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 -fdollars-in-identifiers @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 -fvtable-gc -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}.
196 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
197 -fgnu-runtime -fnext-runtime -gen-decls @gol
198 -Wno-protocol -Wselector -Wundeclared-selector}
200 @item Language Independent Options
201 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
202 @gccoptlist{-fmessage-length=@var{n} @gol
203 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
205 @item Warning Options
206 @xref{Warning Options,,Options to Request or Suppress Warnings}.
207 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
208 -w -Wextra -Wall -Waggregate-return @gol
209 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
210 -Wconversion -Wno-deprecated-declarations @gol
211 -Wdisabled-optimization -Wno-div-by-zero -Werror @gol
212 -Wfloat-equal -Wformat -Wformat=2 @gol
213 -Wformat-nonliteral -Wformat-security @gol
214 -Wimplicit -Wimplicit-int @gol
215 -Wimplicit-function-declaration @gol
216 -Werror-implicit-function-declaration @gol
217 -Wimport -Winline -Winvalid-pch -Wno-endif-labels @gol
218 -Wlarger-than-@var{len} -Wlong-long @gol
219 -Wmain -Wmissing-braces @gol
220 -Wmissing-format-attribute -Wmissing-noreturn @gol
221 -Wno-multichar -Wno-format-extra-args -Wno-format-y2k @gol
222 -Wno-import -Wnonnull -Wpacked -Wpadded @gol
223 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
224 -Wreturn-type -Wsequence-point -Wshadow @gol
225 -Wsign-compare -Wstrict-aliasing @gol
226 -Wswitch -Wswitch-default -Wswitch-enum @gol
227 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
228 -Wunknown-pragmas -Wunreachable-code @gol
229 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
230 -Wunused-value -Wunused-variable -Wwrite-strings}
232 @item C-only Warning Options
233 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
234 -Wmissing-prototypes -Wnested-externs @gol
235 -Wstrict-prototypes -Wtraditional}
237 @item Debugging Options
238 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
239 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
240 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
241 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
242 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
243 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
244 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
245 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
246 -fmem-report -fprofile-arcs @gol
247 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
248 -ftest-coverage -ftime-report @gol
249 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
250 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
251 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
252 -print-multi-directory -print-multi-lib @gol
253 -print-prog-name=@var{program} -print-search-dirs -Q @gol
256 @item Optimization Options
257 @xref{Optimize Options,,Options that Control Optimization}.
258 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
259 -falign-labels=@var{n} -falign-loops=@var{n} @gol
260 -fbranch-probabilities -fcaller-saves -fcprop-registers @gol
261 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
262 -fdelayed-branch -fdelete-null-pointer-checks @gol
263 -fexpensive-optimizations -ffast-math -ffloat-store @gol
264 -fforce-addr -fforce-mem -ffunction-sections @gol
265 -fgcse -fgcse-lm -fgcse-sm -floop-optimize -fcrossjumping @gol
266 -fif-conversion -fif-conversion2 @gol
267 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
268 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
269 -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
270 -fno-default-inline -fno-defer-pop @gol
271 -fno-function-cse -fno-guess-branch-probability @gol
272 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
273 -funsafe-math-optimizations -ffinite-math-only @gol
274 -fno-trapping-math -fno-zero-initialized-in-bss @gol
275 -fomit-frame-pointer -foptimize-register-move @gol
276 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
277 -freduce-all-givs -fregmove -frename-registers @gol
278 -freorder-blocks -freorder-functions @gol
279 -frerun-cse-after-loop -frerun-loop-opt @gol
280 -fschedule-insns -fschedule-insns2 @gol
281 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
282 -fsched-spec-load-dangerous -fsched2-use-superblocks @gol
283 -fsched2-use-traces -fsignaling-nans @gol
284 -fsingle-precision-constant -fssa -fssa-ccp -fssa-dce @gol
285 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
286 -funroll-all-loops -funroll-loops -fpeel-loops @gol
287 -funswitch-loops -fold-unroll-loops -fold-unroll-all-loops @gol
288 --param @var{name}=@var{value}
289 -O -O0 -O1 -O2 -O3 -Os}
291 @item Preprocessor Options
292 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
293 @gccoptlist{-A@var{question}=@var{answer} @gol
294 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
295 -C -dD -dI -dM -dN @gol
296 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
297 -idirafter @var{dir} @gol
298 -include @var{file} -imacros @var{file} @gol
299 -iprefix @var{file} -iwithprefix @var{dir} @gol
300 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
301 -M -MM -MF -MG -MP -MQ -MT -nostdinc -P -remap @gol
302 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
303 -Xpreprocessor @var{option}}
305 @item Assembler Option
306 @xref{Assembler Options,,Passing Options to the Assembler}.
307 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
310 @xref{Link Options,,Options for Linking}.
311 @gccoptlist{@var{object-file-name} -l@var{library} @gol
312 -nostartfiles -nodefaultlibs -nostdlib @gol
313 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
314 -Wl,@var{option} -Xlinker @var{option} @gol
317 @item Directory Options
318 @xref{Directory Options,,Options for Directory Search}.
319 @gccoptlist{-B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
322 @c I wrote this xref this way to avoid overfull hbox. -- rms
323 @xref{Target Options}.
324 @gccoptlist{-V @var{version} -b @var{machine}}
326 @item Machine Dependent Options
327 @xref{Submodel Options,,Hardware Models and Configurations}.
329 @emph{M680x0 Options}
330 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
331 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
332 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
333 -malign-int -mstrict-align}
335 @emph{M68hc1x Options}
336 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
337 -mauto-incdec -minmax -mlong-calls -mshort @gol
338 -msoft-reg-count=@var{count}}
341 @gccoptlist{-mg -mgnu -munix}
344 @gccoptlist{-mcpu=@var{cpu-type} @gol
345 -mtune=@var{cpu-type} @gol
346 -mcmodel=@var{code-model} @gol
348 -mapp-regs -mbroken-saverestore -mcypress @gol
349 -mfaster-structs -mflat @gol
350 -mfpu -mhard-float -mhard-quad-float @gol
351 -mimpure-text -mlive-g0 -mno-app-regs @gol
352 -mno-faster-structs -mno-flat -mno-fpu @gol
353 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
354 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
355 -msupersparc -munaligned-doubles -mv8}
358 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
359 -mapcs-26 -mapcs-32 @gol
360 -mapcs-stack-check -mno-apcs-stack-check @gol
361 -mapcs-float -mno-apcs-float @gol
362 -mapcs-reentrant -mno-apcs-reentrant @gol
363 -msched-prolog -mno-sched-prolog @gol
364 -mlittle-endian -mbig-endian -mwords-little-endian @gol
365 -malignment-traps -mno-alignment-traps @gol
366 -msoft-float -mhard-float -mfpe @gol
367 -mthumb-interwork -mno-thumb-interwork @gol
368 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
369 -mstructure-size-boundary=@var{n} @gol
370 -mabort-on-noreturn @gol
371 -mlong-calls -mno-long-calls @gol
372 -msingle-pic-base -mno-single-pic-base @gol
373 -mpic-register=@var{reg} @gol
374 -mnop-fun-dllimport @gol
375 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
376 -mpoke-function-name @gol
378 -mtpcs-frame -mtpcs-leaf-frame @gol
379 -mcaller-super-interworking -mcallee-super-interworking}
381 @emph{MN10200 Options}
384 @emph{MN10300 Options}
385 @gccoptlist{-mmult-bug -mno-mult-bug @gol
386 -mam33 -mno-am33 @gol
389 @emph{M32R/D Options}
390 @gccoptlist{-m32rx -m32r -mcode-model=@var{model-type} @gol
391 -msdata=@var{sdata-type} -G @var{num}}
394 @gccoptlist{-m88000 -m88100 -m88110 -mbig-pic @gol
395 -mcheck-zero-division -mhandle-large-shift @gol
396 -midentify-revision -mno-check-zero-division @gol
397 -mno-ocs-debug-info -mno-ocs-frame-position @gol
398 -mno-optimize-arg-area -mno-serialize-volatile @gol
399 -mno-underscores -mocs-debug-info @gol
400 -mocs-frame-position -moptimize-arg-area @gol
401 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
402 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
403 -mversion-03.00 -mwarn-passed-structs}
405 @emph{RS/6000 and PowerPC Options}
406 @gccoptlist{-mcpu=@var{cpu-type} @gol
407 -mtune=@var{cpu-type} @gol
408 -mpower -mno-power -mpower2 -mno-power2 @gol
409 -mpowerpc -mpowerpc64 -mno-powerpc @gol
410 -maltivec -mno-altivec @gol
411 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
412 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
413 -mnew-mnemonics -mold-mnemonics @gol
414 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
415 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
416 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
417 -mstring -mno-string -mupdate -mno-update @gol
418 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
419 -mstrict-align -mno-strict-align -mrelocatable @gol
420 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
421 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
422 -mdynamic-no-pic @gol
423 -mcall-sysv -mcall-netbsd @gol
424 -maix-struct-return -msvr4-struct-return @gol
425 -mabi=altivec -mabi=no-altivec @gol
426 -mabi=spe -mabi=no-spe @gol
427 -misel=yes -misel=no @gol
428 -mspe=yes -mspe=no @gol
429 -mfloat-gprs=yes -mfloat-gprs=no @gol
430 -mprototype -mno-prototype @gol
431 -msim -mmvme -mads -myellowknife -memb -msdata @gol
432 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
434 @emph{Darwin Options}
435 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
436 -arch_only -bind_at_load -bundle -bundle_loader @gol
437 -client_name -compatibility_version -current_version @gol
438 -dependency-file -dylib_file -dylinker_install_name @gol
439 -dynamic -dynamiclib -exported_symbols_list @gol
440 -filelist -flat_namespace -force_cpusubtype_ALL @gol
441 -force_flat_namespace -headerpad_max_install_names @gol
442 -image_base -init -install_name -keep_private_externs @gol
443 -multi_module -multiply_defined -multiply_defined_unused @gol
444 -noall_load -nomultidefs -noprebind -noseglinkedit @gol
445 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
446 -private_bundle -read_only_relocs -sectalign @gol
447 -sectobjectsymbols -whyload -seg1addr @gol
448 -sectcreate -sectobjectsymbols -sectorder @gol
449 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
450 -segprot -segs_read_only_addr -segs_read_write_addr @gol
451 -single_module -static -sub_library -sub_umbrella @gol
452 -twolevel_namespace -umbrella -undefined @gol
453 -unexported_symbols_list -weak_reference_mismatches @gol
457 @gccoptlist{-mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
458 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
459 -mminimum-fp-blocks -mnohc-struct-return}
462 @gccoptlist{-mabicalls -march=@var{cpu-type} -mtune=@var{cpu=type} @gol
463 -mcpu=@var{cpu-type} -membedded-data -muninit-const-in-rodata @gol
464 -membedded-pic -mfp32 -mfp64 -mfused-madd -mno-fused-madd @gol
465 -mgas -mgp32 -mgp64 @gol
466 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1 @gol
467 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
468 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
469 -mmips-as -mmips-tfile -mno-abicalls @gol
470 -mno-embedded-data -mno-uninit-const-in-rodata @gol
471 -mno-embedded-pic -mno-gpopt -mno-long-calls @gol
472 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats @gol
473 -mrnames -msoft-float @gol
474 -m4650 -msingle-float -mmad @gol
475 -mstats -EL -EB -G @var{num} -nocpp @gol
476 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi -mabi-fake-default @gol
477 -mfix7000 -mno-crt0 -mflush-func=@var{func} -mno-flush-func @gol
478 -mbranch-likely -mno-branch-likely}
480 @emph{i386 and x86-64 Options}
481 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
482 -mfpmath=@var{unit} @gol
483 -masm=@var{dialect} -mno-fancy-math-387 @gol
484 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
485 -mno-wide-multiply -mrtd -malign-double @gol
486 -mpreferred-stack-boundary=@var{num} @gol
487 -mmmx -msse -msse2 -m3dnow @gol
488 -mthreads -mno-align-stringops -minline-all-stringops @gol
489 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
490 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
492 -mcmodel=@var{code-model} @gol
496 @gccoptlist{-march=@var{architecture-type} @gol
497 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
498 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
499 -mjump-in-delay -mlinker-opt -mlong-calls @gol
500 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
501 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
502 -mno-jump-in-delay -mno-long-load-store @gol
503 -mno-portable-runtime -mno-soft-float @gol
504 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
505 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
506 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
507 -nolibdld -static -threads}
509 @emph{Intel 960 Options}
510 @gccoptlist{-m@var{cpu-type} -masm-compat -mclean-linkage @gol
511 -mcode-align -mcomplex-addr -mleaf-procedures @gol
512 -mic-compat -mic2.0-compat -mic3.0-compat @gol
513 -mintel-asm -mno-clean-linkage -mno-code-align @gol
514 -mno-complex-addr -mno-leaf-procedures @gol
515 -mno-old-align -mno-strict-align -mno-tail-call @gol
516 -mnumerics -mold-align -msoft-float -mstrict-align @gol
519 @emph{DEC Alpha Options}
520 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
521 -mieee -mieee-with-inexact -mieee-conformant @gol
522 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
523 -mtrap-precision=@var{mode} -mbuild-constants @gol
524 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
525 -mbwx -mmax -mfix -mcix @gol
526 -mfloat-vax -mfloat-ieee @gol
527 -mexplicit-relocs -msmall-data -mlarge-data @gol
528 -msmall-text -mlarge-text @gol
529 -mmemory-latency=@var{time}}
531 @emph{DEC Alpha/VMS Options}
532 @gccoptlist{-mvms-return-codes}
534 @emph{H8/300 Options}
535 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
538 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
539 -m4-nofpu -m4-single-only -m4-single -m4 @gol
540 -m5-64media -m5-64media-nofpu @gol
541 -m5-32media -m5-32media-nofpu @gol
542 -m5-compact -m5-compact-nofpu @gol
543 -mb -ml -mdalign -mrelax @gol
544 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
545 -mieee -misize -mpadstruct -mspace @gol
546 -mprefergot -musermode}
548 @emph{System V Options}
549 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
552 @gccoptlist{-EB -EL @gol
553 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
554 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
556 @emph{TMS320C3x/C4x Options}
557 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
558 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
559 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
560 -mparallel-insns -mparallel-mpy -mpreserve-float}
563 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
564 -mprolog-function -mno-prolog-function -mspace @gol
565 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
566 -mapp-regs -mno-app-regs @gol
567 -mdisable-callt -mno-disable-callt @gol
572 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
573 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
574 -mregparam -mnoregparam -msb -mnosb @gol
575 -mbitfield -mnobitfield -mhimem -mnohimem}
578 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
579 -mcall-prologues -mno-tablejump -mtiny-stack}
582 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
583 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
584 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
585 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
586 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
589 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
590 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
591 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
592 -mno-base-addresses -msingle-exit -mno-single-exit}
595 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
596 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
597 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
598 -minline-float-divide-max-throughput @gol
599 -minline-int-divide-min-latency @gol
600 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
601 -mfixed-range=@var{register-range}}
604 @gccoptlist{-mextmem -mextmemory -monchip -mno-asm-optimize @gol
605 -masm-optimize -mbranch-cost=@var{n} -mcond-exec=@var{n}}
607 @emph{S/390 and zSeries Options}
608 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
609 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
610 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
611 -m64 -m31 -mdebug -mno-debug -mesa -mzarch}
614 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
615 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
616 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
617 -mstack-align -mdata-align -mconst-align @gol
618 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
619 -melf -maout -melinux -mlinux -sim -sim2}
621 @emph{PDP-11 Options}
622 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
623 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
624 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
625 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
626 -mbranch-expensive -mbranch-cheap @gol
627 -msplit -mno-split -munix-asm -mdec-asm}
629 @emph{Xstormy16 Options}
632 @emph{Xtensa Options}
633 @gccoptlist{-mbig-endian -mlittle-endian @gol
634 -mdensity -mno-density @gol
635 -mconst16 -mno-const16 @gol
636 -mmac16 -mno-mac16 @gol
637 -mmul16 -mno-mul16 @gol
638 -mmul32 -mno-mul32 @gol
640 -mminmax -mno-minmax @gol
641 -msext -mno-sext @gol
642 -mbooleans -mno-booleans @gol
643 -mhard-float -msoft-float @gol
644 -mfused-madd -mno-fused-madd @gol
645 -mserialize-volatile -mno-serialize-volatile @gol
646 -mtext-section-literals -mno-text-section-literals @gol
647 -mtarget-align -mno-target-align @gol
648 -mlongcalls -mno-longcalls}
651 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
652 -mhard-float -msoft-float @gol
653 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
654 -mdouble -mno-double @gol
655 -mmedia -mno-media -mmuladd -mno-muladd @gol
656 -mlibrary-pic -macc-4 -macc-8 @gol
657 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
658 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
659 -mvliw-branch -mno-vliw-branch @gol
660 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
661 -mno-nested-cond-exec -mtomcat-stats @gol
664 @item Code Generation Options
665 @xref{Code Gen Options,,Options for Code Generation Conventions}.
666 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
667 -ffixed-@var{reg} -fexceptions @gol
668 -fnon-call-exceptions -funwind-tables @gol
669 -fasynchronous-unwind-tables @gol
670 -finhibit-size-directive -finstrument-functions @gol
671 -fno-common -fno-ident -fno-gnu-linker @gol
672 -fpcc-struct-return -fpic -fPIC @gol
673 -freg-struct-return -fshared-data -fshort-enums @gol
674 -fshort-double -fshort-wchar @gol
675 -fverbose-asm -fpack-struct -fstack-check @gol
676 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
677 -fargument-alias -fargument-noalias @gol
678 -fargument-noalias-global -fleading-underscore @gol
679 -ftls-model=@var{model} @gol
680 -ftrapv -fbounds-check}
684 * Overall Options:: Controlling the kind of output:
685 an executable, object files, assembler files,
686 or preprocessed source.
687 * C Dialect Options:: Controlling the variant of C language compiled.
688 * C++ Dialect Options:: Variations on C++.
689 * Objective-C Dialect Options:: Variations on Objective-C.
690 * Language Independent Options:: Controlling how diagnostics should be
692 * Warning Options:: How picky should the compiler be?
693 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
694 * Optimize Options:: How much optimization?
695 * Preprocessor Options:: Controlling header files and macro definitions.
696 Also, getting dependency information for Make.
697 * Assembler Options:: Passing options to the assembler.
698 * Link Options:: Specifying libraries and so on.
699 * Directory Options:: Where to find header files and libraries.
700 Where to find the compiler executable files.
701 * Spec Files:: How to pass switches to sub-processes.
702 * Target Options:: Running a cross-compiler, or an old version of GCC.
705 @node Overall Options
706 @section Options Controlling the Kind of Output
708 Compilation can involve up to four stages: preprocessing, compilation
709 proper, assembly and linking, always in that order. The first three
710 stages apply to an individual source file, and end by producing an
711 object file; linking combines all the object files (those newly
712 compiled, and those specified as input) into an executable file.
714 @cindex file name suffix
715 For any given input file, the file name suffix determines what kind of
720 C source code which must be preprocessed.
723 C source code which should not be preprocessed.
726 C++ source code which should not be preprocessed.
729 Objective-C source code. Note that you must link with the library
730 @file{libobjc.a} to make an Objective-C program work.
733 Objective-C source code which should not be preprocessed.
736 C or C++ header file to be turned into a precompiled header.
740 @itemx @var{file}.cxx
741 @itemx @var{file}.cpp
742 @itemx @var{file}.CPP
743 @itemx @var{file}.c++
745 C++ source code which must be preprocessed. Note that in @samp{.cxx},
746 the last two letters must both be literally @samp{x}. Likewise,
747 @samp{.C} refers to a literal capital C@.
751 C++ header file to be turned into a precompiled header.
754 @itemx @var{file}.for
755 @itemx @var{file}.FOR
756 Fortran source code which should not be preprocessed.
759 @itemx @var{file}.fpp
760 @itemx @var{file}.FPP
761 Fortran source code which must be preprocessed (with the traditional
765 Fortran source code which must be preprocessed with a RATFOR
766 preprocessor (not included with GCC)@.
768 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
769 Using and Porting GNU Fortran}, for more details of the handling of
772 @c FIXME: Descriptions of Java file types.
779 Ada source code file which contains a library unit declaration (a
780 declaration of a package, subprogram, or generic, or a generic
781 instantiation), or a library unit renaming declaration (a package,
782 generic, or subprogram renaming declaration). Such files are also
785 @itemx @var{file}.adb
786 Ada source code file containing a library unit body (a subprogram or
787 package body). Such files are also called @dfn{bodies}.
789 @c GCC also knows about some suffixes for languages not yet included:
798 Assembler code which must be preprocessed.
801 An object file to be fed straight into linking.
802 Any file name with no recognized suffix is treated this way.
806 You can specify the input language explicitly with the @option{-x} option:
809 @item -x @var{language}
810 Specify explicitly the @var{language} for the following input files
811 (rather than letting the compiler choose a default based on the file
812 name suffix). This option applies to all following input files until
813 the next @option{-x} option. Possible values for @var{language} are:
815 c c-header cpp-output
816 c++ c++-header c++-cpp-output
817 objective-c objective-c-header objc-cpp-output
818 assembler assembler-with-cpp
820 f77 f77-cpp-input ratfor
826 Turn off any specification of a language, so that subsequent files are
827 handled according to their file name suffixes (as they are if @option{-x}
828 has not been used at all).
830 @item -pass-exit-codes
831 @opindex pass-exit-codes
832 Normally the @command{gcc} program will exit with the code of 1 if any
833 phase of the compiler returns a non-success return code. If you specify
834 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
835 numerically highest error produced by any phase that returned an error
839 If you only want some of the stages of compilation, you can use
840 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
841 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
842 @command{gcc} is to stop. Note that some combinations (for example,
843 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
848 Compile or assemble the source files, but do not link. The linking
849 stage simply is not done. The ultimate output is in the form of an
850 object file for each source file.
852 By default, the object file name for a source file is made by replacing
853 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
855 Unrecognized input files, not requiring compilation or assembly, are
860 Stop after the stage of compilation proper; do not assemble. The output
861 is in the form of an assembler code file for each non-assembler input
864 By default, the assembler file name for a source file is made by
865 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
867 Input files that don't require compilation are ignored.
871 Stop after the preprocessing stage; do not run the compiler proper. The
872 output is in the form of preprocessed source code, which is sent to the
875 Input files which don't require preprocessing are ignored.
877 @cindex output file option
880 Place output in file @var{file}. This applies regardless to whatever
881 sort of output is being produced, whether it be an executable file,
882 an object file, an assembler file or preprocessed C code.
884 Since only one output file can be specified, it does not make sense to
885 use @option{-o} when compiling more than one input file, unless you are
886 producing an executable file as output.
888 If @option{-o} is not specified, the default is to put an executable file
889 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
890 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
891 all preprocessed C source on standard output.
895 Print (on standard error output) the commands executed to run the stages
896 of compilation. Also print the version number of the compiler driver
897 program and of the preprocessor and the compiler proper.
901 Like @option{-v} except the commands are not executed and all command
902 arguments are quoted. This is useful for shell scripts to capture the
903 driver-generated command lines.
907 Use pipes rather than temporary files for communication between the
908 various stages of compilation. This fails to work on some systems where
909 the assembler is unable to read from a pipe; but the GNU assembler has
914 Print (on the standard output) a description of the command line options
915 understood by @command{gcc}. If the @option{-v} option is also specified
916 then @option{--help} will also be passed on to the various processes
917 invoked by @command{gcc}, so that they can display the command line options
918 they accept. If the @option{-Wextra} option is also specified then command
919 line options which have no documentation associated with them will also
924 Print (on the standard output) a description of target specific command
925 line options for each tool.
929 Display the version number and copyrights of the invoked GCC.
933 @section Compiling C++ Programs
935 @cindex suffixes for C++ source
936 @cindex C++ source file suffixes
937 C++ source files conventionally use one of the suffixes @samp{.C},
938 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
939 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
940 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
941 files with these names and compiles them as C++ programs even if you
942 call the compiler the same way as for compiling C programs (usually
943 with the name @command{gcc}).
947 However, C++ programs often require class libraries as well as a
948 compiler that understands the C++ language---and under some
949 circumstances, you might want to compile programs or header files from
950 standard input, or otherwise without a suffix that flags them as C++
951 programs. You might also like to precompile a C header file with a
952 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
953 program that calls GCC with the default language set to C++, and
954 automatically specifies linking against the C++ library. On many
955 systems, @command{g++} is also installed with the name @command{c++}.
957 @cindex invoking @command{g++}
958 When you compile C++ programs, you may specify many of the same
959 command-line options that you use for compiling programs in any
960 language; or command-line options meaningful for C and related
961 languages; or options that are meaningful only for C++ programs.
962 @xref{C Dialect Options,,Options Controlling C Dialect}, for
963 explanations of options for languages related to C@.
964 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
965 explanations of options that are meaningful only for C++ programs.
967 @node C Dialect Options
968 @section Options Controlling C Dialect
969 @cindex dialect options
970 @cindex language dialect options
971 @cindex options, dialect
973 The following options control the dialect of C (or languages derived
974 from C, such as C++ and Objective-C) that the compiler accepts:
981 In C mode, support all ISO C90 programs. In C++ mode,
982 remove GNU extensions that conflict with ISO C++.
984 This turns off certain features of GCC that are incompatible with ISO
985 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
986 such as the @code{asm} and @code{typeof} keywords, and
987 predefined macros such as @code{unix} and @code{vax} that identify the
988 type of system you are using. It also enables the undesirable and
989 rarely used ISO trigraph feature. For the C compiler,
990 it disables recognition of C++ style @samp{//} comments as well as
991 the @code{inline} keyword.
993 The alternate keywords @code{__asm__}, @code{__extension__},
994 @code{__inline__} and @code{__typeof__} continue to work despite
995 @option{-ansi}. You would not want to use them in an ISO C program, of
996 course, but it is useful to put them in header files that might be included
997 in compilations done with @option{-ansi}. Alternate predefined macros
998 such as @code{__unix__} and @code{__vax__} are also available, with or
999 without @option{-ansi}.
1001 The @option{-ansi} option does not cause non-ISO programs to be
1002 rejected gratuitously. For that, @option{-pedantic} is required in
1003 addition to @option{-ansi}. @xref{Warning Options}.
1005 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1006 option is used. Some header files may notice this macro and refrain
1007 from declaring certain functions or defining certain macros that the
1008 ISO standard doesn't call for; this is to avoid interfering with any
1009 programs that might use these names for other things.
1011 Functions which would normally be built in but do not have semantics
1012 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1013 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1014 built-in functions provided by GCC}, for details of the functions
1019 Determine the language standard. This option is currently only
1020 supported when compiling C or C++. A value for this option must be
1021 provided; possible values are
1026 ISO C90 (same as @option{-ansi}).
1028 @item iso9899:199409
1029 ISO C90 as modified in amendment 1.
1035 ISO C99. Note that this standard is not yet fully supported; see
1036 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1037 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1040 Default, ISO C90 plus GNU extensions (including some C99 features).
1044 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1045 this will become the default. The name @samp{gnu9x} is deprecated.
1048 The 1998 ISO C++ standard plus amendments.
1051 The same as @option{-std=c++98} plus GNU extensions. This is the
1052 default for C++ code.
1055 Even when this option is not specified, you can still use some of the
1056 features of newer standards in so far as they do not conflict with
1057 previous C standards. For example, you may use @code{__restrict__} even
1058 when @option{-std=c99} is not specified.
1060 The @option{-std} options specifying some version of ISO C have the same
1061 effects as @option{-ansi}, except that features that were not in ISO C90
1062 but are in the specified version (for example, @samp{//} comments and
1063 the @code{inline} keyword in ISO C99) are not disabled.
1065 @xref{Standards,,Language Standards Supported by GCC}, for details of
1066 these standard versions.
1068 @item -aux-info @var{filename}
1070 Output to the given filename prototyped declarations for all functions
1071 declared and/or defined in a translation unit, including those in header
1072 files. This option is silently ignored in any language other than C@.
1074 Besides declarations, the file indicates, in comments, the origin of
1075 each declaration (source file and line), whether the declaration was
1076 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1077 @samp{O} for old, respectively, in the first character after the line
1078 number and the colon), and whether it came from a declaration or a
1079 definition (@samp{C} or @samp{F}, respectively, in the following
1080 character). In the case of function definitions, a K&R-style list of
1081 arguments followed by their declarations is also provided, inside
1082 comments, after the declaration.
1086 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1087 keyword, so that code can use these words as identifiers. You can use
1088 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1089 instead. @option{-ansi} implies @option{-fno-asm}.
1091 In C++, this switch only affects the @code{typeof} keyword, since
1092 @code{asm} and @code{inline} are standard keywords. You may want to
1093 use the @option{-fno-gnu-keywords} flag instead, which has the same
1094 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1095 switch only affects the @code{asm} and @code{typeof} keywords, since
1096 @code{inline} is a standard keyword in ISO C99.
1099 @itemx -fno-builtin-@var{function}
1100 @opindex fno-builtin
1101 @cindex built-in functions
1102 Don't recognize built-in functions that do not begin with
1103 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1104 functions provided by GCC}, for details of the functions affected,
1105 including those which are not built-in functions when @option{-ansi} or
1106 @option{-std} options for strict ISO C conformance are used because they
1107 do not have an ISO standard meaning.
1109 GCC normally generates special code to handle certain built-in functions
1110 more efficiently; for instance, calls to @code{alloca} may become single
1111 instructions that adjust the stack directly, and calls to @code{memcpy}
1112 may become inline copy loops. The resulting code is often both smaller
1113 and faster, but since the function calls no longer appear as such, you
1114 cannot set a breakpoint on those calls, nor can you change the behavior
1115 of the functions by linking with a different library.
1117 With the @option{-fno-builtin-@var{function}} option
1118 only the built-in function @var{function} is
1119 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1120 function is named this is not built-in in this version of GCC, this
1121 option is ignored. There is no corresponding
1122 @option{-fbuiltin-@var{function}} option; if you wish to enable
1123 built-in functions selectively when using @option{-fno-builtin} or
1124 @option{-ffreestanding}, you may define macros such as:
1127 #define abs(n) __builtin_abs ((n))
1128 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1133 @cindex hosted environment
1135 Assert that compilation takes place in a hosted environment. This implies
1136 @option{-fbuiltin}. A hosted environment is one in which the
1137 entire standard library is available, and in which @code{main} has a return
1138 type of @code{int}. Examples are nearly everything except a kernel.
1139 This is equivalent to @option{-fno-freestanding}.
1141 @item -ffreestanding
1142 @opindex ffreestanding
1143 @cindex hosted environment
1145 Assert that compilation takes place in a freestanding environment. This
1146 implies @option{-fno-builtin}. A freestanding environment
1147 is one in which the standard library may not exist, and program startup may
1148 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1149 This is equivalent to @option{-fno-hosted}.
1151 @xref{Standards,,Language Standards Supported by GCC}, for details of
1152 freestanding and hosted environments.
1154 @item -fms-extensions
1155 @opindex fms-extensions
1156 Accept some non-standard constructs used in Microsoft header files.
1160 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1161 options for strict ISO C conformance) implies @option{-trigraphs}.
1163 @item -no-integrated-cpp
1164 @opindex no-integrated-cpp
1165 Performs a compilation in two passes: preprocessing and compiling. This
1166 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1167 @option{-B} option. The user supplied compilation step can then add in
1168 an additional preprocessing step after normal preprocessing but before
1169 compiling. The default is to use the integrated cpp (internal cpp)
1171 The semantics of this option will change if "cc1", "cc1plus", and
1172 "cc1obj" are merged.
1174 @cindex traditional C language
1175 @cindex C language, traditional
1177 @itemx -traditional-cpp
1178 @opindex traditional-cpp
1179 @opindex traditional
1180 Formerly, these options caused GCC to attempt to emulate a pre-standard
1181 C compiler. They are now only supported with the @option{-E} switch.
1182 The preprocessor continues to support a pre-standard mode. See the GNU
1183 CPP manual for details.
1185 @item -fcond-mismatch
1186 @opindex fcond-mismatch
1187 Allow conditional expressions with mismatched types in the second and
1188 third arguments. The value of such an expression is void. This option
1189 is not supported for C++.
1191 @item -funsigned-char
1192 @opindex funsigned-char
1193 Let the type @code{char} be unsigned, like @code{unsigned char}.
1195 Each kind of machine has a default for what @code{char} should
1196 be. It is either like @code{unsigned char} by default or like
1197 @code{signed char} by default.
1199 Ideally, a portable program should always use @code{signed char} or
1200 @code{unsigned char} when it depends on the signedness of an object.
1201 But many programs have been written to use plain @code{char} and
1202 expect it to be signed, or expect it to be unsigned, depending on the
1203 machines they were written for. This option, and its inverse, let you
1204 make such a program work with the opposite default.
1206 The type @code{char} is always a distinct type from each of
1207 @code{signed char} or @code{unsigned char}, even though its behavior
1208 is always just like one of those two.
1211 @opindex fsigned-char
1212 Let the type @code{char} be signed, like @code{signed char}.
1214 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1215 the negative form of @option{-funsigned-char}. Likewise, the option
1216 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1218 @item -fsigned-bitfields
1219 @itemx -funsigned-bitfields
1220 @itemx -fno-signed-bitfields
1221 @itemx -fno-unsigned-bitfields
1222 @opindex fsigned-bitfields
1223 @opindex funsigned-bitfields
1224 @opindex fno-signed-bitfields
1225 @opindex fno-unsigned-bitfields
1226 These options control whether a bit-field is signed or unsigned, when the
1227 declaration does not use either @code{signed} or @code{unsigned}. By
1228 default, such a bit-field is signed, because this is consistent: the
1229 basic integer types such as @code{int} are signed types.
1231 @item -fwritable-strings
1232 @opindex fwritable-strings
1233 Store string constants in the writable data segment and don't uniquize
1234 them. This is for compatibility with old programs which assume they can
1235 write into string constants.
1237 Writing into string constants is a very bad idea; ``constants'' should
1241 @node C++ Dialect Options
1242 @section Options Controlling C++ Dialect
1244 @cindex compiler options, C++
1245 @cindex C++ options, command line
1246 @cindex options, C++
1247 This section describes the command-line options that are only meaningful
1248 for C++ programs; but you can also use most of the GNU compiler options
1249 regardless of what language your program is in. For example, you
1250 might compile a file @code{firstClass.C} like this:
1253 g++ -g -frepo -O -c firstClass.C
1257 In this example, only @option{-frepo} is an option meant
1258 only for C++ programs; you can use the other options with any
1259 language supported by GCC@.
1261 Here is a list of options that are @emph{only} for compiling C++ programs:
1265 @item -fabi-version=@var{n}
1266 @opindex fabi-version
1267 Use version @var{n} of the C++ ABI. Version 1 is the version of the C++
1268 ABI that first appeared in G++ 3.2. Version 0 will always be the
1269 version that conforms most closely to the C++ ABI specification.
1270 Therefore, the ABI obtained using version 0 will change as ABI bugs are
1273 The default is version 1.
1275 @item -fno-access-control
1276 @opindex fno-access-control
1277 Turn off all access checking. This switch is mainly useful for working
1278 around bugs in the access control code.
1282 Check that the pointer returned by @code{operator new} is non-null
1283 before attempting to modify the storage allocated. This check is
1284 normally unnecessary because the C++ standard specifies that
1285 @code{operator new} will only return @code{0} if it is declared
1286 @samp{throw()}, in which case the compiler will always check the
1287 return value even without this option. In all other cases, when
1288 @code{operator new} has a non-empty exception specification, memory
1289 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1290 @samp{new (nothrow)}.
1292 @item -fconserve-space
1293 @opindex fconserve-space
1294 Put uninitialized or runtime-initialized global variables into the
1295 common segment, as C does. This saves space in the executable at the
1296 cost of not diagnosing duplicate definitions. If you compile with this
1297 flag and your program mysteriously crashes after @code{main()} has
1298 completed, you may have an object that is being destroyed twice because
1299 two definitions were merged.
1301 This option is no longer useful on most targets, now that support has
1302 been added for putting variables into BSS without making them common.
1304 @item -fno-const-strings
1305 @opindex fno-const-strings
1306 Give string constants type @code{char *} instead of type @code{const
1307 char *}. By default, G++ uses type @code{const char *} as required by
1308 the standard. Even if you use @option{-fno-const-strings}, you cannot
1309 actually modify the value of a string constant, unless you also use
1310 @option{-fwritable-strings}.
1312 This option might be removed in a future release of G++. For maximum
1313 portability, you should structure your code so that it works with
1314 string constants that have type @code{const char *}.
1316 @item -fdollars-in-identifiers
1317 @opindex fdollars-in-identifiers
1318 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1319 @samp{$} with the option @option{-fno-dollars-in-identifiers}. (GNU C allows
1320 @samp{$} by default on most target systems, but there are a few exceptions.)
1321 Traditional C allowed the character @samp{$} to form part of
1322 identifiers. However, ISO C and C++ forbid @samp{$} in identifiers.
1324 @item -fno-elide-constructors
1325 @opindex fno-elide-constructors
1326 The C++ standard allows an implementation to omit creating a temporary
1327 which is only used to initialize another object of the same type.
1328 Specifying this option disables that optimization, and forces G++ to
1329 call the copy constructor in all cases.
1331 @item -fno-enforce-eh-specs
1332 @opindex fno-enforce-eh-specs
1333 Don't check for violation of exception specifications at runtime. This
1334 option violates the C++ standard, but may be useful for reducing code
1335 size in production builds, much like defining @samp{NDEBUG}. The compiler
1336 will still optimize based on the exception specifications.
1338 @item -fexternal-templates
1339 @opindex fexternal-templates
1341 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1342 template instantiation; template instances are emitted or not according
1343 to the location of the template definition. @xref{Template
1344 Instantiation}, for more information.
1346 This option is deprecated.
1348 @item -falt-external-templates
1349 @opindex falt-external-templates
1350 Similar to @option{-fexternal-templates}, but template instances are
1351 emitted or not according to the place where they are first instantiated.
1352 @xref{Template Instantiation}, for more information.
1354 This option is deprecated.
1357 @itemx -fno-for-scope
1359 @opindex fno-for-scope
1360 If @option{-ffor-scope} is specified, the scope of variables declared in
1361 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1362 as specified by the C++ standard.
1363 If @option{-fno-for-scope} is specified, the scope of variables declared in
1364 a @i{for-init-statement} extends to the end of the enclosing scope,
1365 as was the case in old versions of G++, and other (traditional)
1366 implementations of C++.
1368 The default if neither flag is given to follow the standard,
1369 but to allow and give a warning for old-style code that would
1370 otherwise be invalid, or have different behavior.
1372 @item -fno-gnu-keywords
1373 @opindex fno-gnu-keywords
1374 Do not recognize @code{typeof} as a keyword, so that code can use this
1375 word as an identifier. You can use the keyword @code{__typeof__} instead.
1376 @option{-ansi} implies @option{-fno-gnu-keywords}.
1378 @item -fno-implicit-templates
1379 @opindex fno-implicit-templates
1380 Never emit code for non-inline templates which are instantiated
1381 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1382 @xref{Template Instantiation}, for more information.
1384 @item -fno-implicit-inline-templates
1385 @opindex fno-implicit-inline-templates
1386 Don't emit code for implicit instantiations of inline templates, either.
1387 The default is to handle inlines differently so that compiles with and
1388 without optimization will need the same set of explicit instantiations.
1390 @item -fno-implement-inlines
1391 @opindex fno-implement-inlines
1392 To save space, do not emit out-of-line copies of inline functions
1393 controlled by @samp{#pragma implementation}. This will cause linker
1394 errors if these functions are not inlined everywhere they are called.
1396 @item -fms-extensions
1397 @opindex fms-extensions
1398 Disable pedantic warnings about constructs used in MFC, such as implicit
1399 int and getting a pointer to member function via non-standard syntax.
1401 @item -fno-nonansi-builtins
1402 @opindex fno-nonansi-builtins
1403 Disable built-in declarations of functions that are not mandated by
1404 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1405 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1407 @item -fno-operator-names
1408 @opindex fno-operator-names
1409 Do not treat the operator name keywords @code{and}, @code{bitand},
1410 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1411 synonyms as keywords.
1413 @item -fno-optional-diags
1414 @opindex fno-optional-diags
1415 Disable diagnostics that the standard says a compiler does not need to
1416 issue. Currently, the only such diagnostic issued by G++ is the one for
1417 a name having multiple meanings within a class.
1420 @opindex fpermissive
1421 Downgrade messages about nonconformant code from errors to warnings. By
1422 default, G++ effectively sets @option{-pedantic-errors} without
1423 @option{-pedantic}; this option reverses that. This behavior and this
1424 option are superseded by @option{-pedantic}, which works as it does for GNU C@.
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 Emit special relocations for vtables and virtual function references
1464 so that the linker can identify unused virtual functions and zero out
1465 vtable slots that refer to them. This is most useful with
1466 @option{-ffunction-sections} and @option{-Wl,--gc-sections}, in order to
1467 also discard the functions themselves.
1469 This optimization requires GNU as and GNU ld. Not all systems support
1470 this option. @option{-Wl,--gc-sections} is ignored without @option{-static}.
1474 Do not use weak symbol support, even if it is provided by the linker.
1475 By default, G++ will use weak symbols if they are available. This
1476 option exists only for testing, and should not be used by end-users;
1477 it will result in inferior code and has no benefits. This option may
1478 be removed in a future release of G++.
1482 Do not search for header files in the standard directories specific to
1483 C++, but do still search the other standard directories. (This option
1484 is used when building the C++ library.)
1487 In addition, these optimization, warning, and code generation options
1488 have meanings only for C++ programs:
1491 @item -fno-default-inline
1492 @opindex fno-default-inline
1493 Do not assume @samp{inline} for functions defined inside a class scope.
1494 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1495 functions will have linkage like inline functions; they just won't be
1498 @item -Wabi @r{(C++ only)}
1500 Warn when G++ generates code that is probably not compatible with the
1501 vendor-neutral C++ ABI. Although an effort has been made to warn about
1502 all such cases, there are probably some cases that are not warned about,
1503 even though G++ is generating incompatible code. There may also be
1504 cases where warnings are emitted even though the code that is generated
1507 You should rewrite your code to avoid these warnings if you are
1508 concerned about the fact that code generated by G++ may not be binary
1509 compatible with code generated by other compilers.
1511 The known incompatibilities at this point include:
1516 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1517 pack data into the same byte as a base class. For example:
1520 struct A @{ virtual void f(); int f1 : 1; @};
1521 struct B : public A @{ int f2 : 1; @};
1525 In this case, G++ will place @code{B::f2} into the same byte
1526 as@code{A::f1}; other compilers will not. You can avoid this problem
1527 by explicitly padding @code{A} so that its size is a multiple of the
1528 byte size on your platform; that will cause G++ and other compilers to
1529 layout @code{B} identically.
