1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003 Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
8 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997,
9 1998, 1999, 2000, 2001, 2002, 2003 Free Software Foundation, Inc.
11 Permission is granted to copy, distribute and/or modify this document
12 under the terms of the GNU Free Documentation License, Version 1.2 or
13 any later version published by the Free Software Foundation; with the
14 Invariant Sections being ``GNU General Public License'' and ``Funding
15 Free Software'', the Front-Cover texts being (a) (see below), and with
16 the Back-Cover Texts being (b) (see below). A copy of the license is
17 included in the gfdl(7) man page.
19 (a) The FSF's Front-Cover Text is:
23 (b) The FSF's Back-Cover Text is:
25 You have freedom to copy and modify this GNU Manual, like GNU
26 software. Copies published by the Free Software Foundation raise
27 funds for GNU development.
29 @c Set file name and title for the man page.
31 @settitle GNU project C and C++ compiler
33 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
34 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
35 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
36 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
37 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
38 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
39 [@option{-o} @var{outfile}] @var{infile}@dots{}
41 Only the most useful options are listed here; see below for the
42 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
45 gpl(7), gfdl(7), fsf-funding(7),
46 cpp(1), gcov(1), g77(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
47 and the Info entries for @file{gcc}, @file{cpp}, @file{g77}, @file{as},
48 @file{ld}, @file{binutils} and @file{gdb}.
51 For instructions on reporting bugs, see
52 @w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug}
53 script to report bugs is recommended.
56 See the Info entry for @command{gcc}, or
57 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
58 for contributors to GCC@.
63 @chapter GCC Command Options
64 @cindex GCC command options
65 @cindex command options
66 @cindex options, GCC command
68 @c man begin DESCRIPTION
69 When you invoke GCC, it normally does preprocessing, compilation,
70 assembly and linking. The ``overall options'' allow you to stop this
71 process at an intermediate stage. For example, the @option{-c} option
72 says not to run the linker. Then the output consists of object files
73 output by the assembler.
75 Other options are passed on to one stage of processing. Some options
76 control the preprocessor and others the compiler itself. Yet other
77 options control the assembler and linker; most of these are not
78 documented here, since you rarely need to use any of them.
80 @cindex C compilation options
81 Most of the command line options that you can use with GCC are useful
82 for C programs; when an option is only useful with another language
83 (usually C++), the explanation says so explicitly. If the description
84 for a particular option does not mention a source language, you can use
85 that option with all supported languages.
87 @cindex C++ compilation options
88 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
89 options for compiling C++ programs.
91 @cindex grouping options
92 @cindex options, grouping
93 The @command{gcc} program accepts options and file names as operands. Many
94 options have multi-letter names; therefore multiple single-letter options
95 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
98 @cindex order of options
99 @cindex options, order
100 You can mix options and other arguments. For the most part, the order
101 you use doesn't matter. Order does matter when you use several options
102 of the same kind; for example, if you specify @option{-L} more than once,
103 the directories are searched in the order specified.
105 Many options have long names starting with @samp{-f} or with
106 @samp{-W}---for example, @option{-fforce-mem},
107 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
108 these have both positive and negative forms; the negative form of
109 @option{-ffoo} would be @option{-fno-foo}. This manual documents
110 only one of these two forms, whichever one is not the default.
114 @xref{Option Index}, for an index to GCC's options.
117 * Option Summary:: Brief list of all options, without explanations.
118 * Overall Options:: Controlling the kind of output:
119 an executable, object files, assembler files,
120 or preprocessed source.
121 * Invoking G++:: Compiling C++ programs.
122 * C Dialect Options:: Controlling the variant of C language compiled.
123 * C++ Dialect Options:: Variations on C++.
124 * Objective-C Dialect Options:: Variations on Objective-C.
125 * Language Independent Options:: Controlling how diagnostics should be
127 * Warning Options:: How picky should the compiler be?
128 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
129 * Optimize Options:: How much optimization?
130 * Preprocessor Options:: Controlling header files and macro definitions.
131 Also, getting dependency information for Make.
132 * Assembler Options:: Passing options to the assembler.
133 * Link Options:: Specifying libraries and so on.
134 * Directory Options:: Where to find header files and libraries.
135 Where to find the compiler executable files.
136 * Spec Files:: How to pass switches to sub-processes.
137 * Target Options:: Running a cross-compiler, or an old version of GCC.
138 * Submodel Options:: Specifying minor hardware or convention variations,
139 such as 68010 vs 68020.
140 * Code Gen Options:: Specifying conventions for function calls, data layout
142 * Environment Variables:: Env vars that affect GCC.
143 * Precompiled Headers:: Compiling a header once, and using it many times.
144 * Running Protoize:: Automatically adding or removing function prototypes.
150 @section Option Summary
152 Here is a summary of all the options, grouped by type. Explanations are
153 in the following sections.
156 @item Overall Options
157 @xref{Overall Options,,Options Controlling the Kind of Output}.
158 @gccoptlist{-c -S -E -o @var{file} -pipe -pass-exit-codes @gol
159 -x @var{language} -v -### --help --target-help --version}
161 @item C Language Options
162 @xref{C Dialect Options,,Options Controlling C Dialect}.
163 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
164 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
165 -fhosted -ffreestanding -fms-extensions @gol
166 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
167 -fallow-single-precision -fcond-mismatch @gol
168 -fsigned-bitfields -fsigned-char @gol
169 -funsigned-bitfields -funsigned-char @gol
172 @item C++ Language Options
173 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
174 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
175 -fconserve-space -fno-const-strings @gol
176 -fno-elide-constructors @gol
177 -fno-enforce-eh-specs -fexternal-templates @gol
178 -falt-external-templates @gol
179 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
180 -fno-implicit-templates @gol
181 -fno-implicit-inline-templates @gol
182 -fno-implement-inlines -fms-extensions @gol
183 -fno-nonansi-builtins -fno-operator-names @gol
184 -fno-optional-diags -fpermissive @gol
185 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
186 -fuse-cxa-atexit -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 -Wno-invalid-offsetof @gol
219 -Wlarger-than-@var{len} -Wlong-long @gol
220 -Wmain -Wmissing-braces @gol
221 -Wmissing-format-attribute -Wmissing-noreturn @gol
222 -Wno-multichar -Wno-format-extra-args -Wno-format-y2k @gol
223 -Wno-import -Wnonnull -Wpacked -Wpadded @gol
224 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
225 -Wreturn-type -Wsequence-point -Wshadow @gol
226 -Wsign-compare -Wstrict-aliasing @gol
227 -Wswitch -Wswitch-default -Wswitch-enum @gol
228 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
229 -Wunknown-pragmas -Wunreachable-code @gol
230 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
231 -Wunused-value -Wunused-variable -Wwrite-strings}
233 @item C-only Warning Options
234 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
235 -Wmissing-prototypes -Wnested-externs @gol
236 -Wstrict-prototypes -Wtraditional}
238 @item Debugging Options
239 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
240 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
241 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
242 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
243 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
244 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
245 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
246 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
247 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs @gol
248 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
249 -ftest-coverage -ftime-report @gol
250 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
251 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
252 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
253 -print-multi-directory -print-multi-lib @gol
254 -print-prog-name=@var{program} -print-search-dirs -Q @gol
257 @item Optimization Options
258 @xref{Optimize Options,,Options that Control Optimization}.
259 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
260 -falign-labels=@var{n} -falign-loops=@var{n} @gol
261 -fbranch-probabilities -fprofile-values -fbranch-target-load-optimize @gol
262 -fbranch-target-load-optimize2 -fcaller-saves -fcprop-registers @gol
263 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
264 -fdelayed-branch -fdelete-null-pointer-checks @gol
265 -fexpensive-optimizations -ffast-math -ffloat-store @gol
266 -fforce-addr -fforce-mem -ffunction-sections @gol
267 -fgcse -fgcse-lm -fgcse-sm -floop-optimize -fcrossjumping @gol
268 -fif-conversion -fif-conversion2 @gol
269 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
270 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
271 -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
272 -fno-default-inline -fno-defer-pop @gol
273 -fno-function-cse -fno-guess-branch-probability @gol
274 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
275 -funsafe-math-optimizations -ffinite-math-only @gol
276 -fno-trapping-math -fno-zero-initialized-in-bss @gol
277 -fomit-frame-pointer -foptimize-register-move @gol
278 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
279 -freduce-all-givs -fregmove -frename-registers @gol
280 -freorder-blocks -freorder-functions @gol
281 -frerun-cse-after-loop -frerun-loop-opt @gol
282 -fschedule-insns -fschedule-insns2 @gol
283 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
284 -fsched-spec-load-dangerous -fsched2-use-superblocks @gol
285 -fsched2-use-traces -fsignaling-nans @gol
286 -fsingle-precision-constant -fssa -fssa-ccp -fssa-dce @gol
287 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
288 -funroll-all-loops -funroll-loops -fpeel-loops @gol
289 -funswitch-loops -fold-unroll-loops -fold-unroll-all-loops @gol
290 --param @var{name}=@var{value}
291 -O -O0 -O1 -O2 -O3 -Os}
293 @item Preprocessor Options
294 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
295 @gccoptlist{-A@var{question}=@var{answer} @gol
296 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
297 -C -dD -dI -dM -dN @gol
298 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
299 -idirafter @var{dir} @gol
300 -include @var{file} -imacros @var{file} @gol
301 -iprefix @var{file} -iwithprefix @var{dir} @gol
302 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
303 -M -MM -MF -MG -MP -MQ -MT -nostdinc -P -remap @gol
304 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
305 -Xpreprocessor @var{option}}
307 @item Assembler Option
308 @xref{Assembler Options,,Passing Options to the Assembler}.
309 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
312 @xref{Link Options,,Options for Linking}.
313 @gccoptlist{@var{object-file-name} -l@var{library} @gol
314 -nostartfiles -nodefaultlibs -nostdlib -pie @gol
315 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
316 -Wl,@var{option} -Xlinker @var{option} @gol
319 @item Directory Options
320 @xref{Directory Options,,Options for Directory Search}.
321 @gccoptlist{-B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
324 @c I wrote this xref this way to avoid overfull hbox. -- rms
325 @xref{Target Options}.
326 @gccoptlist{-V @var{version} -b @var{machine}}
328 @item Machine Dependent Options
329 @xref{Submodel Options,,Hardware Models and Configurations}.
331 @emph{M680x0 Options}
332 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
333 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
334 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
335 -malign-int -mstrict-align}
337 @emph{M68hc1x Options}
338 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
339 -mauto-incdec -minmax -mlong-calls -mshort @gol
340 -msoft-reg-count=@var{count}}
343 @gccoptlist{-mg -mgnu -munix}
346 @gccoptlist{-mcpu=@var{cpu-type} @gol
347 -mtune=@var{cpu-type} @gol
348 -mcmodel=@var{code-model} @gol
350 -mapp-regs -mbroken-saverestore -mcypress @gol
351 -mfaster-structs -mflat @gol
352 -mfpu -mhard-float -mhard-quad-float @gol
353 -mimpure-text -mlive-g0 -mno-app-regs @gol
354 -mno-faster-structs -mno-flat -mno-fpu @gol
355 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
356 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
357 -msupersparc -munaligned-doubles -mv8}
360 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
361 -mapcs-26 -mapcs-32 @gol
362 -mapcs-stack-check -mno-apcs-stack-check @gol
363 -mapcs-float -mno-apcs-float @gol
364 -mapcs-reentrant -mno-apcs-reentrant @gol
365 -msched-prolog -mno-sched-prolog @gol
366 -mlittle-endian -mbig-endian -mwords-little-endian @gol
367 -malignment-traps -mno-alignment-traps @gol
368 -msoft-float -mhard-float -mfpe @gol
369 -mthumb-interwork -mno-thumb-interwork @gol
370 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
371 -mstructure-size-boundary=@var{n} @gol
372 -mabort-on-noreturn @gol
373 -mlong-calls -mno-long-calls @gol
374 -msingle-pic-base -mno-single-pic-base @gol
375 -mpic-register=@var{reg} @gol
376 -mnop-fun-dllimport @gol
377 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
378 -mpoke-function-name @gol
380 -mtpcs-frame -mtpcs-leaf-frame @gol
381 -mcaller-super-interworking -mcallee-super-interworking}
383 @emph{MN10200 Options}
386 @emph{MN10300 Options}
387 @gccoptlist{-mmult-bug -mno-mult-bug @gol
388 -mam33 -mno-am33 @gol
389 -mam33-2 -mno-am33-2 @gol
392 @emph{M32R/D Options}
393 @gccoptlist{-m32rx -m32r -mcode-model=@var{model-type} @gol
394 -msdata=@var{sdata-type} -G @var{num}}
397 @gccoptlist{-m88000 -m88100 -m88110 -mbig-pic @gol
398 -mcheck-zero-division -mhandle-large-shift @gol
399 -midentify-revision -mno-check-zero-division @gol
400 -mno-ocs-debug-info -mno-ocs-frame-position @gol
401 -mno-optimize-arg-area -mno-serialize-volatile @gol
402 -mno-underscores -mocs-debug-info @gol
403 -mocs-frame-position -moptimize-arg-area @gol
404 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
405 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
406 -mversion-03.00 -mwarn-passed-structs}
408 @emph{RS/6000 and PowerPC Options}
409 @gccoptlist{-mcpu=@var{cpu-type} @gol
410 -mtune=@var{cpu-type} @gol
411 -mpower -mno-power -mpower2 -mno-power2 @gol
412 -mpowerpc -mpowerpc64 -mno-powerpc @gol
413 -maltivec -mno-altivec @gol
414 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
415 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
416 -mnew-mnemonics -mold-mnemonics @gol
417 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
418 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
419 -malign-power -malign-natural @gol
420 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
421 -mstring -mno-string -mupdate -mno-update @gol
422 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
423 -mstrict-align -mno-strict-align -mrelocatable @gol
424 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
425 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
426 -mdynamic-no-pic @gol
427 -mcall-sysv -mcall-netbsd @gol
428 -maix-struct-return -msvr4-struct-return @gol
429 -mabi=altivec -mabi=no-altivec @gol
430 -mabi=spe -mabi=no-spe @gol
431 -misel=yes -misel=no @gol
432 -mspe=yes -mspe=no @gol
433 -mfloat-gprs=yes -mfloat-gprs=no @gol
434 -mprototype -mno-prototype @gol
435 -msim -mmvme -mads -myellowknife -memb -msdata @gol
436 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
438 @emph{Darwin Options}
439 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
440 -arch_only -bind_at_load -bundle -bundle_loader @gol
441 -client_name -compatibility_version -current_version @gol
442 -dependency-file -dylib_file -dylinker_install_name @gol
443 -dynamic -dynamiclib -exported_symbols_list @gol
444 -filelist -flat_namespace -force_cpusubtype_ALL @gol
445 -force_flat_namespace -headerpad_max_install_names @gol
446 -image_base -init -install_name -keep_private_externs @gol
447 -multi_module -multiply_defined -multiply_defined_unused @gol
448 -noall_load -nomultidefs -noprebind -noseglinkedit @gol
449 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
450 -private_bundle -read_only_relocs -sectalign @gol
451 -sectobjectsymbols -whyload -seg1addr @gol
452 -sectcreate -sectobjectsymbols -sectorder @gol
453 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
454 -segprot -segs_read_only_addr -segs_read_write_addr @gol
455 -single_module -static -sub_library -sub_umbrella @gol
456 -twolevel_namespace -umbrella -undefined @gol
457 -unexported_symbols_list -weak_reference_mismatches @gol
461 @gccoptlist{-mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
462 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
463 -mminimum-fp-blocks -mnohc-struct-return}
466 @gccoptlist{-mabicalls -march=@var{cpu-type} -mtune=@var{cpu=type} @gol
467 -mcpu=@var{cpu-type} -membedded-data -muninit-const-in-rodata @gol
468 -membedded-pic -mfp32 -mfp64 -mfused-madd -mno-fused-madd @gol
469 -mgas -mgp32 -mgp64 -mhard-float -mint64 -mips1 @gol
470 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
471 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
472 -mmips-as -mmips-tfile -mno-abicalls @gol
473 -mno-embedded-data -mno-uninit-const-in-rodata @gol
474 -mno-embedded-pic -mno-long-calls @gol
475 -mno-memcpy -mno-mips-tfile -mno-rnames @gol
476 -mrnames -msoft-float @gol
477 -m4650 -msingle-float -mmad @gol
478 -EL -EB -G @var{num} -nocpp @gol
479 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi -mabi-fake-default @gol
480 -mfix7000 -mno-crt0 -mflush-func=@var{func} -mno-flush-func @gol
481 -mbranch-likely -mno-branch-likely}
483 @emph{i386 and x86-64 Options}
484 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
485 -mfpmath=@var{unit} @gol
486 -masm=@var{dialect} -mno-fancy-math-387 @gol
487 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
488 -mno-wide-multiply -mrtd -malign-double @gol
489 -mpreferred-stack-boundary=@var{num} @gol
490 -mmmx -msse -msse2 -mpni -m3dnow @gol
491 -mthreads -mno-align-stringops -minline-all-stringops @gol
492 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
493 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
494 -mno-red-zone -mno-tls-direct-seg-refs @gol
495 -mcmodel=@var{code-model} @gol
499 @gccoptlist{-march=@var{architecture-type} @gol
500 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
501 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
502 -mjump-in-delay -mlinker-opt -mlong-calls @gol
503 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
504 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
505 -mno-jump-in-delay -mno-long-load-store @gol
506 -mno-portable-runtime -mno-soft-float @gol
507 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
508 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
509 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
510 -nolibdld -static -threads}
512 @emph{Intel 960 Options}
513 @gccoptlist{-m@var{cpu-type} -masm-compat -mclean-linkage @gol
514 -mcode-align -mcomplex-addr -mleaf-procedures @gol
515 -mic-compat -mic2.0-compat -mic3.0-compat @gol
516 -mintel-asm -mno-clean-linkage -mno-code-align @gol
517 -mno-complex-addr -mno-leaf-procedures @gol
518 -mno-old-align -mno-strict-align -mno-tail-call @gol
519 -mnumerics -mold-align -msoft-float -mstrict-align @gol
522 @emph{DEC Alpha Options}
523 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
524 -mieee -mieee-with-inexact -mieee-conformant @gol
525 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
526 -mtrap-precision=@var{mode} -mbuild-constants @gol
527 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
528 -mbwx -mmax -mfix -mcix @gol
529 -mfloat-vax -mfloat-ieee @gol
530 -mexplicit-relocs -msmall-data -mlarge-data @gol
531 -msmall-text -mlarge-text @gol
532 -mmemory-latency=@var{time}}
534 @emph{DEC Alpha/VMS Options}
535 @gccoptlist{-mvms-return-codes}
537 @emph{H8/300 Options}
538 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
541 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
542 -m4-nofpu -m4-single-only -m4-single -m4 @gol
543 -m5-64media -m5-64media-nofpu @gol
544 -m5-32media -m5-32media-nofpu @gol
545 -m5-compact -m5-compact-nofpu @gol
546 -mb -ml -mdalign -mrelax @gol
547 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
548 -mieee -misize -mpadstruct -mspace @gol
549 -mprefergot -musermode}
551 @emph{System V Options}
552 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
555 @gccoptlist{-EB -EL @gol
556 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
557 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
559 @emph{TMS320C3x/C4x Options}
560 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
561 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
562 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
563 -mparallel-insns -mparallel-mpy -mpreserve-float}
566 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
567 -mprolog-function -mno-prolog-function -mspace @gol
568 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
569 -mapp-regs -mno-app-regs @gol
570 -mdisable-callt -mno-disable-callt @gol
575 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
576 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
577 -mregparam -mnoregparam -msb -mnosb @gol
578 -mbitfield -mnobitfield -mhimem -mnohimem}
581 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
582 -mcall-prologues -mno-tablejump -mtiny-stack}
585 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
586 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
587 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
588 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
589 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
592 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
593 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
594 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
595 -mno-base-addresses -msingle-exit -mno-single-exit}
598 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
599 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
600 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
601 -minline-float-divide-max-throughput @gol
602 -minline-int-divide-min-latency @gol
603 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
604 -mfixed-range=@var{register-range}}
607 @gccoptlist{-mextmem -mextmemory -monchip -mno-asm-optimize @gol
608 -masm-optimize -mbranch-cost=@var{n} -mcond-exec=@var{n}}
610 @emph{S/390 and zSeries Options}
611 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
612 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
613 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
614 -m64 -m31 -mdebug -mno-debug -mesa -mzarch}
617 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
618 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
619 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
620 -mstack-align -mdata-align -mconst-align @gol
621 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
622 -melf -maout -melinux -mlinux -sim -sim2}
624 @emph{PDP-11 Options}
625 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
626 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
627 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
628 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
629 -mbranch-expensive -mbranch-cheap @gol
630 -msplit -mno-split -munix-asm -mdec-asm}
632 @emph{Xstormy16 Options}
635 @emph{Xtensa Options}
636 @gccoptlist{-mbig-endian -mlittle-endian @gol
637 -mdensity -mno-density @gol
638 -mconst16 -mno-const16 @gol
640 -maddx -mno-addx @gol
641 -mmac16 -mno-mac16 @gol
642 -mmul16 -mno-mul16 @gol
643 -mmul32 -mno-mul32 @gol
645 -mminmax -mno-minmax @gol
646 -msext -mno-sext @gol
647 -mbooleans -mno-booleans @gol
648 -mhard-float -msoft-float @gol
649 -mfused-madd -mno-fused-madd @gol
650 -mtext-section-literals -mno-text-section-literals @gol
651 -mtarget-align -mno-target-align @gol
652 -mlongcalls -mno-longcalls}
655 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
656 -mhard-float -msoft-float @gol
657 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
658 -mdouble -mno-double @gol
659 -mmedia -mno-media -mmuladd -mno-muladd @gol
660 -mlibrary-pic -macc-4 -macc-8 @gol
661 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
662 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
663 -mvliw-branch -mno-vliw-branch @gol
664 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
665 -mno-nested-cond-exec -mtomcat-stats @gol
668 @item Code Generation Options
669 @xref{Code Gen Options,,Options for Code Generation Conventions}.
