Setting this parameter and
4866 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4867 every opportunity. This is extremely slow, but can be useful for
4870 @item ggc-min-heapsize
4872 Minimum size of the garbage collector's heap before it begins bothering
4873 to collect garbage. The first collection occurs after the heap expands
4874 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4875 tuning this may improve compilation speed, and has no effect on code
4878 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
4879 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
4880 available, the notion of "RAM" is the smallest of actual RAM,
4881 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
4882 RAM on a particular platform, the lower bound is used. Setting this
4883 parameter very large effectively disables garbage collection. Setting
4884 this parameter and @option{ggc-min-expand} to zero causes a full
4885 collection to occur at every opportunity.
4887 @item reorder-blocks-duplicate
4888 @itemx reorder-blocks-duplicate-feedback
4890 Used by basic block reordering pass to decide whether to use unconditional
4891 branch or duplicate the code on its destination. Code is duplicated when its
4892 estimated size is smaller than this value multiplied by the estimated size of
4893 unconditional jump in the hot spots of the program.
4895 The @option{reorder-block-duplicate-feedback} is used only when profile
4896 feedback is available and may be set to higher values than
4897 @option{reorder-block-duplicate} since information about the hot spots is more
4902 @node Preprocessor Options
4903 @section Options Controlling the Preprocessor
4904 @cindex preprocessor options
4905 @cindex options, preprocessor
4907 These options control the C preprocessor, which is run on each C source
4908 file before actual compilation.
4910 If you use the @option{-E} option, nothing is done except preprocessing.
4911 Some of these options make sense only together with @option{-E} because
4912 they cause the preprocessor output to be unsuitable for actual
4917 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4918 and pass @var{option} directly through to the preprocessor. If
4919 @var{option} contains commas, it is split into multiple options at the
4920 commas. However, many options are modified, translated or interpreted
4921 by the compiler driver before being passed to the preprocessor, and
4922 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4923 interface is undocumented and subject to change, so whenever possible
4924 you should avoid using @option{-Wp} and let the driver handle the
4927 @item -Xpreprocessor @var{option}
4928 @opindex preprocessor
4929 Pass @var{option} as an option to the preprocessor. You can use this to
4930 supply system-specific preprocessor options which GCC does not know how to
4933 If you want to pass an option that takes an argument, you must use
4934 @option{-Xpreprocessor} twice, once for the option and once for the argument.
4937 @include cppopts.texi
4939 @node Assembler Options
4940 @section Passing Options to the Assembler
4942 @c prevent bad page break with this line
4943 You can pass options to the assembler.
4946 @item -Wa,@var{option}
4948 Pass @var{option} as an option to the assembler. If @var{option}
4949 contains commas, it is split into multiple options at the commas.
4951 @item -Xassembler @var{option}
4953 Pass @var{option} as an option to the assembler. You can use this to
4954 supply system-specific assembler options which GCC does not know how to
4957 If you want to pass an option that takes an argument, you must use
4958 @option{-Xassembler} twice, once for the option and once for the argument.
4963 @section Options for Linking
4964 @cindex link options
4965 @cindex options, linking
4967 These options come into play when the compiler links object files into
4968 an executable output file. They are meaningless if the compiler is
4969 not doing a link step.
4973 @item @var{object-file-name}
4974 A file name that does not end in a special recognized suffix is
4975 considered to name an object file or library. (Object files are
4976 distinguished from libraries by the linker according to the file
4977 contents.) If linking is done, these object files are used as input
4986 If any of these options is used, then the linker is not run, and
4987 object file names should not be used as arguments. @xref{Overall
4991 @item -l@var{library}
4992 @itemx -l @var{library}
4994 Search the library named @var{library} when linking. (The second
4995 alternative with the library as a separate argument is only for
4996 POSIX compliance and is not recommended.)
4998 It makes a difference where in the command you write this option; the
4999 linker searches and processes libraries and object files in the order they
5000 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
5001 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
5002 to functions in @samp{z}, those functions may not be loaded.
5004 The linker searches a standard list of directories for the library,
5005 which is actually a file named @file{lib@var{library}.a}. The linker
5006 then uses this file as if it had been specified precisely by name.
5008 The directories searched include several standard system directories
5009 plus any that you specify with @option{-L}.
5011 Normally the files found this way are library files---archive files
5012 whose members are object files. The linker handles an archive file by
5013 scanning through it for members which define symbols that have so far
5014 been referenced but not defined. But if the file that is found is an
5015 ordinary object file, it is linked in the usual fashion. The only
5016 difference between using an @option{-l} option and specifying a file name
5017 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
5018 and searches several directories.
5022 You need this special case of the @option{-l} option in order to
5023 link an Objective-C program.
5026 @opindex nostartfiles
5027 Do not use the standard system startup files when linking.
5028 The standard system libraries are used normally, unless @option{-nostdlib}
5029 or @option{-nodefaultlibs} is used.
5031 @item -nodefaultlibs
5032 @opindex nodefaultlibs
5033 Do not use the standard system libraries when linking.
5034 Only the libraries you specify will be passed to the linker.
5035 The standard startup files are used normally, unless @option{-nostartfiles}
5036 is used. The compiler may generate calls to memcmp, memset, and memcpy
5037 for System V (and ISO C) environments or to bcopy and bzero for
5038 BSD environments. These entries are usually resolved by entries in
5039 libc. These entry points should be supplied through some other
5040 mechanism when this option is specified.
5044 Do not use the standard system startup files or libraries when linking.
5045 No startup files and only the libraries you specify will be passed to
5046 the linker. The compiler may generate calls to memcmp, memset, and memcpy
5047 for System V (and ISO C) environments or to bcopy and bzero for
5048 BSD environments. These entries are usually resolved by entries in
5049 libc. These entry points should be supplied through some other
5050 mechanism when this option is specified.
5052 @cindex @option{-lgcc}, use with @option{-nostdlib}
5053 @cindex @option{-nostdlib} and unresolved references
5054 @cindex unresolved references and @option{-nostdlib}
5055 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5056 @cindex @option{-nodefaultlibs} and unresolved references
5057 @cindex unresolved references and @option{-nodefaultlibs}
5058 One of the standard libraries bypassed by @option{-nostdlib} and
5059 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5060 that GCC uses to overcome shortcomings of particular machines, or special
5061 needs for some languages.
5062 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5063 Collection (GCC) Internals},
5064 for more discussion of @file{libgcc.a}.)
5065 In most cases, you need @file{libgcc.a} even when you want to avoid
5066 other standard libraries. In other words, when you specify @option{-nostdlib}
5067 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5068 This ensures that you have no unresolved references to internal GCC
5069 library subroutines. (For example, @samp{__main}, used to ensure C++
5070 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5071 GNU Compiler Collection (GCC) Internals}.)
5075 Produce a position independent executable on targets which support it.
5076 For predictable results, you must also specify the same set of options
5077 that were used to generate code (@option{-fpie}, @option{-fPIE},
5078 or model suboptions) when you specify this option.
5082 Remove all symbol table and relocation information from the executable.
5086 On systems that support dynamic linking, this prevents linking with the shared
5087 libraries. On other systems, this option has no effect.
5091 Produce a shared object which can then be linked with other objects to
5092 form an executable. Not all systems support this option. For predictable
5093 results, you must also specify the same set of options that were used to
5094 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5095 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5096 needs to build supplementary stub code for constructors to work. On
5097 multi-libbed systems, @samp{gcc -shared} must select the correct support
5098 libraries to link against. Failing to supply the correct flags may lead
5099 to subtle defects. Supplying them in cases where they are not necessary
5102 @item -shared-libgcc
5103 @itemx -static-libgcc
5104 @opindex shared-libgcc
5105 @opindex static-libgcc
5106 On systems that provide @file{libgcc} as a shared library, these options
5107 force the use of either the shared or static version respectively.
5108 If no shared version of @file{libgcc} was built when the compiler was
5109 configured, these options have no effect.
5111 There are several situations in which an application should use the
5112 shared @file{libgcc} instead of the static version. The most common
5113 of these is when the application wishes to throw and catch exceptions
5114 across different shared libraries. In that case, each of the libraries
5115 as well as the application itself should use the shared @file{libgcc}.
5117 Therefore, the G++ and GCJ drivers automatically add
5118 @option{-shared-libgcc} whenever you build a shared library or a main
5119 executable, because C++ and Java programs typically use exceptions, so
5120 this is the right thing to do.
5122 If, instead, you use the GCC driver to create shared libraries, you may
5123 find that they will not always be linked with the shared @file{libgcc}.
5124 If GCC finds, at its configuration time, that you have a GNU linker that
5125 does not support option @option{--eh-frame-hdr}, it will link the shared
5126 version of @file{libgcc} into shared libraries by default. Otherwise,
5127 it will take advantage of the linker and optimize away the linking with
5128 the shared version of @file{libgcc}, linking with the static version of
5129 libgcc by default. This allows exceptions to propagate through such
5130 shared libraries, without incurring relocation costs at library load
5133 However, if a library or main executable is supposed to throw or catch
5134 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5135 for the languages used in the program, or using the option
5136 @option{-shared-libgcc}, such that it is linked with the shared
5141 Bind references to global symbols when building a shared object. Warn
5142 about any unresolved references (unless overridden by the link editor
5143 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5146 @item -Xlinker @var{option}
5148 Pass @var{option} as an option to the linker. You can use this to
5149 supply system-specific linker options which GCC does not know how to
5152 If you want to pass an option that takes an argument, you must use
5153 @option{-Xlinker} twice, once for the option and once for the argument.
5154 For example, to pass @option{-assert definitions}, you must write
5155 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5156 @option{-Xlinker "-assert definitions"}, because this passes the entire
5157 string as a single argument, which is not what the linker expects.
5159 @item -Wl,@var{option}
5161 Pass @var{option} as an option to the linker. If @var{option} contains
5162 commas, it is split into multiple options at the commas.
5164 @item -u @var{symbol}
5166 Pretend the symbol @var{symbol} is undefined, to force linking of
5167 library modules to define it. You can use @option{-u} multiple times with
5168 different symbols to force loading of additional library modules.
5171 @node Directory Options
5172 @section Options for Directory Search
5173 @cindex directory options
5174 @cindex options, directory search
5177 These options specify directories to search for header files, for
5178 libraries and for parts of the compiler:
5183 Add the directory @var{dir} to the head of the list of directories to be
5184 searched for header files. This can be used to override a system header
5185 file, substituting your own version, since these directories are
5186 searched before the system header file directories. However, you should
5187 not use this option to add directories that contain vendor-supplied
5188 system header files (use @option{-isystem} for that). If you use more than
5189 one @option{-I} option, the directories are scanned in left-to-right
5190 order; the standard system directories come after.
5192 If a standard system include directory, or a directory specified with
5193 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5194 option will be ignored. The directory will still be searched but as a
5195 system directory at its normal position in the system include chain.
5196 This is to ensure that GCC's procedure to fix buggy system headers and
5197 the ordering for the include_next directive are not inadvertently changed.
5198 If you really need to change the search order for system directories,
5199 use the @option{-nostdinc} and/or @option{-isystem} options.
5203 Any directories you specify with @option{-I} options before the @option{-I-}
5204 option are searched only for the case of @samp{#include "@var{file}"};
5205 they are not searched for @samp{#include <@var{file}>}.
5207 If additional directories are specified with @option{-I} options after
5208 the @option{-I-}, these directories are searched for all @samp{#include}
5209 directives. (Ordinarily @emph{all} @option{-I} directories are used
5212 In addition, the @option{-I-} option inhibits the use of the current
5213 directory (where the current input file came from) as the first search
5214 directory for @samp{#include "@var{file}"}. There is no way to
5215 override this effect of @option{-I-}. With @option{-I.} you can specify
5216 searching the directory which was current when the compiler was
5217 invoked. That is not exactly the same as what the preprocessor does
5218 by default, but it is often satisfactory.
5220 @option{-I-} does not inhibit the use of the standard system directories
5221 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5226 Add directory @var{dir} to the list of directories to be searched
5229 @item -B@var{prefix}
5231 This option specifies where to find the executables, libraries,
5232 include files, and data files of the compiler itself.
5234 The compiler driver program runs one or more of the subprograms
5235 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5236 @var{prefix} as a prefix for each program it tries to run, both with and
5237 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5239 For each subprogram to be run, the compiler driver first tries the
5240 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5241 was not specified, the driver tries two standard prefixes, which are
5242 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5243 those results in a file name that is found, the unmodified program
5244 name is searched for using the directories specified in your
5245 @env{PATH} environment variable.
5247 The compiler will check to see if the path provided by the @option{-B}
5248 refers to a directory, and if necessary it will add a directory
5249 separator character at the end of the path.
5251 @option{-B} prefixes that effectively specify directory names also apply
5252 to libraries in the linker, because the compiler translates these
5253 options into @option{-L} options for the linker. They also apply to
5254 includes files in the preprocessor, because the compiler translates these
5255 options into @option{-isystem} options for the preprocessor. In this case,
5256 the compiler appends @samp{include} to the prefix.
5258 The run-time support file @file{libgcc.a} can also be searched for using
5259 the @option{-B} prefix, if needed. If it is not found there, the two
5260 standard prefixes above are tried, and that is all. The file is left
5261 out of the link if it is not found by those means.
5263 Another way to specify a prefix much like the @option{-B} prefix is to use
5264 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5267 As a special kludge, if the path provided by @option{-B} is
5268 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5269 9, then it will be replaced by @file{[dir/]include}. This is to help
5270 with boot-strapping the compiler.
5272 @item -specs=@var{file}
5274 Process @var{file} after the compiler reads in the standard @file{specs}
5275 file, in order to override the defaults that the @file{gcc} driver
5276 program uses when determining what switches to pass to @file{cc1},
5277 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5278 @option{-specs=@var{file}} can be specified on the command line, and they
5279 are processed in order, from left to right.
5285 @section Specifying subprocesses and the switches to pass to them
5288 @command{gcc} is a driver program. It performs its job by invoking a
5289 sequence of other programs to do the work of compiling, assembling and
5290 linking. GCC interprets its command-line parameters and uses these to
5291 deduce which programs it should invoke, and which command-line options
5292 it ought to place on their command lines. This behavior is controlled
5293 by @dfn{spec strings}. In most cases there is one spec string for each
5294 program that GCC can invoke, but a few programs have multiple spec
5295 strings to control their behavior. The spec strings built into GCC can
5296 be overridden by using the @option{-specs=} command-line switch to specify
5299 @dfn{Spec files} are plaintext files that are used to construct spec
5300 strings. They consist of a sequence of directives separated by blank
5301 lines. The type of directive is determined by the first non-whitespace
5302 character on the line and it can be one of the following:
5305 @item %@var{command}
5306 Issues a @var{command} to the spec file processor. The commands that can
5310 @item %include <@var{file}>
5312 Search for @var{file} and insert its text at the current point in the
5315 @item %include_noerr <@var{file}>
5316 @cindex %include_noerr
5317 Just like @samp{%include}, but do not generate an error message if the include
5318 file cannot be found.
5320 @item %rename @var{old_name} @var{new_name}
5322 Rename the spec string @var{old_name} to @var{new_name}.
5326 @item *[@var{spec_name}]:
5327 This tells the compiler to create, override or delete the named spec
5328 string. All lines after this directive up to the next directive or
5329 blank line are considered to be the text for the spec string. If this
5330 results in an empty string then the spec will be deleted. (Or, if the
5331 spec did not exist, then nothing will happened.) Otherwise, if the spec
5332 does not currently exist a new spec will be created. If the spec does
5333 exist then its contents will be overridden by the text of this
5334 directive, unless the first character of that text is the @samp{+}
5335 character, in which case the text will be appended to the spec.
5337 @item [@var{suffix}]:
5338 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5339 and up to the next directive or blank line are considered to make up the
5340 spec string for the indicated suffix. When the compiler encounters an
5341 input file with the named suffix, it will processes the spec string in
5342 order to work out how to compile that file. For example:
5349 This says that any input file whose name ends in @samp{.ZZ} should be
5350 passed to the program @samp{z-compile}, which should be invoked with the
5351 command-line switch @option{-input} and with the result of performing the
5352 @samp{%i} substitution. (See below.)
5354 As an alternative to providing a spec string, the text that follows a
5355 suffix directive can be one of the following:
5358 @item @@@var{language}
5359 This says that the suffix is an alias for a known @var{language}. This is
5360 similar to using the @option{-x} command-line switch to GCC to specify a
5361 language explicitly. For example:
5368 Says that .ZZ files are, in fact, C++ source files.
5371 This causes an error messages saying:
5374 @var{name} compiler not installed on this system.
5378 GCC already has an extensive list of suffixes built into it.
5379 This directive will add an entry to the end of the list of suffixes, but
5380 since the list is searched from the end backwards, it is effectively
5381 possible to override earlier entries using this technique.
5385 GCC has the following spec strings built into it. Spec files can
5386 override these strings or create their own. Note that individual
5387 targets can also add their own spec strings to this list.
5390 asm Options to pass to the assembler
5391 asm_final Options to pass to the assembler post-processor
5392 cpp Options to pass to the C preprocessor
5393 cc1 Options to pass to the C compiler
5394 cc1plus Options to pass to the C++ compiler
5395 endfile Object files to include at the end of the link
5396 link Options to pass to the linker
5397 lib Libraries to include on the command line to the linker
5398 libgcc Decides which GCC support library to pass to the linker
5399 linker Sets the name of the linker
5400 predefines Defines to be passed to the C preprocessor
5401 signed_char Defines to pass to CPP to say whether @code{char} is signed
5403 startfile Object files to include at the start of the link
5406 Here is a small example of a spec file:
5412 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5415 This example renames the spec called @samp{lib} to @samp{old_lib} and
5416 then overrides the previous definition of @samp{lib} with a new one.
5417 The new definition adds in some extra command-line options before
5418 including the text of the old definition.
5420 @dfn{Spec strings} are a list of command-line options to be passed to their
5421 corresponding program. In addition, the spec strings can contain
5422 @samp{%}-prefixed sequences to substitute variable text or to
5423 conditionally insert text into the command line. Using these constructs
5424 it is possible to generate quite complex command lines.
5426 Here is a table of all defined @samp{%}-sequences for spec
5427 strings. Note that spaces are not generated automatically around the
5428 results of expanding these sequences. Therefore you can concatenate them
5429 together or combine them with constant text in a single argument.
5433 Substitute one @samp{%} into the program name or argument.
5436 Substitute the name of the input file being processed.
5439 Substitute the basename of the input file being processed.
5440 This is the substring up to (and not including) the last period
5441 and not including the directory.
5444 This is the same as @samp{%b}, but include the file suffix (text after
5448 Marks the argument containing or following the @samp{%d} as a
5449 temporary file name, so that that file will be deleted if GCC exits
5450 successfully. Unlike @samp{%g}, this contributes no text to the
5453 @item %g@var{suffix}
5454 Substitute a file name that has suffix @var{suffix} and is chosen
5455 once per compilation, and mark the argument in the same way as
5456 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5457 name is now chosen in a way that is hard to predict even when previously
5458 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5459 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5460 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5461 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5462 was simply substituted with a file name chosen once per compilation,
5463 without regard to any appended suffix (which was therefore treated
5464 just like ordinary text), making such attacks more likely to succeed.
5466 @item %u@var{suffix}
5467 Like @samp{%g}, but generates a new temporary file name even if
5468 @samp{%u@var{suffix}} was already seen.
5470 @item %U@var{suffix}
5471 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5472 new one if there is no such last file name. In the absence of any
5473 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5474 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5475 would involve the generation of two distinct file names, one
5476 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5477 simply substituted with a file name chosen for the previous @samp{%u},
5478 without regard to any appended suffix.
5480 @item %j@var{suffix}
5481 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5482 writable, and if save-temps is off; otherwise, substitute the name
5483 of a temporary file, just like @samp{%u}. This temporary file is not
5484 meant for communication between processes, but rather as a junk
5487 @item %|@var{suffix}
5488 @itemx %m@var{suffix}
5489 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5490 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5491 all. These are the two most common ways to instruct a program that it
5492 should read from standard input or write to standard output. If you
5493 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5494 construct: see for example @file{f/lang-specs.h}.
5496 @item %.@var{SUFFIX}
5497 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5498 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5499 terminated by the next space or %.
5502 Marks the argument containing or following the @samp{%w} as the
5503 designated output file of this compilation. This puts the argument
5504 into the sequence of arguments that @samp{%o} will substitute later.
5507 Substitutes the names of all the output files, with spaces
5508 automatically placed around them. You should write spaces
5509 around the @samp{%o} as well or the results are undefined.
5510 @samp{%o} is for use in the specs for running the linker.
5511 Input files whose names have no recognized suffix are not compiled
5512 at all, but they are included among the output files, so they will
5516 Substitutes the suffix for object files. Note that this is
5517 handled specially when it immediately follows @samp{%g, %u, or %U},
5518 because of the need for those to form complete file names. The
5519 handling is such that @samp{%O} is treated exactly as if it had already
5520 been substituted, except that @samp{%g, %u, and %U} do not currently
5521 support additional @var{suffix} characters following @samp{%O} as they would
5522 following, for example, @samp{.o}.
5525 Substitutes the standard macro predefinitions for the
5526 current target machine. Use this when running @code{cpp}.
5529 Like @samp{%p}, but puts @samp{__} before and after the name of each
5530 predefined macro, except for macros that start with @samp{__} or with
5531 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5535 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5536 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5537 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5541 Current argument is the name of a library or startup file of some sort.
5542 Search for that file in a standard list of directories and substitute
5543 the full name found.
5546 Print @var{str} as an error message. @var{str} is terminated by a newline.
5547 Use this when inconsistent options are detected.
5550 Substitute the contents of spec string @var{name} at this point.
5553 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5555 @item %x@{@var{option}@}
5556 Accumulate an option for @samp{%X}.
5559 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5563 Output the accumulated assembler options specified by @option{-Wa}.
5566 Output the accumulated preprocessor options specified by @option{-Wp}.
5569 Process the @code{asm} spec. This is used to compute the
5570 switches to be passed to the assembler.
5573 Process the @code{asm_final} spec. This is a spec string for
5574 passing switches to an assembler post-processor, if such a program is
5578 Process the @code{link} spec. This is the spec for computing the
5579 command line passed to the linker. Typically it will make use of the
5580 @samp{%L %G %S %D and %E} sequences.
5583 Dump out a @option{-L} option for each directory that GCC believes might
5584 contain startup files. If the target supports multilibs then the
5585 current multilib directory will be prepended to each of these paths.
5588 Output the multilib directory with directory separators replaced with
5589 @samp{_}. If multilib directories are not set, or the multilib directory is
5590 @file{.} then this option emits nothing.
5593 Process the @code{lib} spec. This is a spec string for deciding which
5594 libraries should be included on the command line to the linker.
5597 Process the @code{libgcc} spec. This is a spec string for deciding
5598 which GCC support library should be included on the command line to the linker.
5601 Process the @code{startfile} spec. This is a spec for deciding which
5602 object files should be the first ones passed to the linker. Typically
5603 this might be a file named @file{crt0.o}.
5606 Process the @code{endfile} spec. This is a spec string that specifies
5607 the last object files that will be passed to the linker.
5610 Process the @code{cpp} spec. This is used to construct the arguments
5611 to be passed to the C preprocessor.
5614 Process the @code{signed_char} spec. This is intended to be used
5615 to tell cpp whether a char is signed. It typically has the definition:
5617 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
5621 Process the @code{cc1} spec. This is used to construct the options to be
5622 passed to the actual C compiler (@samp{cc1}).
5625 Process the @code{cc1plus} spec. This is used to construct the options to be
5626 passed to the actual C++ compiler (@samp{cc1plus}).
5629 Substitute the variable part of a matched option. See below.
5630 Note that each comma in the substituted string is replaced by
5634 Remove all occurrences of @code{-S} from the command line. Note---this
5635 command is position dependent. @samp{%} commands in the spec string
5636 before this one will see @code{-S}, @samp{%} commands in the spec string
5637 after this one will not.
5639 @item %:@var{function}(@var{args})
5640 Call the named function @var{function}, passing it @var{args}.
5641 @var{args} is first processed as a nested spec string, then split
5642 into an argument vector in the usual fashion. The function returns
5643 a string which is processed as if it had appeared literally as part
5644 of the current spec.
5646 The following built-in spec functions are provided:
5649 @item @code{if-exists}
5650 The @code{if-exists} spec function takes one argument, an absolute
5651 pathname to a file. If the file exists, @code{if-exists} returns the
5652 pathname. Here is a small example of its usage:
5656 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5659 @item @code{if-exists-else}
5660 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5661 spec function, except that it takes two arguments. The first argument is
5662 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5663 returns the pathname. If it does not exist, it returns the second argument.
5664 This way, @code{if-exists-else} can be used to select one file or another,
5665 based on the existence of the first. Here is a small example of its usage:
5669 crt0%O%s %:if-exists(crti%O%s) \
5670 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5675 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5676 If that switch was not specified, this substitutes nothing. Note that
5677 the leading dash is omitted when specifying this option, and it is
5678 automatically inserted if the substitution is performed. Thus the spec
5679 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5680 and would output the command line option @option{-foo}.
5682 @item %W@{@code{S}@}
5683 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5686 @item %@{@code{S}*@}
5687 Substitutes all the switches specified to GCC whose names start
5688 with @code{-S}, but which also take an argument. This is used for
5689 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5690 GCC considers @option{-o foo} as being
5691 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5692 text, including the space. Thus two arguments would be generated.
5694 @item %@{@code{S}*&@code{T}*@}
5695 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5696 (the order of @code{S} and @code{T} in the spec is not significant).
5697 There can be any number of ampersand-separated variables; for each the
5698 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5700 @item %@{@code{S}:@code{X}@}
5701 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
5703 @item %@{!@code{S}:@code{X}@}
5704 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
5706 @item %@{@code{S}*:@code{X}@}
5707 Substitutes @code{X} if one or more switches whose names start with
5708 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
5709 once, no matter how many such switches appeared. However, if @code{%*}
5710 appears somewhere in @code{X}, then @code{X} will be substituted once
5711 for each matching switch, with the @code{%*} replaced by the part of
5712 that switch that matched the @code{*}.