1532 Incorrect handling of tail-padding for virtual bases. G++ does not use
1533 tail padding when laying out virtual bases. For example:
1536 struct A @{ virtual void f(); char c1; @};
1537 struct B @{ B(); char c2; @};
1538 struct C : public A, public virtual B @{@};
1542 In this case, G++ will not place @code{B} into the tail-padding for
1543 @code{A}; other compilers will. You can avoid this problem by
1544 explicitly padding @code{A} so that its size is a multiple of its
1545 alignment (ignoring virtual base classes); that will cause G++ and other
1546 compilers to layout @code{C} identically.
1549 Incorrect handling of bit-fields with declared widths greater than that
1550 of their underlying types, when the bit-fields appear in a union. For
1554 union U @{ int i : 4096; @};
1558 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1559 union too small by the number of bits in an @code{int}.
1562 Empty classes can be placed at incorrect offsets. For example:
1572 struct C : public B, public A @{@};
1576 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1577 it should be placed at offset zero. G++ mistakenly believes that the
1578 @code{A} data member of @code{B} is already at offset zero.
1581 Names of template functions whose types involve @code{typename} or
1582 template template parameters can be mangled incorrectly.
1585 template <typename Q>
1586 void f(typename Q::X) @{@}
1588 template <template <typename> class Q>
1589 void f(typename Q<int>::X) @{@}
1593 Instantiations of these templates may be mangled incorrectly.
1597 @item -Wctor-dtor-privacy @r{(C++ only)}
1598 @opindex Wctor-dtor-privacy
1599 Warn when a class seems unusable, because all the constructors or
1600 destructors in a class are private and the class has no friends or
1601 public static member functions.
1603 @item -Wnon-virtual-dtor @r{(C++ only)}
1604 @opindex Wnon-virtual-dtor
1605 Warn when a class declares a non-virtual destructor that should probably
1606 be virtual, because it looks like the class will be used polymorphically.
1607 This warning is enabled by @option{-Wall}.
1609 @item -Wreorder @r{(C++ only)}
1611 @cindex reordering, warning
1612 @cindex warning for reordering of member initializers
1613 Warn when the order of member initializers given in the code does not
1614 match the order in which they must be executed. For instance:
1620 A(): j (0), i (1) @{ @}
1624 Here the compiler will warn that the member initializers for @samp{i}
1625 and @samp{j} will be rearranged to match the declaration order of the
1626 members. This warning is enabled by @option{-Wall}.
1629 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1632 @item -Weffc++ @r{(C++ only)}
1634 Warn about violations of the following style guidelines from Scott Meyers'
1635 @cite{Effective C++} book:
1639 Item 11: Define a copy constructor and an assignment operator for classes
1640 with dynamically allocated memory.
1643 Item 12: Prefer initialization to assignment in constructors.
1646 Item 14: Make destructors virtual in base classes.
1649 Item 15: Have @code{operator=} return a reference to @code{*this}.
1652 Item 23: Don't try to return a reference when you must return an object.
1656 and about violations of the following style guidelines from Scott Meyers'
1657 @cite{More Effective C++} book:
1661 Item 6: Distinguish between prefix and postfix forms of increment and
1662 decrement operators.
1665 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1669 If you use this option, you should be aware that the standard library
1670 headers do not obey all of these guidelines; you can use @samp{grep -v}
1671 to filter out those warnings.
1673 @item -Wno-deprecated @r{(C++ only)}
1674 @opindex Wno-deprecated
1675 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1677 @item -Wno-non-template-friend @r{(C++ only)}
1678 @opindex Wno-non-template-friend
1679 Disable warnings when non-templatized friend functions are declared
1680 within a template. With the advent of explicit template specification
1681 support in G++, if the name of the friend is an unqualified-id (i.e.,
1682 @samp{friend foo(int)}), the C++ language specification demands that the
1683 friend declare or define an ordinary, nontemplate function. (Section
1684 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1685 could be interpreted as a particular specialization of a templatized
1686 function. Because this non-conforming behavior is no longer the default
1687 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1688 check existing code for potential trouble spots, and is on by default.
1689 This new compiler behavior can be turned off with
1690 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1691 but disables the helpful warning.
1693 @item -Wold-style-cast @r{(C++ only)}
1694 @opindex Wold-style-cast
1695 Warn if an old-style (C-style) cast to a non-void type is used within
1696 a C++ program. The new-style casts (@samp{static_cast},
1697 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1698 unintended effects, and much easier to grep for.
1700 @item -Woverloaded-virtual @r{(C++ only)}
1701 @opindex Woverloaded-virtual
1702 @cindex overloaded virtual fn, warning
1703 @cindex warning for overloaded virtual fn
1704 Warn when a function declaration hides virtual functions from a
1705 base class. For example, in:
1712 struct B: public A @{
1717 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1725 will fail to compile.
1727 @item -Wno-pmf-conversions @r{(C++ only)}
1728 @opindex Wno-pmf-conversions
1729 Disable the diagnostic for converting a bound pointer to member function
1732 @item -Wsign-promo @r{(C++ only)}
1733 @opindex Wsign-promo
1734 Warn when overload resolution chooses a promotion from unsigned or
1735 enumeral type to a signed type over a conversion to an unsigned type of
1736 the same size. Previous versions of G++ would try to preserve
1737 unsignedness, but the standard mandates the current behavior.
1739 @item -Wsynth @r{(C++ only)}
1741 @cindex warning for synthesized methods
1742 @cindex synthesized methods, warning
1743 Warn when G++'s synthesis behavior does not match that of cfront. For
1749 A& operator = (int);
1759 In this example, G++ will synthesize a default @samp{A& operator =
1760 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1763 @node Objective-C Dialect Options
1764 @section Options Controlling Objective-C Dialect
1766 @cindex compiler options, Objective-C
1767 @cindex Objective-C options, command line
1768 @cindex options, Objective-C
1769 This section describes the command-line options that are only meaningful
1770 for Objective-C programs; but you can also use most of the GNU compiler
1771 options regardless of what language your program is in. For example,
1772 you might compile a file @code{some_class.m} like this:
1775 gcc -g -fgnu-runtime -O -c some_class.m
1779 In this example, only @option{-fgnu-runtime} is an option meant only for
1780 Objective-C programs; you can use the other options with any language
1783 Here is a list of options that are @emph{only} for compiling Objective-C
1787 @item -fconstant-string-class=@var{class-name}
1788 @opindex fconstant-string-class
1789 Use @var{class-name} as the name of the class to instantiate for each
1790 literal string specified with the syntax @code{@@"@dots{}"}. The default
1791 class name is @code{NXConstantString}.
1794 @opindex fgnu-runtime
1795 Generate object code compatible with the standard GNU Objective-C
1796 runtime. This is the default for most types of systems.
1798 @item -fnext-runtime
1799 @opindex fnext-runtime
1800 Generate output compatible with the NeXT runtime. This is the default
1801 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1802 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1807 Dump interface declarations for all classes seen in the source file to a
1808 file named @file{@var{sourcename}.decl}.
1811 @opindex Wno-protocol
1812 If a class is declared to implement a protocol, a warning is issued for
1813 every method in the protocol that is not implemented by the class. The
1814 default behavior is to issue a warning for every method not explicitly
1815 implemented in the class, even if a method implementation is inherited
1816 from the superclass. If you use the @code{-Wno-protocol} option, then
1817 methods inherited from the superclass are considered to be implemented,
1818 and no warning is issued for them.
1822 Warn if multiple methods of different types for the same selector are
1823 found during compilation. The check is performed on the list of methods
1824 in the final stage of compilation. Additionally, a check is performed
1825 that for each selector appearing in a @code{@@selector(@dots{})}
1826 expression, a corresponding method with that selector has been found
1827 during compilation. Because these checks scan the method table only at
1828 the end of compilation, these warnings are not produced if the final
1829 stage of compilation is not reached, for example because an error is
1830 found during compilation, or because the @code{-fsyntax-only} option is
1833 @item -Wundeclared-selector
1834 @opindex Wundeclared-selector
1835 Warn if a @code{@@selector(@dots{})} expression referring to an
1836 undeclared selector is found. A selector is considered undeclared if no
1837 method with that name has been declared (explicitly, in an
1838 @code{@@interface} or @code{@@protocol} declaration, or implicitly, in
1839 an @code{@@implementation} section) before the
1840 @code{@@selector(@dots{})} expression. This option always performs its
1841 checks as soon as a @code{@@selector(@dots{})} expression is found
1842 (while @code{-Wselector} only performs its checks in the final stage of
1843 compilation), and so additionally enforces the coding style convention
1844 that methods and selectors must be declared before being used.
1846 @c not documented because only avail via -Wp
1847 @c @item -print-objc-runtime-info
1851 @node Language Independent Options
1852 @section Options to Control Diagnostic Messages Formatting
1853 @cindex options to control diagnostics formatting
1854 @cindex diagnostic messages
1855 @cindex message formatting
1857 Traditionally, diagnostic messages have been formatted irrespective of
1858 the output device's aspect (e.g.@: its width, @dots{}). The options described
1859 below can be used to control the diagnostic messages formatting
1860 algorithm, e.g.@: how many characters per line, how often source location
1861 information should be reported. Right now, only the C++ front end can
1862 honor these options. However it is expected, in the near future, that
1863 the remaining front ends would be able to digest them correctly.
1866 @item -fmessage-length=@var{n}
1867 @opindex fmessage-length
1868 Try to format error messages so that they fit on lines of about @var{n}
1869 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1870 the front ends supported by GCC@. If @var{n} is zero, then no
1871 line-wrapping will be done; each error message will appear on a single
1874 @opindex fdiagnostics-show-location
1875 @item -fdiagnostics-show-location=once
1876 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1877 reporter to emit @emph{once} source location information; that is, in
1878 case the message is too long to fit on a single physical line and has to
1879 be wrapped, the source location won't be emitted (as prefix) again,
1880 over and over, in subsequent continuation lines. This is the default
1883 @item -fdiagnostics-show-location=every-line
1884 Only meaningful in line-wrapping mode. Instructs the diagnostic
1885 messages reporter to emit the same source location information (as
1886 prefix) for physical lines that result from the process of breaking
1887 a message which is too long to fit on a single line.
1891 @node Warning Options
1892 @section Options to Request or Suppress Warnings
1893 @cindex options to control warnings
1894 @cindex warning messages
1895 @cindex messages, warning
1896 @cindex suppressing warnings
1898 Warnings are diagnostic messages that report constructions which
1899 are not inherently erroneous but which are risky or suggest there
1900 may have been an error.
1902 You can request many specific warnings with options beginning @samp{-W},
1903 for example @option{-Wimplicit} to request warnings on implicit
1904 declarations. Each of these specific warning options also has a
1905 negative form beginning @samp{-Wno-} to turn off warnings;
1906 for example, @option{-Wno-implicit}. This manual lists only one of the
1907 two forms, whichever is not the default.
1909 The following options control the amount and kinds of warnings produced
1910 by GCC; for further, language-specific options also refer to
1911 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
1914 @cindex syntax checking
1916 @opindex fsyntax-only
1917 Check the code for syntax errors, but don't do anything beyond that.
1921 Issue all the warnings demanded by strict ISO C and ISO C++;
1922 reject all programs that use forbidden extensions, and some other
1923 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1924 version of the ISO C standard specified by any @option{-std} option used.
1926 Valid ISO C and ISO C++ programs should compile properly with or without
1927 this option (though a rare few will require @option{-ansi} or a
1928 @option{-std} option specifying the required version of ISO C)@. However,
1929 without this option, certain GNU extensions and traditional C and C++
1930 features are supported as well. With this option, they are rejected.
1932 @option{-pedantic} does not cause warning messages for use of the
1933 alternate keywords whose names begin and end with @samp{__}. Pedantic
1934 warnings are also disabled in the expression that follows
1935 @code{__extension__}. However, only system header files should use
1936 these escape routes; application programs should avoid them.
1937 @xref{Alternate Keywords}.
1939 Some users try to use @option{-pedantic} to check programs for strict ISO
1940 C conformance. They soon find that it does not do quite what they want:
1941 it finds some non-ISO practices, but not all---only those for which
1942 ISO C @emph{requires} a diagnostic, and some others for which
1943 diagnostics have been added.
1945 A feature to report any failure to conform to ISO C might be useful in
1946 some instances, but would require considerable additional work and would
1947 be quite different from @option{-pedantic}. We don't have plans to
1948 support such a feature in the near future.
1950 Where the standard specified with @option{-std} represents a GNU
1951 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1952 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1953 extended dialect is based. Warnings from @option{-pedantic} are given
1954 where they are required by the base standard. (It would not make sense
1955 for such warnings to be given only for features not in the specified GNU
1956 C dialect, since by definition the GNU dialects of C include all
1957 features the compiler supports with the given option, and there would be
1958 nothing to warn about.)
1960 @item -pedantic-errors
1961 @opindex pedantic-errors
1962 Like @option{-pedantic}, except that errors are produced rather than
1967 Inhibit all warning messages.
1971 Inhibit warning messages about the use of @samp{#import}.
1973 @item -Wchar-subscripts
1974 @opindex Wchar-subscripts
1975 Warn if an array subscript has type @code{char}. This is a common cause
1976 of error, as programmers often forget that this type is signed on some
1981 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1982 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1986 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1987 the arguments supplied have types appropriate to the format string
1988 specified, and that the conversions specified in the format string make
1989 sense. This includes standard functions, and others specified by format
1990 attributes (@pxref{Function Attributes}), in the @code{printf},
1991 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1992 not in the C standard) families.
1994 The formats are checked against the format features supported by GNU
1995 libc version 2.2. These include all ISO C90 and C99 features, as well
1996 as features from the Single Unix Specification and some BSD and GNU
1997 extensions. Other library implementations may not support all these
1998 features; GCC does not support warning about features that go beyond a
1999 particular library's limitations. However, if @option{-pedantic} is used
2000 with @option{-Wformat}, warnings will be given about format features not
2001 in the selected standard version (but not for @code{strfmon} formats,
2002 since those are not in any version of the C standard). @xref{C Dialect
2003 Options,,Options Controlling C Dialect}.
2005 Since @option{-Wformat} also checks for null format arguments for
2006 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2008 @option{-Wformat} is included in @option{-Wall}. For more control over some
2009 aspects of format checking, the options @option{-Wno-format-y2k},
2010 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2011 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2012 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2014 @item -Wno-format-y2k
2015 @opindex Wno-format-y2k
2016 If @option{-Wformat} is specified, do not warn about @code{strftime}
2017 formats which may yield only a two-digit year.
2019 @item -Wno-format-extra-args
2020 @opindex Wno-format-extra-args
2021 If @option{-Wformat} is specified, do not warn about excess arguments to a
2022 @code{printf} or @code{scanf} format function. The C standard specifies
2023 that such arguments are ignored.
2025 Where the unused arguments lie between used arguments that are
2026 specified with @samp{$} operand number specifications, normally
2027 warnings are still given, since the implementation could not know what
2028 type to pass to @code{va_arg} to skip the unused arguments. However,
2029 in the case of @code{scanf} formats, this option will suppress the
2030 warning if the unused arguments are all pointers, since the Single
2031 Unix Specification says that such unused arguments are allowed.
2033 @item -Wno-format-zero-length
2034 @opindex Wno-format-zero-length
2035 If @option{-Wformat} is specified, do not warn about zero-length formats.
2036 The C standard specifies that zero-length formats are allowed.
2038 @item -Wformat-nonliteral
2039 @opindex Wformat-nonliteral
2040 If @option{-Wformat} is specified, also warn if the format string is not a
2041 string literal and so cannot be checked, unless the format function
2042 takes its format arguments as a @code{va_list}.
2044 @item -Wformat-security
2045 @opindex Wformat-security
2046 If @option{-Wformat} is specified, also warn about uses of format
2047 functions that represent possible security problems. At present, this
2048 warns about calls to @code{printf} and @code{scanf} functions where the
2049 format string is not a string literal and there are no format arguments,
2050 as in @code{printf (foo);}. This may be a security hole if the format
2051 string came from untrusted input and contains @samp{%n}. (This is
2052 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2053 in future warnings may be added to @option{-Wformat-security} that are not
2054 included in @option{-Wformat-nonliteral}.)
2058 Enable @option{-Wformat} plus format checks not included in
2059 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2060 -Wformat-nonliteral -Wformat-security}.
2064 Enable warning about passing a null pointer for arguments marked as
2065 requiring a non-null value by the @code{nonnull} function attribute.
2067 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2068 can be disabled with the @option{-Wno-nonnull} option.
2070 @item -Wimplicit-int
2071 @opindex Wimplicit-int
2072 Warn when a declaration does not specify a type.
2074 @item -Wimplicit-function-declaration
2075 @itemx -Werror-implicit-function-declaration
2076 @opindex Wimplicit-function-declaration
2077 @opindex Werror-implicit-function-declaration
2078 Give a warning (or error) whenever a function is used before being
2083 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2087 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2088 function with external linkage, returning int, taking either zero
2089 arguments, two, or three arguments of appropriate types.
2091 @item -Wmissing-braces
2092 @opindex Wmissing-braces
2093 Warn if an aggregate or union initializer is not fully bracketed. In
2094 the following example, the initializer for @samp{a} is not fully
2095 bracketed, but that for @samp{b} is fully bracketed.
2098 int a[2][2] = @{ 0, 1, 2, 3 @};
2099 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2103 @opindex Wparentheses
2104 Warn if parentheses are omitted in certain contexts, such
2105 as when there is an assignment in a context where a truth value
2106 is expected, or when operators are nested whose precedence people
2107 often get confused about.
2109 Also warn about constructions where there may be confusion to which
2110 @code{if} statement an @code{else} branch belongs. Here is an example of
2125 In C, every @code{else} branch belongs to the innermost possible @code{if}
2126 statement, which in this example is @code{if (b)}. This is often not
2127 what the programmer expected, as illustrated in the above example by
2128 indentation the programmer chose. When there is the potential for this
2129 confusion, GCC will issue a warning when this flag is specified.
2130 To eliminate the warning, add explicit braces around the innermost
2131 @code{if} statement so there is no way the @code{else} could belong to
2132 the enclosing @code{if}. The resulting code would look like this:
2148 @item -Wsequence-point
2149 @opindex Wsequence-point
2150 Warn about code that may have undefined semantics because of violations
2151 of sequence point rules in the C standard.
2153 The C standard defines the order in which expressions in a C program are
2154 evaluated in terms of @dfn{sequence points}, which represent a partial
2155 ordering between the execution of parts of the program: those executed
2156 before the sequence point, and those executed after it. These occur
2157 after the evaluation of a full expression (one which is not part of a
2158 larger expression), after the evaluation of the first operand of a
2159 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2160 function is called (but after the evaluation of its arguments and the
2161 expression denoting the called function), and in certain other places.
2162 Other than as expressed by the sequence point rules, the order of
2163 evaluation of subexpressions of an expression is not specified. All
2164 these rules describe only a partial order rather than a total order,
2165 since, for example, if two functions are called within one expression
2166 with no sequence point between them, the order in which the functions
2167 are called is not specified. However, the standards committee have
2168 ruled that function calls do not overlap.
2170 It is not specified when between sequence points modifications to the
2171 values of objects take effect. Programs whose behavior depends on this
2172 have undefined behavior; the C standard specifies that ``Between the
2173 previous and next sequence point an object shall have its stored value
2174 modified at most once by the evaluation of an expression. Furthermore,
2175 the prior value shall be read only to determine the value to be
2176 stored.''. If a program breaks these rules, the results on any
2177 particular implementation are entirely unpredictable.
2179 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2180 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2181 diagnosed by this option, and it may give an occasional false positive
2182 result, but in general it has been found fairly effective at detecting
2183 this sort of problem in programs.
2185 The present implementation of this option only works for C programs. A
2186 future implementation may also work for C++ programs.
2188 The C standard is worded confusingly, therefore there is some debate
2189 over the precise meaning of the sequence point rules in subtle cases.
2190 Links to discussions of the problem, including proposed formal
2191 definitions, may be found on our readings page, at
2192 @w{@uref{http://gcc.gnu.org/readings.html}}.
2195 @opindex Wreturn-type
2196 Warn whenever a function is defined with a return-type that defaults to
2197 @code{int}. Also warn about any @code{return} statement with no
2198 return-value in a function whose return-type is not @code{void}.
2200 For C++, a function without return type always produces a diagnostic
2201 message, even when @option{-Wno-return-type} is specified. The only
2202 exceptions are @samp{main} and functions defined in system headers.
2206 Warn whenever a @code{switch} statement has an index of enumeral type
2207 and lacks a @code{case} for one or more of the named codes of that
2208 enumeration. (The presence of a @code{default} label prevents this
2209 warning.) @code{case} labels outside the enumeration range also
2210 provoke warnings when this option is used.
2212 @item -Wswitch-default
2213 @opindex Wswitch-switch
2214 Warn whenever a @code{switch} statement does not have a @code{default}
2218 @opindex Wswitch-enum
2219 Warn whenever a @code{switch} statement has an index of enumeral type
2220 and lacks a @code{case} for one or more of the named codes of that
2221 enumeration. @code{case} labels outside the enumeration range also
2222 provoke warnings when this option is used.
2226 Warn if any trigraphs are encountered that might change the meaning of
2227 the program (trigraphs within comments are not warned about).
2229 @item -Wunused-function
2230 @opindex Wunused-function
2231 Warn whenever a static function is declared but not defined or a
2232 non\-inline static function is unused.
2234 @item -Wunused-label
2235 @opindex Wunused-label
2236 Warn whenever a label is declared but not used.
2238 To suppress this warning use the @samp{unused} attribute
2239 (@pxref{Variable Attributes}).
2241 @item -Wunused-parameter
2242 @opindex Wunused-parameter
2243 Warn whenever a function parameter is unused aside from its declaration.
2245 To suppress this warning use the @samp{unused} attribute
2246 (@pxref{Variable Attributes}).
2248 @item -Wunused-variable
2249 @opindex Wunused-variable
2250 Warn whenever a local variable or non-constant static variable is unused
2251 aside from its declaration
2253 To suppress this warning use the @samp{unused} attribute
2254 (@pxref{Variable Attributes}).
2256 @item -Wunused-value
2257 @opindex Wunused-value
2258 Warn whenever a statement computes a result that is explicitly not used.
2260 To suppress this warning cast the expression to @samp{void}.
2264 All the above @option{-Wunused} options combined.
2266 In order to get a warning about an unused function parameter, you must
2267 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2268 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2270 @item -Wuninitialized
2271 @opindex Wuninitialized
2272 Warn if an automatic variable is used without first being initialized or
2273 if a variable may be clobbered by a @code{setjmp} call.
2275 These warnings are possible only in optimizing compilation,
2276 because they require data flow information that is computed only
2277 when optimizing. If you don't specify @option{-O}, you simply won't
2280 These warnings occur only for variables that are candidates for
2281 register allocation. Therefore, they do not occur for a variable that
2282 is declared @code{volatile}, or whose address is taken, or whose size
2283 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2284 structures, unions or arrays, even when they are in registers.
2286 Note that there may be no warning about a variable that is used only
2287 to compute a value that itself is never used, because such
2288 computations may be deleted by data flow analysis before the warnings
2291 These warnings are made optional because GCC is not smart
2292 enough to see all the reasons why the code might be correct
2293 despite appearing to have an error. Here is one example of how
2314 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2315 always initialized, but GCC doesn't know this. Here is
2316 another common case:
2321 if (change_y) save_y = y, y = new_y;
2323 if (change_y) y = save_y;
2328 This has no bug because @code{save_y} is used only if it is set.
2330 @cindex @code{longjmp} warnings
2331 This option also warns when a non-volatile automatic variable might be
2332 changed by a call to @code{longjmp}. These warnings as well are possible
2333 only in optimizing compilation.
2335 The compiler sees only the calls to @code{setjmp}. It cannot know
2336 where @code{longjmp} will be called; in fact, a signal handler could
2337 call it at any point in the code. As a result, you may get a warning
2338 even when there is in fact no problem because @code{longjmp} cannot
2339 in fact be called at the place which would cause a problem.
2341 Some spurious warnings can be avoided if you declare all the functions
2342 you use that never return as @code{noreturn}. @xref{Function
2345 @item -Wunknown-pragmas
2346 @opindex Wunknown-pragmas
2347 @cindex warning for unknown pragmas
2348 @cindex unknown pragmas, warning
2349 @cindex pragmas, warning of unknown
2350 Warn when a #pragma directive is encountered which is not understood by
2351 GCC@. If this command line option is used, warnings will even be issued
2352 for unknown pragmas in system header files. This is not the case if
2353 the warnings were only enabled by the @option{-Wall} command line option.
2355 @item -Wstrict-aliasing
2356 @opindex Wstrict-aliasing
2357 This option is only active when @option{-fstrict-aliasing} is active.
2358 It warns about code which might break the strict aliasing rules that the
2359 compiler is using for optimization. The warning does not catch all
2360 cases, but does attempt to catch the more common pitfalls. It is
2361 included in @option{-Wall}.
2365 All of the above @samp{-W} options combined. This enables all the
2366 warnings about constructions that some users consider questionable, and
2367 that are easy to avoid (or modify to prevent the warning), even in
2368 conjunction with macros. This also enables some language-specific
2369 warnings described in @ref{C++ Dialect Options} and
2370 @ref{Objective-C Dialect Options}.
2373 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2374 Some of them warn about constructions that users generally do not
2375 consider questionable, but which occasionally you might wish to check
2376 for; others warn about constructions that are necessary or hard to avoid
2377 in some cases, and there is no simple way to modify the code to suppress
2384 (This option used to be called @option{-W}. The older name is still
2385 supported, but the newer name is more descriptive.) Print extra warning
2386 messages for these events:
2390 A function can return either with or without a value. (Falling
2391 off the end of the function body is considered returning without
2392 a value.) For example, this function would evoke such a
2406 An expression-statement or the left-hand side of a comma expression
2407 contains no side effects.
2408 To suppress the warning, cast the unused expression to void.
2409 For example, an expression such as @samp{x[i,j]} will cause a warning,
2410 but @samp{x[(void)i,j]} will not.
2413 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2416 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2417 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2418 that of ordinary mathematical notation.
2421 Storage-class specifiers like @code{static} are not the first things in
2422 a declaration. According to the C Standard, this usage is obsolescent.
2425 The return type of a function has a type qualifier such as @code{const}.
2426 Such a type qualifier has no effect, since the value returned by a
2427 function is not an lvalue. (But don't warn about the GNU extension of
2428 @code{volatile void} return types. That extension will be warned about
2429 if @option{-pedantic} is specified.)
2432 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2436 A comparison between signed and unsigned values could produce an
2437 incorrect result when the signed value is converted to unsigned.
2438 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2441 An aggregate has an initializer which does not initialize all members.
2442 For example, the following code would cause such a warning, because
2443 @code{x.h} would be implicitly initialized to zero:
2446 struct s @{ int f, g, h; @};
2447 struct s x = @{ 3, 4 @};
2451 A function parameter is declared without a type specifier in K&R-style
2459 An empty body occurs in an @samp{if} or @samp{else} statement.
2462 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2463 @samp{>}, or @samp{>=}.
2466 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2469 Any of several floating-point events that often indicate errors, such as
2470 overflow, underflow, loss of precision, etc.
2472 @item @r{(C++ only)}
2473 An enumerator and a non-enumerator both appear in a conditional expression.
2475 @item @r{(C++ only)}
2476 A non-static reference or non-static @samp{const} member appears in a
2477 class without constructors.
2479 @item @r{(C++ only)}
2480 Ambiguous virtual bases.
2482 @item @r{(C++ only)}
2483 Subscripting an array which has been declared @samp{register}.
2485 @item @r{(C++ only)}
2486 Taking the address of a variable which has been declared @samp{register}.
2488 @item @r{(C++ only)}
2489 A base class is not initialized in a derived class' copy constructor.
2492 @item -Wno-div-by-zero
2493 @opindex Wno-div-by-zero
2494 @opindex Wdiv-by-zero
2495 Do not warn about compile-time integer division by zero. Floating point
2496 division by zero is not warned about, as it can be a legitimate way of
2497 obtaining infinities and NaNs.
2499 @item -Wsystem-headers
2500 @opindex Wsystem-headers
2501 @cindex warnings from system headers
2502 @cindex system headers, warnings from
2503 Print warning messages for constructs found in system header files.
2504 Warnings from system headers are normally suppressed, on the assumption
2505 that they usually do not indicate real problems and would only make the
2506 compiler output harder to read. Using this command line option tells
2507 GCC to emit warnings from system headers as if they occurred in user
2508 code. However, note that using @option{-Wall} in conjunction with this
2509 option will @emph{not} warn about unknown pragmas in system
2510 headers---for that, @option{-Wunknown-pragmas} must also be used.
2513 @opindex Wfloat-equal
2514 Warn if floating point values are used in equality comparisons.
2516 The idea behind this is that sometimes it is convenient (for the
2517 programmer) to consider floating-point values as approximations to
2518 infinitely precise real numbers. If you are doing this, then you need
2519 to compute (by analyzing the code, or in some other way) the maximum or
2520 likely maximum error that the computation introduces, and allow for it
2521 when performing comparisons (and when producing output, but that's a
2522 different problem). In particular, instead of testing for equality, you
2523 would check to see whether the two values have ranges that overlap; and
2524 this is done with the relational operators, so equality comparisons are
2527 @item -Wtraditional @r{(C only)}
2528 @opindex Wtraditional
2529 Warn about certain constructs that behave differently in traditional and
2530 ISO C@. Also warn about ISO C constructs that have no traditional C
2531 equivalent, and/or problematic constructs which should be avoided.
2535 Macro parameters that appear within string literals in the macro body.
2536 In traditional C macro replacement takes place within string literals,
2537 but does not in ISO C@.
2540 In traditional C, some preprocessor directives did not exist.
2541 Traditional preprocessors would only consider a line to be a directive
2542 if the @samp{#} appeared in column 1 on the line. Therefore
2543 @option{-Wtraditional} warns about directives that traditional C
2544 understands but would ignore because the @samp{#} does not appear as the
2545 first character on the line. It also suggests you hide directives like
2546 @samp{#pragma} not understood by traditional C by indenting them. Some
2547 traditional implementations would not recognize @samp{#elif}, so it
2548 suggests avoiding it altogether.
2551 A function-like macro that appears without arguments.
2554 The unary plus operator.
2557 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2558 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2559 constants.) Note, these suffixes appear in macros defined in the system
2560 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2561 Use of these macros in user code might normally lead to spurious
2562 warnings, however gcc's integrated preprocessor has enough context to
2563 avoid warning in these cases.
2566 A function declared external in one block and then used after the end of
2570 A @code{switch} statement has an operand of type @code{long}.
2573 A non-@code{static} function declaration follows a @code{static} one.
2574 This construct is not accepted by some traditional C compilers.
2577 The ISO type of an integer constant has a different width or
2578 signedness from its traditional type. This warning is only issued if
2579 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2580 typically represent bit patterns, are not warned about.
2583 Usage of ISO string concatenation is detected.
2586 Initialization of automatic aggregates.
2589 Identifier conflicts with labels. Traditional C lacks a separate
2590 namespace for labels.
2593 Initialization of unions. If the initializer is zero, the warning is
2594 omitted. This is done under the assumption that the zero initializer in
2595 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2596 initializer warnings and relies on default initialization to zero in the
2600 Conversions by prototypes between fixed/floating point values and vice
2601 versa. The absence of these prototypes when compiling with traditional
2602 C would cause serious problems. This is a subset of the possible
2603 conversion warnings, for the full set use @option{-Wconversion}.
2606 Use of ISO C style function definitions. This warning intentionally is
2607 @emph{not} issued for prototype declarations or variadic functions
2608 because these ISO C features will appear in your code when using
2609 libiberty's traditional C compatibility macros, @code{PARAMS} and
2610 @code{VPARAMS}. This warning is also bypassed for nested functions
2611 because that feature is already a gcc extension and thus not relevant to
2612 traditional C compatibility.
2617 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2619 @item -Wendif-labels
2620 @opindex Wendif-labels
2621 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2625 Warn whenever a local variable shadows another local variable, parameter or
2626 global variable or whenever a built-in function is shadowed.
2628 @item -Wlarger-than-@var{len}
2629 @opindex Wlarger-than
2630 Warn whenever an object of larger than @var{len} bytes is defined.
2632 @item -Wpointer-arith
2633 @opindex Wpointer-arith
2634 Warn about anything that depends on the ``size of'' a function type or
2635 of @code{void}. GNU C assigns these types a size of 1, for
2636 convenience in calculations with @code{void *} pointers and pointers
2639 @item -Wbad-function-cast @r{(C only)}
2640 @opindex Wbad-function-cast
2641 Warn whenever a function call is cast to a non-matching type.
2642 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2646 Warn whenever a pointer is cast so as to remove a type qualifier from
2647 the target type. For example, warn if a @code{const char *} is cast
2648 to an ordinary @code{char *}.
2651 @opindex Wcast-align
2652 Warn whenever a pointer is cast such that the required alignment of the
2653 target is increased. For example, warn if a @code{char *} is cast to
2654 an @code{int *} on machines where integers can only be accessed at
2655 two- or four-byte boundaries.
2657 @item -Wwrite-strings
2658 @opindex Wwrite-strings
2659 When compiling C, give string constants the type @code{const
2660 char[@var{length}]} so that
2661 copying the address of one into a non-@code{const} @code{char *}
2662 pointer will get a warning; when compiling C++, warn about the
2663 deprecated conversion from string constants to @code{char *}.
2664 These warnings will help you find at
2665 compile time code that can try to write into a string constant, but
2666 only if you have been very careful about using @code{const} in
2667 declarations and prototypes. Otherwise, it will just be a nuisance;
2668 this is why we did not make @option{-Wall} request these warnings.
2671 @opindex Wconversion
2672 Warn if a prototype causes a type conversion that is different from what
2673 would happen to the same argument in the absence of a prototype. This
2674 includes conversions of fixed point to floating and vice versa, and
2675 conversions changing the width or signedness of a fixed point argument
2676 except when the same as the default promotion.
2678 Also, warn if a negative integer constant expression is implicitly
2679 converted to an unsigned type. For example, warn about the assignment
2680 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2681 casts like @code{(unsigned) -1}.
2683 @item -Wsign-compare
2684 @opindex Wsign-compare
2685 @cindex warning for comparison of signed and unsigned values
2686 @cindex comparison of signed and unsigned values, warning
2687 @cindex signed and unsigned values, comparison warning
2688 Warn when a comparison between signed and unsigned values could produce
2689 an incorrect result when the signed value is converted to unsigned.
2690 This warning is also enabled by @option{-Wextra}; to get the other warnings
2691 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2693 @item -Waggregate-return
2694 @opindex Waggregate-return
2695 Warn if any functions that return structures or unions are defined or
2696 called. (In languages where you can return an array, this also elicits
2699 @item -Wstrict-prototypes @r{(C only)}
2700 @opindex Wstrict-prototypes
2701 Warn if a function is declared or defined without specifying the
2702 argument types. (An old-style function definition is permitted without
2703 a warning if preceded by a declaration which specifies the argument
2706 @item -Wmissing-prototypes @r{(C only)}
2707 @opindex Wmissing-prototypes
2708 Warn if a global function is defined without a previous prototype
2709 declaration. This warning is issued even if the definition itself
2710 provides a prototype. The aim is to detect global functions that fail
2711 to be declared in header files.
2713 @item -Wmissing-declarations @r{(C only)}
2714 @opindex Wmissing-declarations
2715 Warn if a global function is defined without a previous declaration.
2716 Do so even if the definition itself provides a prototype.
2717 Use this option to detect global functions that are not declared in
2720 @item -Wmissing-noreturn
2721 @opindex Wmissing-noreturn
2722 Warn about functions which might be candidates for attribute @code{noreturn}.
2723 Note these are only possible candidates, not absolute ones. Care should
2724 be taken to manually verify functions actually do not ever return before
2725 adding the @code{noreturn} attribute, otherwise subtle code generation
2726 bugs could be introduced. You will not get a warning for @code{main} in
2727 hosted C environments.
2729 @item -Wmissing-format-attribute
2730 @opindex Wmissing-format-attribute
2732 If @option{-Wformat} is enabled, also warn about functions which might be
2733 candidates for @code{format} attributes. Note these are only possible
2734 candidates, not absolute ones. GCC will guess that @code{format}
2735 attributes might be appropriate for any function that calls a function
2736 like @code{vprintf} or @code{vscanf}, but this might not always be the
2737 case, and some functions for which @code{format} attributes are
2738 appropriate may not be detected. This option has no effect unless
2739 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2741 @item -Wno-multichar
2742 @opindex Wno-multichar
2744 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2745 Usually they indicate a typo in the user's code, as they have
2746 implementation-defined values, and should not be used in portable code.
2748 @item -Wno-deprecated-declarations
2749 @opindex Wno-deprecated-declarations
2750 Do not warn about uses of functions, variables, and types marked as
2751 deprecated by using the @code{deprecated} attribute.
2752 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2753 @pxref{Type Attributes}.)
2757 Warn if a structure is given the packed attribute, but the packed
2758 attribute has no effect on the layout or size of the structure.
2759 Such structures may be mis-aligned for little benefit. For
2760 instance, in this code, the variable @code{f.x} in @code{struct bar}
2761 will be misaligned even though @code{struct bar} does not itself
2762 have the packed attribute:
2769 @} __attribute__((packed));
2779 Warn if padding is included in a structure, either to align an element
2780 of the structure or to align the whole structure. Sometimes when this
2781 happens it is possible to rearrange the fields of the structure to
2782 reduce the padding and so make the structure smaller.
2784 @item -Wredundant-decls
2785 @opindex Wredundant-decls
2786 Warn if anything is declared more than once in the same scope, even in
2787 cases where multiple declaration is valid and changes nothing.
2789 @item -Wnested-externs @r{(C only)}
2790 @opindex Wnested-externs
2791 Warn if an @code{extern} declaration is encountered within a function.
2793 @item -Wunreachable-code
2794 @opindex Wunreachable-code
2795 Warn if the compiler detects that code will never be executed.
2797 This option is intended to warn when the compiler detects that at
2798 least a whole line of source code will never be executed, because
2799 some condition is never satisfied or because it is after a
2800 procedure that never returns.
2802 It is possible for this option to produce a warning even though there
2803 are circumstances under which part of the affected line can be executed,
2804 so care should be taken when removing apparently-unreachable code.
2806 For instance, when a function is inlined, a warning may mean that the
2807 line is unreachable in only one inlined copy of the function.
2809 This option is not made part of @option{-Wall} because in a debugging
2810 version of a program there is often substantial code which checks
2811 correct functioning of the program and is, hopefully, unreachable
2812 because the program does work. Another common use of unreachable
2813 code is to provide behavior which is selectable at compile-time.
2817 Warn if a function can not be inlined and it was declared as inline.
2820 @opindex Winvalid-pch
2821 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
2822 the search path but can't be used.
2826 @opindex Wno-long-long
2827 Warn if @samp{long long} type is used. This is default. To inhibit
2828 the warning messages, use @option{-Wno-long-long}. Flags
2829 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2830 only when @option{-pedantic} flag is used.
2832 @item -Wdisabled-optimization
2833 @opindex Wdisabled-optimization
2834 Warn if a requested optimization pass is disabled. This warning does
2835 not generally indicate that there is anything wrong with your code; it
2836 merely indicates that GCC's optimizers were unable to handle the code
2837 effectively. Often, the problem is that your code is too big or too
2838 complex; GCC will refuse to optimize programs when the optimization
2839 itself is likely to take inordinate amounts of time.
2843 Make all warnings into errors.
2846 @node Debugging Options
2847 @section Options for Debugging Your Program or GCC
2848 @cindex options, debugging
2849 @cindex debugging information options
2851 GCC has various special options that are used for debugging
2852 either your program or GCC:
2857 Produce debugging information in the operating system's native format
2858 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2861 On most systems that use stabs format, @option{-g} enables use of extra
2862 debugging information that only GDB can use; this extra information
2863 makes debugging work better in GDB but will probably make other debuggers
2865 refuse to read the program. If you want to control for certain whether
2866 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
2867 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, @option{-gdwarf-1},
2868 or @option{-gvms} (see below).
2870 Unlike most other C compilers, GCC allows you to use @option{-g} with
2871 @option{-O}. The shortcuts taken by optimized code may occasionally
2872 produce surprising results: some variables you declared may not exist
2873 at all; flow of control may briefly move where you did not expect it;
2874 some statements may not be executed because they compute constant
2875 results or their values were already at hand; some statements may
2876 execute in different places because they were moved out of loops.
2878 Nevertheless it proves possible to debug optimized output. This makes
2879 it reasonable to use the optimizer for programs that might have bugs.
2881 The following options are useful when GCC is generated with the
2882 capability for more than one debugging format.
2886 Produce debugging information for use by GDB@. This means to use the
2887 most expressive format available (DWARF 2, stabs, or the native format
2888 if neither of those are supported), including GDB extensions if at all
2893 Produce debugging information in stabs format (if that is supported),
2894 without GDB extensions. This is the format used by DBX on most BSD
2895 systems. On MIPS, Alpha and System V Release 4 systems this option
2896 produces stabs debugging output which is not understood by DBX or SDB@.
2897 On System V Release 4 systems this option requires the GNU assembler.
2901 Produce debugging information in stabs format (if that is supported),
2902 using GNU extensions understood only by the GNU debugger (GDB)@. The
2903 use of these extensions is likely to make other debuggers crash or
2904 refuse to read the program.
2908 Produce debugging information in COFF format (if that is supported).
2909 This is the format used by SDB on most System V systems prior to
2914 Produce debugging information in XCOFF format (if that is supported).
2915 This is the format used by the DBX debugger on IBM RS/6000 systems.
2919 Produce debugging information in XCOFF format (if that is supported),
2920 using GNU extensions understood only by the GNU debugger (GDB)@. The
2921 use of these extensions is likely to make other debuggers crash or
2922 refuse to read the program, and may cause assemblers other than the GNU
2923 assembler (GAS) to fail with an error.
2927 Produce debugging information in DWARF version 1 format (if that is
2928 supported). This is the format used by SDB on most System V Release 4
2931 This option is deprecated.
2935 Produce debugging information in DWARF version 1 format (if that is
2936 supported), using GNU extensions understood only by the GNU debugger
2937 (GDB)@. The use of these extensions is likely to make other debuggers
2938 crash or refuse to read the program.
2940 This option is deprecated.
2944 Produce debugging information in DWARF version 2 format (if that is
2945 supported). This is the format used by DBX on IRIX 6.
2949 Produce debugging information in VMS debug format (if that is
2950 supported). This is the format used by DEBUG on VMS systems.