670 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
671 -ffixed-@var{reg} -fexceptions @gol
672 -fnon-call-exceptions -funwind-tables @gol
673 -fasynchronous-unwind-tables @gol
674 -finhibit-size-directive -finstrument-functions @gol
675 -fno-common -fno-ident -fno-gnu-linker @gol
676 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
677 -freg-struct-return -fshared-data -fshort-enums @gol
678 -fshort-double -fshort-wchar @gol
679 -fverbose-asm -fpack-struct -fstack-check @gol
680 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
681 -fargument-alias -fargument-noalias @gol
682 -fargument-noalias-global -fleading-underscore @gol
683 -ftls-model=@var{model} @gol
684 -ftrapv -fwrapv -fbounds-check}
688 * Overall Options:: Controlling the kind of output:
689 an executable, object files, assembler files,
690 or preprocessed source.
691 * C Dialect Options:: Controlling the variant of C language compiled.
692 * C++ Dialect Options:: Variations on C++.
693 * Objective-C Dialect Options:: Variations on Objective-C.
694 * Language Independent Options:: Controlling how diagnostics should be
696 * Warning Options:: How picky should the compiler be?
697 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
698 * Optimize Options:: How much optimization?
699 * Preprocessor Options:: Controlling header files and macro definitions.
700 Also, getting dependency information for Make.
701 * Assembler Options:: Passing options to the assembler.
702 * Link Options:: Specifying libraries and so on.
703 * Directory Options:: Where to find header files and libraries.
704 Where to find the compiler executable files.
705 * Spec Files:: How to pass switches to sub-processes.
706 * Target Options:: Running a cross-compiler, or an old version of GCC.
709 @node Overall Options
710 @section Options Controlling the Kind of Output
712 Compilation can involve up to four stages: preprocessing, compilation
713 proper, assembly and linking, always in that order. GCC is capable of
714 preprocessing and compiling several files either into several
715 assembler input files, or into one assembler input file; then each
716 assembler input file produces an object file, and linking combines all
717 the object files (those newly compiled, and those specified as input)
718 into an executable file.
720 @cindex file name suffix
721 For any given input file, the file name suffix determines what kind of
726 C source code which must be preprocessed.
729 C source code which should not be preprocessed.
732 C++ source code which should not be preprocessed.
735 Objective-C source code. Note that you must link with the library
736 @file{libobjc.a} to make an Objective-C program work.
739 Objective-C source code which should not be preprocessed.
742 C or C++ header file to be turned into a precompiled header.
746 @itemx @var{file}.cxx
747 @itemx @var{file}.cpp
748 @itemx @var{file}.CPP
749 @itemx @var{file}.c++
751 C++ source code which must be preprocessed. Note that in @samp{.cxx},
752 the last two letters must both be literally @samp{x}. Likewise,
753 @samp{.C} refers to a literal capital C@.
757 C++ header file to be turned into a precompiled header.
760 @itemx @var{file}.for
761 @itemx @var{file}.FOR
762 Fortran source code which should not be preprocessed.
765 @itemx @var{file}.fpp
766 @itemx @var{file}.FPP
767 Fortran source code which must be preprocessed (with the traditional
771 Fortran source code which must be preprocessed with a RATFOR
772 preprocessor (not included with GCC)@.
774 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
775 Using and Porting GNU Fortran}, for more details of the handling of
778 @c FIXME: Descriptions of Java file types.
785 Ada source code file which contains a library unit declaration (a
786 declaration of a package, subprogram, or generic, or a generic
787 instantiation), or a library unit renaming declaration (a package,
788 generic, or subprogram renaming declaration). Such files are also
791 @itemx @var{file}.adb
792 Ada source code file containing a library unit body (a subprogram or
793 package body). Such files are also called @dfn{bodies}.
795 @c GCC also knows about some suffixes for languages not yet included:
804 Assembler code which must be preprocessed.
807 An object file to be fed straight into linking.
808 Any file name with no recognized suffix is treated this way.
812 You can specify the input language explicitly with the @option{-x} option:
815 @item -x @var{language}
816 Specify explicitly the @var{language} for the following input files
817 (rather than letting the compiler choose a default based on the file
818 name suffix). This option applies to all following input files until
819 the next @option{-x} option. Possible values for @var{language} are:
821 c c-header cpp-output
822 c++ c++-header c++-cpp-output
823 objective-c objective-c-header objc-cpp-output
824 assembler assembler-with-cpp
826 f77 f77-cpp-input ratfor
832 Turn off any specification of a language, so that subsequent files are
833 handled according to their file name suffixes (as they are if @option{-x}
834 has not been used at all).
836 @item -pass-exit-codes
837 @opindex pass-exit-codes
838 Normally the @command{gcc} program will exit with the code of 1 if any
839 phase of the compiler returns a non-success return code. If you specify
840 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
841 numerically highest error produced by any phase that returned an error
845 If you only want some of the stages of compilation, you can use
846 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
847 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
848 @command{gcc} is to stop. Note that some combinations (for example,
849 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
854 Compile or assemble the source files, but do not link. The linking
855 stage simply is not done. The ultimate output is in the form of an
856 object file for each source file.
858 By default, the object file name for a source file is made by replacing
859 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
861 Unrecognized input files, not requiring compilation or assembly, are
866 Stop after the stage of compilation proper; do not assemble. The output
867 is in the form of an assembler code file for each non-assembler input
870 By default, the assembler file name for a source file is made by
871 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
873 Input files that don't require compilation are ignored.
877 Stop after the preprocessing stage; do not run the compiler proper. The
878 output is in the form of preprocessed source code, which is sent to the
881 Input files which don't require preprocessing are ignored.
883 @cindex output file option
886 Place output in file @var{file}. This applies regardless to whatever
887 sort of output is being produced, whether it be an executable file,
888 an object file, an assembler file or preprocessed C code.
890 If you specify @option{-o} when compiling more than one input file, or
891 you are producing an executable file as output, all the source files
892 on the command line will be compiled at once.
894 If @option{-o} is not specified, the default is to put an executable file
895 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
896 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
897 all preprocessed C source on standard output.
901 Print (on standard error output) the commands executed to run the stages
902 of compilation. Also print the version number of the compiler driver
903 program and of the preprocessor and the compiler proper.
907 Like @option{-v} except the commands are not executed and all command
908 arguments are quoted. This is useful for shell scripts to capture the
909 driver-generated command lines.
913 Use pipes rather than temporary files for communication between the
914 various stages of compilation. This fails to work on some systems where
915 the assembler is unable to read from a pipe; but the GNU assembler has
920 Print (on the standard output) a description of the command line options
921 understood by @command{gcc}. If the @option{-v} option is also specified
922 then @option{--help} will also be passed on to the various processes
923 invoked by @command{gcc}, so that they can display the command line options
924 they accept. If the @option{-Wextra} option is also specified then command
925 line options which have no documentation associated with them will also
930 Print (on the standard output) a description of target specific command
931 line options for each tool.
935 Display the version number and copyrights of the invoked GCC.
939 @section Compiling C++ Programs
941 @cindex suffixes for C++ source
942 @cindex C++ source file suffixes
943 C++ source files conventionally use one of the suffixes @samp{.C},
944 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
945 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
946 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
947 files with these names and compiles them as C++ programs even if you
948 call the compiler the same way as for compiling C programs (usually
949 with the name @command{gcc}).
953 However, C++ programs often require class libraries as well as a
954 compiler that understands the C++ language---and under some
955 circumstances, you might want to compile programs or header files from
956 standard input, or otherwise without a suffix that flags them as C++
957 programs. You might also like to precompile a C header file with a
958 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
959 program that calls GCC with the default language set to C++, and
960 automatically specifies linking against the C++ library. On many
961 systems, @command{g++} is also installed with the name @command{c++}.
963 @cindex invoking @command{g++}
964 When you compile C++ programs, you may specify many of the same
965 command-line options that you use for compiling programs in any
966 language; or command-line options meaningful for C and related
967 languages; or options that are meaningful only for C++ programs.
968 @xref{C Dialect Options,,Options Controlling C Dialect}, for
969 explanations of options for languages related to C@.
970 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
971 explanations of options that are meaningful only for C++ programs.
973 @node C Dialect Options
974 @section Options Controlling C Dialect
975 @cindex dialect options
976 @cindex language dialect options
977 @cindex options, dialect
979 The following options control the dialect of C (or languages derived
980 from C, such as C++ and Objective-C) that the compiler accepts:
987 In C mode, support all ISO C90 programs. In C++ mode,
988 remove GNU extensions that conflict with ISO C++.
990 This turns off certain features of GCC that are incompatible with ISO
991 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
992 such as the @code{asm} and @code{typeof} keywords, and
993 predefined macros such as @code{unix} and @code{vax} that identify the
994 type of system you are using. It also enables the undesirable and
995 rarely used ISO trigraph feature. For the C compiler,
996 it disables recognition of C++ style @samp{//} comments as well as
997 the @code{inline} keyword.
999 The alternate keywords @code{__asm__}, @code{__extension__},
1000 @code{__inline__} and @code{__typeof__} continue to work despite
1001 @option{-ansi}. You would not want to use them in an ISO C program, of
1002 course, but it is useful to put them in header files that might be included
1003 in compilations done with @option{-ansi}. Alternate predefined macros
1004 such as @code{__unix__} and @code{__vax__} are also available, with or
1005 without @option{-ansi}.
1007 The @option{-ansi} option does not cause non-ISO programs to be
1008 rejected gratuitously. For that, @option{-pedantic} is required in
1009 addition to @option{-ansi}. @xref{Warning Options}.
1011 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1012 option is used. Some header files may notice this macro and refrain
1013 from declaring certain functions or defining certain macros that the
1014 ISO standard doesn't call for; this is to avoid interfering with any
1015 programs that might use these names for other things.
1017 Functions which would normally be built in but do not have semantics
1018 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1019 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1020 built-in functions provided by GCC}, for details of the functions
1025 Determine the language standard. This option is currently only
1026 supported when compiling C or C++. A value for this option must be
1027 provided; possible values are
1032 ISO C90 (same as @option{-ansi}).
1034 @item iso9899:199409
1035 ISO C90 as modified in amendment 1.
1041 ISO C99. Note that this standard is not yet fully supported; see
1042 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1043 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1046 Default, ISO C90 plus GNU extensions (including some C99 features).
1050 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1051 this will become the default. The name @samp{gnu9x} is deprecated.
1054 The 1998 ISO C++ standard plus amendments.
1057 The same as @option{-std=c++98} plus GNU extensions. This is the
1058 default for C++ code.
1061 Even when this option is not specified, you can still use some of the
1062 features of newer standards in so far as they do not conflict with
1063 previous C standards. For example, you may use @code{__restrict__} even
1064 when @option{-std=c99} is not specified.
1066 The @option{-std} options specifying some version of ISO C have the same
1067 effects as @option{-ansi}, except that features that were not in ISO C90
1068 but are in the specified version (for example, @samp{//} comments and
1069 the @code{inline} keyword in ISO C99) are not disabled.
1071 @xref{Standards,,Language Standards Supported by GCC}, for details of
1072 these standard versions.
1074 @item -aux-info @var{filename}
1076 Output to the given filename prototyped declarations for all functions
1077 declared and/or defined in a translation unit, including those in header
1078 files. This option is silently ignored in any language other than C@.
1080 Besides declarations, the file indicates, in comments, the origin of
1081 each declaration (source file and line), whether the declaration was
1082 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1083 @samp{O} for old, respectively, in the first character after the line
1084 number and the colon), and whether it came from a declaration or a
1085 definition (@samp{C} or @samp{F}, respectively, in the following
1086 character). In the case of function definitions, a K&R-style list of
1087 arguments followed by their declarations is also provided, inside
1088 comments, after the declaration.
1092 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1093 keyword, so that code can use these words as identifiers. You can use
1094 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1095 instead. @option{-ansi} implies @option{-fno-asm}.
1097 In C++, this switch only affects the @code{typeof} keyword, since
1098 @code{asm} and @code{inline} are standard keywords. You may want to
1099 use the @option{-fno-gnu-keywords} flag instead, which has the same
1100 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1101 switch only affects the @code{asm} and @code{typeof} keywords, since
1102 @code{inline} is a standard keyword in ISO C99.
1105 @itemx -fno-builtin-@var{function}
1106 @opindex fno-builtin
1107 @cindex built-in functions
1108 Don't recognize built-in functions that do not begin with
1109 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1110 functions provided by GCC}, for details of the functions affected,
1111 including those which are not built-in functions when @option{-ansi} or
1112 @option{-std} options for strict ISO C conformance are used because they
1113 do not have an ISO standard meaning.
1115 GCC normally generates special code to handle certain built-in functions
1116 more efficiently; for instance, calls to @code{alloca} may become single
1117 instructions that adjust the stack directly, and calls to @code{memcpy}
1118 may become inline copy loops. The resulting code is often both smaller
1119 and faster, but since the function calls no longer appear as such, you
1120 cannot set a breakpoint on those calls, nor can you change the behavior
1121 of the functions by linking with a different library.
1123 With the @option{-fno-builtin-@var{function}} option
1124 only the built-in function @var{function} is
1125 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1126 function is named this is not built-in in this version of GCC, this
1127 option is ignored. There is no corresponding
1128 @option{-fbuiltin-@var{function}} option; if you wish to enable
1129 built-in functions selectively when using @option{-fno-builtin} or
1130 @option{-ffreestanding}, you may define macros such as:
1133 #define abs(n) __builtin_abs ((n))
1134 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1139 @cindex hosted environment
1141 Assert that compilation takes place in a hosted environment. This implies
1142 @option{-fbuiltin}. A hosted environment is one in which the
1143 entire standard library is available, and in which @code{main} has a return
1144 type of @code{int}. Examples are nearly everything except a kernel.
1145 This is equivalent to @option{-fno-freestanding}.
1147 @item -ffreestanding
1148 @opindex ffreestanding
1149 @cindex hosted environment
1151 Assert that compilation takes place in a freestanding environment. This
1152 implies @option{-fno-builtin}. A freestanding environment
1153 is one in which the standard library may not exist, and program startup may
1154 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1155 This is equivalent to @option{-fno-hosted}.
1157 @xref{Standards,,Language Standards Supported by GCC}, for details of
1158 freestanding and hosted environments.
1160 @item -fms-extensions
1161 @opindex fms-extensions
1162 Accept some non-standard constructs used in Microsoft header files.
1166 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1167 options for strict ISO C conformance) implies @option{-trigraphs}.
1169 @item -no-integrated-cpp
1170 @opindex no-integrated-cpp
1171 Performs a compilation in two passes: preprocessing and compiling. This
1172 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1173 @option{-B} option. The user supplied compilation step can then add in
1174 an additional preprocessing step after normal preprocessing but before
1175 compiling. The default is to use the integrated cpp (internal cpp)
1177 The semantics of this option will change if "cc1", "cc1plus", and
1178 "cc1obj" are merged.
1180 @cindex traditional C language
1181 @cindex C language, traditional
1183 @itemx -traditional-cpp
1184 @opindex traditional-cpp
1185 @opindex traditional
1186 Formerly, these options caused GCC to attempt to emulate a pre-standard
1187 C compiler. They are now only supported with the @option{-E} switch.
1188 The preprocessor continues to support a pre-standard mode. See the GNU
1189 CPP manual for details.
1191 @item -fcond-mismatch
1192 @opindex fcond-mismatch
1193 Allow conditional expressions with mismatched types in the second and
1194 third arguments. The value of such an expression is void. This option
1195 is not supported for C++.
1197 @item -funsigned-char
1198 @opindex funsigned-char
1199 Let the type @code{char} be unsigned, like @code{unsigned char}.
1201 Each kind of machine has a default for what @code{char} should
1202 be. It is either like @code{unsigned char} by default or like
1203 @code{signed char} by default.
1205 Ideally, a portable program should always use @code{signed char} or
1206 @code{unsigned char} when it depends on the signedness of an object.
1207 But many programs have been written to use plain @code{char} and
1208 expect it to be signed, or expect it to be unsigned, depending on the
1209 machines they were written for. This option, and its inverse, let you
1210 make such a program work with the opposite default.
1212 The type @code{char} is always a distinct type from each of
1213 @code{signed char} or @code{unsigned char}, even though its behavior
1214 is always just like one of those two.
1217 @opindex fsigned-char
1218 Let the type @code{char} be signed, like @code{signed char}.
1220 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1221 the negative form of @option{-funsigned-char}. Likewise, the option
1222 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1224 @item -fsigned-bitfields
1225 @itemx -funsigned-bitfields
1226 @itemx -fno-signed-bitfields
1227 @itemx -fno-unsigned-bitfields
1228 @opindex fsigned-bitfields
1229 @opindex funsigned-bitfields
1230 @opindex fno-signed-bitfields
1231 @opindex fno-unsigned-bitfields
1232 These options control whether a bit-field is signed or unsigned, when the
1233 declaration does not use either @code{signed} or @code{unsigned}. By
1234 default, such a bit-field is signed, because this is consistent: the
1235 basic integer types such as @code{int} are signed types.
1237 @item -fwritable-strings
1238 @opindex fwritable-strings
1239 Store string constants in the writable data segment and don't uniquize
1240 them. This is for compatibility with old programs which assume they can
1241 write into string constants.
1243 Writing into string constants is a very bad idea; ``constants'' should
1247 @node C++ Dialect Options
1248 @section Options Controlling C++ Dialect
1250 @cindex compiler options, C++
1251 @cindex C++ options, command line
1252 @cindex options, C++
1253 This section describes the command-line options that are only meaningful
1254 for C++ programs; but you can also use most of the GNU compiler options
1255 regardless of what language your program is in. For example, you
1256 might compile a file @code{firstClass.C} like this:
1259 g++ -g -frepo -O -c firstClass.C
1263 In this example, only @option{-frepo} is an option meant
1264 only for C++ programs; you can use the other options with any
1265 language supported by GCC@.