5714 @item %@{.@code{S}:@code{X}@}
5715 Substitutes @code{X}, if processing a file with suffix @code{S}.
5717 @item %@{!.@code{S}:@code{X}@}
5718 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
5720 @item %@{@code{S}|@code{P}:@code{X}@}
5721 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
5722 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
5723 although they have a stronger binding than the @samp{|}. If @code{%*}
5724 appears in @code{X}, all of the alternatives must be starred, and only
5725 the first matching alternative is substituted.
5727 For example, a spec string like this:
5730 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5733 will output the following command-line options from the following input
5734 command-line options:
5739 -d fred.c -foo -baz -boggle
5740 -d jim.d -bar -baz -boggle
5743 @item %@{S:X; T:Y; :D@}
5745 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
5746 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
5747 be as many clauses as you need. This may be combined with @code{.},
5748 @code{!}, @code{|}, and @code{*} as needed.
5753 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
5754 construct may contain other nested @samp{%} constructs or spaces, or
5755 even newlines. They are processed as usual, as described above.
5756 Trailing white space in @code{X} is ignored. White space may also
5757 appear anywhere on the left side of the colon in these constructs,
5758 except between @code{.} or @code{*} and the corresponding word.
5760 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
5761 handled specifically in these constructs. If another value of
5762 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5763 @option{-W} switch is found later in the command line, the earlier
5764 switch value is ignored, except with @{@code{S}*@} where @code{S} is
5765 just one letter, which passes all matching options.
5767 The character @samp{|} at the beginning of the predicate text is used to
5768 indicate that a command should be piped to the following command, but
5769 only if @option{-pipe} is specified.
5771 It is built into GCC which switches take arguments and which do not.
5772 (You might think it would be useful to generalize this to allow each
5773 compiler's spec to say which switches take arguments. But this cannot
5774 be done in a consistent fashion. GCC cannot even decide which input
5775 files have been specified without knowing which switches take arguments,
5776 and it must know which input files to compile in order to tell which
5779 GCC also knows implicitly that arguments starting in @option{-l} are to be
5780 treated as compiler output files, and passed to the linker in their
5781 proper position among the other output files.
5783 @c man begin OPTIONS
5785 @node Target Options
5786 @section Specifying Target Machine and Compiler Version
5787 @cindex target options
5788 @cindex cross compiling
5789 @cindex specifying machine version
5790 @cindex specifying compiler version and target machine
5791 @cindex compiler version, specifying
5792 @cindex target machine, specifying
5794 The usual way to run GCC is to run the executable called @file{gcc}, or
5795 @file{<machine>-gcc} when cross-compiling, or
5796 @file{<machine>-gcc-<version>} to run a version other than the one that
5797 was installed last. Sometimes this is inconvenient, so GCC provides
5798 options that will switch to another cross-compiler or version.
5801 @item -b @var{machine}
5803 The argument @var{machine} specifies the target machine for compilation.
5805 The value to use for @var{machine} is the same as was specified as the
5806 machine type when configuring GCC as a cross-compiler. For
5807 example, if a cross-compiler was configured with @samp{configure
5808 i386v}, meaning to compile for an 80386 running System V, then you
5809 would specify @option{-b i386v} to run that cross compiler.
5811 @item -V @var{version}
5813 The argument @var{version} specifies which version of GCC to run.
5814 This is useful when multiple versions are installed. For example,
5815 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5818 The @option{-V} and @option{-b} options work by running the
5819 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5820 use them if you can just run that directly.
5822 @node Submodel Options
5823 @section Hardware Models and Configurations
5824 @cindex submodel options
5825 @cindex specifying hardware config
5826 @cindex hardware models and configurations, specifying
5827 @cindex machine dependent options
5829 Earlier we discussed the standard option @option{-b} which chooses among
5830 different installed compilers for completely different target
5831 machines, such as VAX vs.@: 68000 vs.@: 80386.
5833 In addition, each of these target machine types can have its own
5834 special options, starting with @samp{-m}, to choose among various
5835 hardware models or configurations---for example, 68010 vs 68020,
5836 floating coprocessor or none. A single installed version of the
5837 compiler can compile for any model or configuration, according to the
5840 Some configurations of the compiler also support additional special
5841 options, usually for compatibility with other compilers on the same
5844 These options are defined by the macro @code{TARGET_SWITCHES} in the
5845 machine description. The default for the options is also defined by
5846 that macro, which enables you to change the defaults.
5858 * RS/6000 and PowerPC Options::
5862 * i386 and x86-64 Options::
5864 * Intel 960 Options::
5865 * DEC Alpha Options::
5866 * DEC Alpha/VMS Options::
5869 * System V Options::
5870 * TMS320C3x/C4x Options::
5878 * S/390 and zSeries Options::
5882 * Xstormy16 Options::
5887 @node M680x0 Options
5888 @subsection M680x0 Options
5889 @cindex M680x0 options
5891 These are the @samp{-m} options defined for the 68000 series. The default
5892 values for these options depends on which style of 68000 was selected when
5893 the compiler was configured; the defaults for the most common choices are
5901 Generate output for a 68000. This is the default
5902 when the compiler is configured for 68000-based systems.
5904 Use this option for microcontrollers with a 68000 or EC000 core,
5905 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5911 Generate output for a 68020. This is the default
5912 when the compiler is configured for 68020-based systems.
5916 Generate output containing 68881 instructions for floating point.
5917 This is the default for most 68020 systems unless @option{--nfp} was
5918 specified when the compiler was configured.
5922 Generate output for a 68030. This is the default when the compiler is
5923 configured for 68030-based systems.
5927 Generate output for a 68040. This is the default when the compiler is
5928 configured for 68040-based systems.
5930 This option inhibits the use of 68881/68882 instructions that have to be
5931 emulated by software on the 68040. Use this option if your 68040 does not
5932 have code to emulate those instructions.
5936 Generate output for a 68060. This is the default when the compiler is
5937 configured for 68060-based systems.
5939 This option inhibits the use of 68020 and 68881/68882 instructions that
5940 have to be emulated by software on the 68060. Use this option if your 68060
5941 does not have code to emulate those instructions.
5945 Generate output for a CPU32. This is the default
5946 when the compiler is configured for CPU32-based systems.
5948 Use this option for microcontrollers with a
5949 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5950 68336, 68340, 68341, 68349 and 68360.
5954 Generate output for a 520X ``coldfire'' family cpu. This is the default
5955 when the compiler is configured for 520X-based systems.
5957 Use this option for microcontroller with a 5200 core, including
5958 the MCF5202, MCF5203, MCF5204 and MCF5202.
5963 Generate output for a 68040, without using any of the new instructions.
5964 This results in code which can run relatively efficiently on either a
5965 68020/68881 or a 68030 or a 68040. The generated code does use the
5966 68881 instructions that are emulated on the 68040.
5970 Generate output for a 68060, without using any of the new instructions.
5971 This results in code which can run relatively efficiently on either a
5972 68020/68881 or a 68030 or a 68040. The generated code does use the
5973 68881 instructions that are emulated on the 68060.
5976 @opindex msoft-float
5977 Generate output containing library calls for floating point.
5978 @strong{Warning:} the requisite libraries are not available for all m68k
5979 targets. Normally the facilities of the machine's usual C compiler are
5980 used, but this can't be done directly in cross-compilation. You must
5981 make your own arrangements to provide suitable library functions for
5982 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5983 @samp{m68k-*-coff} do provide software floating point support.
5987 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5990 @opindex mnobitfield
5991 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5992 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5996 Do use the bit-field instructions. The @option{-m68020} option implies
5997 @option{-mbitfield}. This is the default if you use a configuration
5998 designed for a 68020.
6002 Use a different function-calling convention, in which functions
6003 that take a fixed number of arguments return with the @code{rtd}
6004 instruction, which pops their arguments while returning. This
6005 saves one instruction in the caller since there is no need to pop
6006 the arguments there.
6008 This calling convention is incompatible with the one normally
6009 used on Unix, so you cannot use it if you need to call libraries
6010 compiled with the Unix compiler.
6012 Also, you must provide function prototypes for all functions that
6013 take variable numbers of arguments (including @code{printf});
6014 otherwise incorrect code will be generated for calls to those
6017 In addition, seriously incorrect code will result if you call a
6018 function with too many arguments. (Normally, extra arguments are
6019 harmlessly ignored.)
6021 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
6022 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
6025 @itemx -mno-align-int
6027 @opindex mno-align-int
6028 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
6029 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
6030 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
6031 Aligning variables on 32-bit boundaries produces code that runs somewhat
6032 faster on processors with 32-bit busses at the expense of more memory.
6034 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
6035 align structures containing the above types differently than
6036 most published application binary interface specifications for the m68k.
6040 Use the pc-relative addressing mode of the 68000 directly, instead of
6041 using a global offset table. At present, this option implies @option{-fpic},
6042 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
6043 not presently supported with @option{-mpcrel}, though this could be supported for
6044 68020 and higher processors.
6046 @item -mno-strict-align
6047 @itemx -mstrict-align
6048 @opindex mno-strict-align
6049 @opindex mstrict-align
6050 Do not (do) assume that unaligned memory references will be handled by
6054 Generate code that allows the data segment to be located in a different
6055 area of memory from the text segment. This allows for execute in place in
6056 an environment without virtual memory management. This option implies -fPIC.
6059 Generate code that assumes that the data segment follows the text segment.
6060 This is the default.
6062 @item -mid-shared-library
6063 Generate code that supports shared libraries via the library ID method.
6064 This allows for execute in place and shared libraries in an environment
6065 without virtual memory management. This option implies -fPIC.
6067 @item -mno-id-shared-library
6068 Generate code that doesn't assume ID based shared libraries are being used.
6069 This is the default.
6071 @item -mshared-library-id=n
6072 Specified the identification number of the ID based shared library being
6073 compiled. Specifying a value of 0 will generate more compact code, specifying
6074 other values will force the allocation of that number to the current
6075 library but is no more space or time efficient than omitting this option.
6079 @node M68hc1x Options
6080 @subsection M68hc1x Options
6081 @cindex M68hc1x options
6083 These are the @samp{-m} options defined for the 68hc11 and 68hc12
6084 microcontrollers. The default values for these options depends on
6085 which style of microcontroller was selected when the compiler was configured;
6086 the defaults for the most common choices are given below.
6093 Generate output for a 68HC11. This is the default
6094 when the compiler is configured for 68HC11-based systems.
6100 Generate output for a 68HC12. This is the default
6101 when the compiler is configured for 68HC12-based systems.
6107 Generate output for a 68HCS12.
6110 @opindex mauto-incdec
6111 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
6118 Enable the use of 68HC12 min and max instructions.
6121 @itemx -mno-long-calls
6122 @opindex mlong-calls
6123 @opindex mno-long-calls
6124 Treat all calls as being far away (near). If calls are assumed to be
6125 far away, the compiler will use the @code{call} instruction to
6126 call a function and the @code{rtc} instruction for returning.
6130 Consider type @code{int} to be 16 bits wide, like @code{short int}.
6132 @item -msoft-reg-count=@var{count}
6133 @opindex msoft-reg-count
6134 Specify the number of pseudo-soft registers which are used for the
6135 code generation. The maximum number is 32. Using more pseudo-soft
6136 register may or may not result in better code depending on the program.
6137 The default is 4 for 68HC11 and 2 for 68HC12.
6142 @subsection VAX Options
6145 These @samp{-m} options are defined for the VAX:
6150 Do not output certain jump instructions (@code{aobleq} and so on)
6151 that the Unix assembler for the VAX cannot handle across long
6156 Do output those jump instructions, on the assumption that you
6157 will assemble with the GNU assembler.
6161 Output code for g-format floating point numbers instead of d-format.
6165 @subsection SPARC Options
6166 @cindex SPARC options
6168 These @samp{-m} switches are supported on the SPARC:
6173 @opindex mno-app-regs
6175 Specify @option{-mapp-regs} to generate output using the global registers
6176 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
6179 To be fully SVR4 ABI compliant at the cost of some performance loss,
6180 specify @option{-mno-app-regs}. You should compile libraries and system
6181 software with this option.
6186 @opindex mhard-float
6187 Generate output containing floating point instructions. This is the
6193 @opindex msoft-float
6194 Generate output containing library calls for floating point.
6195 @strong{Warning:} the requisite libraries are not available for all SPARC
6196 targets. Normally the facilities of the machine's usual C compiler are
6197 used, but this cannot be done directly in cross-compilation. You must make
6198 your own arrangements to provide suitable library functions for
6199 cross-compilation. The embedded targets @samp{sparc-*-aout} and
6200 @samp{sparclite-*-*} do provide software floating point support.
6202 @option{-msoft-float} changes the calling convention in the output file;
6203 therefore, it is only useful if you compile @emph{all} of a program with
6204 this option. In particular, you need to compile @file{libgcc.a}, the
6205 library that comes with GCC, with @option{-msoft-float} in order for
6208 @item -mhard-quad-float
6209 @opindex mhard-quad-float
6210 Generate output containing quad-word (long double) floating point
6213 @item -msoft-quad-float
6214 @opindex msoft-quad-float
6215 Generate output containing library calls for quad-word (long double)
6216 floating point instructions. The functions called are those specified
6217 in the SPARC ABI@. This is the default.
6219 As of this writing, there are no sparc implementations that have hardware
6220 support for the quad-word floating point instructions. They all invoke
6221 a trap handler for one of these instructions, and then the trap handler
6222 emulates the effect of the instruction. Because of the trap handler overhead,
6223 this is much slower than calling the ABI library routines. Thus the
6224 @option{-msoft-quad-float} option is the default.
6230 With @option{-mflat}, the compiler does not generate save/restore instructions
6231 and will use a ``flat'' or single register window calling convention.
6232 This model uses %i7 as the frame pointer and is compatible with the normal
6233 register window model. Code from either may be intermixed.
6234 The local registers and the input registers (0--5) are still treated as
6235 ``call saved'' registers and will be saved on the stack as necessary.
6237 With @option{-mno-flat} (the default), the compiler emits save/restore
6238 instructions (except for leaf functions) and is the normal mode of operation.
6240 @item -mno-unaligned-doubles
6241 @itemx -munaligned-doubles
6242 @opindex mno-unaligned-doubles
6243 @opindex munaligned-doubles
6244 Assume that doubles have 8 byte alignment. This is the default.
6246 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
6247 alignment only if they are contained in another type, or if they have an
6248 absolute address. Otherwise, it assumes they have 4 byte alignment.
6249 Specifying this option avoids some rare compatibility problems with code
6250 generated by other compilers. It is not the default because it results
6251 in a performance loss, especially for floating point code.
6253 @item -mno-faster-structs
6254 @itemx -mfaster-structs
6255 @opindex mno-faster-structs
6256 @opindex mfaster-structs
6257 With @option{-mfaster-structs}, the compiler assumes that structures
6258 should have 8 byte alignment. This enables the use of pairs of
6259 @code{ldd} and @code{std} instructions for copies in structure
6260 assignment, in place of twice as many @code{ld} and @code{st} pairs.
6261 However, the use of this changed alignment directly violates the SPARC
6262 ABI@. Thus, it's intended only for use on targets where the developer
6263 acknowledges that their resulting code will not be directly in line with
6264 the rules of the ABI@.
6267 @opindex mimpure-text
6268 @option{-mimpure-text}, used in addition to @option{-shared}, tells
6269 the compiler to not pass @option{-z text} to the linker when linking a
6270 shared object. Using this option, you can link position-dependent
6271 code into a shared object.
6273 @option{-mimpure-text} suppresses the ``relocations remain against
6274 allocatable but non-writable sections'' linker error message.
6275 However, the necessary relocations will trigger copy-on-write, and the
6276 shared object is not actually shared across processes. Instead of
6277 using @option{-mimpure-text}, you should compile all source code with
6278 @option{-fpic} or @option{-fPIC}.
6280 This option is only available on SunOS and Solaris.
6286 These two options select variations on the SPARC architecture.
6288 By default (unless specifically configured for the Fujitsu SPARClite),
6289 GCC generates code for the v7 variant of the SPARC architecture.
6291 @option{-mv8} will give you SPARC v8 code. The only difference from v7
6292 code is that the compiler emits the integer multiply and integer
6293 divide instructions which exist in SPARC v8 but not in SPARC v7.
6295 @option{-msparclite} will give you SPARClite code. This adds the integer
6296 multiply, integer divide step and scan (@code{ffs}) instructions which
6297 exist in SPARClite but not in SPARC v7.
6299 These options are deprecated and will be deleted in a future GCC release.
6300 They have been replaced with @option{-mcpu=xxx}.
6305 @opindex msupersparc
6306 These two options select the processor for which the code is optimized.
6308 With @option{-mcypress} (the default), the compiler optimizes code for the
6309 Cypress CY7C602 chip, as used in the SPARCStation/SPARCServer 3xx series.
6310 This is also appropriate for the older SPARCStation 1, 2, IPX etc.
6312 With @option{-msupersparc} the compiler optimizes code for the SuperSPARC cpu, as
6313 used in the SPARCStation 10, 1000 and 2000 series. This flag also enables use
6314 of the full SPARC v8 instruction set.
6316 These options are deprecated and will be deleted in a future GCC release.
6317 They have been replaced with @option{-mcpu=xxx}.
6319 @item -mcpu=@var{cpu_type}
6321 Set the instruction set, register set, and instruction scheduling parameters
6322 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
6323 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
6324 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
6325 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
6328 Default instruction scheduling parameters are used for values that select
6329 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
6330 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
6332 Here is a list of each supported architecture and their supported
6337 v8: supersparc, hypersparc
6338 sparclite: f930, f934, sparclite86x
6340 v9: ultrasparc, ultrasparc3
6343 @item -mtune=@var{cpu_type}
6345 Set the instruction scheduling parameters for machine type
6346 @var{cpu_type}, but do not set the instruction set or register set that the
6347 option @option{-mcpu=@var{cpu_type}} would.
6349 The same values for @option{-mcpu=@var{cpu_type}} can be used for
6350 @option{-mtune=@var{cpu_type}}, but the only useful values are those
6351 that select a particular cpu implementation. Those are @samp{cypress},
6352 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
6353 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
6358 These @samp{-m} switches are supported in addition to the above
6359 on the SPARCLET processor.
6362 @item -mlittle-endian
6363 @opindex mlittle-endian
6364 Generate code for a processor running in little-endian mode.
6368 Treat register @code{%g0} as a normal register.
6369 GCC will continue to clobber it as necessary but will not assume
6370 it always reads as 0.
6372 @item -mbroken-saverestore
6373 @opindex mbroken-saverestore
6374 Generate code that does not use non-trivial forms of the @code{save} and
6375 @code{restore} instructions. Early versions of the SPARCLET processor do
6376 not correctly handle @code{save} and @code{restore} instructions used with
6377 arguments. They correctly handle them used without arguments. A @code{save}
6378 instruction used without arguments increments the current window pointer
6379 but does not allocate a new stack frame. It is assumed that the window
6380 overflow trap handler will properly handle this case as will interrupt
6384 These @samp{-m} switches are supported in addition to the above
6385 on SPARC V9 processors in 64-bit environments.
6388 @item -mlittle-endian
6389 @opindex mlittle-endian
6390 Generate code for a processor running in little-endian mode. It is only
6391 available for a few configurations and most notably not on Solaris.
6397 Generate code for a 32-bit or 64-bit environment.
6398 The 32-bit environment sets int, long and pointer to 32 bits.
6399 The 64-bit environment sets int to 32 bits and long and pointer
6402 @item -mcmodel=medlow
6403 @opindex mcmodel=medlow
6404 Generate code for the Medium/Low code model: the program must be linked
6405 in the low 32 bits of the address space. Pointers are 64 bits.
6406 Programs can be statically or dynamically linked.
6408 @item -mcmodel=medmid
6409 @opindex mcmodel=medmid
6410 Generate code for the Medium/Middle code model: the program must be linked
6411 in the low 44 bits of the address space, the text segment must be less than
6412 2G bytes, and data segment must be within 2G of the text segment.
6413 Pointers are 64 bits.
6415 @item -mcmodel=medany
6416 @opindex mcmodel=medany
6417 Generate code for the Medium/Anywhere code model: the program may be linked
6418 anywhere in the address space, the text segment must be less than
6419 2G bytes, and data segment must be within 2G of the text segment.
6420 Pointers are 64 bits.
6422 @item -mcmodel=embmedany
6423 @opindex mcmodel=embmedany
6424 Generate code for the Medium/Anywhere code model for embedded systems:
6425 assume a 32-bit text and a 32-bit data segment, both starting anywhere
6426 (determined at link time). Register %g4 points to the base of the
6427 data segment. Pointers are still 64 bits.
6428 Programs are statically linked, PIC is not supported.
6431 @itemx -mno-stack-bias
6432 @opindex mstack-bias
6433 @opindex mno-stack-bias
6434 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
6435 frame pointer if present, are offset by @minus{}2047 which must be added back
6436 when making stack frame references.
6437 Otherwise, assume no such offset is present.
6441 @subsection ARM Options
6444 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6449 @opindex mapcs-frame
6450 Generate a stack frame that is compliant with the ARM Procedure Call
6451 Standard for all functions, even if this is not strictly necessary for
6452 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6453 with this option will cause the stack frames not to be generated for
6454 leaf functions. The default is @option{-mno-apcs-frame}.
6458 This is a synonym for @option{-mapcs-frame}.
6462 Generate code for a processor running with a 26-bit program counter,
6463 and conforming to the function calling standards for the APCS 26-bit
6464 option. This option replaces the @option{-m2} and @option{-m3} options
6465 of previous releases of the compiler.
6469 Generate code for a processor running with a 32-bit program counter,
6470 and conforming to the function calling standards for the APCS 32-bit
6471 option. This option replaces the @option{-m6} option of previous releases
6475 @c not currently implemented
6476 @item -mapcs-stack-check
6477 @opindex mapcs-stack-check
6478 Generate code to check the amount of stack space available upon entry to
6479 every function (that actually uses some stack space). If there is
6480 insufficient space available then either the function
6481 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6482 called, depending upon the amount of stack space required. The run time
6483 system is required to provide these functions. The default is
6484 @option{-mno-apcs-stack-check}, since this produces smaller code.
6486 @c not currently implemented
6488 @opindex mapcs-float
6489 Pass floating point arguments using the float point registers. This is
6490 one of the variants of the APCS@. This option is recommended if the
6491 target hardware has a floating point unit or if a lot of floating point
6492 arithmetic is going to be performed by the code. The default is
6493 @option{-mno-apcs-float}, since integer only code is slightly increased in
6494 size if @option{-mapcs-float} is used.
6496 @c not currently implemented
6497 @item -mapcs-reentrant
6498 @opindex mapcs-reentrant
6499 Generate reentrant, position independent code. The default is
6500 @option{-mno-apcs-reentrant}.
6503 @item -mthumb-interwork
6504 @opindex mthumb-interwork
6505 Generate code which supports calling between the ARM and Thumb
6506 instruction sets. Without this option the two instruction sets cannot
6507 be reliably used inside one program. The default is
6508 @option{-mno-thumb-interwork}, since slightly larger code is generated
6509 when @option{-mthumb-interwork} is specified.
6511 @item -mno-sched-prolog
6512 @opindex mno-sched-prolog
6513 Prevent the reordering of instructions in the function prolog, or the
6514 merging of those instruction with the instructions in the function's
6515 body. This means that all functions will start with a recognizable set
6516 of instructions (or in fact one of a choice from a small set of
6517 different function prologues), and this information can be used to
6518 locate the start if functions inside an executable piece of code. The
6519 default is @option{-msched-prolog}.
6522 @opindex mhard-float
6523 Generate output containing floating point instructions. This is the
6527 @opindex msoft-float
6528 Generate output containing library calls for floating point.
6529 @strong{Warning:} the requisite libraries are not available for all ARM
6530 targets. Normally the facilities of the machine's usual C compiler are
6531 used, but this cannot be done directly in cross-compilation. You must make
6532 your own arrangements to provide suitable library functions for
6535 @option{-msoft-float} changes the calling convention in the output file;
6536 therefore, it is only useful if you compile @emph{all} of a program with
6537 this option. In particular, you need to compile @file{libgcc.a}, the
6538 library that comes with GCC, with @option{-msoft-float} in order for
6541 @item -mlittle-endian
6542 @opindex mlittle-endian
6543 Generate code for a processor running in little-endian mode. This is
6544 the default for all standard configurations.
6547 @opindex mbig-endian
6548 Generate code for a processor running in big-endian mode; the default is
6549 to compile code for a little-endian processor.
6551 @item -mwords-little-endian
6552 @opindex mwords-little-endian
6553 This option only applies when generating code for big-endian processors.
6554 Generate code for a little-endian word order but a big-endian byte
6555 order. That is, a byte order of the form @samp{32107654}. Note: this
6556 option should only be used if you require compatibility with code for
6557 big-endian ARM processors generated by versions of the compiler prior to
6560 @item -malignment-traps
6561 @opindex malignment-traps
6562 Generate code that will not trap if the MMU has alignment traps enabled.
6563 On ARM architectures prior to ARMv4, there were no instructions to
6564 access half-word objects stored in memory. However, when reading from
6565 memory a feature of the ARM architecture allows a word load to be used,
6566 even if the address is unaligned, and the processor core will rotate the
6567 data as it is being loaded. This option tells the compiler that such
6568 misaligned accesses will cause a MMU trap and that it should instead
6569 synthesize the access as a series of byte accesses. The compiler can
6570 still use word accesses to load half-word data if it knows that the
6571 address is aligned to a word boundary.
6573 This option is ignored when compiling for ARM architecture 4 or later,
6574 since these processors have instructions to directly access half-word
6577 @item -mno-alignment-traps
6578 @opindex mno-alignment-traps
6579 Generate code that assumes that the MMU will not trap unaligned
6580 accesses. This produces better code when the target instruction set
6581 does not have half-word memory operations (i.e.@: implementations prior to
6584 Note that you cannot use this option to access unaligned word objects,
6585 since the processor will only fetch one 32-bit aligned object from
6588 The default setting for most targets is @option{-mno-alignment-traps}, since
6589 this produces better code when there are no half-word memory
6590 instructions available.