2953 @itemx -ggdb@var{level}
2954 @itemx -gstabs@var{level}
2955 @itemx -gcoff@var{level}
2956 @itemx -gxcoff@var{level}
2957 @itemx -gvms@var{level}
2958 Request debugging information and also use @var{level} to specify how
2959 much information. The default level is 2.
2961 Level 1 produces minimal information, enough for making backtraces in
2962 parts of the program that you don't plan to debug. This includes
2963 descriptions of functions and external variables, but no information
2964 about local variables and no line numbers.
2966 Level 3 includes extra information, such as all the macro definitions
2967 present in the program. Some debuggers support macro expansion when
2968 you use @option{-g3}.
2970 Note that in order to avoid confusion between DWARF1 debug level 2,
2971 and DWARF2, neither @option{-gdwarf} nor @option{-gdwarf-2} accept
2972 a concatenated debug level. Instead use an additional @option{-g@var{level}}
2973 option to change the debug level for DWARF1 or DWARF2.
2975 @item -feliminate-dwarf2-dups
2976 @opindex feliminate-dwarf2-dups
2977 Compress DWARF2 debugging information by eliminating duplicated
2978 information about each symbol. This option only makes sense when
2979 generating DWARF2 debugging information with @option{-gdwarf-2}.
2981 @cindex @command{prof}
2984 Generate extra code to write profile information suitable for the
2985 analysis program @command{prof}. You must use this option when compiling
2986 the source files you want data about, and you must also use it when
2989 @cindex @command{gprof}
2992 Generate extra code to write profile information suitable for the
2993 analysis program @command{gprof}. You must use this option when compiling
2994 the source files you want data about, and you must also use it when
2999 Makes the compiler print out each function name as it is compiled, and
3000 print some statistics about each pass when it finishes.
3003 @opindex ftime-report
3004 Makes the compiler print some statistics about the time consumed by each
3005 pass when it finishes.
3008 @opindex fmem-report
3009 Makes the compiler print some statistics about permanent memory
3010 allocation when it finishes.
3012 @item -fprofile-arcs
3013 @opindex fprofile-arcs
3014 Add code so that program flow @dfn{arcs} are instrumented. During
3015 execution the program records how many times each branch and call is
3016 executed and how many times it is taken or returns. When the compiled
3017 program exits it saves this data to a file called
3018 @file{@var{auxname}.da} for each source file. The data may be used for
3019 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3020 test coverage analysis (@option{-ftest-coverage}). Each object file's
3021 @var{auxname} is generated from the name of the output file, if
3022 explicitly specified and it is not the final executable, otherwise it is
3023 the basename of the source file. In both cases any suffix is removed
3024 (e.g. @file{foo.da} for input file @file{dir/foo.c}, or
3025 @file{dir/foo.da} for output file specified as @option{-o dir/foo.o}).
3030 Compile the source files with @option{-fprofile-arcs} plus optimization
3031 and code generation options. For test coverage analysis, use the
3032 additional @option{-ftest-coverage} option. You do not need to profile
3033 every source file in a program.
3036 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3037 (the latter implies the former).
3040 Run the program on a representative workload to generate the arc profile
3041 information. This may be repeated any number of times. You can run
3042 concurrent instances of your program, and provided that the file system
3043 supports locking, the data files will be correctly updated. Also
3044 @code{fork} calls are detected and correctly handled (double counting
3048 For profile-directed optimizations, compile the source files again with
3049 the same optimization and code generation options plus
3050 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3051 Control Optimization}).
3054 For test coverage analysis, use @command{gcov} to produce human readable
3055 information from the @file{.bbg} and @file{.da} files. Refer to the
3056 @command{gcov} documentation for further information.
3060 With @option{-fprofile-arcs}, for each function of your program GCC
3061 creates a program flow graph, then finds a spanning tree for the graph.
3062 Only arcs that are not on the spanning tree have to be instrumented: the
3063 compiler adds code to count the number of times that these arcs are
3064 executed. When an arc is the only exit or only entrance to a block, the
3065 instrumentation code can be added to the block; otherwise, a new basic
3066 block must be created to hold the instrumentation code.
3069 @item -ftest-coverage
3070 @opindex ftest-coverage
3071 Produce a graph file that the @command{gcov} code-coverage utility
3072 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3073 show program coverage. Each source file's data file is called
3074 @file{@var{auxname}.bbg}. Refer to the @option{-fprofile-arcs} option
3075 above for a description of @var{auxname} and instructions on how to
3076 generate test coverage data. Coverage data will match the source files
3077 more closely, if you do not optimize.
3079 @item -d@var{letters}
3081 Says to make debugging dumps during compilation at times specified by
3082 @var{letters}. This is used for debugging the compiler. The file names
3083 for most of the dumps are made by appending a pass number and a word to
3084 the @var{dumpname}. @var{dumpname} is generated from the name of the
3085 output file, if explicitly specified and it is not an executable,
3086 otherwise it is the basename of the source file. In both cases any
3087 suffix is removed (e.g. @file{foo.00.rtl} or @file{foo.01.sibling}).
3088 Here are the possible letters for use in @var{letters}, and their
3094 Annotate the assembler output with miscellaneous debugging information.
3097 Dump after computing branch probabilities, to @file{@var{file}.14.bp}.
3100 Dump after block reordering, to @file{@var{file}.32.bbro}.
3103 Dump after instruction combination, to the file @file{@var{file}.19.combine}.
3106 Dump after the first if conversion, to the file @file{@var{file}.15.ce1}.
3109 Dump after delayed branch scheduling, to @file{@var{file}.34.dbr}.
3112 Dump all macro definitions, at the end of preprocessing, in addition to
3116 Dump after SSA optimizations, to @file{@var{file}.04.ssa} and
3117 @file{@var{file}.07.ussa}.
3120 Dump after the second if conversion, to @file{@var{file}.29.ce3}.
3123 Dump after life analysis, to @file{@var{file}.18.life}.
3126 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.10.addressof}.
3129 Dump after global register allocation, to @file{@var{file}.24.greg}.
3132 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3135 Dump after reg-to-stack conversion, to @file{@var{file}.31.stack}.
3138 Dump after post-reload optimizations, to @file{@var{file}.25.postreload}.
3141 Dump after GCSE, to @file{@var{file}.11.gcse}.
3144 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
3147 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
3150 Dump after conversion from registers to stack, to @file{@var{file}.31.stack}.
3153 Dump after local register allocation, to @file{@var{file}.23.lreg}.
3156 Dump after loop optimization passes, to @file{@var{file}.12.loop} and
3157 @file{@var{file}.18.loop2}.
3160 Dump after performing the machine dependent reorganization pass, to
3161 @file{@var{file}.33.mach}.
3164 Dump after register renumbering, to @file{@var{file}.28.rnreg}.
3167 Dump after the register move pass, to @file{@var{file}.21.regmove}.
3170 Dump after RTL generation, to @file{@var{file}.00.rtl}.
3173 Dump after the second scheduling pass, to @file{@var{file}.30.sched2}.
3176 Dump after CSE (including the jump optimization that sometimes follows
3177 CSE), to @file{@var{file}.09.cse}.
3180 Dump after the first scheduling pass, to @file{@var{file}.22.sched}.
3183 Dump after the second CSE pass (including the jump optimization that
3184 sometimes follows CSE), to @file{@var{file}.17.cse2}.
3187 Dump after null pointer elimination pass to @file{@var{file}.08.null}.
3190 Dump after the second flow pass, to @file{@var{file}.26.flow2}.
3193 Dump after SSA dead code elimination, to @file{@var{file}.06.ssadce}.
3196 Dump after the peephole pass, to @file{@var{file}.27.peephole2}.
3199 Produce all the dumps listed above.
3202 Produce a core dump whenever an error occurs.
3205 Print statistics on memory usage, at the end of the run, to
3209 Annotate the assembler output with a comment indicating which
3210 pattern and alternative was used. The length of each instruction is
3214 Dump the RTL in the assembler output as a comment before each instruction.
3215 Also turns on @option{-dp} annotation.
3218 For each of the other indicated dump files (except for
3219 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
3220 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3223 Just generate RTL for a function instead of compiling it. Usually used
3227 Dump debugging information during parsing, to standard error.
3230 @item -fdump-unnumbered
3231 @opindex fdump-unnumbered
3232 When doing debugging dumps (see @option{-d} option above), suppress instruction
3233 numbers and line number note output. This makes it more feasible to
3234 use diff on debugging dumps for compiler invocations with different
3235 options, in particular with and without @option{-g}.
3237 @item -fdump-translation-unit @r{(C and C++ only)}
3238 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3239 @opindex fdump-translation-unit
3240 Dump a representation of the tree structure for the entire translation
3241 unit to a file. The file name is made by appending @file{.tu} to the
3242 source file name. If the @samp{-@var{options}} form is used, @var{options}
3243 controls the details of the dump as described for the
3244 @option{-fdump-tree} options.
3246 @item -fdump-class-hierarchy @r{(C++ only)}
3247 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3248 @opindex fdump-class-hierarchy
3249 Dump a representation of each class's hierarchy and virtual function
3250 table layout to a file. The file name is made by appending @file{.class}
3251 to the source file name. If the @samp{-@var{options}} form is used,
3252 @var{options} controls the details of the dump as described for the
3253 @option{-fdump-tree} options.
3255 @item -fdump-tree-@var{switch} @r{(C++ only)}
3256 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3258 Control the dumping at various stages of processing the intermediate
3259 language tree to a file. The file name is generated by appending a switch
3260 specific suffix to the source file name. If the @samp{-@var{options}}
3261 form is used, @var{options} is a list of @samp{-} separated options that
3262 control the details of the dump. Not all options are applicable to all
3263 dumps, those which are not meaningful will be ignored. The following
3264 options are available
3268 Print the address of each node. Usually this is not meaningful as it
3269 changes according to the environment and source file. Its primary use
3270 is for tying up a dump file with a debug environment.
3272 Inhibit dumping of members of a scope or body of a function merely
3273 because that scope has been reached. Only dump such items when they
3274 are directly reachable by some other path.
3276 Turn on all options.
3279 The following tree dumps are possible:
3282 Dump before any tree based optimization, to @file{@var{file}.original}.
3284 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3286 Dump after function inlining, to @file{@var{file}.inlined}.
3289 @item -frandom-seed=@var{string}
3290 @opindex frandom-string
3291 This option provides a seed that GCC uses when it would otherwise use
3292 random numbers. At present, this is used to generate certain symbol names
3293 that have to be different in every compiled file.
3295 The @var{string} should be different for every file you compile.
3297 @item -fsched-verbose=@var{n}
3298 @opindex fsched-verbose
3299 On targets that use instruction scheduling, this option controls the
3300 amount of debugging output the scheduler prints. This information is
3301 written to standard error, unless @option{-dS} or @option{-dR} is
3302 specified, in which case it is output to the usual dump
3303 listing file, @file{.sched} or @file{.sched2} respectively. However
3304 for @var{n} greater than nine, the output is always printed to standard
3307 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3308 same information as @option{-dRS}. For @var{n} greater than one, it
3309 also output basic block probabilities, detailed ready list information
3310 and unit/insn info. For @var{n} greater than two, it includes RTL
3311 at abort point, control-flow and regions info. And for @var{n} over
3312 four, @option{-fsched-verbose} also includes dependence info.
3316 Store the usual ``temporary'' intermediate files permanently; place them
3317 in the current directory and name them based on the source file. Thus,
3318 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3319 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3320 preprocessed @file{foo.i} output file even though the compiler now
3321 normally uses an integrated preprocessor.
3325 Report the CPU time taken by each subprocess in the compilation
3326 sequence. For C source files, this is the compiler proper and assembler
3327 (plus the linker if linking is done). The output looks like this:
3334 The first number on each line is the ``user time,'' that is time spent
3335 executing the program itself. The second number is ``system time,''
3336 time spent executing operating system routines on behalf of the program.
3337 Both numbers are in seconds.
3339 @item -print-file-name=@var{library}
3340 @opindex print-file-name
3341 Print the full absolute name of the library file @var{library} that
3342 would be used when linking---and don't do anything else. With this
3343 option, GCC does not compile or link anything; it just prints the
3346 @item -print-multi-directory
3347 @opindex print-multi-directory
3348 Print the directory name corresponding to the multilib selected by any
3349 other switches present in the command line. This directory is supposed
3350 to exist in @env{GCC_EXEC_PREFIX}.
3352 @item -print-multi-lib
3353 @opindex print-multi-lib
3354 Print the mapping from multilib directory names to compiler switches
3355 that enable them. The directory name is separated from the switches by
3356 @samp{;}, and each switch starts with an @samp{@@} instead of the
3357 @samp{-}, without spaces between multiple switches. This is supposed to
3358 ease shell-processing.
3360 @item -print-prog-name=@var{program}
3361 @opindex print-prog-name
3362 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3364 @item -print-libgcc-file-name
3365 @opindex print-libgcc-file-name
3366 Same as @option{-print-file-name=libgcc.a}.
3368 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3369 but you do want to link with @file{libgcc.a}. You can do
3372 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3375 @item -print-search-dirs
3376 @opindex print-search-dirs
3377 Print the name of the configured installation directory and a list of
3378 program and library directories gcc will search---and don't do anything else.
3380 This is useful when gcc prints the error message
3381 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3382 To resolve this you either need to put @file{cpp0} and the other compiler
3383 components where gcc expects to find them, or you can set the environment
3384 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3385 Don't forget the trailing '/'.
3386 @xref{Environment Variables}.
3389 @opindex dumpmachine
3390 Print the compiler's target machine (for example,
3391 @samp{i686-pc-linux-gnu})---and don't do anything else.
3394 @opindex dumpversion
3395 Print the compiler version (for example, @samp{3.0})---and don't do
3400 Print the compiler's built-in specs---and don't do anything else. (This
3401 is used when GCC itself is being built.) @xref{Spec Files}.
3403 @item -feliminate-unused-debug-types
3404 @opindex feliminate-unused-debug-types
3405 Normally, when producing DWARF2 output, GCC will emit debugging
3406 information for all types declared in a compilation
3407 unit, regardless of whether or not they are actually used
3408 in that compilation unit. Sometimes this is useful, such as
3409 if, in the debugger, you want to cast a value to a type that is
3410 not actually used in your program (but is declared). More often,
3411 however, this results in a significant amount of wasted space.
3412 With this option, GCC will avoid producing debug symbol output
3413 for types that are nowhere used in the source file being compiled.
3416 @node Optimize Options
3417 @section Options That Control Optimization
3418 @cindex optimize options
3419 @cindex options, optimization
3421 These options control various sorts of optimizations.
3423 Without any optimization option, the compiler's goal is to reduce the
3424 cost of compilation and to make debugging produce the expected
3425 results. Statements are independent: if you stop the program with a
3426 breakpoint between statements, you can then assign a new value to any
3427 variable or change the program counter to any other statement in the
3428 function and get exactly the results you would expect from the source
3431 Turning on optimization flags makes the compiler attempt to improve
3432 the performance and/or code size at the expense of compilation time
3433 and possibly the ability to debug the program.
3435 Not all optimizations are controlled directly by a flag. Only
3436 optimizations that have a flag are listed.
3443 Optimize. Optimizing compilation takes somewhat more time, and a lot
3444 more memory for a large function.
3446 With @option{-O}, the compiler tries to reduce code size and execution
3447 time, without performing any optimizations that take a great deal of
3450 @option{-O} turns on the following optimization flags:
3451 @gccoptlist{-fdefer-pop @gol
3452 -fmerge-constants @gol
3454 -floop-optimize @gol
3456 -fif-conversion @gol
3457 -fif-conversion2 @gol
3458 -fdelayed-branch @gol
3459 -fguess-branch-probability @gol
3462 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3463 where doing so does not interfere with debugging.
3467 Optimize even more. GCC performs nearly all supported optimizations
3468 that do not involve a space-speed tradeoff. The compiler does not
3469 perform loop unrolling or function inlining when you specify @option{-O2}.
3470 As compared to @option{-O}, this option increases both compilation time
3471 and the performance of the generated code.
3473 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3474 also turns on the following optimization flags:
3475 @gccoptlist{-fforce-mem @gol
3476 -foptimize-sibling-calls @gol
3477 -fstrength-reduce @gol
3478 -fcse-follow-jumps -fcse-skip-blocks @gol
3479 -frerun-cse-after-loop -frerun-loop-opt @gol
3480 -fgcse -fgcse-lm -fgcse-sm @gol
3481 -fdelete-null-pointer-checks @gol
3482 -fexpensive-optimizations @gol
3484 -fschedule-insns -fschedule-insns2 @gol
3485 -fsched-interblock -fsched-spec @gol
3488 -freorder-blocks -freorder-functions @gol
3489 -fstrict-aliasing @gol
3490 -falign-functions -falign-jumps @gol
3491 -falign-loops -falign-labels}
3493 Please note the warning under @option{-fgcse} about
3494 invoking @option{-O2} on programs that use computed gotos.
3498 Optimize yet more. @option{-O3} turns on all optimizations specified by
3499 @option{-O2} and also turns on the @option{-finline-functions},
3500 @option{-funit-at-a-time} and @option{-frename-registers} options.
3504 Do not optimize. This is the default.
3508 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3509 do not typically increase code size. It also performs further
3510 optimizations designed to reduce code size.
3512 @option{-Os} disables the following optimization flags:
3513 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3514 -falign-labels -freorder-blocks -fprefetch-loop-arrays}
3516 If you use multiple @option{-O} options, with or without level numbers,
3517 the last such option is the one that is effective.
3520 Options of the form @option{-f@var{flag}} specify machine-independent
3521 flags. Most flags have both positive and negative forms; the negative
3522 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3523 below, only one of the forms is listed---the one you typically will
3524 use. You can figure out the other form by either removing @samp{no-}
3527 The following options control specific optimizations. They are either
3528 activated by @option{-O} options or are related to ones that are. You
3529 can use the following flags in the rare cases when ``fine-tuning'' of
3530 optimizations to be performed is desired.
3533 @item -fno-default-inline
3534 @opindex fno-default-inline
3535 Do not make member functions inline by default merely because they are
3536 defined inside the class scope (C++ only). Otherwise, when you specify
3537 @w{@option{-O}}, member functions defined inside class scope are compiled
3538 inline by default; i.e., you don't need to add @samp{inline} in front of
3539 the member function name.
3541 @item -fno-defer-pop
3542 @opindex fno-defer-pop
3543 Always pop the arguments to each function call as soon as that function
3544 returns. For machines which must pop arguments after a function call,
3545 the compiler normally lets arguments accumulate on the stack for several
3546 function calls and pops them all at once.
3548 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3552 Force memory operands to be copied into registers before doing
3553 arithmetic on them. This produces better code by making all memory
3554 references potential common subexpressions. When they are not common
3555 subexpressions, instruction combination should eliminate the separate
3558 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3561 @opindex fforce-addr
3562 Force memory address constants to be copied into registers before
3563 doing arithmetic on them. This may produce better code just as
3564 @option{-fforce-mem} may.
3566 @item -fomit-frame-pointer
3567 @opindex fomit-frame-pointer
3568 Don't keep the frame pointer in a register for functions that
3569 don't need one. This avoids the instructions to save, set up and
3570 restore frame pointers; it also makes an extra register available
3571 in many functions. @strong{It also makes debugging impossible on
3574 On some machines, such as the VAX, this flag has no effect, because
3575 the standard calling sequence automatically handles the frame pointer
3576 and nothing is saved by pretending it doesn't exist. The
3577 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3578 whether a target machine supports this flag. @xref{Registers,,Register
3579 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3581 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3583 @item -foptimize-sibling-calls
3584 @opindex foptimize-sibling-calls
3585 Optimize sibling and tail recursive calls.
3587 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3591 Don't pay attention to the @code{inline} keyword. Normally this option
3592 is used to keep the compiler from expanding any functions inline.
3593 Note that if you are not optimizing, no functions can be expanded inline.
3595 @item -finline-functions
3596 @opindex finline-functions
3597 Integrate all simple functions into their callers. The compiler
3598 heuristically decides which functions are simple enough to be worth
3599 integrating in this way.
3601 If all calls to a given function are integrated, and the function is
3602 declared @code{static}, then the function is normally not output as
3603 assembler code in its own right.
3605 Enabled at level @option{-O3}.
3607 @item -finline-limit=@var{n}
3608 @opindex finline-limit
3609 By default, gcc limits the size of functions that can be inlined. This flag
3610 allows the control of this limit for functions that are explicitly marked as
3611 inline (i.e., marked with the inline keyword or defined within the class
3612 definition in c++). @var{n} is the size of functions that can be inlined in
3613 number of pseudo instructions (not counting parameter handling). The default
3614 value of @var{n} is 600.
3615 Increasing this value can result in more inlined code at
3616 the cost of compilation time and memory consumption. Decreasing usually makes
3617 the compilation faster and less code will be inlined (which presumably
3618 means slower programs). This option is particularly useful for programs that
3619 use inlining heavily such as those based on recursive templates with C++.
3621 Inlining is actually controlled by a number of parameters, which may be
3622 specified individually by using @option{--param @var{name}=@var{value}}.
3623 The @option{-finline-limit=@var{n}} option sets some of these parameters
3627 @item max-inline-insns
3629 @item max-inline-insns-single
3630 is set to @var{n}/2.
3631 @item max-inline-insns-auto
3632 is set to @var{n}/2.
3633 @item min-inline-insns
3634 is set to 130 or @var{n}/4, whichever is smaller.
3635 @item max-inline-insns-rtl
3639 Using @option{-finline-limit=600} thus results in the default settings
3640 for these parameters. See below for a documentation of the individual
3641 parameters controlling inlining.
3643 @emph{Note:} pseudo instruction represents, in this particular context, an
3644 abstract measurement of function's size. In no way, it represents a count
3645 of assembly instructions and as such its exact meaning might change from one
3646 release to an another.
3648 @item -fkeep-inline-functions
3649 @opindex fkeep-inline-functions
3650 Even if all calls to a given function are integrated, and the function
3651 is declared @code{static}, nevertheless output a separate run-time
3652 callable version of the function. This switch does not affect
3653 @code{extern inline} functions.
3655 @item -fkeep-static-consts
3656 @opindex fkeep-static-consts
3657 Emit variables declared @code{static const} when optimization isn't turned
3658 on, even if the variables aren't referenced.
3660 GCC enables this option by default. If you want to force the compiler to
3661 check if the variable was referenced, regardless of whether or not
3662 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3664 @item -fmerge-constants
3665 Attempt to merge identical constants (string constants and floating point
3666 constants) across compilation units.
3668 This option is the default for optimized compilation if the assembler and
3669 linker support it. Use @option{-fno-merge-constants} to inhibit this
3672 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3674 @item -fmerge-all-constants
3675 Attempt to merge identical constants and identical variables.
3677 This option implies @option{-fmerge-constants}. In addition to
3678 @option{-fmerge-constants} this considers e.g. even constant initialized
3679 arrays or initialized constant variables with integral or floating point
3680 types. Languages like C or C++ require each non-automatic variable to
3681 have distinct location, so using this option will result in non-conforming
3686 Use a graph coloring register allocator. Currently this option is meant
3687 for testing, so we are interested to hear about miscompilations with
3690 @item -fno-branch-count-reg
3691 @opindex fno-branch-count-reg
3692 Do not use ``decrement and branch'' instructions on a count register,
3693 but instead generate a sequence of instructions that decrement a
3694 register, compare it against zero, then branch based upon the result.
3695 This option is only meaningful on architectures that support such
3696 instructions, which include x86, PowerPC, IA-64 and S/390.
3698 The default is @option{-fbranch-count-reg}, enabled when
3699 @option{-fstrength-reduce} is enabled.
3701 @item -fno-function-cse
3702 @opindex fno-function-cse
3703 Do not put function addresses in registers; make each instruction that
3704 calls a constant function contain the function's address explicitly.
3706 This option results in less efficient code, but some strange hacks
3707 that alter the assembler output may be confused by the optimizations
3708 performed when this option is not used.
3710 The default is @option{-ffunction-cse}
3712 @item -fno-zero-initialized-in-bss
3713 @opindex fno-zero-initialized-in-bss
3714 If the target supports a BSS section, GCC by default puts variables that
3715 are initialized to zero into BSS@. This can save space in the resulting
3718 This option turns off this behavior because some programs explicitly
3719 rely on variables going to the data section. E.g., so that the
3720 resulting executable can find the beginning of that section and/or make
3721 assumptions based on that.
3723 The default is @option{-fzero-initialized-in-bss}.
3725 @item -fstrength-reduce
3726 @opindex fstrength-reduce
3727 Perform the optimizations of loop strength reduction and
3728 elimination of iteration variables.
3730 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3732 @item -fthread-jumps
3733 @opindex fthread-jumps
3734 Perform optimizations where we check to see if a jump branches to a
3735 location where another comparison subsumed by the first is found. If
3736 so, the first branch is redirected to either the destination of the
3737 second branch or a point immediately following it, depending on whether
3738 the condition is known to be true or false.
3740 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3742 @item -fcse-follow-jumps
3743 @opindex fcse-follow-jumps
3744 In common subexpression elimination, scan through jump instructions
3745 when the target of the jump is not reached by any other path. For
3746 example, when CSE encounters an @code{if} statement with an
3747 @code{else} clause, CSE will follow the jump when the condition
3750 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3752 @item -fcse-skip-blocks
3753 @opindex fcse-skip-blocks
3754 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3755 follow jumps which conditionally skip over blocks. When CSE
3756 encounters a simple @code{if} statement with no else clause,
3757 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3758 body of the @code{if}.
3760 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3762 @item -frerun-cse-after-loop
3763 @opindex frerun-cse-after-loop
3764 Re-run common subexpression elimination after loop optimizations has been
3767 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3769 @item -frerun-loop-opt
3770 @opindex frerun-loop-opt
3771 Run the loop optimizer twice.
3773 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3777 Perform a global common subexpression elimination pass.
3778 This pass also performs global constant and copy propagation.
3780 @emph{Note:} When compiling a program using computed gotos, a GCC
3781 extension, you may get better runtime performance if you disable
3782 the global common subexpression elimination pass by adding
3783 @option{-fno-gcse} to the command line.
3785 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3789 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3790 attempt to move loads which are only killed by stores into themselves. This
3791 allows a loop containing a load/store sequence to be changed to a load outside
3792 the loop, and a copy/store within the loop.
3794 Enabled by default when gcse is enabled.
3798 When @option{-fgcse-sm} is enabled, A store motion pass is run after global common
3799 subexpression elimination. This pass will attempt to move stores out of loops.
3800 When used in conjunction with @option{-fgcse-lm}, loops containing a load/store sequence
3801 can be changed to a load before the loop and a store after the loop.
3803 Enabled by default when gcse is enabled.
3805 @item -floop-optimize
3806 @opindex floop-optimize
3807 Perform loop optimizations: move constant expressions out of loops, simplify
3808 exit test conditions and optionally do strength-reduction and loop unrolling as
3811 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3813 @item -fcrossjumping
3814 @opindex crossjumping
3815 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
3816 resulting code may or may not perform better than without cross-jumping.
3818 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3820 @item -fif-conversion
3821 @opindex if-conversion
3822 Attempt to transform conditional jumps into branch-less equivalents. This
3823 include use of conditional moves, min, max, set flags and abs instructions, and
3824 some tricks doable by standard arithmetics. The use of conditional execution
3825 on chips where it is available is controlled by @code{if-conversion2}.
3827 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3829 @item -fif-conversion2
3830 @opindex if-conversion2
3831 Use conditional execution (where available) to transform conditional jumps into
3832 branch-less equivalents.
3834 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3836 @item -fdelete-null-pointer-checks
3837 @opindex fdelete-null-pointer-checks
3838 Use global dataflow analysis to identify and eliminate useless checks
3839 for null pointers. The compiler assumes that dereferencing a null
3840 pointer would have halted the program. If a pointer is checked after
3841 it has already been dereferenced, it cannot be null.
3843 In some environments, this assumption is not true, and programs can
3844 safely dereference null pointers. Use
3845 @option{-fno-delete-null-pointer-checks} to disable this optimization
3846 for programs which depend on that behavior.
3848 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3850 @item -fexpensive-optimizations
3851 @opindex fexpensive-optimizations
3852 Perform a number of minor optimizations that are relatively expensive.
3854 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3856 @item -foptimize-register-move
3858 @opindex foptimize-register-move
3860 Attempt to reassign register numbers in move instructions and as
3861 operands of other simple instructions in order to maximize the amount of
3862 register tying. This is especially helpful on machines with two-operand
3865 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3868 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3870 @item -fdelayed-branch
3871 @opindex fdelayed-branch
3872 If supported for the target machine, attempt to reorder instructions
3873 to exploit instruction slots available after delayed branch
3876 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3878 @item -fschedule-insns
3879 @opindex fschedule-insns
3880 If supported for the target machine, attempt to reorder instructions to
3881 eliminate execution stalls due to required data being unavailable. This
3882 helps machines that have slow floating point or memory load instructions
3883 by allowing other instructions to be issued until the result of the load
3884 or floating point instruction is required.
3886 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3888 @item -fschedule-insns2
3889 @opindex fschedule-insns2
3890 Similar to @option{-fschedule-insns}, but requests an additional pass of
3891 instruction scheduling after register allocation has been done. This is
3892 especially useful on machines with a relatively small number of
3893 registers and where memory load instructions take more than one cycle.
3895 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3897 @item -fno-sched-interblock
3898 @opindex fno-sched-interblock
3899 Don't schedule instructions across basic blocks. This is normally
3900 enabled by default when scheduling before register allocation, i.e.@:
3901 with @option{-fschedule-insns} or at @option{-O2} or higher.
3903 @item -fno-sched-spec
3904 @opindex fno-sched-spec
3905 Don't allow speculative motion of non-load instructions. This is normally
3906 enabled by default when scheduling before register allocation, i.e.@:
3907 with @option{-fschedule-insns} or at @option{-O2} or higher.
3909 @item -fsched-spec-load
3910 @opindex fsched-spec-load
3911 Allow speculative motion of some load instructions. This only makes
3912 sense when scheduling before register allocation, i.e.@: with
3913 @option{-fschedule-insns} or at @option{-O2} or higher.
3915 @item -fsched-spec-load-dangerous
3916 @opindex fsched-spec-load-dangerous
3917 Allow speculative motion of more load instructions. This only makes
3918 sense when scheduling before register allocation, i.e.@: with
3919 @option{-fschedule-insns} or at @option{-O2} or higher.
3921 @item -fsched2-use-superblocks
3922 @opindex fsched2-use-superblocks
3923 When schedulilng after register allocation, do use superblock scheduling
3924 algorithm. Superblock scheduling allows motion across basic block boundaries
3925 resulting on faster schedules. This option is experimental, as not all machine
3926 descriptions used by GCC model the CPU closely enough to avoid unreliable
3927 results from the algorithm.
3929 This only makes sense when scheduling after register allocation, i.e.@: with
3930 @option{-fschedule-insns2} or at @option{-O2} or higher.
3932 @item -fsched2-use-traces
3933 @opindex fsched2-use-traces
3934 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
3935 allocation and additionally perform code duplication in order to increase the
3936 size of superblocks using tracer pass. See @option{-ftracer} for details on
3939 This mode should produce faster but significantly longer programs. Also
3940 without @code{-fbranch-probabilities} the traces constructed may not match the
3941 reality and hurt the performance. This only makes
3942 sense when scheduling after register allocation, i.e.@: with
3943 @option{-fschedule-insns2} or at @option{-O2} or higher.
3945 @item -fcaller-saves
3946 @opindex fcaller-saves
3947 Enable values to be allocated in registers that will be clobbered by
3948 function calls, by emitting extra instructions to save and restore the
3949 registers around such calls. Such allocation is done only when it
3950 seems to result in better code than would otherwise be produced.
3952 This option is always enabled by default on certain machines, usually
3953 those which have no call-preserved registers to use instead.
3955 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3957 @item -fmove-all-movables
3958 @opindex fmove-all-movables
3959 Forces all invariant computations in loops to be moved
3962 @item -freduce-all-givs
3963 @opindex freduce-all-givs
3964 Forces all general-induction variables in loops to be
3967 @emph{Note:} When compiling programs written in Fortran,
3968 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
3969 by default when you use the optimizer.
3971 These options may generate better or worse code; results are highly
3972 dependent on the structure of loops within the source code.
3974 These two options are intended to be removed someday, once
3975 they have helped determine the efficacy of various
3976 approaches to improving loop optimizations.
3978 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
3979 know how use of these options affects
3980 the performance of your production code.
3981 We're very interested in code that runs @emph{slower}
3982 when these options are @emph{enabled}.
3985 @itemx -fno-peephole2
3986 @opindex fno-peephole
3987 @opindex fno-peephole2
3988 Disable any machine-specific peephole optimizations. The difference
3989 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
3990 are implemented in the compiler; some targets use one, some use the
3991 other, a few use both.
3993 @option{-fpeephole} is enabled by default.
3994 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3996 @item -fno-guess-branch-probability
3997 @opindex fno-guess-branch-probability
3998 Do not guess branch probabilities using a randomized model.
4000 Sometimes gcc will opt to use a randomized model to guess branch
4001 probabilities, when none are available from either profiling feedback
4002 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4003 different runs of the compiler on the same program may produce different
4006 In a hard real-time system, people don't want different runs of the
4007 compiler to produce code that has different behavior; minimizing
4008 non-determinism is of paramount import. This switch allows users to
4009 reduce non-determinism, possibly at the expense of inferior
4012 The default is @option{-fguess-branch-probability} at levels
4013 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4015 @item -freorder-blocks
4016 @opindex freorder-blocks
4017 Reorder basic blocks in the compiled function in order to reduce number of
4018 taken branches and improve code locality.
4020 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4022 @item -freorder-functions
4023 @opindex freorder-functions
4024 Reorder basic blocks in the compiled function in order to reduce number of
4025 taken branches and improve code locality. This is implemented by using special
4026 subsections @code{text.hot} for most frequently executed functions and
4027 @code{text.unlikely} for unlikely executed functions. Reordering is done by
4028 the linker so object file format must support named sections and linker must
4029 place them in a reasonable way.
4031 Also profile feedback must be available in to make this option effective. See
4032 @option{-fprofile-arcs} for details.
4034 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4036 @item -fstrict-aliasing
4037 @opindex fstrict-aliasing
4038 Allows the compiler to assume the strictest aliasing rules applicable to
4039 the language being compiled. For C (and C++), this activates
4040 optimizations based on the type of expressions. In particular, an
4041 object of one type is assumed never to reside at the same address as an
4042 object of a different type, unless the types are almost the same. For
4043 example, an @code{unsigned int} can alias an @code{int}, but not a
4044 @code{void*} or a @code{double}. A character type may alias any other
4047 Pay special attention to code like this:
4060 The practice of reading from a different union member than the one most
4061 recently written to (called ``type-punning'') is common. Even with
4062 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4063 is accessed through the union type. So, the code above will work as
4064 expected. However, this code might not:
4075 Every language that wishes to perform language-specific alias analysis
4076 should define a function that computes, given an @code{tree}
4077 node, an alias set for the node. Nodes in different alias sets are not
4078 allowed to alias. For an example, see the C front-end function
4079 @code{c_get_alias_set}.
4081 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4083 @item -falign-functions
4084 @itemx -falign-functions=@var{n}
4085 @opindex falign-functions
4086 Align the start of functions to the next power-of-two greater than
4087 @var{n}, skipping up to @var{n} bytes. For instance,
4088 @option{-falign-functions=32} aligns functions to the next 32-byte
4089 boundary, but @option{-falign-functions=24} would align to the next
4090 32-byte boundary only if this can be done by skipping 23 bytes or less.
4092 @option{-fno-align-functions} and @option{-falign-functions=1} are
4093 equivalent and mean that functions will not be aligned.
4095 Some assemblers only support this flag when @var{n} is a power of two;
4096 in that case, it is rounded up.
4098 If @var{n} is not specified, use a machine-dependent default.
4100 Enabled at levels @option{-O2}, @option{-O3}.
4102 @item -falign-labels
4103 @itemx -falign-labels=@var{n}
4104 @opindex falign-labels
4105 Align all branch targets to a power-of-two boundary, skipping up to
4106 @var{n} bytes like @option{-falign-functions}. This option can easily
4107 make code slower, because it must insert dummy operations for when the
4108 branch target is reached in the usual flow of the code.
4110 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4111 are greater than this value, then their values are used instead.
4113 If @var{n} is not specified, use a machine-dependent default which is
4114 very likely to be @samp{1}, meaning no alignment.
4116 Enabled at levels @option{-O2}, @option{-O3}.
4119 @itemx -falign-loops=@var{n}
4120 @opindex falign-loops
4121 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4122 like @option{-falign-functions}. The hope is that the loop will be
4123 executed many times, which will make up for any execution of the dummy
4126 If @var{n} is not specified, use a machine-dependent default.
4128 Enabled at levels @option{-O2}, @option{-O3}.
4131 @itemx -falign-jumps=@var{n}
4132 @opindex falign-jumps
4133 Align branch targets to a power-of-two boundary, for branch targets
4134 where the targets can only be reached by jumping, skipping up to @var{n}
4135 bytes like @option{-falign-functions}. In this case, no dummy operations
4138 If @var{n} is not specified, use a machine-dependent default.
4140 Enabled at levels @option{-O2}, @option{-O3}.
4142 @item -frename-registers
4143 @opindex frename-registers
4144 Attempt to avoid false dependencies in scheduled code by making use
4145 of registers left over after register allocation. This optimization
4146 will most benefit processors with lots of registers. It can, however,
4147 make debugging impossible, since variables will no longer stay in
4148 a ``home register''.
4150 Enabled at levels @option{-O3}.
4152 @item -fno-cprop-registers
4153 @opindex fno-cprop-registers
4154 After register allocation and post-register allocation instruction splitting,
4155 we perform a copy-propagation pass to try to reduce scheduling dependencies
4156 and occasionally eliminate the copy.
4158 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4162 The following options control compiler behavior regarding floating
4163 point arithmetic. These options trade off between speed and
4164 correctness. All must be specifically enabled.
4168 @opindex ffloat-store
4169 Do not store floating point variables in registers, and inhibit other
4170 options that might change whether a floating point value is taken from a
4173 @cindex floating point precision
4174 This option prevents undesirable excess precision on machines such as
4175 the 68000 where the floating registers (of the 68881) keep more
4176 precision than a @code{double} is supposed to have. Similarly for the
4177 x86 architecture. For most programs, the excess precision does only
4178 good, but a few programs rely on the precise definition of IEEE floating
4179 point. Use @option{-ffloat-store} for such programs, after modifying
4180 them to store all pertinent intermediate computations into variables.
4184 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4185 @option{-fno-trapping-math}, @option{-ffinite-math-only} and @*
4186 @option{-fno-signaling-nans}.
4188 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4190 This option should never be turned on by any @option{-O} option since
4191 it can result in incorrect output for programs which depend on
4192 an exact implementation of IEEE or ISO rules/specifications for
4195 @item -fno-math-errno
4196 @opindex fno-math-errno
4197 Do not set ERRNO after calling math functions that are executed
4198 with a single instruction, e.g., sqrt. A program that relies on
4199 IEEE exceptions for math error handling may want to use this flag
4200 for speed while maintaining IEEE arithmetic compatibility.
4202 This option should never be turned on by any @option{-O} option since
4203 it can result in incorrect output for programs which depend on
4204 an exact implementation of IEEE or ISO rules/specifications for
4207 The default is @option{-fmath-errno}.
4209 @item -funsafe-math-optimizations
4210 @opindex funsafe-math-optimizations
4211 Allow optimizations for floating-point arithmetic that (a) assume
4212 that arguments and results are valid and (b) may violate IEEE or
4213 ANSI standards. When used at link-time, it may include libraries
4214 or startup files that change the default FPU control word or other
4215 similar optimizations.
4217 This option should never be turned on by any @option{-O} option since
4218 it can result in incorrect output for programs which depend on
4219 an exact implementation of IEEE or ISO rules/specifications for
4222 The default is @option{-fno-unsafe-math-optimizations}.
4224 @item -ffinite-math-only
4225 @opindex ffinite-math-only
4226 Allow optimizations for floating-point arithmetic that assume
4227 that arguments and results are not NaNs or +-Infs.
4229 This option should never be turned on by any @option{-O} option since
4230 it can result in incorrect output for programs which depend on
4231 an exact implementation of IEEE or ISO rules/specifications.
4233 The default is @option{-fno-finite-math-only}.
4235 @item -fno-trapping-math
4236 @opindex fno-trapping-math
4237 Compile code assuming that floating-point operations cannot generate
4238 user-visible traps. These traps include division by zero, overflow,
4239 underflow, inexact result and invalid operation. This option implies
4240 @option{-fno-signaling-nans}. Setting this option may allow faster
4241 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4243 This option should never be turned on by any @option{-O} option since
4244 it can result in incorrect output for programs which depend on
4245 an exact implementation of IEEE or ISO rules/specifications for
4248 The default is @option{-ftrapping-math}.
4250 @item -fsignaling-nans
4251 @opindex fsignaling-nans
4252 Compile code assuming that IEEE signaling NaNs may generate user-visible
4253 traps during floating-point operations. Setting this option disables
4254 optimizations that may change the number of exceptions visible with
4255 signaling NaNs. This option implies @option{-ftrapping-math}.
4257 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4260 The default is @option{-fno-signaling-nans}.
4262 This option is experimental and does not currently guarantee to
4263 disable all GCC optimizations that affect signaling NaN behavior.
4265 @item -fsingle-precision-constant
4266 @opindex fsingle-precision-constant
4267 Treat floating point constant as single precision constant instead of
4268 implicitly converting it to double precision constant.
4273 The following options control optimizations that may improve
4274 performance, but are not enabled by any @option{-O} options. This
4275 section includes experimental options that may produce broken code.
4278 @item -fbranch-probabilities
4279 @opindex fbranch-probabilities
4280 After running a program compiled with @option{-fprofile-arcs}
4281 (@pxref{Debugging Options,, Options for Debugging Your Program or
4282 @command{gcc}}), you can compile it a second time using
4283 @option{-fbranch-probabilities}, to improve optimizations based on
4284 the number of times each branch was taken. When the program
4285 compiled with @option{-fprofile-arcs} exits it saves arc execution
4286 counts to a file called @file{@var{sourcename}.da} for each source
4287 file The information in this data file is very dependent on the
4288 structure of the generated code, so you must use the same source code
4289 and the same optimization options for both compilations.
4291 With @option{-fbranch-probabilities}, GCC puts a
4292 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4293 These can be used to improve optimization. Currently, they are only
4294 used in one place: in @file{reorg.c}, instead of guessing which path a
4295 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4296 exactly determine which path is taken more often.
4300 Use a graph coloring register allocator. Currently this option is meant
4301 for testing, so we are interested to hear about miscompilations with
4306 Perform tail duplication to enlarge superblock size. This transformation
4307 simplifies the control flow of the function allowing other optimizations to do
4310 @item -funit-at-a-time
4311 @opindex funit-at-a-time
4312 Parse the whole compilation unit before starting to produce code. This allows some
4313 extra optimizations to take place but consumes more memory.