1267 Here is a list of options that are @emph{only} for compiling C++ programs:
1271 @item -fabi-version=@var{n}
1272 @opindex fabi-version
1273 Use version @var{n} of the C++ ABI. Version 1 is the version of the C++
1274 ABI that first appeared in G++ 3.2. Version 0 will always be the
1275 version that conforms most closely to the C++ ABI specification.
1276 Therefore, the ABI obtained using version 0 will change as ABI bugs are
1279 The default is version 1.
1281 @item -fno-access-control
1282 @opindex fno-access-control
1283 Turn off all access checking. This switch is mainly useful for working
1284 around bugs in the access control code.
1288 Check that the pointer returned by @code{operator new} is non-null
1289 before attempting to modify the storage allocated. This check is
1290 normally unnecessary because the C++ standard specifies that
1291 @code{operator new} will only return @code{0} if it is declared
1292 @samp{throw()}, in which case the compiler will always check the
1293 return value even without this option. In all other cases, when
1294 @code{operator new} has a non-empty exception specification, memory
1295 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1296 @samp{new (nothrow)}.
1298 @item -fconserve-space
1299 @opindex fconserve-space
1300 Put uninitialized or runtime-initialized global variables into the
1301 common segment, as C does. This saves space in the executable at the
1302 cost of not diagnosing duplicate definitions. If you compile with this
1303 flag and your program mysteriously crashes after @code{main()} has
1304 completed, you may have an object that is being destroyed twice because
1305 two definitions were merged.
1307 This option is no longer useful on most targets, now that support has
1308 been added for putting variables into BSS without making them common.
1310 @item -fno-const-strings
1311 @opindex fno-const-strings
1312 Give string constants type @code{char *} instead of type @code{const
1313 char *}. By default, G++ uses type @code{const char *} as required by
1314 the standard. Even if you use @option{-fno-const-strings}, you cannot
1315 actually modify the value of a string constant, unless you also use
1316 @option{-fwritable-strings}.
1318 This option might be removed in a future release of G++. For maximum
1319 portability, you should structure your code so that it works with
1320 string constants that have type @code{const char *}.
1322 @item -fno-elide-constructors
1323 @opindex fno-elide-constructors
1324 The C++ standard allows an implementation to omit creating a temporary
1325 which is only used to initialize another object of the same type.
1326 Specifying this option disables that optimization, and forces G++ to
1327 call the copy constructor in all cases.
1329 @item -fno-enforce-eh-specs
1330 @opindex fno-enforce-eh-specs
1331 Don't check for violation of exception specifications at runtime. This
1332 option violates the C++ standard, but may be useful for reducing code
1333 size in production builds, much like defining @samp{NDEBUG}. The compiler
1334 will still optimize based on the exception specifications.
1336 @item -fexternal-templates
1337 @opindex fexternal-templates
1339 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1340 template instantiation; template instances are emitted or not according
1341 to the location of the template definition. @xref{Template
1342 Instantiation}, for more information.
1344 This option is deprecated.
1346 @item -falt-external-templates
1347 @opindex falt-external-templates
1348 Similar to @option{-fexternal-templates}, but template instances are
1349 emitted or not according to the place where they are first instantiated.
1350 @xref{Template Instantiation}, for more information.
1352 This option is deprecated.
1355 @itemx -fno-for-scope
1357 @opindex fno-for-scope
1358 If @option{-ffor-scope} is specified, the scope of variables declared in
1359 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1360 as specified by the C++ standard.
1361 If @option{-fno-for-scope} is specified, the scope of variables declared in
1362 a @i{for-init-statement} extends to the end of the enclosing scope,
1363 as was the case in old versions of G++, and other (traditional)
1364 implementations of C++.
1366 The default if neither flag is given to follow the standard,
1367 but to allow and give a warning for old-style code that would
1368 otherwise be invalid, or have different behavior.
1370 @item -fno-gnu-keywords
1371 @opindex fno-gnu-keywords
1372 Do not recognize @code{typeof} as a keyword, so that code can use this
1373 word as an identifier. You can use the keyword @code{__typeof__} instead.
1374 @option{-ansi} implies @option{-fno-gnu-keywords}.
1376 @item -fno-implicit-templates
1377 @opindex fno-implicit-templates
1378 Never emit code for non-inline templates which are instantiated
1379 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1380 @xref{Template Instantiation}, for more information.
1382 @item -fno-implicit-inline-templates
1383 @opindex fno-implicit-inline-templates
1384 Don't emit code for implicit instantiations of inline templates, either.
1385 The default is to handle inlines differently so that compiles with and
1386 without optimization will need the same set of explicit instantiations.
1388 @item -fno-implement-inlines
1389 @opindex fno-implement-inlines
1390 To save space, do not emit out-of-line copies of inline functions
1391 controlled by @samp{#pragma implementation}. This will cause linker
1392 errors if these functions are not inlined everywhere they are called.
1394 @item -fms-extensions
1395 @opindex fms-extensions
1396 Disable pedantic warnings about constructs used in MFC, such as implicit
1397 int and getting a pointer to member function via non-standard syntax.
1399 @item -fno-nonansi-builtins
1400 @opindex fno-nonansi-builtins
1401 Disable built-in declarations of functions that are not mandated by
1402 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1403 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1405 @item -fno-operator-names
1406 @opindex fno-operator-names
1407 Do not treat the operator name keywords @code{and}, @code{bitand},
1408 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1409 synonyms as keywords.
1411 @item -fno-optional-diags
1412 @opindex fno-optional-diags
1413 Disable diagnostics that the standard says a compiler does not need to
1414 issue. Currently, the only such diagnostic issued by G++ is the one for
1415 a name having multiple meanings within a class.
1418 @opindex fpermissive
1419 Downgrade some diagnostics about nonconformant code from errors to
1420 warnings. Thus, using @option{-fpermissive} will allow some
1421 nonconforming code to compile.
1425 Enable automatic template instantiation at link time. This option also
1426 implies @option{-fno-implicit-templates}. @xref{Template
1427 Instantiation}, for more information.
1431 Disable generation of information about every class with virtual
1432 functions for use by the C++ runtime type identification features
1433 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1434 of the language, you can save some space by using this flag. Note that
1435 exception handling uses the same information, but it will generate it as
1440 Emit statistics about front-end processing at the end of the compilation.
1441 This information is generally only useful to the G++ development team.
1443 @item -ftemplate-depth-@var{n}
1444 @opindex ftemplate-depth
1445 Set the maximum instantiation depth for template classes to @var{n}.
1446 A limit on the template instantiation depth is needed to detect
1447 endless recursions during template class instantiation. ANSI/ISO C++
1448 conforming programs must not rely on a maximum depth greater than 17.
1450 @item -fuse-cxa-atexit
1451 @opindex fuse-cxa-atexit
1452 Register destructors for objects with static storage duration with the
1453 @code{__cxa_atexit} function rather than the @code{atexit} function.
1454 This option is required for fully standards-compliant handling of static
1455 destructors, but will only work if your C library supports
1456 @code{__cxa_atexit}.
1460 Emit special relocations for vtables and virtual function references
1461 so that the linker can identify unused virtual functions and zero out
1462 vtable slots that refer to them. This is most useful with
1463 @option{-ffunction-sections} and @option{-Wl,--gc-sections}, in order to
1464 also discard the functions themselves.
1466 This optimization requires GNU as and GNU ld. Not all systems support
1467 this option. @option{-Wl,--gc-sections} is ignored without @option{-static}.
1471 Do not use weak symbol support, even if it is provided by the linker.
1472 By default, G++ will use weak symbols if they are available. This
1473 option exists only for testing, and should not be used by end-users;
1474 it will result in inferior code and has no benefits. This option may
1475 be removed in a future release of G++.
1479 Do not search for header files in the standard directories specific to
1480 C++, but do still search the other standard directories. (This option
1481 is used when building the C++ library.)
1484 In addition, these optimization, warning, and code generation options
1485 have meanings only for C++ programs:
1488 @item -fno-default-inline
1489 @opindex fno-default-inline
1490 Do not assume @samp{inline} for functions defined inside a class scope.
1491 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1492 functions will have linkage like inline functions; they just won't be
1495 @item -Wabi @r{(C++ only)}
1497 Warn when G++ generates code that is probably not compatible with the
1498 vendor-neutral C++ ABI. Although an effort has been made to warn about
1499 all such cases, there are probably some cases that are not warned about,
1500 even though G++ is generating incompatible code. There may also be
1501 cases where warnings are emitted even though the code that is generated
1504 You should rewrite your code to avoid these warnings if you are
1505 concerned about the fact that code generated by G++ may not be binary
1506 compatible with code generated by other compilers.
1508 The known incompatibilities at this point include:
1513 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1514 pack data into the same byte as a base class. For example:
1517 struct A @{ virtual void f(); int f1 : 1; @};
1518 struct B : public A @{ int f2 : 1; @};
1522 In this case, G++ will place @code{B::f2} into the same byte
1523 as@code{A::f1}; other compilers will not. You can avoid this problem
1524 by explicitly padding @code{A} so that its size is a multiple of the
1525 byte size on your platform; that will cause G++ and other compilers to
1526 layout @code{B} identically.
1529 Incorrect handling of tail-padding for virtual bases. G++ does not use
1530 tail padding when laying out virtual bases. For example:
1533 struct A @{ virtual void f(); char c1; @};
1534 struct B @{ B(); char c2; @};
1535 struct C : public A, public virtual B @{@};
1539 In this case, G++ will not place @code{B} into the tail-padding for
1540 @code{A}; other compilers will. You can avoid this problem by
1541 explicitly padding @code{A} so that its size is a multiple of its
1542 alignment (ignoring virtual base classes); that will cause G++ and other
1543 compilers to layout @code{C} identically.
1546 Incorrect handling of bit-fields with declared widths greater than that
1547 of their underlying types, when the bit-fields appear in a union. For
1551 union U @{ int i : 4096; @};
1555 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1556 union too small by the number of bits in an @code{int}.
1559 Empty classes can be placed at incorrect offsets. For example:
1569 struct C : public B, public A @{@};
1573 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1574 it should be placed at offset zero. G++ mistakenly believes that the
1575 @code{A} data member of @code{B} is already at offset zero.
1578 Names of template functions whose types involve @code{typename} or
1579 template template parameters can be mangled incorrectly.
1582 template <typename Q>
1583 void f(typename Q::X) @{@}
1585 template <template <typename> class Q>
1586 void f(typename Q<int>::X) @{@}
1590 Instantiations of these templates may be mangled incorrectly.
1594 @item -Wctor-dtor-privacy @r{(C++ only)}
1595 @opindex Wctor-dtor-privacy
1596 Warn when a class seems unusable, because all the constructors or
1597 destructors in a class are private and the class has no friends or
1598 public static member functions.
1600 @item -Wnon-virtual-dtor @r{(C++ only)}
1601 @opindex Wnon-virtual-dtor
1602 Warn when a class declares a non-virtual destructor that should probably
1603 be virtual, because it looks like the class will be used polymorphically.
1604 This warning is enabled by @option{-Wall}.
1606 @item -Wreorder @r{(C++ only)}
1608 @cindex reordering, warning
1609 @cindex warning for reordering of member initializers
1610 Warn when the order of member initializers given in the code does not
1611 match the order in which they must be executed. For instance:
1617 A(): j (0), i (1) @{ @}
1621 Here the compiler will warn that the member initializers for @samp{i}
1622 and @samp{j} will be rearranged to match the declaration order of the
1623 members. This warning is enabled by @option{-Wall}.
1626 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1629 @item -Weffc++ @r{(C++ only)}
1631 Warn about violations of the following style guidelines from Scott Meyers'
1632 @cite{Effective C++} book:
1636 Item 11: Define a copy constructor and an assignment operator for classes
1637 with dynamically allocated memory.
1640 Item 12: Prefer initialization to assignment in constructors.
1643 Item 14: Make destructors virtual in base classes.
1646 Item 15: Have @code{operator=} return a reference to @code{*this}.
1649 Item 23: Don't try to return a reference when you must return an object.
1653 and about violations of the following style guidelines from Scott Meyers'
1654 @cite{More Effective C++} book:
1658 Item 6: Distinguish between prefix and postfix forms of increment and
1659 decrement operators.
1662 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1666 If you use this option, you should be aware that the standard library
1667 headers do not obey all of these guidelines; you can use @samp{grep -v}
1668 to filter out those warnings.
1670 @item -Wno-deprecated @r{(C++ only)}
1671 @opindex Wno-deprecated
1672 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1674 @item -Wno-non-template-friend @r{(C++ only)}
1675 @opindex Wno-non-template-friend
1676 Disable warnings when non-templatized friend functions are declared
1677 within a template. With the advent of explicit template specification
1678 support in G++, if the name of the friend is an unqualified-id (i.e.,
1679 @samp{friend foo(int)}), the C++ language specification demands that the
1680 friend declare or define an ordinary, nontemplate function. (Section
1681 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1682 could be interpreted as a particular specialization of a templatized
1683 function. Because this non-conforming behavior is no longer the default
1684 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1685 check existing code for potential trouble spots, and is on by default.
1686 This new compiler behavior can be turned off with
1687 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1688 but disables the helpful warning.
1690 @item -Wold-style-cast @r{(C++ only)}
1691 @opindex Wold-style-cast
1692 Warn if an old-style (C-style) cast to a non-void type is used within
1693 a C++ program. The new-style casts (@samp{static_cast},
1694 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1695 unintended effects, and much easier to grep for.
1697 @item -Woverloaded-virtual @r{(C++ only)}
1698 @opindex Woverloaded-virtual
1699 @cindex overloaded virtual fn, warning
1700 @cindex warning for overloaded virtual fn
1701 Warn when a function declaration hides virtual functions from a
1702 base class. For example, in:
1709 struct B: public A @{
1714 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1722 will fail to compile.
1724 @item -Wno-pmf-conversions @r{(C++ only)}
1725 @opindex Wno-pmf-conversions
1726 Disable the diagnostic for converting a bound pointer to member function
1729 @item -Wsign-promo @r{(C++ only)}
1730 @opindex Wsign-promo
1731 Warn when overload resolution chooses a promotion from unsigned or
1732 enumeral type to a signed type over a conversion to an unsigned type of
1733 the same size. Previous versions of G++ would try to preserve
1734 unsignedness, but the standard mandates the current behavior.
1736 @item -Wsynth @r{(C++ only)}
1738 @cindex warning for synthesized methods
1739 @cindex synthesized methods, warning
1740 Warn when G++'s synthesis behavior does not match that of cfront. For
1746 A& operator = (int);
1756 In this example, G++ will synthesize a default @samp{A& operator =
1757 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1760 @node Objective-C Dialect Options
1761 @section Options Controlling Objective-C Dialect
1763 @cindex compiler options, Objective-C
1764 @cindex Objective-C options, command line
1765 @cindex options, Objective-C
1766 This section describes the command-line options that are only meaningful
1767 for Objective-C programs; but you can also use most of the GNU compiler
1768 options regardless of what language your program is in. For example,
1769 you might compile a file @code{some_class.m} like this:
1772 gcc -g -fgnu-runtime -O -c some_class.m
1776 In this example, only @option{-fgnu-runtime} is an option meant only for
1777 Objective-C programs; you can use the other options with any language
1780 Here is a list of options that are @emph{only} for compiling Objective-C
1784 @item -fconstant-string-class=@var{class-name}
1785 @opindex fconstant-string-class
1786 Use @var{class-name} as the name of the class to instantiate for each
1787 literal string specified with the syntax @code{@@"@dots{}"}. The default
1788 class name is @code{NXConstantString}.
1791 @opindex fgnu-runtime
1792 Generate object code compatible with the standard GNU Objective-C
1793 runtime. This is the default for most types of systems.
1795 @item -fnext-runtime
1796 @opindex fnext-runtime
1797 Generate output compatible with the NeXT runtime. This is the default
1798 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1799 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1804 Dump interface declarations for all classes seen in the source file to a
1805 file named @file{@var{sourcename}.decl}.
1808 @opindex Wno-protocol
1809 If a class is declared to implement a protocol, a warning is issued for
1810 every method in the protocol that is not implemented by the class. The
1811 default behavior is to issue a warning for every method not explicitly
1812 implemented in the class, even if a method implementation is inherited
1813 from the superclass. If you use the @code{-Wno-protocol} option, then
1814 methods inherited from the superclass are considered to be implemented,
1815 and no warning is issued for them.
1819 Warn if multiple methods of different types for the same selector are
1820 found during compilation. The check is performed on the list of methods
1821 in the final stage of compilation. Additionally, a check is performed
1822 that for each selector appearing in a @code{@@selector(@dots{})}
1823 expression, a corresponding method with that selector has been found
1824 during compilation. Because these checks scan the method table only at
1825 the end of compilation, these warnings are not produced if the final
1826 stage of compilation is not reached, for example because an error is
1827 found during compilation, or because the @code{-fsyntax-only} option is
1830 @item -Wundeclared-selector
1831 @opindex Wundeclared-selector
1832 Warn if a @code{@@selector(@dots{})} expression referring to an
1833 undeclared selector is found. A selector is considered undeclared if no
1834 method with that name has been declared (explicitly, in an
1835 @code{@@interface} or @code{@@protocol} declaration, or implicitly, in
1836 an @code{@@implementation} section) before the
1837 @code{@@selector(@dots{})} expression. This option always performs its
1838 checks as soon as a @code{@@selector(@dots{})} expression is found
1839 (while @code{-Wselector} only performs its checks in the final stage of
1840 compilation), and so additionally enforces the coding style convention
1841 that methods and selectors must be declared before being used.
1843 @c not documented because only avail via -Wp
1844 @c @item -print-objc-runtime-info
1848 @node Language Independent Options
1849 @section Options to Control Diagnostic Messages Formatting
1850 @cindex options to control diagnostics formatting
1851 @cindex diagnostic messages
1852 @cindex message formatting
1854 Traditionally, diagnostic messages have been formatted irrespective of
1855 the output device's aspect (e.g.@: its width, @dots{}). The options described
1856 below can be used to control the diagnostic messages formatting
1857 algorithm, e.g.@: how many characters per line, how often source location
1858 information should be reported. Right now, only the C++ front end can
1859 honor these options. However it is expected, in the near future, that
1860 the remaining front ends would be able to digest them correctly.
1863 @item -fmessage-length=@var{n}
1864 @opindex fmessage-length
1865 Try to format error messages so that they fit on lines of about @var{n}
1866 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1867 the front ends supported by GCC@. If @var{n} is zero, then no
1868 line-wrapping will be done; each error message will appear on a single
1871 @opindex fdiagnostics-show-location
1872 @item -fdiagnostics-show-location=once
1873 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1874 reporter to emit @emph{once} source location information; that is, in
1875 case the message is too long to fit on a single physical line and has to
1876 be wrapped, the source location won't be emitted (as prefix) again,
1877 over and over, in subsequent continuation lines. This is the default
1880 @item -fdiagnostics-show-location=every-line
1881 Only meaningful in line-wrapping mode. Instructs the diagnostic
1882 messages reporter to emit the same source location information (as
1883 prefix) for physical lines that result from the process of breaking
1884 a message which is too long to fit on a single line.
1888 @node Warning Options
1889 @section Options to Request or Suppress Warnings
1890 @cindex options to control warnings
1891 @cindex warning messages
1892 @cindex messages, warning
1893 @cindex suppressing warnings
1895 Warnings are diagnostic messages that report constructions which
1896 are not inherently erroneous but which are risky or suggest there
1897 may have been an error.
1899 You can request many specific warnings with options beginning @samp{-W},
1900 for example @option{-Wimplicit} to request warnings on implicit
1901 declarations. Each of these specific warning options also has a
1902 negative form beginning @samp{-Wno-} to turn off warnings;
1903 for example, @option{-Wno-implicit}. This manual lists only one of the
1904 two forms, whichever is not the default.