6592 @item -mshort-load-bytes
6593 @itemx -mno-short-load-words
6594 @opindex mshort-load-bytes
6595 @opindex mno-short-load-words
6596 These are deprecated aliases for @option{-malignment-traps}.
6598 @item -mno-short-load-bytes
6599 @itemx -mshort-load-words
6600 @opindex mno-short-load-bytes
6601 @opindex mshort-load-words
6602 This are deprecated aliases for @option{-mno-alignment-traps}.
6604 @item -mcpu=@var{name}
6606 This specifies the name of the target ARM processor. GCC uses this name
6607 to determine what kind of instructions it can emit when generating
6608 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6609 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6610 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6611 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6612 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6613 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6614 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6615 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6616 @samp{arm920t}, @samp{arm926ejs}, @samp{arm940t}, @samp{arm9tdmi},
6617 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ejs},
6618 @samp{arm1136js}, @samp{arm1136jfs} ,@samp{xscale}, @samp{iwmmxt},
6621 @itemx -mtune=@var{name}
6623 This option is very similar to the @option{-mcpu=} option, except that
6624 instead of specifying the actual target processor type, and hence
6625 restricting which instructions can be used, it specifies that GCC should
6626 tune the performance of the code as if the target were of the type
6627 specified in this option, but still choosing the instructions that it
6628 will generate based on the cpu specified by a @option{-mcpu=} option.
6629 For some ARM implementations better performance can be obtained by using
6632 @item -march=@var{name}
6634 This specifies the name of the target ARM architecture. GCC uses this
6635 name to determine what kind of instructions it can emit when generating
6636 assembly code. This option can be used in conjunction with or instead
6637 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6638 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6639 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6j},
6640 @samp{iwmmxt}, @samp{ep9312}.
6642 @item -mfpe=@var{number}
6643 @itemx -mfp=@var{number}
6646 This specifies the version of the floating point emulation available on
6647 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6648 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6650 @item -mstructure-size-boundary=@var{n}
6651 @opindex mstructure-size-boundary
6652 The size of all structures and unions will be rounded up to a multiple
6653 of the number of bits set by this option. Permissible values are 8 and
6654 32. The default value varies for different toolchains. For the COFF
6655 targeted toolchain the default value is 8. Specifying the larger number
6656 can produce faster, more efficient code, but can also increase the size
6657 of the program. The two values are potentially incompatible. Code
6658 compiled with one value cannot necessarily expect to work with code or
6659 libraries compiled with the other value, if they exchange information
6660 using structures or unions.
6662 @item -mabort-on-noreturn
6663 @opindex mabort-on-noreturn
6664 Generate a call to the function @code{abort} at the end of a
6665 @code{noreturn} function. It will be executed if the function tries to
6669 @itemx -mno-long-calls
6670 @opindex mlong-calls
6671 @opindex mno-long-calls
6672 Tells the compiler to perform function calls by first loading the
6673 address of the function into a register and then performing a subroutine
6674 call on this register. This switch is needed if the target function
6675 will lie outside of the 64 megabyte addressing range of the offset based
6676 version of subroutine call instruction.
6678 Even if this switch is enabled, not all function calls will be turned
6679 into long calls. The heuristic is that static functions, functions
6680 which have the @samp{short-call} attribute, functions that are inside
6681 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6682 definitions have already been compiled within the current compilation
6683 unit, will not be turned into long calls. The exception to this rule is
6684 that weak function definitions, functions with the @samp{long-call}
6685 attribute or the @samp{section} attribute, and functions that are within
6686 the scope of a @samp{#pragma long_calls} directive, will always be
6687 turned into long calls.
6689 This feature is not enabled by default. Specifying
6690 @option{-mno-long-calls} will restore the default behavior, as will
6691 placing the function calls within the scope of a @samp{#pragma
6692 long_calls_off} directive. Note these switches have no effect on how
6693 the compiler generates code to handle function calls via function
6696 @item -mnop-fun-dllimport
6697 @opindex mnop-fun-dllimport
6698 Disable support for the @code{dllimport} attribute.
6700 @item -msingle-pic-base
6701 @opindex msingle-pic-base
6702 Treat the register used for PIC addressing as read-only, rather than
6703 loading it in the prologue for each function. The run-time system is
6704 responsible for initializing this register with an appropriate value
6705 before execution begins.
6707 @item -mpic-register=@var{reg}
6708 @opindex mpic-register
6709 Specify the register to be used for PIC addressing. The default is R10
6710 unless stack-checking is enabled, when R9 is used.
6712 @item -mcirrus-fix-invalid-insns
6713 @opindex mcirrus-fix-invalid-insns
6714 @opindex mno-cirrus-fix-invalid-insns
6715 Insert NOPs into the instruction stream to in order to work around
6716 problems with invalid Maverick instruction combinations. This option
6717 is only valid if the @option{-mcpu=ep9312} option has been used to
6718 enable generation of instructions for the Cirrus Maverick floating
6719 point co-processor. This option is not enabled by default, since the
6720 problem is only present in older Maverick implementations. The default
6721 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6724 @item -mpoke-function-name
6725 @opindex mpoke-function-name
6726 Write the name of each function into the text section, directly
6727 preceding the function prologue. The generated code is similar to this:
6731 .ascii "arm_poke_function_name", 0
6734 .word 0xff000000 + (t1 - t0)
6735 arm_poke_function_name
6737 stmfd sp!, @{fp, ip, lr, pc@}
6741 When performing a stack backtrace, code can inspect the value of
6742 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6743 location @code{pc - 12} and the top 8 bits are set, then we know that
6744 there is a function name embedded immediately preceding this location
6745 and has length @code{((pc[-3]) & 0xff000000)}.
6749 Generate code for the 16-bit Thumb instruction set. The default is to
6750 use the 32-bit ARM instruction set.
6753 @opindex mtpcs-frame
6754 Generate a stack frame that is compliant with the Thumb Procedure Call
6755 Standard for all non-leaf functions. (A leaf function is one that does
6756 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6758 @item -mtpcs-leaf-frame
6759 @opindex mtpcs-leaf-frame
6760 Generate a stack frame that is compliant with the Thumb Procedure Call
6761 Standard for all leaf functions. (A leaf function is one that does
6762 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6764 @item -mcallee-super-interworking
6765 @opindex mcallee-super-interworking
6766 Gives all externally visible functions in the file being compiled an ARM
6767 instruction set header which switches to Thumb mode before executing the
6768 rest of the function. This allows these functions to be called from
6769 non-interworking code.
6771 @item -mcaller-super-interworking
6772 @opindex mcaller-super-interworking
6773 Allows calls via function pointers (including virtual functions) to
6774 execute correctly regardless of whether the target code has been
6775 compiled for interworking or not. There is a small overhead in the cost
6776 of executing a function pointer if this option is enabled.
6780 @node MN10200 Options
6781 @subsection MN10200 Options
6782 @cindex MN10200 options
6784 These @option{-m} options are defined for Matsushita MN10200 architectures:
6789 Indicate to the linker that it should perform a relaxation optimization pass
6790 to shorten branches, calls and absolute memory addresses. This option only
6791 has an effect when used on the command line for the final link step.
6793 This option makes symbolic debugging impossible.
6796 @node MN10300 Options
6797 @subsection MN10300 Options
6798 @cindex MN10300 options
6800 These @option{-m} options are defined for Matsushita MN10300 architectures:
6805 Generate code to avoid bugs in the multiply instructions for the MN10300
6806 processors. This is the default.
6809 @opindex mno-mult-bug
6810 Do not generate code to avoid bugs in the multiply instructions for the
6815 Generate code which uses features specific to the AM33 processor.
6819 Do not generate code which uses features specific to the AM33 processor. This
6824 Do not link in the C run-time initialization object file.
6828 Indicate to the linker that it should perform a relaxation optimization pass
6829 to shorten branches, calls and absolute memory addresses. This option only
6830 has an effect when used on the command line for the final link step.
6832 This option makes symbolic debugging impossible.
6836 @node M32R/D Options
6837 @subsection M32R/D Options
6838 @cindex M32R/D options
6840 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6845 Generate code for the M32R/X@.
6849 Generate code for the M32R@. This is the default.
6851 @item -mcode-model=small
6852 @opindex mcode-model=small
6853 Assume all objects live in the lower 16MB of memory (so that their addresses
6854 can be loaded with the @code{ld24} instruction), and assume all subroutines
6855 are reachable with the @code{bl} instruction.
6856 This is the default.
6858 The addressability of a particular object can be set with the
6859 @code{model} attribute.
6861 @item -mcode-model=medium
6862 @opindex mcode-model=medium
6863 Assume objects may be anywhere in the 32-bit address space (the compiler
6864 will generate @code{seth/add3} instructions to load their addresses), and
6865 assume all subroutines are reachable with the @code{bl} instruction.
6867 @item -mcode-model=large
6868 @opindex mcode-model=large
6869 Assume objects may be anywhere in the 32-bit address space (the compiler
6870 will generate @code{seth/add3} instructions to load their addresses), and
6871 assume subroutines may not be reachable with the @code{bl} instruction
6872 (the compiler will generate the much slower @code{seth/add3/jl}
6873 instruction sequence).
6876 @opindex msdata=none
6877 Disable use of the small data area. Variables will be put into
6878 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6879 @code{section} attribute has been specified).
6880 This is the default.
6882 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6883 Objects may be explicitly put in the small data area with the
6884 @code{section} attribute using one of these sections.
6887 @opindex msdata=sdata
6888 Put small global and static data in the small data area, but do not
6889 generate special code to reference them.
6893 Put small global and static data in the small data area, and generate
6894 special instructions to reference them.
6898 @cindex smaller data references
6899 Put global and static objects less than or equal to @var{num} bytes
6900 into the small data or bss sections instead of the normal data or bss
6901 sections. The default value of @var{num} is 8.
6902 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6903 for this option to have any effect.
6905 All modules should be compiled with the same @option{-G @var{num}} value.
6906 Compiling with different values of @var{num} may or may not work; if it
6907 doesn't the linker will give an error message---incorrect code will not be
6913 @subsection M88K Options
6914 @cindex M88k options
6916 These @samp{-m} options are defined for Motorola 88k architectures:
6921 Generate code that works well on both the m88100 and the
6926 Generate code that works best for the m88100, but that also
6931 Generate code that works best for the m88110, and may not run
6936 Obsolete option to be removed from the next revision.
6939 @item -midentify-revision
6940 @opindex midentify-revision
6941 @cindex identifying source, compiler (88k)
6942 Include an @code{ident} directive in the assembler output recording the
6943 source file name, compiler name and version, timestamp, and compilation
6946 @item -mno-underscores
6947 @opindex mno-underscores
6948 @cindex underscores, avoiding (88k)
6949 In assembler output, emit symbol names without adding an underscore
6950 character at the beginning of each name. The default is to use an
6951 underscore as prefix on each name.
6953 @item -mocs-debug-info
6954 @itemx -mno-ocs-debug-info
6955 @opindex mocs-debug-info
6956 @opindex mno-ocs-debug-info
6958 @cindex debugging, 88k OCS
6959 Include (or omit) additional debugging information (about registers used
6960 in each stack frame) as specified in the 88open Object Compatibility
6961 Standard, ``OCS''@. This extra information allows debugging of code that
6962 has had the frame pointer eliminated. The default for SVr4 and Delta 88
6963 SVr3.2 is to include this information; other 88k configurations omit this
6964 information by default.
6966 @item -mocs-frame-position
6967 @opindex mocs-frame-position
6968 @cindex register positions in frame (88k)
6969 When emitting COFF debugging information for automatic variables and
6970 parameters stored on the stack, use the offset from the canonical frame
6971 address, which is the stack pointer (register 31) on entry to the
6972 function. The SVr4 and Delta88 SVr3.2, and BCS configurations use
6973 @option{-mocs-frame-position}; other 88k configurations have the default
6974 @option{-mno-ocs-frame-position}.
6976 @item -mno-ocs-frame-position
6977 @opindex mno-ocs-frame-position
6978 @cindex register positions in frame (88k)
6979 When emitting COFF debugging information for automatic variables and
6980 parameters stored on the stack, use the offset from the frame pointer
6981 register (register 30). When this option is in effect, the frame
6982 pointer is not eliminated when debugging information is selected by the
6985 @item -moptimize-arg-area
6986 @opindex moptimize-arg-area
6987 @cindex arguments in frame (88k)
6988 Save space by reorganizing the stack frame. This option generates code
6989 that does not agree with the 88open specifications, but uses less
6992 @itemx -mno-optimize-arg-area
6993 @opindex mno-optimize-arg-area
6994 Do not reorganize the stack frame to save space. This is the default.
6995 The generated conforms to the specification, but uses more memory.
6997 @item -mshort-data-@var{num}
6998 @opindex mshort-data
6999 @cindex smaller data references (88k)
7000 @cindex r0-relative references (88k)
7001 Generate smaller data references by making them relative to @code{r0},
7002 which allows loading a value using a single instruction (rather than the
7003 usual two). You control which data references are affected by
7004 specifying @var{num} with this option. For example, if you specify
7005 @option{-mshort-data-512}, then the data references affected are those
7006 involving displacements of less than 512 bytes.
7007 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
7010 @item -mserialize-volatile
7011 @opindex mserialize-volatile
7012 @itemx -mno-serialize-volatile
7013 @opindex mno-serialize-volatile
7014 @cindex sequential consistency on 88k
7015 Do, or don't, generate code to guarantee sequential consistency
7016 of volatile memory references. By default, consistency is
7019 The order of memory references made by the MC88110 processor does
7020 not always match the order of the instructions requesting those
7021 references. In particular, a load instruction may execute before
7022 a preceding store instruction. Such reordering violates
7023 sequential consistency of volatile memory references, when there
7024 are multiple processors. When consistency must be guaranteed,
7025 GCC generates special instructions, as needed, to force
7026 execution in the proper order.
7028 The MC88100 processor does not reorder memory references and so
7029 always provides sequential consistency. However, by default, GCC
7030 generates the special instructions to guarantee consistency
7031 even when you use @option{-m88100}, so that the code may be run on an
7032 MC88110 processor. If you intend to run your code only on the
7033 MC88100 processor, you may use @option{-mno-serialize-volatile}.
7035 The extra code generated to guarantee consistency may affect the
7036 performance of your application. If you know that you can safely
7037 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
7043 @cindex assembler syntax, 88k
7045 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
7046 related to System V release 4 (SVr4). This controls the following:
7050 Which variant of the assembler syntax to emit.
7052 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
7053 that is used on System V release 4.
7055 @option{-msvr4} makes GCC issue additional declaration directives used in
7059 @option{-msvr4} is the default for the m88k-motorola-sysv4 configuration.
7060 @option{-msvr3} is the default for all other m88k configurations.
7062 @item -mversion-03.00
7063 @opindex mversion-03.00
7064 This option is obsolete, and is ignored.
7065 @c ??? which asm syntax better for GAS? option there too?
7067 @item -mno-check-zero-division
7068 @itemx -mcheck-zero-division
7069 @opindex mno-check-zero-division
7070 @opindex mcheck-zero-division
7071 @cindex zero division on 88k
7072 Do, or don't, generate code to guarantee that integer division by
7073 zero will be detected. By default, detection is guaranteed.
7075 Some models of the MC88100 processor fail to trap upon integer
7076 division by zero under certain conditions. By default, when
7077 compiling code that might be run on such a processor, GCC
7078 generates code that explicitly checks for zero-valued divisors
7079 and traps with exception number 503 when one is detected. Use of
7080 @option{-mno-check-zero-division} suppresses such checking for code
7081 generated to run on an MC88100 processor.
7083 GCC assumes that the MC88110 processor correctly detects all instances
7084 of integer division by zero. When @option{-m88110} is specified, no
7085 explicit checks for zero-valued divisors are generated, and both
7086 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
7089 @item -muse-div-instruction
7090 @opindex muse-div-instruction
7091 @cindex divide instruction, 88k
7092 Use the div instruction for signed integer division on the
7093 MC88100 processor. By default, the div instruction is not used.
7095 On the MC88100 processor the signed integer division instruction
7096 div) traps to the operating system on a negative operand. The
7097 operating system transparently completes the operation, but at a
7098 large cost in execution time. By default, when compiling code
7099 that might be run on an MC88100 processor, GCC emulates signed
7100 integer division using the unsigned integer division instruction
7101 divu), thereby avoiding the large penalty of a trap to the
7102 operating system. Such emulation has its own, smaller, execution
7103 cost in both time and space. To the extent that your code's
7104 important signed integer division operations are performed on two
7105 nonnegative operands, it may be desirable to use the div
7106 instruction directly.
7108 On the MC88110 processor the div instruction (also known as the
7109 divs instruction) processes negative operands without trapping to
7110 the operating system. When @option{-m88110} is specified,
7111 @option{-muse-div-instruction} is ignored, and the div instruction is used
7112 for signed integer division.
7114 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
7115 particular, the behavior of such a division with and without
7116 @option{-muse-div-instruction} may differ.
7118 @item -mtrap-large-shift
7119 @itemx -mhandle-large-shift
7120 @opindex mtrap-large-shift
7121 @opindex mhandle-large-shift
7122 @cindex bit shift overflow (88k)
7123 @cindex large bit shifts (88k)
7124 Include code to detect bit-shifts of more than 31 bits; respectively,
7125 trap such shifts or emit code to handle them properly. By default GCC
7126 makes no special provision for large bit shifts.
7128 @item -mwarn-passed-structs
7129 @opindex mwarn-passed-structs
7130 @cindex structure passing (88k)
7131 Warn when a function passes a struct as an argument or result.
7132 Structure-passing conventions have changed during the evolution of the C
7133 language, and are often the source of portability problems. By default,
7134 GCC issues no such warning.
7137 @c break page here to avoid unsightly interparagraph stretch.
7141 @node RS/6000 and PowerPC Options
7142 @subsection IBM RS/6000 and PowerPC Options
7143 @cindex RS/6000 and PowerPC Options
7144 @cindex IBM RS/6000 and PowerPC Options
7146 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
7154 @itemx -mpowerpc-gpopt
7155 @itemx -mno-powerpc-gpopt
7156 @itemx -mpowerpc-gfxopt
7157 @itemx -mno-powerpc-gfxopt
7159 @itemx -mno-powerpc64
7165 @opindex mno-powerpc
7166 @opindex mpowerpc-gpopt
7167 @opindex mno-powerpc-gpopt
7168 @opindex mpowerpc-gfxopt
7169 @opindex mno-powerpc-gfxopt
7171 @opindex mno-powerpc64
7172 GCC supports two related instruction set architectures for the
7173 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
7174 instructions supported by the @samp{rios} chip set used in the original
7175 RS/6000 systems and the @dfn{PowerPC} instruction set is the
7176 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
7177 the IBM 4xx microprocessors.
7179 Neither architecture is a subset of the other. However there is a
7180 large common subset of instructions supported by both. An MQ
7181 register is included in processors supporting the POWER architecture.
7183 You use these options to specify which instructions are available on the
7184 processor you are using. The default value of these options is
7185 determined when configuring GCC@. Specifying the
7186 @option{-mcpu=@var{cpu_type}} overrides the specification of these
7187 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
7188 rather than the options listed above.
7190 The @option{-mpower} option allows GCC to generate instructions that
7191 are found only in the POWER architecture and to use the MQ register.
7192 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
7193 to generate instructions that are present in the POWER2 architecture but
7194 not the original POWER architecture.
7196 The @option{-mpowerpc} option allows GCC to generate instructions that
7197 are found only in the 32-bit subset of the PowerPC architecture.
7198 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
7199 GCC to use the optional PowerPC architecture instructions in the
7200 General Purpose group, including floating-point square root. Specifying
7201 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
7202 use the optional PowerPC architecture instructions in the Graphics
7203 group, including floating-point select.
7205 The @option{-mpowerpc64} option allows GCC to generate the additional
7206 64-bit instructions that are found in the full PowerPC64 architecture
7207 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
7208 @option{-mno-powerpc64}.
7210 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
7211 will use only the instructions in the common subset of both
7212 architectures plus some special AIX common-mode calls, and will not use
7213 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
7214 permits GCC to use any instruction from either architecture and to
7215 allow use of the MQ register; specify this for the Motorola MPC601.
7217 @item -mnew-mnemonics
7218 @itemx -mold-mnemonics
7219 @opindex mnew-mnemonics
7220 @opindex mold-mnemonics
7221 Select which mnemonics to use in the generated assembler code. With
7222 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
7223 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
7224 assembler mnemonics defined for the POWER architecture. Instructions
7225 defined in only one architecture have only one mnemonic; GCC uses that
7226 mnemonic irrespective of which of these options is specified.
7228 GCC defaults to the mnemonics appropriate for the architecture in
7229 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
7230 value of these option. Unless you are building a cross-compiler, you
7231 should normally not specify either @option{-mnew-mnemonics} or
7232 @option{-mold-mnemonics}, but should instead accept the default.
7234 @item -mcpu=@var{cpu_type}
7236 Set architecture type, register usage, choice of mnemonics, and
7237 instruction scheduling parameters for machine type @var{cpu_type}.
7238 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
7239 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
7240 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
7241 @samp{630}, @samp{740}, @samp{7400}, @samp{7450}, @samp{G4},
7242 @samp{750}, @samp{G3}, @samp{power}, @samp{power2}, @samp{powerpc},
7243 @samp{403}, @samp{505}, @samp{801}, @samp{821}, @samp{823}, @samp{860},
7244 @samp{970}, @samp{G5} and @samp{common}.
7246 @option{-mcpu=common} selects a completely generic processor. Code
7247 generated under this option will run on any POWER or PowerPC processor.
7248 GCC will use only the instructions in the common subset of both
7249 architectures, and will not use the MQ register. GCC assumes a generic
7250 processor model for scheduling purposes.
7252 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
7253 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
7254 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
7255 types, with an appropriate, generic processor model assumed for
7256 scheduling purposes.
7258 The other options specify a specific processor. Code generated under
7259 those options will run best on that processor, and may not run at all on
7262 The @option{-mcpu} options automatically enable or disable other
7263 @option{-m} options as follows:
7267 @option{-mno-power}, @option{-mno-powerpc}
7274 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
7293 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
7296 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
7301 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
7304 @item -mtune=@var{cpu_type}
7306 Set the instruction scheduling parameters for machine type
7307 @var{cpu_type}, but do not set the architecture type, register usage, or
7308 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
7309 values for @var{cpu_type} are used for @option{-mtune} as for
7310 @option{-mcpu}. If both are specified, the code generated will use the
7311 architecture, registers, and mnemonics set by @option{-mcpu}, but the
7312 scheduling parameters set by @option{-mtune}.
7317 @opindex mno-altivec
7318 These switches enable or disable the use of built-in functions that
7319 allow access to the AltiVec instruction set. You may also need to set
7320 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
7325 Extend the current ABI with SPE ABI extensions. This does not change
7326 the default ABI, instead it adds the SPE ABI extensions to the current
7330 @opindex mabi=no-spe
7331 Disable Booke SPE ABI extensions for the current ABI.
7333 @item -misel=@var{yes/no}
7336 This switch enables or disables the generation of ISEL instructions.
7338 @item -mspe=@var{yes/no}
7341 This switch enables or disables the generation of SPE simd
7344 @item -mfloat-gprs=@var{yes/no}
7346 @opindex mfloat-gprs
7347 This switch enables or disables the generation of floating point
7348 operations on the general purpose registers for architectures that
7349 support it. This option is currently only available on the MPC8540.
7352 @itemx -mno-fp-in-toc
7353 @itemx -mno-sum-in-toc
7354 @itemx -mminimal-toc
7356 @opindex mno-fp-in-toc
7357 @opindex mno-sum-in-toc
7358 @opindex mminimal-toc
7359 Modify generation of the TOC (Table Of Contents), which is created for
7360 every executable file. The @option{-mfull-toc} option is selected by
7361 default. In that case, GCC will allocate at least one TOC entry for
7362 each unique non-automatic variable reference in your program. GCC
7363 will also place floating-point constants in the TOC@. However, only
7364 16,384 entries are available in the TOC@.
7366 If you receive a linker error message that saying you have overflowed
7367 the available TOC space, you can reduce the amount of TOC space used
7368 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
7369 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
7370 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
7371 generate code to calculate the sum of an address and a constant at
7372 run-time instead of putting that sum into the TOC@. You may specify one
7373 or both of these options. Each causes GCC to produce very slightly
7374 slower and larger code at the expense of conserving TOC space.
7376 If you still run out of space in the TOC even when you specify both of
7377 these options, specify @option{-mminimal-toc} instead. This option causes
7378 GCC to make only one TOC entry for every file. When you specify this
7379 option, GCC will produce code that is slower and larger but which
7380 uses extremely little TOC space. You may wish to use this option
7381 only on files that contain less frequently executed code.
7387 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
7388 @code{long} type, and the infrastructure needed to support them.
7389 Specifying @option{-maix64} implies @option{-mpowerpc64} and
7390 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
7391 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
7396 @opindex mno-xl-call
7397 On AIX, pass floating-point arguments to prototyped functions beyond the
7398 register save area (RSA) on the stack in addition to argument FPRs. The
7399 AIX calling convention was extended but not initially documented to
7400 handle an obscure K&R C case of calling a function that takes the
7401 address of its arguments with fewer arguments than declared. AIX XL
7402 compilers access floating point arguments which do not fit in the
7403 RSA from the stack when a subroutine is compiled without
7404 optimization. Because always storing floating-point arguments on the
7405 stack is inefficient and rarely needed, this option is not enabled by
7406 default and only is necessary when calling subroutines compiled by AIX
7407 XL compilers without optimization.
7411 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
7412 application written to use message passing with special startup code to
7413 enable the application to run. The system must have PE installed in the
7414 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
7415 must be overridden with the @option{-specs=} option to specify the
7416 appropriate directory location. The Parallel Environment does not
7417 support threads, so the @option{-mpe} option and the @option{-pthread}
7418 option are incompatible.