4315 @item -funroll-loops
4316 @opindex funroll-loops
4317 Unroll loops whose number of iterations can be determined at compile time or
4318 upon entry to the loop. @option{-funroll-loops} implies
4319 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4320 (i.e. complete removal of loops with small constant number of iterations).
4321 This option makes code larger, and may or may not make it run faster.
4323 @item -funroll-all-loops
4324 @opindex funroll-all-loops
4325 Unroll all loops, even if their number of iterations is uncertain when
4326 the loop is entered. This usually makes programs run more slowly.
4327 @option{-funroll-all-loops} implies the same options as
4328 @option{-funroll-loops}.
4331 @opindex fpeel-loops
4332 Peels the loops for that there is enough information that they do not
4333 roll much (from profile feedback). It also turns on complete loop peeling
4334 (i.e. complete removal of loops with small constant number of iterations).
4336 @item -funswitch-loops
4337 @opindex funswitch-loops
4338 Move branches with loop invariant conditions out of the loop, with duplicates
4339 of the loop on both branches (modified according to result of the condition).
4341 @item -fold-unroll-loops
4342 @opindex fold-unroll-loops
4343 Unroll loops whose number of iterations can be determined at compile
4344 time or upon entry to the loop, using the old loop unroller whose loop
4345 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4346 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4347 option makes code larger, and may or may not make it run faster.
4349 @item -fold-unroll-all-loops
4350 @opindex fold-unroll-all-loops
4351 Unroll all loops, even if their number of iterations is uncertain when
4352 the loop is entered. This is done using the old loop unroller whose loop
4353 recognition is based on notes from frontend. This usually makes programs run more slowly.
4354 @option{-fold-unroll-all-loops} implies the same options as
4355 @option{-fold-unroll-loops}.
4357 @item -funswitch-loops
4358 @opindex funswitch-loops
4359 Move branches with loop invariant conditions out of the loop, with duplicates
4360 of the loop on both branches (modified according to result of the condition).
4362 @item -funswitch-loops
4363 @opindex funswitch-loops
4364 Move branches with loop invariant conditions out of the loop, with duplicates
4365 of the loop on both branches (modified according to result of the condition).
4367 @item -fprefetch-loop-arrays
4368 @opindex fprefetch-loop-arrays
4369 If supported by the target machine, generate instructions to prefetch
4370 memory to improve the performance of loops that access large arrays.
4372 Disabled at level @option{-Os}.
4374 @item -ffunction-sections
4375 @itemx -fdata-sections
4376 @opindex ffunction-sections
4377 @opindex fdata-sections
4378 Place each function or data item into its own section in the output
4379 file if the target supports arbitrary sections. The name of the
4380 function or the name of the data item determines the section's name
4383 Use these options on systems where the linker can perform optimizations
4384 to improve locality of reference in the instruction space. Most systems
4385 using the ELF object format and SPARC processors running Solaris 2 have
4386 linkers with such optimizations. AIX may have these optimizations in
4389 Only use these options when there are significant benefits from doing
4390 so. When you specify these options, the assembler and linker will
4391 create larger object and executable files and will also be slower.
4392 You will not be able to use @code{gprof} on all systems if you
4393 specify this option and you may have problems with debugging if
4394 you specify both this option and @option{-g}.
4398 Perform optimizations in static single assignment form. Each function's
4399 flow graph is translated into SSA form, optimizations are performed, and
4400 the flow graph is translated back from SSA form. Users should not
4401 specify this option, since it is not yet ready for production use.
4405 Perform Sparse Conditional Constant Propagation in SSA form. Requires
4406 @option{-fssa}. Like @option{-fssa}, this is an experimental feature.
4410 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
4411 Like @option{-fssa}, this is an experimental feature.
4416 @item --param @var{name}=@var{value}
4418 In some places, GCC uses various constants to control the amount of
4419 optimization that is done. For example, GCC will not inline functions
4420 that contain more that a certain number of instructions. You can
4421 control some of these constants on the command-line using the
4422 @option{--param} option.
4424 In each case, the @var{value} is an integer. The allowable choices for
4425 @var{name} are given in the following table:
4428 @item max-crossjump-edges
4429 The maximum number of incoming edges to consider for crossjumping.
4430 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
4431 the number of edges incoming to each block. Increasing values mean
4432 more aggressive optimization, making the compile time increase with
4433 probably small improvement in executable size.
4435 @item max-delay-slot-insn-search
4436 The maximum number of instructions to consider when looking for an
4437 instruction to fill a delay slot. If more than this arbitrary number of
4438 instructions is searched, the time savings from filling the delay slot
4439 will be minimal so stop searching. Increasing values mean more
4440 aggressive optimization, making the compile time increase with probably
4441 small improvement in executable run time.
4443 @item max-delay-slot-live-search
4444 When trying to fill delay slots, the maximum number of instructions to
4445 consider when searching for a block with valid live register
4446 information. Increasing this arbitrarily chosen value means more
4447 aggressive optimization, increasing the compile time. This parameter
4448 should be removed when the delay slot code is rewritten to maintain the
4451 @item max-gcse-memory
4452 The approximate maximum amount of memory that will be allocated in
4453 order to perform the global common subexpression elimination
4454 optimization. If more memory than specified is required, the
4455 optimization will not be done.
4457 @item max-gcse-passes
4458 The maximum number of passes of GCSE to run.
4460 @item max-pending-list-length
4461 The maximum number of pending dependencies scheduling will allow
4462 before flushing the current state and starting over. Large functions
4463 with few branches or calls can create excessively large lists which
4464 needlessly consume memory and resources.
4466 @item max-inline-insns-single
4467 Several parameters control the tree inliner used in gcc.
4468 This number sets the maximum number of instructions (counted in gcc's
4469 internal representation) in a single function that the tree inliner
4470 will consider for inlining. This only affects functions declared
4471 inline and methods implemented in a class declaration (C++).
4472 The default value is 300.
4474 @item max-inline-insns-auto
4475 When you use @option{-finline-functions} (included in @option{-O3}),
4476 a lot of functions that would otherwise not be considered for inlining
4477 by the compiler will be investigated. To those functions, a different
4478 (more restrictive) limit compared to functions declared inline can
4480 The default value is 300.
4482 @item max-inline-insns
4483 The tree inliner does decrease the allowable size for single functions
4484 to be inlined after we already inlined the number of instructions
4485 given here by repeated inlining. This number should be a factor of
4486 two or more larger than the single function limit.
4487 Higher numbers result in better runtime performance, but incur higher
4488 compile-time resource (CPU time, memory) requirements and result in
4489 larger binaries. Very high values are not advisable, as too large
4490 binaries may adversely affect runtime performance.
4491 The default value is 600.
4493 @item max-inline-slope
4494 After exceeding the maximum number of inlined instructions by repeated
4495 inlining, a linear function is used to decrease the allowable size
4496 for single functions. The slope of that function is the negative
4497 reciprocal of the number specified here.
4498 The default value is 32.
4500 @item min-inline-insns
4501 The repeated inlining is throttled more and more by the linear function
4502 after exceeding the limit. To avoid too much throttling, a minimum for
4503 this function is specified here to allow repeated inlining for very small
4504 functions even when a lot of repeated inlining already has been done.
4505 The default value is 130.
4507 @item max-inline-insns-rtl
4508 For languages that use the RTL inliner (this happens at a later stage
4509 than tree inlining), you can set the maximum allowable size (counted
4510 in RTL instructions) for the RTL inliner with this parameter.
4511 The default value is 600.
4514 @item max-unrolled-insns
4515 The maximum number of instructions that a loop should have if that loop
4516 is unrolled, and if the loop is unrolled, it determines how many times
4517 the loop code is unrolled.
4519 @item max-average-unrolled-insns
4520 The maximum number of instructions biased by probabilities of their execution
4521 that a loop should have if that loop is unrolled, and if the loop is unrolled,
4522 it determines how many times the loop code is unrolled.
4524 @item max-unroll-times
4525 The maximum number of unrollings of a single loop.
4527 @item max-peeled-insns
4528 The maximum number of instructions that a loop should have if that loop
4529 is peeled, and if the loop is peeled, it determines how many times
4530 the loop code is peeled.
4532 @item max-peel-times
4533 The maximum number of peelings of a single loop.
4535 @item max-completely-peeled-insns
4536 The maximum number of insns of a completely peeled loop.
4538 @item max-completely-peel-times
4539 The maximum number of iterations of a loop to be suitable for complete peeling.
4541 @item max-unswitch-insns
4542 The maximum number of insns of an unswitched loop.
4544 @item max-unswitch-level
4545 The maximum number of branches unswitched in a single loop.
4547 @item hot-bb-count-fraction
4548 Select fraction of the maximal count of repetitions of basic block in program
4549 given basic block needs to have to be considered hot.
4551 @item hot-bb-frequency-fraction
4552 Select fraction of the maximal frequency of executions of basic block in
4553 function given basic block needs to have to be considered hot
4555 @item tracer-dynamic-coverage
4556 @itemx tracer-dynamic-coverage-feedback
4558 This value is used to limit superblock formation once the given percentage of
4559 executed instructions is covered. This limits unnecessary code size
4562 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4563 feedback is available. The real profiles (as opposed to statically estimated
4564 ones) are much less balanced allowing the threshold to be larger value.
4566 @item tracer-max-code-growth
4567 Stop tail duplication once code growth has reached given percentage. This is
4568 rather hokey argument, as most of the duplicates will be eliminated later in
4569 cross jumping, so it may be set to much higher values than is the desired code
4572 @item tracer-min-branch-ratio
4574 Stop reverse growth when the reverse probability of best edge is less than this
4575 threshold (in percent).
4577 @item tracer-min-branch-ratio
4578 @itemx tracer-min-branch-ratio-feedback
4580 Stop forward growth if the best edge do have probability lower than this
4583 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
4584 compilation for profile feedback and one for compilation without. The value
4585 for compilation with profile feedback needs to be more conservative (higher) in
4586 order to make tracer effective.
4588 @item ggc-min-expand
4590 GCC uses a garbage collector to manage its own memory allocation. This
4591 parameter specifies the minimum percentage by which the garbage
4592 collector's heap should be allowed to expand between collections.
4593 Tuning this may improve compilation speed; it has no effect on code
4596 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
4597 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
4598 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
4599 GCC is not able to calculate RAM on a particular platform, the lower
4600 bound of 30% is used. Setting this parameter and
4601 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4602 every opportunity. This is extremely slow, but can be useful for
4605 @item ggc-min-heapsize
4607 Minimum size of the garbage collector's heap before it begins bothering
4608 to collect garbage. The first collection occurs after the heap expands
4609 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4610 tuning this may improve compilation speed, and has no effect on code
4613 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
4614 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
4615 available, the notion of "RAM" is the smallest of actual RAM,
4616 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
4617 RAM on a particular platform, the lower bound is used. Setting this
4618 parameter very large effectively disables garbage collection. Setting
4619 this parameter and @option{ggc-min-expand} to zero causes a full
4620 collection to occur at every opportunity.
4622 @item reorder-blocks-duplicate
4623 @itemx reorder-blocks-duplicate-feedback
4625 Used by basic block reordering pass to decide whether to use uncondtional
4626 branch or duplicate the code on it's destination. Code is duplicated when it's
4627 estimated size is smaller than this value multiplied by the estimated size of
4628 unconditinal jump in the hot spots of the program.
4630 The @option{reorder-block-duplicate-feedback} is used only when profile
4631 feedback is available and may be set to higher values than
4632 @option{reorder-block-duplicate} since information about the hot spots is more
4637 @node Preprocessor Options
4638 @section Options Controlling the Preprocessor
4639 @cindex preprocessor options
4640 @cindex options, preprocessor
4642 These options control the C preprocessor, which is run on each C source
4643 file before actual compilation.
4645 If you use the @option{-E} option, nothing is done except preprocessing.
4646 Some of these options make sense only together with @option{-E} because
4647 they cause the preprocessor output to be unsuitable for actual
4652 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4653 and pass @var{option} directly through to the preprocessor. If
4654 @var{option} contains commas, it is split into multiple options at the
4655 commas. However, many options are modified, translated or interpreted
4656 by the compiler driver before being passed to the preprocessor, and
4657 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4658 interface is undocumented and subject to change, so whenever possible
4659 you should avoid using @option{-Wp} and let the driver handle the
4662 @item -Xpreprocessor @var{option}
4663 @opindex preprocessor
4664 Pass @var{option} as an option to the preprocessor. You can use this to
4665 supply system-specific preprocessor options which GCC does not know how to
4668 If you want to pass an option that takes an argument, you must use
4669 @option{-Xpreprocessor} twice, once for the option and once for the argument.
4672 @include cppopts.texi
4674 @node Assembler Options
4675 @section Passing Options to the Assembler
4677 @c prevent bad page break with this line
4678 You can pass options to the assembler.
4681 @item -Wa,@var{option}
4683 Pass @var{option} as an option to the assembler. If @var{option}
4684 contains commas, it is split into multiple options at the commas.
4686 @item -Xassembler @var{option}
4688 Pass @var{option} as an option to the assembler. You can use this to
4689 supply system-specific assembler options which GCC does not know how to
4692 If you want to pass an option that takes an argument, you must use
4693 @option{-Xassembler} twice, once for the option and once for the argument.
4698 @section Options for Linking
4699 @cindex link options
4700 @cindex options, linking
4702 These options come into play when the compiler links object files into
4703 an executable output file. They are meaningless if the compiler is
4704 not doing a link step.
4708 @item @var{object-file-name}
4709 A file name that does not end in a special recognized suffix is
4710 considered to name an object file or library. (Object files are
4711 distinguished from libraries by the linker according to the file
4712 contents.) If linking is done, these object files are used as input
4721 If any of these options is used, then the linker is not run, and
4722 object file names should not be used as arguments. @xref{Overall
4726 @item -l@var{library}
4727 @itemx -l @var{library}
4729 Search the library named @var{library} when linking. (The second
4730 alternative with the library as a separate argument is only for
4731 POSIX compliance and is not recommended.)
4733 It makes a difference where in the command you write this option; the
4734 linker searches and processes libraries and object files in the order they
4735 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4736 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4737 to functions in @samp{z}, those functions may not be loaded.
4739 The linker searches a standard list of directories for the library,
4740 which is actually a file named @file{lib@var{library}.a}. The linker
4741 then uses this file as if it had been specified precisely by name.
4743 The directories searched include several standard system directories
4744 plus any that you specify with @option{-L}.
4746 Normally the files found this way are library files---archive files
4747 whose members are object files. The linker handles an archive file by
4748 scanning through it for members which define symbols that have so far
4749 been referenced but not defined. But if the file that is found is an
4750 ordinary object file, it is linked in the usual fashion. The only
4751 difference between using an @option{-l} option and specifying a file name
4752 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4753 and searches several directories.
4757 You need this special case of the @option{-l} option in order to
4758 link an Objective-C program.
4761 @opindex nostartfiles
4762 Do not use the standard system startup files when linking.
4763 The standard system libraries are used normally, unless @option{-nostdlib}
4764 or @option{-nodefaultlibs} is used.
4766 @item -nodefaultlibs
4767 @opindex nodefaultlibs
4768 Do not use the standard system libraries when linking.
4769 Only the libraries you specify will be passed to the linker.
4770 The standard startup files are used normally, unless @option{-nostartfiles}
4771 is used. The compiler may generate calls to memcmp, memset, and memcpy
4772 for System V (and ISO C) environments or to bcopy and bzero for
4773 BSD environments. These entries are usually resolved by entries in
4774 libc. These entry points should be supplied through some other
4775 mechanism when this option is specified.
4779 Do not use the standard system startup files or libraries when linking.
4780 No startup files and only the libraries you specify will be passed to
4781 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4782 for System V (and ISO C) environments or to bcopy and bzero for
4783 BSD environments. These entries are usually resolved by entries in
4784 libc. These entry points should be supplied through some other
4785 mechanism when this option is specified.
4787 @cindex @option{-lgcc}, use with @option{-nostdlib}
4788 @cindex @option{-nostdlib} and unresolved references
4789 @cindex unresolved references and @option{-nostdlib}
4790 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
4791 @cindex @option{-nodefaultlibs} and unresolved references
4792 @cindex unresolved references and @option{-nodefaultlibs}
4793 One of the standard libraries bypassed by @option{-nostdlib} and
4794 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4795 that GCC uses to overcome shortcomings of particular machines, or special
4796 needs for some languages.
4797 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
4798 Collection (GCC) Internals},
4799 for more discussion of @file{libgcc.a}.)
4800 In most cases, you need @file{libgcc.a} even when you want to avoid
4801 other standard libraries. In other words, when you specify @option{-nostdlib}
4802 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
4803 This ensures that you have no unresolved references to internal GCC
4804 library subroutines. (For example, @samp{__main}, used to ensure C++
4805 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
4806 GNU Compiler Collection (GCC) Internals}.)
4810 Remove all symbol table and relocation information from the executable.
4814 On systems that support dynamic linking, this prevents linking with the shared
4815 libraries. On other systems, this option has no effect.
4819 Produce a shared object which can then be linked with other objects to
4820 form an executable. Not all systems support this option. For predictable
4821 results, you must also specify the same set of options that were used to
4822 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
4823 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4824 needs to build supplementary stub code for constructors to work. On
4825 multi-libbed systems, @samp{gcc -shared} must select the correct support
4826 libraries to link against. Failing to supply the correct flags may lead
4827 to subtle defects. Supplying them in cases where they are not necessary
4830 @item -shared-libgcc
4831 @itemx -static-libgcc
4832 @opindex shared-libgcc
4833 @opindex static-libgcc
4834 On systems that provide @file{libgcc} as a shared library, these options
4835 force the use of either the shared or static version respectively.
4836 If no shared version of @file{libgcc} was built when the compiler was
4837 configured, these options have no effect.
4839 There are several situations in which an application should use the
4840 shared @file{libgcc} instead of the static version. The most common
4841 of these is when the application wishes to throw and catch exceptions
4842 across different shared libraries. In that case, each of the libraries
4843 as well as the application itself should use the shared @file{libgcc}.
4845 Therefore, the G++ and GCJ drivers automatically add
4846 @option{-shared-libgcc} whenever you build a shared library or a main
4847 executable, because C++ and Java programs typically use exceptions, so
4848 this is the right thing to do.
4850 If, instead, you use the GCC driver to create shared libraries, you may
4851 find that they will not always be linked with the shared @file{libgcc}.
4852 If GCC finds, at its configuration time, that you have a GNU linker that
4853 does not support option @option{--eh-frame-hdr}, it will link the shared
4854 version of @file{libgcc} into shared libraries by default. Otherwise,
4855 it will take advantage of the linker and optimize away the linking with
4856 the shared version of @file{libgcc}, linking with the static version of
4857 libgcc by default. This allows exceptions to propagate through such
4858 shared libraries, without incurring relocation costs at library load
4861 However, if a library or main executable is supposed to throw or catch
4862 exceptions, you must link it using the G++ or GCJ driver, as appropriate
4863 for the languages used in the program, or using the option
4864 @option{-shared-libgcc}, such that it is linked with the shared
4869 Bind references to global symbols when building a shared object. Warn
4870 about any unresolved references (unless overridden by the link editor
4871 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
4874 @item -Xlinker @var{option}
4876 Pass @var{option} as an option to the linker. You can use this to
4877 supply system-specific linker options which GCC does not know how to
4880 If you want to pass an option that takes an argument, you must use
4881 @option{-Xlinker} twice, once for the option and once for the argument.
4882 For example, to pass @option{-assert definitions}, you must write
4883 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
4884 @option{-Xlinker "-assert definitions"}, because this passes the entire
4885 string as a single argument, which is not what the linker expects.
4887 @item -Wl,@var{option}
4889 Pass @var{option} as an option to the linker. If @var{option} contains
4890 commas, it is split into multiple options at the commas.
4892 @item -u @var{symbol}
4894 Pretend the symbol @var{symbol} is undefined, to force linking of
4895 library modules to define it. You can use @option{-u} multiple times with
4896 different symbols to force loading of additional library modules.
4899 @node Directory Options
4900 @section Options for Directory Search
4901 @cindex directory options
4902 @cindex options, directory search
4905 These options specify directories to search for header files, for
4906 libraries and for parts of the compiler:
4911 Add the directory @var{dir} to the head of the list of directories to be
4912 searched for header files. This can be used to override a system header
4913 file, substituting your own version, since these directories are
4914 searched before the system header file directories. However, you should
4915 not use this option to add directories that contain vendor-supplied
4916 system header files (use @option{-isystem} for that). If you use more than
4917 one @option{-I} option, the directories are scanned in left-to-right
4918 order; the standard system directories come after.
4920 If a standard system include directory, or a directory specified with
4921 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
4922 option will be ignored. The directory will still be searched but as a
4923 system directory at its normal position in the system include chain.
4924 This is to ensure that GCC's procedure to fix buggy system headers and
4925 the ordering for the include_next directive are not inadvertently changed.
4926 If you really need to change the search order for system directories,
4927 use the @option{-nostdinc} and/or @option{-isystem} options.
4931 Any directories you specify with @option{-I} options before the @option{-I-}
4932 option are searched only for the case of @samp{#include "@var{file}"};
4933 they are not searched for @samp{#include <@var{file}>}.
4935 If additional directories are specified with @option{-I} options after
4936 the @option{-I-}, these directories are searched for all @samp{#include}
4937 directives. (Ordinarily @emph{all} @option{-I} directories are used
4940 In addition, the @option{-I-} option inhibits the use of the current
4941 directory (where the current input file came from) as the first search
4942 directory for @samp{#include "@var{file}"}. There is no way to
4943 override this effect of @option{-I-}. With @option{-I.} you can specify
4944 searching the directory which was current when the compiler was
4945 invoked. That is not exactly the same as what the preprocessor does
4946 by default, but it is often satisfactory.
4948 @option{-I-} does not inhibit the use of the standard system directories
4949 for header files. Thus, @option{-I-} and @option{-nostdinc} are
4954 Add directory @var{dir} to the list of directories to be searched
4957 @item -B@var{prefix}
4959 This option specifies where to find the executables, libraries,
4960 include files, and data files of the compiler itself.
4962 The compiler driver program runs one or more of the subprograms
4963 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
4964 @var{prefix} as a prefix for each program it tries to run, both with and
4965 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
4967 For each subprogram to be run, the compiler driver first tries the
4968 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
4969 was not specified, the driver tries two standard prefixes, which are
4970 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
4971 those results in a file name that is found, the unmodified program
4972 name is searched for using the directories specified in your
4973 @env{PATH} environment variable.
4975 The compiler will check to see if the path provided by the @option{-B}
4976 refers to a directory, and if necessary it will add a directory
4977 separator character at the end of the path.
4979 @option{-B} prefixes that effectively specify directory names also apply
4980 to libraries in the linker, because the compiler translates these
4981 options into @option{-L} options for the linker. They also apply to
4982 includes files in the preprocessor, because the compiler translates these
4983 options into @option{-isystem} options for the preprocessor. In this case,
4984 the compiler appends @samp{include} to the prefix.
4986 The run-time support file @file{libgcc.a} can also be searched for using
4987 the @option{-B} prefix, if needed. If it is not found there, the two
4988 standard prefixes above are tried, and that is all. The file is left
4989 out of the link if it is not found by those means.
4991 Another way to specify a prefix much like the @option{-B} prefix is to use
4992 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
4995 As a special kludge, if the path provided by @option{-B} is
4996 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
4997 9, then it will be replaced by @file{[dir/]include}. This is to help
4998 with boot-strapping the compiler.
5000 @item -specs=@var{file}
5002 Process @var{file} after the compiler reads in the standard @file{specs}
5003 file, in order to override the defaults that the @file{gcc} driver
5004 program uses when determining what switches to pass to @file{cc1},
5005 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5006 @option{-specs=@var{file}} can be specified on the command line, and they
5007 are processed in order, from left to right.
5013 @section Specifying subprocesses and the switches to pass to them
5016 @command{gcc} is a driver program. It performs its job by invoking a
5017 sequence of other programs to do the work of compiling, assembling and
5018 linking. GCC interprets its command-line parameters and uses these to
5019 deduce which programs it should invoke, and which command-line options
5020 it ought to place on their command lines. This behavior is controlled
5021 by @dfn{spec strings}. In most cases there is one spec string for each
5022 program that GCC can invoke, but a few programs have multiple spec
5023 strings to control their behavior. The spec strings built into GCC can
5024 be overridden by using the @option{-specs=} command-line switch to specify
5027 @dfn{Spec files} are plaintext files that are used to construct spec
5028 strings. They consist of a sequence of directives separated by blank
5029 lines. The type of directive is determined by the first non-whitespace
5030 character on the line and it can be one of the following:
5033 @item %@var{command}
5034 Issues a @var{command} to the spec file processor. The commands that can
5038 @item %include <@var{file}>
5040 Search for @var{file} and insert its text at the current point in the
5043 @item %include_noerr <@var{file}>
5044 @cindex %include_noerr
5045 Just like @samp{%include}, but do not generate an error message if the include
5046 file cannot be found.
5048 @item %rename @var{old_name} @var{new_name}
5050 Rename the spec string @var{old_name} to @var{new_name}.
5054 @item *[@var{spec_name}]:
5055 This tells the compiler to create, override or delete the named spec
5056 string. All lines after this directive up to the next directive or
5057 blank line are considered to be the text for the spec string. If this
5058 results in an empty string then the spec will be deleted. (Or, if the
5059 spec did not exist, then nothing will happened.) Otherwise, if the spec
5060 does not currently exist a new spec will be created. If the spec does
5061 exist then its contents will be overridden by the text of this
5062 directive, unless the first character of that text is the @samp{+}
5063 character, in which case the text will be appended to the spec.
5065 @item [@var{suffix}]:
5066 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5067 and up to the next directive or blank line are considered to make up the
5068 spec string for the indicated suffix. When the compiler encounters an
5069 input file with the named suffix, it will processes the spec string in
5070 order to work out how to compile that file. For example:
5077 This says that any input file whose name ends in @samp{.ZZ} should be
5078 passed to the program @samp{z-compile}, which should be invoked with the
5079 command-line switch @option{-input} and with the result of performing the
5080 @samp{%i} substitution. (See below.)
5082 As an alternative to providing a spec string, the text that follows a
5083 suffix directive can be one of the following:
5086 @item @@@var{language}
5087 This says that the suffix is an alias for a known @var{language}. This is
5088 similar to using the @option{-x} command-line switch to GCC to specify a
5089 language explicitly. For example:
5096 Says that .ZZ files are, in fact, C++ source files.
5099 This causes an error messages saying:
5102 @var{name} compiler not installed on this system.
5106 GCC already has an extensive list of suffixes built into it.
5107 This directive will add an entry to the end of the list of suffixes, but
5108 since the list is searched from the end backwards, it is effectively
5109 possible to override earlier entries using this technique.
5113 GCC has the following spec strings built into it. Spec files can
5114 override these strings or create their own. Note that individual
5115 targets can also add their own spec strings to this list.
5118 asm Options to pass to the assembler
5119 asm_final Options to pass to the assembler post-processor
5120 cpp Options to pass to the C preprocessor
5121 cc1 Options to pass to the C compiler
5122 cc1plus Options to pass to the C++ compiler
5123 endfile Object files to include at the end of the link
5124 link Options to pass to the linker
5125 lib Libraries to include on the command line to the linker
5126 libgcc Decides which GCC support library to pass to the linker
5127 linker Sets the name of the linker
5128 predefines Defines to be passed to the C preprocessor
5129 signed_char Defines to pass to CPP to say whether @code{char} is signed
5131 startfile Object files to include at the start of the link
5134 Here is a small example of a spec file:
5140 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5143 This example renames the spec called @samp{lib} to @samp{old_lib} and
5144 then overrides the previous definition of @samp{lib} with a new one.
5145 The new definition adds in some extra command-line options before
5146 including the text of the old definition.
5148 @dfn{Spec strings} are a list of command-line options to be passed to their
5149 corresponding program. In addition, the spec strings can contain
5150 @samp{%}-prefixed sequences to substitute variable text or to
5151 conditionally insert text into the command line. Using these constructs
5152 it is possible to generate quite complex command lines.
5154 Here is a table of all defined @samp{%}-sequences for spec
5155 strings. Note that spaces are not generated automatically around the
5156 results of expanding these sequences. Therefore you can concatenate them
5157 together or combine them with constant text in a single argument.
5161 Substitute one @samp{%} into the program name or argument.
5164 Substitute the name of the input file being processed.
5167 Substitute the basename of the input file being processed.
5168 This is the substring up to (and not including) the last period
5169 and not including the directory.
5172 This is the same as @samp{%b}, but include the file suffix (text after
5176 Marks the argument containing or following the @samp{%d} as a
5177 temporary file name, so that that file will be deleted if GCC exits
5178 successfully. Unlike @samp{%g}, this contributes no text to the
5181 @item %g@var{suffix}
5182 Substitute a file name that has suffix @var{suffix} and is chosen
5183 once per compilation, and mark the argument in the same way as
5184 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5185 name is now chosen in a way that is hard to predict even when previously
5186 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5187 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5188 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5189 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5190 was simply substituted with a file name chosen once per compilation,
5191 without regard to any appended suffix (which was therefore treated
5192 just like ordinary text), making such attacks more likely to succeed.
5194 @item %u@var{suffix}
5195 Like @samp{%g}, but generates a new temporary file name even if
5196 @samp{%u@var{suffix}} was already seen.
5198 @item %U@var{suffix}
5199 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5200 new one if there is no such last file name. In the absence of any
5201 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5202 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5203 would involve the generation of two distinct file names, one
5204 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5205 simply substituted with a file name chosen for the previous @samp{%u},
5206 without regard to any appended suffix.
5208 @item %j@var{suffix}
5209 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5210 writable, and if save-temps is off; otherwise, substitute the name
5211 of a temporary file, just like @samp{%u}. This temporary file is not
5212 meant for communication between processes, but rather as a junk
5215 @item %|@var{suffix}
5216 @itemx %m@var{suffix}
5217 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5218 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5219 all. These are the two most common ways to instruct a program that it
5220 should read from standard input or write to standard output. If you
5221 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5222 construct: see for example @file{f/lang-specs.h}.
5224 @item %.@var{SUFFIX}
5225 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5226 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5227 terminated by the next space or %.
5230 Marks the argument containing or following the @samp{%w} as the
5231 designated output file of this compilation. This puts the argument
5232 into the sequence of arguments that @samp{%o} will substitute later.
5235 Substitutes the names of all the output files, with spaces
5236 automatically placed around them. You should write spaces
5237 around the @samp{%o} as well or the results are undefined.
5238 @samp{%o} is for use in the specs for running the linker.
5239 Input files whose names have no recognized suffix are not compiled
5240 at all, but they are included among the output files, so they will
5244 Substitutes the suffix for object files. Note that this is
5245 handled specially when it immediately follows @samp{%g, %u, or %U},
5246 because of the need for those to form complete file names. The
5247 handling is such that @samp{%O} is treated exactly as if it had already
5248 been substituted, except that @samp{%g, %u, and %U} do not currently
5249 support additional @var{suffix} characters following @samp{%O} as they would
5250 following, for example, @samp{.o}.
5253 Substitutes the standard macro predefinitions for the
5254 current target machine. Use this when running @code{cpp}.
5257 Like @samp{%p}, but puts @samp{__} before and after the name of each
5258 predefined macro, except for macros that start with @samp{__} or with
5259 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5263 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5264 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5265 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5269 Current argument is the name of a library or startup file of some sort.
5270 Search for that file in a standard list of directories and substitute
5271 the full name found.
5274 Print @var{str} as an error message. @var{str} is terminated by a newline.
5275 Use this when inconsistent options are detected.
5278 Substitute the contents of spec string @var{name} at this point.
5281 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5283 @item %x@{@var{option}@}
5284 Accumulate an option for @samp{%X}.
5287 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5291 Output the accumulated assembler options specified by @option{-Wa}.
5294 Output the accumulated preprocessor options specified by @option{-Wp}.
5297 Process the @code{asm} spec. This is used to compute the
5298 switches to be passed to the assembler.
5301 Process the @code{asm_final} spec. This is a spec string for
5302 passing switches to an assembler post-processor, if such a program is
5306 Process the @code{link} spec. This is the spec for computing the
5307 command line passed to the linker. Typically it will make use of the
5308 @samp{%L %G %S %D and %E} sequences.
5311 Dump out a @option{-L} option for each directory that GCC believes might
5312 contain startup files. If the target supports multilibs then the
5313 current multilib directory will be prepended to each of these paths.
5316 Output the multilib directory with directory separators replaced with
5317 @samp{_}. If multilib directories are not set, or the multilib directory is
5318 @file{.} then this option emits nothing.
5321 Process the @code{lib} spec. This is a spec string for deciding which
5322 libraries should be included on the command line to the linker.
5325 Process the @code{libgcc} spec. This is a spec string for deciding
5326 which GCC support library should be included on the command line to the linker.
5329 Process the @code{startfile} spec. This is a spec for deciding which
5330 object files should be the first ones passed to the linker. Typically
5331 this might be a file named @file{crt0.o}.
5334 Process the @code{endfile} spec. This is a spec string that specifies
5335 the last object files that will be passed to the linker.
5338 Process the @code{cpp} spec. This is used to construct the arguments
5339 to be passed to the C preprocessor.
5342 Process the @code{signed_char} spec. This is intended to be used
5343 to tell cpp whether a char is signed. It typically has the definition:
5345 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
5349 Process the @code{cc1} spec. This is used to construct the options to be
5350 passed to the actual C compiler (@samp{cc1}).
5353 Process the @code{cc1plus} spec. This is used to construct the options to be
5354 passed to the actual C++ compiler (@samp{cc1plus}).
5357 Substitute the variable part of a matched option. See below.
5358 Note that each comma in the substituted string is replaced by
5362 Remove all occurrences of @code{-S} from the command line. Note---this
5363 command is position dependent. @samp{%} commands in the spec string
5364 before this one will see @code{-S}, @samp{%} commands in the spec string
5365 after this one will not.
5367 @item %:@var{function}(@var{args})
5368 Call the named function @var{function}, passing it @var{args}.
5369 @var{args} is first processed as a nested spec string, then split
5370 into an argument vector in the usual fashion. The function returns
5371 a string which is processed as if it had appeared literally as part
5372 of the current spec.
5374 The following built-in spec functions are provided:
5377 @item @code{if-exists}
5378 The @code{if-exists} spec function takes one argument, an absolute
5379 pathname to a file. If the file exists, @code{if-exists} returns the
5380 pathname. Here is a small example of its usage:
5384 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5387 @item @code{if-exists-else}
5388 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5389 spec function, except that it takes two arguments. The first argument is
5390 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5391 returns the pathname. If it does not exist, it returns the second argument.
5392 This way, @code{if-exists-else} can be used to select one file or another,
5393 based on the existence of the first. Here is a small example of its usage:
5397 crt0%O%s %:if-exists(crti%O%s) \
5398 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5403 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5404 If that switch was not specified, this substitutes nothing. Note that
5405 the leading dash is omitted when specifying this option, and it is
5406 automatically inserted if the substitution is performed. Thus the spec
5407 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5408 and would output the command line option @option{-foo}.
5410 @item %W@{@code{S}@}
5411 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5414 @item %@{@code{S}*@}
5415 Substitutes all the switches specified to GCC whose names start
5416 with @code{-S}, but which also take an argument. This is used for
5417 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5418 GCC considers @option{-o foo} as being
5419 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5420 text, including the space. Thus two arguments would be generated.
5422 @item %@{@code{S}*&@code{T}*@}
5423 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5424 (the order of @code{S} and @code{T} in the spec is not significant).
5425 There can be any number of ampersand-separated variables; for each the
5426 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5428 @item %@{@code{S}:@code{X}@}
5429 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
5431 @item %@{!@code{S}:@code{X}@}
5432 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
5434 @item %@{@code{S}*:@code{X}@}
5435 Substitutes @code{X} if one or more switches whose names start with
5436 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
5437 once, no matter how many such switches appeared. However, if @code{%*}
5438 appears somewhere in @code{X}, then @code{X} will be substituted once
5439 for each matching switch, with the @code{%*} replaced by the part of
5440 that switch that matched the @code{*}.
5442 @item %@{.@code{S}:@code{X}@}
5443 Substitutes @code{X}, if processing a file with suffix @code{S}.
5445 @item %@{!.@code{S}:@code{X}@}
5446 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
5448 @item %@{@code{S}|@code{P}:@code{X}@}
5449 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
5450 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
5451 although they have a stronger binding than the @samp{|}. If @code{%*}
5452 appears in @code{X}, all of the alternatives must be starred, and only
5453 the first matching alternative is substituted.
5455 For example, a spec string like this:
5458 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5461 will output the following command-line options from the following input
5462 command-line options:
5467 -d fred.c -foo -baz -boggle
5468 -d jim.d -bar -baz -boggle
5471 @item %@{S:X; T:Y; :D@}
5473 If @code{S} was given to GCC, substitues @code{X}; else if @code{T} was
5474 given to GCC, substitues @code{Y}; else substitutes @code{D}. There can
5475 be as many clauses as you need. This may be combined with @code{.},
5476 @code{!}, @code{|}, and @code{*} as needed.
5481 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
5482 construct may contain other nested @samp{%} constructs or spaces, or
5483 even newlines. They are processed as usual, as described above.
5484 Trailing white space in @code{X} is ignored. White space may also
5485 appear anywhere on the left side of the colon in these constructs,
5486 except between @code{.} or @code{*} and the corresponding word.
5488 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
5489 handled specifically in these constructs. If another value of
5490 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5491 @option{-W} switch is found later in the command line, the earlier
5492 switch value is ignored, except with @{@code{S}*@} where @code{S} is
5493 just one letter, which passes all matching options.
5495 The character @samp{|} at the beginning of the predicate text is used to
5496 indicate that a command should be piped to the following command, but
5497 only if @option{-pipe} is specified.
5499 It is built into GCC which switches take arguments and which do not.
5500 (You might think it would be useful to generalize this to allow each
5501 compiler's spec to say which switches take arguments. But this cannot
5502 be done in a consistent fashion. GCC cannot even decide which input
5503 files have been specified without knowing which switches take arguments,
5504 and it must know which input files to compile in order to tell which
5507 GCC also knows implicitly that arguments starting in @option{-l} are to be
5508 treated as compiler output files, and passed to the linker in their
5509 proper position among the other output files.
5511 @c man begin OPTIONS
5513 @node Target Options
5514 @section Specifying Target Machine and Compiler Version
5515 @cindex target options
5516 @cindex cross compiling
5517 @cindex specifying machine version
5518 @cindex specifying compiler version and target machine
5519 @cindex compiler version, specifying
5520 @cindex target machine, specifying
5522 The usual way to run GCC is to run the executable called @file{gcc}, or
5523 @file{<machine>-gcc} when cross-compiling, or
5524 @file{<machine>-gcc-<version>} to run a version other than the one that
5525 was installed last. Sometimes this is inconvenient, so GCC provides
5526 options that will switch to another cross-compiler or version.
5529 @item -b @var{machine}
5531 The argument @var{machine} specifies the target machine for compilation.
5533 The value to use for @var{machine} is the same as was specified as the
5534 machine type when configuring GCC as a cross-compiler. For
5535 example, if a cross-compiler was configured with @samp{configure
5536 i386v}, meaning to compile for an 80386 running System V, then you
5537 would specify @option{-b i386v} to run that cross compiler.
5539 @item -V @var{version}
5541 The argument @var{version} specifies which version of GCC to run.
5542 This is useful when multiple versions are installed. For example,
5543 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5546 The @option{-V} and @option{-b} options work by running the
5547 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5548 use them if you can just run that directly.
5550 @node Submodel Options
5551 @section Hardware Models and Configurations
5552 @cindex submodel options
5553 @cindex specifying hardware config
5554 @cindex hardware models and configurations, specifying
5555 @cindex machine dependent options
5557 Earlier we discussed the standard option @option{-b} which chooses among
5558 different installed compilers for completely different target
5559 machines, such as VAX vs.@: 68000 vs.@: 80386.
5561 In addition, each of these target machine types can have its own
5562 special options, starting with @samp{-m}, to choose among various
5563 hardware models or configurations---for example, 68010 vs 68020,
5564 floating coprocessor or none. A single installed version of the
5565 compiler can compile for any model or configuration, according to the
5568 Some configurations of the compiler also support additional special
5569 options, usually for compatibility with other compilers on the same
5572 These options are defined by the macro @code{TARGET_SWITCHES} in the
5573 machine description. The default for the options is also defined by
5574 that macro, which enables you to change the defaults.
5586 * RS/6000 and PowerPC Options::
5590 * i386 and x86-64 Options::
5592 * Intel 960 Options::
5593 * DEC Alpha Options::
5594 * DEC Alpha/VMS Options::
5597 * System V Options::
5598 * TMS320C3x/C4x Options::
5606 * S/390 and zSeries Options::
5610 * Xstormy16 Options::
5615 @node M680x0 Options
5616 @subsection M680x0 Options
5617 @cindex M680x0 options
5619 These are the @samp{-m} options defined for the 68000 series. The default
5620 values for these options depends on which style of 68000 was selected when
5621 the compiler was configured; the defaults for the most common choices are
5629 Generate output for a 68000. This is the default
5630 when the compiler is configured for 68000-based systems.
5632 Use this option for microcontrollers with a 68000 or EC000 core,
5633 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5639 Generate output for a 68020. This is the default
5640 when the compiler is configured for 68020-based systems.
5644 Generate output containing 68881 instructions for floating point.
5645 This is the default for most 68020 systems unless @option{--nfp} was
5646 specified when the compiler was configured.
5650 Generate output for a 68030. This is the default when the compiler is
5651 configured for 68030-based systems.
5655 Generate output for a 68040. This is the default when the compiler is
5656 configured for 68040-based systems.
5658 This option inhibits the use of 68881/68882 instructions that have to be
5659 emulated by software on the 68040. Use this option if your 68040 does not
5660 have code to emulate those instructions.
5664 Generate output for a 68060. This is the default when the compiler is
5665 configured for 68060-based systems.
5667 This option inhibits the use of 68020 and 68881/68882 instructions that
5668 have to be emulated by software on the 68060. Use this option if your 68060
5669 does not have code to emulate those instructions.
5673 Generate output for a CPU32. This is the default
5674 when the compiler is configured for CPU32-based systems.
5676 Use this option for microcontrollers with a
5677 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5678 68336, 68340, 68341, 68349 and 68360.
5682 Generate output for a 520X ``coldfire'' family cpu. This is the default
5683 when the compiler is configured for 520X-based systems.
5685 Use this option for microcontroller with a 5200 core, including
5686 the MCF5202, MCF5203, MCF5204 and MCF5202.