1906 The following options control the amount and kinds of warnings produced
1907 by GCC; for further, language-specific options also refer to
1908 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
1911 @cindex syntax checking
1913 @opindex fsyntax-only
1914 Check the code for syntax errors, but don't do anything beyond that.
1918 Issue all the warnings demanded by strict ISO C and ISO C++;
1919 reject all programs that use forbidden extensions, and some other
1920 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1921 version of the ISO C standard specified by any @option{-std} option used.
1923 Valid ISO C and ISO C++ programs should compile properly with or without
1924 this option (though a rare few will require @option{-ansi} or a
1925 @option{-std} option specifying the required version of ISO C)@. However,
1926 without this option, certain GNU extensions and traditional C and C++
1927 features are supported as well. With this option, they are rejected.
1929 @option{-pedantic} does not cause warning messages for use of the
1930 alternate keywords whose names begin and end with @samp{__}. Pedantic
1931 warnings are also disabled in the expression that follows
1932 @code{__extension__}. However, only system header files should use
1933 these escape routes; application programs should avoid them.
1934 @xref{Alternate Keywords}.
1936 Some users try to use @option{-pedantic} to check programs for strict ISO
1937 C conformance. They soon find that it does not do quite what they want:
1938 it finds some non-ISO practices, but not all---only those for which
1939 ISO C @emph{requires} a diagnostic, and some others for which
1940 diagnostics have been added.
1942 A feature to report any failure to conform to ISO C might be useful in
1943 some instances, but would require considerable additional work and would
1944 be quite different from @option{-pedantic}. We don't have plans to
1945 support such a feature in the near future.
1947 Where the standard specified with @option{-std} represents a GNU
1948 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1949 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1950 extended dialect is based. Warnings from @option{-pedantic} are given
1951 where they are required by the base standard. (It would not make sense
1952 for such warnings to be given only for features not in the specified GNU
1953 C dialect, since by definition the GNU dialects of C include all
1954 features the compiler supports with the given option, and there would be
1955 nothing to warn about.)
1957 @item -pedantic-errors
1958 @opindex pedantic-errors
1959 Like @option{-pedantic}, except that errors are produced rather than
1964 Inhibit all warning messages.
1968 Inhibit warning messages about the use of @samp{#import}.
1970 @item -Wchar-subscripts
1971 @opindex Wchar-subscripts
1972 Warn if an array subscript has type @code{char}. This is a common cause
1973 of error, as programmers often forget that this type is signed on some
1978 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1979 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1983 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1984 the arguments supplied have types appropriate to the format string
1985 specified, and that the conversions specified in the format string make
1986 sense. This includes standard functions, and others specified by format
1987 attributes (@pxref{Function Attributes}), in the @code{printf},
1988 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1989 not in the C standard) families.
1991 The formats are checked against the format features supported by GNU
1992 libc version 2.2. These include all ISO C90 and C99 features, as well
1993 as features from the Single Unix Specification and some BSD and GNU
1994 extensions. Other library implementations may not support all these
1995 features; GCC does not support warning about features that go beyond a
1996 particular library's limitations. However, if @option{-pedantic} is used
1997 with @option{-Wformat}, warnings will be given about format features not
1998 in the selected standard version (but not for @code{strfmon} formats,
1999 since those are not in any version of the C standard). @xref{C Dialect
2000 Options,,Options Controlling C Dialect}.
2002 Since @option{-Wformat} also checks for null format arguments for
2003 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2005 @option{-Wformat} is included in @option{-Wall}. For more control over some
2006 aspects of format checking, the options @option{-Wno-format-y2k},
2007 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2008 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2009 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2011 @item -Wno-format-y2k
2012 @opindex Wno-format-y2k
2013 If @option{-Wformat} is specified, do not warn about @code{strftime}
2014 formats which may yield only a two-digit year.
2016 @item -Wno-format-extra-args
2017 @opindex Wno-format-extra-args
2018 If @option{-Wformat} is specified, do not warn about excess arguments to a
2019 @code{printf} or @code{scanf} format function. The C standard specifies
2020 that such arguments are ignored.
2022 Where the unused arguments lie between used arguments that are
2023 specified with @samp{$} operand number specifications, normally
2024 warnings are still given, since the implementation could not know what
2025 type to pass to @code{va_arg} to skip the unused arguments. However,
2026 in the case of @code{scanf} formats, this option will suppress the
2027 warning if the unused arguments are all pointers, since the Single
2028 Unix Specification says that such unused arguments are allowed.
2030 @item -Wno-format-zero-length
2031 @opindex Wno-format-zero-length
2032 If @option{-Wformat} is specified, do not warn about zero-length formats.
2033 The C standard specifies that zero-length formats are allowed.
2035 @item -Wformat-nonliteral
2036 @opindex Wformat-nonliteral
2037 If @option{-Wformat} is specified, also warn if the format string is not a
2038 string literal and so cannot be checked, unless the format function
2039 takes its format arguments as a @code{va_list}.
2041 @item -Wformat-security
2042 @opindex Wformat-security
2043 If @option{-Wformat} is specified, also warn about uses of format
2044 functions that represent possible security problems. At present, this
2045 warns about calls to @code{printf} and @code{scanf} functions where the
2046 format string is not a string literal and there are no format arguments,
2047 as in @code{printf (foo);}. This may be a security hole if the format
2048 string came from untrusted input and contains @samp{%n}. (This is
2049 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2050 in future warnings may be added to @option{-Wformat-security} that are not
2051 included in @option{-Wformat-nonliteral}.)
2055 Enable @option{-Wformat} plus format checks not included in
2056 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2057 -Wformat-nonliteral -Wformat-security}.
2061 Enable warning about passing a null pointer for arguments marked as
2062 requiring a non-null value by the @code{nonnull} function attribute.
2064 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2065 can be disabled with the @option{-Wno-nonnull} option.
2067 @item -Wimplicit-int
2068 @opindex Wimplicit-int
2069 Warn when a declaration does not specify a type.
2071 @item -Wimplicit-function-declaration
2072 @itemx -Werror-implicit-function-declaration
2073 @opindex Wimplicit-function-declaration
2074 @opindex Werror-implicit-function-declaration
2075 Give a warning (or error) whenever a function is used before being
2080 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2084 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2085 function with external linkage, returning int, taking either zero
2086 arguments, two, or three arguments of appropriate types.
2088 @item -Wmissing-braces
2089 @opindex Wmissing-braces
2090 Warn if an aggregate or union initializer is not fully bracketed. In
2091 the following example, the initializer for @samp{a} is not fully
2092 bracketed, but that for @samp{b} is fully bracketed.
2095 int a[2][2] = @{ 0, 1, 2, 3 @};
2096 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2100 @opindex Wparentheses
2101 Warn if parentheses are omitted in certain contexts, such
2102 as when there is an assignment in a context where a truth value
2103 is expected, or when operators are nested whose precedence people
2104 often get confused about.
2106 Also warn about constructions where there may be confusion to which
2107 @code{if} statement an @code{else} branch belongs. Here is an example of
2122 In C, every @code{else} branch belongs to the innermost possible @code{if}
2123 statement, which in this example is @code{if (b)}. This is often not
2124 what the programmer expected, as illustrated in the above example by
2125 indentation the programmer chose. When there is the potential for this
2126 confusion, GCC will issue a warning when this flag is specified.
2127 To eliminate the warning, add explicit braces around the innermost
2128 @code{if} statement so there is no way the @code{else} could belong to
2129 the enclosing @code{if}. The resulting code would look like this:
2145 @item -Wsequence-point
2146 @opindex Wsequence-point
2147 Warn about code that may have undefined semantics because of violations
2148 of sequence point rules in the C standard.
2150 The C standard defines the order in which expressions in a C program are
2151 evaluated in terms of @dfn{sequence points}, which represent a partial
2152 ordering between the execution of parts of the program: those executed
2153 before the sequence point, and those executed after it. These occur
2154 after the evaluation of a full expression (one which is not part of a
2155 larger expression), after the evaluation of the first operand of a
2156 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2157 function is called (but after the evaluation of its arguments and the
2158 expression denoting the called function), and in certain other places.
2159 Other than as expressed by the sequence point rules, the order of
2160 evaluation of subexpressions of an expression is not specified. All
2161 these rules describe only a partial order rather than a total order,
2162 since, for example, if two functions are called within one expression
2163 with no sequence point between them, the order in which the functions
2164 are called is not specified. However, the standards committee have
2165 ruled that function calls do not overlap.
2167 It is not specified when between sequence points modifications to the
2168 values of objects take effect. Programs whose behavior depends on this
2169 have undefined behavior; the C standard specifies that ``Between the
2170 previous and next sequence point an object shall have its stored value
2171 modified at most once by the evaluation of an expression. Furthermore,
2172 the prior value shall be read only to determine the value to be
2173 stored.''. If a program breaks these rules, the results on any
2174 particular implementation are entirely unpredictable.
2176 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2177 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2178 diagnosed by this option, and it may give an occasional false positive
2179 result, but in general it has been found fairly effective at detecting
2180 this sort of problem in programs.
2182 The present implementation of this option only works for C programs. A
2183 future implementation may also work for C++ programs.
2185 The C standard is worded confusingly, therefore there is some debate
2186 over the precise meaning of the sequence point rules in subtle cases.
2187 Links to discussions of the problem, including proposed formal
2188 definitions, may be found on our readings page, at
2189 @w{@uref{http://gcc.gnu.org/readings.html}}.
2192 @opindex Wreturn-type
2193 Warn whenever a function is defined with a return-type that defaults to
2194 @code{int}. Also warn about any @code{return} statement with no
2195 return-value in a function whose return-type is not @code{void}.
2197 For C++, a function without return type always produces a diagnostic
2198 message, even when @option{-Wno-return-type} is specified. The only
2199 exceptions are @samp{main} and functions defined in system headers.
2203 Warn whenever a @code{switch} statement has an index of enumeral type
2204 and lacks a @code{case} for one or more of the named codes of that
2205 enumeration. (The presence of a @code{default} label prevents this
2206 warning.) @code{case} labels outside the enumeration range also
2207 provoke warnings when this option is used.
2209 @item -Wswitch-default
2210 @opindex Wswitch-switch
2211 Warn whenever a @code{switch} statement does not have a @code{default}
2215 @opindex Wswitch-enum
2216 Warn whenever a @code{switch} statement has an index of enumeral type
2217 and lacks a @code{case} for one or more of the named codes of that
2218 enumeration. @code{case} labels outside the enumeration range also
2219 provoke warnings when this option is used.
2223 Warn if any trigraphs are encountered that might change the meaning of
2224 the program (trigraphs within comments are not warned about).
2226 @item -Wunused-function
2227 @opindex Wunused-function
2228 Warn whenever a static function is declared but not defined or a
2229 non\-inline static function is unused.
2231 @item -Wunused-label
2232 @opindex Wunused-label
2233 Warn whenever a label is declared but not used.
2235 To suppress this warning use the @samp{unused} attribute
2236 (@pxref{Variable Attributes}).
2238 @item -Wunused-parameter
2239 @opindex Wunused-parameter
2240 Warn whenever a function parameter is unused aside from its declaration.
2242 To suppress this warning use the @samp{unused} attribute
2243 (@pxref{Variable Attributes}).
2245 @item -Wunused-variable
2246 @opindex Wunused-variable
2247 Warn whenever a local variable or non-constant static variable is unused
2248 aside from its declaration
2250 To suppress this warning use the @samp{unused} attribute
2251 (@pxref{Variable Attributes}).
2253 @item -Wunused-value
2254 @opindex Wunused-value
2255 Warn whenever a statement computes a result that is explicitly not used.
2257 To suppress this warning cast the expression to @samp{void}.
2261 All the above @option{-Wunused} options combined.
2263 In order to get a warning about an unused function parameter, you must
2264 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2265 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2267 @item -Wuninitialized
2268 @opindex Wuninitialized
2269 Warn if an automatic variable is used without first being initialized or
2270 if a variable may be clobbered by a @code{setjmp} call.
2272 These warnings are possible only in optimizing compilation,
2273 because they require data flow information that is computed only
2274 when optimizing. If you don't specify @option{-O}, you simply won't
2277 These warnings occur only for variables that are candidates for
2278 register allocation. Therefore, they do not occur for a variable that
2279 is declared @code{volatile}, or whose address is taken, or whose size
2280 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2281 structures, unions or arrays, even when they are in registers.
2283 Note that there may be no warning about a variable that is used only
2284 to compute a value that itself is never used, because such
2285 computations may be deleted by data flow analysis before the warnings
2288 These warnings are made optional because GCC is not smart
2289 enough to see all the reasons why the code might be correct
2290 despite appearing to have an error. Here is one example of how
2311 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2312 always initialized, but GCC doesn't know this. Here is
2313 another common case:
2318 if (change_y) save_y = y, y = new_y;
2320 if (change_y) y = save_y;
2325 This has no bug because @code{save_y} is used only if it is set.
2327 @cindex @code{longjmp} warnings
2328 This option also warns when a non-volatile automatic variable might be
2329 changed by a call to @code{longjmp}. These warnings as well are possible
2330 only in optimizing compilation.
2332 The compiler sees only the calls to @code{setjmp}. It cannot know
2333 where @code{longjmp} will be called; in fact, a signal handler could
2334 call it at any point in the code. As a result, you may get a warning
2335 even when there is in fact no problem because @code{longjmp} cannot
2336 in fact be called at the place which would cause a problem.
2338 Some spurious warnings can be avoided if you declare all the functions
2339 you use that never return as @code{noreturn}. @xref{Function
2342 @item -Wunknown-pragmas
2343 @opindex Wunknown-pragmas
2344 @cindex warning for unknown pragmas
2345 @cindex unknown pragmas, warning
2346 @cindex pragmas, warning of unknown
2347 Warn when a #pragma directive is encountered which is not understood by
2348 GCC@. If this command line option is used, warnings will even be issued
2349 for unknown pragmas in system header files. This is not the case if
2350 the warnings were only enabled by the @option{-Wall} command line option.
2352 @item -Wstrict-aliasing
2353 @opindex Wstrict-aliasing
2354 This option is only active when @option{-fstrict-aliasing} is active.
2355 It warns about code which might break the strict aliasing rules that the
2356 compiler is using for optimization. The warning does not catch all
2357 cases, but does attempt to catch the more common pitfalls. It is
2358 included in @option{-Wall}.
2362 All of the above @samp{-W} options combined. This enables all the
2363 warnings about constructions that some users consider questionable, and
2364 that are easy to avoid (or modify to prevent the warning), even in
2365 conjunction with macros. This also enables some language-specific
2366 warnings described in @ref{C++ Dialect Options} and
2367 @ref{Objective-C Dialect Options}.
2370 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2371 Some of them warn about constructions that users generally do not
2372 consider questionable, but which occasionally you might wish to check
2373 for; others warn about constructions that are necessary or hard to avoid
2374 in some cases, and there is no simple way to modify the code to suppress
2381 (This option used to be called @option{-W}. The older name is still
2382 supported, but the newer name is more descriptive.) Print extra warning
2383 messages for these events:
2387 A function can return either with or without a value. (Falling
2388 off the end of the function body is considered returning without
2389 a value.) For example, this function would evoke such a
2403 An expression-statement or the left-hand side of a comma expression
2404 contains no side effects.
2405 To suppress the warning, cast the unused expression to void.
2406 For example, an expression such as @samp{x[i,j]} will cause a warning,
2407 but @samp{x[(void)i,j]} will not.
2410 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2413 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2414 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2415 that of ordinary mathematical notation.
2418 Storage-class specifiers like @code{static} are not the first things in
2419 a declaration. According to the C Standard, this usage is obsolescent.
2422 The return type of a function has a type qualifier such as @code{const}.
2423 Such a type qualifier has no effect, since the value returned by a
2424 function is not an lvalue. (But don't warn about the GNU extension of
2425 @code{volatile void} return types. That extension will be warned about
2426 if @option{-pedantic} is specified.)
2429 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2433 A comparison between signed and unsigned values could produce an
2434 incorrect result when the signed value is converted to unsigned.
2435 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2438 An aggregate has an initializer which does not initialize all members.
2439 For example, the following code would cause such a warning, because
2440 @code{x.h} would be implicitly initialized to zero:
2443 struct s @{ int f, g, h; @};
2444 struct s x = @{ 3, 4 @};
2448 A function parameter is declared without a type specifier in K&R-style
2456 An empty body occurs in an @samp{if} or @samp{else} statement.
2459 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2460 @samp{>}, or @samp{>=}.
2463 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2466 Any of several floating-point events that often indicate errors, such as
2467 overflow, underflow, loss of precision, etc.
2469 @item @r{(C++ only)}
2470 An enumerator and a non-enumerator both appear in a conditional expression.
2472 @item @r{(C++ only)}
2473 A non-static reference or non-static @samp{const} member appears in a
2474 class without constructors.
2476 @item @r{(C++ only)}
2477 Ambiguous virtual bases.
2479 @item @r{(C++ only)}
2480 Subscripting an array which has been declared @samp{register}.
2482 @item @r{(C++ only)}
2483 Taking the address of a variable which has been declared @samp{register}.
2485 @item @r{(C++ only)}
2486 A base class is not initialized in a derived class' copy constructor.
2489 @item -Wno-div-by-zero
2490 @opindex Wno-div-by-zero
2491 @opindex Wdiv-by-zero
2492 Do not warn about compile-time integer division by zero. Floating point
2493 division by zero is not warned about, as it can be a legitimate way of
2494 obtaining infinities and NaNs.
2496 @item -Wsystem-headers
2497 @opindex Wsystem-headers
2498 @cindex warnings from system headers
2499 @cindex system headers, warnings from
2500 Print warning messages for constructs found in system header files.
2501 Warnings from system headers are normally suppressed, on the assumption
2502 that they usually do not indicate real problems and would only make the
2503 compiler output harder to read. Using this command line option tells
2504 GCC to emit warnings from system headers as if they occurred in user
2505 code. However, note that using @option{-Wall} in conjunction with this
2506 option will @emph{not} warn about unknown pragmas in system
2507 headers---for that, @option{-Wunknown-pragmas} must also be used.
2510 @opindex Wfloat-equal
2511 Warn if floating point values are used in equality comparisons.
2513 The idea behind this is that sometimes it is convenient (for the
2514 programmer) to consider floating-point values as approximations to
2515 infinitely precise real numbers. If you are doing this, then you need
2516 to compute (by analyzing the code, or in some other way) the maximum or
2517 likely maximum error that the computation introduces, and allow for it
2518 when performing comparisons (and when producing output, but that's a
2519 different problem). In particular, instead of testing for equality, you
2520 would check to see whether the two values have ranges that overlap; and
2521 this is done with the relational operators, so equality comparisons are
2524 @item -Wtraditional @r{(C only)}
2525 @opindex Wtraditional
2526 Warn about certain constructs that behave differently in traditional and
2527 ISO C@. Also warn about ISO C constructs that have no traditional C
2528 equivalent, and/or problematic constructs which should be avoided.
2532 Macro parameters that appear within string literals in the macro body.
2533 In traditional C macro replacement takes place within string literals,
2534 but does not in ISO C@.
2537 In traditional C, some preprocessor directives did not exist.
2538 Traditional preprocessors would only consider a line to be a directive
2539 if the @samp{#} appeared in column 1 on the line. Therefore
2540 @option{-Wtraditional} warns about directives that traditional C
2541 understands but would ignore because the @samp{#} does not appear as the
2542 first character on the line. It also suggests you hide directives like
2543 @samp{#pragma} not understood by traditional C by indenting them. Some
2544 traditional implementations would not recognize @samp{#elif}, so it
2545 suggests avoiding it altogether.
2548 A function-like macro that appears without arguments.
2551 The unary plus operator.
2554 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2555 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2556 constants.) Note, these suffixes appear in macros defined in the system
2557 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2558 Use of these macros in user code might normally lead to spurious
2559 warnings, however gcc's integrated preprocessor has enough context to
2560 avoid warning in these cases.
2563 A function declared external in one block and then used after the end of
2567 A @code{switch} statement has an operand of type @code{long}.
2570 A non-@code{static} function declaration follows a @code{static} one.
2571 This construct is not accepted by some traditional C compilers.
2574 The ISO type of an integer constant has a different width or
2575 signedness from its traditional type. This warning is only issued if
2576 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2577 typically represent bit patterns, are not warned about.