7420 @item -malign-natural
7421 @itemx -malign-power
7422 @opindex malign-natural
7423 @opindex malign-power
7424 On AIX, Darwin, and 64-bit PowerPC Linux, the option
7425 @option{-malign-natural} overrides the ABI-defined alignment of larger
7426 types, such as floating-point doubles, on their natural size-based boundary.
7427 The option @option{-malign-power} instructs GCC to follow the ABI-specified
7428 alignment rules. GCC defaults to the standard alignment defined in the ABI.
7432 @opindex msoft-float
7433 @opindex mhard-float
7434 Generate code that does not use (uses) the floating-point register set.
7435 Software floating point emulation is provided if you use the
7436 @option{-msoft-float} option, and pass the option to GCC when linking.
7439 @itemx -mno-multiple
7441 @opindex mno-multiple
7442 Generate code that uses (does not use) the load multiple word
7443 instructions and the store multiple word instructions. These
7444 instructions are generated by default on POWER systems, and not
7445 generated on PowerPC systems. Do not use @option{-mmultiple} on little
7446 endian PowerPC systems, since those instructions do not work when the
7447 processor is in little endian mode. The exceptions are PPC740 and
7448 PPC750 which permit the instructions usage in little endian mode.
7454 Generate code that uses (does not use) the load string instructions
7455 and the store string word instructions to save multiple registers and
7456 do small block moves. These instructions are generated by default on
7457 POWER systems, and not generated on PowerPC systems. Do not use
7458 @option{-mstring} on little endian PowerPC systems, since those
7459 instructions do not work when the processor is in little endian mode.
7460 The exceptions are PPC740 and PPC750 which permit the instructions
7461 usage in little endian mode.
7467 Generate code that uses (does not use) the load or store instructions
7468 that update the base register to the address of the calculated memory
7469 location. These instructions are generated by default. If you use
7470 @option{-mno-update}, there is a small window between the time that the
7471 stack pointer is updated and the address of the previous frame is
7472 stored, which means code that walks the stack frame across interrupts or
7473 signals may get corrupted data.
7476 @itemx -mno-fused-madd
7477 @opindex mfused-madd
7478 @opindex mno-fused-madd
7479 Generate code that uses (does not use) the floating point multiply and
7480 accumulate instructions. These instructions are generated by default if
7481 hardware floating is used.
7483 @item -mno-bit-align
7485 @opindex mno-bit-align
7487 On System V.4 and embedded PowerPC systems do not (do) force structures
7488 and unions that contain bit-fields to be aligned to the base type of the
7491 For example, by default a structure containing nothing but 8
7492 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7493 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7494 the structure would be aligned to a 1 byte boundary and be one byte in
7497 @item -mno-strict-align
7498 @itemx -mstrict-align
7499 @opindex mno-strict-align
7500 @opindex mstrict-align
7501 On System V.4 and embedded PowerPC systems do not (do) assume that
7502 unaligned memory references will be handled by the system.
7505 @itemx -mno-relocatable
7506 @opindex mrelocatable
7507 @opindex mno-relocatable
7508 On embedded PowerPC systems generate code that allows (does not allow)
7509 the program to be relocated to a different address at runtime. If you
7510 use @option{-mrelocatable} on any module, all objects linked together must
7511 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7513 @item -mrelocatable-lib
7514 @itemx -mno-relocatable-lib
7515 @opindex mrelocatable-lib
7516 @opindex mno-relocatable-lib
7517 On embedded PowerPC systems generate code that allows (does not allow)
7518 the program to be relocated to a different address at runtime. Modules
7519 compiled with @option{-mrelocatable-lib} can be linked with either modules
7520 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7521 with modules compiled with the @option{-mrelocatable} options.
7527 On System V.4 and embedded PowerPC systems do not (do) assume that
7528 register 2 contains a pointer to a global area pointing to the addresses
7529 used in the program.
7532 @itemx -mlittle-endian
7534 @opindex mlittle-endian
7535 On System V.4 and embedded PowerPC systems compile code for the
7536 processor in little endian mode. The @option{-mlittle-endian} option is
7537 the same as @option{-mlittle}.
7542 @opindex mbig-endian
7543 On System V.4 and embedded PowerPC systems compile code for the
7544 processor in big endian mode. The @option{-mbig-endian} option is
7545 the same as @option{-mbig}.
7547 @item -mdynamic-no-pic
7548 @opindex mdynamic-no-pic
7549 On Darwin and Mac OS X systems, compile code so that it is not
7550 relocatable, but that its external references are relocatable. The
7551 resulting code is suitable for applications, but not shared
7554 @item -mprioritize-restricted-insns=@var{priority}
7555 @opindex mprioritize-restricted-insns
7556 This option controls the priority that is assigned to
7557 dispatch-slot restricted instructions during the second scheduling
7558 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
7559 @var{no/highest/second-highest} priority to dispatch slot restricted
7562 @item -msched-costly-dep=@var{dependence_type}
7563 @opindex msched-costly-dep
7564 This option controls which dependences are considered costly
7565 by the target during instruction scheduling. The argument
7566 @var{dependence_type} takes one of the following values:
7567 @var{no}: no dependence is costly,
7568 @var{all}: all dependences are costly,
7569 @var{true_store_to_load}: a true dependence from store to load is costly,
7570 @var{store_to_load}: any dependence from store to load is costly,
7571 @var{number}: any dependence which latency >= @var{number} is costly.
7573 @item -minsert-sched-nops=@var{scheme}
7574 @opindex minsert-sched-nops
7575 This option controls which nop insertion scheme will be used during
7576 the second scheduling pass. The argument @var{scheme} takes one of the
7578 @var{no}: Don't insert nops.
7579 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
7580 according to the scheduler's grouping.
7581 @var{regroup_exact}: Insert nops to force costly dependent insns into
7582 separate groups. Insert exactly as many nops as needed to force an insn
7583 to a new group, according to the estimatied processor grouping.
7584 @var{number}: Insert nops to force costly dependent insns into
7585 separate groups. Insert @var{number} nops to force an insn to a new group.
7589 On System V.4 and embedded PowerPC systems compile code using calling
7590 conventions that adheres to the March 1995 draft of the System V
7591 Application Binary Interface, PowerPC processor supplement. This is the
7592 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7594 @item -mcall-sysv-eabi
7595 @opindex mcall-sysv-eabi
7596 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7598 @item -mcall-sysv-noeabi
7599 @opindex mcall-sysv-noeabi
7600 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7602 @item -mcall-solaris
7603 @opindex mcall-solaris
7604 On System V.4 and embedded PowerPC systems compile code for the Solaris
7608 @opindex mcall-linux
7609 On System V.4 and embedded PowerPC systems compile code for the
7610 Linux-based GNU system.
7614 On System V.4 and embedded PowerPC systems compile code for the
7615 Hurd-based GNU system.
7618 @opindex mcall-netbsd
7619 On System V.4 and embedded PowerPC systems compile code for the
7620 NetBSD operating system.
7622 @item -maix-struct-return
7623 @opindex maix-struct-return
7624 Return all structures in memory (as specified by the AIX ABI)@.
7626 @item -msvr4-struct-return
7627 @opindex msvr4-struct-return
7628 Return structures smaller than 8 bytes in registers (as specified by the
7632 @opindex mabi=altivec
7633 Extend the current ABI with AltiVec ABI extensions. This does not
7634 change the default ABI, instead it adds the AltiVec ABI extensions to
7637 @item -mabi=no-altivec
7638 @opindex mabi=no-altivec
7639 Disable AltiVec ABI extensions for the current ABI.
7642 @itemx -mno-prototype
7644 @opindex mno-prototype
7645 On System V.4 and embedded PowerPC systems assume that all calls to
7646 variable argument functions are properly prototyped. Otherwise, the
7647 compiler must insert an instruction before every non prototyped call to
7648 set or clear bit 6 of the condition code register (@var{CR}) to
7649 indicate whether floating point values were passed in the floating point
7650 registers in case the function takes a variable arguments. With
7651 @option{-mprototype}, only calls to prototyped variable argument functions
7652 will set or clear the bit.
7656 On embedded PowerPC systems, assume that the startup module is called
7657 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7658 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7663 On embedded PowerPC systems, assume that the startup module is called
7664 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7669 On embedded PowerPC systems, assume that the startup module is called
7670 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7674 @opindex myellowknife
7675 On embedded PowerPC systems, assume that the startup module is called
7676 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7681 On System V.4 and embedded PowerPC systems, specify that you are
7682 compiling for a VxWorks system.
7686 Specify that you are compiling for the WindISS simulation environment.
7690 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7691 header to indicate that @samp{eabi} extended relocations are used.
7697 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7698 Embedded Applications Binary Interface (eabi) which is a set of
7699 modifications to the System V.4 specifications. Selecting @option{-meabi}
7700 means that the stack is aligned to an 8 byte boundary, a function
7701 @code{__eabi} is called to from @code{main} to set up the eabi
7702 environment, and the @option{-msdata} option can use both @code{r2} and
7703 @code{r13} to point to two separate small data areas. Selecting
7704 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7705 do not call an initialization function from @code{main}, and the
7706 @option{-msdata} option will only use @code{r13} to point to a single
7707 small data area. The @option{-meabi} option is on by default if you
7708 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7711 @opindex msdata=eabi
7712 On System V.4 and embedded PowerPC systems, put small initialized
7713 @code{const} global and static data in the @samp{.sdata2} section, which
7714 is pointed to by register @code{r2}. Put small initialized
7715 non-@code{const} global and static data in the @samp{.sdata} section,
7716 which is pointed to by register @code{r13}. Put small uninitialized
7717 global and static data in the @samp{.sbss} section, which is adjacent to
7718 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7719 incompatible with the @option{-mrelocatable} option. The
7720 @option{-msdata=eabi} option also sets the @option{-memb} option.
7723 @opindex msdata=sysv
7724 On System V.4 and embedded PowerPC systems, put small global and static
7725 data in the @samp{.sdata} section, which is pointed to by register
7726 @code{r13}. Put small uninitialized global and static data in the
7727 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7728 The @option{-msdata=sysv} option is incompatible with the
7729 @option{-mrelocatable} option.
7731 @item -msdata=default
7733 @opindex msdata=default
7735 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7736 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7737 same as @option{-msdata=sysv}.
7740 @opindex msdata-data
7741 On System V.4 and embedded PowerPC systems, put small global and static
7742 data in the @samp{.sdata} section. Put small uninitialized global and
7743 static data in the @samp{.sbss} section. Do not use register @code{r13}
7744 to address small data however. This is the default behavior unless
7745 other @option{-msdata} options are used.
7749 @opindex msdata=none
7751 On embedded PowerPC systems, put all initialized global and static data
7752 in the @samp{.data} section, and all uninitialized data in the
7753 @samp{.bss} section.
7757 @cindex smaller data references (PowerPC)
7758 @cindex .sdata/.sdata2 references (PowerPC)
7759 On embedded PowerPC systems, put global and static items less than or
7760 equal to @var{num} bytes into the small data or bss sections instead of
7761 the normal data or bss section. By default, @var{num} is 8. The
7762 @option{-G @var{num}} switch is also passed to the linker.
7763 All modules should be compiled with the same @option{-G @var{num}} value.
7766 @itemx -mno-regnames
7768 @opindex mno-regnames
7769 On System V.4 and embedded PowerPC systems do (do not) emit register
7770 names in the assembly language output using symbolic forms.
7773 @itemx -mno-longcall
7775 @opindex mno-longcall
7776 Default to making all function calls via pointers, so that functions
7777 which reside further than 64 megabytes (67,108,864 bytes) from the
7778 current location can be called. This setting can be overridden by the
7779 @code{shortcall} function attribute, or by @code{#pragma longcall(0)}.
7781 Some linkers are capable of detecting out-of-range calls and generating
7782 glue code on the fly. On these systems, long calls are unnecessary and
7783 generate slower code. As of this writing, the AIX linker can do this,
7784 as can the GNU linker for PowerPC/64. It is planned to add this feature
7785 to the GNU linker for 32-bit PowerPC systems as well.
7787 In the future, we may cause GCC to ignore all longcall specifications
7788 when the linker is known to generate glue.
7792 Adds support for multithreading with the @dfn{pthreads} library.
7793 This option sets flags for both the preprocessor and linker.
7797 @node Darwin Options
7798 @subsection Darwin Options
7799 @cindex Darwin options
7801 These options are defined for all architectures running the Darwin operating
7802 system. They are useful for compatibility with other Mac OS compilers.
7807 Loads all members of static archive libraries.
7808 See man ld(1) for more information.
7810 @item -arch_errors_fatal
7811 @opindex arch_errors_fatal
7812 Cause the errors having to do with files that have the wrong architecture
7816 @opindex bind_at_load
7817 Causes the output file to be marked such that the dynamic linker will
7818 bind all undefined references when the file is loaded or launched.
7822 Produce a Mach-o bundle format file.
7823 See man ld(1) for more information.
7825 @item -bundle_loader @var{executable}
7826 @opindex bundle_loader
7827 This specifies the @var{executable} that will be loading the build
7828 output file being linked. See man ld(1) for more information.
7830 @item -allowable_client @var{client_name}
7834 @item -compatibility_version
7835 @item -current_version
7836 @item -dependency-file
7838 @item -dylinker_install_name
7841 @item -exported_symbols_list
7843 @item -flat_namespace
7844 @item -force_cpusubtype_ALL
7845 @item -force_flat_namespace
7846 @item -headerpad_max_install_names
7850 @item -keep_private_externs
7852 @item -multiply_defined
7853 @item -multiply_defined_unused
7855 @item -nofixprebinding
7858 @item -noseglinkedit
7859 @item -pagezero_size
7861 @item -prebind_all_twolevel_modules
7862 @item -private_bundle
7863 @item -read_only_relocs
7865 @item -sectobjectsymbols
7869 @item -sectobjectsymbols
7871 @item -seg_addr_table
7872 @item -seg_addr_table_filename
7875 @item -segs_read_only_addr
7876 @item -segs_read_write_addr
7877 @item -single_module
7881 @item -twolevel_namespace
7884 @item -unexported_symbols_list
7885 @item -weak_reference_mismatches
7888 @opindex allowable_client
7890 @opindex client_name
7891 @opindex compatibility_version
7892 @opindex current_version
7893 @opindex dependency-file
7895 @opindex dylinker_install_name
7898 @opindex exported_symbols_list
7900 @opindex flat_namespace
7901 @opindex force_cpusubtype_ALL
7902 @opindex force_flat_namespace
7903 @opindex headerpad_max_install_names
7906 @opindex install_name
7907 @opindex keep_private_externs
7908 @opindex multi_module
7909 @opindex multiply_defined
7910 @opindex multiply_defined_unused
7912 @opindex nofixprebinding
7913 @opindex nomultidefs
7915 @opindex noseglinkedit
7916 @opindex pagezero_size
7918 @opindex prebind_all_twolevel_modules
7919 @opindex private_bundle
7920 @opindex read_only_relocs
7922 @opindex sectobjectsymbols
7926 @opindex sectobjectsymbols
7928 @opindex seg_addr_table
7929 @opindex seg_addr_table_filename
7930 @opindex seglinkedit
7932 @opindex segs_read_only_addr
7933 @opindex segs_read_write_addr
7934 @opindex single_module
7936 @opindex sub_library
7937 @opindex sub_umbrella
7938 @opindex twolevel_namespace
7941 @opindex unexported_symbols_list
7942 @opindex weak_reference_mismatches
7943 @opindex whatsloaded
7945 This options are available for Darwin linker. Darwin linker man page
7946 describes them in detail.
7951 @subsection IBM RT Options
7953 @cindex IBM RT options
7955 These @samp{-m} options are defined for the IBM RT PC:
7959 @opindex min-line-mul
7960 Use an in-line code sequence for integer multiplies. This is the
7963 @item -mcall-lib-mul
7964 @opindex mcall-lib-mul
7965 Call @code{lmul$$} for integer multiples.
7967 @item -mfull-fp-blocks
7968 @opindex mfull-fp-blocks
7969 Generate full-size floating point data blocks, including the minimum
7970 amount of scratch space recommended by IBM@. This is the default.
7972 @item -mminimum-fp-blocks
7973 @opindex mminimum-fp-blocks
7974 Do not include extra scratch space in floating point data blocks. This
7975 results in smaller code, but slower execution, since scratch space must
7976 be allocated dynamically.
7978 @cindex @file{stdarg.h} and RT PC
7979 @item -mfp-arg-in-fpregs
7980 @opindex mfp-arg-in-fpregs
7981 Use a calling sequence incompatible with the IBM calling convention in
7982 which floating point arguments are passed in floating point registers.
7983 Note that @code{stdarg.h} will not work with floating point operands
7984 if this option is specified.
7986 @item -mfp-arg-in-gregs
7987 @opindex mfp-arg-in-gregs
7988 Use the normal calling convention for floating point arguments. This is
7991 @item -mhc-struct-return
7992 @opindex mhc-struct-return
7993 Return structures of more than one word in memory, rather than in a
7994 register. This provides compatibility with the MetaWare HighC (hc)
7995 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7996 with the Portable C Compiler (pcc).
7998 @item -mnohc-struct-return
7999 @opindex mnohc-struct-return
8000 Return some structures of more than one word in registers, when
8001 convenient. This is the default. For compatibility with the
8002 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
8003 option @option{-mhc-struct-return}.
8007 @subsection MIPS Options
8008 @cindex MIPS options
8010 These @samp{-m} options are defined for the MIPS family of computers:
8014 @item -march=@var{arch}
8016 Generate code that will run on @var{arch}, which can be the name of a
8017 generic MIPS ISA, or the name of a particular processor.
8019 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
8020 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
8021 The processor names are:
8022 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
8024 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
8025 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
8029 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4300},
8030 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
8031 The special value @samp{from-abi} selects the
8032 most compatible architecture for the selected ABI (that is,
8033 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
8035 In processor names, a final @samp{000} can be abbreviated as @samp{k}
8036 (for example, @samp{-march=r2k}). Prefixes are optional, and
8037 @samp{vr} may be written @samp{r}.
8039 GCC defines two macros based on the value of this option. The first
8040 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
8041 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
8042 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
8043 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
8044 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
8046 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
8047 above. In other words, it will have the full prefix and will not
8048 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
8049 the macro names the resolved architecture (either @samp{"mips1"} or
8050 @samp{"mips3"}). It names the default architecture when no
8051 @option{-march} option is given.
8053 @item -mtune=@var{arch}
8055 Optimize for @var{arch}. Among other things, this option controls
8056 the way instructions are scheduled, and the perceived cost of arithmetic
8057 operations. The list of @var{arch} values is the same as for
8060 When this option is not used, GCC will optimize for the processor
8061 specified by @option{-march}. By using @option{-march} and
8062 @option{-mtune} together, it is possible to generate code that will
8063 run on a family of processors, but optimize the code for one
8064 particular member of that family.
8066 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
8067 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
8068 @samp{-march} ones described above.
8072 Equivalent to @samp{-march=mips1}.
8076 Equivalent to @samp{-march=mips2}.
8080 Equivalent to @samp{-march=mips3}.
8084 Equivalent to @samp{-march=mips4}.
8088 Equivalent to @samp{-march=mips32}.
8092 Equivalent to @samp{-march=mips32r2}.
8096 Equivalent to @samp{-march=mips64}.
8099 @itemx -mno-fused-madd
8100 @opindex mfused-madd
8101 @opindex mno-fused-madd
8102 Generate code that uses (does not use) the floating point multiply and
8103 accumulate instructions, when they are available. These instructions
8104 are generated by default if they are available, but this may be
8105 undesirable if the extra precision causes problems or on certain chips
8106 in the mode where denormals are rounded to zero where denormals
8107 generated by multiply and accumulate instructions cause exceptions
8112 Assume that floating point registers are 32 bits wide.
8116 Assume that floating point registers are 64 bits wide.
8120 Assume that general purpose registers are 32 bits wide.
8124 Assume that general purpose registers are 64 bits wide.
8128 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
8129 explanation of the default, and the width of pointers.
8133 Force long types to be 64 bits wide. See @option{-mlong32} for an
8134 explanation of the default, and the width of pointers.
8138 Force long, int, and pointer types to be 32 bits wide.
8140 The default size of ints, longs and pointers depends on the ABI@. All
8141 the supported ABIs use 32-bit ints. The n64 ABI uses 64-bit longs, as
8142 does the 64-bit Cygnus EABI; the others use 32-bit longs. Pointers
8143 are the same size as longs, or the same size as integer registers,
8144 whichever is smaller.
8158 Generate code for the given ABI@.
8160 Note that there are two embedded ABIs: @option{-mabi=eabi}
8161 selects the one defined by Cygnus while @option{-meabi=meabi}
8162 selects the one defined by MIPS@. Both these ABIs have
8163 32-bit and 64-bit variants. Normally, GCC will generate
8164 64-bit code when you select a 64-bit architecture, but you
8165 can use @option{-mgp32} to get 32-bit code instead.
8167 @item -mabi-fake-default
8168 @opindex mabi-fake-default
8169 You don't want to know what this option does. No, really. I mean
8170 it. Move on to the next option.
8172 What? You're still here? Oh, well@enddots{} Ok, here's the deal. GCC
8173 wants the default set of options to get the root of the multilib tree,
8174 and the shared library SONAMEs without any multilib-indicating
8175 suffixes. This is not convenience for @samp{mips64-linux-gnu}, since
8176 we want to default to the N32 ABI, while still being binary-compatible
8177 with @samp{mips-linux-gnu} if you stick to the O32 ABI@. Being
8178 binary-compatible means shared libraries should have the same SONAMEs,
8179 and libraries should live in the same location. Having O32 libraries
8180 in a sub-directory named say @file{o32} is not acceptable.
8182 So we trick GCC into believing that O32 is the default ABI, except
8183 that we override the default with some internal command-line
8184 processing magic. Problem is, if we stopped at that, and you then
8185 created a multilib-aware package that used the output of @command{gcc
8186 -print-multi-lib} to decide which multilibs to build, and how, and
8187 you'd find yourself in an awkward situation when you found out that
8188 some of the options listed ended up mapping to the same multilib, and
8189 none of your libraries was actually built for the multilib that
8190 @option{-print-multi-lib} claims to be the default. So we added this
8191 option that disables the default switcher, falling back to GCC's
8192 original notion of the default library. Confused yet?
8194 For short: don't ever use this option, unless you find it in the list
8195 of additional options to be used when building for multilibs, in the
8196 output of @option{gcc -print-multi-lib}.
8200 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
8201 add normal debug information. This is the default for all
8202 platforms except for the OSF/1 reference platform, using the OSF/rose
8203 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
8204 switches are used, the @file{mips-tfile} program will encapsulate the
8205 stabs within MIPS ECOFF@.
8209 Generate code for the GNU assembler. This is the default on the OSF/1
8210 reference platform, using the OSF/rose object format. Also, this is
8211 the default if the configure option @option{--with-gnu-as} is used.
8213 @item -msplit-addresses
8214 @itemx -mno-split-addresses
8215 @opindex msplit-addresses
8216 @opindex mno-split-addresses
8217 Generate code to load the high and low parts of address constants separately.
8218 This allows GCC to optimize away redundant loads of the high order
8219 bits of addresses. This optimization requires GNU as and GNU ld.
8220 This optimization is enabled by default for some embedded targets where
8221 GNU as and GNU ld are standard.
8227 The @option{-mrnames} switch says to output code using the MIPS software
8228 names for the registers, instead of the hardware names (ie, @var{a0}
8229 instead of @var{$4}). The only known assembler that supports this option
8230 is the Algorithmics assembler.
8236 The @option{-mmemcpy} switch makes all block moves call the appropriate
8237 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
8238 generating inline code.
8241 @itemx -mno-mips-tfile
8242 @opindex mmips-tfile
8243 @opindex mno-mips-tfile
8244 The @option{-mno-mips-tfile} switch causes the compiler not
8245 postprocess the object file with the @file{mips-tfile} program,
8246 after the MIPS assembler has generated it to add debug support. If
8247 @file{mips-tfile} is not run, then no local variables will be
8248 available to the debugger. In addition, @file{stage2} and
8249 @file{stage3} objects will have the temporary file names passed to the
8250 assembler embedded in the object file, which means the objects will
8251 not compare the same. The @option{-mno-mips-tfile} switch should only
8252 be used when there are bugs in the @file{mips-tfile} program that
8253 prevents compilation.
8256 @opindex msoft-float
8257 Generate output containing library calls for floating point.
8258 @strong{Warning:} the requisite libraries are not part of GCC@.
8259 Normally the facilities of the machine's usual C compiler are used, but
8260 this can't be done directly in cross-compilation. You must make your
8261 own arrangements to provide suitable library functions for
8265 @opindex mhard-float
8266 Generate output containing floating point instructions. This is the
8267 default if you use the unmodified sources.
8270 @itemx -mno-abicalls
8272 @opindex mno-abicalls
8273 Emit (or do not emit) the pseudo operations @samp{.abicalls},
8274 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
8275 position independent code.
8281 Lift (or do not lift) the usual restrictions on the size of the global
8284 GCC normally uses a single instruction to load values from the GOT.
8285 While this is relatively efficient, it will only work if the GOT
8286 is smaller than about 64k. Anything larger will cause the linker
8287 to report an error such as:
8289 @cindex relocation truncated to fit (MIPS)
8291 relocation truncated to fit: R_MIPS_GOT16 foobar
8294 If this happens, you should recompile your code with @option{-mxgot}.
8295 It should then work with very large GOTs, although it will also be
8296 less efficient, since it will take three instructions to fetch the
8297 value of a global symbol.
8299 Note that some linkers can create multiple GOTs. If you have such a
8300 linker, you should only need to use @option{-mxgot} when a single object
8301 file accesses more than 64k's worth of GOT entries. Very few do.
8303 These options have no effect unless GCC is generating position
8307 @itemx -mno-long-calls
8308 @opindex mlong-calls
8309 @opindex mno-long-calls
8310 Do all calls with the @samp{JALR} instruction, which requires
8311 loading up a function's address into a register before the call.
8312 You need to use this switch, if you call outside of the current
8313 512 megabyte segment to functions that are not through pointers.