5691 Generate output for a 68040, without using any of the new instructions.
5692 This results in code which can run relatively efficiently on either a
5693 68020/68881 or a 68030 or a 68040. The generated code does use the
5694 68881 instructions that are emulated on the 68040.
5698 Generate output for a 68060, without using any of the new instructions.
5699 This results in code which can run relatively efficiently on either a
5700 68020/68881 or a 68030 or a 68040. The generated code does use the
5701 68881 instructions that are emulated on the 68060.
5704 @opindex msoft-float
5705 Generate output containing library calls for floating point.
5706 @strong{Warning:} the requisite libraries are not available for all m68k
5707 targets. Normally the facilities of the machine's usual C compiler are
5708 used, but this can't be done directly in cross-compilation. You must
5709 make your own arrangements to provide suitable library functions for
5710 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5711 @samp{m68k-*-coff} do provide software floating point support.
5715 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5718 @opindex mnobitfield
5719 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5720 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5724 Do use the bit-field instructions. The @option{-m68020} option implies
5725 @option{-mbitfield}. This is the default if you use a configuration
5726 designed for a 68020.
5730 Use a different function-calling convention, in which functions
5731 that take a fixed number of arguments return with the @code{rtd}
5732 instruction, which pops their arguments while returning. This
5733 saves one instruction in the caller since there is no need to pop
5734 the arguments there.
5736 This calling convention is incompatible with the one normally
5737 used on Unix, so you cannot use it if you need to call libraries
5738 compiled with the Unix compiler.
5740 Also, you must provide function prototypes for all functions that
5741 take variable numbers of arguments (including @code{printf});
5742 otherwise incorrect code will be generated for calls to those
5745 In addition, seriously incorrect code will result if you call a
5746 function with too many arguments. (Normally, extra arguments are
5747 harmlessly ignored.)
5749 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5750 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5753 @itemx -mno-align-int
5755 @opindex mno-align-int
5756 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5757 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5758 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
5759 Aligning variables on 32-bit boundaries produces code that runs somewhat
5760 faster on processors with 32-bit busses at the expense of more memory.
5762 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
5763 align structures containing the above types differently than
5764 most published application binary interface specifications for the m68k.
5768 Use the pc-relative addressing mode of the 68000 directly, instead of
5769 using a global offset table. At present, this option implies @option{-fpic},
5770 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
5771 not presently supported with @option{-mpcrel}, though this could be supported for
5772 68020 and higher processors.
5774 @item -mno-strict-align
5775 @itemx -mstrict-align
5776 @opindex mno-strict-align
5777 @opindex mstrict-align
5778 Do not (do) assume that unaligned memory references will be handled by
5783 @node M68hc1x Options
5784 @subsection M68hc1x Options
5785 @cindex M68hc1x options
5787 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5788 microcontrollers. The default values for these options depends on
5789 which style of microcontroller was selected when the compiler was configured;
5790 the defaults for the most common choices are given below.
5797 Generate output for a 68HC11. This is the default
5798 when the compiler is configured for 68HC11-based systems.
5804 Generate output for a 68HC12. This is the default
5805 when the compiler is configured for 68HC12-based systems.
5811 Generate output for a 68HCS12.
5814 @opindex mauto-incdec
5815 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5822 Enable the use of 68HC12 min and max instructions.
5825 @itemx -mno-long-calls
5826 @opindex mlong-calls
5827 @opindex mno-long-calls
5828 Treat all calls as being far away (near). If calls are assumed to be
5829 far away, the compiler will use the @code{call} instruction to
5830 call a function and the @code{rtc} instruction for returning.
5834 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5836 @item -msoft-reg-count=@var{count}
5837 @opindex msoft-reg-count
5838 Specify the number of pseudo-soft registers which are used for the
5839 code generation. The maximum number is 32. Using more pseudo-soft
5840 register may or may not result in better code depending on the program.
5841 The default is 4 for 68HC11 and 2 for 68HC12.
5846 @subsection VAX Options
5849 These @samp{-m} options are defined for the VAX:
5854 Do not output certain jump instructions (@code{aobleq} and so on)
5855 that the Unix assembler for the VAX cannot handle across long
5860 Do output those jump instructions, on the assumption that you
5861 will assemble with the GNU assembler.
5865 Output code for g-format floating point numbers instead of d-format.
5869 @subsection SPARC Options
5870 @cindex SPARC options
5872 These @samp{-m} switches are supported on the SPARC:
5877 @opindex mno-app-regs
5879 Specify @option{-mapp-regs} to generate output using the global registers
5880 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
5883 To be fully SVR4 ABI compliant at the cost of some performance loss,
5884 specify @option{-mno-app-regs}. You should compile libraries and system
5885 software with this option.
5890 @opindex mhard-float
5891 Generate output containing floating point instructions. This is the
5897 @opindex msoft-float
5898 Generate output containing library calls for floating point.
5899 @strong{Warning:} the requisite libraries are not available for all SPARC
5900 targets. Normally the facilities of the machine's usual C compiler are
5901 used, but this cannot be done directly in cross-compilation. You must make
5902 your own arrangements to provide suitable library functions for
5903 cross-compilation. The embedded targets @samp{sparc-*-aout} and
5904 @samp{sparclite-*-*} do provide software floating point support.
5906 @option{-msoft-float} changes the calling convention in the output file;
5907 therefore, it is only useful if you compile @emph{all} of a program with
5908 this option. In particular, you need to compile @file{libgcc.a}, the
5909 library that comes with GCC, with @option{-msoft-float} in order for
5912 @item -mhard-quad-float
5913 @opindex mhard-quad-float
5914 Generate output containing quad-word (long double) floating point
5917 @item -msoft-quad-float
5918 @opindex msoft-quad-float
5919 Generate output containing library calls for quad-word (long double)
5920 floating point instructions. The functions called are those specified
5921 in the SPARC ABI@. This is the default.
5923 As of this writing, there are no sparc implementations that have hardware
5924 support for the quad-word floating point instructions. They all invoke
5925 a trap handler for one of these instructions, and then the trap handler
5926 emulates the effect of the instruction. Because of the trap handler overhead,
5927 this is much slower than calling the ABI library routines. Thus the
5928 @option{-msoft-quad-float} option is the default.
5934 With @option{-mflat}, the compiler does not generate save/restore instructions
5935 and will use a ``flat'' or single register window calling convention.
5936 This model uses %i7 as the frame pointer and is compatible with the normal
5937 register window model. Code from either may be intermixed.
5938 The local registers and the input registers (0--5) are still treated as
5939 ``call saved'' registers and will be saved on the stack as necessary.
5941 With @option{-mno-flat} (the default), the compiler emits save/restore
5942 instructions (except for leaf functions) and is the normal mode of operation.
5944 @item -mno-unaligned-doubles
5945 @itemx -munaligned-doubles
5946 @opindex mno-unaligned-doubles
5947 @opindex munaligned-doubles
5948 Assume that doubles have 8 byte alignment. This is the default.
5950 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
5951 alignment only if they are contained in another type, or if they have an
5952 absolute address. Otherwise, it assumes they have 4 byte alignment.
5953 Specifying this option avoids some rare compatibility problems with code
5954 generated by other compilers. It is not the default because it results
5955 in a performance loss, especially for floating point code.
5957 @item -mno-faster-structs
5958 @itemx -mfaster-structs
5959 @opindex mno-faster-structs
5960 @opindex mfaster-structs
5961 With @option{-mfaster-structs}, the compiler assumes that structures
5962 should have 8 byte alignment. This enables the use of pairs of
5963 @code{ldd} and @code{std} instructions for copies in structure
5964 assignment, in place of twice as many @code{ld} and @code{st} pairs.
5965 However, the use of this changed alignment directly violates the SPARC
5966 ABI@. Thus, it's intended only for use on targets where the developer
5967 acknowledges that their resulting code will not be directly in line with
5968 the rules of the ABI@.
5974 These two options select variations on the SPARC architecture.
5976 By default (unless specifically configured for the Fujitsu SPARClite),
5977 GCC generates code for the v7 variant of the SPARC architecture.
5979 @option{-mv8} will give you SPARC v8 code. The only difference from v7
5980 code is that the compiler emits the integer multiply and integer
5981 divide instructions which exist in SPARC v8 but not in SPARC v7.
5983 @option{-msparclite} will give you SPARClite code. This adds the integer
5984 multiply, integer divide step and scan (@code{ffs}) instructions which
5985 exist in SPARClite but not in SPARC v7.
5987 These options are deprecated and will be deleted in a future GCC release.
5988 They have been replaced with @option{-mcpu=xxx}.
5993 @opindex msupersparc
5994 These two options select the processor for which the code is optimized.
5996 With @option{-mcypress} (the default), the compiler optimizes code for the
5997 Cypress CY7C602 chip, as used in the SPARCStation/SPARCServer 3xx series.
5998 This is also appropriate for the older SPARCStation 1, 2, IPX etc.
6000 With @option{-msupersparc} the compiler optimizes code for the SuperSPARC cpu, as
6001 used in the SPARCStation 10, 1000 and 2000 series. This flag also enables use
6002 of the full SPARC v8 instruction set.
6004 These options are deprecated and will be deleted in a future GCC release.
6005 They have been replaced with @option{-mcpu=xxx}.
6007 @item -mcpu=@var{cpu_type}
6009 Set the instruction set, register set, and instruction scheduling parameters
6010 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
6011 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
6012 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
6013 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
6016 Default instruction scheduling parameters are used for values that select
6017 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
6018 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
6020 Here is a list of each supported architecture and their supported
6025 v8: supersparc, hypersparc
6026 sparclite: f930, f934, sparclite86x
6028 v9: ultrasparc, ultrasparc3
6031 @item -mtune=@var{cpu_type}
6033 Set the instruction scheduling parameters for machine type
6034 @var{cpu_type}, but do not set the instruction set or register set that the
6035 option @option{-mcpu=@var{cpu_type}} would.
6037 The same values for @option{-mcpu=@var{cpu_type}} can be used for
6038 @option{-mtune=@var{cpu_type}}, but the only useful values are those
6039 that select a particular cpu implementation. Those are @samp{cypress},
6040 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
6041 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
6046 These @samp{-m} switches are supported in addition to the above
6047 on the SPARCLET processor.
6050 @item -mlittle-endian
6051 @opindex mlittle-endian
6052 Generate code for a processor running in little-endian mode.
6056 Treat register @code{%g0} as a normal register.
6057 GCC will continue to clobber it as necessary but will not assume
6058 it always reads as 0.
6060 @item -mbroken-saverestore
6061 @opindex mbroken-saverestore
6062 Generate code that does not use non-trivial forms of the @code{save} and
6063 @code{restore} instructions. Early versions of the SPARCLET processor do
6064 not correctly handle @code{save} and @code{restore} instructions used with
6065 arguments. They correctly handle them used without arguments. A @code{save}
6066 instruction used without arguments increments the current window pointer
6067 but does not allocate a new stack frame. It is assumed that the window
6068 overflow trap handler will properly handle this case as will interrupt
6072 These @samp{-m} switches are supported in addition to the above
6073 on SPARC V9 processors in 64-bit environments.
6076 @item -mlittle-endian
6077 @opindex mlittle-endian
6078 Generate code for a processor running in little-endian mode.
6084 Generate code for a 32-bit or 64-bit environment.
6085 The 32-bit environment sets int, long and pointer to 32 bits.
6086 The 64-bit environment sets int to 32 bits and long and pointer
6089 @item -mcmodel=medlow
6090 @opindex mcmodel=medlow
6091 Generate code for the Medium/Low code model: the program must be linked
6092 in the low 32 bits of the address space. Pointers are 64 bits.
6093 Programs can be statically or dynamically linked.
6095 @item -mcmodel=medmid
6096 @opindex mcmodel=medmid
6097 Generate code for the Medium/Middle code model: the program must be linked
6098 in the low 44 bits of the address space, the text segment must be less than
6099 2G bytes, and data segment must be within 2G of the text segment.
6100 Pointers are 64 bits.
6102 @item -mcmodel=medany
6103 @opindex mcmodel=medany
6104 Generate code for the Medium/Anywhere code model: the program may be linked
6105 anywhere in the address space, the text segment must be less than
6106 2G bytes, and data segment must be within 2G of the text segment.
6107 Pointers are 64 bits.
6109 @item -mcmodel=embmedany
6110 @opindex mcmodel=embmedany
6111 Generate code for the Medium/Anywhere code model for embedded systems:
6112 assume a 32-bit text and a 32-bit data segment, both starting anywhere
6113 (determined at link time). Register %g4 points to the base of the
6114 data segment. Pointers are still 64 bits.
6115 Programs are statically linked, PIC is not supported.
6118 @itemx -mno-stack-bias
6119 @opindex mstack-bias
6120 @opindex mno-stack-bias
6121 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
6122 frame pointer if present, are offset by @minus{}2047 which must be added back
6123 when making stack frame references.
6124 Otherwise, assume no such offset is present.
6128 @subsection ARM Options
6131 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6136 @opindex mapcs-frame
6137 Generate a stack frame that is compliant with the ARM Procedure Call
6138 Standard for all functions, even if this is not strictly necessary for
6139 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6140 with this option will cause the stack frames not to be generated for
6141 leaf functions. The default is @option{-mno-apcs-frame}.
6145 This is a synonym for @option{-mapcs-frame}.
6149 Generate code for a processor running with a 26-bit program counter,
6150 and conforming to the function calling standards for the APCS 26-bit
6151 option. This option replaces the @option{-m2} and @option{-m3} options
6152 of previous releases of the compiler.
6156 Generate code for a processor running with a 32-bit program counter,
6157 and conforming to the function calling standards for the APCS 32-bit
6158 option. This option replaces the @option{-m6} option of previous releases
6162 @c not currently implemented
6163 @item -mapcs-stack-check
6164 @opindex mapcs-stack-check
6165 Generate code to check the amount of stack space available upon entry to
6166 every function (that actually uses some stack space). If there is
6167 insufficient space available then either the function
6168 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6169 called, depending upon the amount of stack space required. The run time
6170 system is required to provide these functions. The default is
6171 @option{-mno-apcs-stack-check}, since this produces smaller code.
6173 @c not currently implemented
6175 @opindex mapcs-float
6176 Pass floating point arguments using the float point registers. This is
6177 one of the variants of the APCS@. This option is recommended if the
6178 target hardware has a floating point unit or if a lot of floating point
6179 arithmetic is going to be performed by the code. The default is
6180 @option{-mno-apcs-float}, since integer only code is slightly increased in
6181 size if @option{-mapcs-float} is used.
6183 @c not currently implemented
6184 @item -mapcs-reentrant
6185 @opindex mapcs-reentrant
6186 Generate reentrant, position independent code. The default is
6187 @option{-mno-apcs-reentrant}.
6190 @item -mthumb-interwork
6191 @opindex mthumb-interwork
6192 Generate code which supports calling between the ARM and Thumb
6193 instruction sets. Without this option the two instruction sets cannot
6194 be reliably used inside one program. The default is
6195 @option{-mno-thumb-interwork}, since slightly larger code is generated
6196 when @option{-mthumb-interwork} is specified.
6198 @item -mno-sched-prolog
6199 @opindex mno-sched-prolog
6200 Prevent the reordering of instructions in the function prolog, or the
6201 merging of those instruction with the instructions in the function's
6202 body. This means that all functions will start with a recognizable set
6203 of instructions (or in fact one of a choice from a small set of
6204 different function prologues), and this information can be used to
6205 locate the start if functions inside an executable piece of code. The
6206 default is @option{-msched-prolog}.
6209 @opindex mhard-float
6210 Generate output containing floating point instructions. This is the
6214 @opindex msoft-float
6215 Generate output containing library calls for floating point.
6216 @strong{Warning:} the requisite libraries are not available for all ARM
6217 targets. Normally the facilities of the machine's usual C compiler are
6218 used, but this cannot be done directly in cross-compilation. You must make
6219 your own arrangements to provide suitable library functions for
6222 @option{-msoft-float} changes the calling convention in the output file;
6223 therefore, it is only useful if you compile @emph{all} of a program with
6224 this option. In particular, you need to compile @file{libgcc.a}, the
6225 library that comes with GCC, with @option{-msoft-float} in order for
6228 @item -mlittle-endian
6229 @opindex mlittle-endian
6230 Generate code for a processor running in little-endian mode. This is
6231 the default for all standard configurations.
6234 @opindex mbig-endian
6235 Generate code for a processor running in big-endian mode; the default is
6236 to compile code for a little-endian processor.
6238 @item -mwords-little-endian
6239 @opindex mwords-little-endian
6240 This option only applies when generating code for big-endian processors.
6241 Generate code for a little-endian word order but a big-endian byte
6242 order. That is, a byte order of the form @samp{32107654}. Note: this
6243 option should only be used if you require compatibility with code for
6244 big-endian ARM processors generated by versions of the compiler prior to
6247 @item -malignment-traps
6248 @opindex malignment-traps
6249 Generate code that will not trap if the MMU has alignment traps enabled.
6250 On ARM architectures prior to ARMv4, there were no instructions to
6251 access half-word objects stored in memory. However, when reading from
6252 memory a feature of the ARM architecture allows a word load to be used,
6253 even if the address is unaligned, and the processor core will rotate the
6254 data as it is being loaded. This option tells the compiler that such
6255 misaligned accesses will cause a MMU trap and that it should instead
6256 synthesize the access as a series of byte accesses. The compiler can
6257 still use word accesses to load half-word data if it knows that the
6258 address is aligned to a word boundary.
6260 This option is ignored when compiling for ARM architecture 4 or later,
6261 since these processors have instructions to directly access half-word
6264 @item -mno-alignment-traps
6265 @opindex mno-alignment-traps
6266 Generate code that assumes that the MMU will not trap unaligned
6267 accesses. This produces better code when the target instruction set
6268 does not have half-word memory operations (i.e.@: implementations prior to
6271 Note that you cannot use this option to access unaligned word objects,
6272 since the processor will only fetch one 32-bit aligned object from
6275 The default setting for most targets is @option{-mno-alignment-traps}, since
6276 this produces better code when there are no half-word memory
6277 instructions available.
6279 @item -mshort-load-bytes
6280 @itemx -mno-short-load-words
6281 @opindex mshort-load-bytes
6282 @opindex mno-short-load-words
6283 These are deprecated aliases for @option{-malignment-traps}.
6285 @item -mno-short-load-bytes
6286 @itemx -mshort-load-words
6287 @opindex mno-short-load-bytes
6288 @opindex mshort-load-words
6289 This are deprecated aliases for @option{-mno-alignment-traps}.
6291 @item -mcpu=@var{name}
6293 This specifies the name of the target ARM processor. GCC uses this name
6294 to determine what kind of instructions it can emit when generating
6295 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6296 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6297 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6298 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6299 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6300 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6301 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6302 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6303 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
6304 @samp{arm1020t}, @samp{xscale}, @samp{ep9312}.
6306 @itemx -mtune=@var{name}
6308 This option is very similar to the @option{-mcpu=} option, except that
6309 instead of specifying the actual target processor type, and hence
6310 restricting which instructions can be used, it specifies that GCC should
6311 tune the performance of the code as if the target were of the type
6312 specified in this option, but still choosing the instructions that it
6313 will generate based on the cpu specified by a @option{-mcpu=} option.
6314 For some ARM implementations better performance can be obtained by using
6317 @item -march=@var{name}
6319 This specifies the name of the target ARM architecture. GCC uses this
6320 name to determine what kind of instructions it can emit when generating
6321 assembly code. This option can be used in conjunction with or instead
6322 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6323 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6324 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{ep9312}.
6326 @item -mfpe=@var{number}
6327 @itemx -mfp=@var{number}
6330 This specifies the version of the floating point emulation available on
6331 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6332 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6334 @item -mstructure-size-boundary=@var{n}
6335 @opindex mstructure-size-boundary
6336 The size of all structures and unions will be rounded up to a multiple
6337 of the number of bits set by this option. Permissible values are 8 and
6338 32. The default value varies for different toolchains. For the COFF
6339 targeted toolchain the default value is 8. Specifying the larger number
6340 can produce faster, more efficient code, but can also increase the size
6341 of the program. The two values are potentially incompatible. Code
6342 compiled with one value cannot necessarily expect to work with code or
6343 libraries compiled with the other value, if they exchange information
6344 using structures or unions.
6346 @item -mabort-on-noreturn
6347 @opindex mabort-on-noreturn
6348 Generate a call to the function @code{abort} at the end of a
6349 @code{noreturn} function. It will be executed if the function tries to
6353 @itemx -mno-long-calls
6354 @opindex mlong-calls
6355 @opindex mno-long-calls
6356 Tells the compiler to perform function calls by first loading the
6357 address of the function into a register and then performing a subroutine
6358 call on this register. This switch is needed if the target function
6359 will lie outside of the 64 megabyte addressing range of the offset based
6360 version of subroutine call instruction.
6362 Even if this switch is enabled, not all function calls will be turned
6363 into long calls. The heuristic is that static functions, functions
6364 which have the @samp{short-call} attribute, functions that are inside
6365 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6366 definitions have already been compiled within the current compilation
6367 unit, will not be turned into long calls. The exception to this rule is
6368 that weak function definitions, functions with the @samp{long-call}
6369 attribute or the @samp{section} attribute, and functions that are within
6370 the scope of a @samp{#pragma long_calls} directive, will always be
6371 turned into long calls.
6373 This feature is not enabled by default. Specifying
6374 @option{-mno-long-calls} will restore the default behavior, as will
6375 placing the function calls within the scope of a @samp{#pragma
6376 long_calls_off} directive. Note these switches have no effect on how
6377 the compiler generates code to handle function calls via function
6380 @item -mnop-fun-dllimport
6381 @opindex mnop-fun-dllimport
6382 Disable support for the @code{dllimport} attribute.
6384 @item -msingle-pic-base
6385 @opindex msingle-pic-base
6386 Treat the register used for PIC addressing as read-only, rather than
6387 loading it in the prologue for each function. The run-time system is
6388 responsible for initializing this register with an appropriate value
6389 before execution begins.
6391 @item -mpic-register=@var{reg}
6392 @opindex mpic-register
6393 Specify the register to be used for PIC addressing. The default is R10
6394 unless stack-checking is enabled, when R9 is used.
6396 @item -mcirrus-fix-invalid-insns
6397 @opindex -mcirrus-fix-invalid-insns
6398 @opindex -mno-cirrus-fix-invalid-insns
6399 Insert NOPs into the instruction stream to in order to work around
6400 problems with invalid Maverick instruction combinations. This option
6401 is only valid if the @option{-mcpu=ep9312} option has been used to
6402 enable generation of instructions for the Cirrus Maverick floating
6403 point co-processor. This option is not enabled by default, since the
6404 problem is only present in older Maverick implementations. The default
6405 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6408 @item -mpoke-function-name
6409 @opindex mpoke-function-name
6410 Write the name of each function into the text section, directly
6411 preceding the function prologue. The generated code is similar to this:
6415 .ascii "arm_poke_function_name", 0
6418 .word 0xff000000 + (t1 - t0)
6419 arm_poke_function_name
6421 stmfd sp!, @{fp, ip, lr, pc@}
6425 When performing a stack backtrace, code can inspect the value of
6426 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6427 location @code{pc - 12} and the top 8 bits are set, then we know that
6428 there is a function name embedded immediately preceding this location
6429 and has length @code{((pc[-3]) & 0xff000000)}.
6433 Generate code for the 16-bit Thumb instruction set. The default is to
6434 use the 32-bit ARM instruction set.
6437 @opindex mtpcs-frame
6438 Generate a stack frame that is compliant with the Thumb Procedure Call
6439 Standard for all non-leaf functions. (A leaf function is one that does
6440 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6442 @item -mtpcs-leaf-frame
6443 @opindex mtpcs-leaf-frame
6444 Generate a stack frame that is compliant with the Thumb Procedure Call
6445 Standard for all leaf functions. (A leaf function is one that does
6446 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6448 @item -mcallee-super-interworking
6449 @opindex mcallee-super-interworking
6450 Gives all externally visible functions in the file being compiled an ARM
6451 instruction set header which switches to Thumb mode before executing the
6452 rest of the function. This allows these functions to be called from
6453 non-interworking code.
6455 @item -mcaller-super-interworking
6456 @opindex mcaller-super-interworking
6457 Allows calls via function pointers (including virtual functions) to
6458 execute correctly regardless of whether the target code has been
6459 compiled for interworking or not. There is a small overhead in the cost
6460 of executing a function pointer if this option is enabled.
6464 @node MN10200 Options
6465 @subsection MN10200 Options
6466 @cindex MN10200 options
6468 These @option{-m} options are defined for Matsushita MN10200 architectures:
6473 Indicate to the linker that it should perform a relaxation optimization pass
6474 to shorten branches, calls and absolute memory addresses. This option only
6475 has an effect when used on the command line for the final link step.
6477 This option makes symbolic debugging impossible.
6480 @node MN10300 Options
6481 @subsection MN10300 Options
6482 @cindex MN10300 options
6484 These @option{-m} options are defined for Matsushita MN10300 architectures:
6489 Generate code to avoid bugs in the multiply instructions for the MN10300
6490 processors. This is the default.
6493 @opindex mno-mult-bug
6494 Do not generate code to avoid bugs in the multiply instructions for the
6499 Generate code which uses features specific to the AM33 processor.
6503 Do not generate code which uses features specific to the AM33 processor. This
6508 Do not link in the C run-time initialization object file.
6512 Indicate to the linker that it should perform a relaxation optimization pass
6513 to shorten branches, calls and absolute memory addresses. This option only
6514 has an effect when used on the command line for the final link step.
6516 This option makes symbolic debugging impossible.
6520 @node M32R/D Options
6521 @subsection M32R/D Options
6522 @cindex M32R/D options
6524 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6529 Generate code for the M32R/X@.
6533 Generate code for the M32R@. This is the default.
6535 @item -mcode-model=small
6536 @opindex mcode-model=small
6537 Assume all objects live in the lower 16MB of memory (so that their addresses
6538 can be loaded with the @code{ld24} instruction), and assume all subroutines
6539 are reachable with the @code{bl} instruction.
6540 This is the default.
6542 The addressability of a particular object can be set with the
6543 @code{model} attribute.
6545 @item -mcode-model=medium
6546 @opindex mcode-model=medium
6547 Assume objects may be anywhere in the 32-bit address space (the compiler
6548 will generate @code{seth/add3} instructions to load their addresses), and
6549 assume all subroutines are reachable with the @code{bl} instruction.
6551 @item -mcode-model=large
6552 @opindex mcode-model=large
6553 Assume objects may be anywhere in the 32-bit address space (the compiler
6554 will generate @code{seth/add3} instructions to load their addresses), and
6555 assume subroutines may not be reachable with the @code{bl} instruction
6556 (the compiler will generate the much slower @code{seth/add3/jl}
6557 instruction sequence).
6560 @opindex msdata=none
6561 Disable use of the small data area. Variables will be put into
6562 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6563 @code{section} attribute has been specified).
6564 This is the default.
6566 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6567 Objects may be explicitly put in the small data area with the
6568 @code{section} attribute using one of these sections.
6571 @opindex msdata=sdata
6572 Put small global and static data in the small data area, but do not
6573 generate special code to reference them.
6577 Put small global and static data in the small data area, and generate
6578 special instructions to reference them.
6582 @cindex smaller data references
6583 Put global and static objects less than or equal to @var{num} bytes
6584 into the small data or bss sections instead of the normal data or bss
6585 sections. The default value of @var{num} is 8.
6586 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6587 for this option to have any effect.
6589 All modules should be compiled with the same @option{-G @var{num}} value.
6590 Compiling with different values of @var{num} may or may not work; if it
6591 doesn't the linker will give an error message---incorrect code will not be
6597 @subsection M88K Options
6598 @cindex M88k options
6600 These @samp{-m} options are defined for Motorola 88k architectures:
6605 Generate code that works well on both the m88100 and the
6610 Generate code that works best for the m88100, but that also
6615 Generate code that works best for the m88110, and may not run
6620 Obsolete option to be removed from the next revision.
6623 @item -midentify-revision
6624 @opindex midentify-revision
6625 @cindex identifying source, compiler (88k)
6626 Include an @code{ident} directive in the assembler output recording the
6627 source file name, compiler name and version, timestamp, and compilation
6630 @item -mno-underscores
6631 @opindex mno-underscores
6632 @cindex underscores, avoiding (88k)
6633 In assembler output, emit symbol names without adding an underscore
6634 character at the beginning of each name. The default is to use an
6635 underscore as prefix on each name.
6637 @item -mocs-debug-info
6638 @itemx -mno-ocs-debug-info
6639 @opindex mocs-debug-info
6640 @opindex mno-ocs-debug-info
6642 @cindex debugging, 88k OCS
6643 Include (or omit) additional debugging information (about registers used
6644 in each stack frame) as specified in the 88open Object Compatibility
6645 Standard, ``OCS''@. This extra information allows debugging of code that
6646 has had the frame pointer eliminated. The default for SVr4 and Delta 88
6647 SVr3.2 is to include this information; other 88k configurations omit this
6648 information by default.
6650 @item -mocs-frame-position
6651 @opindex mocs-frame-position
6652 @cindex register positions in frame (88k)
6653 When emitting COFF debugging information for automatic variables and
6654 parameters stored on the stack, use the offset from the canonical frame
6655 address, which is the stack pointer (register 31) on entry to the
6656 function. The SVr4 and Delta88 SVr3.2, and BCS configurations use
6657 @option{-mocs-frame-position}; other 88k configurations have the default
6658 @option{-mno-ocs-frame-position}.
6660 @item -mno-ocs-frame-position
6661 @opindex mno-ocs-frame-position
6662 @cindex register positions in frame (88k)
6663 When emitting COFF debugging information for automatic variables and
6664 parameters stored on the stack, use the offset from the frame pointer
6665 register (register 30). When this option is in effect, the frame
6666 pointer is not eliminated when debugging information is selected by the
6669 @item -moptimize-arg-area
6670 @opindex moptimize-arg-area
6671 @cindex arguments in frame (88k)
6672 Save space by reorganizing the stack frame. This option generates code
6673 that does not agree with the 88open specifications, but uses less
6676 @itemx -mno-optimize-arg-area
6677 @opindex mno-optimize-arg-area
6678 Do not reorganize the stack frame to save space. This is the default.
6679 The generated conforms to the specification, but uses more memory.
6681 @item -mshort-data-@var{num}
6682 @opindex mshort-data
6683 @cindex smaller data references (88k)
6684 @cindex r0-relative references (88k)
6685 Generate smaller data references by making them relative to @code{r0},
6686 which allows loading a value using a single instruction (rather than the
6687 usual two). You control which data references are affected by
6688 specifying @var{num} with this option. For example, if you specify
6689 @option{-mshort-data-512}, then the data references affected are those
6690 involving displacements of less than 512 bytes.
6691 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6694 @item -mserialize-volatile
6695 @opindex mserialize-volatile
6696 @itemx -mno-serialize-volatile
6697 @opindex mno-serialize-volatile
6698 @cindex sequential consistency on 88k
6699 Do, or don't, generate code to guarantee sequential consistency
6700 of volatile memory references. By default, consistency is
6703 The order of memory references made by the MC88110 processor does
6704 not always match the order of the instructions requesting those
6705 references. In particular, a load instruction may execute before
6706 a preceding store instruction. Such reordering violates
6707 sequential consistency of volatile memory references, when there
6708 are multiple processors. When consistency must be guaranteed,
6709 GCC generates special instructions, as needed, to force
6710 execution in the proper order.
6712 The MC88100 processor does not reorder memory references and so
6713 always provides sequential consistency. However, by default, GCC
6714 generates the special instructions to guarantee consistency
6715 even when you use @option{-m88100}, so that the code may be run on an
6716 MC88110 processor. If you intend to run your code only on the
6717 MC88100 processor, you may use @option{-mno-serialize-volatile}.
6719 The extra code generated to guarantee consistency may affect the
6720 performance of your application. If you know that you can safely
6721 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
6727 @cindex assembler syntax, 88k
6729 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
6730 related to System V release 4 (SVr4). This controls the following:
6734 Which variant of the assembler syntax to emit.
6736 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6737 that is used on System V release 4.
6739 @option{-msvr4} makes GCC issue additional declaration directives used in
6743 @option{-msvr4} is the default for the m88k-motorola-sysv4 configuration.
6744 @option{-msvr3} is the default for all other m88k configurations.
6746 @item -mversion-03.00
6747 @opindex mversion-03.00
6748 This option is obsolete, and is ignored.
6749 @c ??? which asm syntax better for GAS? option there too?
6751 @item -mno-check-zero-division
6752 @itemx -mcheck-zero-division
6753 @opindex mno-check-zero-division
6754 @opindex mcheck-zero-division
6755 @cindex zero division on 88k
6756 Do, or don't, generate code to guarantee that integer division by
6757 zero will be detected. By default, detection is guaranteed.
6759 Some models of the MC88100 processor fail to trap upon integer
6760 division by zero under certain conditions. By default, when
6761 compiling code that might be run on such a processor, GCC
6762 generates code that explicitly checks for zero-valued divisors
6763 and traps with exception number 503 when one is detected. Use of
6764 @option{-mno-check-zero-division} suppresses such checking for code
6765 generated to run on an MC88100 processor.
6767 GCC assumes that the MC88110 processor correctly detects all instances
6768 of integer division by zero. When @option{-m88110} is specified, no
6769 explicit checks for zero-valued divisors are generated, and both
6770 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
6773 @item -muse-div-instruction
6774 @opindex muse-div-instruction
6775 @cindex divide instruction, 88k
6776 Use the div instruction for signed integer division on the
6777 MC88100 processor. By default, the div instruction is not used.
6779 On the MC88100 processor the signed integer division instruction
6780 div) traps to the operating system on a negative operand. The
6781 operating system transparently completes the operation, but at a
6782 large cost in execution time. By default, when compiling code
6783 that might be run on an MC88100 processor, GCC emulates signed
6784 integer division using the unsigned integer division instruction
6785 divu), thereby avoiding the large penalty of a trap to the
6786 operating system. Such emulation has its own, smaller, execution
6787 cost in both time and space. To the extent that your code's
6788 important signed integer division operations are performed on two
6789 nonnegative operands, it may be desirable to use the div
6790 instruction directly.
6792 On the MC88110 processor the div instruction (also known as the
6793 divs instruction) processes negative operands without trapping to
6794 the operating system. When @option{-m88110} is specified,
6795 @option{-muse-div-instruction} is ignored, and the div instruction is used
6796 for signed integer division.
6798 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
6799 particular, the behavior of such a division with and without
6800 @option{-muse-div-instruction} may differ.
6802 @item -mtrap-large-shift
6803 @itemx -mhandle-large-shift
6804 @opindex mtrap-large-shift
6805 @opindex mhandle-large-shift
6806 @cindex bit shift overflow (88k)
6807 @cindex large bit shifts (88k)
6808 Include code to detect bit-shifts of more than 31 bits; respectively,
6809 trap such shifts or emit code to handle them properly. By default GCC
6810 makes no special provision for large bit shifts.
6812 @item -mwarn-passed-structs
6813 @opindex mwarn-passed-structs
6814 @cindex structure passing (88k)
6815 Warn when a function passes a struct as an argument or result.
6816 Structure-passing conventions have changed during the evolution of the C
6817 language, and are often the source of portability problems. By default,
6818 GCC issues no such warning.
6821 @c break page here to avoid unsightly interparagraph stretch.
6825 @node RS/6000 and PowerPC Options
6826 @subsection IBM RS/6000 and PowerPC Options
6827 @cindex RS/6000 and PowerPC Options
6828 @cindex IBM RS/6000 and PowerPC Options
6830 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6838 @itemx -mpowerpc-gpopt
6839 @itemx -mno-powerpc-gpopt
6840 @itemx -mpowerpc-gfxopt
6841 @itemx -mno-powerpc-gfxopt
6843 @itemx -mno-powerpc64
6849 @opindex mno-powerpc
6850 @opindex mpowerpc-gpopt
6851 @opindex mno-powerpc-gpopt
6852 @opindex mpowerpc-gfxopt
6853 @opindex mno-powerpc-gfxopt
6855 @opindex mno-powerpc64
6856 GCC supports two related instruction set architectures for the
6857 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6858 instructions supported by the @samp{rios} chip set used in the original
6859 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6860 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6861 the IBM 4xx microprocessors.
6863 Neither architecture is a subset of the other. However there is a
6864 large common subset of instructions supported by both. An MQ
6865 register is included in processors supporting the POWER architecture.
6867 You use these options to specify which instructions are available on the
6868 processor you are using. The default value of these options is
6869 determined when configuring GCC@. Specifying the
6870 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6871 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6872 rather than the options listed above.
6874 The @option{-mpower} option allows GCC to generate instructions that
6875 are found only in the POWER architecture and to use the MQ register.
6876 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
6877 to generate instructions that are present in the POWER2 architecture but
6878 not the original POWER architecture.
6880 The @option{-mpowerpc} option allows GCC to generate instructions that
6881 are found only in the 32-bit subset of the PowerPC architecture.
6882 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
6883 GCC to use the optional PowerPC architecture instructions in the
6884 General Purpose group, including floating-point square root. Specifying
6885 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
6886 use the optional PowerPC architecture instructions in the Graphics
6887 group, including floating-point select.
6889 The @option{-mpowerpc64} option allows GCC to generate the additional
6890 64-bit instructions that are found in the full PowerPC64 architecture
6891 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6892 @option{-mno-powerpc64}.
6894 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
6895 will use only the instructions in the common subset of both
6896 architectures plus some special AIX common-mode calls, and will not use
6897 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
6898 permits GCC to use any instruction from either architecture and to
6899 allow use of the MQ register; specify this for the Motorola MPC601.
6901 @item -mnew-mnemonics
6902 @itemx -mold-mnemonics
6903 @opindex mnew-mnemonics
6904 @opindex mold-mnemonics
6905 Select which mnemonics to use in the generated assembler code. With
6906 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
6907 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
6908 assembler mnemonics defined for the POWER architecture. Instructions
6909 defined in only one architecture have only one mnemonic; GCC uses that
6910 mnemonic irrespective of which of these options is specified.
6912 GCC defaults to the mnemonics appropriate for the architecture in
6913 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
6914 value of these option. Unless you are building a cross-compiler, you
6915 should normally not specify either @option{-mnew-mnemonics} or
6916 @option{-mold-mnemonics}, but should instead accept the default.
6918 @item -mcpu=@var{cpu_type}
6920 Set architecture type, register usage, choice of mnemonics, and
6921 instruction scheduling parameters for machine type @var{cpu_type}.
6922 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
6923 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
6924 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
6925 @samp{630}, @samp{740}, @samp{7400}, @samp{7450}, @samp{750},
6926 @samp{power}, @samp{power2}, @samp{powerpc}, @samp{403}, @samp{505},
6927 @samp{801}, @samp{821}, @samp{823}, and @samp{860} and @samp{common}.
6929 @option{-mcpu=common} selects a completely generic processor. Code
6930 generated under this option will run on any POWER or PowerPC processor.
6931 GCC will use only the instructions in the common subset of both
6932 architectures, and will not use the MQ register. GCC assumes a generic
6933 processor model for scheduling purposes.
6935 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
6936 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
6937 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
6938 types, with an appropriate, generic processor model assumed for
6939 scheduling purposes.
6941 The other options specify a specific processor. Code generated under
6942 those options will run best on that processor, and may not run at all on
6945 The @option{-mcpu} options automatically enable or disable other
6946 @option{-m} options as follows:
6950 @option{-mno-power}, @option{-mno-powerpc}
6957 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
6972 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6975 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6980 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
6983 @item -mtune=@var{cpu_type}
6985 Set the instruction scheduling parameters for machine type
6986 @var{cpu_type}, but do not set the architecture type, register usage, or
6987 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
6988 values for @var{cpu_type} are used for @option{-mtune} as for
6989 @option{-mcpu}. If both are specified, the code generated will use the
6990 architecture, registers, and mnemonics set by @option{-mcpu}, but the
6991 scheduling parameters set by @option{-mtune}.
6996 @opindex mno-altivec
6997 These switches enable or disable the use of built-in functions that
6998 allow access to the AltiVec instruction set. You may also need to set
6999 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
7004 Extend the current ABI with SPE ABI extensions. This does not change
7005 the default ABI, instead it adds the SPE ABI extensions to the current
7009 @opindex mabi=no-spe
7010 Disable Booke SPE ABI extensions for the current ABI.
7012 @item -misel=@var{yes/no}
7015 This switch enables or disables the generation of ISEL instructions.
7017 @item -mspe=@var{yes/no}
7020 This switch enables or disables the generation of SPE simd
7023 @item -mfloat-gprs=@var{yes/no}
7025 @opindex mfloat-gprs
7026 This switch enables or disables the generation of floating point
7027 operations on the general purpose registers for architectures that
7028 support it. This option is currently only available on the MPC8540.
7031 @itemx -mno-fp-in-toc
7032 @itemx -mno-sum-in-toc
7033 @itemx -mminimal-toc
7035 @opindex mno-fp-in-toc
7036 @opindex mno-sum-in-toc
7037 @opindex mminimal-toc
7038 Modify generation of the TOC (Table Of Contents), which is created for
7039 every executable file. The @option{-mfull-toc} option is selected by
7040 default. In that case, GCC will allocate at least one TOC entry for
7041 each unique non-automatic variable reference in your program. GCC
7042 will also place floating-point constants in the TOC@. However, only
7043 16,384 entries are available in the TOC@.
7045 If you receive a linker error message that saying you have overflowed
7046 the available TOC space, you can reduce the amount of TOC space used
7047 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
7048 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
7049 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
7050 generate code to calculate the sum of an address and a constant at
7051 run-time instead of putting that sum into the TOC@. You may specify one
7052 or both of these options. Each causes GCC to produce very slightly
7053 slower and larger code at the expense of conserving TOC space.
7055 If you still run out of space in the TOC even when you specify both of
7056 these options, specify @option{-mminimal-toc} instead. This option causes
7057 GCC to make only one TOC entry for every file. When you specify this
7058 option, GCC will produce code that is slower and larger but which
7059 uses extremely little TOC space. You may wish to use this option
7060 only on files that contain less frequently executed code.
7066 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
7067 @code{long} type, and the infrastructure needed to support them.
7068 Specifying @option{-maix64} implies @option{-mpowerpc64} and
7069 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
7070 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
7075 @opindex mno-xl-call
7076 On AIX, pass floating-point arguments to prototyped functions beyond the
7077 register save area (RSA) on the stack in addition to argument FPRs. The
7078 AIX calling convention was extended but not initially documented to
7079 handle an obscure K&R C case of calling a function that takes the
7080 address of its arguments with fewer arguments than declared. AIX XL
7081 compilers access floating point arguments which do not fit in the
7082 RSA from the stack when a subroutine is compiled without
7083 optimization. Because always storing floating-point arguments on the
7084 stack is inefficient and rarely needed, this option is not enabled by
7085 default and only is necessary when calling subroutines compiled by AIX
7086 XL compilers without optimization.
7090 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
7091 application written to use message passing with special startup code to
7092 enable the application to run. The system must have PE installed in the
7093 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
7094 must be overridden with the @option{-specs=} option to specify the
7095 appropriate directory location. The Parallel Environment does not
7096 support threads, so the @option{-mpe} option and the @option{-pthread}
7097 option are incompatible.