2580 Usage of ISO string concatenation is detected.
2583 Initialization of automatic aggregates.
2586 Identifier conflicts with labels. Traditional C lacks a separate
2587 namespace for labels.
2590 Initialization of unions. If the initializer is zero, the warning is
2591 omitted. This is done under the assumption that the zero initializer in
2592 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2593 initializer warnings and relies on default initialization to zero in the
2597 Conversions by prototypes between fixed/floating point values and vice
2598 versa. The absence of these prototypes when compiling with traditional
2599 C would cause serious problems. This is a subset of the possible
2600 conversion warnings, for the full set use @option{-Wconversion}.
2603 Use of ISO C style function definitions. This warning intentionally is
2604 @emph{not} issued for prototype declarations or variadic functions
2605 because these ISO C features will appear in your code when using
2606 libiberty's traditional C compatibility macros, @code{PARAMS} and
2607 @code{VPARAMS}. This warning is also bypassed for nested functions
2608 because that feature is already a gcc extension and thus not relevant to
2609 traditional C compatibility.
2614 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2616 @item -Wendif-labels
2617 @opindex Wendif-labels
2618 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2622 Warn whenever a local variable shadows another local variable, parameter or
2623 global variable or whenever a built-in function is shadowed.
2625 @item -Wlarger-than-@var{len}
2626 @opindex Wlarger-than
2627 Warn whenever an object of larger than @var{len} bytes is defined.
2629 @item -Wpointer-arith
2630 @opindex Wpointer-arith
2631 Warn about anything that depends on the ``size of'' a function type or
2632 of @code{void}. GNU C assigns these types a size of 1, for
2633 convenience in calculations with @code{void *} pointers and pointers
2636 @item -Wbad-function-cast @r{(C only)}
2637 @opindex Wbad-function-cast
2638 Warn whenever a function call is cast to a non-matching type.
2639 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2643 Warn whenever a pointer is cast so as to remove a type qualifier from
2644 the target type. For example, warn if a @code{const char *} is cast
2645 to an ordinary @code{char *}.
2648 @opindex Wcast-align
2649 Warn whenever a pointer is cast such that the required alignment of the
2650 target is increased. For example, warn if a @code{char *} is cast to
2651 an @code{int *} on machines where integers can only be accessed at
2652 two- or four-byte boundaries.
2654 @item -Wwrite-strings
2655 @opindex Wwrite-strings
2656 When compiling C, give string constants the type @code{const
2657 char[@var{length}]} so that
2658 copying the address of one into a non-@code{const} @code{char *}
2659 pointer will get a warning; when compiling C++, warn about the
2660 deprecated conversion from string constants to @code{char *}.
2661 These warnings will help you find at
2662 compile time code that can try to write into a string constant, but
2663 only if you have been very careful about using @code{const} in
2664 declarations and prototypes. Otherwise, it will just be a nuisance;
2665 this is why we did not make @option{-Wall} request these warnings.
2668 @opindex Wconversion
2669 Warn if a prototype causes a type conversion that is different from what
2670 would happen to the same argument in the absence of a prototype. This
2671 includes conversions of fixed point to floating and vice versa, and
2672 conversions changing the width or signedness of a fixed point argument
2673 except when the same as the default promotion.
2675 Also, warn if a negative integer constant expression is implicitly
2676 converted to an unsigned type. For example, warn about the assignment
2677 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2678 casts like @code{(unsigned) -1}.
2680 @item -Wsign-compare
2681 @opindex Wsign-compare
2682 @cindex warning for comparison of signed and unsigned values
2683 @cindex comparison of signed and unsigned values, warning
2684 @cindex signed and unsigned values, comparison warning
2685 Warn when a comparison between signed and unsigned values could produce
2686 an incorrect result when the signed value is converted to unsigned.
2687 This warning is also enabled by @option{-Wextra}; to get the other warnings
2688 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2690 @item -Waggregate-return
2691 @opindex Waggregate-return
2692 Warn if any functions that return structures or unions are defined or
2693 called. (In languages where you can return an array, this also elicits
2696 @item -Wstrict-prototypes @r{(C only)}
2697 @opindex Wstrict-prototypes
2698 Warn if a function is declared or defined without specifying the
2699 argument types. (An old-style function definition is permitted without
2700 a warning if preceded by a declaration which specifies the argument
2703 @item -Wmissing-prototypes @r{(C only)}
2704 @opindex Wmissing-prototypes
2705 Warn if a global function is defined without a previous prototype
2706 declaration. This warning is issued even if the definition itself
2707 provides a prototype. The aim is to detect global functions that fail
2708 to be declared in header files.
2710 @item -Wmissing-declarations @r{(C only)}
2711 @opindex Wmissing-declarations
2712 Warn if a global function is defined without a previous declaration.
2713 Do so even if the definition itself provides a prototype.
2714 Use this option to detect global functions that are not declared in
2717 @item -Wmissing-noreturn
2718 @opindex Wmissing-noreturn
2719 Warn about functions which might be candidates for attribute @code{noreturn}.
2720 Note these are only possible candidates, not absolute ones. Care should
2721 be taken to manually verify functions actually do not ever return before
2722 adding the @code{noreturn} attribute, otherwise subtle code generation
2723 bugs could be introduced. You will not get a warning for @code{main} in
2724 hosted C environments.
2726 @item -Wmissing-format-attribute
2727 @opindex Wmissing-format-attribute
2729 If @option{-Wformat} is enabled, also warn about functions which might be
2730 candidates for @code{format} attributes. Note these are only possible
2731 candidates, not absolute ones. GCC will guess that @code{format}
2732 attributes might be appropriate for any function that calls a function
2733 like @code{vprintf} or @code{vscanf}, but this might not always be the
2734 case, and some functions for which @code{format} attributes are
2735 appropriate may not be detected. This option has no effect unless
2736 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2738 @item -Wno-multichar
2739 @opindex Wno-multichar
2741 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2742 Usually they indicate a typo in the user's code, as they have
2743 implementation-defined values, and should not be used in portable code.
2745 @item -Wno-deprecated-declarations
2746 @opindex Wno-deprecated-declarations
2747 Do not warn about uses of functions, variables, and types marked as
2748 deprecated by using the @code{deprecated} attribute.
2749 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2750 @pxref{Type Attributes}.)
2754 Warn if a structure is given the packed attribute, but the packed
2755 attribute has no effect on the layout or size of the structure.
2756 Such structures may be mis-aligned for little benefit. For
2757 instance, in this code, the variable @code{f.x} in @code{struct bar}
2758 will be misaligned even though @code{struct bar} does not itself
2759 have the packed attribute:
2766 @} __attribute__((packed));
2776 Warn if padding is included in a structure, either to align an element
2777 of the structure or to align the whole structure. Sometimes when this
2778 happens it is possible to rearrange the fields of the structure to
2779 reduce the padding and so make the structure smaller.
2781 @item -Wredundant-decls
2782 @opindex Wredundant-decls
2783 Warn if anything is declared more than once in the same scope, even in
2784 cases where multiple declaration is valid and changes nothing.
2786 @item -Wnested-externs @r{(C only)}
2787 @opindex Wnested-externs
2788 Warn if an @code{extern} declaration is encountered within a function.
2790 @item -Wunreachable-code
2791 @opindex Wunreachable-code
2792 Warn if the compiler detects that code will never be executed.
2794 This option is intended to warn when the compiler detects that at
2795 least a whole line of source code will never be executed, because
2796 some condition is never satisfied or because it is after a
2797 procedure that never returns.
2799 It is possible for this option to produce a warning even though there
2800 are circumstances under which part of the affected line can be executed,
2801 so care should be taken when removing apparently-unreachable code.
2803 For instance, when a function is inlined, a warning may mean that the
2804 line is unreachable in only one inlined copy of the function.
2806 This option is not made part of @option{-Wall} because in a debugging
2807 version of a program there is often substantial code which checks
2808 correct functioning of the program and is, hopefully, unreachable
2809 because the program does work. Another common use of unreachable
2810 code is to provide behavior which is selectable at compile-time.
2814 Warn if a function can not be inlined and it was declared as inline.
2815 Even with this option, the compiler will not warn about failures to
2816 inline functions declared in system headers.
2818 The compiler uses a variety of heuristics to determine whether or not
2819 to inline a function. For example, the compiler takes into account
2820 the size of the function being inlined and the the amount of inlining
2821 that has already been done in the current function. Therefore,
2822 seemingly insignificant changes in the source program can cause the
2823 warnings produced by @option{-Winline} to appear or disappear.
2825 @item -Wno-invalid-offsetof @r{(C++ only)}
2826 @opindex Wno-invalid-offsetof
2827 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
2828 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
2829 to a non-POD type is undefined. In existing C++ implementations,
2830 however, @samp{offsetof} typically gives meaningful results even when
2831 applied to certain kinds of non-POD types. (Such as a simple
2832 @samp{struct} that fails to be a POD type only by virtue of having a
2833 constructor.) This flag is for users who are aware that they are
2834 writing nonportable code and who have deliberately chosen to ignore the
2837 The restrictions on @samp{offsetof} may be relaxed in a future version
2838 of the C++ standard.
2841 @opindex Winvalid-pch
2842 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
2843 the search path but can't be used.
2847 @opindex Wno-long-long
2848 Warn if @samp{long long} type is used. This is default. To inhibit
2849 the warning messages, use @option{-Wno-long-long}. Flags
2850 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2851 only when @option{-pedantic} flag is used.
2853 @item -Wdisabled-optimization
2854 @opindex Wdisabled-optimization
2855 Warn if a requested optimization pass is disabled. This warning does
2856 not generally indicate that there is anything wrong with your code; it
2857 merely indicates that GCC's optimizers were unable to handle the code
2858 effectively. Often, the problem is that your code is too big or too
2859 complex; GCC will refuse to optimize programs when the optimization
2860 itself is likely to take inordinate amounts of time.
2864 Make all warnings into errors.
2867 @node Debugging Options
2868 @section Options for Debugging Your Program or GCC
2869 @cindex options, debugging
2870 @cindex debugging information options
2872 GCC has various special options that are used for debugging
2873 either your program or GCC:
2878 Produce debugging information in the operating system's native format
2879 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2882 On most systems that use stabs format, @option{-g} enables use of extra
2883 debugging information that only GDB can use; this extra information
2884 makes debugging work better in GDB but will probably make other debuggers
2886 refuse to read the program. If you want to control for certain whether
2887 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
2888 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, @option{-gdwarf-1},
2889 or @option{-gvms} (see below).
2891 Unlike most other C compilers, GCC allows you to use @option{-g} with
2892 @option{-O}. The shortcuts taken by optimized code may occasionally
2893 produce surprising results: some variables you declared may not exist
2894 at all; flow of control may briefly move where you did not expect it;
2895 some statements may not be executed because they compute constant
2896 results or their values were already at hand; some statements may
2897 execute in different places because they were moved out of loops.
2899 Nevertheless it proves possible to debug optimized output. This makes
2900 it reasonable to use the optimizer for programs that might have bugs.
2902 The following options are useful when GCC is generated with the
2903 capability for more than one debugging format.
2907 Produce debugging information for use by GDB@. This means to use the
2908 most expressive format available (DWARF 2, stabs, or the native format
2909 if neither of those are supported), including GDB extensions if at all
2914 Produce debugging information in stabs format (if that is supported),
2915 without GDB extensions. This is the format used by DBX on most BSD
2916 systems. On MIPS, Alpha and System V Release 4 systems this option
2917 produces stabs debugging output which is not understood by DBX or SDB@.
2918 On System V Release 4 systems this option requires the GNU assembler.
2920 @item -feliminate-unused-debug-symbols
2921 @opindex feliminate-unused-debug-symbols
2922 Produce debugging information in stabs format (if that is supported),
2923 for only symbols that are actually used.
2927 Produce debugging information in stabs format (if that is supported),
2928 using GNU extensions understood only by the GNU debugger (GDB)@. The
2929 use of these extensions is likely to make other debuggers crash or
2930 refuse to read the program.
2934 Produce debugging information in COFF format (if that is supported).
2935 This is the format used by SDB on most System V systems prior to
2940 Produce debugging information in XCOFF format (if that is supported).
2941 This is the format used by the DBX debugger on IBM RS/6000 systems.
2945 Produce debugging information in XCOFF format (if that is supported),
2946 using GNU extensions understood only by the GNU debugger (GDB)@. The
2947 use of these extensions is likely to make other debuggers crash or
2948 refuse to read the program, and may cause assemblers other than the GNU
2949 assembler (GAS) to fail with an error.
2953 Produce debugging information in DWARF version 1 format (if that is
2954 supported). This is the format used by SDB on most System V Release 4
2957 This option is deprecated.
2961 Produce debugging information in DWARF version 1 format (if that is
2962 supported), using GNU extensions understood only by the GNU debugger
2963 (GDB)@. The use of these extensions is likely to make other debuggers
2964 crash or refuse to read the program.
2966 This option is deprecated.
2970 Produce debugging information in DWARF version 2 format (if that is
2971 supported). This is the format used by DBX on IRIX 6.
2975 Produce debugging information in VMS debug format (if that is
2976 supported). This is the format used by DEBUG on VMS systems.
2979 @itemx -ggdb@var{level}
2980 @itemx -gstabs@var{level}
2981 @itemx -gcoff@var{level}
2982 @itemx -gxcoff@var{level}
2983 @itemx -gvms@var{level}
2984 Request debugging information and also use @var{level} to specify how
2985 much information. The default level is 2.
2987 Level 1 produces minimal information, enough for making backtraces in
2988 parts of the program that you don't plan to debug. This includes
2989 descriptions of functions and external variables, but no information
2990 about local variables and no line numbers.
2992 Level 3 includes extra information, such as all the macro definitions
2993 present in the program. Some debuggers support macro expansion when
2994 you use @option{-g3}.
2996 Note that in order to avoid confusion between DWARF1 debug level 2,
2997 and DWARF2, neither @option{-gdwarf} nor @option{-gdwarf-2} accept
2998 a concatenated debug level. Instead use an additional @option{-g@var{level}}
2999 option to change the debug level for DWARF1 or DWARF2.
3001 @item -feliminate-dwarf2-dups
3002 @opindex feliminate-dwarf2-dups
3003 Compress DWARF2 debugging information by eliminating duplicated
3004 information about each symbol. This option only makes sense when
3005 generating DWARF2 debugging information with @option{-gdwarf-2}.
3007 @cindex @command{prof}
3010 Generate extra code to write profile information suitable for the
3011 analysis program @command{prof}. You must use this option when compiling
3012 the source files you want data about, and you must also use it when
3015 @cindex @command{gprof}
3018 Generate extra code to write profile information suitable for the
3019 analysis program @command{gprof}. You must use this option when compiling
3020 the source files you want data about, and you must also use it when
3025 Makes the compiler print out each function name as it is compiled, and
3026 print some statistics about each pass when it finishes.
3029 @opindex ftime-report
3030 Makes the compiler print some statistics about the time consumed by each
3031 pass when it finishes.
3034 @opindex fmem-report
3035 Makes the compiler print some statistics about permanent memory
3036 allocation when it finishes.
3038 @item -fprofile-arcs
3039 @opindex fprofile-arcs
3040 Add code so that program flow @dfn{arcs} are instrumented. During
3041 execution the program records how many times each branch and call is
3042 executed and how many times it is taken or returns. When the compiled
3043 program exits it saves this data to a file called
3044 @file{@var{auxname}.da} for each source file. The data may be used for
3045 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3046 test coverage analysis (@option{-ftest-coverage}). Each object file's
3047 @var{auxname} is generated from the name of the output file, if
3048 explicitly specified and it is not the final executable, otherwise it is
3049 the basename of the source file. In both cases any suffix is removed
3050 (e.g. @file{foo.da} for input file @file{dir/foo.c}, or
3051 @file{dir/foo.da} for output file specified as @option{-o dir/foo.o}).
3056 Compile the source files with @option{-fprofile-arcs} plus optimization
3057 and code generation options. For test coverage analysis, use the
3058 additional @option{-ftest-coverage} option. You do not need to profile
3059 every source file in a program.
3062 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3063 (the latter implies the former).
3066 Run the program on a representative workload to generate the arc profile
3067 information. This may be repeated any number of times. You can run
3068 concurrent instances of your program, and provided that the file system
3069 supports locking, the data files will be correctly updated. Also
3070 @code{fork} calls are detected and correctly handled (double counting
3074 For profile-directed optimizations, compile the source files again with
3075 the same optimization and code generation options plus
3076 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3077 Control Optimization}).
3080 For test coverage analysis, use @command{gcov} to produce human readable
3081 information from the @file{.bbg} and @file{.da} files. Refer to the
3082 @command{gcov} documentation for further information.
3086 With @option{-fprofile-arcs}, for each function of your program GCC
3087 creates a program flow graph, then finds a spanning tree for the graph.
3088 Only arcs that are not on the spanning tree have to be instrumented: the
3089 compiler adds code to count the number of times that these arcs are
3090 executed. When an arc is the only exit or only entrance to a block, the
3091 instrumentation code can be added to the block; otherwise, a new basic
3092 block must be created to hold the instrumentation code.
3095 @item -ftest-coverage
3096 @opindex ftest-coverage
3097 Produce a graph file that the @command{gcov} code-coverage utility
3098 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3099 show program coverage. Each source file's data file is called
3100 @file{@var{auxname}.bbg}. Refer to the @option{-fprofile-arcs} option
3101 above for a description of @var{auxname} and instructions on how to
3102 generate test coverage data. Coverage data will match the source files
3103 more closely, if you do not optimize.
3105 @item -d@var{letters}
3107 Says to make debugging dumps during compilation at times specified by
3108 @var{letters}. This is used for debugging the compiler. The file names
3109 for most of the dumps are made by appending a pass number and a word to
3110 the @var{dumpname}. @var{dumpname} is generated from the name of the
3111 output file, if explicitly specified and it is not an executable,
3112 otherwise it is the basename of the source file. In both cases any
3113 suffix is removed (e.g. @file{foo.00.rtl} or @file{foo.01.sibling}).
3114 Here are the possible letters for use in @var{letters}, and their
3120 Annotate the assembler output with miscellaneous debugging information.
3123 Dump after computing branch probabilities, to @file{@var{file}.16.bp}.
3126 Dump after block reordering, to @file{@var{file}.32.bbro}.
3129 Dump after instruction combination, to the file @file{@var{file}.22.combine}.
3132 Dump after the first if conversion, to the file @file{@var{file}.17.ce1}.
3133 Also dump after the second if conversion, to the file @file{@var{file}.23.ce2}.
3136 Dump after branch target load optimization, to to @file{@var{file}.34.btl}.
3137 Also dump after delayed branch scheduling, to @file{@var{file}.37.dbr}.
3140 Dump all macro definitions, at the end of preprocessing, in addition to
3144 Dump after SSA optimizations, to @file{@var{file}.05.ssa} and
3145 @file{@var{file}.010.ussa}.
3148 Dump after the second if conversion, to @file{@var{file}.33.ce3}.
3151 Dump after control and data flow analysis, to @file{@var{file}.15.cfg}.
3152 Also dump after life analysis, to @file{@var{file}.21.life}.
3155 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.11.addressof}.
3158 Dump after global register allocation, to @file{@var{file}.27.greg}.
3161 Dump after GCSE, to @file{@var{file}.12.gcse}.
3162 Also dump after jump bypassing and control flow optimizations, to
3163 @file{@var{file}.14.bypass}.
3166 Dump after finalization of EH handling code, to @file{@var{file}.03.eh}.
3169 Dump after sibling call optimizations, to @file{@var{file}.02.sibling}.
3172 Dump after the first jump optimization, to @file{@var{file}.04.jump}.
3175 Dump after conversion from registers to stack, to @file{@var{file}.36.stack}.
3178 Dump after local register allocation, to @file{@var{file}.26.lreg}.
3181 Dump after loop optimization passes, to @file{@var{file}.13.loop} and
3182 @file{@var{file}.19.loop2}.