8315 @item -membedded-pic
8316 @itemx -mno-embedded-pic
8317 @opindex membedded-pic
8318 @opindex mno-embedded-pic
8319 Generate PIC code suitable for some embedded systems. All calls are
8320 made using PC relative address, and all data is addressed using the $gp
8321 register. No more than 65536 bytes of global data may be used. This
8322 requires GNU as and GNU ld which do most of the work. This currently
8323 only works on targets which use ECOFF; it does not work with ELF@.
8325 @item -membedded-data
8326 @itemx -mno-embedded-data
8327 @opindex membedded-data
8328 @opindex mno-embedded-data
8329 Allocate variables to the read-only data section first if possible, then
8330 next in the small data section if possible, otherwise in data. This gives
8331 slightly slower code than the default, but reduces the amount of RAM required
8332 when executing, and thus may be preferred for some embedded systems.
8334 @item -muninit-const-in-rodata
8335 @itemx -mno-uninit-const-in-rodata
8336 @opindex muninit-const-in-rodata
8337 @opindex mno-uninit-const-in-rodata
8338 When used together with @option{-membedded-data}, it will always store uninitialized
8339 const variables in the read-only data section.
8341 @item -msingle-float
8342 @itemx -mdouble-float
8343 @opindex msingle-float
8344 @opindex mdouble-float
8345 The @option{-msingle-float} switch tells gcc to assume that the floating
8346 point coprocessor only supports single precision operations, as on the
8347 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
8348 double precision operations. This is the default.
8354 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
8355 as on the @samp{r4650} chip.
8359 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
8360 @option{-mcpu=r4650}.
8366 Enable 16-bit instructions.
8370 Compile code for the processor in little endian mode.
8371 The requisite libraries are assumed to exist.
8375 Compile code for the processor in big endian mode.
8376 The requisite libraries are assumed to exist.
8380 @cindex smaller data references (MIPS)
8381 @cindex gp-relative references (MIPS)
8382 Put global and static items less than or equal to @var{num} bytes into
8383 the small data or bss sections instead of the normal data or bss
8384 section. This allows the assembler to emit one word memory reference
8385 instructions based on the global pointer (@var{gp} or @var{$28}),
8386 instead of the normal two words used. By default, @var{num} is 8 when
8387 the MIPS assembler is used, and 0 when the GNU assembler is used. The
8388 @option{-G @var{num}} switch is also passed to the assembler and linker.
8389 All modules should be compiled with the same @option{-G @var{num}}
8394 Tell the MIPS assembler to not run its preprocessor over user
8395 assembler files (with a @samp{.s} suffix) when assembling them.
8399 Pass an option to gas which will cause nops to be inserted if
8400 the read of the destination register of an mfhi or mflo instruction
8401 occurs in the following two instructions.
8406 Work around certain SB-1 CPU core errata.
8407 (This flag currently works around the SB-1 revision 2
8408 ``F1'' and ``F2'' floating point errata.)
8412 Do not include the default crt0.
8414 @item -mflush-func=@var{func}
8415 @itemx -mno-flush-func
8416 @opindex mflush-func
8417 Specifies the function to call to flush the I and D caches, or to not
8418 call any such function. If called, the function must take the same
8419 arguments as the common @code{_flush_func()}, that is, the address of the
8420 memory range for which the cache is being flushed, the size of the
8421 memory range, and the number 3 (to flush both caches). The default
8422 depends on the target gcc was configured for, but commonly is either
8423 @samp{_flush_func} or @samp{__cpu_flush}.
8425 @item -mbranch-likely
8426 @itemx -mno-branch-likely
8427 @opindex mbranch-likely
8428 @opindex mno-branch-likely
8429 Enable or disable use of Branch Likely instructions, regardless of the
8430 default for the selected architecture. By default, Branch Likely
8431 instructions may be generated if they are supported by the selected
8432 architecture. An exception is for the MIPS32 and MIPS64 architectures
8433 and processors which implement those architectures; for those, Branch
8434 Likely instructions will not be generated by default because the MIPS32
8435 and MIPS64 architectures specifically deprecate their use.
8438 @node i386 and x86-64 Options
8439 @subsection Intel 386 and AMD x86-64 Options
8440 @cindex i386 Options
8441 @cindex x86-64 Options
8442 @cindex Intel 386 Options
8443 @cindex AMD x86-64 Options
8445 These @samp{-m} options are defined for the i386 and x86-64 family of
8449 @item -mtune=@var{cpu-type}
8451 Tune to @var{cpu-type} everything applicable about the generated code, except
8452 for the ABI and the set of available instructions. The choices for
8453 @var{cpu-type} are @samp{i386}, @samp{i486}, @samp{i586}, @samp{i686},
8454 @samp{pentium}, @samp{pentium-mmx}, @samp{pentiumpro}, @samp{pentium2},
8455 @samp{pentium3}, @samp{pentium4}, @samp{k6}, @samp{k6-2}, @samp{k6-3},
8456 @samp{athlon}, @samp{athlon-tbird}, @samp{athlon-4}, @samp{athlon-xp},
8457 @samp{athlon-mp}, @samp{winchip-c6}, @samp{winchip2}, @samp{k8}, @samp{c3}
8460 While picking a specific @var{cpu-type} will schedule things appropriately
8461 for that particular chip, the compiler will not generate any code that
8462 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8463 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
8464 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
8465 AMD chips as opposed to the Intel ones.
8467 @item -march=@var{cpu-type}
8469 Generate instructions for the machine type @var{cpu-type}. The choices
8470 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8471 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8473 @item -mcpu=@var{cpu-type}
8475 A deprecated synonym for @option{-mtune}.
8484 @opindex mpentiumpro
8485 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8486 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8487 These synonyms are deprecated.
8489 @item -mfpmath=@var{unit}
8491 generate floating point arithmetics for selected unit @var{unit}. the choices
8496 Use the standard 387 floating point coprocessor present majority of chips and
8497 emulated otherwise. Code compiled with this option will run almost everywhere.
8498 The temporary results are computed in 80bit precision instead of precision
8499 specified by the type resulting in slightly different results compared to most
8500 of other chips. See @option{-ffloat-store} for more detailed description.
8502 This is the default choice for i386 compiler.
8505 Use scalar floating point instructions present in the SSE instruction set.
8506 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8507 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8508 instruction set supports only single precision arithmetics, thus the double and
8509 extended precision arithmetics is still done using 387. Later version, present
8510 only in Pentium4 and the future AMD x86-64 chips supports double precision
8513 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8514 @option{-msse2} switches to enable SSE extensions and make this option
8515 effective. For x86-64 compiler, these extensions are enabled by default.
8517 The resulting code should be considerably faster in majority of cases and avoid
8518 the numerical instability problems of 387 code, but may break some existing
8519 code that expects temporaries to be 80bit.
8521 This is the default choice for x86-64 compiler.
8524 Use all SSE extensions enabled by @option{-msse2} as well as the new
8525 SSE extensions in Prescott New Instructions. @option{-mpni} also
8526 enables 2 builtin functions, @code{__builtin_ia32_monitor} and
8527 @code{__builtin_ia32_mwait}, for new instructions @code{monitor} and
8531 Attempt to utilize both instruction sets at once. This effectively double the
8532 amount of available registers and on chips with separate execution units for
8533 387 and SSE the execution resources too. Use this option with care, as it is
8534 still experimental, because gcc register allocator does not model separate
8535 functional units well resulting in instable performance.
8538 @item -masm=@var{dialect}
8539 @opindex masm=@var{dialect}
8540 Output asm instructions using selected @var{dialect}. Supported choices are
8541 @samp{intel} or @samp{att} (the default one).
8546 @opindex mno-ieee-fp
8547 Control whether or not the compiler uses IEEE floating point
8548 comparisons. These handle correctly the case where the result of a
8549 comparison is unordered.
8552 @opindex msoft-float
8553 Generate output containing library calls for floating point.
8554 @strong{Warning:} the requisite libraries are not part of GCC@.
8555 Normally the facilities of the machine's usual C compiler are used, but
8556 this can't be done directly in cross-compilation. You must make your
8557 own arrangements to provide suitable library functions for
8560 On machines where a function returns floating point results in the 80387
8561 register stack, some floating point opcodes may be emitted even if
8562 @option{-msoft-float} is used.
8564 @item -mno-fp-ret-in-387
8565 @opindex mno-fp-ret-in-387
8566 Do not use the FPU registers for return values of functions.
8568 The usual calling convention has functions return values of types
8569 @code{float} and @code{double} in an FPU register, even if there
8570 is no FPU@. The idea is that the operating system should emulate
8573 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8574 in ordinary CPU registers instead.
8576 @item -mno-fancy-math-387
8577 @opindex mno-fancy-math-387
8578 Some 387 emulators do not support the @code{sin}, @code{cos} and
8579 @code{sqrt} instructions for the 387. Specify this option to avoid
8580 generating those instructions. This option is the default on FreeBSD,
8581 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8582 indicates that the target cpu will always have an FPU and so the
8583 instruction will not need emulation. As of revision 2.6.1, these
8584 instructions are not generated unless you also use the
8585 @option{-funsafe-math-optimizations} switch.
8587 @item -malign-double
8588 @itemx -mno-align-double
8589 @opindex malign-double
8590 @opindex mno-align-double
8591 Control whether GCC aligns @code{double}, @code{long double}, and
8592 @code{long long} variables on a two word boundary or a one word
8593 boundary. Aligning @code{double} variables on a two word boundary will
8594 produce code that runs somewhat faster on a @samp{Pentium} at the
8595 expense of more memory.
8597 @strong{Warning:} if you use the @option{-malign-double} switch,
8598 structures containing the above types will be aligned differently than
8599 the published application binary interface specifications for the 386
8600 and will not be binary compatible with structures in code compiled
8601 without that switch.
8603 @item -m96bit-long-double
8604 @item -m128bit-long-double
8605 @opindex m96bit-long-double
8606 @opindex m128bit-long-double
8607 These switches control the size of @code{long double} type. The i386
8608 application binary interface specifies the size to be 96 bits,
8609 so @option{-m96bit-long-double} is the default in 32 bit mode.
8611 Modern architectures (Pentium and newer) would prefer @code{long double}
8612 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8613 conforming to the ABI, this would not be possible. So specifying a
8614 @option{-m128bit-long-double} will align @code{long double}
8615 to a 16 byte boundary by padding the @code{long double} with an additional
8618 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8619 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8621 Notice that neither of these options enable any extra precision over the x87
8622 standard of 80 bits for a @code{long double}.
8624 @strong{Warning:} if you override the default value for your target ABI, the
8625 structures and arrays containing @code{long double} will change their size as
8626 well as function calling convention for function taking @code{long double}
8627 will be modified. Hence they will not be binary compatible with arrays or
8628 structures in code compiled without that switch.
8632 @itemx -mno-svr3-shlib
8633 @opindex msvr3-shlib
8634 @opindex mno-svr3-shlib
8635 Control whether GCC places uninitialized local variables into the
8636 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8637 into @code{bss}. These options are meaningful only on System V Release 3.
8641 Use a different function-calling convention, in which functions that
8642 take a fixed number of arguments return with the @code{ret} @var{num}
8643 instruction, which pops their arguments while returning. This saves one
8644 instruction in the caller since there is no need to pop the arguments
8647 You can specify that an individual function is called with this calling
8648 sequence with the function attribute @samp{stdcall}. You can also
8649 override the @option{-mrtd} option by using the function attribute
8650 @samp{cdecl}. @xref{Function Attributes}.
8652 @strong{Warning:} this calling convention is incompatible with the one
8653 normally used on Unix, so you cannot use it if you need to call
8654 libraries compiled with the Unix compiler.
8656 Also, you must provide function prototypes for all functions that
8657 take variable numbers of arguments (including @code{printf});
8658 otherwise incorrect code will be generated for calls to those
8661 In addition, seriously incorrect code will result if you call a
8662 function with too many arguments. (Normally, extra arguments are
8663 harmlessly ignored.)
8665 @item -mregparm=@var{num}
8667 Control how many registers are used to pass integer arguments. By
8668 default, no registers are used to pass arguments, and at most 3
8669 registers can be used. You can control this behavior for a specific
8670 function by using the function attribute @samp{regparm}.
8671 @xref{Function Attributes}.
8673 @strong{Warning:} if you use this switch, and
8674 @var{num} is nonzero, then you must build all modules with the same
8675 value, including any libraries. This includes the system libraries and
8678 @item -mpreferred-stack-boundary=@var{num}
8679 @opindex mpreferred-stack-boundary
8680 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8681 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8682 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8683 size (@option{-Os}), in which case the default is the minimum correct
8684 alignment (4 bytes for x86, and 8 bytes for x86-64).
8686 On Pentium and PentiumPro, @code{double} and @code{long double} values
8687 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8688 suffer significant run time performance penalties. On Pentium III, the
8689 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8690 penalties if it is not 16 byte aligned.
8692 To ensure proper alignment of this values on the stack, the stack boundary
8693 must be as aligned as that required by any value stored on the stack.
8694 Further, every function must be generated such that it keeps the stack
8695 aligned. Thus calling a function compiled with a higher preferred
8696 stack boundary from a function compiled with a lower preferred stack
8697 boundary will most likely misalign the stack. It is recommended that
8698 libraries that use callbacks always use the default setting.
8700 This extra alignment does consume extra stack space, and generally
8701 increases code size. Code that is sensitive to stack space usage, such
8702 as embedded systems and operating system kernels, may want to reduce the
8703 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8721 These switches enable or disable the use of built-in functions that allow
8722 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
8724 @xref{X86 Built-in Functions}, for details of the functions enabled
8725 and disabled by these switches.
8727 To have SSE/SSE2 instructions generated automatically from floating-point
8728 code, see @option{-mfpmath=sse}.
8731 @itemx -mno-push-args
8733 @opindex mno-push-args
8734 Use PUSH operations to store outgoing parameters. This method is shorter
8735 and usually equally fast as method using SUB/MOV operations and is enabled
8736 by default. In some cases disabling it may improve performance because of
8737 improved scheduling and reduced dependencies.
8739 @item -maccumulate-outgoing-args
8740 @opindex maccumulate-outgoing-args
8741 If enabled, the maximum amount of space required for outgoing arguments will be
8742 computed in the function prologue. This is faster on most modern CPUs
8743 because of reduced dependencies, improved scheduling and reduced stack usage
8744 when preferred stack boundary is not equal to 2. The drawback is a notable
8745 increase in code size. This switch implies @option{-mno-push-args}.
8749 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8750 on thread-safe exception handling must compile and link all code with the
8751 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8752 @option{-D_MT}; when linking, it links in a special thread helper library
8753 @option{-lmingwthrd} which cleans up per thread exception handling data.
8755 @item -mno-align-stringops
8756 @opindex mno-align-stringops
8757 Do not align destination of inlined string operations. This switch reduces
8758 code size and improves performance in case the destination is already aligned,
8759 but gcc don't know about it.
8761 @item -minline-all-stringops
8762 @opindex minline-all-stringops
8763 By default GCC inlines string operations only when destination is known to be
8764 aligned at least to 4 byte boundary. This enables more inlining, increase code
8765 size, but may improve performance of code that depends on fast memcpy, strlen
8766 and memset for short lengths.
8768 @item -momit-leaf-frame-pointer
8769 @opindex momit-leaf-frame-pointer
8770 Don't keep the frame pointer in a register for leaf functions. This
8771 avoids the instructions to save, set up and restore frame pointers and
8772 makes an extra register available in leaf functions. The option
8773 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8774 which might make debugging harder.
8776 @item -mtls-direct-seg-refs
8777 @itemx -mno-tls-direct-seg-refs
8778 @opindex mtls-direct-seg-refs
8779 Controls whether TLS variables may be accessed with offsets from the
8780 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8781 or whether the thread base pointer must be added. Whether or not this
8782 is legal depends on the operating system, and whether it maps the
8783 segment to cover the entire TLS area.
8785 For systems that use GNU libc, the default is on.
8788 These @samp{-m} switches are supported in addition to the above
8789 on AMD x86-64 processors in 64-bit environments.
8796 Generate code for a 32-bit or 64-bit environment.
8797 The 32-bit environment sets int, long and pointer to 32 bits and
8798 generates code that runs on any i386 system.
8799 The 64-bit environment sets int to 32 bits and long and pointer
8800 to 64 bits and generates code for AMD's x86-64 architecture.
8803 @opindex no-red-zone
8804 Do not use a so called red zone for x86-64 code. The red zone is mandated
8805 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8806 stack pointer that will not be modified by signal or interrupt handlers
8807 and therefore can be used for temporary data without adjusting the stack
8808 pointer. The flag @option{-mno-red-zone} disables this red zone.
8810 @item -mcmodel=small
8811 @opindex mcmodel=small
8812 Generate code for the small code model: the program and its symbols must
8813 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8814 Programs can be statically or dynamically linked. This is the default
8817 @item -mcmodel=kernel
8818 @opindex mcmodel=kernel
8819 Generate code for the kernel code model. The kernel runs in the
8820 negative 2 GB of the address space.
8821 This model has to be used for Linux kernel code.
8823 @item -mcmodel=medium
8824 @opindex mcmodel=medium
8825 Generate code for the medium model: The program is linked in the lower 2
8826 GB of the address space but symbols can be located anywhere in the
8827 address space. Programs can be statically or dynamically linked, but
8828 building of shared libraries are not supported with the medium model.
8830 @item -mcmodel=large
8831 @opindex mcmodel=large
8832 Generate code for the large model: This model makes no assumptions
8833 about addresses and sizes of sections. Currently GCC does not implement
8838 @subsection HPPA Options
8839 @cindex HPPA Options
8841 These @samp{-m} options are defined for the HPPA family of computers:
8844 @item -march=@var{architecture-type}
8846 Generate code for the specified architecture. The choices for
8847 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8848 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8849 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8850 architecture option for your machine. Code compiled for lower numbered
8851 architectures will run on higher numbered architectures, but not the
8854 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8855 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8859 @itemx -mpa-risc-1-1
8860 @itemx -mpa-risc-2-0
8861 @opindex mpa-risc-1-0
8862 @opindex mpa-risc-1-1
8863 @opindex mpa-risc-2-0
8864 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8867 @opindex mbig-switch
8868 Generate code suitable for big switch tables. Use this option only if
8869 the assembler/linker complain about out of range branches within a switch
8872 @item -mjump-in-delay
8873 @opindex mjump-in-delay
8874 Fill delay slots of function calls with unconditional jump instructions
8875 by modifying the return pointer for the function call to be the target
8876 of the conditional jump.
8878 @item -mdisable-fpregs
8879 @opindex mdisable-fpregs
8880 Prevent floating point registers from being used in any manner. This is
8881 necessary for compiling kernels which perform lazy context switching of
8882 floating point registers. If you use this option and attempt to perform
8883 floating point operations, the compiler will abort.
8885 @item -mdisable-indexing
8886 @opindex mdisable-indexing
8887 Prevent the compiler from using indexing address modes. This avoids some
8888 rather obscure problems when compiling MIG generated code under MACH@.
8890 @item -mno-space-regs
8891 @opindex mno-space-regs
8892 Generate code that assumes the target has no space registers. This allows
8893 GCC to generate faster indirect calls and use unscaled index address modes.
8895 Such code is suitable for level 0 PA systems and kernels.
8897 @item -mfast-indirect-calls
8898 @opindex mfast-indirect-calls
8899 Generate code that assumes calls never cross space boundaries. This
8900 allows GCC to emit code which performs faster indirect calls.
8902 This option will not work in the presence of shared libraries or nested
8905 @item -mlong-load-store
8906 @opindex mlong-load-store
8907 Generate 3-instruction load and store sequences as sometimes required by
8908 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8911 @item -mportable-runtime
8912 @opindex mportable-runtime
8913 Use the portable calling conventions proposed by HP for ELF systems.
8917 Enable the use of assembler directives only GAS understands.
8919 @item -mschedule=@var{cpu-type}
8921 Schedule code according to the constraints for the machine type
8922 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8923 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8924 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8925 proper scheduling option for your machine. The default scheduling is
8929 @opindex mlinker-opt
8930 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8931 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8932 linkers in which they give bogus error messages when linking some programs.
8935 @opindex msoft-float
8936 Generate output containing library calls for floating point.
8937 @strong{Warning:} the requisite libraries are not available for all HPPA
8938 targets. Normally the facilities of the machine's usual C compiler are
8939 used, but this cannot be done directly in cross-compilation. You must make
8940 your own arrangements to provide suitable library functions for
8941 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8942 does provide software floating point support.
8944 @option{-msoft-float} changes the calling convention in the output file;
8945 therefore, it is only useful if you compile @emph{all} of a program with
8946 this option. In particular, you need to compile @file{libgcc.a}, the
8947 library that comes with GCC, with @option{-msoft-float} in order for
8952 Generate the predefine, @code{_SIO}, for server IO. The default is
8953 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8954 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8955 options are available under HP-UX and HI-UX.
8959 Use GNU ld specific options. This passes @option{-shared} to ld when
8960 building a shared library. It is the default when GCC is configured,
8961 explicitly or implicitly, with the GNU linker. This option does not
8962 have any affect on which ld is called, it only changes what parameters
8963 are passed to that ld. The ld that is called is determined by the
8964 @option{--with-ld} configure option, gcc's program search path, and
8965 finally by the user's @env{PATH}. The linker used by GCC can be printed
8966 using @samp{which `gcc -print-prog-name=ld`}.
8970 Use HP ld specific options. This passes @option{-b} to ld when building
8971 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8972 links. It is the default when GCC is configured, explicitly or
8973 implicitly, with the HP linker. This option does not have any affect on
8974 which ld is called, it only changes what parameters are passed to that
8975 ld. The ld that is called is determined by the @option{--with-ld}
8976 configure option, gcc's program search path, and finally by the user's
8977 @env{PATH}. The linker used by GCC can be printed using @samp{which
8978 `gcc -print-prog-name=ld`}.
8981 @opindex mno-long-calls
8982 Generate code that uses long call sequences. This ensures that a call
8983 is always able to reach linker generated stubs. The default is to generate
8984 long calls only when the distance from the call site to the beginning
8985 of the function or translation unit, as the case may be, exceeds a
8986 predefined limit set by the branch type being used. The limits for
8987 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8988 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8991 Distances are measured from the beginning of functions when using the
8992 @option{-ffunction-sections} option, or when using the @option{-mgas}
8993 and @option{-mno-portable-runtime} options together under HP-UX with
8996 It is normally not desirable to use this option as it will degrade
8997 performance. However, it may be useful in large applications,
8998 particularly when partial linking is used to build the application.
9000 The types of long calls used depends on the capabilities of the
9001 assembler and linker, and the type of code being generated. The
9002 impact on systems that support long absolute calls, and long pic
9003 symbol-difference or pc-relative calls should be relatively small.
9004 However, an indirect call is used on 32-bit ELF systems in pic code
9005 and it is quite long.
9009 Suppress the generation of link options to search libdld.sl when the
9010 @option{-static} option is specified on HP-UX 10 and later.
9014 The HP-UX implementation of setlocale in libc has a dependency on
9015 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9016 when the @option{-static} option is specified, special link options
9017 are needed to resolve this dependency.
9019 On HP-UX 10 and later, the GCC driver adds the necessary options to
9020 link with libdld.sl when the @option{-static} option is specified.
9021 This causes the resulting binary to be dynamic. On the 64-bit port,
9022 the linkers generate dynamic binaries by default in any case. The
9023 @option{-nolibdld} option can be used to prevent the GCC driver from
9024 adding these link options.
9028 Add support for multithreading with the @dfn{dce thread} library
9029 under HP-UX. This option sets flags for both the preprocessor and
9033 @node Intel 960 Options
9034 @subsection Intel 960 Options
9036 These @samp{-m} options are defined for the Intel 960 implementations:
9039 @item -m@var{cpu-type}
9047 Assume the defaults for the machine type @var{cpu-type} for some of
9048 the other options, including instruction scheduling, floating point
9049 support, and addressing modes. The choices for @var{cpu-type} are
9050 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
9051 @samp{sa}, and @samp{sb}.
9058 @opindex msoft-float
9059 The @option{-mnumerics} option indicates that the processor does support
9060 floating-point instructions. The @option{-msoft-float} option indicates
9061 that floating-point support should not be assumed.
9063 @item -mleaf-procedures
9064 @itemx -mno-leaf-procedures
9065 @opindex mleaf-procedures
9066 @opindex mno-leaf-procedures
9067 Do (or do not) attempt to alter leaf procedures to be callable with the
9068 @code{bal} instruction as well as @code{call}. This will result in more
9069 efficient code for explicit calls when the @code{bal} instruction can be
9070 substituted by the assembler or linker, but less efficient code in other
9071 cases, such as calls via function pointers, or using a linker that doesn't
9072 support this optimization.
9075 @itemx -mno-tail-call
9077 @opindex mno-tail-call
9078 Do (or do not) make additional attempts (beyond those of the
9079 machine-independent portions of the compiler) to optimize tail-recursive
9080 calls into branches. You may not want to do this because the detection of
9081 cases where this is not valid is not totally complete. The default is
9082 @option{-mno-tail-call}.
9084 @item -mcomplex-addr
9085 @itemx -mno-complex-addr
9086 @opindex mcomplex-addr
9087 @opindex mno-complex-addr
9088 Assume (or do not assume) that the use of a complex addressing mode is a
9089 win on this implementation of the i960. Complex addressing modes may not
9090 be worthwhile on the K-series, but they definitely are on the C-series.
9091 The default is currently @option{-mcomplex-addr} for all processors except
9095 @itemx -mno-code-align
9096 @opindex mcode-align
9097 @opindex mno-code-align
9098 Align code to 8-byte boundaries for faster fetching (or don't bother).
9099 Currently turned on by default for C-series implementations only.
9102 @item -mclean-linkage
9103 @itemx -mno-clean-linkage
9104 @opindex mclean-linkage
9105 @opindex mno-clean-linkage
9106 These options are not fully implemented.
9110 @itemx -mic2.0-compat
9111 @itemx -mic3.0-compat
9113 @opindex mic2.0-compat
9114 @opindex mic3.0-compat
9115 Enable compatibility with iC960 v2.0 or v3.0.
9119 @opindex masm-compat
9121 Enable compatibility with the iC960 assembler.
9123 @item -mstrict-align
9124 @itemx -mno-strict-align
9125 @opindex mstrict-align
9126 @opindex mno-strict-align
9127 Do not permit (do permit) unaligned accesses.
9131 Enable structure-alignment compatibility with Intel's gcc release version
9132 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
9134 @item -mlong-double-64
9135 @opindex mlong-double-64
9136 Implement type @samp{long double} as 64-bit floating point numbers.
9137 Without the option @samp{long double} is implemented by 80-bit
9138 floating point numbers. The only reason we have it because there is
9139 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
9140 is only useful for people using soft-float targets. Otherwise, we
9141 should recommend against use of it.