7101 @opindex msoft-float
7102 @opindex mhard-float
7103 Generate code that does not use (uses) the floating-point register set.
7104 Software floating point emulation is provided if you use the
7105 @option{-msoft-float} option, and pass the option to GCC when linking.
7108 @itemx -mno-multiple
7110 @opindex mno-multiple
7111 Generate code that uses (does not use) the load multiple word
7112 instructions and the store multiple word instructions. These
7113 instructions are generated by default on POWER systems, and not
7114 generated on PowerPC systems. Do not use @option{-mmultiple} on little
7115 endian PowerPC systems, since those instructions do not work when the
7116 processor is in little endian mode. The exceptions are PPC740 and
7117 PPC750 which permit the instructions usage in little endian mode.
7123 Generate code that uses (does not use) the load string instructions
7124 and the store string word instructions to save multiple registers and
7125 do small block moves. These instructions are generated by default on
7126 POWER systems, and not generated on PowerPC systems. Do not use
7127 @option{-mstring} on little endian PowerPC systems, since those
7128 instructions do not work when the processor is in little endian mode.
7129 The exceptions are PPC740 and PPC750 which permit the instructions
7130 usage in little endian mode.
7136 Generate code that uses (does not use) the load or store instructions
7137 that update the base register to the address of the calculated memory
7138 location. These instructions are generated by default. If you use
7139 @option{-mno-update}, there is a small window between the time that the
7140 stack pointer is updated and the address of the previous frame is
7141 stored, which means code that walks the stack frame across interrupts or
7142 signals may get corrupted data.
7145 @itemx -mno-fused-madd
7146 @opindex mfused-madd
7147 @opindex mno-fused-madd
7148 Generate code that uses (does not use) the floating point multiply and
7149 accumulate instructions. These instructions are generated by default if
7150 hardware floating is used.
7152 @item -mno-bit-align
7154 @opindex mno-bit-align
7156 On System V.4 and embedded PowerPC systems do not (do) force structures
7157 and unions that contain bit-fields to be aligned to the base type of the
7160 For example, by default a structure containing nothing but 8
7161 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7162 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7163 the structure would be aligned to a 1 byte boundary and be one byte in
7166 @item -mno-strict-align
7167 @itemx -mstrict-align
7168 @opindex mno-strict-align
7169 @opindex mstrict-align
7170 On System V.4 and embedded PowerPC systems do not (do) assume that
7171 unaligned memory references will be handled by the system.
7174 @itemx -mno-relocatable
7175 @opindex mrelocatable
7176 @opindex mno-relocatable
7177 On embedded PowerPC systems generate code that allows (does not allow)
7178 the program to be relocated to a different address at runtime. If you
7179 use @option{-mrelocatable} on any module, all objects linked together must
7180 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7182 @item -mrelocatable-lib
7183 @itemx -mno-relocatable-lib
7184 @opindex mrelocatable-lib
7185 @opindex mno-relocatable-lib
7186 On embedded PowerPC systems generate code that allows (does not allow)
7187 the program to be relocated to a different address at runtime. Modules
7188 compiled with @option{-mrelocatable-lib} can be linked with either modules
7189 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7190 with modules compiled with the @option{-mrelocatable} options.
7196 On System V.4 and embedded PowerPC systems do not (do) assume that
7197 register 2 contains a pointer to a global area pointing to the addresses
7198 used in the program.
7201 @itemx -mlittle-endian
7203 @opindex mlittle-endian
7204 On System V.4 and embedded PowerPC systems compile code for the
7205 processor in little endian mode. The @option{-mlittle-endian} option is
7206 the same as @option{-mlittle}.
7211 @opindex mbig-endian
7212 On System V.4 and embedded PowerPC systems compile code for the
7213 processor in big endian mode. The @option{-mbig-endian} option is
7214 the same as @option{-mbig}.
7216 @item -mdynamic-no-pic
7217 @opindex mdynamic-no-pic
7218 On Darwin and Mac OS X systems, compile code so that it is not
7219 relocatable, but that its external references are relocatable. The
7220 resulting code is suitable for applications, but not shared
7225 On System V.4 and embedded PowerPC systems compile code using calling
7226 conventions that adheres to the March 1995 draft of the System V
7227 Application Binary Interface, PowerPC processor supplement. This is the
7228 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7230 @item -mcall-sysv-eabi
7231 @opindex mcall-sysv-eabi
7232 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7234 @item -mcall-sysv-noeabi
7235 @opindex mcall-sysv-noeabi
7236 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7238 @item -mcall-solaris
7239 @opindex mcall-solaris
7240 On System V.4 and embedded PowerPC systems compile code for the Solaris
7244 @opindex mcall-linux
7245 On System V.4 and embedded PowerPC systems compile code for the
7246 Linux-based GNU system.
7250 On System V.4 and embedded PowerPC systems compile code for the
7251 Hurd-based GNU system.
7254 @opindex mcall-netbsd
7255 On System V.4 and embedded PowerPC systems compile code for the
7256 NetBSD operating system.
7258 @item -maix-struct-return
7259 @opindex maix-struct-return
7260 Return all structures in memory (as specified by the AIX ABI)@.
7262 @item -msvr4-struct-return
7263 @opindex msvr4-struct-return
7264 Return structures smaller than 8 bytes in registers (as specified by the
7268 @opindex mabi=altivec
7269 Extend the current ABI with AltiVec ABI extensions. This does not
7270 change the default ABI, instead it adds the AltiVec ABI extensions to
7273 @item -mabi=no-altivec
7274 @opindex mabi=no-altivec
7275 Disable AltiVec ABI extensions for the current ABI.
7278 @itemx -mno-prototype
7280 @opindex mno-prototype
7281 On System V.4 and embedded PowerPC systems assume that all calls to
7282 variable argument functions are properly prototyped. Otherwise, the
7283 compiler must insert an instruction before every non prototyped call to
7284 set or clear bit 6 of the condition code register (@var{CR}) to
7285 indicate whether floating point values were passed in the floating point
7286 registers in case the function takes a variable arguments. With
7287 @option{-mprototype}, only calls to prototyped variable argument functions
7288 will set or clear the bit.
7292 On embedded PowerPC systems, assume that the startup module is called
7293 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7294 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7299 On embedded PowerPC systems, assume that the startup module is called
7300 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7305 On embedded PowerPC systems, assume that the startup module is called
7306 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7310 @opindex myellowknife
7311 On embedded PowerPC systems, assume that the startup module is called
7312 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7317 On System V.4 and embedded PowerPC systems, specify that you are
7318 compiling for a VxWorks system.
7322 Specify that you are compiling for the WindISS simulation environment.
7326 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7327 header to indicate that @samp{eabi} extended relocations are used.
7333 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7334 Embedded Applications Binary Interface (eabi) which is a set of
7335 modifications to the System V.4 specifications. Selecting @option{-meabi}
7336 means that the stack is aligned to an 8 byte boundary, a function
7337 @code{__eabi} is called to from @code{main} to set up the eabi
7338 environment, and the @option{-msdata} option can use both @code{r2} and
7339 @code{r13} to point to two separate small data areas. Selecting
7340 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7341 do not call an initialization function from @code{main}, and the
7342 @option{-msdata} option will only use @code{r13} to point to a single
7343 small data area. The @option{-meabi} option is on by default if you
7344 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7347 @opindex msdata=eabi
7348 On System V.4 and embedded PowerPC systems, put small initialized
7349 @code{const} global and static data in the @samp{.sdata2} section, which
7350 is pointed to by register @code{r2}. Put small initialized
7351 non-@code{const} global and static data in the @samp{.sdata} section,
7352 which is pointed to by register @code{r13}. Put small uninitialized
7353 global and static data in the @samp{.sbss} section, which is adjacent to
7354 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7355 incompatible with the @option{-mrelocatable} option. The
7356 @option{-msdata=eabi} option also sets the @option{-memb} option.
7359 @opindex msdata=sysv
7360 On System V.4 and embedded PowerPC systems, put small global and static
7361 data in the @samp{.sdata} section, which is pointed to by register
7362 @code{r13}. Put small uninitialized global and static data in the
7363 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7364 The @option{-msdata=sysv} option is incompatible with the
7365 @option{-mrelocatable} option.
7367 @item -msdata=default
7369 @opindex msdata=default
7371 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7372 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7373 same as @option{-msdata=sysv}.
7376 @opindex msdata-data
7377 On System V.4 and embedded PowerPC systems, put small global and static
7378 data in the @samp{.sdata} section. Put small uninitialized global and
7379 static data in the @samp{.sbss} section. Do not use register @code{r13}
7380 to address small data however. This is the default behavior unless
7381 other @option{-msdata} options are used.
7385 @opindex msdata=none
7387 On embedded PowerPC systems, put all initialized global and static data
7388 in the @samp{.data} section, and all uninitialized data in the
7389 @samp{.bss} section.
7393 @cindex smaller data references (PowerPC)
7394 @cindex .sdata/.sdata2 references (PowerPC)
7395 On embedded PowerPC systems, put global and static items less than or
7396 equal to @var{num} bytes into the small data or bss sections instead of
7397 the normal data or bss section. By default, @var{num} is 8. The
7398 @option{-G @var{num}} switch is also passed to the linker.
7399 All modules should be compiled with the same @option{-G @var{num}} value.
7402 @itemx -mno-regnames
7404 @opindex mno-regnames
7405 On System V.4 and embedded PowerPC systems do (do not) emit register
7406 names in the assembly language output using symbolic forms.
7409 @itemx -mno-longcall
7411 @opindex mno-longcall
7412 Default to making all function calls via pointers, so that functions
7413 which reside further than 64 megabytes (67,108,864 bytes) from the
7414 current location can be called. This setting can be overridden by the
7415 @code{shortcall} function attribute, or by @code{#pragma longcall(0)}.
7417 Some linkers are capable of detecting out-of-range calls and generating
7418 glue code on the fly. On these systems, long calls are unnecessary and
7419 generate slower code. As of this writing, the AIX linker can do this,
7420 as can the GNU linker for PowerPC/64. It is planned to add this feature
7421 to the GNU linker for 32-bit PowerPC systems as well.
7423 In the future, we may cause GCC to ignore all longcall specifications
7424 when the linker is known to generate glue.
7428 Adds support for multithreading with the @dfn{pthreads} library.
7429 This option sets flags for both the preprocessor and linker.
7433 @node Darwin Options
7434 @subsection Darwin Options
7435 @cindex Darwin options
7437 These options are defined for all architectures running the Darwin operating
7438 system. They are useful for compatibility with other Mac OS compilers.
7443 Loads all members of static archive libraries.
7444 See man ld(1) for more information.
7446 @item -arch_errors_fatal
7447 @opindex arch_errors_fatal
7448 Cause the errors having to do with files that have the wrong architecture
7452 @opindex bind_at_load
7453 Causes the output file to be marked such that the dynamic linker will
7454 bind all undefined references when the file is loaded or launched.
7458 Produce a Mach-o bundle format file.
7459 See man ld(1) for more information.
7461 @item -bundle_loader @var{executable}
7462 @opindex bundle_loader
7463 This specifies the @var{executable} that will be loading the build
7464 output file being linked. See man ld(1) for more information.
7466 @item -allowable_client @var{client_name}
7470 @item -compatibility_version
7471 @item -current_version
7472 @item -dependency-file
7474 @item -dylinker_install_name
7477 @item -exported_symbols_list
7479 @item -flat_namespace
7480 @item -force_cpusubtype_ALL
7481 @item -force_flat_namespace
7482 @item -headerpad_max_install_names
7486 @item -keep_private_externs
7488 @item -multiply_defined
7489 @item -multiply_defined_unused
7493 @item -noseglinkedit
7494 @item -pagezero_size
7496 @item -prebind_all_twolevel_modules
7497 @item -private_bundle
7498 @item -read_only_relocs
7500 @item -sectobjectsymbols
7504 @item -sectobjectsymbols
7506 @item -seg_addr_table
7507 @item -seg_addr_table_filename
7510 @item -segs_read_only_addr
7511 @item -segs_read_write_addr
7512 @item -single_module
7516 @item -twolevel_namespace
7519 @item -unexported_symbols_list
7520 @item -weak_reference_mismatches
7523 @opindex allowable_client
7525 @opindex client_name
7526 @opindex compatibility_version
7527 @opindex current_version
7528 @opindex dependency-file
7530 @opindex dylinker_install_name
7533 @opindex exported_symbols_list
7535 @opindex flat_namespace
7536 @opindex force_cpusubtype_ALL
7537 @opindex force_flat_namespace
7538 @opindex headerpad_max_install_names
7541 @opindex install_name
7542 @opindex keep_private_externs
7543 @opindex multi_module
7544 @opindex multiply_defined
7545 @opindex multiply_defined_unused
7547 @opindex nomultidefs
7549 @opindex noseglinkedit
7550 @opindex pagezero_size
7552 @opindex prebind_all_twolevel_modules
7553 @opindex private_bundle
7554 @opindex read_only_relocs
7556 @opindex sectobjectsymbols
7560 @opindex sectobjectsymbols
7562 @opindex seg_addr_table
7563 @opindex seg_addr_table_filename
7564 @opindex seglinkedit
7566 @opindex segs_read_only_addr
7567 @opindex segs_read_write_addr
7568 @opindex single_module
7570 @opindex sub_library
7571 @opindex sub_umbrella
7572 @opindex twolevel_namespace
7575 @opindex unexported_symbols_list
7576 @opindex weak_reference_mismatches
7577 @opindex whatsloaded
7579 This options are available for Darwin linker. Darwin linker man page
7580 describes them in detail.
7585 @subsection IBM RT Options
7587 @cindex IBM RT options
7589 These @samp{-m} options are defined for the IBM RT PC:
7593 @opindex min-line-mul
7594 Use an in-line code sequence for integer multiplies. This is the
7597 @item -mcall-lib-mul
7598 @opindex mcall-lib-mul
7599 Call @code{lmul$$} for integer multiples.
7601 @item -mfull-fp-blocks
7602 @opindex mfull-fp-blocks
7603 Generate full-size floating point data blocks, including the minimum
7604 amount of scratch space recommended by IBM@. This is the default.
7606 @item -mminimum-fp-blocks
7607 @opindex mminimum-fp-blocks
7608 Do not include extra scratch space in floating point data blocks. This
7609 results in smaller code, but slower execution, since scratch space must
7610 be allocated dynamically.
7612 @cindex @file{stdarg.h} and RT PC
7613 @item -mfp-arg-in-fpregs
7614 @opindex mfp-arg-in-fpregs
7615 Use a calling sequence incompatible with the IBM calling convention in
7616 which floating point arguments are passed in floating point registers.
7617 Note that @code{stdarg.h} will not work with floating point operands
7618 if this option is specified.
7620 @item -mfp-arg-in-gregs
7621 @opindex mfp-arg-in-gregs
7622 Use the normal calling convention for floating point arguments. This is
7625 @item -mhc-struct-return
7626 @opindex mhc-struct-return
7627 Return structures of more than one word in memory, rather than in a
7628 register. This provides compatibility with the MetaWare HighC (hc)
7629 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7630 with the Portable C Compiler (pcc).
7632 @item -mnohc-struct-return
7633 @opindex mnohc-struct-return
7634 Return some structures of more than one word in registers, when
7635 convenient. This is the default. For compatibility with the
7636 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7637 option @option{-mhc-struct-return}.
7641 @subsection MIPS Options
7642 @cindex MIPS options
7644 These @samp{-m} options are defined for the MIPS family of computers:
7648 @item -march=@var{arch}
7650 Generate code that will run on @var{arch}, which can be the name of a
7651 generic MIPS ISA, or the name of a particular processor.
7653 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
7654 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
7655 The processor names are:
7656 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
7658 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
7659 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
7662 @samp{vr4100}, @samp{vr4300}, and @samp{vr5000}.
7663 The special value @samp{from-abi} selects the
7664 most compatible architecture for the selected ABI (that is,
7665 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
7667 In processor names, a final @samp{000} can be abbreviated as @samp{k}
7668 (for example, @samp{-march=r2k}). Prefixes are optional, and
7669 @samp{vr} may be written @samp{r}.
7671 GCC defines two macros based on the value of this option. The first
7672 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
7673 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
7674 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
7675 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
7676 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
7678 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
7679 above. In other words, it will have the full prefix and will not
7680 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
7681 the macro names the resolved architecture (either @samp{"mips1"} or
7682 @samp{"mips3"}). It names the default architecture when no
7683 @option{-march} option is given.
7685 @item -mtune=@var{arch}
7687 Optimize for @var{arch}. Among other things, this option controls
7688 the way instructions are scheduled, and the perceived cost of arithmetic
7689 operations. The list of @var{arch} values is the same as for
7692 When this option is not used, GCC will optimize for the processor
7693 specified by @option{-march}. By using @option{-march} and
7694 @option{-mtune} together, it is possible to generate code that will
7695 run on a family of processors, but optimize the code for one
7696 particular member of that family.
7698 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
7699 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
7700 @samp{-march} ones described above.
7704 Equivalent to @samp{-march=mips1}.
7708 Equivalent to @samp{-march=mips2}.
7712 Equivalent to @samp{-march=mips3}.
7716 Equivalent to @samp{-march=mips4}.
7720 Equivalent to @samp{-march=mips32}.
7724 Equivalent to @samp{-march=mips32r2}.
7728 Equivalent to @samp{-march=mips64}.
7731 @itemx -mno-fused-madd
7732 @opindex mfused-madd
7733 @opindex mno-fused-madd
7734 Generate code that uses (does not use) the floating point multiply and
7735 accumulate instructions, when they are available. These instructions
7736 are generated by default if they are available, but this may be
7737 undesirable if the extra precision causes problems or on certain chips
7738 in the mode where denormals are rounded to zero where denormals
7739 generated by multiply and accumulate instructions cause exceptions
7744 Assume that floating point registers are 32 bits wide.
7748 Assume that floating point registers are 64 bits wide.
7752 Assume that general purpose registers are 32 bits wide.
7756 Assume that general purpose registers are 64 bits wide.
7760 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
7761 explanation of the default, and the width of pointers.
7765 Force long types to be 64 bits wide. See @option{-mlong32} for an
7766 explanation of the default, and the width of pointers.
7770 Force long, int, and pointer types to be 32 bits wide.
7772 The default size of ints, longs and pointers depends on the ABI@. All
7773 the supported ABIs use 32-bit ints. The n64 ABI uses 64-bit longs, as
7774 does the 64-bit Cygnus EABI; the others use 32-bit longs. Pointers
7775 are the same size as longs, or the same size as integer registers,
7776 whichever is smaller.
7790 Generate code for the given ABI@.
7792 Note that there are two embedded ABIs: @option{-mabi=eabi}
7793 selects the one defined by Cygnus while @option{-meabi=meabi}
7794 selects the one defined by MIPS@. Both these ABIs have
7795 32-bit and 64-bit variants. Normally, GCC will generate
7796 64-bit code when you select a 64-bit architecture, but you
7797 can use @option{-mgp32} to get 32-bit code instead.
7799 @item -mabi-fake-default
7800 @opindex mabi-fake-default
7801 You don't want to know what this option does. No, really. I mean
7802 it. Move on to the next option.
7804 What? You're still here? Oh, well@enddots{} Ok, here's the deal. GCC
7805 wants the default set of options to get the root of the multilib tree,
7806 and the shared library SONAMEs without any multilib-indicating
7807 suffixes. This is not convenience for @samp{mips64-linux-gnu}, since
7808 we want to default to the N32 ABI, while still being binary-compatible
7809 with @samp{mips-linux-gnu} if you stick to the O32 ABI@. Being
7810 binary-compatible means shared libraries should have the same SONAMEs,
7811 and libraries should live in the same location. Having O32 libraries
7812 in a sub-directory named say @file{o32} is not acceptable.
7814 So we trick GCC into believing that O32 is the default ABI, except
7815 that we override the default with some internal command-line
7816 processing magic. Problem is, if we stopped at that, and you then
7817 created a multilib-aware package that used the output of @command{gcc
7818 -print-multi-lib} to decide which multilibs to build, and how, and
7819 you'd find yourself in an awkward situation when you found out that
7820 some of the options listed ended up mapping to the same multilib, and
7821 none of your libraries was actually built for the multilib that
7822 @option{-print-multi-lib} claims to be the default. So we added this
7823 option that disables the default switcher, falling back to GCC's
7824 original notion of the default library. Confused yet?
7826 For short: don't ever use this option, unless you find it in the list
7827 of additional options to be used when building for multilibs, in the
7828 output of @option{gcc -print-multi-lib}.
7832 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
7833 add normal debug information. This is the default for all
7834 platforms except for the OSF/1 reference platform, using the OSF/rose
7835 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
7836 switches are used, the @file{mips-tfile} program will encapsulate the
7837 stabs within MIPS ECOFF@.
7841 Generate code for the GNU assembler. This is the default on the OSF/1
7842 reference platform, using the OSF/rose object format. Also, this is
7843 the default if the configure option @option{--with-gnu-as} is used.
7845 @item -msplit-addresses
7846 @itemx -mno-split-addresses
7847 @opindex msplit-addresses
7848 @opindex mno-split-addresses
7849 Generate code to load the high and low parts of address constants separately.
7850 This allows GCC to optimize away redundant loads of the high order
7851 bits of addresses. This optimization requires GNU as and GNU ld.
7852 This optimization is enabled by default for some embedded targets where
7853 GNU as and GNU ld are standard.
7859 The @option{-mrnames} switch says to output code using the MIPS software
7860 names for the registers, instead of the hardware names (ie, @var{a0}
7861 instead of @var{$4}). The only known assembler that supports this option
7862 is the Algorithmics assembler.
7868 The @option{-mgpopt} switch says to write all of the data declarations
7869 before the instructions in the text section, this allows the MIPS
7870 assembler to generate one word memory references instead of using two
7871 words for short global or static data items. This is on by default if
7872 optimization is selected.
7878 For each non-inline function processed, the @option{-mstats} switch
7879 causes the compiler to emit one line to the standard error file to
7880 print statistics about the program (number of registers saved, stack
7887 The @option{-mmemcpy} switch makes all block moves call the appropriate
7888 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
7889 generating inline code.
7892 @itemx -mno-mips-tfile
7893 @opindex mmips-tfile
7894 @opindex mno-mips-tfile
7895 The @option{-mno-mips-tfile} switch causes the compiler not
7896 postprocess the object file with the @file{mips-tfile} program,
7897 after the MIPS assembler has generated it to add debug support. If
7898 @file{mips-tfile} is not run, then no local variables will be
7899 available to the debugger. In addition, @file{stage2} and
7900 @file{stage3} objects will have the temporary file names passed to the
7901 assembler embedded in the object file, which means the objects will
7902 not compare the same. The @option{-mno-mips-tfile} switch should only
7903 be used when there are bugs in the @file{mips-tfile} program that
7904 prevents compilation.
7907 @opindex msoft-float
7908 Generate output containing library calls for floating point.
7909 @strong{Warning:} the requisite libraries are not part of GCC@.
7910 Normally the facilities of the machine's usual C compiler are used, but
7911 this can't be done directly in cross-compilation. You must make your
7912 own arrangements to provide suitable library functions for
7916 @opindex mhard-float
7917 Generate output containing floating point instructions. This is the
7918 default if you use the unmodified sources.
7921 @itemx -mno-abicalls
7923 @opindex mno-abicalls
7924 Emit (or do not emit) the pseudo operations @samp{.abicalls},
7925 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
7926 position independent code.
7929 @itemx -mno-long-calls
7930 @opindex mlong-calls
7931 @opindex mno-long-calls
7932 Do all calls with the @samp{JALR} instruction, which requires
7933 loading up a function's address into a register before the call.
7934 You need to use this switch, if you call outside of the current
7935 512 megabyte segment to functions that are not through pointers.
7938 @itemx -mno-half-pic
7940 @opindex mno-half-pic
7941 Put pointers to extern references into the data section and load them
7942 up, rather than put the references in the text section.
7944 @item -membedded-pic
7945 @itemx -mno-embedded-pic
7946 @opindex membedded-pic
7947 @opindex mno-embedded-pic
7948 Generate PIC code suitable for some embedded systems. All calls are
7949 made using PC relative address, and all data is addressed using the $gp
7950 register. No more than 65536 bytes of global data may be used. This
7951 requires GNU as and GNU ld which do most of the work. This currently
7952 only works on targets which use ECOFF; it does not work with ELF@.
7954 @item -membedded-data
7955 @itemx -mno-embedded-data
7956 @opindex membedded-data
7957 @opindex mno-embedded-data
7958 Allocate variables to the read-only data section first if possible, then
7959 next in the small data section if possible, otherwise in data. This gives
7960 slightly slower code than the default, but reduces the amount of RAM required
7961 when executing, and thus may be preferred for some embedded systems.
7963 @item -muninit-const-in-rodata
7964 @itemx -mno-uninit-const-in-rodata
7965 @opindex muninit-const-in-rodata
7966 @opindex mno-uninit-const-in-rodata
7967 When used together with @option{-membedded-data}, it will always store uninitialized
7968 const variables in the read-only data section.
7970 @item -msingle-float
7971 @itemx -mdouble-float
7972 @opindex msingle-float
7973 @opindex mdouble-float
7974 The @option{-msingle-float} switch tells gcc to assume that the floating
7975 point coprocessor only supports single precision operations, as on the
7976 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
7977 double precision operations. This is the default.
7983 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
7984 as on the @samp{r4650} chip.
7988 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
7989 @option{-mcpu=r4650}.
7995 Enable 16-bit instructions.
7999 Use the entry and exit pseudo ops. This option can only be used with
8004 Compile code for the processor in little endian mode.
8005 The requisite libraries are assumed to exist.
8009 Compile code for the processor in big endian mode.
8010 The requisite libraries are assumed to exist.
8014 @cindex smaller data references (MIPS)
8015 @cindex gp-relative references (MIPS)
8016 Put global and static items less than or equal to @var{num} bytes into
8017 the small data or bss sections instead of the normal data or bss
8018 section. This allows the assembler to emit one word memory reference
8019 instructions based on the global pointer (@var{gp} or @var{$28}),
8020 instead of the normal two words used. By default, @var{num} is 8 when
8021 the MIPS assembler is used, and 0 when the GNU assembler is used. The
8022 @option{-G @var{num}} switch is also passed to the assembler and linker.
8023 All modules should be compiled with the same @option{-G @var{num}}
8028 Tell the MIPS assembler to not run its preprocessor over user
8029 assembler files (with a @samp{.s} suffix) when assembling them.
8033 Pass an option to gas which will cause nops to be inserted if
8034 the read of the destination register of an mfhi or mflo instruction
8035 occurs in the following two instructions.
8039 Do not include the default crt0.
8041 @item -mflush-func=@var{func}
8042 @itemx -mno-flush-func
8043 @opindex mflush-func
8044 Specifies the function to call to flush the I and D caches, or to not
8045 call any such function. If called, the function must take the same
8046 arguments as the common @code{_flush_func()}, that is, the address of the
8047 memory range for which the cache is being flushed, the size of the
8048 memory range, and the number 3 (to flush both caches). The default
8049 depends on the target gcc was configured for, but commonly is either
8050 @samp{_flush_func} or @samp{__cpu_flush}.
8052 @item -mbranch-likely
8053 @itemx -mno-branch-likely
8054 @opindex mbranch-likely
8055 @opindex mno-branch-likely
8056 Enable or disable use of Branch Likely instructions, regardless of the
8057 default for the selected architecture. By default, Branch Likely
8058 instructions may be generated if they are supported by the selected
8059 architecture. An exception is for the MIPS32 and MIPS64 architectures
8060 and processors which implement those architectures; for those, Branch
8061 Likely instructions will not be generated by default because the MIPS32
8062 and MIPS64 architectures specifically deprecate their use.
8065 @node i386 and x86-64 Options
8066 @subsection Intel 386 and AMD x86-64 Options
8067 @cindex i386 Options
8068 @cindex x86-64 Options
8069 @cindex Intel 386 Options
8070 @cindex AMD x86-64 Options
8072 These @samp{-m} options are defined for the i386 and x86-64 family of
8076 @item -mtune=@var{cpu-type}
8078 Tune to @var{cpu-type} everything applicable about the generated code, except
8079 for the ABI and the set of available instructions. The choices for
8080 @var{cpu-type} are @samp{i386}, @samp{i486}, @samp{i586}, @samp{i686},
8081 @samp{pentium}, @samp{pentium-mmx}, @samp{pentiumpro}, @samp{pentium2},
8082 @samp{pentium3}, @samp{pentium4}, @samp{k6}, @samp{k6-2}, @samp{k6-3},
8083 @samp{athlon}, @samp{athlon-tbird}, @samp{athlon-4}, @samp{athlon-xp},
8084 @samp{athlon-mp}, @samp{winchip-c6}, @samp{winchip2}, @samp{k8}, @samp{c3}
8087 While picking a specific @var{cpu-type} will schedule things appropriately
8088 for that particular chip, the compiler will not generate any code that
8089 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8090 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
8091 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
8092 AMD chips as opposed to the Intel ones.
8094 @item -march=@var{cpu-type}
8096 Generate instructions for the machine type @var{cpu-type}. The choices
8097 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8098 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8100 @item -mcpu=@var{cpu-type}
8102 A deprecated synonym for @option{-mtune}.
8111 @opindex mpentiumpro
8112 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8113 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8114 These synonyms are deprecated.
8116 @item -mfpmath=@var{unit}
8118 generate floating point arithmetics for selected unit @var{unit}. the choices
8123 Use the standard 387 floating point coprocessor present majority of chips and
8124 emulated otherwise. Code compiled with this option will run almost everywhere.
8125 The temporary results are computed in 80bit precision instead of precision
8126 specified by the type resulting in slightly different results compared to most
8127 of other chips. See @option{-ffloat-store} for more detailed description.
8129 This is the default choice for i386 compiler.
8132 Use scalar floating point instructions present in the SSE instruction set.
8133 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8134 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8135 instruction set supports only single precision arithmetics, thus the double and
8136 extended precision arithmetics is still done using 387. Later version, present
8137 only in Pentium4 and the future AMD x86-64 chips supports double precision
8140 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8141 @option{-msse2} switches to enable SSE extensions and make this option
8142 effective. For x86-64 compiler, these extensions are enabled by default.
8144 The resulting code should be considerably faster in majority of cases and avoid
8145 the numerical instability problems of 387 code, but may break some existing
8146 code that expects temporaries to be 80bit.
8148 This is the default choice for x86-64 compiler.
8151 Attempt to utilize both instruction sets at once. This effectively double the
8152 amount of available registers and on chips with separate execution units for
8153 387 and SSE the execution resources too. Use this option with care, as it is
8154 still experimental, because gcc register allocator does not model separate
8155 functional units well resulting in instable performance.
8158 @item -masm=@var{dialect}
8159 @opindex masm=@var{dialect}
8160 Output asm instructions using selected @var{dialect}. Supported choices are
8161 @samp{intel} or @samp{att} (the default one).
8166 @opindex mno-ieee-fp
8167 Control whether or not the compiler uses IEEE floating point
8168 comparisons. These handle correctly the case where the result of a
8169 comparison is unordered.
8172 @opindex msoft-float
8173 Generate output containing library calls for floating point.
8174 @strong{Warning:} the requisite libraries are not part of GCC@.
8175 Normally the facilities of the machine's usual C compiler are used, but
8176 this can't be done directly in cross-compilation. You must make your
8177 own arrangements to provide suitable library functions for
8180 On machines where a function returns floating point results in the 80387
8181 register stack, some floating point opcodes may be emitted even if
8182 @option{-msoft-float} is used.
8184 @item -mno-fp-ret-in-387
8185 @opindex mno-fp-ret-in-387
8186 Do not use the FPU registers for return values of functions.
8188 The usual calling convention has functions return values of types
8189 @code{float} and @code{double} in an FPU register, even if there
8190 is no FPU@. The idea is that the operating system should emulate
8193 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8194 in ordinary CPU registers instead.
8196 @item -mno-fancy-math-387
8197 @opindex mno-fancy-math-387
8198 Some 387 emulators do not support the @code{sin}, @code{cos} and
8199 @code{sqrt} instructions for the 387. Specify this option to avoid
8200 generating those instructions. This option is the default on FreeBSD,
8201 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8202 indicates that the target cpu will always have an FPU and so the
8203 instruction will not need emulation. As of revision 2.6.1, these
8204 instructions are not generated unless you also use the
8205 @option{-funsafe-math-optimizations} switch.
8207 @item -malign-double
8208 @itemx -mno-align-double
8209 @opindex malign-double
8210 @opindex mno-align-double
8211 Control whether GCC aligns @code{double}, @code{long double}, and
8212 @code{long long} variables on a two word boundary or a one word
8213 boundary. Aligning @code{double} variables on a two word boundary will
8214 produce code that runs somewhat faster on a @samp{Pentium} at the
8215 expense of more memory.
8217 @strong{Warning:} if you use the @option{-malign-double} switch,
8218 structures containing the above types will be aligned differently than
8219 the published application binary interface specifications for the 386
8220 and will not be binary compatible with structures in code compiled
8221 without that switch.
8223 @item -m128bit-long-double
8224 @opindex m128bit-long-double
8225 Control the size of @code{long double} type. i386 application binary interface
8226 specify the size to be 12 bytes, while modern architectures (Pentium and newer)
8227 prefer @code{long double} aligned to 8 or 16 byte boundary. This is
8228 impossible to reach with 12 byte long doubles in the array accesses.
8230 @strong{Warning:} if you use the @option{-m128bit-long-double} switch, the
8231 structures and arrays containing @code{long double} will change their size as
8232 well as function calling convention for function taking @code{long double}
8235 @item -m96bit-long-double
8236 @opindex m96bit-long-double
8237 Set the size of @code{long double} to 96 bits as required by the i386
8238 application binary interface. This is the default.
8241 @itemx -mno-svr3-shlib
8242 @opindex msvr3-shlib
8243 @opindex mno-svr3-shlib
8244 Control whether GCC places uninitialized local variables into the
8245 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8246 into @code{bss}. These options are meaningful only on System V Release 3.
8250 Use a different function-calling convention, in which functions that
8251 take a fixed number of arguments return with the @code{ret} @var{num}
8252 instruction, which pops their arguments while returning. This saves one
8253 instruction in the caller since there is no need to pop the arguments
8256 You can specify that an individual function is called with this calling
8257 sequence with the function attribute @samp{stdcall}. You can also
8258 override the @option{-mrtd} option by using the function attribute
8259 @samp{cdecl}. @xref{Function Attributes}.
8261 @strong{Warning:} this calling convention is incompatible with the one
8262 normally used on Unix, so you cannot use it if you need to call
8263 libraries compiled with the Unix compiler.
8265 Also, you must provide function prototypes for all functions that
8266 take variable numbers of arguments (including @code{printf});
8267 otherwise incorrect code will be generated for calls to those
8270 In addition, seriously incorrect code will result if you call a
8271 function with too many arguments. (Normally, extra arguments are
8272 harmlessly ignored.)
8274 @item -mregparm=@var{num}
8276 Control how many registers are used to pass integer arguments. By
8277 default, no registers are used to pass arguments, and at most 3
8278 registers can be used. You can control this behavior for a specific
8279 function by using the function attribute @samp{regparm}.
8280 @xref{Function Attributes}.
8282 @strong{Warning:} if you use this switch, and
8283 @var{num} is nonzero, then you must build all modules with the same
8284 value, including any libraries. This includes the system libraries and
8287 @item -mpreferred-stack-boundary=@var{num}
8288 @opindex mpreferred-stack-boundary
8289 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8290 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8291 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8292 size (@option{-Os}), in which case the default is the minimum correct
8293 alignment (4 bytes for x86, and 8 bytes for x86-64).
8295 On Pentium and PentiumPro, @code{double} and @code{long double} values
8296 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8297 suffer significant run time performance penalties. On Pentium III, the
8298 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8299 penalties if it is not 16 byte aligned.
8301 To ensure proper alignment of this values on the stack, the stack boundary
8302 must be as aligned as that required by any value stored on the stack.
8303 Further, every function must be generated such that it keeps the stack
8304 aligned. Thus calling a function compiled with a higher preferred
8305 stack boundary from a function compiled with a lower preferred stack
8306 boundary will most likely misalign the stack. It is recommended that
8307 libraries that use callbacks always use the default setting.
8309 This extra alignment does consume extra stack space, and generally
8310 increases code size. Code that is sensitive to stack space usage, such
8311 as embedded systems and operating system kernels, may want to reduce the
8312 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8328 These switches enable or disable the use of built-in functions that allow
8329 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
8331 @xref{X86 Built-in Functions}, for details of the functions enabled
8332 and disabled by these switches.
8334 To have SSE/SSE2 instructions generated automatically from floating-point
8335 code, see @option{-mfpmath=sse}.
8338 @itemx -mno-push-args
8340 @opindex mno-push-args
8341 Use PUSH operations to store outgoing parameters. This method is shorter
8342 and usually equally fast as method using SUB/MOV operations and is enabled
8343 by default. In some cases disabling it may improve performance because of
8344 improved scheduling and reduced dependencies.
8346 @item -maccumulate-outgoing-args
8347 @opindex maccumulate-outgoing-args
8348 If enabled, the maximum amount of space required for outgoing arguments will be
8349 computed in the function prologue. This is faster on most modern CPUs
8350 because of reduced dependencies, improved scheduling and reduced stack usage
8351 when preferred stack boundary is not equal to 2. The drawback is a notable
8352 increase in code size. This switch implies @option{-mno-push-args}.
8356 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8357 on thread-safe exception handling must compile and link all code with the
8358 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8359 @option{-D_MT}; when linking, it links in a special thread helper library
8360 @option{-lmingwthrd} which cleans up per thread exception handling data.
8362 @item -mno-align-stringops
8363 @opindex mno-align-stringops
8364 Do not align destination of inlined string operations. This switch reduces
8365 code size and improves performance in case the destination is already aligned,
8366 but gcc don't know about it.
8368 @item -minline-all-stringops
8369 @opindex minline-all-stringops
8370 By default GCC inlines string operations only when destination is known to be
8371 aligned at least to 4 byte boundary. This enables more inlining, increase code
8372 size, but may improve performance of code that depends on fast memcpy, strlen
8373 and memset for short lengths.
8375 @item -momit-leaf-frame-pointer
8376 @opindex momit-leaf-frame-pointer
8377 Don't keep the frame pointer in a register for leaf functions. This
8378 avoids the instructions to save, set up and restore frame pointers and
8379 makes an extra register available in leaf functions. The option
8380 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8381 which might make debugging harder.
8384 These @samp{-m} switches are supported in addition to the above
8385 on AMD x86-64 processors in 64-bit environments.
8392 Generate code for a 32-bit or 64-bit environment.
8393 The 32-bit environment sets int, long and pointer to 32 bits and
8394 generates code that runs on any i386 system.
8395 The 64-bit environment sets int to 32 bits and long and pointer
8396 to 64 bits and generates code for AMD's x86-64 architecture.
8399 @opindex no-red-zone
8400 Do not use a so called red zone for x86-64 code. The red zone is mandated
8401 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8402 stack pointer that will not be modified by signal or interrupt handlers
8403 and therefore can be used for temporary data without adjusting the stack
8404 pointer. The flag @option{-mno-red-zone} disables this red zone.
8406 @item -mcmodel=small
8407 @opindex mcmodel=small
8408 Generate code for the small code model: the program and its symbols must
8409 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8410 Programs can be statically or dynamically linked. This is the default
8413 @item -mcmodel=kernel
8414 @opindex mcmodel=kernel
8415 Generate code for the kernel code model. The kernel runs in the
8416 negative 2 GB of the address space.
8417 This model has to be used for Linux kernel code.
8419 @item -mcmodel=medium
8420 @opindex mcmodel=medium
8421 Generate code for the medium model: The program is linked in the lower 2
8422 GB of the address space but symbols can be located anywhere in the
8423 address space. Programs can be statically or dynamically linked, but
8424 building of shared libraries are not supported with the medium model.
8426 @item -mcmodel=large
8427 @opindex mcmodel=large
8428 Generate code for the large model: This model makes no assumptions
8429 about addresses and sizes of sections. Currently GCC does not implement
8434 @subsection HPPA Options
8435 @cindex HPPA Options
8437 These @samp{-m} options are defined for the HPPA family of computers:
8440 @item -march=@var{architecture-type}
8442 Generate code for the specified architecture. The choices for
8443 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8444 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8445 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8446 architecture option for your machine. Code compiled for lower numbered
8447 architectures will run on higher numbered architectures, but not the
8450 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8451 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8455 @itemx -mpa-risc-1-1
8456 @itemx -mpa-risc-2-0
8457 @opindex mpa-risc-1-0
8458 @opindex mpa-risc-1-1
8459 @opindex mpa-risc-2-0
8460 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8463 @opindex mbig-switch
8464 Generate code suitable for big switch tables. Use this option only if
8465 the assembler/linker complain about out of range branches within a switch
8468 @item -mjump-in-delay
8469 @opindex mjump-in-delay
8470 Fill delay slots of function calls with unconditional jump instructions
8471 by modifying the return pointer for the function call to be the target
8472 of the conditional jump.
8474 @item -mdisable-fpregs
8475 @opindex mdisable-fpregs
8476 Prevent floating point registers from being used in any manner. This is
8477 necessary for compiling kernels which perform lazy context switching of
8478 floating point registers. If you use this option and attempt to perform
8479 floating point operations, the compiler will abort.
8481 @item -mdisable-indexing
8482 @opindex mdisable-indexing
8483 Prevent the compiler from using indexing address modes. This avoids some
8484 rather obscure problems when compiling MIG generated code under MACH@.
8486 @item -mno-space-regs
8487 @opindex mno-space-regs
8488 Generate code that assumes the target has no space registers. This allows
8489 GCC to generate faster indirect calls and use unscaled index address modes.
8491 Such code is suitable for level 0 PA systems and kernels.
8493 @item -mfast-indirect-calls
8494 @opindex mfast-indirect-calls
8495 Generate code that assumes calls never cross space boundaries. This
8496 allows GCC to emit code which performs faster indirect calls.
8498 This option will not work in the presence of shared libraries or nested
8501 @item -mlong-load-store
8502 @opindex mlong-load-store
8503 Generate 3-instruction load and store sequences as sometimes required by
8504 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8507 @item -mportable-runtime
8508 @opindex mportable-runtime
8509 Use the portable calling conventions proposed by HP for ELF systems.
8513 Enable the use of assembler directives only GAS understands.
8515 @item -mschedule=@var{cpu-type}
8517 Schedule code according to the constraints for the machine type
8518 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8519 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8520 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8521 proper scheduling option for your machine. The default scheduling is
8525 @opindex mlinker-opt
8526 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8527 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8528 linkers in which they give bogus error messages when linking some programs.