3185 Dump after performing the machine dependent reorganization pass, to
3186 @file{@var{file}.37.mach}.
3189 Dump after register renumbering, to @file{@var{file}.31.rnreg}.
3192 Dump after the register move pass, to @file{@var{file}.24.regmove}.
3195 Dump after post-reload optimizations, to @file{@var{file}.28.postreload}.
3198 Dump after RTL generation, to @file{@var{file}.01.rtl}.
3201 Dump after the second scheduling pass, to @file{@var{file}.35.sched2}.
3204 Dump after CSE (including the jump optimization that sometimes follows
3205 CSE), to @file{@var{file}.019.cse}.
3208 Dump after the first scheduling pass, to @file{@var{file}.25.sched}.
3211 Dump after the second CSE pass (including the jump optimization that
3212 sometimes follows CSE), to @file{@var{file}.20.cse2}.
3215 Dump after running tracer, to @file{@var{file}.18.tracer}.
3218 Dump after null pointer elimination pass to @file{@var{file}.018.null}.
3221 Dump callgraph and unit-at-a-time optimization @file{@var{file}.00.unit}.
3224 Dump after the second flow pass, to @file{@var{file}.29.flow2}.
3227 Dump after SSA conditional constant propagation, to
3228 @file{@var{file}.06.ssaccp}.
3231 Dump after SSA dead code elimination, to @file{@var{file}.07.ssadce}.
3234 Dump after the peephole pass, to @file{@var{file}.30.peephole2}.
3237 Produce all the dumps listed above.
3240 Produce a core dump whenever an error occurs.
3243 Print statistics on memory usage, at the end of the run, to
3247 Annotate the assembler output with a comment indicating which
3248 pattern and alternative was used. The length of each instruction is
3252 Dump the RTL in the assembler output as a comment before each instruction.
3253 Also turns on @option{-dp} annotation.
3256 For each of the other indicated dump files (except for
3257 @file{@var{file}.01.rtl}), dump a representation of the control flow graph
3258 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3261 Just generate RTL for a function instead of compiling it. Usually used
3265 Dump debugging information during parsing, to standard error.
3268 @item -fdump-unnumbered
3269 @opindex fdump-unnumbered
3270 When doing debugging dumps (see @option{-d} option above), suppress instruction
3271 numbers and line number note output. This makes it more feasible to
3272 use diff on debugging dumps for compiler invocations with different
3273 options, in particular with and without @option{-g}.
3275 @item -fdump-translation-unit @r{(C and C++ only)}
3276 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3277 @opindex fdump-translation-unit
3278 Dump a representation of the tree structure for the entire translation
3279 unit to a file. The file name is made by appending @file{.tu} to the
3280 source file name. If the @samp{-@var{options}} form is used, @var{options}
3281 controls the details of the dump as described for the
3282 @option{-fdump-tree} options.
3284 @item -fdump-class-hierarchy @r{(C++ only)}
3285 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3286 @opindex fdump-class-hierarchy
3287 Dump a representation of each class's hierarchy and virtual function
3288 table layout to a file. The file name is made by appending @file{.class}
3289 to the source file name. If the @samp{-@var{options}} form is used,
3290 @var{options} controls the details of the dump as described for the
3291 @option{-fdump-tree} options.
3293 @item -fdump-tree-@var{switch} @r{(C++ only)}
3294 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3296 Control the dumping at various stages of processing the intermediate
3297 language tree to a file. The file name is generated by appending a switch
3298 specific suffix to the source file name. If the @samp{-@var{options}}
3299 form is used, @var{options} is a list of @samp{-} separated options that
3300 control the details of the dump. Not all options are applicable to all
3301 dumps, those which are not meaningful will be ignored. The following
3302 options are available
3306 Print the address of each node. Usually this is not meaningful as it
3307 changes according to the environment and source file. Its primary use
3308 is for tying up a dump file with a debug environment.
3310 Inhibit dumping of members of a scope or body of a function merely
3311 because that scope has been reached. Only dump such items when they
3312 are directly reachable by some other path.
3314 Turn on all options.
3317 The following tree dumps are possible:
3320 Dump before any tree based optimization, to @file{@var{file}.original}.
3322 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3324 Dump after function inlining, to @file{@var{file}.inlined}.
3327 @item -frandom-seed=@var{string}
3328 @opindex frandom-string
3329 This option provides a seed that GCC uses when it would otherwise use
3330 random numbers. It is used to generate certain symbol names
3331 that have to be different in every compiled file. It is also used to
3332 place unique stamps in coverage data files and the object files that
3333 produce them. You can use the @option{-frandom-seed} option to produce
3334 reproducibly identical object files.
3336 The @var{string} should be different for every file you compile.
3338 @item -fsched-verbose=@var{n}
3339 @opindex fsched-verbose
3340 On targets that use instruction scheduling, this option controls the
3341 amount of debugging output the scheduler prints. This information is
3342 written to standard error, unless @option{-dS} or @option{-dR} is
3343 specified, in which case it is output to the usual dump
3344 listing file, @file{.sched} or @file{.sched2} respectively. However
3345 for @var{n} greater than nine, the output is always printed to standard
3348 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3349 same information as @option{-dRS}. For @var{n} greater than one, it
3350 also output basic block probabilities, detailed ready list information
3351 and unit/insn info. For @var{n} greater than two, it includes RTL
3352 at abort point, control-flow and regions info. And for @var{n} over
3353 four, @option{-fsched-verbose} also includes dependence info.
3357 Store the usual ``temporary'' intermediate files permanently; place them
3358 in the current directory and name them based on the source file. Thus,
3359 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3360 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3361 preprocessed @file{foo.i} output file even though the compiler now
3362 normally uses an integrated preprocessor.
3366 Report the CPU time taken by each subprocess in the compilation
3367 sequence. For C source files, this is the compiler proper and assembler
3368 (plus the linker if linking is done). The output looks like this:
3375 The first number on each line is the ``user time,'' that is time spent
3376 executing the program itself. The second number is ``system time,''
3377 time spent executing operating system routines on behalf of the program.
3378 Both numbers are in seconds.
3380 @item -print-file-name=@var{library}
3381 @opindex print-file-name
3382 Print the full absolute name of the library file @var{library} that
3383 would be used when linking---and don't do anything else. With this
3384 option, GCC does not compile or link anything; it just prints the
3387 @item -print-multi-directory
3388 @opindex print-multi-directory
3389 Print the directory name corresponding to the multilib selected by any
3390 other switches present in the command line. This directory is supposed
3391 to exist in @env{GCC_EXEC_PREFIX}.
3393 @item -print-multi-lib
3394 @opindex print-multi-lib
3395 Print the mapping from multilib directory names to compiler switches
3396 that enable them. The directory name is separated from the switches by
3397 @samp{;}, and each switch starts with an @samp{@@} instead of the
3398 @samp{-}, without spaces between multiple switches. This is supposed to
3399 ease shell-processing.
3401 @item -print-prog-name=@var{program}
3402 @opindex print-prog-name
3403 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3405 @item -print-libgcc-file-name
3406 @opindex print-libgcc-file-name
3407 Same as @option{-print-file-name=libgcc.a}.
3409 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3410 but you do want to link with @file{libgcc.a}. You can do
3413 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3416 @item -print-search-dirs
3417 @opindex print-search-dirs
3418 Print the name of the configured installation directory and a list of
3419 program and library directories gcc will search---and don't do anything else.
3421 This is useful when gcc prints the error message
3422 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3423 To resolve this you either need to put @file{cpp0} and the other compiler
3424 components where gcc expects to find them, or you can set the environment
3425 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3426 Don't forget the trailing '/'.
3427 @xref{Environment Variables}.
3430 @opindex dumpmachine
3431 Print the compiler's target machine (for example,
3432 @samp{i686-pc-linux-gnu})---and don't do anything else.
3435 @opindex dumpversion
3436 Print the compiler version (for example, @samp{3.0})---and don't do
3441 Print the compiler's built-in specs---and don't do anything else. (This
3442 is used when GCC itself is being built.) @xref{Spec Files}.
3444 @item -feliminate-unused-debug-types
3445 @opindex feliminate-unused-debug-types
3446 Normally, when producing DWARF2 output, GCC will emit debugging
3447 information for all types declared in a compilation
3448 unit, regardless of whether or not they are actually used
3449 in that compilation unit. Sometimes this is useful, such as
3450 if, in the debugger, you want to cast a value to a type that is
3451 not actually used in your program (but is declared). More often,
3452 however, this results in a significant amount of wasted space.
3453 With this option, GCC will avoid producing debug symbol output
3454 for types that are nowhere used in the source file being compiled.
3457 @node Optimize Options
3458 @section Options That Control Optimization
3459 @cindex optimize options
3460 @cindex options, optimization
3462 These options control various sorts of optimizations.
3464 Without any optimization option, the compiler's goal is to reduce the
3465 cost of compilation and to make debugging produce the expected
3466 results. Statements are independent: if you stop the program with a
3467 breakpoint between statements, you can then assign a new value to any
3468 variable or change the program counter to any other statement in the
3469 function and get exactly the results you would expect from the source
3472 Turning on optimization flags makes the compiler attempt to improve
3473 the performance and/or code size at the expense of compilation time
3474 and possibly the ability to debug the program.
3476 The compiler performs optimisation based on the knowledge it has of
3477 the program. Using the @option{-funit-at-a-time} flag will allow the
3478 compiler to consider information gained from later functions in the
3479 file when compiling a function. Compiling multiple files at once to a
3480 single output file (and using @option{-funit-at-a-time}) will allow
3481 the compiler to use information gained from all of the files when
3482 compiling each of them.
3484 Not all optimizations are controlled directly by a flag. Only
3485 optimizations that have a flag are listed.
3492 Optimize. Optimizing compilation takes somewhat more time, and a lot
3493 more memory for a large function.
3495 With @option{-O}, the compiler tries to reduce code size and execution
3496 time, without performing any optimizations that take a great deal of
3499 @option{-O} turns on the following optimization flags:
3500 @gccoptlist{-fdefer-pop @gol
3501 -fmerge-constants @gol
3503 -floop-optimize @gol
3505 -fif-conversion @gol
3506 -fif-conversion2 @gol
3507 -fdelayed-branch @gol
3508 -fguess-branch-probability @gol
3511 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3512 where doing so does not interfere with debugging.
3516 Optimize even more. GCC performs nearly all supported optimizations
3517 that do not involve a space-speed tradeoff. The compiler does not
3518 perform loop unrolling or function inlining when you specify @option{-O2}.
3519 As compared to @option{-O}, this option increases both compilation time
3520 and the performance of the generated code.
3522 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3523 also turns on the following optimization flags:
3524 @gccoptlist{-fforce-mem @gol
3525 -foptimize-sibling-calls @gol
3526 -fstrength-reduce @gol
3527 -fcse-follow-jumps -fcse-skip-blocks @gol
3528 -frerun-cse-after-loop -frerun-loop-opt @gol
3529 -fgcse -fgcse-lm -fgcse-sm @gol
3530 -fdelete-null-pointer-checks @gol
3531 -fexpensive-optimizations @gol
3533 -fschedule-insns -fschedule-insns2 @gol
3534 -fsched-interblock -fsched-spec @gol
3537 -freorder-blocks -freorder-functions @gol
3538 -fstrict-aliasing @gol
3539 -falign-functions -falign-jumps @gol
3540 -falign-loops -falign-labels}
3542 Please note the warning under @option{-fgcse} about
3543 invoking @option{-O2} on programs that use computed gotos.
3547 Optimize yet more. @option{-O3} turns on all optimizations specified by
3548 @option{-O2} and also turns on the @option{-finline-functions},
3549 @option{-funit-at-a-time} and @option{-frename-registers} options.
3553 Do not optimize. This is the default.
3557 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3558 do not typically increase code size. It also performs further
3559 optimizations designed to reduce code size.
3561 @option{-Os} disables the following optimization flags:
3562 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3563 -falign-labels -freorder-blocks -fprefetch-loop-arrays}
3565 If you use multiple @option{-O} options, with or without level numbers,
3566 the last such option is the one that is effective.
3569 Options of the form @option{-f@var{flag}} specify machine-independent
3570 flags. Most flags have both positive and negative forms; the negative
3571 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3572 below, only one of the forms is listed---the one you typically will
3573 use. You can figure out the other form by either removing @samp{no-}
3576 The following options control specific optimizations. They are either
3577 activated by @option{-O} options or are related to ones that are. You
3578 can use the following flags in the rare cases when ``fine-tuning'' of
3579 optimizations to be performed is desired.
3582 @item -fno-default-inline
3583 @opindex fno-default-inline
3584 Do not make member functions inline by default merely because they are
3585 defined inside the class scope (C++ only). Otherwise, when you specify
3586 @w{@option{-O}}, member functions defined inside class scope are compiled
3587 inline by default; i.e., you don't need to add @samp{inline} in front of
3588 the member function name.
3590 @item -fno-defer-pop
3591 @opindex fno-defer-pop
3592 Always pop the arguments to each function call as soon as that function
3593 returns. For machines which must pop arguments after a function call,
3594 the compiler normally lets arguments accumulate on the stack for several
3595 function calls and pops them all at once.
3597 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3601 Force memory operands to be copied into registers before doing
3602 arithmetic on them. This produces better code by making all memory
3603 references potential common subexpressions. When they are not common
3604 subexpressions, instruction combination should eliminate the separate
3607 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3610 @opindex fforce-addr
3611 Force memory address constants to be copied into registers before
3612 doing arithmetic on them. This may produce better code just as
3613 @option{-fforce-mem} may.
3615 @item -fomit-frame-pointer
3616 @opindex fomit-frame-pointer
3617 Don't keep the frame pointer in a register for functions that
3618 don't need one. This avoids the instructions to save, set up and
3619 restore frame pointers; it also makes an extra register available
3620 in many functions. @strong{It also makes debugging impossible on
3623 On some machines, such as the VAX, this flag has no effect, because
3624 the standard calling sequence automatically handles the frame pointer
3625 and nothing is saved by pretending it doesn't exist. The
3626 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3627 whether a target machine supports this flag. @xref{Registers,,Register
3628 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3630 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3632 @item -foptimize-sibling-calls
3633 @opindex foptimize-sibling-calls
3634 Optimize sibling and tail recursive calls.
3636 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3640 Don't pay attention to the @code{inline} keyword. Normally this option
3641 is used to keep the compiler from expanding any functions inline.
3642 Note that if you are not optimizing, no functions can be expanded inline.
3644 @item -finline-functions
3645 @opindex finline-functions
3646 Integrate all simple functions into their callers. The compiler
3647 heuristically decides which functions are simple enough to be worth
3648 integrating in this way.
3650 If all calls to a given function are integrated, and the function is
3651 declared @code{static}, then the function is normally not output as
3652 assembler code in its own right.
3654 Enabled at level @option{-O3}.
3656 @item -finline-limit=@var{n}
3657 @opindex finline-limit
3658 By default, gcc limits the size of functions that can be inlined. This flag
3659 allows the control of this limit for functions that are explicitly marked as
3660 inline (i.e., marked with the inline keyword or defined within the class
3661 definition in c++). @var{n} is the size of functions that can be inlined in
3662 number of pseudo instructions (not counting parameter handling). The default
3663 value of @var{n} is 600.
3664 Increasing this value can result in more inlined code at
3665 the cost of compilation time and memory consumption. Decreasing usually makes
3666 the compilation faster and less code will be inlined (which presumably
3667 means slower programs). This option is particularly useful for programs that
3668 use inlining heavily such as those based on recursive templates with C++.
3670 Inlining is actually controlled by a number of parameters, which may be
3671 specified individually by using @option{--param @var{name}=@var{value}}.
3672 The @option{-finline-limit=@var{n}} option sets some of these parameters
3676 @item max-inline-insns
3678 @item max-inline-insns-single
3679 is set to @var{n}/2.
3680 @item max-inline-insns-auto
3681 is set to @var{n}/2.
3682 @item min-inline-insns
3683 is set to 130 or @var{n}/4, whichever is smaller.
3684 @item max-inline-insns-rtl
3688 Using @option{-finline-limit=600} thus results in the default settings
3689 for these parameters. See below for a documentation of the individual
3690 parameters controlling inlining.
3692 @emph{Note:} pseudo instruction represents, in this particular context, an
3693 abstract measurement of function's size. In no way, it represents a count
3694 of assembly instructions and as such its exact meaning might change from one
3695 release to an another.
3697 @item -fkeep-inline-functions
3698 @opindex fkeep-inline-functions
3699 Even if all calls to a given function are integrated, and the function
3700 is declared @code{static}, nevertheless output a separate run-time
3701 callable version of the function. This switch does not affect
3702 @code{extern inline} functions.
3704 @item -fkeep-static-consts
3705 @opindex fkeep-static-consts
3706 Emit variables declared @code{static const} when optimization isn't turned
3707 on, even if the variables aren't referenced.
3709 GCC enables this option by default. If you want to force the compiler to
3710 check if the variable was referenced, regardless of whether or not
3711 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3713 @item -fmerge-constants
3714 Attempt to merge identical constants (string constants and floating point
3715 constants) across compilation units.
3717 This option is the default for optimized compilation if the assembler and
3718 linker support it. Use @option{-fno-merge-constants} to inhibit this
3721 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3723 @item -fmerge-all-constants
3724 Attempt to merge identical constants and identical variables.
3726 This option implies @option{-fmerge-constants}. In addition to
3727 @option{-fmerge-constants} this considers e.g. even constant initialized
3728 arrays or initialized constant variables with integral or floating point
3729 types. Languages like C or C++ require each non-automatic variable to
3730 have distinct location, so using this option will result in non-conforming
3735 Use a graph coloring register allocator. Currently this option is meant
3736 for testing, so we are interested to hear about miscompilations with
3739 @item -fno-branch-count-reg
3740 @opindex fno-branch-count-reg
3741 Do not use ``decrement and branch'' instructions on a count register,
3742 but instead generate a sequence of instructions that decrement a
3743 register, compare it against zero, then branch based upon the result.
3744 This option is only meaningful on architectures that support such
3745 instructions, which include x86, PowerPC, IA-64 and S/390.
3747 The default is @option{-fbranch-count-reg}, enabled when
3748 @option{-fstrength-reduce} is enabled.
3750 @item -fno-function-cse
3751 @opindex fno-function-cse
3752 Do not put function addresses in registers; make each instruction that
3753 calls a constant function contain the function's address explicitly.
3755 This option results in less efficient code, but some strange hacks
3756 that alter the assembler output may be confused by the optimizations
3757 performed when this option is not used.
3759 The default is @option{-ffunction-cse}
3761 @item -fno-zero-initialized-in-bss
3762 @opindex fno-zero-initialized-in-bss
3763 If the target supports a BSS section, GCC by default puts variables that
3764 are initialized to zero into BSS@. This can save space in the resulting
3767 This option turns off this behavior because some programs explicitly
3768 rely on variables going to the data section. E.g., so that the
3769 resulting executable can find the beginning of that section and/or make
3770 assumptions based on that.
3772 The default is @option{-fzero-initialized-in-bss}.
3774 @item -fstrength-reduce
3775 @opindex fstrength-reduce
3776 Perform the optimizations of loop strength reduction and
3777 elimination of iteration variables.
3779 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3781 @item -fthread-jumps
3782 @opindex fthread-jumps
3783 Perform optimizations where we check to see if a jump branches to a
3784 location where another comparison subsumed by the first is found. If
3785 so, the first branch is redirected to either the destination of the
3786 second branch or a point immediately following it, depending on whether
3787 the condition is known to be true or false.
3789 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3791 @item -fcse-follow-jumps
3792 @opindex fcse-follow-jumps
3793 In common subexpression elimination, scan through jump instructions
3794 when the target of the jump is not reached by any other path. For
3795 example, when CSE encounters an @code{if} statement with an
3796 @code{else} clause, CSE will follow the jump when the condition
3799 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3801 @item -fcse-skip-blocks
3802 @opindex fcse-skip-blocks
3803 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3804 follow jumps which conditionally skip over blocks. When CSE
3805 encounters a simple @code{if} statement with no else clause,
3806 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3807 body of the @code{if}.