9145 @node DEC Alpha Options
9146 @subsection DEC Alpha Options
9148 These @samp{-m} options are defined for the DEC Alpha implementations:
9151 @item -mno-soft-float
9153 @opindex mno-soft-float
9154 @opindex msoft-float
9155 Use (do not use) the hardware floating-point instructions for
9156 floating-point operations. When @option{-msoft-float} is specified,
9157 functions in @file{libgcc.a} will be used to perform floating-point
9158 operations. Unless they are replaced by routines that emulate the
9159 floating-point operations, or compiled in such a way as to call such
9160 emulations routines, these routines will issue floating-point
9161 operations. If you are compiling for an Alpha without floating-point
9162 operations, you must ensure that the library is built so as not to call
9165 Note that Alpha implementations without floating-point operations are
9166 required to have floating-point registers.
9171 @opindex mno-fp-regs
9172 Generate code that uses (does not use) the floating-point register set.
9173 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9174 register set is not used, floating point operands are passed in integer
9175 registers as if they were integers and floating-point results are passed
9176 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9177 so any function with a floating-point argument or return value called by code
9178 compiled with @option{-mno-fp-regs} must also be compiled with that
9181 A typical use of this option is building a kernel that does not use,
9182 and hence need not save and restore, any floating-point registers.
9186 The Alpha architecture implements floating-point hardware optimized for
9187 maximum performance. It is mostly compliant with the IEEE floating
9188 point standard. However, for full compliance, software assistance is
9189 required. This option generates code fully IEEE compliant code
9190 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9191 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9192 defined during compilation. The resulting code is less efficient but is
9193 able to correctly support denormalized numbers and exceptional IEEE
9194 values such as not-a-number and plus/minus infinity. Other Alpha
9195 compilers call this option @option{-ieee_with_no_inexact}.
9197 @item -mieee-with-inexact
9198 @opindex mieee-with-inexact
9199 This is like @option{-mieee} except the generated code also maintains
9200 the IEEE @var{inexact-flag}. Turning on this option causes the
9201 generated code to implement fully-compliant IEEE math. In addition to
9202 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9203 macro. On some Alpha implementations the resulting code may execute
9204 significantly slower than the code generated by default. Since there is
9205 very little code that depends on the @var{inexact-flag}, you should
9206 normally not specify this option. Other Alpha compilers call this
9207 option @option{-ieee_with_inexact}.
9209 @item -mfp-trap-mode=@var{trap-mode}
9210 @opindex mfp-trap-mode
9211 This option controls what floating-point related traps are enabled.
9212 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9213 The trap mode can be set to one of four values:
9217 This is the default (normal) setting. The only traps that are enabled
9218 are the ones that cannot be disabled in software (e.g., division by zero
9222 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9226 Like @samp{su}, but the instructions are marked to be safe for software
9227 completion (see Alpha architecture manual for details).
9230 Like @samp{su}, but inexact traps are enabled as well.
9233 @item -mfp-rounding-mode=@var{rounding-mode}
9234 @opindex mfp-rounding-mode
9235 Selects the IEEE rounding mode. Other Alpha compilers call this option
9236 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9241 Normal IEEE rounding mode. Floating point numbers are rounded towards
9242 the nearest machine number or towards the even machine number in case
9246 Round towards minus infinity.
9249 Chopped rounding mode. Floating point numbers are rounded towards zero.
9252 Dynamic rounding mode. A field in the floating point control register
9253 (@var{fpcr}, see Alpha architecture reference manual) controls the
9254 rounding mode in effect. The C library initializes this register for
9255 rounding towards plus infinity. Thus, unless your program modifies the
9256 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9259 @item -mtrap-precision=@var{trap-precision}
9260 @opindex mtrap-precision
9261 In the Alpha architecture, floating point traps are imprecise. This
9262 means without software assistance it is impossible to recover from a
9263 floating trap and program execution normally needs to be terminated.
9264 GCC can generate code that can assist operating system trap handlers
9265 in determining the exact location that caused a floating point trap.
9266 Depending on the requirements of an application, different levels of
9267 precisions can be selected:
9271 Program precision. This option is the default and means a trap handler
9272 can only identify which program caused a floating point exception.
9275 Function precision. The trap handler can determine the function that
9276 caused a floating point exception.
9279 Instruction precision. The trap handler can determine the exact
9280 instruction that caused a floating point exception.
9283 Other Alpha compilers provide the equivalent options called
9284 @option{-scope_safe} and @option{-resumption_safe}.
9286 @item -mieee-conformant
9287 @opindex mieee-conformant
9288 This option marks the generated code as IEEE conformant. You must not
9289 use this option unless you also specify @option{-mtrap-precision=i} and either
9290 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9291 is to emit the line @samp{.eflag 48} in the function prologue of the
9292 generated assembly file. Under DEC Unix, this has the effect that
9293 IEEE-conformant math library routines will be linked in.
9295 @item -mbuild-constants
9296 @opindex mbuild-constants
9297 Normally GCC examines a 32- or 64-bit integer constant to
9298 see if it can construct it from smaller constants in two or three
9299 instructions. If it cannot, it will output the constant as a literal and
9300 generate code to load it from the data segment at runtime.
9302 Use this option to require GCC to construct @emph{all} integer constants
9303 using code, even if it takes more instructions (the maximum is six).
9305 You would typically use this option to build a shared library dynamic
9306 loader. Itself a shared library, it must relocate itself in memory
9307 before it can find the variables and constants in its own data segment.
9313 Select whether to generate code to be assembled by the vendor-supplied
9314 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9332 Indicate whether GCC should generate code to use the optional BWX,
9333 CIX, FIX and MAX instruction sets. The default is to use the instruction
9334 sets supported by the CPU type specified via @option{-mcpu=} option or that
9335 of the CPU on which GCC was built if none was specified.
9340 @opindex mfloat-ieee
9341 Generate code that uses (does not use) VAX F and G floating point
9342 arithmetic instead of IEEE single and double precision.
9344 @item -mexplicit-relocs
9345 @itemx -mno-explicit-relocs
9346 @opindex mexplicit-relocs
9347 @opindex mno-explicit-relocs
9348 Older Alpha assemblers provided no way to generate symbol relocations
9349 except via assembler macros. Use of these macros does not allow
9350 optimal instruction scheduling. GNU binutils as of version 2.12
9351 supports a new syntax that allows the compiler to explicitly mark
9352 which relocations should apply to which instructions. This option
9353 is mostly useful for debugging, as GCC detects the capabilities of
9354 the assembler when it is built and sets the default accordingly.
9358 @opindex msmall-data
9359 @opindex mlarge-data
9360 When @option{-mexplicit-relocs} is in effect, static data is
9361 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9362 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9363 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9364 16-bit relocations off of the @code{$gp} register. This limits the
9365 size of the small data area to 64KB, but allows the variables to be
9366 directly accessed via a single instruction.
9368 The default is @option{-mlarge-data}. With this option the data area
9369 is limited to just below 2GB. Programs that require more than 2GB of
9370 data must use @code{malloc} or @code{mmap} to allocate the data in the
9371 heap instead of in the program's data segment.
9373 When generating code for shared libraries, @option{-fpic} implies
9374 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9378 @opindex msmall-text
9379 @opindex mlarge-text
9380 When @option{-msmall-text} is used, the compiler assumes that the
9381 code of the entire program (or shared library) fits in 4MB, and is
9382 thus reachable with a branch instruction. When @option{-msmall-data}
9383 is used, the compiler can assume that all local symbols share the
9384 same @code{$gp} value, and thus reduce the number of instructions
9385 required for a function call from 4 to 1.
9387 The default is @option{-mlarge-text}.
9389 @item -mcpu=@var{cpu_type}
9391 Set the instruction set and instruction scheduling parameters for
9392 machine type @var{cpu_type}. You can specify either the @samp{EV}
9393 style name or the corresponding chip number. GCC supports scheduling
9394 parameters for the EV4, EV5 and EV6 family of processors and will
9395 choose the default values for the instruction set from the processor
9396 you specify. If you do not specify a processor type, GCC will default
9397 to the processor on which the compiler was built.
9399 Supported values for @var{cpu_type} are
9405 Schedules as an EV4 and has no instruction set extensions.
9409 Schedules as an EV5 and has no instruction set extensions.
9413 Schedules as an EV5 and supports the BWX extension.
9418 Schedules as an EV5 and supports the BWX and MAX extensions.
9422 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9426 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9429 @item -mtune=@var{cpu_type}
9431 Set only the instruction scheduling parameters for machine type
9432 @var{cpu_type}. The instruction set is not changed.
9434 @item -mmemory-latency=@var{time}
9435 @opindex mmemory-latency
9436 Sets the latency the scheduler should assume for typical memory
9437 references as seen by the application. This number is highly
9438 dependent on the memory access patterns used by the application
9439 and the size of the external cache on the machine.
9441 Valid options for @var{time} are
9445 A decimal number representing clock cycles.
9451 The compiler contains estimates of the number of clock cycles for
9452 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9453 (also called Dcache, Scache, and Bcache), as well as to main memory.
9454 Note that L3 is only valid for EV5.
9459 @node DEC Alpha/VMS Options
9460 @subsection DEC Alpha/VMS Options
9462 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9465 @item -mvms-return-codes
9466 @opindex mvms-return-codes
9467 Return VMS condition codes from main. The default is to return POSIX
9468 style condition (e.g.@ error) codes.
9471 @node H8/300 Options
9472 @subsection H8/300 Options
9474 These @samp{-m} options are defined for the H8/300 implementations:
9479 Shorten some address references at link time, when possible; uses the
9480 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9481 ld, Using ld}, for a fuller description.
9485 Generate code for the H8/300H@.
9489 Generate code for the H8S@.
9493 Generate code for the H8S and H8/300H in the normal mode. This switch
9494 must be used either with -mh or -ms.
9498 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9502 Make @code{int} data 32 bits by default.
9506 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9507 The default for the H8/300H and H8S is to align longs and floats on 4
9509 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9510 This option has no effect on the H8/300.
9514 @subsection SH Options
9516 These @samp{-m} options are defined for the SH implementations:
9521 Generate code for the SH1.
9525 Generate code for the SH2.
9528 Generate code for the SH2e.
9532 Generate code for the SH3.
9536 Generate code for the SH3e.
9540 Generate code for the SH4 without a floating-point unit.
9542 @item -m4-single-only
9543 @opindex m4-single-only
9544 Generate code for the SH4 with a floating-point unit that only
9545 supports single-precision arithmetic.
9549 Generate code for the SH4 assuming the floating-point unit is in
9550 single-precision mode by default.
9554 Generate code for the SH4.
9558 Compile code for the processor in big endian mode.
9562 Compile code for the processor in little endian mode.
9566 Align doubles at 64-bit boundaries. Note that this changes the calling
9567 conventions, and thus some functions from the standard C library will
9568 not work unless you recompile it first with @option{-mdalign}.
9572 Shorten some address references at link time, when possible; uses the
9573 linker option @option{-relax}.
9577 Use 32-bit offsets in @code{switch} tables. The default is to use
9582 Enable the use of the instruction @code{fmovd}.
9586 Comply with the calling conventions defined by Renesas.
9590 Mark the @code{MAC} register as call-clobbered, even if
9591 @option{-mhitachi} is given.
9595 Increase IEEE-compliance of floating-point code.
9599 Dump instruction size and location in the assembly code.
9603 This option is deprecated. It pads structures to multiple of 4 bytes,
9604 which is incompatible with the SH ABI@.
9608 Optimize for space instead of speed. Implied by @option{-Os}.
9612 When generating position-independent code, emit function calls using
9613 the Global Offset Table instead of the Procedure Linkage Table.
9617 Generate a library function call to invalidate instruction cache
9618 entries, after fixing up a trampoline. This library function call
9619 doesn't assume it can write to the whole memory address space. This
9620 is the default when the target is @code{sh-*-linux*}.
9623 @node System V Options
9624 @subsection Options for System V
9626 These additional options are available on System V Release 4 for
9627 compatibility with other compilers on those systems:
9632 Create a shared object.
9633 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9637 Identify the versions of each tool used by the compiler, in a
9638 @code{.ident} assembler directive in the output.
9642 Refrain from adding @code{.ident} directives to the output file (this is
9645 @item -YP,@var{dirs}
9647 Search the directories @var{dirs}, and no others, for libraries
9648 specified with @option{-l}.
9652 Look in the directory @var{dir} to find the M4 preprocessor.
9653 The assembler uses this option.
9654 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9655 @c the generic assembler that comes with Solaris takes just -Ym.
9658 @node TMS320C3x/C4x Options
9659 @subsection TMS320C3x/C4x Options
9660 @cindex TMS320C3x/C4x Options
9662 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9666 @item -mcpu=@var{cpu_type}
9668 Set the instruction set, register set, and instruction scheduling
9669 parameters for machine type @var{cpu_type}. Supported values for
9670 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9671 @samp{c44}. The default is @samp{c40} to generate code for the
9676 @itemx -msmall-memory
9678 @opindex mbig-memory
9680 @opindex msmall-memory
9682 Generates code for the big or small memory model. The small memory
9683 model assumed that all data fits into one 64K word page. At run-time
9684 the data page (DP) register must be set to point to the 64K page
9685 containing the .bss and .data program sections. The big memory model is
9686 the default and requires reloading of the DP register for every direct
9693 Allow (disallow) allocation of general integer operands into the block
9700 Enable (disable) generation of code using decrement and branch,
9701 DBcond(D), instructions. This is enabled by default for the C4x. To be
9702 on the safe side, this is disabled for the C3x, since the maximum
9703 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9704 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9705 that it can utilize the decrement and branch instruction, but will give
9706 up if there is more than one memory reference in the loop. Thus a loop
9707 where the loop counter is decremented can generate slightly more
9708 efficient code, in cases where the RPTB instruction cannot be utilized.
9710 @item -mdp-isr-reload
9712 @opindex mdp-isr-reload
9714 Force the DP register to be saved on entry to an interrupt service
9715 routine (ISR), reloaded to point to the data section, and restored on
9716 exit from the ISR@. This should not be required unless someone has
9717 violated the small memory model by modifying the DP register, say within
9724 For the C3x use the 24-bit MPYI instruction for integer multiplies
9725 instead of a library call to guarantee 32-bit results. Note that if one
9726 of the operands is a constant, then the multiplication will be performed
9727 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9728 then squaring operations are performed inline instead of a library call.
9731 @itemx -mno-fast-fix
9733 @opindex mno-fast-fix
9734 The C3x/C4x FIX instruction to convert a floating point value to an
9735 integer value chooses the nearest integer less than or equal to the
9736 floating point value rather than to the nearest integer. Thus if the
9737 floating point number is negative, the result will be incorrectly
9738 truncated an additional code is necessary to detect and correct this
9739 case. This option can be used to disable generation of the additional
9740 code required to correct the result.
9746 Enable (disable) generation of repeat block sequences using the RPTB
9747 instruction for zero overhead looping. The RPTB construct is only used
9748 for innermost loops that do not call functions or jump across the loop
9749 boundaries. There is no advantage having nested RPTB loops due to the
9750 overhead required to save and restore the RC, RS, and RE registers.
9751 This is enabled by default with @option{-O2}.
9753 @item -mrpts=@var{count}
9757 Enable (disable) the use of the single instruction repeat instruction
9758 RPTS@. If a repeat block contains a single instruction, and the loop
9759 count can be guaranteed to be less than the value @var{count}, GCC will
9760 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9761 then a RPTS will be emitted even if the loop count cannot be determined
9762 at compile time. Note that the repeated instruction following RPTS does
9763 not have to be reloaded from memory each iteration, thus freeing up the
9764 CPU buses for operands. However, since interrupts are blocked by this
9765 instruction, it is disabled by default.
9767 @item -mloop-unsigned
9768 @itemx -mno-loop-unsigned
9769 @opindex mloop-unsigned
9770 @opindex mno-loop-unsigned
9771 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9772 is @math{2^{31} + 1} since these instructions test if the iteration count is
9773 negative to terminate the loop. If the iteration count is unsigned
9774 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9775 exceeded. This switch allows an unsigned iteration count.
9779 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9780 with. This also enforces compatibility with the API employed by the TI
9781 C3x C compiler. For example, long doubles are passed as structures
9782 rather than in floating point registers.
9788 Generate code that uses registers (stack) for passing arguments to functions.
9789 By default, arguments are passed in registers where possible rather
9790 than by pushing arguments on to the stack.
9792 @item -mparallel-insns
9793 @itemx -mno-parallel-insns
9794 @opindex mparallel-insns
9795 @opindex mno-parallel-insns
9796 Allow the generation of parallel instructions. This is enabled by
9797 default with @option{-O2}.
9799 @item -mparallel-mpy
9800 @itemx -mno-parallel-mpy
9801 @opindex mparallel-mpy
9802 @opindex mno-parallel-mpy
9803 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9804 provided @option{-mparallel-insns} is also specified. These instructions have
9805 tight register constraints which can pessimize the code generation
9811 @subsection V850 Options
9812 @cindex V850 Options
9814 These @samp{-m} options are defined for V850 implementations:
9818 @itemx -mno-long-calls
9819 @opindex mlong-calls
9820 @opindex mno-long-calls
9821 Treat all calls as being far away (near). If calls are assumed to be
9822 far away, the compiler will always load the functions address up into a
9823 register, and call indirect through the pointer.
9829 Do not optimize (do optimize) basic blocks that use the same index
9830 pointer 4 or more times to copy pointer into the @code{ep} register, and
9831 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9832 option is on by default if you optimize.
9834 @item -mno-prolog-function
9835 @itemx -mprolog-function
9836 @opindex mno-prolog-function
9837 @opindex mprolog-function
9838 Do not use (do use) external functions to save and restore registers
9839 at the prologue and epilogue of a function. The external functions
9840 are slower, but use less code space if more than one function saves
9841 the same number of registers. The @option{-mprolog-function} option
9842 is on by default if you optimize.
9846 Try to make the code as small as possible. At present, this just turns
9847 on the @option{-mep} and @option{-mprolog-function} options.
9851 Put static or global variables whose size is @var{n} bytes or less into
9852 the tiny data area that register @code{ep} points to. The tiny data
9853 area can hold up to 256 bytes in total (128 bytes for byte references).
9857 Put static or global variables whose size is @var{n} bytes or less into
9858 the small data area that register @code{gp} points to. The small data
9859 area can hold up to 64 kilobytes.
9863 Put static or global variables whose size is @var{n} bytes or less into
9864 the first 32 kilobytes of memory.
9868 Specify that the target processor is the V850.
9871 @opindex mbig-switch
9872 Generate code suitable for big switch tables. Use this option only if
9873 the assembler/linker complain about out of range branches within a switch
9878 This option will cause r2 and r5 to be used in the code generated by
9879 the compiler. This setting is the default.
9882 @opindex mno-app-regs
9883 This option will cause r2 and r5 to be treated as fixed registers.
9887 Specify that the target processor is the V850E1. The preprocessor
9888 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
9889 this option is used.
9893 Specify that the target processor is the V850E. The preprocessor
9894 constant @samp{__v850e__} will be defined if this option is used.
9896 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
9897 are defined then a default target processor will be chosen and the
9898 relevant @samp{__v850*__} preprocessor constant will be defined.
9900 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9901 defined, regardless of which processor variant is the target.
9903 @item -mdisable-callt
9904 @opindex mdisable-callt
9905 This option will suppress generation of the CALLT instruction for the
9906 v850e and v850e1 flavors of the v850 architecture. The default is
9907 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9912 @subsection ARC Options
9915 These options are defined for ARC implementations:
9920 Compile code for little endian mode. This is the default.
9924 Compile code for big endian mode.
9927 @opindex mmangle-cpu
9928 Prepend the name of the cpu to all public symbol names.
9929 In multiple-processor systems, there are many ARC variants with different
9930 instruction and register set characteristics. This flag prevents code
9931 compiled for one cpu to be linked with code compiled for another.
9932 No facility exists for handling variants that are ``almost identical''.
9933 This is an all or nothing option.
9935 @item -mcpu=@var{cpu}
9937 Compile code for ARC variant @var{cpu}.
9938 Which variants are supported depend on the configuration.
9939 All variants support @option{-mcpu=base}, this is the default.
9941 @item -mtext=@var{text-section}
9942 @itemx -mdata=@var{data-section}
9943 @itemx -mrodata=@var{readonly-data-section}
9947 Put functions, data, and readonly data in @var{text-section},
9948 @var{data-section}, and @var{readonly-data-section} respectively
9949 by default. This can be overridden with the @code{section} attribute.
9950 @xref{Variable Attributes}.
9955 @subsection NS32K Options
9956 @cindex NS32K options
9958 These are the @samp{-m} options defined for the 32000 series. The default
9959 values for these options depends on which style of 32000 was selected when
9960 the compiler was configured; the defaults for the most common choices are
9968 Generate output for a 32032. This is the default
9969 when the compiler is configured for 32032 and 32016 based systems.
9975 Generate output for a 32332. This is the default
9976 when the compiler is configured for 32332-based systems.
9982 Generate output for a 32532. This is the default
9983 when the compiler is configured for 32532-based systems.
9987 Generate output containing 32081 instructions for floating point.
9988 This is the default for all systems.
9992 Generate output containing 32381 instructions for floating point. This
9993 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9994 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9998 Try and generate multiply-add floating point instructions @code{polyF}
9999 and @code{dotF}. This option is only available if the @option{-m32381}
10000 option is in effect. Using these instructions requires changes to
10001 register allocation which generally has a negative impact on
10002 performance. This option should only be enabled when compiling code
10003 particularly likely to make heavy use of multiply-add instructions.
10005 @item -mnomulti-add
10006 @opindex mnomulti-add
10007 Do not try and generate multiply-add floating point instructions
10008 @code{polyF} and @code{dotF}. This is the default on all platforms.
10011 @opindex msoft-float
10012 Generate output containing library calls for floating point.
10013 @strong{Warning:} the requisite libraries may not be available.
10015 @item -mieee-compare
10016 @itemx -mno-ieee-compare
10017 @opindex mieee-compare
10018 @opindex mno-ieee-compare
10019 Control whether or not the compiler uses IEEE floating point
10020 comparisons. These handle correctly the case where the result of a
10021 comparison is unordered.
10022 @strong{Warning:} the requisite kernel support may not be available.
10025 @opindex mnobitfield
10026 Do not use the bit-field instructions. On some machines it is faster to
10027 use shifting and masking operations. This is the default for the pc532.
10031 Do use the bit-field instructions. This is the default for all platforms
10036 Use a different function-calling convention, in which functions
10037 that take a fixed number of arguments return pop their
10038 arguments on return with the @code{ret} instruction.
10040 This calling convention is incompatible with the one normally
10041 used on Unix, so you cannot use it if you need to call libraries
10042 compiled with the Unix compiler.
10044 Also, you must provide function prototypes for all functions that
10045 take variable numbers of arguments (including @code{printf});
10046 otherwise incorrect code will be generated for calls to those
10049 In addition, seriously incorrect code will result if you call a
10050 function with too many arguments. (Normally, extra arguments are
10051 harmlessly ignored.)
10053 This option takes its name from the 680x0 @code{rtd} instruction.
10058 Use a different function-calling convention where the first two arguments
10059 are passed in registers.
10061 This calling convention is incompatible with the one normally
10062 used on Unix, so you cannot use it if you need to call libraries
10063 compiled with the Unix compiler.
10066 @opindex mnoregparam
10067 Do not pass any arguments in registers. This is the default for all
10072 It is OK to use the sb as an index register which is always loaded with
10073 zero. This is the default for the pc532-netbsd target.
10077 The sb register is not available for use or has not been initialized to
10078 zero by the run time system. This is the default for all targets except
10079 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
10080 @option{-fpic} is set.
10084 Many ns32000 series addressing modes use displacements of up to 512MB@.
10085 If an address is above 512MB then displacements from zero can not be used.
10086 This option causes code to be generated which can be loaded above 512MB@.
10087 This may be useful for operating systems or ROM code.
10091 Assume code will be loaded in the first 512MB of virtual address space.
10092 This is the default for all platforms.
10098 @subsection AVR Options
10099 @cindex AVR Options
10101 These options are defined for AVR implementations:
10104 @item -mmcu=@var{mcu}
10106 Specify ATMEL AVR instruction set or MCU type.
10108 Instruction set avr1 is for the minimal AVR core, not supported by the C
10109 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
10110 attiny11, attiny12, attiny15, attiny28).
10112 Instruction set avr2 (default) is for the classic AVR core with up to
10113 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
10114 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
10115 at90c8534, at90s8535).
10117 Instruction set avr3 is for the classic AVR core with up to 128K program
10118 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
10120 Instruction set avr4 is for the enhanced AVR core with up to 8K program
10121 memory space (MCU types: atmega8, atmega83, atmega85).
10123 Instruction set avr5 is for the enhanced AVR core with up to 128K program
10124 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
10125 atmega64, atmega128, at43usb355, at94k).
10129 Output instruction sizes to the asm file.
10131 @item -minit-stack=@var{N}
10132 @opindex minit-stack
10133 Specify the initial stack address, which may be a symbol or numeric value,
10134 @samp{__stack} is the default.
10136 @item -mno-interrupts
10137 @opindex mno-interrupts
10138 Generated code is not compatible with hardware interrupts.
10139 Code size will be smaller.
10141 @item -mcall-prologues
10142 @opindex mcall-prologues
10143 Functions prologues/epilogues expanded as call to appropriate
10144 subroutines. Code size will be smaller.