8531 @opindex msoft-float
8532 Generate output containing library calls for floating point.
8533 @strong{Warning:} the requisite libraries are not available for all HPPA
8534 targets. Normally the facilities of the machine's usual C compiler are
8535 used, but this cannot be done directly in cross-compilation. You must make
8536 your own arrangements to provide suitable library functions for
8537 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8538 does provide software floating point support.
8540 @option{-msoft-float} changes the calling convention in the output file;
8541 therefore, it is only useful if you compile @emph{all} of a program with
8542 this option. In particular, you need to compile @file{libgcc.a}, the
8543 library that comes with GCC, with @option{-msoft-float} in order for
8548 Generate the predefine, @code{_SIO}, for server IO. The default is
8549 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8550 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8551 options are available under HP-UX and HI-UX.
8555 Use GNU ld specific options. This passes @option{-shared} to ld when
8556 building a shared library. It is the default when GCC is configured,
8557 explicitly or implicitly, with the GNU linker. This option does not
8558 have any affect on which ld is called, it only changes what parameters
8559 are passed to that ld. The ld that is called is determined by the
8560 @option{--with-ld} configure option, gcc's program search path, and
8561 finally by the user's @env{PATH}. The linker used by GCC can be printed
8562 using @samp{which `gcc -print-prog-name=ld`}.
8566 Use HP ld specific options. This passes @option{-b} to ld when building
8567 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8568 links. It is the default when GCC is configured, explicitly or
8569 implicitly, with the HP linker. This option does not have any affect on
8570 which ld is called, it only changes what parameters are passed to that
8571 ld. The ld that is called is determined by the @option{--with-ld}
8572 configure option, gcc's program search path, and finally by the user's
8573 @env{PATH}. The linker used by GCC can be printed using @samp{which
8574 `gcc -print-prog-name=ld`}.
8577 @opindex mno-long-calls
8578 Generate code that uses long call sequences. This ensures that a call
8579 is always able to reach linker generated stubs. The default is to generate
8580 long calls only when the distance from the call site to the beginning
8581 of the function or translation unit, as the case may be, exceeds a
8582 predefined limit set by the branch type being used. The limits for
8583 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8584 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8587 Distances are measured from the beginning of functions when using the
8588 @option{-ffunction-sections} option, or when using the @option{-mgas}
8589 and @option{-mno-portable-runtime} options together under HP-UX with
8592 It is normally not desirable to use this option as it will degrade
8593 performance. However, it may be useful in large applications,
8594 particularly when partial linking is used to build the application.
8596 The types of long calls used depends on the capabilities of the
8597 assembler and linker, and the type of code being generated. The
8598 impact on systems that support long absolute calls, and long pic
8599 symbol-difference or pc-relative calls should be relatively small.
8600 However, an indirect call is used on 32-bit ELF systems in pic code
8601 and it is quite long.
8605 Suppress the generation of link options to search libdld.sl when the
8606 @option{-static} option is specified on HP-UX 10 and later.
8610 The HP-UX implementation of setlocale in libc has a dependency on
8611 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8612 when the @option{-static} option is specified, special link options
8613 are needed to resolve this dependency.
8615 On HP-UX 10 and later, the GCC driver adds the necessary options to
8616 link with libdld.sl when the @option{-static} option is specified.
8617 This causes the resulting binary to be dynamic. On the 64-bit port,
8618 the linkers generate dynamic binaries by default in any case. The
8619 @option{-nolibdld} option can be used to prevent the GCC driver from
8620 adding these link options.
8624 Add support for multithreading with the @dfn{dce thread} library
8625 under HP-UX. This option sets flags for both the preprocessor and
8629 @node Intel 960 Options
8630 @subsection Intel 960 Options
8632 These @samp{-m} options are defined for the Intel 960 implementations:
8635 @item -m@var{cpu-type}
8643 Assume the defaults for the machine type @var{cpu-type} for some of
8644 the other options, including instruction scheduling, floating point
8645 support, and addressing modes. The choices for @var{cpu-type} are
8646 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
8647 @samp{sa}, and @samp{sb}.
8654 @opindex msoft-float
8655 The @option{-mnumerics} option indicates that the processor does support
8656 floating-point instructions. The @option{-msoft-float} option indicates
8657 that floating-point support should not be assumed.
8659 @item -mleaf-procedures
8660 @itemx -mno-leaf-procedures
8661 @opindex mleaf-procedures
8662 @opindex mno-leaf-procedures
8663 Do (or do not) attempt to alter leaf procedures to be callable with the
8664 @code{bal} instruction as well as @code{call}. This will result in more
8665 efficient code for explicit calls when the @code{bal} instruction can be
8666 substituted by the assembler or linker, but less efficient code in other
8667 cases, such as calls via function pointers, or using a linker that doesn't
8668 support this optimization.
8671 @itemx -mno-tail-call
8673 @opindex mno-tail-call
8674 Do (or do not) make additional attempts (beyond those of the
8675 machine-independent portions of the compiler) to optimize tail-recursive
8676 calls into branches. You may not want to do this because the detection of
8677 cases where this is not valid is not totally complete. The default is
8678 @option{-mno-tail-call}.
8680 @item -mcomplex-addr
8681 @itemx -mno-complex-addr
8682 @opindex mcomplex-addr
8683 @opindex mno-complex-addr
8684 Assume (or do not assume) that the use of a complex addressing mode is a
8685 win on this implementation of the i960. Complex addressing modes may not
8686 be worthwhile on the K-series, but they definitely are on the C-series.
8687 The default is currently @option{-mcomplex-addr} for all processors except
8691 @itemx -mno-code-align
8692 @opindex mcode-align
8693 @opindex mno-code-align
8694 Align code to 8-byte boundaries for faster fetching (or don't bother).
8695 Currently turned on by default for C-series implementations only.
8698 @item -mclean-linkage
8699 @itemx -mno-clean-linkage
8700 @opindex mclean-linkage
8701 @opindex mno-clean-linkage
8702 These options are not fully implemented.
8706 @itemx -mic2.0-compat
8707 @itemx -mic3.0-compat
8709 @opindex mic2.0-compat
8710 @opindex mic3.0-compat
8711 Enable compatibility with iC960 v2.0 or v3.0.
8715 @opindex masm-compat
8717 Enable compatibility with the iC960 assembler.
8719 @item -mstrict-align
8720 @itemx -mno-strict-align
8721 @opindex mstrict-align
8722 @opindex mno-strict-align
8723 Do not permit (do permit) unaligned accesses.
8727 Enable structure-alignment compatibility with Intel's gcc release version
8728 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
8730 @item -mlong-double-64
8731 @opindex mlong-double-64
8732 Implement type @samp{long double} as 64-bit floating point numbers.
8733 Without the option @samp{long double} is implemented by 80-bit
8734 floating point numbers. The only reason we have it because there is
8735 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
8736 is only useful for people using soft-float targets. Otherwise, we
8737 should recommend against use of it.
8741 @node DEC Alpha Options
8742 @subsection DEC Alpha Options
8744 These @samp{-m} options are defined for the DEC Alpha implementations:
8747 @item -mno-soft-float
8749 @opindex mno-soft-float
8750 @opindex msoft-float
8751 Use (do not use) the hardware floating-point instructions for
8752 floating-point operations. When @option{-msoft-float} is specified,
8753 functions in @file{libgcc.a} will be used to perform floating-point
8754 operations. Unless they are replaced by routines that emulate the
8755 floating-point operations, or compiled in such a way as to call such
8756 emulations routines, these routines will issue floating-point
8757 operations. If you are compiling for an Alpha without floating-point
8758 operations, you must ensure that the library is built so as not to call
8761 Note that Alpha implementations without floating-point operations are
8762 required to have floating-point registers.
8767 @opindex mno-fp-regs
8768 Generate code that uses (does not use) the floating-point register set.
8769 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8770 register set is not used, floating point operands are passed in integer
8771 registers as if they were integers and floating-point results are passed
8772 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8773 so any function with a floating-point argument or return value called by code
8774 compiled with @option{-mno-fp-regs} must also be compiled with that
8777 A typical use of this option is building a kernel that does not use,
8778 and hence need not save and restore, any floating-point registers.
8782 The Alpha architecture implements floating-point hardware optimized for
8783 maximum performance. It is mostly compliant with the IEEE floating
8784 point standard. However, for full compliance, software assistance is
8785 required. This option generates code fully IEEE compliant code
8786 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8787 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8788 defined during compilation. The resulting code is less efficient but is
8789 able to correctly support denormalized numbers and exceptional IEEE
8790 values such as not-a-number and plus/minus infinity. Other Alpha
8791 compilers call this option @option{-ieee_with_no_inexact}.
8793 @item -mieee-with-inexact
8794 @opindex mieee-with-inexact
8795 This is like @option{-mieee} except the generated code also maintains
8796 the IEEE @var{inexact-flag}. Turning on this option causes the
8797 generated code to implement fully-compliant IEEE math. In addition to
8798 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8799 macro. On some Alpha implementations the resulting code may execute
8800 significantly slower than the code generated by default. Since there is
8801 very little code that depends on the @var{inexact-flag}, you should
8802 normally not specify this option. Other Alpha compilers call this
8803 option @option{-ieee_with_inexact}.
8805 @item -mfp-trap-mode=@var{trap-mode}
8806 @opindex mfp-trap-mode
8807 This option controls what floating-point related traps are enabled.
8808 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8809 The trap mode can be set to one of four values:
8813 This is the default (normal) setting. The only traps that are enabled
8814 are the ones that cannot be disabled in software (e.g., division by zero
8818 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8822 Like @samp{su}, but the instructions are marked to be safe for software
8823 completion (see Alpha architecture manual for details).
8826 Like @samp{su}, but inexact traps are enabled as well.
8829 @item -mfp-rounding-mode=@var{rounding-mode}
8830 @opindex mfp-rounding-mode
8831 Selects the IEEE rounding mode. Other Alpha compilers call this option
8832 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8837 Normal IEEE rounding mode. Floating point numbers are rounded towards
8838 the nearest machine number or towards the even machine number in case
8842 Round towards minus infinity.
8845 Chopped rounding mode. Floating point numbers are rounded towards zero.
8848 Dynamic rounding mode. A field in the floating point control register
8849 (@var{fpcr}, see Alpha architecture reference manual) controls the
8850 rounding mode in effect. The C library initializes this register for
8851 rounding towards plus infinity. Thus, unless your program modifies the
8852 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8855 @item -mtrap-precision=@var{trap-precision}
8856 @opindex mtrap-precision
8857 In the Alpha architecture, floating point traps are imprecise. This
8858 means without software assistance it is impossible to recover from a
8859 floating trap and program execution normally needs to be terminated.
8860 GCC can generate code that can assist operating system trap handlers
8861 in determining the exact location that caused a floating point trap.
8862 Depending on the requirements of an application, different levels of
8863 precisions can be selected:
8867 Program precision. This option is the default and means a trap handler
8868 can only identify which program caused a floating point exception.
8871 Function precision. The trap handler can determine the function that
8872 caused a floating point exception.
8875 Instruction precision. The trap handler can determine the exact
8876 instruction that caused a floating point exception.
8879 Other Alpha compilers provide the equivalent options called
8880 @option{-scope_safe} and @option{-resumption_safe}.
8882 @item -mieee-conformant
8883 @opindex mieee-conformant
8884 This option marks the generated code as IEEE conformant. You must not
8885 use this option unless you also specify @option{-mtrap-precision=i} and either
8886 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8887 is to emit the line @samp{.eflag 48} in the function prologue of the
8888 generated assembly file. Under DEC Unix, this has the effect that
8889 IEEE-conformant math library routines will be linked in.
8891 @item -mbuild-constants
8892 @opindex mbuild-constants
8893 Normally GCC examines a 32- or 64-bit integer constant to
8894 see if it can construct it from smaller constants in two or three
8895 instructions. If it cannot, it will output the constant as a literal and
8896 generate code to load it from the data segment at runtime.
8898 Use this option to require GCC to construct @emph{all} integer constants
8899 using code, even if it takes more instructions (the maximum is six).
8901 You would typically use this option to build a shared library dynamic
8902 loader. Itself a shared library, it must relocate itself in memory
8903 before it can find the variables and constants in its own data segment.
8909 Select whether to generate code to be assembled by the vendor-supplied
8910 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8928 Indicate whether GCC should generate code to use the optional BWX,
8929 CIX, FIX and MAX instruction sets. The default is to use the instruction
8930 sets supported by the CPU type specified via @option{-mcpu=} option or that
8931 of the CPU on which GCC was built if none was specified.
8936 @opindex mfloat-ieee
8937 Generate code that uses (does not use) VAX F and G floating point
8938 arithmetic instead of IEEE single and double precision.
8940 @item -mexplicit-relocs
8941 @itemx -mno-explicit-relocs
8942 @opindex mexplicit-relocs
8943 @opindex mno-explicit-relocs
8944 Older Alpha assemblers provided no way to generate symbol relocations
8945 except via assembler macros. Use of these macros does not allow
8946 optimal instruction scheduling. GNU binutils as of version 2.12
8947 supports a new syntax that allows the compiler to explicitly mark
8948 which relocations should apply to which instructions. This option
8949 is mostly useful for debugging, as GCC detects the capabilities of
8950 the assembler when it is built and sets the default accordingly.
8954 @opindex msmall-data
8955 @opindex mlarge-data
8956 When @option{-mexplicit-relocs} is in effect, static data is
8957 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8958 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8959 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8960 16-bit relocations off of the @code{$gp} register. This limits the
8961 size of the small data area to 64KB, but allows the variables to be
8962 directly accessed via a single instruction.
8964 The default is @option{-mlarge-data}. With this option the data area
8965 is limited to just below 2GB. Programs that require more than 2GB of
8966 data must use @code{malloc} or @code{mmap} to allocate the data in the
8967 heap instead of in the program's data segment.
8969 When generating code for shared libraries, @option{-fpic} implies
8970 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8974 @opindex msmall-text
8975 @opindex mlarge-text
8976 When @option{-msmall-text} is used, the compiler assumes that the
8977 code of the entire program (or shared library) fits in 4MB, and is
8978 thus reachable with a branch instruction. When @option{-msmall-data}
8979 is used, the compiler can assume that all local symbols share the
8980 same @code{$gp} value, and thus reduce the number of instructions
8981 required for a function call from 4 to 1.
8983 The default is @option{-mlarge-text}.
8985 @item -mcpu=@var{cpu_type}
8987 Set the instruction set and instruction scheduling parameters for
8988 machine type @var{cpu_type}. You can specify either the @samp{EV}
8989 style name or the corresponding chip number. GCC supports scheduling
8990 parameters for the EV4, EV5 and EV6 family of processors and will
8991 choose the default values for the instruction set from the processor
8992 you specify. If you do not specify a processor type, GCC will default
8993 to the processor on which the compiler was built.
8995 Supported values for @var{cpu_type} are
9001 Schedules as an EV4 and has no instruction set extensions.
9005 Schedules as an EV5 and has no instruction set extensions.
9009 Schedules as an EV5 and supports the BWX extension.
9014 Schedules as an EV5 and supports the BWX and MAX extensions.
9018 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9022 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9025 @item -mtune=@var{cpu_type}
9027 Set only the instruction scheduling parameters for machine type
9028 @var{cpu_type}. The instruction set is not changed.
9030 @item -mmemory-latency=@var{time}
9031 @opindex mmemory-latency
9032 Sets the latency the scheduler should assume for typical memory
9033 references as seen by the application. This number is highly
9034 dependent on the memory access patterns used by the application
9035 and the size of the external cache on the machine.
9037 Valid options for @var{time} are
9041 A decimal number representing clock cycles.
9047 The compiler contains estimates of the number of clock cycles for
9048 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9049 (also called Dcache, Scache, and Bcache), as well as to main memory.
9050 Note that L3 is only valid for EV5.
9055 @node DEC Alpha/VMS Options
9056 @subsection DEC Alpha/VMS Options
9058 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9061 @item -mvms-return-codes
9062 @opindex mvms-return-codes
9063 Return VMS condition codes from main. The default is to return POSIX
9064 style condition (e.g.@ error) codes.
9067 @node H8/300 Options
9068 @subsection H8/300 Options
9070 These @samp{-m} options are defined for the H8/300 implementations:
9075 Shorten some address references at link time, when possible; uses the
9076 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9077 ld.info, Using ld}, for a fuller description.
9081 Generate code for the H8/300H@.
9085 Generate code for the H8S@.
9089 Generate code for the H8S and H8/300H in the normal mode. This switch
9090 must be used either with -mh or -ms.
9094 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9098 Make @code{int} data 32 bits by default.
9102 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9103 The default for the H8/300H and H8S is to align longs and floats on 4
9105 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9106 This option has no effect on the H8/300.
9110 @subsection SH Options
9112 These @samp{-m} options are defined for the SH implementations:
9117 Generate code for the SH1.
9121 Generate code for the SH2.
9124 Generate code for the SH2e.
9128 Generate code for the SH3.
9132 Generate code for the SH3e.
9136 Generate code for the SH4 without a floating-point unit.
9138 @item -m4-single-only
9139 @opindex m4-single-only
9140 Generate code for the SH4 with a floating-point unit that only
9141 supports single-precision arithmetic.
9145 Generate code for the SH4 assuming the floating-point unit is in
9146 single-precision mode by default.
9150 Generate code for the SH4.
9154 Compile code for the processor in big endian mode.
9158 Compile code for the processor in little endian mode.
9162 Align doubles at 64-bit boundaries. Note that this changes the calling
9163 conventions, and thus some functions from the standard C library will
9164 not work unless you recompile it first with @option{-mdalign}.
9168 Shorten some address references at link time, when possible; uses the
9169 linker option @option{-relax}.
9173 Use 32-bit offsets in @code{switch} tables. The default is to use
9178 Enable the use of the instruction @code{fmovd}.
9182 Comply with the calling conventions defined by Hitachi.
9186 Mark the @code{MAC} register as call-clobbered, even if
9187 @option{-mhitachi} is given.
9191 Increase IEEE-compliance of floating-point code.
9195 Dump instruction size and location in the assembly code.
9199 This option is deprecated. It pads structures to multiple of 4 bytes,
9200 which is incompatible with the SH ABI@.
9204 Optimize for space instead of speed. Implied by @option{-Os}.
9208 When generating position-independent code, emit function calls using
9209 the Global Offset Table instead of the Procedure Linkage Table.
9213 Generate a library function call to invalidate instruction cache
9214 entries, after fixing up a trampoline. This library function call
9215 doesn't assume it can write to the whole memory address space. This
9216 is the default when the target is @code{sh-*-linux*}.
9219 @node System V Options
9220 @subsection Options for System V
9222 These additional options are available on System V Release 4 for
9223 compatibility with other compilers on those systems:
9228 Create a shared object.
9229 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9233 Identify the versions of each tool used by the compiler, in a
9234 @code{.ident} assembler directive in the output.
9238 Refrain from adding @code{.ident} directives to the output file (this is
9241 @item -YP,@var{dirs}
9243 Search the directories @var{dirs}, and no others, for libraries
9244 specified with @option{-l}.
9248 Look in the directory @var{dir} to find the M4 preprocessor.
9249 The assembler uses this option.
9250 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9251 @c the generic assembler that comes with Solaris takes just -Ym.
9254 @node TMS320C3x/C4x Options
9255 @subsection TMS320C3x/C4x Options
9256 @cindex TMS320C3x/C4x Options
9258 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9262 @item -mcpu=@var{cpu_type}
9264 Set the instruction set, register set, and instruction scheduling
9265 parameters for machine type @var{cpu_type}. Supported values for
9266 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9267 @samp{c44}. The default is @samp{c40} to generate code for the
9272 @itemx -msmall-memory
9274 @opindex mbig-memory
9276 @opindex msmall-memory
9278 Generates code for the big or small memory model. The small memory
9279 model assumed that all data fits into one 64K word page. At run-time
9280 the data page (DP) register must be set to point to the 64K page
9281 containing the .bss and .data program sections. The big memory model is
9282 the default and requires reloading of the DP register for every direct
9289 Allow (disallow) allocation of general integer operands into the block
9296 Enable (disable) generation of code using decrement and branch,
9297 DBcond(D), instructions. This is enabled by default for the C4x. To be
9298 on the safe side, this is disabled for the C3x, since the maximum
9299 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9300 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9301 that it can utilize the decrement and branch instruction, but will give
9302 up if there is more than one memory reference in the loop. Thus a loop
9303 where the loop counter is decremented can generate slightly more
9304 efficient code, in cases where the RPTB instruction cannot be utilized.
9306 @item -mdp-isr-reload
9308 @opindex mdp-isr-reload
9310 Force the DP register to be saved on entry to an interrupt service
9311 routine (ISR), reloaded to point to the data section, and restored on
9312 exit from the ISR@. This should not be required unless someone has
9313 violated the small memory model by modifying the DP register, say within
9320 For the C3x use the 24-bit MPYI instruction for integer multiplies
9321 instead of a library call to guarantee 32-bit results. Note that if one
9322 of the operands is a constant, then the multiplication will be performed
9323 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9324 then squaring operations are performed inline instead of a library call.
9327 @itemx -mno-fast-fix
9329 @opindex mno-fast-fix
9330 The C3x/C4x FIX instruction to convert a floating point value to an
9331 integer value chooses the nearest integer less than or equal to the
9332 floating point value rather than to the nearest integer. Thus if the
9333 floating point number is negative, the result will be incorrectly
9334 truncated an additional code is necessary to detect and correct this
9335 case. This option can be used to disable generation of the additional
9336 code required to correct the result.
9342 Enable (disable) generation of repeat block sequences using the RPTB
9343 instruction for zero overhead looping. The RPTB construct is only used
9344 for innermost loops that do not call functions or jump across the loop
9345 boundaries. There is no advantage having nested RPTB loops due to the
9346 overhead required to save and restore the RC, RS, and RE registers.
9347 This is enabled by default with @option{-O2}.
9349 @item -mrpts=@var{count}
9353 Enable (disable) the use of the single instruction repeat instruction
9354 RPTS@. If a repeat block contains a single instruction, and the loop
9355 count can be guaranteed to be less than the value @var{count}, GCC will
9356 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9357 then a RPTS will be emitted even if the loop count cannot be determined
9358 at compile time. Note that the repeated instruction following RPTS does
9359 not have to be reloaded from memory each iteration, thus freeing up the
9360 CPU buses for operands. However, since interrupts are blocked by this
9361 instruction, it is disabled by default.
9363 @item -mloop-unsigned
9364 @itemx -mno-loop-unsigned
9365 @opindex mloop-unsigned
9366 @opindex mno-loop-unsigned
9367 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9368 is @math{2^{31} + 1} since these instructions test if the iteration count is
9369 negative to terminate the loop. If the iteration count is unsigned
9370 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9371 exceeded. This switch allows an unsigned iteration count.
9375 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9376 with. This also enforces compatibility with the API employed by the TI
9377 C3x C compiler. For example, long doubles are passed as structures
9378 rather than in floating point registers.
9384 Generate code that uses registers (stack) for passing arguments to functions.
9385 By default, arguments are passed in registers where possible rather
9386 than by pushing arguments on to the stack.
9388 @item -mparallel-insns
9389 @itemx -mno-parallel-insns
9390 @opindex mparallel-insns
9391 @opindex mno-parallel-insns
9392 Allow the generation of parallel instructions. This is enabled by
9393 default with @option{-O2}.
9395 @item -mparallel-mpy
9396 @itemx -mno-parallel-mpy
9397 @opindex mparallel-mpy
9398 @opindex mno-parallel-mpy
9399 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9400 provided @option{-mparallel-insns} is also specified. These instructions have
9401 tight register constraints which can pessimize the code generation
9407 @subsection V850 Options
9408 @cindex V850 Options
9410 These @samp{-m} options are defined for V850 implementations:
9414 @itemx -mno-long-calls
9415 @opindex mlong-calls
9416 @opindex mno-long-calls
9417 Treat all calls as being far away (near). If calls are assumed to be
9418 far away, the compiler will always load the functions address up into a
9419 register, and call indirect through the pointer.
9425 Do not optimize (do optimize) basic blocks that use the same index
9426 pointer 4 or more times to copy pointer into the @code{ep} register, and
9427 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9428 option is on by default if you optimize.
9430 @item -mno-prolog-function
9431 @itemx -mprolog-function
9432 @opindex mno-prolog-function
9433 @opindex mprolog-function
9434 Do not use (do use) external functions to save and restore registers at
9435 the prolog and epilog of a function. The external functions are slower,
9436 but use less code space if more than one function saves the same number
9437 of registers. The @option{-mprolog-function} option is on by default if
9442 Try to make the code as small as possible. At present, this just turns
9443 on the @option{-mep} and @option{-mprolog-function} options.
9447 Put static or global variables whose size is @var{n} bytes or less into
9448 the tiny data area that register @code{ep} points to. The tiny data
9449 area can hold up to 256 bytes in total (128 bytes for byte references).
9453 Put static or global variables whose size is @var{n} bytes or less into
9454 the small data area that register @code{gp} points to. The small data
9455 area can hold up to 64 kilobytes.
9459 Put static or global variables whose size is @var{n} bytes or less into
9460 the first 32 kilobytes of memory.
9464 Specify that the target processor is the V850.
9467 @opindex mbig-switch
9468 Generate code suitable for big switch tables. Use this option only if
9469 the assembler/linker complain about out of range branches within a switch
9474 This option will cause r2 and r5 to be used in the code generated by
9475 the compiler. This setting is the default.
9478 @opindex -mno-app-regs
9479 This option will cause r2 and r5 to be treated as fixed registers.
9483 Specify that the target processor is the V850E. The preprocessor
9484 constant @samp{__v850e__} will be defined if this option is used.
9486 If neither @option{-mv850} nor @option{-mv850e} are defined
9487 then a default target processor will be chosen and the relevant
9488 @samp{__v850*__} preprocessor constant will be defined.
9490 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9491 defined, regardless of which processor variant is the target.
9493 @item -mdisable-callt
9494 @opindex -mdisable-callt
9495 This option will suppress generation of the CALLT instruction for the
9496 v850e flavors of the v850 architecture. The default is
9497 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9502 @subsection ARC Options
9505 These options are defined for ARC implementations:
9510 Compile code for little endian mode. This is the default.
9514 Compile code for big endian mode.
9517 @opindex mmangle-cpu
9518 Prepend the name of the cpu to all public symbol names.
9519 In multiple-processor systems, there are many ARC variants with different
9520 instruction and register set characteristics. This flag prevents code
9521 compiled for one cpu to be linked with code compiled for another.
9522 No facility exists for handling variants that are ``almost identical''.
9523 This is an all or nothing option.
9525 @item -mcpu=@var{cpu}
9527 Compile code for ARC variant @var{cpu}.
9528 Which variants are supported depend on the configuration.
9529 All variants support @option{-mcpu=base}, this is the default.
9531 @item -mtext=@var{text-section}
9532 @itemx -mdata=@var{data-section}
9533 @itemx -mrodata=@var{readonly-data-section}
9537 Put functions, data, and readonly data in @var{text-section},
9538 @var{data-section}, and @var{readonly-data-section} respectively
9539 by default. This can be overridden with the @code{section} attribute.
9540 @xref{Variable Attributes}.
9545 @subsection NS32K Options
9546 @cindex NS32K options
9548 These are the @samp{-m} options defined for the 32000 series. The default
9549 values for these options depends on which style of 32000 was selected when
9550 the compiler was configured; the defaults for the most common choices are
9558 Generate output for a 32032. This is the default
9559 when the compiler is configured for 32032 and 32016 based systems.
9565 Generate output for a 32332. This is the default
9566 when the compiler is configured for 32332-based systems.
9572 Generate output for a 32532. This is the default
9573 when the compiler is configured for 32532-based systems.
9577 Generate output containing 32081 instructions for floating point.
9578 This is the default for all systems.
9582 Generate output containing 32381 instructions for floating point. This
9583 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9584 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9588 Try and generate multiply-add floating point instructions @code{polyF}
9589 and @code{dotF}. This option is only available if the @option{-m32381}
9590 option is in effect. Using these instructions requires changes to
9591 register allocation which generally has a negative impact on
9592 performance. This option should only be enabled when compiling code
9593 particularly likely to make heavy use of multiply-add instructions.
9596 @opindex mnomulti-add
9597 Do not try and generate multiply-add floating point instructions
9598 @code{polyF} and @code{dotF}. This is the default on all platforms.
9601 @opindex msoft-float
9602 Generate output containing library calls for floating point.
9603 @strong{Warning:} the requisite libraries may not be available.
9605 @item -mieee-compare
9606 @itemx -mno-ieee-compare
9607 @opindex mieee-compare
9608 @opindex mno-ieee-compare
9609 Control whether or not the compiler uses IEEE floating point
9610 comparisons. These handle correctly the case where the result of a
9611 comparison is unordered.
9612 @strong{Warning:} the requisite kernel support may not be available.
9615 @opindex mnobitfield
9616 Do not use the bit-field instructions. On some machines it is faster to
9617 use shifting and masking operations. This is the default for the pc532.
9621 Do use the bit-field instructions. This is the default for all platforms
9626 Use a different function-calling convention, in which functions
9627 that take a fixed number of arguments return pop their
9628 arguments on return with the @code{ret} instruction.
9630 This calling convention is incompatible with the one normally
9631 used on Unix, so you cannot use it if you need to call libraries
9632 compiled with the Unix compiler.
9634 Also, you must provide function prototypes for all functions that
9635 take variable numbers of arguments (including @code{printf});
9636 otherwise incorrect code will be generated for calls to those
9639 In addition, seriously incorrect code will result if you call a
9640 function with too many arguments. (Normally, extra arguments are
9641 harmlessly ignored.)
9643 This option takes its name from the 680x0 @code{rtd} instruction.
9648 Use a different function-calling convention where the first two arguments
9649 are passed in registers.
9651 This calling convention is incompatible with the one normally
9652 used on Unix, so you cannot use it if you need to call libraries
9653 compiled with the Unix compiler.
9656 @opindex mnoregparam
9657 Do not pass any arguments in registers. This is the default for all
9662 It is OK to use the sb as an index register which is always loaded with
9663 zero. This is the default for the pc532-netbsd target.
9667 The sb register is not available for use or has not been initialized to
9668 zero by the run time system. This is the default for all targets except
9669 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9670 @option{-fpic} is set.
9674 Many ns32000 series addressing modes use displacements of up to 512MB@.
9675 If an address is above 512MB then displacements from zero can not be used.
9676 This option causes code to be generated which can be loaded above 512MB@.
9677 This may be useful for operating systems or ROM code.
9681 Assume code will be loaded in the first 512MB of virtual address space.
9682 This is the default for all platforms.
9688 @subsection AVR Options
9691 These options are defined for AVR implementations:
9694 @item -mmcu=@var{mcu}
9696 Specify ATMEL AVR instruction set or MCU type.
9698 Instruction set avr1 is for the minimal AVR core, not supported by the C
9699 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9700 attiny11, attiny12, attiny15, attiny28).
9702 Instruction set avr2 (default) is for the classic AVR core with up to
9703 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9704 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9705 at90c8534, at90s8535).
9707 Instruction set avr3 is for the classic AVR core with up to 128K program
9708 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9710 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9711 memory space (MCU types: atmega8, atmega83, atmega85).
9713 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9714 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9715 atmega64, atmega128, at43usb355, at94k).
9719 Output instruction sizes to the asm file.
9721 @item -minit-stack=@var{N}
9722 @opindex minit-stack
9723 Specify the initial stack address, which may be a symbol or numeric value,
9724 @samp{__stack} is the default.
9726 @item -mno-interrupts
9727 @opindex mno-interrupts
9728 Generated code is not compatible with hardware interrupts.
9729 Code size will be smaller.
9731 @item -mcall-prologues
9732 @opindex mcall-prologues
9733 Functions prologues/epilogues expanded as call to appropriate
9734 subroutines. Code size will be smaller.
9736 @item -mno-tablejump
9737 @opindex mno-tablejump
9738 Do not generate tablejump insns which sometimes increase code size.
9741 @opindex mtiny-stack
9742 Change only the low 8 bits of the stack pointer.
9746 @subsection MCore Options
9747 @cindex MCore options
9749 These are the @samp{-m} options defined for the Motorola M*Core
9757 @opindex mno-hardlit
9758 Inline constants into the code stream if it can be done in two
9759 instructions or less.
9765 Use the divide instruction. (Enabled by default).
9767 @item -mrelax-immediate
9768 @itemx -mno-relax-immediate
9769 @opindex mrelax-immediate
9770 @opindex mno-relax-immediate
9771 Allow arbitrary sized immediates in bit operations.
9773 @item -mwide-bitfields
9774 @itemx -mno-wide-bitfields
9775 @opindex mwide-bitfields
9776 @opindex mno-wide-bitfields
9777 Always treat bit-fields as int-sized.
9779 @item -m4byte-functions
9780 @itemx -mno-4byte-functions
9781 @opindex m4byte-functions
9782 @opindex mno-4byte-functions
9783 Force all functions to be aligned to a four byte boundary.
9785 @item -mcallgraph-data
9786 @itemx -mno-callgraph-data
9787 @opindex mcallgraph-data
9788 @opindex mno-callgraph-data
9789 Emit callgraph information.
9792 @itemx -mno-slow-bytes
9793 @opindex mslow-bytes
9794 @opindex mno-slow-bytes
9795 Prefer word access when reading byte quantities.
9797 @item -mlittle-endian
9799 @opindex mlittle-endian
9800 @opindex mbig-endian
9801 Generate code for a little endian target.
9807 Generate code for the 210 processor.
9811 @subsection IA-64 Options
9812 @cindex IA-64 Options
9814 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9818 @opindex mbig-endian
9819 Generate code for a big endian target. This is the default for HP-UX@.
9821 @item -mlittle-endian
9822 @opindex mlittle-endian
9823 Generate code for a little endian target. This is the default for AIX5
9830 Generate (or don't) code for the GNU assembler. This is the default.
9831 @c Also, this is the default if the configure option @option{--with-gnu-as}
9838 Generate (or don't) code for the GNU linker. This is the default.
9839 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9844 Generate code that does not use a global pointer register. The result
9845 is not position independent code, and violates the IA-64 ABI@.
9847 @item -mvolatile-asm-stop
9848 @itemx -mno-volatile-asm-stop
9849 @opindex mvolatile-asm-stop
9850 @opindex mno-volatile-asm-stop
9851 Generate (or don't) a stop bit immediately before and after volatile asm
9856 Generate code that works around Itanium B step errata.
9858 @item -mregister-names
9859 @itemx -mno-register-names
9860 @opindex mregister-names
9861 @opindex mno-register-names
9862 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9863 the stacked registers. This may make assembler output more readable.
9869 Disable (or enable) optimizations that use the small data section. This may
9870 be useful for working around optimizer bugs.
9873 @opindex mconstant-gp
9874 Generate code that uses a single constant global pointer value. This is
9875 useful when compiling kernel code.
9879 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9880 This is useful when compiling firmware code.
9882 @item -minline-float-divide-min-latency
9883 @opindex minline-float-divide-min-latency
9884 Generate code for inline divides of floating point values
9885 using the minimum latency algorithm.
9887 @item -minline-float-divide-max-throughput
9888 @opindex minline-float-divide-max-throughput
9889 Generate code for inline divides of floating point values
9890 using the maximum throughput algorithm.
9892 @item -minline-int-divide-min-latency
9893 @opindex minline-int-divide-min-latency
9894 Generate code for inline divides of integer values
9895 using the minimum latency algorithm.
9897 @item -minline-int-divide-max-throughput
9898 @opindex minline-int-divide-max-throughput
9899 Generate code for inline divides of integer values
9900 using the maximum throughput algorithm.
9902 @item -mno-dwarf2-asm
9904 @opindex mno-dwarf2-asm
9905 @opindex mdwarf2-asm
9906 Don't (or do) generate assembler code for the DWARF2 line number debugging
9907 info. This may be useful when not using the GNU assembler.
9909 @item -mfixed-range=@var{register-range}
9910 @opindex mfixed-range
9911 Generate code treating the given register range as fixed registers.
9912 A fixed register is one that the register allocator can not use. This is
9913 useful when compiling kernel code. A register range is specified as
9914 two registers separated by a dash. Multiple register ranges can be
9915 specified separated by a comma.
9917 @item -mearly-stop-bits
9918 @itemx -mno-early-stop-bits
9919 @opindex mearly-stop-bits
9920 @opindex mno-early-stop-bits
9921 Allow stop bits to be placed earlier than immediately preceding the
9922 instruction that triggered the stop bit. This can improve instruction
9923 scheduling, but does not always do so.
9927 @subsection D30V Options
9928 @cindex D30V Options
9930 These @samp{-m} options are defined for D30V implementations:
9935 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
9936 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
9937 memory, which starts at location @code{0x80000000}.
9941 Same as the @option{-mextmem} switch.
9945 Link the @samp{.text} section into onchip text memory, which starts at
9946 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
9947 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
9948 into onchip data memory, which starts at location @code{0x20000000}.
9950 @item -mno-asm-optimize
9951 @itemx -masm-optimize
9952 @opindex mno-asm-optimize
9953 @opindex masm-optimize
9954 Disable (enable) passing @option{-O} to the assembler when optimizing.
9955 The assembler uses the @option{-O} option to automatically parallelize
9956 adjacent short instructions where possible.
9958 @item -mbranch-cost=@var{n}
9959 @opindex mbranch-cost
9960 Increase the internal costs of branches to @var{n}. Higher costs means
9961 that the compiler will issue more instructions to avoid doing a branch.
9964 @item -mcond-exec=@var{n}
9966 Specify the maximum number of conditionally executed instructions that
9967 replace a branch. The default is 4.
9970 @node S/390 and zSeries Options
9971 @subsection S/390 and zSeries Options
9972 @cindex S/390 and zSeries Options
9974 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
9979 @opindex mhard-float
9980 @opindex msoft-float
9981 Use (do not use) the hardware floating-point instructions and registers
9982 for floating-point operations. When @option{-msoft-float} is specified,
9983 functions in @file{libgcc.a} will be used to perform floating-point
9984 operations. When @option{-mhard-float} is specified, the compiler
9985 generates IEEE floating-point instructions. This is the default.
9988 @itemx -mno-backchain
9990 @opindex mno-backchain
9991 Generate (or do not generate) code which maintains an explicit
9992 backchain within the stack frame that points to the caller's frame.
9993 This is currently needed to allow debugging. The default is to
9994 generate the backchain.
9997 @itemx -mno-small-exec
9998 @opindex msmall-exec
9999 @opindex mno-small-exec
10000 Generate (or do not generate) code using the @code{bras} instruction
10001 to do subroutine calls.
10002 This only works reliably if the total executable size does not
10003 exceed 64k. The default is to use the @code{basr} instruction instead,
10004 which does not have this limitation.
10010 When @option{-m31} is specified, generate code compliant to the
10011 Linux for S/390 ABI@. When @option{-m64} is specified, generate
10012 code compliant to the Linux for zSeries ABI@. This allows GCC in
10013 particular to generate 64-bit instructions. For the @samp{s390}
10014 targets, the default is @option{-m31}, while the @samp{s390x}
10015 targets default to @option{-m64}.
10021 When @option{-mzarch} is specified, generate code using the
10022 instructions available on z/Architecture.
10023 When @option{-mesa} is specified, generate code using the
10024 instructions available on ESA/390. Note that @option{-mesa} is
10025 not possible with @option{-m64}.
10026 For the @samp{s390} targets, the default is @option{-mesa},
10027 while the @samp{s390x} targets default to @option{-mzarch}.
10033 Generate (or do not generate) code using the @code{mvcle} instruction
10034 to perform block moves. When @option{-mno-mvcle} is specified,
10035 use a @code{mvc} loop instead. This is the default.
10041 Print (or do not print) additional debug information when compiling.
10042 The default is to not print debug information.
10044 @item -march=@var{arch}
10046 Generate code that will run on @var{arch}, which is the name of system
10047 representing a certain processor type. Possible values for
10048 @var{cpu-type} are @samp{g5}, @samp{g6} and @samp{z900}.
10050 @item -mtune=@var{arch}
10052 Tune to @var{cpu-type} everything applicable about the generated code,
10053 except for the ABI and the set of available instructions.
10054 The list of @var{arch} values is the same as for @option{-march}.
10059 @subsection CRIS Options
10060 @cindex CRIS Options
10062 These options are defined specifically for the CRIS ports.
10065 @item -march=@var{architecture-type}
10066 @itemx -mcpu=@var{architecture-type}
10069 Generate code for the specified architecture. The choices for
10070 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10071 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
10072 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10075 @item -mtune=@var{architecture-type}
10077 Tune to @var{architecture-type} everything applicable about the generated
10078 code, except for the ABI and the set of available instructions. The
10079 choices for @var{architecture-type} are the same as for
10080 @option{-march=@var{architecture-type}}.
10082 @item -mmax-stack-frame=@var{n}
10083 @opindex mmax-stack-frame
10084 Warn when the stack frame of a function exceeds @var{n} bytes.
10086 @item -melinux-stacksize=@var{n}
10087 @opindex melinux-stacksize
10088 Only available with the @samp{cris-axis-aout} target. Arranges for
10089 indications in the program to the kernel loader that the stack of the
10090 program should be set to @var{n} bytes.
10096 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10097 @option{-march=v3} and @option{-march=v8} respectively.
10101 Enable CRIS-specific verbose debug-related information in the assembly
10102 code. This option also has the effect to turn off the @samp{#NO_APP}
10103 formatted-code indicator to the assembler at the beginning of the
10108 Do not use condition-code results from previous instruction; always emit
10109 compare and test instructions before use of condition codes.
10111 @item -mno-side-effects
10112 @opindex mno-side-effects
10113 Do not emit instructions with side-effects in addressing modes other than
10116 @item -mstack-align
10117 @itemx -mno-stack-align
10118 @itemx -mdata-align
10119 @itemx -mno-data-align
10120 @itemx -mconst-align
10121 @itemx -mno-const-align
10122 @opindex mstack-align
10123 @opindex mno-stack-align
10124 @opindex mdata-align
10125 @opindex mno-data-align
10126 @opindex mconst-align
10127 @opindex mno-const-align
10128 These options (no-options) arranges (eliminate arrangements) for the
10129 stack-frame, individual data and constants to be aligned for the maximum
10130 single data access size for the chosen CPU model. The default is to
10131 arrange for 32-bit alignment. ABI details such as structure layout are
10132 not affected by these options.
10140 Similar to the stack- data- and const-align options above, these options
10141 arrange for stack-frame, writable data and constants to all be 32-bit,
10142 16-bit or 8-bit aligned. The default is 32-bit alignment.