3809 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3811 @item -frerun-cse-after-loop
3812 @opindex frerun-cse-after-loop
3813 Re-run common subexpression elimination after loop optimizations has been
3816 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3818 @item -frerun-loop-opt
3819 @opindex frerun-loop-opt
3820 Run the loop optimizer twice.
3822 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3826 Perform a global common subexpression elimination pass.
3827 This pass also performs global constant and copy propagation.
3829 @emph{Note:} When compiling a program using computed gotos, a GCC
3830 extension, you may get better runtime performance if you disable
3831 the global common subexpression elimination pass by adding
3832 @option{-fno-gcse} to the command line.
3834 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3838 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3839 attempt to move loads which are only killed by stores into themselves. This
3840 allows a loop containing a load/store sequence to be changed to a load outside
3841 the loop, and a copy/store within the loop.
3843 Enabled by default when gcse is enabled.
3847 When @option{-fgcse-sm} is enabled, A store motion pass is run after global common
3848 subexpression elimination. This pass will attempt to move stores out of loops.
3849 When used in conjunction with @option{-fgcse-lm}, loops containing a load/store sequence
3850 can be changed to a load before the loop and a store after the loop.
3852 Enabled by default when gcse is enabled.
3854 @item -floop-optimize
3855 @opindex floop-optimize
3856 Perform loop optimizations: move constant expressions out of loops, simplify
3857 exit test conditions and optionally do strength-reduction and loop unrolling as
3860 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3862 @item -fcrossjumping
3863 @opindex crossjumping
3864 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
3865 resulting code may or may not perform better than without cross-jumping.
3867 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3869 @item -fif-conversion
3870 @opindex if-conversion
3871 Attempt to transform conditional jumps into branch-less equivalents. This
3872 include use of conditional moves, min, max, set flags and abs instructions, and
3873 some tricks doable by standard arithmetics. The use of conditional execution
3874 on chips where it is available is controlled by @code{if-conversion2}.
3876 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3878 @item -fif-conversion2
3879 @opindex if-conversion2
3880 Use conditional execution (where available) to transform conditional jumps into
3881 branch-less equivalents.
3883 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3885 @item -fdelete-null-pointer-checks
3886 @opindex fdelete-null-pointer-checks
3887 Use global dataflow analysis to identify and eliminate useless checks
3888 for null pointers. The compiler assumes that dereferencing a null
3889 pointer would have halted the program. If a pointer is checked after
3890 it has already been dereferenced, it cannot be null.
3892 In some environments, this assumption is not true, and programs can
3893 safely dereference null pointers. Use
3894 @option{-fno-delete-null-pointer-checks} to disable this optimization
3895 for programs which depend on that behavior.
3897 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3899 @item -fexpensive-optimizations
3900 @opindex fexpensive-optimizations
3901 Perform a number of minor optimizations that are relatively expensive.
3903 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3905 @item -foptimize-register-move
3907 @opindex foptimize-register-move
3909 Attempt to reassign register numbers in move instructions and as
3910 operands of other simple instructions in order to maximize the amount of
3911 register tying. This is especially helpful on machines with two-operand
3914 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3917 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3919 @item -fdelayed-branch
3920 @opindex fdelayed-branch
3921 If supported for the target machine, attempt to reorder instructions
3922 to exploit instruction slots available after delayed branch
3925 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3927 @item -fschedule-insns
3928 @opindex fschedule-insns
3929 If supported for the target machine, attempt to reorder instructions to
3930 eliminate execution stalls due to required data being unavailable. This
3931 helps machines that have slow floating point or memory load instructions
3932 by allowing other instructions to be issued until the result of the load
3933 or floating point instruction is required.
3935 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3937 @item -fschedule-insns2
3938 @opindex fschedule-insns2
3939 Similar to @option{-fschedule-insns}, but requests an additional pass of
3940 instruction scheduling after register allocation has been done. This is
3941 especially useful on machines with a relatively small number of
3942 registers and where memory load instructions take more than one cycle.
3944 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3946 @item -fno-sched-interblock
3947 @opindex fno-sched-interblock
3948 Don't schedule instructions across basic blocks. This is normally
3949 enabled by default when scheduling before register allocation, i.e.@:
3950 with @option{-fschedule-insns} or at @option{-O2} or higher.
3952 @item -fno-sched-spec
3953 @opindex fno-sched-spec
3954 Don't allow speculative motion of non-load instructions. This is normally
3955 enabled by default when scheduling before register allocation, i.e.@:
3956 with @option{-fschedule-insns} or at @option{-O2} or higher.
3958 @item -fsched-spec-load
3959 @opindex fsched-spec-load
3960 Allow speculative motion of some load instructions. This only makes
3961 sense when scheduling before register allocation, i.e.@: with
3962 @option{-fschedule-insns} or at @option{-O2} or higher.
3964 @item -fsched-spec-load-dangerous
3965 @opindex fsched-spec-load-dangerous
3966 Allow speculative motion of more load instructions. This only makes
3967 sense when scheduling before register allocation, i.e.@: with
3968 @option{-fschedule-insns} or at @option{-O2} or higher.
3970 @item -fsched2-use-superblocks
3971 @opindex fsched2-use-superblocks
3972 When scheduling after register allocation, do use superblock scheduling
3973 algorithm. Superblock scheduling allows motion across basic block boundaries
3974 resulting on faster schedules. This option is experimental, as not all machine
3975 descriptions used by GCC model the CPU closely enough to avoid unreliable
3976 results from the algorithm.
3978 This only makes sense when scheduling after register allocation, i.e.@: with
3979 @option{-fschedule-insns2} or at @option{-O2} or higher.
3981 @item -fsched2-use-traces
3982 @opindex fsched2-use-traces
3983 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
3984 allocation and additionally perform code duplication in order to increase the
3985 size of superblocks using tracer pass. See @option{-ftracer} for details on
3988 This mode should produce faster but significantly longer programs. Also
3989 without @code{-fbranch-probabilities} the traces constructed may not match the
3990 reality and hurt the performance. This only makes
3991 sense when scheduling after register allocation, i.e.@: with
3992 @option{-fschedule-insns2} or at @option{-O2} or higher.
3994 @item -fcaller-saves
3995 @opindex fcaller-saves
3996 Enable values to be allocated in registers that will be clobbered by
3997 function calls, by emitting extra instructions to save and restore the
3998 registers around such calls. Such allocation is done only when it
3999 seems to result in better code than would otherwise be produced.
4001 This option is always enabled by default on certain machines, usually
4002 those which have no call-preserved registers to use instead.
4004 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4006 @item -fmove-all-movables
4007 @opindex fmove-all-movables
4008 Forces all invariant computations in loops to be moved
4011 @item -freduce-all-givs
4012 @opindex freduce-all-givs
4013 Forces all general-induction variables in loops to be
4016 @emph{Note:} When compiling programs written in Fortran,
4017 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4018 by default when you use the optimizer.
4020 These options may generate better or worse code; results are highly
4021 dependent on the structure of loops within the source code.
4023 These two options are intended to be removed someday, once
4024 they have helped determine the efficacy of various
4025 approaches to improving loop optimizations.
4027 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
4028 know how use of these options affects
4029 the performance of your production code.
4030 We're very interested in code that runs @emph{slower}
4031 when these options are @emph{enabled}.
4034 @itemx -fno-peephole2
4035 @opindex fno-peephole
4036 @opindex fno-peephole2
4037 Disable any machine-specific peephole optimizations. The difference
4038 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4039 are implemented in the compiler; some targets use one, some use the
4040 other, a few use both.
4042 @option{-fpeephole} is enabled by default.
4043 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4045 @item -fno-guess-branch-probability
4046 @opindex fno-guess-branch-probability
4047 Do not guess branch probabilities using a randomized model.
4049 Sometimes gcc will opt to use a randomized model to guess branch
4050 probabilities, when none are available from either profiling feedback
4051 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4052 different runs of the compiler on the same program may produce different
4055 In a hard real-time system, people don't want different runs of the
4056 compiler to produce code that has different behavior; minimizing
4057 non-determinism is of paramount import. This switch allows users to
4058 reduce non-determinism, possibly at the expense of inferior
4061 The default is @option{-fguess-branch-probability} at levels
4062 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4064 @item -freorder-blocks
4065 @opindex freorder-blocks
4066 Reorder basic blocks in the compiled function in order to reduce number of
4067 taken branches and improve code locality.
4069 Enabled at levels @option{-O2}, @option{-O3}.
4071 @item -freorder-functions
4072 @opindex freorder-functions
4073 Reorder basic blocks in the compiled function in order to reduce number of
4074 taken branches and improve code locality. This is implemented by using special
4075 subsections @code{text.hot} for most frequently executed functions and
4076 @code{text.unlikely} for unlikely executed functions. Reordering is done by
4077 the linker so object file format must support named sections and linker must
4078 place them in a reasonable way.
4080 Also profile feedback must be available in to make this option effective. See
4081 @option{-fprofile-arcs} for details.
4083 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4085 @item -fstrict-aliasing
4086 @opindex fstrict-aliasing
4087 Allows the compiler to assume the strictest aliasing rules applicable to
4088 the language being compiled. For C (and C++), this activates
4089 optimizations based on the type of expressions. In particular, an
4090 object of one type is assumed never to reside at the same address as an
4091 object of a different type, unless the types are almost the same. For
4092 example, an @code{unsigned int} can alias an @code{int}, but not a
4093 @code{void*} or a @code{double}. A character type may alias any other
4096 Pay special attention to code like this:
4109 The practice of reading from a different union member than the one most
4110 recently written to (called ``type-punning'') is common. Even with
4111 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4112 is accessed through the union type. So, the code above will work as
4113 expected. However, this code might not:
4124 Every language that wishes to perform language-specific alias analysis
4125 should define a function that computes, given an @code{tree}
4126 node, an alias set for the node. Nodes in different alias sets are not
4127 allowed to alias. For an example, see the C front-end function
4128 @code{c_get_alias_set}.
4130 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4132 @item -falign-functions
4133 @itemx -falign-functions=@var{n}
4134 @opindex falign-functions
4135 Align the start of functions to the next power-of-two greater than
4136 @var{n}, skipping up to @var{n} bytes. For instance,
4137 @option{-falign-functions=32} aligns functions to the next 32-byte
4138 boundary, but @option{-falign-functions=24} would align to the next
4139 32-byte boundary only if this can be done by skipping 23 bytes or less.
4141 @option{-fno-align-functions} and @option{-falign-functions=1} are
4142 equivalent and mean that functions will not be aligned.
4144 Some assemblers only support this flag when @var{n} is a power of two;
4145 in that case, it is rounded up.
4147 If @var{n} is not specified or is zero, use a machine-dependent default.
4149 Enabled at levels @option{-O2}, @option{-O3}.
4151 @item -falign-labels
4152 @itemx -falign-labels=@var{n}
4153 @opindex falign-labels
4154 Align all branch targets to a power-of-two boundary, skipping up to
4155 @var{n} bytes like @option{-falign-functions}. This option can easily
4156 make code slower, because it must insert dummy operations for when the
4157 branch target is reached in the usual flow of the code.
4159 @option{-fno-align-labels} and @option{-falign-labels=1} are
4160 equivalent and mean that labels will not be aligned.
4162 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4163 are greater than this value, then their values are used instead.
4165 If @var{n} is not specified or is zero, use a machine-dependent default
4166 which is very likely to be @samp{1}, meaning no alignment.
4168 Enabled at levels @option{-O2}, @option{-O3}.
4171 @itemx -falign-loops=@var{n}
4172 @opindex falign-loops
4173 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4174 like @option{-falign-functions}. The hope is that the loop will be
4175 executed many times, which will make up for any execution of the dummy
4178 @option{-fno-align-loops} and @option{-falign-loops=1} are
4179 equivalent and mean that loops will not be aligned.
4181 If @var{n} is not specified or is zero, use a machine-dependent default.
4183 Enabled at levels @option{-O2}, @option{-O3}.
4186 @itemx -falign-jumps=@var{n}
4187 @opindex falign-jumps
4188 Align branch targets to a power-of-two boundary, for branch targets
4189 where the targets can only be reached by jumping, skipping up to @var{n}
4190 bytes like @option{-falign-functions}. In this case, no dummy operations
4193 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4194 equivalent and mean that loops will not be aligned.
4196 If @var{n} is not specified or is zero, use a machine-dependent default.
4198 Enabled at levels @option{-O2}, @option{-O3}.
4200 @item -frename-registers
4201 @opindex frename-registers
4202 Attempt to avoid false dependencies in scheduled code by making use
4203 of registers left over after register allocation. This optimization
4204 will most benefit processors with lots of registers. It can, however,
4205 make debugging impossible, since variables will no longer stay in
4206 a ``home register''.
4208 Enabled at levels @option{-O3}.
4210 @item -fno-cprop-registers
4211 @opindex fno-cprop-registers
4212 After register allocation and post-register allocation instruction splitting,
4213 we perform a copy-propagation pass to try to reduce scheduling dependencies
4214 and occasionally eliminate the copy.
4216 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4220 The following options control compiler behavior regarding floating
4221 point arithmetic. These options trade off between speed and
4222 correctness. All must be specifically enabled.
4226 @opindex ffloat-store
4227 Do not store floating point variables in registers, and inhibit other
4228 options that might change whether a floating point value is taken from a
4231 @cindex floating point precision
4232 This option prevents undesirable excess precision on machines such as
4233 the 68000 where the floating registers (of the 68881) keep more
4234 precision than a @code{double} is supposed to have. Similarly for the
4235 x86 architecture. For most programs, the excess precision does only
4236 good, but a few programs rely on the precise definition of IEEE floating
4237 point. Use @option{-ffloat-store} for such programs, after modifying
4238 them to store all pertinent intermediate computations into variables.
4242 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4243 @option{-fno-trapping-math}, @option{-ffinite-math-only} and @*
4244 @option{-fno-signaling-nans}.
4246 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4248 This option should never be turned on by any @option{-O} option since
4249 it can result in incorrect output for programs which depend on
4250 an exact implementation of IEEE or ISO rules/specifications for
4253 @item -fno-math-errno
4254 @opindex fno-math-errno
4255 Do not set ERRNO after calling math functions that are executed
4256 with a single instruction, e.g., sqrt. A program that relies on
4257 IEEE exceptions for math error handling may want to use this flag
4258 for speed while maintaining IEEE arithmetic compatibility.
4260 This option should never be turned on by any @option{-O} option since
4261 it can result in incorrect output for programs which depend on
4262 an exact implementation of IEEE or ISO rules/specifications for
4265 The default is @option{-fmath-errno}.
4267 @item -funsafe-math-optimizations
4268 @opindex funsafe-math-optimizations
4269 Allow optimizations for floating-point arithmetic that (a) assume
4270 that arguments and results are valid and (b) may violate IEEE or
4271 ANSI standards. When used at link-time, it may include libraries
4272 or startup files that change the default FPU control word or other
4273 similar optimizations.
4275 This option should never be turned on by any @option{-O} option since
4276 it can result in incorrect output for programs which depend on
4277 an exact implementation of IEEE or ISO rules/specifications for
4280 The default is @option{-fno-unsafe-math-optimizations}.
4282 @item -ffinite-math-only
4283 @opindex ffinite-math-only
4284 Allow optimizations for floating-point arithmetic that assume
4285 that arguments and results are not NaNs or +-Infs.
4287 This option should never be turned on by any @option{-O} option since
4288 it can result in incorrect output for programs which depend on
4289 an exact implementation of IEEE or ISO rules/specifications.
4291 The default is @option{-fno-finite-math-only}.
4293 @item -fno-trapping-math
4294 @opindex fno-trapping-math
4295 Compile code assuming that floating-point operations cannot generate
4296 user-visible traps. These traps include division by zero, overflow,
4297 underflow, inexact result and invalid operation. This option implies
4298 @option{-fno-signaling-nans}. Setting this option may allow faster
4299 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4301 This option should never be turned on by any @option{-O} option since
4302 it can result in incorrect output for programs which depend on
4303 an exact implementation of IEEE or ISO rules/specifications for
4306 The default is @option{-ftrapping-math}.
4308 @item -fsignaling-nans
4309 @opindex fsignaling-nans
4310 Compile code assuming that IEEE signaling NaNs may generate user-visible
4311 traps during floating-point operations. Setting this option disables
4312 optimizations that may change the number of exceptions visible with
4313 signaling NaNs. This option implies @option{-ftrapping-math}.
4315 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4318 The default is @option{-fno-signaling-nans}.
4320 This option is experimental and does not currently guarantee to
4321 disable all GCC optimizations that affect signaling NaN behavior.
4323 @item -fsingle-precision-constant
4324 @opindex fsingle-precision-constant
4325 Treat floating point constant as single precision constant instead of
4326 implicitly converting it to double precision constant.
4331 The following options control optimizations that may improve
4332 performance, but are not enabled by any @option{-O} options. This
4333 section includes experimental options that may produce broken code.
4336 @item -fbranch-probabilities
4337 @opindex fbranch-probabilities
4338 After running a program compiled with @option{-fprofile-arcs}
4339 (@pxref{Debugging Options,, Options for Debugging Your Program or
4340 @command{gcc}}), you can compile it a second time using
4341 @option{-fbranch-probabilities}, to improve optimizations based on
4342 the number of times each branch was taken. When the program
4343 compiled with @option{-fprofile-arcs} exits it saves arc execution
4344 counts to a file called @file{@var{sourcename}.da} for each source
4345 file The information in this data file is very dependent on the
4346 structure of the generated code, so you must use the same source code
4347 and the same optimization options for both compilations.
4349 With @option{-fbranch-probabilities}, GCC puts a
4350 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4351 These can be used to improve optimization. Currently, they are only
4352 used in one place: in @file{reorg.c}, instead of guessing which path a
4353 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4354 exactly determine which path is taken more often.
4356 @item -fprofile-values
4357 @opindex fprofile-values
4358 If combined with @option{-fprofile-arcs}, it adds code so that some
4359 data about values of expressions in the program is gathered.
4363 Use a graph coloring register allocator. Currently this option is meant
4364 for testing, so we are interested to hear about miscompilations with
4369 Perform tail duplication to enlarge superblock size. This transformation
4370 simplifies the control flow of the function allowing other optimizations to do
4373 @item -funit-at-a-time
4374 @opindex funit-at-a-time
4375 Parse the whole compilation unit before starting to produce code.
4376 This allows some extra optimizations to take place but consumes more
4379 @item -funroll-loops
4380 @opindex funroll-loops
4381 Unroll loops whose number of iterations can be determined at compile time or
4382 upon entry to the loop. @option{-funroll-loops} implies
4383 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4384 (i.e. complete removal of loops with small constant number of iterations).
4385 This option makes code larger, and may or may not make it run faster.
4387 @item -funroll-all-loops
4388 @opindex funroll-all-loops
4389 Unroll all loops, even if their number of iterations is uncertain when
4390 the loop is entered. This usually makes programs run more slowly.
4391 @option{-funroll-all-loops} implies the same options as
4392 @option{-funroll-loops}.
4395 @opindex fpeel-loops
4396 Peels the loops for that there is enough information that they do not
4397 roll much (from profile feedback). It also turns on complete loop peeling
4398 (i.e. complete removal of loops with small constant number of iterations).
4400 @item -funswitch-loops
4401 @opindex funswitch-loops
4402 Move branches with loop invariant conditions out of the loop, with duplicates
4403 of the loop on both branches (modified according to result of the condition).
4405 @item -fold-unroll-loops
4406 @opindex fold-unroll-loops
4407 Unroll loops whose number of iterations can be determined at compile
4408 time or upon entry to the loop, using the old loop unroller whose loop
4409 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4410 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4411 option makes code larger, and may or may not make it run faster.