10146 @item -mno-tablejump
10147 @opindex mno-tablejump
10148 Do not generate tablejump insns which sometimes increase code size.
10151 @opindex mtiny-stack
10152 Change only the low 8 bits of the stack pointer.
10155 @node MCore Options
10156 @subsection MCore Options
10157 @cindex MCore options
10159 These are the @samp{-m} options defined for the Motorola M*Core
10165 @itemx -mno-hardlit
10167 @opindex mno-hardlit
10168 Inline constants into the code stream if it can be done in two
10169 instructions or less.
10175 Use the divide instruction. (Enabled by default).
10177 @item -mrelax-immediate
10178 @itemx -mno-relax-immediate
10179 @opindex mrelax-immediate
10180 @opindex mno-relax-immediate
10181 Allow arbitrary sized immediates in bit operations.
10183 @item -mwide-bitfields
10184 @itemx -mno-wide-bitfields
10185 @opindex mwide-bitfields
10186 @opindex mno-wide-bitfields
10187 Always treat bit-fields as int-sized.
10189 @item -m4byte-functions
10190 @itemx -mno-4byte-functions
10191 @opindex m4byte-functions
10192 @opindex mno-4byte-functions
10193 Force all functions to be aligned to a four byte boundary.
10195 @item -mcallgraph-data
10196 @itemx -mno-callgraph-data
10197 @opindex mcallgraph-data
10198 @opindex mno-callgraph-data
10199 Emit callgraph information.
10202 @itemx -mno-slow-bytes
10203 @opindex mslow-bytes
10204 @opindex mno-slow-bytes
10205 Prefer word access when reading byte quantities.
10207 @item -mlittle-endian
10208 @itemx -mbig-endian
10209 @opindex mlittle-endian
10210 @opindex mbig-endian
10211 Generate code for a little endian target.
10217 Generate code for the 210 processor.
10220 @node IA-64 Options
10221 @subsection IA-64 Options
10222 @cindex IA-64 Options
10224 These are the @samp{-m} options defined for the Intel IA-64 architecture.
10228 @opindex mbig-endian
10229 Generate code for a big endian target. This is the default for HP-UX@.
10231 @item -mlittle-endian
10232 @opindex mlittle-endian
10233 Generate code for a little endian target. This is the default for AIX5
10239 @opindex mno-gnu-as
10240 Generate (or don't) code for the GNU assembler. This is the default.
10241 @c Also, this is the default if the configure option @option{--with-gnu-as}
10247 @opindex mno-gnu-ld
10248 Generate (or don't) code for the GNU linker. This is the default.
10249 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10254 Generate code that does not use a global pointer register. The result
10255 is not position independent code, and violates the IA-64 ABI@.
10257 @item -mvolatile-asm-stop
10258 @itemx -mno-volatile-asm-stop
10259 @opindex mvolatile-asm-stop
10260 @opindex mno-volatile-asm-stop
10261 Generate (or don't) a stop bit immediately before and after volatile asm
10266 Generate code that works around Itanium B step errata.
10268 @item -mregister-names
10269 @itemx -mno-register-names
10270 @opindex mregister-names
10271 @opindex mno-register-names
10272 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10273 the stacked registers. This may make assembler output more readable.
10279 Disable (or enable) optimizations that use the small data section. This may
10280 be useful for working around optimizer bugs.
10282 @item -mconstant-gp
10283 @opindex mconstant-gp
10284 Generate code that uses a single constant global pointer value. This is
10285 useful when compiling kernel code.
10289 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10290 This is useful when compiling firmware code.
10292 @item -minline-float-divide-min-latency
10293 @opindex minline-float-divide-min-latency
10294 Generate code for inline divides of floating point values
10295 using the minimum latency algorithm.
10297 @item -minline-float-divide-max-throughput
10298 @opindex minline-float-divide-max-throughput
10299 Generate code for inline divides of floating point values
10300 using the maximum throughput algorithm.
10302 @item -minline-int-divide-min-latency
10303 @opindex minline-int-divide-min-latency
10304 Generate code for inline divides of integer values
10305 using the minimum latency algorithm.
10307 @item -minline-int-divide-max-throughput
10308 @opindex minline-int-divide-max-throughput
10309 Generate code for inline divides of integer values
10310 using the maximum throughput algorithm.
10312 @item -mno-dwarf2-asm
10313 @itemx -mdwarf2-asm
10314 @opindex mno-dwarf2-asm
10315 @opindex mdwarf2-asm
10316 Don't (or do) generate assembler code for the DWARF2 line number debugging
10317 info. This may be useful when not using the GNU assembler.
10319 @item -mfixed-range=@var{register-range}
10320 @opindex mfixed-range
10321 Generate code treating the given register range as fixed registers.
10322 A fixed register is one that the register allocator can not use. This is
10323 useful when compiling kernel code. A register range is specified as
10324 two registers separated by a dash. Multiple register ranges can be
10325 specified separated by a comma.
10327 @item -mearly-stop-bits
10328 @itemx -mno-early-stop-bits
10329 @opindex mearly-stop-bits
10330 @opindex mno-early-stop-bits
10331 Allow stop bits to be placed earlier than immediately preceding the
10332 instruction that triggered the stop bit. This can improve instruction
10333 scheduling, but does not always do so.
10337 @subsection D30V Options
10338 @cindex D30V Options
10340 These @samp{-m} options are defined for D30V implementations:
10345 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
10346 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
10347 memory, which starts at location @code{0x80000000}.
10350 @opindex mextmemory
10351 Same as the @option{-mextmem} switch.
10355 Link the @samp{.text} section into onchip text memory, which starts at
10356 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
10357 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
10358 into onchip data memory, which starts at location @code{0x20000000}.
10360 @item -mno-asm-optimize
10361 @itemx -masm-optimize
10362 @opindex mno-asm-optimize
10363 @opindex masm-optimize
10364 Disable (enable) passing @option{-O} to the assembler when optimizing.
10365 The assembler uses the @option{-O} option to automatically parallelize
10366 adjacent short instructions where possible.
10368 @item -mbranch-cost=@var{n}
10369 @opindex mbranch-cost
10370 Increase the internal costs of branches to @var{n}. Higher costs means
10371 that the compiler will issue more instructions to avoid doing a branch.
10374 @item -mcond-exec=@var{n}
10375 @opindex mcond-exec
10376 Specify the maximum number of conditionally executed instructions that
10377 replace a branch. The default is 4.
10380 @node S/390 and zSeries Options
10381 @subsection S/390 and zSeries Options
10382 @cindex S/390 and zSeries Options
10384 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10388 @itemx -msoft-float
10389 @opindex mhard-float
10390 @opindex msoft-float
10391 Use (do not use) the hardware floating-point instructions and registers
10392 for floating-point operations. When @option{-msoft-float} is specified,
10393 functions in @file{libgcc.a} will be used to perform floating-point
10394 operations. When @option{-mhard-float} is specified, the compiler
10395 generates IEEE floating-point instructions. This is the default.
10398 @itemx -mno-backchain
10399 @opindex mbackchain
10400 @opindex mno-backchain
10401 Generate (or do not generate) code which maintains an explicit
10402 backchain within the stack frame that points to the caller's frame.
10403 This is currently needed to allow debugging. The default is to
10404 generate the backchain.
10407 @itemx -mno-small-exec
10408 @opindex msmall-exec
10409 @opindex mno-small-exec
10410 Generate (or do not generate) code using the @code{bras} instruction
10411 to do subroutine calls.
10412 This only works reliably if the total executable size does not
10413 exceed 64k. The default is to use the @code{basr} instruction instead,
10414 which does not have this limitation.
10420 When @option{-m31} is specified, generate code compliant to the
10421 Linux for S/390 ABI@. When @option{-m64} is specified, generate
10422 code compliant to the Linux for zSeries ABI@. This allows GCC in
10423 particular to generate 64-bit instructions. For the @samp{s390}
10424 targets, the default is @option{-m31}, while the @samp{s390x}
10425 targets default to @option{-m64}.
10431 When @option{-mzarch} is specified, generate code using the
10432 instructions available on z/Architecture.
10433 When @option{-mesa} is specified, generate code using the
10434 instructions available on ESA/390. Note that @option{-mesa} is
10435 not possible with @option{-m64}.
10436 When generating code compliant to the Linux for S/390 ABI,
10437 the default is @option{-mesa}. When generating code compliant
10438 to the Linux for zSeries ABI, the default is @option{-mzarch}.
10444 Generate (or do not generate) code using the @code{mvcle} instruction
10445 to perform block moves. When @option{-mno-mvcle} is specified,
10446 use a @code{mvc} loop instead. This is the default.
10452 Print (or do not print) additional debug information when compiling.
10453 The default is to not print debug information.
10455 @item -march=@var{cpu-type}
10457 Generate code that will run on @var{cpu-type}, which is the name of a system
10458 representing a certain processor type. Possible values for
10459 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10460 When generating code using the instructions available on z/Architecture,
10461 the default is @option{-march=z900}. Otherwise, the default is
10462 @option{-march=g5}.
10464 @item -mtune=@var{cpu-type}
10466 Tune to @var{cpu-type} everything applicable about the generated code,
10467 except for the ABI and the set of available instructions.
10468 The list of @var{cpu-type} values is the same as for @option{-march}.
10469 The default is the value used for @option{-march}.
10472 @itemx -mno-fused-madd
10473 @opindex mfused-madd
10474 @opindex mno-fused-madd
10475 Generate code that uses (does not use) the floating point multiply and
10476 accumulate instructions. These instructions are generated by default if
10477 hardware floating point is used.
10481 @subsection CRIS Options
10482 @cindex CRIS Options
10484 These options are defined specifically for the CRIS ports.
10487 @item -march=@var{architecture-type}
10488 @itemx -mcpu=@var{architecture-type}
10491 Generate code for the specified architecture. The choices for
10492 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10493 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
10494 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10497 @item -mtune=@var{architecture-type}
10499 Tune to @var{architecture-type} everything applicable about the generated
10500 code, except for the ABI and the set of available instructions. The
10501 choices for @var{architecture-type} are the same as for
10502 @option{-march=@var{architecture-type}}.
10504 @item -mmax-stack-frame=@var{n}
10505 @opindex mmax-stack-frame
10506 Warn when the stack frame of a function exceeds @var{n} bytes.
10508 @item -melinux-stacksize=@var{n}
10509 @opindex melinux-stacksize
10510 Only available with the @samp{cris-axis-aout} target. Arranges for
10511 indications in the program to the kernel loader that the stack of the
10512 program should be set to @var{n} bytes.
10518 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10519 @option{-march=v3} and @option{-march=v8} respectively.
10523 Enable CRIS-specific verbose debug-related information in the assembly
10524 code. This option also has the effect to turn off the @samp{#NO_APP}
10525 formatted-code indicator to the assembler at the beginning of the
10530 Do not use condition-code results from previous instruction; always emit
10531 compare and test instructions before use of condition codes.
10533 @item -mno-side-effects
10534 @opindex mno-side-effects
10535 Do not emit instructions with side-effects in addressing modes other than
10538 @item -mstack-align
10539 @itemx -mno-stack-align
10540 @itemx -mdata-align
10541 @itemx -mno-data-align
10542 @itemx -mconst-align
10543 @itemx -mno-const-align
10544 @opindex mstack-align
10545 @opindex mno-stack-align
10546 @opindex mdata-align
10547 @opindex mno-data-align
10548 @opindex mconst-align
10549 @opindex mno-const-align
10550 These options (no-options) arranges (eliminate arrangements) for the
10551 stack-frame, individual data and constants to be aligned for the maximum
10552 single data access size for the chosen CPU model. The default is to
10553 arrange for 32-bit alignment. ABI details such as structure layout are
10554 not affected by these options.
10562 Similar to the stack- data- and const-align options above, these options
10563 arrange for stack-frame, writable data and constants to all be 32-bit,
10564 16-bit or 8-bit aligned. The default is 32-bit alignment.
10566 @item -mno-prologue-epilogue
10567 @itemx -mprologue-epilogue
10568 @opindex mno-prologue-epilogue
10569 @opindex mprologue-epilogue
10570 With @option{-mno-prologue-epilogue}, the normal function prologue and
10571 epilogue that sets up the stack-frame are omitted and no return
10572 instructions or return sequences are generated in the code. Use this
10573 option only together with visual inspection of the compiled code: no
10574 warnings or errors are generated when call-saved registers must be saved,
10575 or storage for local variable needs to be allocated.
10579 @opindex mno-gotplt
10581 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10582 instruction sequences that load addresses for functions from the PLT part
10583 of the GOT rather than (traditional on other architectures) calls to the
10584 PLT. The default is @option{-mgotplt}.
10588 Legacy no-op option only recognized with the cris-axis-aout target.
10592 Legacy no-op option only recognized with the cris-axis-elf and
10593 cris-axis-linux-gnu targets.
10597 Only recognized with the cris-axis-aout target, where it selects a
10598 GNU/linux-like multilib, include files and instruction set for
10599 @option{-march=v8}.
10603 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10607 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
10608 to link with input-output functions from a simulator library. Code,
10609 initialized data and zero-initialized data are allocated consecutively.
10613 Like @option{-sim}, but pass linker options to locate initialized data at
10614 0x40000000 and zero-initialized data at 0x80000000.
10618 @subsection MMIX Options
10619 @cindex MMIX Options
10621 These options are defined for the MMIX:
10625 @itemx -mno-libfuncs
10627 @opindex mno-libfuncs
10628 Specify that intrinsic library functions are being compiled, passing all
10629 values in registers, no matter the size.
10632 @itemx -mno-epsilon
10634 @opindex mno-epsilon
10635 Generate floating-point comparison instructions that compare with respect
10636 to the @code{rE} epsilon register.
10638 @item -mabi=mmixware
10640 @opindex mabi-mmixware
10642 Generate code that passes function parameters and return values that (in
10643 the called function) are seen as registers @code{$0} and up, as opposed to
10644 the GNU ABI which uses global registers @code{$231} and up.
10646 @item -mzero-extend
10647 @itemx -mno-zero-extend
10648 @opindex mzero-extend
10649 @opindex mno-zero-extend
10650 When reading data from memory in sizes shorter than 64 bits, use (do not
10651 use) zero-extending load instructions by default, rather than
10652 sign-extending ones.
10655 @itemx -mno-knuthdiv
10657 @opindex mno-knuthdiv
10658 Make the result of a division yielding a remainder have the same sign as
10659 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10660 remainder follows the sign of the dividend. Both methods are
10661 arithmetically valid, the latter being almost exclusively used.
10663 @item -mtoplevel-symbols
10664 @itemx -mno-toplevel-symbols
10665 @opindex mtoplevel-symbols
10666 @opindex mno-toplevel-symbols
10667 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10668 code can be used with the @code{PREFIX} assembly directive.
10672 Generate an executable in the ELF format, rather than the default
10673 @samp{mmo} format used by the @command{mmix} simulator.
10675 @item -mbranch-predict
10676 @itemx -mno-branch-predict
10677 @opindex mbranch-predict
10678 @opindex mno-branch-predict
10679 Use (do not use) the probable-branch instructions, when static branch
10680 prediction indicates a probable branch.
10682 @item -mbase-addresses
10683 @itemx -mno-base-addresses
10684 @opindex mbase-addresses
10685 @opindex mno-base-addresses
10686 Generate (do not generate) code that uses @emph{base addresses}. Using a
10687 base address automatically generates a request (handled by the assembler
10688 and the linker) for a constant to be set up in a global register. The
10689 register is used for one or more base address requests within the range 0
10690 to 255 from the value held in the register. The generally leads to short
10691 and fast code, but the number of different data items that can be
10692 addressed is limited. This means that a program that uses lots of static
10693 data may require @option{-mno-base-addresses}.
10695 @item -msingle-exit
10696 @itemx -mno-single-exit
10697 @opindex msingle-exit
10698 @opindex mno-single-exit
10699 Force (do not force) generated code to have a single exit point in each
10703 @node PDP-11 Options
10704 @subsection PDP-11 Options
10705 @cindex PDP-11 Options
10707 These options are defined for the PDP-11:
10712 Use hardware FPP floating point. This is the default. (FIS floating
10713 point on the PDP-11/40 is not supported.)
10716 @opindex msoft-float
10717 Do not use hardware floating point.
10721 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10725 Return floating-point results in memory. This is the default.
10729 Generate code for a PDP-11/40.
10733 Generate code for a PDP-11/45. This is the default.
10737 Generate code for a PDP-11/10.
10739 @item -mbcopy-builtin
10740 @opindex bcopy-builtin
10741 Use inline @code{movstrhi} patterns for copying memory. This is the
10746 Do not use inline @code{movstrhi} patterns for copying memory.
10752 Use 16-bit @code{int}. This is the default.
10758 Use 32-bit @code{int}.
10761 @itemx -mno-float32
10763 @opindex mno-float32
10764 Use 64-bit @code{float}. This is the default.
10769 @opindex mno-float64
10770 Use 32-bit @code{float}.
10774 Use @code{abshi2} pattern. This is the default.
10778 Do not use @code{abshi2} pattern.
10780 @item -mbranch-expensive
10781 @opindex mbranch-expensive
10782 Pretend that branches are expensive. This is for experimenting with
10783 code generation only.
10785 @item -mbranch-cheap
10786 @opindex mbranch-cheap
10787 Do not pretend that branches are expensive. This is the default.
10791 Generate code for a system with split I&D.
10795 Generate code for a system without split I&D. This is the default.
10799 Use Unix assembler syntax. This is the default when configured for
10800 @samp{pdp11-*-bsd}.
10804 Use DEC assembler syntax. This is the default when configured for any
10805 PDP-11 target other than @samp{pdp11-*-bsd}.
10808 @node Xstormy16 Options
10809 @subsection Xstormy16 Options
10810 @cindex Xstormy16 Options
10812 These options are defined for Xstormy16:
10817 Choose startup files and linker script suitable for the simulator.
10821 @subsection FRV Options
10822 @cindex FRV Options
10828 Only use the first 32 general purpose registers.
10833 Use all 64 general purpose registers.
10838 Use only the first 32 floating point registers.
10843 Use all 64 floating point registers
10846 @opindex mhard-float
10848 Use hardware instructions for floating point operations.
10851 @opindex msoft-float
10853 Use library routines for floating point operations.
10858 Dynamically allocate condition code registers.
10863 Do not try to dynamically allocate condition code registers, only
10864 use @code{icc0} and @code{fcc0}.
10869 Change ABI to use double word insns.
10874 Do not use double word instructions.
10879 Use floating point double instructions.
10882 @opindex mno-double
10884 Do not use floating point double instructions.
10889 Use media instructions.
10894 Do not use media instructions.
10899 Use multiply and add/subtract instructions.
10902 @opindex mno-muladd
10904 Do not use multiply and add/subtract instructions.
10906 @item -mlibrary-pic
10907 @opindex mlibrary-pic
10909 Enable PIC support for building libraries
10914 Use only the first four media accumulator registers.
10919 Use all eight media accumulator registers.
10924 Pack VLIW instructions.
10929 Do not pack VLIW instructions.
10932 @opindex mno-eflags
10934 Do not mark ABI switches in e_flags.
10937 @opindex mcond-move
10939 Enable the use of conditional-move instructions (default).
10941 This switch is mainly for debugging the compiler and will likely be removed
10942 in a future version.
10944 @item -mno-cond-move
10945 @opindex mno-cond-move
10947 Disable the use of conditional-move instructions.
10949 This switch is mainly for debugging the compiler and will likely be removed
10950 in a future version.
10955 Enable the use of conditional set instructions (default).
10957 This switch is mainly for debugging the compiler and will likely be removed
10958 in a future version.
10963 Disable the use of conditional set instructions.
10965 This switch is mainly for debugging the compiler and will likely be removed
10966 in a future version.
10969 @opindex mcond-exec
10971 Enable the use of conditional execution (default).
10973 This switch is mainly for debugging the compiler and will likely be removed
10974 in a future version.
10976 @item -mno-cond-exec
10977 @opindex mno-cond-exec
10979 Disable the use of conditional execution.
10981 This switch is mainly for debugging the compiler and will likely be removed
10982 in a future version.
10984 @item -mvliw-branch
10985 @opindex mvliw-branch
10987 Run a pass to pack branches into VLIW instructions (default).
10989 This switch is mainly for debugging the compiler and will likely be removed
10990 in a future version.
10992 @item -mno-vliw-branch
10993 @opindex mno-vliw-branch
10995 Do not run a pass to pack branches into VLIW instructions.
10997 This switch is mainly for debugging the compiler and will likely be removed
10998 in a future version.
11000 @item -mmulti-cond-exec
11001 @opindex mmulti-cond-exec
11003 Enable optimization of @code{&&} and @code{||} in conditional execution
11006 This switch is mainly for debugging the compiler and will likely be removed
11007 in a future version.
11009 @item -mno-multi-cond-exec
11010 @opindex mno-multi-cond-exec
11012 Disable optimization of @code{&&} and @code{||} in conditional execution.
11014 This switch is mainly for debugging the compiler and will likely be removed
11015 in a future version.
11017 @item -mnested-cond-exec
11018 @opindex mnested-cond-exec
11020 Enable nested conditional execution optimizations (default).
11022 This switch is mainly for debugging the compiler and will likely be removed
11023 in a future version.
11025 @item -mno-nested-cond-exec
11026 @opindex mno-nested-cond-exec
11028 Disable nested conditional execution optimizations.
11030 This switch is mainly for debugging the compiler and will likely be removed
11031 in a future version.
11033 @item -mtomcat-stats
11034 @opindex mtomcat-stats
11036 Cause gas to print out tomcat statistics.
11038 @item -mcpu=@var{cpu}
11041 Select the processor type for which to generate code. Possible values are
11042 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
11047 @node Xtensa Options
11048 @subsection Xtensa Options
11049 @cindex Xtensa Options
11051 These options are supported for Xtensa targets:
11055 @itemx -mno-const16
11057 @opindex mno-const16
11058 Enable or disable use of @code{CONST16} instructions for loading
11059 constant values. The @code{CONST16} instruction is currently not a
11060 standard option from Tensilica. When enabled, @code{CONST16}
11061 instructions are always used in place of the standard @code{L32R}
11062 instructions. The use of @code{CONST16} is enabled by default only if
11063 the @code{L32R} instruction is not available.
11066 @itemx -mno-fused-madd
11067 @opindex mfused-madd
11068 @opindex mno-fused-madd
11069 Enable or disable use of fused multiply/add and multiply/subtract
11070 instructions in the floating-point option. This has no effect if the
11071 floating-point option is not also enabled. Disabling fused multiply/add
11072 and multiply/subtract instructions forces the compiler to use separate
11073 instructions for the multiply and add/subtract operations. This may be
11074 desirable in some cases where strict IEEE 754-compliant results are
11075 required: the fused multiply add/subtract instructions do not round the
11076 intermediate result, thereby producing results with @emph{more} bits of
11077 precision than specified by the IEEE standard. Disabling fused multiply
11078 add/subtract instructions also ensures that the program output is not
11079 sensitive to the compiler's ability to combine multiply and add/subtract
11082 @item -mtext-section-literals
11083 @itemx -mno-text-section-literals
11084 @opindex mtext-section-literals
11085 @opindex mno-text-section-literals
11086 Control the treatment of literal pools. The default is
11087 @option{-mno-text-section-literals}, which places literals in a separate
11088 section in the output file. This allows the literal pool to be placed
11089 in a data RAM/ROM, and it also allows the linker to combine literal
11090 pools from separate object files to remove redundant literals and
11091 improve code size. With @option{-mtext-section-literals}, the literals
11092 are interspersed in the text section in order to keep them as close as
11093 possible to their references. This may be necessary for large assembly
11096 @item -mtarget-align
11097 @itemx -mno-target-align
11098 @opindex mtarget-align
11099 @opindex mno-target-align
11100 When this option is enabled, GCC instructs the assembler to
11101 automatically align instructions to reduce branch penalties at the
11102 expense of some code density. The assembler attempts to widen density
11103 instructions to align branch targets and the instructions following call
11104 instructions. If there are not enough preceding safe density
11105 instructions to align a target, no widening will be performed. The
11106 default is @option{-mtarget-align}. These options do not affect the
11107 treatment of auto-aligned instructions like @code{LOOP}, which the
11108 assembler will always align, either by widening density instructions or
11109 by inserting no-op instructions.
11112 @itemx -mno-longcalls
11113 @opindex mlongcalls
11114 @opindex mno-longcalls
11115 When this option is enabled, GCC instructs the assembler to translate
11116 direct calls to indirect calls unless it can determine that the target
11117 of a direct call is in the range allowed by the call instruction. This
11118 translation typically occurs for calls to functions in other source
11119 files. Specifically, the assembler translates a direct @code{CALL}
11120 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
11121 The default is @option{-mno-longcalls}. This option should be used in
11122 programs where the call target can potentially be out of range. This
11123 option is implemented in the assembler, not the compiler, so the
11124 assembly code generated by GCC will still show direct call
11125 instructions---look at the disassembled object code to see the actual
11126 instructions. Note that the assembler will use an indirect call for
11127 every cross-file call, not just those that really will be out of range.
11130 @node Code Gen Options
11131 @section Options for Code Generation Conventions
11132 @cindex code generation conventions
11133 @cindex options, code generation
11134 @cindex run-time options
11136 These machine-independent options control the interface conventions
11137 used in code generation.
11139 Most of them have both positive and negative forms; the negative form
11140 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
11141 one of the forms is listed---the one which is not the default. You
11142 can figure out the other form by either removing @samp{no-} or adding
11146 @item -fbounds-check
11147 @opindex fbounds-check
11148 For front-ends that support it, generate additional code to check that
11149 indices used to access arrays are within the declared range. This is
11150 currently only supported by the Java and Fortran 77 front-ends, where
11151 this option defaults to true and false respectively.
11155 This option generates traps for signed overflow on addition, subtraction,
11156 multiplication operations.
11160 This option instructs the compiler to assume that signed arithmetic
11161 overflow of addition, subtraction and multiplication wraps around
11162 using twos-complement representation. This flag enables some optimizations
11163 and disables other. This option is enabled by default for the Java
11164 front-end, as required by the Java language specification.