10144 @item -mno-prologue-epilogue
10145 @itemx -mprologue-epilogue
10146 @opindex mno-prologue-epilogue
10147 @opindex mprologue-epilogue
10148 With @option{-mno-prologue-epilogue}, the normal function prologue and
10149 epilogue that sets up the stack-frame are omitted and no return
10150 instructions or return sequences are generated in the code. Use this
10151 option only together with visual inspection of the compiled code: no
10152 warnings or errors are generated when call-saved registers must be saved,
10153 or storage for local variable needs to be allocated.
10157 @opindex mno-gotplt
10159 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10160 instruction sequences that load addresses for functions from the PLT part
10161 of the GOT rather than (traditional on other architectures) calls to the
10162 PLT. The default is @option{-mgotplt}.
10166 Legacy no-op option only recognized with the cris-axis-aout target.
10170 Legacy no-op option only recognized with the cris-axis-elf and
10171 cris-axis-linux-gnu targets.
10175 Only recognized with the cris-axis-aout target, where it selects a
10176 GNU/linux-like multilib, include files and instruction set for
10177 @option{-march=v8}.
10181 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10185 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
10186 to link with input-output functions from a simulator library. Code,
10187 initialized data and zero-initialized data are allocated consecutively.
10191 Like @option{-sim}, but pass linker options to locate initialized data at
10192 0x40000000 and zero-initialized data at 0x80000000.
10196 @subsection MMIX Options
10197 @cindex MMIX Options
10199 These options are defined for the MMIX:
10203 @itemx -mno-libfuncs
10205 @opindex mno-libfuncs
10206 Specify that intrinsic library functions are being compiled, passing all
10207 values in registers, no matter the size.
10210 @itemx -mno-epsilon
10212 @opindex mno-epsilon
10213 Generate floating-point comparison instructions that compare with respect
10214 to the @code{rE} epsilon register.
10216 @item -mabi=mmixware
10218 @opindex mabi-mmixware
10220 Generate code that passes function parameters and return values that (in
10221 the called function) are seen as registers @code{$0} and up, as opposed to
10222 the GNU ABI which uses global registers @code{$231} and up.
10224 @item -mzero-extend
10225 @itemx -mno-zero-extend
10226 @opindex mzero-extend
10227 @opindex mno-zero-extend
10228 When reading data from memory in sizes shorter than 64 bits, use (do not
10229 use) zero-extending load instructions by default, rather than
10230 sign-extending ones.
10233 @itemx -mno-knuthdiv
10235 @opindex mno-knuthdiv
10236 Make the result of a division yielding a remainder have the same sign as
10237 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10238 remainder follows the sign of the dividend. Both methods are
10239 arithmetically valid, the latter being almost exclusively used.
10241 @item -mtoplevel-symbols
10242 @itemx -mno-toplevel-symbols
10243 @opindex mtoplevel-symbols
10244 @opindex mno-toplevel-symbols
10245 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10246 code can be used with the @code{PREFIX} assembly directive.
10250 Generate an executable in the ELF format, rather than the default
10251 @samp{mmo} format used by the @command{mmix} simulator.
10253 @item -mbranch-predict
10254 @itemx -mno-branch-predict
10255 @opindex mbranch-predict
10256 @opindex mno-branch-predict
10257 Use (do not use) the probable-branch instructions, when static branch
10258 prediction indicates a probable branch.
10260 @item -mbase-addresses
10261 @itemx -mno-base-addresses
10262 @opindex mbase-addresses
10263 @opindex mno-base-addresses
10264 Generate (do not generate) code that uses @emph{base addresses}. Using a
10265 base address automatically generates a request (handled by the assembler
10266 and the linker) for a constant to be set up in a global register. The
10267 register is used for one or more base address requests within the range 0
10268 to 255 from the value held in the register. The generally leads to short
10269 and fast code, but the number of different data items that can be
10270 addressed is limited. This means that a program that uses lots of static
10271 data may require @option{-mno-base-addresses}.
10273 @item -msingle-exit
10274 @itemx -mno-single-exit
10275 @opindex msingle-exit
10276 @opindex mno-single-exit
10277 Force (do not force) generated code to have a single exit point in each
10281 @node PDP-11 Options
10282 @subsection PDP-11 Options
10283 @cindex PDP-11 Options
10285 These options are defined for the PDP-11:
10290 Use hardware FPP floating point. This is the default. (FIS floating
10291 point on the PDP-11/40 is not supported.)
10294 @opindex msoft-float
10295 Do not use hardware floating point.
10299 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10303 Return floating-point results in memory. This is the default.
10307 Generate code for a PDP-11/40.
10311 Generate code for a PDP-11/45. This is the default.
10315 Generate code for a PDP-11/10.
10317 @item -mbcopy-builtin
10318 @opindex bcopy-builtin
10319 Use inline @code{movstrhi} patterns for copying memory. This is the
10324 Do not use inline @code{movstrhi} patterns for copying memory.
10330 Use 16-bit @code{int}. This is the default.
10336 Use 32-bit @code{int}.
10339 @itemx -mno-float32
10341 @opindex mno-float32
10342 Use 64-bit @code{float}. This is the default.
10347 @opindex mno-float64
10348 Use 32-bit @code{float}.
10352 Use @code{abshi2} pattern. This is the default.
10356 Do not use @code{abshi2} pattern.
10358 @item -mbranch-expensive
10359 @opindex mbranch-expensive
10360 Pretend that branches are expensive. This is for experimenting with
10361 code generation only.
10363 @item -mbranch-cheap
10364 @opindex mbranch-cheap
10365 Do not pretend that branches are expensive. This is the default.
10369 Generate code for a system with split I&D.
10373 Generate code for a system without split I&D. This is the default.
10377 Use Unix assembler syntax. This is the default when configured for
10378 @samp{pdp11-*-bsd}.
10382 Use DEC assembler syntax. This is the default when configured for any
10383 PDP-11 target other than @samp{pdp11-*-bsd}.
10386 @node Xstormy16 Options
10387 @subsection Xstormy16 Options
10388 @cindex Xstormy16 Options
10390 These options are defined for Xstormy16:
10395 Choose startup files and linker script suitable for the simulator.
10399 @subsection FRV Options
10400 @cindex FRV Options
10406 Only use the first 32 general purpose registers.
10411 Use all 64 general purpose registers.
10416 Use only the first 32 floating point registers.
10421 Use all 64 floating point registers
10424 @opindex mhard-float
10426 Use hardware instructions for floating point operations.
10429 @opindex msoft-float
10431 Use library routines for floating point operations.
10436 Dynamically allocate condition code registers.
10441 Do not try to dynamically allocate condition code registers, only
10442 use @code{icc0} and @code{fcc0}.
10447 Change ABI to use double word insns.
10452 Do not use double word instructions.
10457 Use floating point double instructions.
10460 @opindex mno-double
10462 Do not use floating point double instructions.
10467 Use media instructions.
10472 Do not use media instructions.
10477 Use multiply and add/subtract instructions.
10480 @opindex mno-muladd
10482 Do not use multiply and add/subtract instructions.
10484 @item -mlibrary-pic
10485 @opindex mlibrary-pic
10487 Enable PIC support for building libraries
10492 Use only the first four media accumulator registers.
10497 Use all eight media accumulator registers.
10502 Pack VLIW instructions.
10507 Do not pack VLIW instructions.
10510 @opindex mno-eflags
10512 Do not mark ABI switches in e_flags.
10515 @opindex mcond-move
10517 Enable the use of conditional-move instructions (default).
10519 This switch is mainly for debugging the compiler and will likely be removed
10520 in a future version.
10522 @item -mno-cond-move
10523 @opindex mno-cond-move
10525 Disable the use of conditional-move instructions.
10527 This switch is mainly for debugging the compiler and will likely be removed
10528 in a future version.
10533 Enable the use of conditional set instructions (default).
10535 This switch is mainly for debugging the compiler and will likely be removed
10536 in a future version.
10541 Disable the use of conditional set instructions.
10543 This switch is mainly for debugging the compiler and will likely be removed
10544 in a future version.
10547 @opindex mcond-exec
10549 Enable the use of conditional execution (default).
10551 This switch is mainly for debugging the compiler and will likely be removed
10552 in a future version.
10554 @item -mno-cond-exec
10555 @opindex mno-cond-exec
10557 Disable the use of conditional execution.
10559 This switch is mainly for debugging the compiler and will likely be removed
10560 in a future version.
10562 @item -mvliw-branch
10563 @opindex mvliw-branch
10565 Run a pass to pack branches into VLIW instructions (default).
10567 This switch is mainly for debugging the compiler and will likely be removed
10568 in a future version.
10570 @item -mno-vliw-branch
10571 @opindex mno-vliw-branch
10573 Do not run a pass to pack branches into VLIW instructions.
10575 This switch is mainly for debugging the compiler and will likely be removed
10576 in a future version.
10578 @item -mmulti-cond-exec
10579 @opindex mmulti-cond-exec
10581 Enable optimization of @code{&&} and @code{||} in conditional execution
10584 This switch is mainly for debugging the compiler and will likely be removed
10585 in a future version.
10587 @item -mno-multi-cond-exec
10588 @opindex mno-multi-cond-exec
10590 Disable optimization of @code{&&} and @code{||} in conditional execution.
10592 This switch is mainly for debugging the compiler and will likely be removed
10593 in a future version.
10595 @item -mnested-cond-exec
10596 @opindex mnested-cond-exec
10598 Enable nested conditional execution optimizations (default).
10600 This switch is mainly for debugging the compiler and will likely be removed
10601 in a future version.
10603 @item -mno-nested-cond-exec
10604 @opindex mno-nested-cond-exec
10606 Disable nested conditional execution optimizations.
10608 This switch is mainly for debugging the compiler and will likely be removed
10609 in a future version.
10611 @item -mtomcat-stats
10612 @opindex mtomcat-stats
10614 Cause gas to print out tomcat statistics.
10616 @item -mcpu=@var{cpu}
10619 Select the processor type for which to generate code. Possible values are
10620 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
10625 @node Xtensa Options
10626 @subsection Xtensa Options
10627 @cindex Xtensa Options
10629 The Xtensa architecture is designed to support many different
10630 configurations. The compiler's default options can be set to match a
10631 particular Xtensa configuration by copying a configuration file into the
10632 GCC sources when building GCC@. The options below may be used to
10633 override the default options.
10637 @itemx -mlittle-endian
10638 @opindex mbig-endian
10639 @opindex mlittle-endian
10640 Specify big-endian or little-endian byte ordering for the target Xtensa
10644 @itemx -mno-density
10646 @opindex mno-density
10647 Enable or disable use of the optional Xtensa code density instructions.
10650 @itemx -mno-const16
10652 @opindex mno-const16
10653 Enable or disable use of CONST16 instructions for loading constant values.
10654 The CONST16 instruction is currently not a standard option from Tensilica.
10655 When enabled, CONST16 instructions are always used in place of the standard
10656 L32R instructions. The use of CONST16 is enabled by default only if the
10657 L32R instruction is not available.
10663 Enable or disable use of the Xtensa MAC16 option. When enabled, GCC
10664 will generate MAC16 instructions from standard C code, with the
10665 limitation that it will use neither the MR register file nor any
10666 instruction that operates on the MR registers. When this option is
10667 disabled, GCC will translate 16-bit multiply/accumulate operations to a
10668 combination of core instructions and library calls, depending on whether
10669 any other multiplier options are enabled.
10675 Enable or disable use of the 16-bit integer multiplier option. When
10676 enabled, the compiler will generate 16-bit multiply instructions for
10677 multiplications of 16 bits or smaller in standard C code. When this
10678 option is disabled, the compiler will either use 32-bit multiply or
10679 MAC16 instructions if they are available or generate library calls to
10680 perform the multiply operations using shifts and adds.
10686 Enable or disable use of the 32-bit integer multiplier option. When
10687 enabled, the compiler will generate 32-bit multiply instructions for
10688 multiplications of 32 bits or smaller in standard C code. When this
10689 option is disabled, the compiler will generate library calls to perform
10690 the multiply operations using either shifts and adds or 16-bit multiply
10691 instructions if they are available.
10697 Enable or disable use of the optional normalization shift amount
10698 (@code{NSA}) instructions to implement the built-in @code{ffs} function.
10703 @opindex mno-minmax
10704 Enable or disable use of the optional minimum and maximum value
10711 Enable or disable use of the optional sign extend (@code{SEXT})
10715 @itemx -mno-booleans
10717 @opindex mno-booleans
10718 Enable or disable support for the boolean register file used by Xtensa
10719 coprocessors. This is not typically useful by itself but may be
10720 required for other options that make use of the boolean registers (e.g.,
10721 the floating-point option).
10724 @itemx -msoft-float
10725 @opindex mhard-float
10726 @opindex msoft-float
10727 Enable or disable use of the floating-point option. When enabled, GCC
10728 generates floating-point instructions for 32-bit @code{float}
10729 operations. When this option is disabled, GCC generates library calls
10730 to emulate 32-bit floating-point operations using integer instructions.
10731 Regardless of this option, 64-bit @code{double} operations are always
10732 emulated with calls to library functions.
10735 @itemx -mno-fused-madd
10736 @opindex mfused-madd
10737 @opindex mno-fused-madd
10738 Enable or disable use of fused multiply/add and multiply/subtract
10739 instructions in the floating-point option. This has no effect if the
10740 floating-point option is not also enabled. Disabling fused multiply/add
10741 and multiply/subtract instructions forces the compiler to use separate
10742 instructions for the multiply and add/subtract operations. This may be
10743 desirable in some cases where strict IEEE 754-compliant results are
10744 required: the fused multiply add/subtract instructions do not round the
10745 intermediate result, thereby producing results with @emph{more} bits of
10746 precision than specified by the IEEE standard. Disabling fused multiply
10747 add/subtract instructions also ensures that the program output is not
10748 sensitive to the compiler's ability to combine multiply and add/subtract
10751 @item -mserialize-volatile
10752 @itemx -mno-serialize-volatile
10753 @opindex mserialize-volatile
10754 @opindex mno-serialize-volatile
10755 When this option is enabled, GCC inserts @code{MEMW} instructions before
10756 @code{volatile} memory references to guarantee sequential consistency.
10757 The default is @option{-mserialize-volatile}. Use
10758 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
10760 @item -mtext-section-literals
10761 @itemx -mno-text-section-literals
10762 @opindex mtext-section-literals
10763 @opindex mno-text-section-literals
10764 Control the treatment of literal pools. The default is
10765 @option{-mno-text-section-literals}, which places literals in a separate
10766 section in the output file. This allows the literal pool to be placed
10767 in a data RAM/ROM, and it also allows the linker to combine literal
10768 pools from separate object files to remove redundant literals and
10769 improve code size. With @option{-mtext-section-literals}, the literals
10770 are interspersed in the text section in order to keep them as close as
10771 possible to their references. This may be necessary for large assembly
10774 @item -mtarget-align
10775 @itemx -mno-target-align
10776 @opindex mtarget-align
10777 @opindex mno-target-align
10778 When this option is enabled, GCC instructs the assembler to
10779 automatically align instructions to reduce branch penalties at the
10780 expense of some code density. The assembler attempts to widen density
10781 instructions to align branch targets and the instructions following call
10782 instructions. If there are not enough preceding safe density
10783 instructions to align a target, no widening will be performed. The
10784 default is @option{-mtarget-align}. These options do not affect the
10785 treatment of auto-aligned instructions like @code{LOOP}, which the
10786 assembler will always align, either by widening density instructions or
10787 by inserting no-op instructions.
10790 @itemx -mno-longcalls
10791 @opindex mlongcalls
10792 @opindex mno-longcalls
10793 When this option is enabled, GCC instructs the assembler to translate
10794 direct calls to indirect calls unless it can determine that the target
10795 of a direct call is in the range allowed by the call instruction. This
10796 translation typically occurs for calls to functions in other source
10797 files. Specifically, the assembler translates a direct @code{CALL}
10798 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
10799 The default is @option{-mno-longcalls}. This option should be used in
10800 programs where the call target can potentially be out of range. This
10801 option is implemented in the assembler, not the compiler, so the
10802 assembly code generated by GCC will still show direct call
10803 instructions---look at the disassembled object code to see the actual
10804 instructions. Note that the assembler will use an indirect call for
10805 every cross-file call, not just those that really will be out of range.
10808 @node Code Gen Options
10809 @section Options for Code Generation Conventions
10810 @cindex code generation conventions
10811 @cindex options, code generation
10812 @cindex run-time options
10814 These machine-independent options control the interface conventions
10815 used in code generation.
10817 Most of them have both positive and negative forms; the negative form
10818 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
10819 one of the forms is listed---the one which is not the default. You
10820 can figure out the other form by either removing @samp{no-} or adding
10824 @item -fbounds-check
10825 @opindex fbounds-check
10826 For front-ends that support it, generate additional code to check that
10827 indices used to access arrays are within the declared range. This is
10828 currently only supported by the Java and Fortran 77 front-ends, where
10829 this option defaults to true and false respectively.
10833 This option generates traps for signed overflow on addition, subtraction,
10834 multiplication operations.
10837 @opindex fexceptions
10838 Enable exception handling. Generates extra code needed to propagate
10839 exceptions. For some targets, this implies GCC will generate frame
10840 unwind information for all functions, which can produce significant data
10841 size overhead, although it does not affect execution. If you do not
10842 specify this option, GCC will enable it by default for languages like
10843 C++ which normally require exception handling, and disable it for
10844 languages like C that do not normally require it. However, you may need
10845 to enable this option when compiling C code that needs to interoperate
10846 properly with exception handlers written in C++. You may also wish to
10847 disable this option if you are compiling older C++ programs that don't
10848 use exception handling.
10850 @item -fnon-call-exceptions
10851 @opindex fnon-call-exceptions
10852 Generate code that allows trapping instructions to throw exceptions.
10853 Note that this requires platform-specific runtime support that does
10854 not exist everywhere. Moreover, it only allows @emph{trapping}
10855 instructions to throw exceptions, i.e.@: memory references or floating
10856 point instructions. It does not allow exceptions to be thrown from
10857 arbitrary signal handlers such as @code{SIGALRM}.
10859 @item -funwind-tables
10860 @opindex funwind-tables
10861 Similar to @option{-fexceptions}, except that it will just generate any needed
10862 static data, but will not affect the generated code in any other way.
10863 You will normally not enable this option; instead, a language processor
10864 that needs this handling would enable it on your behalf.
10866 @item -fasynchronous-unwind-tables
10867 @opindex funwind-tables
10868 Generate unwind table in dwarf2 format, if supported by target machine. The
10869 table is exact at each instruction boundary, so it can be used for stack
10870 unwinding from asynchronous events (such as debugger or garbage collector).
10872 @item -fpcc-struct-return
10873 @opindex fpcc-struct-return
10874 Return ``short'' @code{struct} and @code{union} values in memory like
10875 longer ones, rather than in registers. This convention is less
10876 efficient, but it has the advantage of allowing intercallability between
10877 GCC-compiled files and files compiled with other compilers, particularly
10878 the Portable C Compiler (pcc).
10880 The precise convention for returning structures in memory depends
10881 on the target configuration macros.
10883 Short structures and unions are those whose size and alignment match
10884 that of some integer type.
10886 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
10887 switch is not binary compatible with code compiled with the
10888 @option{-freg-struct-return} switch.
10889 Use it to conform to a non-default application binary interface.
10891 @item -freg-struct-return
10892 @opindex freg-struct-return
10893 Return @code{struct} and @code{union} values in registers when possible.
10894 This is more efficient for small structures than
10895 @option{-fpcc-struct-return}.
10897 If you specify neither @option{-fpcc-struct-return} nor
10898 @option{-freg-struct-return}, GCC defaults to whichever convention is
10899 standard for the target. If there is no standard convention, GCC
10900 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
10901 the principal compiler. In those cases, we can choose the standard, and
10902 we chose the more efficient register return alternative.
10904 @strong{Warning:} code compiled with the @option{-freg-struct-return}
10905 switch is not binary compatible with code compiled with the
10906 @option{-fpcc-struct-return} switch.
10907 Use it to conform to a non-default application binary interface.
10909 @item -fshort-enums
10910 @opindex fshort-enums
10911 Allocate to an @code{enum} type only as many bytes as it needs for the
10912 declared range of possible values. Specifically, the @code{enum} type
10913 will be equivalent to the smallest integer type which has enough room.
10915 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
10916 code that is not binary compatible with code generated without that switch.
10917 Use it to conform to a non-default application binary interface.
10919 @item -fshort-double
10920 @opindex fshort-double
10921 Use the same size for @code{double} as for @code{float}.
10923 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
10924 code that is not binary compatible with code generated without that switch.
10925 Use it to conform to a non-default application binary interface.
10927 @item -fshort-wchar
10928 @opindex fshort-wchar
10929 Override the underlying type for @samp{wchar_t} to be @samp{short
10930 unsigned int} instead of the default for the target. This option is
10931 useful for building programs to run under WINE@.
10933 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
10934 code that is not binary compatible with code generated without that switch.
10935 Use it to conform to a non-default application binary interface.
10937 @item -fshared-data
10938 @opindex fshared-data
10939 Requests that the data and non-@code{const} variables of this
10940 compilation be shared data rather than private data. The distinction
10941 makes sense only on certain operating systems, where shared data is
10942 shared between processes running the same program, while private data
10943 exists in one copy per process.
10946 @opindex fno-common
10947 In C, allocate even uninitialized global variables in the data section of the
10948 object file, rather than generating them as common blocks. This has the
10949 effect that if the same variable is declared (without @code{extern}) in
10950 two different compilations, you will get an error when you link them.
10951 The only reason this might be useful is if you wish to verify that the
10952 program will work on other systems which always work this way.
10956 Ignore the @samp{#ident} directive.
10958 @item -fno-gnu-linker
10959 @opindex fno-gnu-linker
10960 Do not output global initializations (such as C++ constructors and
10961 destructors) in the form used by the GNU linker (on systems where the GNU
10962 linker is the standard method of handling them). Use this option when
10963 you want to use a non-GNU linker, which also requires using the
10964 @command{collect2} program to make sure the system linker includes
10965 constructors and destructors. (@command{collect2} is included in the GCC
10966 distribution.) For systems which @emph{must} use @command{collect2}, the
10967 compiler driver @command{gcc} is configured to do this automatically.
10969 @item -finhibit-size-directive
10970 @opindex finhibit-size-directive
10971 Don't output a @code{.size} assembler directive, or anything else that
10972 would cause trouble if the function is split in the middle, and the
10973 two halves are placed at locations far apart in memory. This option is
10974 used when compiling @file{crtstuff.c}; you should not need to use it
10977 @item -fverbose-asm
10978 @opindex fverbose-asm
10979 Put extra commentary information in the generated assembly code to
10980 make it more readable. This option is generally only of use to those
10981 who actually need to read the generated assembly code (perhaps while
10982 debugging the compiler itself).
10984 @option{-fno-verbose-asm}, the default, causes the
10985 extra information to be omitted and is useful when comparing two assembler
10990 @cindex global offset table
10992 Generate position-independent code (PIC) suitable for use in a shared
10993 library, if supported for the target machine. Such code accesses all
10994 constant addresses through a global offset table (GOT)@. The dynamic
10995 loader resolves the GOT entries when the program starts (the dynamic
10996 loader is not part of GCC; it is part of the operating system). If
10997 the GOT size for the linked executable exceeds a machine-specific
10998 maximum size, you get an error message from the linker indicating that
10999 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11000 instead. (These maximums are 16k on the m88k, 8k on the SPARC, and 32k
11001 on the m68k and RS/6000. The 386 has no such limit.)
11003 Position-independent code requires special support, and therefore works
11004 only on certain machines. For the 386, GCC supports PIC for System V
11005 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11006 position-independent.
11010 If supported for the target machine, emit position-independent code,
11011 suitable for dynamic linking and avoiding any limit on the size of the
11012 global offset table. This option makes a difference on the m68k, m88k,
11015 Position-independent code requires special support, and therefore works
11016 only on certain machines.
11018 @item -ffixed-@var{reg}
11020 Treat the register named @var{reg} as a fixed register; generated code
11021 should never refer to it (except perhaps as a stack pointer, frame
11022 pointer or in some other fixed role).
11024 @var{reg} must be the name of a register. The register names accepted
11025 are machine-specific and are defined in the @code{REGISTER_NAMES}
11026 macro in the machine description macro file.
11028 This flag does not have a negative form, because it specifies a
11031 @item -fcall-used-@var{reg}
11032 @opindex fcall-used
11033 Treat the register named @var{reg} as an allocable register that is
11034 clobbered by function calls. It may be allocated for temporaries or
11035 variables that do not live across a call. Functions compiled this way
11036 will not save and restore the register @var{reg}.
11038 It is an error to used this flag with the frame pointer or stack pointer.
11039 Use of this flag for other registers that have fixed pervasive roles in
11040 the machine's execution model will produce disastrous results.
11042 This flag does not have a negative form, because it specifies a
11045 @item -fcall-saved-@var{reg}
11046 @opindex fcall-saved
11047 Treat the register named @var{reg} as an allocable register saved by
11048 functions. It may be allocated even for temporaries or variables that
11049 live across a call. Functions compiled this way will save and restore
11050 the register @var{reg} if they use it.
11052 It is an error to used this flag with the frame pointer or stack pointer.
11053 Use of this flag for other registers that have fixed pervasive roles in
11054 the machine's execution model will produce disastrous results.
11056 A different sort of disaster will result from the use of this flag for
11057 a register in which function values may be returned.
11059 This flag does not have a negative form, because it specifies a
11062 @item -fpack-struct
11063 @opindex fpack-struct
11064 Pack all structure members together without holes.
11066 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11067 code that is not binary compatible with code generated without that switch.
11068 Additionally, it makes the code suboptimal.
11069 Use it to conform to a non-default application binary interface.
11071 @item -finstrument-functions
11072 @opindex finstrument-functions
11073 Generate instrumentation calls for entry and exit to functions. Just
11074 after function entry and just before function exit, the following
11075 profiling functions will be called with the address of the current
11076 function and its call site. (On some platforms,
11077 @code{__builtin_return_address} does not work beyond the current
11078 function, so the call site information may not be available to the
11079 profiling functions otherwise.)
11082 void __cyg_profile_func_enter (void *this_fn,
11084 void __cyg_profile_func_exit (void *this_fn,
11088 The first argument is the address of the start of the current function,
11089 which may be looked up exactly in the symbol table.
11091 This instrumentation is also done for functions expanded inline in other
11092 functions. The profiling calls will indicate where, conceptually, the
11093 inline function is entered and exited. This means that addressable
11094 versions of such functions must be available. If all your uses of a
11095 function are expanded inline, this may mean an additional expansion of
11096 code size. If you use @samp{extern inline} in your C code, an
11097 addressable version of such functions must be provided. (This is
11098 normally the case anyways, but if you get lucky and the optimizer always
11099 expands the functions inline, you might have gotten away without
11100 providing static copies.)
11102 A function may be given the attribute @code{no_instrument_function}, in
11103 which case this instrumentation will not be done. This can be used, for
11104 example, for the profiling functions listed above, high-priority
11105 interrupt routines, and any functions from which the profiling functions
11106 cannot safely be called (perhaps signal handlers, if the profiling
11107 routines generate output or allocate memory).
11109 @item -fstack-check
11110 @opindex fstack-check
11111 Generate code to verify that you do not go beyond the boundary of the
11112 stack. You should specify this flag if you are running in an
11113 environment with multiple threads, but only rarely need to specify it in
11114 a single-threaded environment since stack overflow is automatically
11115 detected on nearly all systems if there is only one stack.
11117 Note that this switch does not actually cause checking to be done; the
11118 operating system must do that. The switch causes generation of code
11119 to ensure that the operating system sees the stack being extended.
11121 @item -fstack-limit-register=@var{reg}
11122 @itemx -fstack-limit-symbol=@var{sym}
11123 @itemx -fno-stack-limit
11124 @opindex fstack-limit-register
11125 @opindex fstack-limit-symbol
11126 @opindex fno-stack-limit
11127 Generate code to ensure that the stack does not grow beyond a certain value,
11128 either the value of a register or the address of a symbol. If the stack
11129 would grow beyond the value, a signal is raised. For most targets,
11130 the signal is raised before the stack overruns the boundary, so
11131 it is possible to catch the signal without taking special precautions.
11133 For instance, if the stack starts at absolute address @samp{0x80000000}
11134 and grows downwards, you can use the flags
11135 @option{-fstack-limit-symbol=__stack_limit} and
11136 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11137 of 128KB@. Note that this may only work with the GNU linker.
11139 @cindex aliasing of parameters
11140 @cindex parameters, aliased
11141 @item -fargument-alias
11142 @itemx -fargument-noalias
11143 @itemx -fargument-noalias-global
11144 @opindex fargument-alias
11145 @opindex fargument-noalias
11146 @opindex fargument-noalias-global
11147 Specify the possible relationships among parameters and between
11148 parameters and global data.
11150 @option{-fargument-alias} specifies that arguments (parameters) may
11151 alias each other and may alias global storage.@*
11152 @option{-fargument-noalias} specifies that arguments do not alias
11153 each other, but may alias global storage.@*
11154 @option{-fargument-noalias-global} specifies that arguments do not
11155 alias each other and do not alias global storage.
11157 Each language will automatically use whatever option is required by
11158 the language standard. You should not need to use these options yourself.
11160 @item -fleading-underscore
11161 @opindex fleading-underscore
11162 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11163 change the way C symbols are represented in the object file. One use
11164 is to help link with legacy assembly code.
11166 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11167 generate code that is not binary compatible with code generated without that
11168 switch. Use it to conform to a non-default application binary interface.
11169 Not all targets provide complete support for this switch.
11171 @item -ftls-model=@var{model}
11172 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11173 The @var{model} argument should be one of @code{global-dynamic},
11174 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11176 The default without @option{-fpic} is @code{initial-exec}; with
11177 @option{-fpic} the default is @code{global-dynamic}.
11182 @node Environment Variables
11183 @section Environment Variables Affecting GCC
11184 @cindex environment variables
11186 @c man begin ENVIRONMENT
11187 This section describes several environment variables that affect how GCC
11188 operates. Some of them work by specifying directories or prefixes to use
11189 when searching for various kinds of files. Some are used to specify other
11190 aspects of the compilation environment.
11192 Note that you can also specify places to search using options such as
11193 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11194 take precedence over places specified using environment variables, which
11195 in turn take precedence over those specified by the configuration of GCC@.
11196 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11197 GNU Compiler Collection (GCC) Internals}.
11202 @c @itemx LC_COLLATE
11204 @c @itemx LC_MONETARY
11205 @c @itemx LC_NUMERIC
11210 @c @findex LC_COLLATE
11211 @findex LC_MESSAGES
11212 @c @findex LC_MONETARY
11213 @c @findex LC_NUMERIC
11217 These environment variables control the way that GCC uses
11218 localization information that allow GCC to work with different
11219 national conventions. GCC inspects the locale categories
11220 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11221 so. These locale categories can be set to any value supported by your
11222 installation. A typical value is @samp{en_UK} for English in the United
11225 The @env{LC_CTYPE} environment variable specifies character
11226 classification. GCC uses it to determine the character boundaries in
11227 a string; this is needed for some multibyte encodings that contain quote
11228 and escape characters that would otherwise be interpreted as a string
11231 The @env{LC_MESSAGES} environment variable specifies the language to
11232 use in diagnostic messages.
11234 If the @env{LC_ALL} environment variable is set, it overrides the value
11235 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11236 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11237 environment variable. If none of these variables are set, GCC
11238 defaults to traditional C English behavior.
11242 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11243 files. GCC uses temporary files to hold the output of one stage of
11244 compilation which is to be used as input to the next stage: for example,
11245 the output of the preprocessor, which is the input to the compiler
11248 @item GCC_EXEC_PREFIX
11249 @findex GCC_EXEC_PREFIX
11250 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11251 names of the subprograms executed by the compiler. No slash is added
11252 when this prefix is combined with the name of a subprogram, but you can
11253 specify a prefix that ends with a slash if you wish.
11255 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11256 an appropriate prefix to use based on the pathname it was invoked with.
11258 If GCC cannot find the subprogram using the specified prefix, it
11259 tries looking in the usual places for the subprogram.
11261 The default value of @env{GCC_EXEC_PREFIX} is
11262 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
11263 of @code{prefix} when you ran the @file{configure} script.
11265 Other prefixes specified with @option{-B} take precedence over this prefix.
11267 This prefix is also used for finding files such as @file{crt0.o} that are
11270 In addition, the prefix is used in an unusual way in finding the
11271 directories to search for header files. For each of the standard
11272 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
11273 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11274 replacing that beginning with the specified prefix to produce an
11275 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11276 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11277 These alternate directories are searched first; the standard directories
11280 @item COMPILER_PATH
11281 @findex COMPILER_PATH
11282 The value of @env{COMPILER_PATH} is a colon-separated list of
11283 directories, much like @env{PATH}. GCC tries the directories thus
11284 specified when searching for subprograms, if it can't find the
11285 subprograms using @env{GCC_EXEC_PREFIX}.
11288 @findex LIBRARY_PATH
11289 The value of @env{LIBRARY_PATH} is a colon-separated list of
11290 directories, much like @env{PATH}. When configured as a native compiler,
11291 GCC tries the directories thus specified when searching for special
11292 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11293 using GCC also uses these directories when searching for ordinary
11294 libraries for the @option{-l} option (but directories specified with
11295 @option{-L} come first).
11299 @cindex locale definition
11300 This variable is used to pass locale information to the compiler. One way in
11301 which this information is used is to determine the character set to be used
11302 when character literals, string literals and comments are parsed in C and C++.
11303 When the compiler is configured to allow multibyte characters,
11304 the following values for @env{LANG} are recognized:
11308 Recognize JIS characters.
11310 Recognize SJIS characters.
11312 Recognize EUCJP characters.
11315 If @env{LANG} is not defined, or if it has some other value, then the
11316 compiler will use mblen and mbtowc as defined by the default locale to
11317 recognize and translate multibyte characters.
11321 Some additional environments variables affect the behavior of the
11324 @include cppenv.texi
11328 @node Precompiled Headers
11329 @section Using Precompiled Headers
11330 @cindex precompiled headers
11331 @cindex speed of compilation
11333 Often large projects have many header files that are included in every
11334 source file. The time the compiler takes to process these header files
11335 over and over again can account for nearly all of the time required to
11336 build the project. To make builds faster, GCC allows users to
11337 `precompile' a header file; then, if builds can use the precompiled
11338 header file they will be much faster.
11340 To create a precompiled header file, simply compile it as you would any
11341 other file, if necessary using the @option{-x} option to make the driver
11342 treat it as a C or C++ header file. You will probably want to use a
11343 tool like @command{make} to keep the precompiled header up-to-date when
11344 the headers it contains change.
11346 A precompiled header file will be searched for when @code{#include} is
11347 seen in the compilation. As it searches for the included file
11348 (@pxref{Search Path,,Search Path,cpp.info,The C Preprocessor}) the
11349 compiler looks for a precompiled header in each directory just before it
11350 looks for the include file in that directory. The name searched for is
11351 the name specified in the @code{#include} with @samp{.gch} appended. If
11352 the precompiled header file can't be used, it is ignored.
11354 For instance, if you have @code{#include "all.h"}, and you have
11355 @file{all.h.gch} in the same directory as @file{all.h}, then the
11356 precompiled header file will be used if possible, and the original
11357 header will be used otherwise.
11359 Alternatively, you might decide to put the precompiled header file in a
11360 directory and use @option{-I} to ensure that directory is searched
11361 before (or instead of) the directory containing the original header.
11362 Then, if you want to check that the precompiled header file is always
11363 used, you can put a file of the same name as the original header in this
11364 directory containing an @code{#error} command.
11366 This also works with @option{-include}. So yet another way to use
11367 precompiled headers, good for projects not designed with precompiled
11368 header files in mind, is to simply take most of the header files used by
11369 a project, include them from another header file, precompile that header
11370 file, and @option{-include} the precompiled header. If the header files
11371 have guards against multiple inclusion, they will be skipped because
11372 they've already been included (in the precompiled header).
11374 If you need to precompile the same header file for different
11375 languages, targets, or compiler options, you can instead make a
11376 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11377 header in the directory. (It doesn't matter what you call the files
11378 in the directory, every precompiled header in the directory will be
11379 considered.) The first precompiled header encountered in the
11380 directory that is valid for this compilation will be used; they're
11381 searched in no particular order.
11383 There are many other possibilities, limited only by your imagination,
11384 good sense, and the constraints of your build system.
11386 A precompiled header file can be used only when these conditions apply:
11390 Only one precompiled header can be used in a particular compilation.
11392 A precompiled header can't be used once the first C token is seen. You
11393 can have preprocessor directives before a precompiled header; you can
11394 even include a precompiled header from inside another header, so long as
11395 there are no C tokens before the @code{#include}.
11397 The precompiled header file must be produced for the same language as
11398 the current compilation. You can't use a C precompiled header for a C++
11401 The precompiled header file must be produced by the same compiler
11402 version and configuration as the current compilation is using.
11403 The easiest way to guarantee this is to use the same compiler binary
11404 for creating and using precompiled headers.
11406 Any macros defined before the precompiled header (including with
11407 @option{-D}) must either be defined in the same way as when the
11408 precompiled header was generated, or must not affect the precompiled
11409 header, which usually means that the they don't appear in the
11410 precompiled header at all.
11412 Certain command-line options must be defined in the same way as when the
11413 precompiled header was generated. At present, it's not clear which
11414 options are safe to change and which are not; the safest choice is to
11415 use exactly the same options when generating and using the precompiled
11419 For all of these but the last, the compiler will automatically ignore
11420 the precompiled header if the conditions aren't met. For the last item,
11421 some option changes will cause the precompiled header to be rejected,
11422 but not all incompatible option combinations have yet been found. If
11423 you find a new incompatible combination, please consider filing a bug
11424 report, see @ref{Bugs}.
11426 @node Running Protoize
11427 @section Running Protoize
11429 The program @code{protoize} is an optional part of GCC@. You can use
11430 it to add prototypes to a program, thus converting the program to ISO
11431 C in one respect. The companion program @code{unprotoize} does the
11432 reverse: it removes argument types from any prototypes that are found.
11434 When you run these programs, you must specify a set of source files as
11435 command line arguments. The conversion programs start out by compiling
11436 these files to see what functions they define. The information gathered
11437 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11439 After scanning comes actual conversion. The specified files are all
11440 eligible to be converted; any files they include (whether sources or
11441 just headers) are eligible as well.
11443 But not all the eligible files are converted. By default,
11444 @code{protoize} and @code{unprotoize} convert only source and header
11445 files in the current directory. You can specify additional directories
11446 whose files should be converted with the @option{-d @var{directory}}
11447 option. You can also specify particular files to exclude with the
11448 @option{-x @var{file}} option. A file is converted if it is eligible, its
11449 directory name matches one of the specified directory names, and its
11450 name within the directory has not been excluded.
11452 Basic conversion with @code{protoize} consists of rewriting most
11453 function definitions and function declarations to specify the types of
11454 the arguments. The only ones not rewritten are those for varargs
11457 @code{protoize} optionally inserts prototype declarations at the
11458 beginning of the source file, to make them available for any calls that
11459 precede the function's definition. Or it can insert prototype
11460 declarations with block scope in the blocks where undeclared functions
11463 Basic conversion with @code{unprotoize} consists of rewriting most
11464 function declarations to remove any argument types, and rewriting
11465 function definitions to the old-style pre-ISO form.
11467 Both conversion programs print a warning for any function declaration or
11468 definition that they can't convert. You can suppress these warnings
11471 The output from @code{protoize} or @code{unprotoize} replaces the
11472 original source file. The original file is renamed to a name ending
11473 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11474 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11475 for DOS) file already exists, then the source file is simply discarded.
11477 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11478 scan the program and collect information about the functions it uses.
11479 So neither of these programs will work until GCC is installed.
11481 Here is a table of the options you can use with @code{protoize} and
11482 @code{unprotoize}. Each option works with both programs unless
11486 @item -B @var{directory}
11487 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11488 usual directory (normally @file{/usr/local/lib}). This file contains
11489 prototype information about standard system functions. This option
11490 applies only to @code{protoize}.
11492 @item -c @var{compilation-options}
11493 Use @var{compilation-options} as the options when running @command{gcc} to
11494 produce the @samp{.X} files. The special option @option{-aux-info} is
11495 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11497 Note that the compilation options must be given as a single argument to
11498 @code{protoize} or @code{unprotoize}. If you want to specify several
11499 @command{gcc} options, you must quote the entire set of compilation options
11500 to make them a single word in the shell.
11502 There are certain @command{gcc} arguments that you cannot use, because they
11503 would produce the wrong kind of output. These include @option{-g},
11504 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11505 the @var{compilation-options}, they are ignored.
11508 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11509 systems) instead of @samp{.c}. This is convenient if you are converting
11510 a C program to C++. This option applies only to @code{protoize}.
11513 Add explicit global declarations. This means inserting explicit
11514 declarations at the beginning of each source file for each function
11515 that is called in the file and was not declared. These declarations
11516 precede the first function definition that contains a call to an
11517 undeclared function. This option applies only to @code{protoize}.
11519 @item -i @var{string}
11520 Indent old-style parameter declarations with the string @var{string}.
11521 This option applies only to @code{protoize}.
11523 @code{unprotoize} converts prototyped function definitions to old-style
11524 function definitions, where the arguments are declared between the
11525 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11526 uses five spaces as the indentation. If you want to indent with just
11527 one space instead, use @option{-i " "}.
11530 Keep the @samp{.X} files. Normally, they are deleted after conversion
11534 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11535 a prototype declaration for each function in each block which calls the
11536 function without any declaration. This option applies only to
11540 Make no real changes. This mode just prints information about the conversions
11541 that would have been done without @option{-n}.
11544 Make no @samp{.save} files. The original files are simply deleted.
11545 Use this option with caution.
11547 @item -p @var{program}
11548 Use the program @var{program} as the compiler. Normally, the name
11549 @file{gcc} is used.
11552 Work quietly. Most warnings are suppressed.
11555 Print the version number, just like @option{-v} for @command{gcc}.
11558 If you need special compiler options to compile one of your program's
11559 source files, then you should generate that file's @samp{.X} file
11560 specially, by running @command{gcc} on that source file with the
11561 appropriate options and the option @option{-aux-info}. Then run
11562 @code{protoize} on the entire set of files. @code{protoize} will use
11563 the existing @samp{.X} file because it is newer than the source file.
11567 gcc -Dfoo=bar file1.c -aux-info file1.X
11572 You need to include the special files along with the rest in the
11573 @code{protoize} command, even though their @samp{.X} files already
11574 exist, because otherwise they won't get converted.
11576 @xref{Protoize Caveats}, for more information on how to use
11577 @code{protoize} successfully.