11167 @opindex fexceptions
11168 Enable exception handling. Generates extra code needed to propagate
11169 exceptions. For some targets, this implies GCC will generate frame
11170 unwind information for all functions, which can produce significant data
11171 size overhead, although it does not affect execution. If you do not
11172 specify this option, GCC will enable it by default for languages like
11173 C++ which normally require exception handling, and disable it for
11174 languages like C that do not normally require it. However, you may need
11175 to enable this option when compiling C code that needs to interoperate
11176 properly with exception handlers written in C++. You may also wish to
11177 disable this option if you are compiling older C++ programs that don't
11178 use exception handling.
11180 @item -fnon-call-exceptions
11181 @opindex fnon-call-exceptions
11182 Generate code that allows trapping instructions to throw exceptions.
11183 Note that this requires platform-specific runtime support that does
11184 not exist everywhere. Moreover, it only allows @emph{trapping}
11185 instructions to throw exceptions, i.e.@: memory references or floating
11186 point instructions. It does not allow exceptions to be thrown from
11187 arbitrary signal handlers such as @code{SIGALRM}.
11189 @item -funwind-tables
11190 @opindex funwind-tables
11191 Similar to @option{-fexceptions}, except that it will just generate any needed
11192 static data, but will not affect the generated code in any other way.
11193 You will normally not enable this option; instead, a language processor
11194 that needs this handling would enable it on your behalf.
11196 @item -fasynchronous-unwind-tables
11197 @opindex funwind-tables
11198 Generate unwind table in dwarf2 format, if supported by target machine. The
11199 table is exact at each instruction boundary, so it can be used for stack
11200 unwinding from asynchronous events (such as debugger or garbage collector).
11202 @item -fpcc-struct-return
11203 @opindex fpcc-struct-return
11204 Return ``short'' @code{struct} and @code{union} values in memory like
11205 longer ones, rather than in registers. This convention is less
11206 efficient, but it has the advantage of allowing intercallability between
11207 GCC-compiled files and files compiled with other compilers, particularly
11208 the Portable C Compiler (pcc).
11210 The precise convention for returning structures in memory depends
11211 on the target configuration macros.
11213 Short structures and unions are those whose size and alignment match
11214 that of some integer type.
11216 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11217 switch is not binary compatible with code compiled with the
11218 @option{-freg-struct-return} switch.
11219 Use it to conform to a non-default application binary interface.
11221 @item -freg-struct-return
11222 @opindex freg-struct-return
11223 Return @code{struct} and @code{union} values in registers when possible.
11224 This is more efficient for small structures than
11225 @option{-fpcc-struct-return}.
11227 If you specify neither @option{-fpcc-struct-return} nor
11228 @option{-freg-struct-return}, GCC defaults to whichever convention is
11229 standard for the target. If there is no standard convention, GCC
11230 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11231 the principal compiler. In those cases, we can choose the standard, and
11232 we chose the more efficient register return alternative.
11234 @strong{Warning:} code compiled with the @option{-freg-struct-return}
11235 switch is not binary compatible with code compiled with the
11236 @option{-fpcc-struct-return} switch.
11237 Use it to conform to a non-default application binary interface.
11239 @item -fshort-enums
11240 @opindex fshort-enums
11241 Allocate to an @code{enum} type only as many bytes as it needs for the
11242 declared range of possible values. Specifically, the @code{enum} type
11243 will be equivalent to the smallest integer type which has enough room.
11245 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11246 code that is not binary compatible with code generated without that switch.
11247 Use it to conform to a non-default application binary interface.
11249 @item -fshort-double
11250 @opindex fshort-double
11251 Use the same size for @code{double} as for @code{float}.
11253 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
11254 code that is not binary compatible with code generated without that switch.
11255 Use it to conform to a non-default application binary interface.
11257 @item -fshort-wchar
11258 @opindex fshort-wchar
11259 Override the underlying type for @samp{wchar_t} to be @samp{short
11260 unsigned int} instead of the default for the target. This option is
11261 useful for building programs to run under WINE@.
11263 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11264 code that is not binary compatible with code generated without that switch.
11265 Use it to conform to a non-default application binary interface.
11267 @item -fshared-data
11268 @opindex fshared-data
11269 Requests that the data and non-@code{const} variables of this
11270 compilation be shared data rather than private data. The distinction
11271 makes sense only on certain operating systems, where shared data is
11272 shared between processes running the same program, while private data
11273 exists in one copy per process.
11276 @opindex fno-common
11277 In C, allocate even uninitialized global variables in the data section of the
11278 object file, rather than generating them as common blocks. This has the
11279 effect that if the same variable is declared (without @code{extern}) in
11280 two different compilations, you will get an error when you link them.
11281 The only reason this might be useful is if you wish to verify that the
11282 program will work on other systems which always work this way.
11286 Ignore the @samp{#ident} directive.
11288 @item -fno-gnu-linker
11289 @opindex fno-gnu-linker
11290 Do not output global initializations (such as C++ constructors and
11291 destructors) in the form used by the GNU linker (on systems where the GNU
11292 linker is the standard method of handling them). Use this option when
11293 you want to use a non-GNU linker, which also requires using the
11294 @command{collect2} program to make sure the system linker includes
11295 constructors and destructors. (@command{collect2} is included in the GCC
11296 distribution.) For systems which @emph{must} use @command{collect2}, the
11297 compiler driver @command{gcc} is configured to do this automatically.
11299 @item -finhibit-size-directive
11300 @opindex finhibit-size-directive
11301 Don't output a @code{.size} assembler directive, or anything else that
11302 would cause trouble if the function is split in the middle, and the
11303 two halves are placed at locations far apart in memory. This option is
11304 used when compiling @file{crtstuff.c}; you should not need to use it
11307 @item -fverbose-asm
11308 @opindex fverbose-asm
11309 Put extra commentary information in the generated assembly code to
11310 make it more readable. This option is generally only of use to those
11311 who actually need to read the generated assembly code (perhaps while
11312 debugging the compiler itself).
11314 @option{-fno-verbose-asm}, the default, causes the
11315 extra information to be omitted and is useful when comparing two assembler
11320 @cindex global offset table
11322 Generate position-independent code (PIC) suitable for use in a shared
11323 library, if supported for the target machine. Such code accesses all
11324 constant addresses through a global offset table (GOT)@. The dynamic
11325 loader resolves the GOT entries when the program starts (the dynamic
11326 loader is not part of GCC; it is part of the operating system). If
11327 the GOT size for the linked executable exceeds a machine-specific
11328 maximum size, you get an error message from the linker indicating that
11329 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11330 instead. (These maximums are 16k on the m88k, 8k on the SPARC, and 32k
11331 on the m68k and RS/6000. The 386 has no such limit.)
11333 Position-independent code requires special support, and therefore works
11334 only on certain machines. For the 386, GCC supports PIC for System V
11335 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11336 position-independent.
11340 If supported for the target machine, emit position-independent code,
11341 suitable for dynamic linking and avoiding any limit on the size of the
11342 global offset table. This option makes a difference on the m68k, m88k,
11345 Position-independent code requires special support, and therefore works
11346 only on certain machines.
11352 These options are similar to @option{-fpic} and @option{-fPIC}, but
11353 generated position independent code can be only linked into executables.
11354 Usually these options are used when @option{-pie} GCC option will be
11355 used during linking.
11357 @item -ffixed-@var{reg}
11359 Treat the register named @var{reg} as a fixed register; generated code
11360 should never refer to it (except perhaps as a stack pointer, frame
11361 pointer or in some other fixed role).
11363 @var{reg} must be the name of a register. The register names accepted
11364 are machine-specific and are defined in the @code{REGISTER_NAMES}
11365 macro in the machine description macro file.
11367 This flag does not have a negative form, because it specifies a
11370 @item -fcall-used-@var{reg}
11371 @opindex fcall-used
11372 Treat the register named @var{reg} as an allocable register that is
11373 clobbered by function calls. It may be allocated for temporaries or
11374 variables that do not live across a call. Functions compiled this way
11375 will not save and restore the register @var{reg}.
11377 It is an error to used this flag with the frame pointer or stack pointer.
11378 Use of this flag for other registers that have fixed pervasive roles in
11379 the machine's execution model will produce disastrous results.
11381 This flag does not have a negative form, because it specifies a
11384 @item -fcall-saved-@var{reg}
11385 @opindex fcall-saved
11386 Treat the register named @var{reg} as an allocable register saved by
11387 functions. It may be allocated even for temporaries or variables that
11388 live across a call. Functions compiled this way will save and restore
11389 the register @var{reg} if they use it.
11391 It is an error to used this flag with the frame pointer or stack pointer.
11392 Use of this flag for other registers that have fixed pervasive roles in
11393 the machine's execution model will produce disastrous results.
11395 A different sort of disaster will result from the use of this flag for
11396 a register in which function values may be returned.
11398 This flag does not have a negative form, because it specifies a
11401 @item -fpack-struct
11402 @opindex fpack-struct
11403 Pack all structure members together without holes.
11405 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11406 code that is not binary compatible with code generated without that switch.
11407 Additionally, it makes the code suboptimal.
11408 Use it to conform to a non-default application binary interface.
11410 @item -finstrument-functions
11411 @opindex finstrument-functions
11412 Generate instrumentation calls for entry and exit to functions. Just
11413 after function entry and just before function exit, the following
11414 profiling functions will be called with the address of the current
11415 function and its call site. (On some platforms,
11416 @code{__builtin_return_address} does not work beyond the current
11417 function, so the call site information may not be available to the
11418 profiling functions otherwise.)
11421 void __cyg_profile_func_enter (void *this_fn,
11423 void __cyg_profile_func_exit (void *this_fn,
11427 The first argument is the address of the start of the current function,
11428 which may be looked up exactly in the symbol table.
11430 This instrumentation is also done for functions expanded inline in other
11431 functions. The profiling calls will indicate where, conceptually, the
11432 inline function is entered and exited. This means that addressable
11433 versions of such functions must be available. If all your uses of a
11434 function are expanded inline, this may mean an additional expansion of
11435 code size. If you use @samp{extern inline} in your C code, an
11436 addressable version of such functions must be provided. (This is
11437 normally the case anyways, but if you get lucky and the optimizer always
11438 expands the functions inline, you might have gotten away without
11439 providing static copies.)
11441 A function may be given the attribute @code{no_instrument_function}, in
11442 which case this instrumentation will not be done. This can be used, for
11443 example, for the profiling functions listed above, high-priority
11444 interrupt routines, and any functions from which the profiling functions
11445 cannot safely be called (perhaps signal handlers, if the profiling
11446 routines generate output or allocate memory).
11448 @item -fstack-check
11449 @opindex fstack-check
11450 Generate code to verify that you do not go beyond the boundary of the
11451 stack. You should specify this flag if you are running in an
11452 environment with multiple threads, but only rarely need to specify it in
11453 a single-threaded environment since stack overflow is automatically
11454 detected on nearly all systems if there is only one stack.
11456 Note that this switch does not actually cause checking to be done; the
11457 operating system must do that. The switch causes generation of code
11458 to ensure that the operating system sees the stack being extended.
11460 @item -fstack-limit-register=@var{reg}
11461 @itemx -fstack-limit-symbol=@var{sym}
11462 @itemx -fno-stack-limit
11463 @opindex fstack-limit-register
11464 @opindex fstack-limit-symbol
11465 @opindex fno-stack-limit
11466 Generate code to ensure that the stack does not grow beyond a certain value,
11467 either the value of a register or the address of a symbol. If the stack
11468 would grow beyond the value, a signal is raised. For most targets,
11469 the signal is raised before the stack overruns the boundary, so
11470 it is possible to catch the signal without taking special precautions.
11472 For instance, if the stack starts at absolute address @samp{0x80000000}
11473 and grows downwards, you can use the flags
11474 @option{-fstack-limit-symbol=__stack_limit} and
11475 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11476 of 128KB@. Note that this may only work with the GNU linker.
11478 @cindex aliasing of parameters
11479 @cindex parameters, aliased
11480 @item -fargument-alias
11481 @itemx -fargument-noalias
11482 @itemx -fargument-noalias-global
11483 @opindex fargument-alias
11484 @opindex fargument-noalias
11485 @opindex fargument-noalias-global
11486 Specify the possible relationships among parameters and between
11487 parameters and global data.
11489 @option{-fargument-alias} specifies that arguments (parameters) may
11490 alias each other and may alias global storage.@*
11491 @option{-fargument-noalias} specifies that arguments do not alias
11492 each other, but may alias global storage.@*
11493 @option{-fargument-noalias-global} specifies that arguments do not
11494 alias each other and do not alias global storage.
11496 Each language will automatically use whatever option is required by
11497 the language standard. You should not need to use these options yourself.
11499 @item -fleading-underscore
11500 @opindex fleading-underscore
11501 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11502 change the way C symbols are represented in the object file. One use
11503 is to help link with legacy assembly code.
11505 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11506 generate code that is not binary compatible with code generated without that
11507 switch. Use it to conform to a non-default application binary interface.
11508 Not all targets provide complete support for this switch.
11510 @item -ftls-model=@var{model}
11511 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11512 The @var{model} argument should be one of @code{global-dynamic},
11513 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11515 The default without @option{-fpic} is @code{initial-exec}; with
11516 @option{-fpic} the default is @code{global-dynamic}.
11521 @node Environment Variables
11522 @section Environment Variables Affecting GCC
11523 @cindex environment variables
11525 @c man begin ENVIRONMENT
11526 This section describes several environment variables that affect how GCC
11527 operates. Some of them work by specifying directories or prefixes to use
11528 when searching for various kinds of files. Some are used to specify other
11529 aspects of the compilation environment.
11531 Note that you can also specify places to search using options such as
11532 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11533 take precedence over places specified using environment variables, which
11534 in turn take precedence over those specified by the configuration of GCC@.
11535 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11536 GNU Compiler Collection (GCC) Internals}.
11541 @c @itemx LC_COLLATE
11543 @c @itemx LC_MONETARY
11544 @c @itemx LC_NUMERIC
11549 @c @findex LC_COLLATE
11550 @findex LC_MESSAGES
11551 @c @findex LC_MONETARY
11552 @c @findex LC_NUMERIC
11556 These environment variables control the way that GCC uses
11557 localization information that allow GCC to work with different
11558 national conventions. GCC inspects the locale categories
11559 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11560 so. These locale categories can be set to any value supported by your
11561 installation. A typical value is @samp{en_UK} for English in the United
11564 The @env{LC_CTYPE} environment variable specifies character
11565 classification. GCC uses it to determine the character boundaries in
11566 a string; this is needed for some multibyte encodings that contain quote
11567 and escape characters that would otherwise be interpreted as a string
11570 The @env{LC_MESSAGES} environment variable specifies the language to
11571 use in diagnostic messages.
11573 If the @env{LC_ALL} environment variable is set, it overrides the value
11574 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11575 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11576 environment variable. If none of these variables are set, GCC
11577 defaults to traditional C English behavior.
11581 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11582 files. GCC uses temporary files to hold the output of one stage of
11583 compilation which is to be used as input to the next stage: for example,
11584 the output of the preprocessor, which is the input to the compiler
11587 @item GCC_EXEC_PREFIX
11588 @findex GCC_EXEC_PREFIX
11589 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11590 names of the subprograms executed by the compiler. No slash is added
11591 when this prefix is combined with the name of a subprogram, but you can
11592 specify a prefix that ends with a slash if you wish.
11594 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11595 an appropriate prefix to use based on the pathname it was invoked with.
11597 If GCC cannot find the subprogram using the specified prefix, it
11598 tries looking in the usual places for the subprogram.
11600 The default value of @env{GCC_EXEC_PREFIX} is
11601 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
11602 of @code{prefix} when you ran the @file{configure} script.
11604 Other prefixes specified with @option{-B} take precedence over this prefix.
11606 This prefix is also used for finding files such as @file{crt0.o} that are
11609 In addition, the prefix is used in an unusual way in finding the
11610 directories to search for header files. For each of the standard
11611 directories whose name normally begins with @samp{/usr/local/lib/gcc}
11612 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11613 replacing that beginning with the specified prefix to produce an
11614 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11615 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11616 These alternate directories are searched first; the standard directories
11619 @item COMPILER_PATH
11620 @findex COMPILER_PATH
11621 The value of @env{COMPILER_PATH} is a colon-separated list of
11622 directories, much like @env{PATH}. GCC tries the directories thus
11623 specified when searching for subprograms, if it can't find the
11624 subprograms using @env{GCC_EXEC_PREFIX}.
11627 @findex LIBRARY_PATH
11628 The value of @env{LIBRARY_PATH} is a colon-separated list of
11629 directories, much like @env{PATH}. When configured as a native compiler,
11630 GCC tries the directories thus specified when searching for special
11631 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11632 using GCC also uses these directories when searching for ordinary
11633 libraries for the @option{-l} option (but directories specified with
11634 @option{-L} come first).
11638 @cindex locale definition
11639 This variable is used to pass locale information to the compiler. One way in
11640 which this information is used is to determine the character set to be used
11641 when character literals, string literals and comments are parsed in C and C++.
11642 When the compiler is configured to allow multibyte characters,
11643 the following values for @env{LANG} are recognized:
11647 Recognize JIS characters.
11649 Recognize SJIS characters.
11651 Recognize EUCJP characters.
11654 If @env{LANG} is not defined, or if it has some other value, then the
11655 compiler will use mblen and mbtowc as defined by the default locale to
11656 recognize and translate multibyte characters.
11660 Some additional environments variables affect the behavior of the
11663 @include cppenv.texi
11667 @node Precompiled Headers
11668 @section Using Precompiled Headers
11669 @cindex precompiled headers
11670 @cindex speed of compilation
11672 Often large projects have many header files that are included in every
11673 source file. The time the compiler takes to process these header files
11674 over and over again can account for nearly all of the time required to
11675 build the project. To make builds faster, GCC allows users to
11676 `precompile' a header file; then, if builds can use the precompiled
11677 header file they will be much faster.
11679 To create a precompiled header file, simply compile it as you would any
11680 other file, if necessary using the @option{-x} option to make the driver
11681 treat it as a C or C++ header file. You will probably want to use a
11682 tool like @command{make} to keep the precompiled header up-to-date when
11683 the headers it contains change.
11685 A precompiled header file will be searched for when @code{#include} is
11686 seen in the compilation. As it searches for the included file
11687 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
11688 compiler looks for a precompiled header in each directory just before it
11689 looks for the include file in that directory. The name searched for is
11690 the name specified in the @code{#include} with @samp{.gch} appended. If
11691 the precompiled header file can't be used, it is ignored.
11693 For instance, if you have @code{#include "all.h"}, and you have
11694 @file{all.h.gch} in the same directory as @file{all.h}, then the
11695 precompiled header file will be used if possible, and the original
11696 header will be used otherwise.
11698 Alternatively, you might decide to put the precompiled header file in a
11699 directory and use @option{-I} to ensure that directory is searched
11700 before (or instead of) the directory containing the original header.
11701 Then, if you want to check that the precompiled header file is always
11702 used, you can put a file of the same name as the original header in this
11703 directory containing an @code{#error} command.
11705 This also works with @option{-include}. So yet another way to use
11706 precompiled headers, good for projects not designed with precompiled
11707 header files in mind, is to simply take most of the header files used by
11708 a project, include them from another header file, precompile that header
11709 file, and @option{-include} the precompiled header. If the header files
11710 have guards against multiple inclusion, they will be skipped because
11711 they've already been included (in the precompiled header).
11713 If you need to precompile the same header file for different
11714 languages, targets, or compiler options, you can instead make a
11715 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11716 header in the directory. (It doesn't matter what you call the files
11717 in the directory, every precompiled header in the directory will be
11718 considered.) The first precompiled header encountered in the
11719 directory that is valid for this compilation will be used; they're
11720 searched in no particular order.
11722 There are many other possibilities, limited only by your imagination,
11723 good sense, and the constraints of your build system.
11725 A precompiled header file can be used only when these conditions apply:
11729 Only one precompiled header can be used in a particular compilation.
11731 A precompiled header can't be used once the first C token is seen. You
11732 can have preprocessor directives before a precompiled header; you can
11733 even include a precompiled header from inside another header, so long as
11734 there are no C tokens before the @code{#include}.
11736 The precompiled header file must be produced for the same language as
11737 the current compilation. You can't use a C precompiled header for a C++
11740 The precompiled header file must be produced by the same compiler
11741 version and configuration as the current compilation is using.
11742 The easiest way to guarantee this is to use the same compiler binary
11743 for creating and using precompiled headers.
11745 Any macros defined before the precompiled header (including with
11746 @option{-D}) must either be defined in the same way as when the
11747 precompiled header was generated, or must not affect the precompiled
11748 header, which usually means that the they don't appear in the
11749 precompiled header at all.
11751 Certain command-line options must be defined in the same way as when the
11752 precompiled header was generated. At present, it's not clear which
11753 options are safe to change and which are not; the safest choice is to
11754 use exactly the same options when generating and using the precompiled
11758 For all of these but the last, the compiler will automatically ignore
11759 the precompiled header if the conditions aren't met. For the last item,
11760 some option changes will cause the precompiled header to be rejected,
11761 but not all incompatible option combinations have yet been found. If
11762 you find a new incompatible combination, please consider filing a bug
11763 report, see @ref{Bugs}.
11765 @node Running Protoize
11766 @section Running Protoize
11768 The program @code{protoize} is an optional part of GCC@. You can use
11769 it to add prototypes to a program, thus converting the program to ISO
11770 C in one respect. The companion program @code{unprotoize} does the
11771 reverse: it removes argument types from any prototypes that are found.
11773 When you run these programs, you must specify a set of source files as
11774 command line arguments. The conversion programs start out by compiling
11775 these files to see what functions they define. The information gathered
11776 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11778 After scanning comes actual conversion. The specified files are all
11779 eligible to be converted; any files they include (whether sources or
11780 just headers) are eligible as well.
11782 But not all the eligible files are converted. By default,
11783 @code{protoize} and @code{unprotoize} convert only source and header
11784 files in the current directory. You can specify additional directories
11785 whose files should be converted with the @option{-d @var{directory}}
11786 option. You can also specify particular files to exclude with the
11787 @option{-x @var{file}} option. A file is converted if it is eligible, its
11788 directory name matches one of the specified directory names, and its
11789 name within the directory has not been excluded.
11791 Basic conversion with @code{protoize} consists of rewriting most
11792 function definitions and function declarations to specify the types of
11793 the arguments. The only ones not rewritten are those for varargs
11796 @code{protoize} optionally inserts prototype declarations at the
11797 beginning of the source file, to make them available for any calls that
11798 precede the function's definition. Or it can insert prototype
11799 declarations with block scope in the blocks where undeclared functions
11802 Basic conversion with @code{unprotoize} consists of rewriting most
11803 function declarations to remove any argument types, and rewriting
11804 function definitions to the old-style pre-ISO form.
11806 Both conversion programs print a warning for any function declaration or
11807 definition that they can't convert. You can suppress these warnings
11810 The output from @code{protoize} or @code{unprotoize} replaces the
11811 original source file. The original file is renamed to a name ending
11812 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11813 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11814 for DOS) file already exists, then the source file is simply discarded.
11816 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11817 scan the program and collect information about the functions it uses.
11818 So neither of these programs will work until GCC is installed.
11820 Here is a table of the options you can use with @code{protoize} and
11821 @code{unprotoize}. Each option works with both programs unless
11825 @item -B @var{directory}
11826 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11827 usual directory (normally @file{/usr/local/lib}). This file contains
11828 prototype information about standard system functions. This option
11829 applies only to @code{protoize}.
11831 @item -c @var{compilation-options}
11832 Use @var{compilation-options} as the options when running @command{gcc} to
11833 produce the @samp{.X} files. The special option @option{-aux-info} is
11834 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11836 Note that the compilation options must be given as a single argument to
11837 @code{protoize} or @code{unprotoize}. If you want to specify several
11838 @command{gcc} options, you must quote the entire set of compilation options
11839 to make them a single word in the shell.
11841 There are certain @command{gcc} arguments that you cannot use, because they
11842 would produce the wrong kind of output. These include @option{-g},
11843 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11844 the @var{compilation-options}, they are ignored.
11847 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11848 systems) instead of @samp{.c}. This is convenient if you are converting
11849 a C program to C++. This option applies only to @code{protoize}.
11852 Add explicit global declarations. This means inserting explicit
11853 declarations at the beginning of each source file for each function
11854 that is called in the file and was not declared. These declarations
11855 precede the first function definition that contains a call to an
11856 undeclared function. This option applies only to @code{protoize}.
11858 @item -i @var{string}
11859 Indent old-style parameter declarations with the string @var{string}.
11860 This option applies only to @code{protoize}.
11862 @code{unprotoize} converts prototyped function definitions to old-style
11863 function definitions, where the arguments are declared between the
11864 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11865 uses five spaces as the indentation. If you want to indent with just
11866 one space instead, use @option{-i " "}.
11869 Keep the @samp{.X} files. Normally, they are deleted after conversion
11873 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11874 a prototype declaration for each function in each block which calls the
11875 function without any declaration. This option applies only to
11879 Make no real changes. This mode just prints information about the conversions
11880 that would have been done without @option{-n}.
11883 Make no @samp{.save} files. The original files are simply deleted.
11884 Use this option with caution.
11886 @item -p @var{program}
11887 Use the program @var{program} as the compiler. Normally, the name
11888 @file{gcc} is used.
11891 Work quietly. Most warnings are suppressed.
11894 Print the version number, just like @option{-v} for @command{gcc}.
11897 If you need special compiler options to compile one of your program's
11898 source files, then you should generate that file's @samp{.X} file
11899 specially, by running @command{gcc} on that source file with the
11900 appropriate options and the option @option{-aux-info}. Then run
11901 @code{protoize} on the entire set of files. @code{protoize} will use
11902 the existing @samp{.X} file because it is newer than the source file.
11906 gcc -Dfoo=bar file1.c -aux-info file1.X
11911 You need to include the special files along with the rest in the
11912 @code{protoize} command, even though their @samp{.X} files already
11913 exist, because otherwise they won't get converted.
11915 @xref{Protoize Caveats}, for more information on how to use
11916 @code{protoize} successfully.