4413 @item -fold-unroll-all-loops
4414 @opindex fold-unroll-all-loops
4415 Unroll all loops, even if their number of iterations is uncertain when
4416 the loop is entered. This is done using the old loop unroller whose loop
4417 recognition is based on notes from frontend. This usually makes programs run more slowly.
4418 @option{-fold-unroll-all-loops} implies the same options as
4419 @option{-fold-unroll-loops}.
4421 @item -funswitch-loops
4422 @opindex funswitch-loops
4423 Move branches with loop invariant conditions out of the loop, with duplicates
4424 of the loop on both branches (modified according to result of the condition).
4426 @item -funswitch-loops
4427 @opindex funswitch-loops
4428 Move branches with loop invariant conditions out of the loop, with duplicates
4429 of the loop on both branches (modified according to result of the condition).
4431 @item -fprefetch-loop-arrays
4432 @opindex fprefetch-loop-arrays
4433 If supported by the target machine, generate instructions to prefetch
4434 memory to improve the performance of loops that access large arrays.
4436 Disabled at level @option{-Os}.
4438 @item -ffunction-sections
4439 @itemx -fdata-sections
4440 @opindex ffunction-sections
4441 @opindex fdata-sections
4442 Place each function or data item into its own section in the output
4443 file if the target supports arbitrary sections. The name of the
4444 function or the name of the data item determines the section's name
4447 Use these options on systems where the linker can perform optimizations
4448 to improve locality of reference in the instruction space. Most systems
4449 using the ELF object format and SPARC processors running Solaris 2 have
4450 linkers with such optimizations. AIX may have these optimizations in
4453 Only use these options when there are significant benefits from doing
4454 so. When you specify these options, the assembler and linker will
4455 create larger object and executable files and will also be slower.
4456 You will not be able to use @code{gprof} on all systems if you
4457 specify this option and you may have problems with debugging if
4458 you specify both this option and @option{-g}.
4462 Perform optimizations in static single assignment form. Each function's
4463 flow graph is translated into SSA form, optimizations are performed, and
4464 the flow graph is translated back from SSA form. Users should not
4465 specify this option, since it is not yet ready for production use.
4469 Perform Sparse Conditional Constant Propagation in SSA form. Requires
4470 @option{-fssa}. Like @option{-fssa}, this is an experimental feature.
4474 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
4475 Like @option{-fssa}, this is an experimental feature.
4477 @item -fbranch-target-load-optimize
4478 @opindex fbranch-target-load-optimize
4479 Perform branch target register load optimization before prologue / epilogue
4481 The use of target registers can typically be exposed only during reload,
4482 thus hoisting loads out of loops and doing inter-block scheduling needs
4483 a separate optimization pass.
4485 @item -fbranch-target-load-optimize2
4486 @opindex fbranch-target-load-optimize2
4487 Perform branch target register load optimization after prologue / epilogue
4493 @item --param @var{name}=@var{value}
4495 In some places, GCC uses various constants to control the amount of
4496 optimization that is done. For example, GCC will not inline functions
4497 that contain more that a certain number of instructions. You can
4498 control some of these constants on the command-line using the
4499 @option{--param} option.
4501 In each case, the @var{value} is an integer. The allowable choices for
4502 @var{name} are given in the following table:
4505 @item max-crossjump-edges
4506 The maximum number of incoming edges to consider for crossjumping.
4507 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
4508 the number of edges incoming to each block. Increasing values mean
4509 more aggressive optimization, making the compile time increase with
4510 probably small improvement in executable size.
4512 @item max-delay-slot-insn-search
4513 The maximum number of instructions to consider when looking for an
4514 instruction to fill a delay slot. If more than this arbitrary number of
4515 instructions is searched, the time savings from filling the delay slot
4516 will be minimal so stop searching. Increasing values mean more
4517 aggressive optimization, making the compile time increase with probably
4518 small improvement in executable run time.
4520 @item max-delay-slot-live-search
4521 When trying to fill delay slots, the maximum number of instructions to
4522 consider when searching for a block with valid live register
4523 information. Increasing this arbitrarily chosen value means more
4524 aggressive optimization, increasing the compile time. This parameter
4525 should be removed when the delay slot code is rewritten to maintain the
4528 @item max-gcse-memory
4529 The approximate maximum amount of memory that will be allocated in
4530 order to perform the global common subexpression elimination
4531 optimization. If more memory than specified is required, the
4532 optimization will not be done.
4534 @item max-gcse-passes
4535 The maximum number of passes of GCSE to run.
4537 @item max-pending-list-length
4538 The maximum number of pending dependencies scheduling will allow
4539 before flushing the current state and starting over. Large functions
4540 with few branches or calls can create excessively large lists which
4541 needlessly consume memory and resources.
4543 @item max-inline-insns-single
4544 Several parameters control the tree inliner used in gcc.
4545 This number sets the maximum number of instructions (counted in gcc's
4546 internal representation) in a single function that the tree inliner
4547 will consider for inlining. This only affects functions declared
4548 inline and methods implemented in a class declaration (C++).
4549 The default value is 100.
4551 @item max-inline-insns-auto
4552 When you use @option{-finline-functions} (included in @option{-O3}),
4553 a lot of functions that would otherwise not be considered for inlining
4554 by the compiler will be investigated. To those functions, a different
4555 (more restrictive) limit compared to functions declared inline can
4557 The default value is 100.
4559 @item max-inline-insns
4560 The tree inliner does decrease the allowable size for single functions
4561 to be inlined after we already inlined the number of instructions
4562 given here by repeated inlining. This number should be a factor of
4563 two or more larger than the single function limit.
4564 Higher numbers result in better runtime performance, but incur higher
4565 compile-time resource (CPU time, memory) requirements and result in
4566 larger binaries. Very high values are not advisable, as too large
4567 binaries may adversely affect runtime performance.
4568 The default value is 200.
4570 @item max-inline-slope
4571 After exceeding the maximum number of inlined instructions by repeated
4572 inlining, a linear function is used to decrease the allowable size
4573 for single functions. The slope of that function is the negative
4574 reciprocal of the number specified here.
4575 This parameter is ignored when @option{-funit-at-a-time} is used.
4576 The default value is 32.
4578 @item min-inline-insns
4579 The repeated inlining is throttled more and more by the linear function
4580 after exceeding the limit. To avoid too much throttling, a minimum for
4581 this function is specified here to allow repeated inlining for very small
4582 functions even when a lot of repeated inlining already has been done.
4583 This parameter is ignored when @option{-funit-at-a-time} is used.
4584 The default value is 10.
4586 @item large-function-insns
4587 The limit specifying really large functions. For functions greater than this
4588 limit inlining is constrained by @option{--param large-function-growth}.
4589 This parameter is usefull primarily to avoid extreme compilation time caused by non-linear
4590 algorithms used by the backend.
4591 This parameter is ignored when @option{-funit-at-a-time} is not used.
4592 The default value is 30000.
4594 @item large-function-growth
4595 Specifies maximal growth of large functtion caused by inlining in percents.
4596 This parameter is ignored when @option{-funit-at-a-time} is not used.
4597 The default value is 200.
4599 @item inline-unit-growth
4600 Specifies maximal overall growth of the compilation unit caused by inlining.
4601 This parameter is ignored when @option{-funit-at-a-time} is not used.
4602 The default value is 150.
4604 @item max-inline-insns-rtl
4605 For languages that use the RTL inliner (this happens at a later stage
4606 than tree inlining), you can set the maximum allowable size (counted
4607 in RTL instructions) for the RTL inliner with this parameter.
4608 The default value is 600.
4611 @item max-unrolled-insns
4612 The maximum number of instructions that a loop should have if that loop
4613 is unrolled, and if the loop is unrolled, it determines how many times
4614 the loop code is unrolled.
4616 @item max-average-unrolled-insns
4617 The maximum number of instructions biased by probabilities of their execution
4618 that a loop should have if that loop is unrolled, and if the loop is unrolled,
4619 it determines how many times the loop code is unrolled.
4621 @item max-unroll-times
4622 The maximum number of unrollings of a single loop.
4624 @item max-peeled-insns
4625 The maximum number of instructions that a loop should have if that loop
4626 is peeled, and if the loop is peeled, it determines how many times
4627 the loop code is peeled.
4629 @item max-peel-times
4630 The maximum number of peelings of a single loop.
4632 @item max-completely-peeled-insns
4633 The maximum number of insns of a completely peeled loop.
4635 @item max-completely-peel-times
4636 The maximum number of iterations of a loop to be suitable for complete peeling.
4638 @item max-unswitch-insns
4639 The maximum number of insns of an unswitched loop.
4641 @item max-unswitch-level
4642 The maximum number of branches unswitched in a single loop.
4644 @item hot-bb-count-fraction
4645 Select fraction of the maximal count of repetitions of basic block in program
4646 given basic block needs to have to be considered hot.
4648 @item hot-bb-frequency-fraction
4649 Select fraction of the maximal frequency of executions of basic block in
4650 function given basic block needs to have to be considered hot
4652 @item tracer-dynamic-coverage
4653 @itemx tracer-dynamic-coverage-feedback
4655 This value is used to limit superblock formation once the given percentage of
4656 executed instructions is covered. This limits unnecessary code size
4659 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4660 feedback is available. The real profiles (as opposed to statically estimated
4661 ones) are much less balanced allowing the threshold to be larger value.
4663 @item tracer-max-code-growth
4664 Stop tail duplication once code growth has reached given percentage. This is
4665 rather hokey argument, as most of the duplicates will be eliminated later in
4666 cross jumping, so it may be set to much higher values than is the desired code
4669 @item tracer-min-branch-ratio
4671 Stop reverse growth when the reverse probability of best edge is less than this
4672 threshold (in percent).
4674 @item tracer-min-branch-ratio
4675 @itemx tracer-min-branch-ratio-feedback
4677 Stop forward growth if the best edge do have probability lower than this
4680 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
4681 compilation for profile feedback and one for compilation without. The value
4682 for compilation with profile feedback needs to be more conservative (higher) in
4683 order to make tracer effective.
4685 @item max-cse-path-length
4687 Maximum number of basic blocks on path that cse considers.
4689 @item ggc-min-expand
4691 GCC uses a garbage collector to manage its own memory allocation. This
4692 parameter specifies the minimum percentage by which the garbage
4693 collector's heap should be allowed to expand between collections.
4694 Tuning this may improve compilation speed; it has no effect on code
4697 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
4698 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
4699 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
4700 GCC is not able to calculate RAM on a particular platform, the lower
4701 bound of 30% is used. Setting this parameter and
4702 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4703 every opportunity. This is extremely slow, but can be useful for
4706 @item ggc-min-heapsize
4708 Minimum size of the garbage collector's heap before it begins bothering
4709 to collect garbage. The first collection occurs after the heap expands
4710 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4711 tuning this may improve compilation speed, and has no effect on code
4714 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
4715 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
4716 available, the notion of "RAM" is the smallest of actual RAM,
4717 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
4718 RAM on a particular platform, the lower bound is used. Setting this
4719 parameter very large effectively disables garbage collection. Setting
4720 this parameter and @option{ggc-min-expand} to zero causes a full
4721 collection to occur at every opportunity.
4723 @item reorder-blocks-duplicate
4724 @itemx reorder-blocks-duplicate-feedback
4726 Used by basic block reordering pass to decide whether to use unconditional
4727 branch or duplicate the code on it's destination. Code is duplicated when it's
4728 estimated size is smaller than this value multiplied by the estimated size of
4729 unconditional jump in the hot spots of the program.
4731 The @option{reorder-block-duplicate-feedback} is used only when profile
4732 feedback is available and may be set to higher values than
4733 @option{reorder-block-duplicate} since information about the hot spots is more
4738 @node Preprocessor Options
4739 @section Options Controlling the Preprocessor
4740 @cindex preprocessor options
4741 @cindex options, preprocessor
4743 These options control the C preprocessor, which is run on each C source
4744 file before actual compilation.
4746 If you use the @option{-E} option, nothing is done except preprocessing.
4747 Some of these options make sense only together with @option{-E} because
4748 they cause the preprocessor output to be unsuitable for actual
4753 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4754 and pass @var{option} directly through to the preprocessor. If
4755 @var{option} contains commas, it is split into multiple options at the
4756 commas. However, many options are modified, translated or interpreted
4757 by the compiler driver before being passed to the preprocessor, and
4758 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4759 interface is undocumented and subject to change, so whenever possible
4760 you should avoid using @option{-Wp} and let the driver handle the
4763 @item -Xpreprocessor @var{option}
4764 @opindex preprocessor
4765 Pass @var{option} as an option to the preprocessor. You can use this to
4766 supply system-specific preprocessor options which GCC does not know how to
4769 If you want to pass an option that takes an argument, you must use
4770 @option{-Xpreprocessor} twice, once for the option and once for the argument.
4773 @include cppopts.texi
4775 @node Assembler Options
4776 @section Passing Options to the Assembler
4778 @c prevent bad page break with this line
4779 You can pass options to the assembler.
4782 @item -Wa,@var{option}
4784 Pass @var{option} as an option to the assembler. If @var{option}
4785 contains commas, it is split into multiple options at the commas.
4787 @item -Xassembler @var{option}
4789 Pass @var{option} as an option to the assembler. You can use this to
4790 supply system-specific assembler options which GCC does not know how to
4793 If you want to pass an option that takes an argument, you must use
4794 @option{-Xassembler} twice, once for the option and once for the argument.
4799 @section Options for Linking
4800 @cindex link options
4801 @cindex options, linking
4803 These options come into play when the compiler links object files into
4804 an executable output file. They are meaningless if the compiler is
4805 not doing a link step.
4809 @item @var{object-file-name}
4810 A file name that does not end in a special recognized suffix is
4811 considered to name an object file or library. (Object files are
4812 distinguished from libraries by the linker according to the file
4813 contents.) If linking is done, these object files are used as input
4822 If any of these options is used, then the linker is not run, and
4823 object file names should not be used as arguments. @xref{Overall
4827 @item -l@var{library}
4828 @itemx -l @var{library}
4830 Search the library named @var{library} when linking. (The second
4831 alternative with the library as a separate argument is only for
4832 POSIX compliance and is not recommended.)
4834 It makes a difference where in the command you write this option; the
4835 linker searches and processes libraries and object files in the order they
4836 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4837 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4838 to functions in @samp{z}, those functions may not be loaded.
4840 The linker searches a standard list of directories for the library,
4841 which is actually a file named @file{lib@var{library}.a}. The linker
4842 then uses this file as if it had been specified precisely by name.
4844 The directories searched include several standard system directories
4845 plus any that you specify with @option{-L}.
4847 Normally the files found this way are library files---archive files
4848 whose members are object files. The linker handles an archive file by
4849 scanning through it for members which define symbols that have so far
4850 been referenced but not defined. But if the file that is found is an
4851 ordinary object file, it is linked in the usual fashion. The only
4852 difference between using an @option{-l} option and specifying a file name
4853 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4854 and searches several directories.
4858 You need this special case of the @option{-l} option in order to
4859 link an Objective-C program.
4862 @opindex nostartfiles
4863 Do not use the standard system startup files when linking.
4864 The standard system libraries are used normally, unless @option{-nostdlib}
4865 or @option{-nodefaultlibs} is used.
4867 @item -nodefaultlibs
4868 @opindex nodefaultlibs
4869 Do not use the standard system libraries when linking.
4870 Only the libraries you specify will be passed to the linker.
4871 The standard startup files are used normally, unless @option{-nostartfiles}
4872 is used. The compiler may generate calls to memcmp, memset, and memcpy
4873 for System V (and ISO C) environments or to bcopy and bzero for
4874 BSD environments. These entries are usually resolved by entries in
4875 libc. These entry points should be supplied through some other
4876 mechanism when this option is specified.
4880 Do not use the standard system startup files or libraries when linking.
4881 No startup files and only the libraries you specify will be passed to
4882 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4883 for System V (and ISO C) environments or to bcopy and bzero for
4884 BSD environments. These entries are usually resolved by entries in
4885 libc. These entry points should be supplied through some other
4886 mechanism when this option is specified.
4888 @cindex @option{-lgcc}, use with @option{-nostdlib}
4889 @cindex @option{-nostdlib} and unresolved references
4890 @cindex unresolved references and @option{-nostdlib}
4891 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
4892 @cindex @option{-nodefaultlibs} and unresolved references
4893 @cindex unresolved references and @option{-nodefaultlibs}
4894 One of the standard libraries bypassed by @option{-nostdlib} and
4895 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4896 that GCC uses to overcome shortcomings of particular machines, or special
4897 needs for some languages.
4898 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
4899 Collection (GCC) Internals},
4900 for more discussion of @file{libgcc.a}.)
4901 In most cases, you need @file{libgcc.a} even when you want to avoid
4902 other standard libraries. In other words, when you specify @option{-nostdlib}
4903 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
4904 This ensures that you have no unresolved references to internal GCC
4905 library subroutines. (For example, @samp{__main}, used to ensure C++
4906 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
4907 GNU Compiler Collection (GCC) Internals}.)
4911 Produce a position independent executable on targets which support it.
4912 For predictable results, you must also specify the same set of options
4913 that were used to generate code (@option{-fpie}, @option{-fPIE},
4914 or model suboptions) when you specify this option.
4918 Remove all symbol table and relocation information from the executable.
4922 On systems that support dynamic linking, this prevents linking with the shared
4923 libraries. On other systems, this option has no effect.
4927 Produce a shared object which can then be linked with other objects to
4928 form an executable. Not all systems support this option. For predictable
4929 results, you must also specify the same set of options that were used to
4930 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
4931 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4932 needs to build supplementary stub code for constructors to work. On
4933 multi-libbed systems, @samp{gcc -shared} must select the correct support
4934 libraries to link against. Failing to supply the correct flags may lead
4935 to subtle defects. Supplying them in cases where they are not necessary
4938 @item -shared-libgcc
4939 @itemx -static-libgcc
4940 @opindex shared-libgcc
4941 @opindex static-libgcc
4942 On systems that provide @file{libgcc} as a shared library, these options
4943 force the use of either the shared or static version respectively.
4944 If no shared version of @file{libgcc} was built when the compiler was
4945 configured, these options have no effect.
4947 There are several situations in which an application should use the
4948 shared @file{libgcc} instead of the static version. The most common
4949 of these is when the application wishes to throw and catch exceptions
4950 across different shared libraries. In that case, each of the libraries
4951 as well as the application itself should use the shared @file{libgcc}.
4953 Therefore, the G++ and GCJ drivers automatically add
4954 @option{-shared-libgcc} whenever you build a shared library or a main
4955 executable, because C++ and Java programs typically use exceptions, so
4956 this is the right thing to do.
4958 If, instead, you use the GCC driver to create shared libraries, you may
4959 find that they will not always be linked with the shared @file{libgcc}.
4960 If GCC finds, at its configuration time, that you have a GNU linker that
4961 does not support option @option{--eh-frame-hdr}, it will link the shared
4962 version of @file{libgcc} into shared libraries by default. Otherwise,
4963 it will take advantage of the linker and optimize away the linking with
4964 the shared version of @file{libgcc}, linking with the static version of
4965 libgcc by default. This allows exceptions to propagate through such
4966 shared libraries, without incurring relocation costs at library load
4969 However, if a library or main executable is supposed to throw or catch
4970 exceptions, you must link it using the G++ or GCJ driver, as appropriate
4971 for the languages used in the program, or using the option
4972 @option{-shared-libgcc}, such that it is linked with the shared
4977 Bind references to global symbols when building a shared object. Warn
4978 about any unresolved references (unless overridden by the link editor
4979 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
4982 @item -Xlinker @var{option}
4984 Pass @var{option} as an option to the linker. You can use this to
4985 supply system-specific linker options which GCC does not know how to
4988 If you want to pass an option that takes an argument, you must use
4989 @option{-Xlinker} twice, once for the option and once for the argument.
4990 For example, to pass @option{-assert definitions}, you must write
4991 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
4992 @option{-Xlinker "-assert definitions"}, because this passes the entire
4993 string as a single argument, which is not what the linker expects.
4995 @item -Wl,@var{option}
4997 Pass @var{option} as an option to the linker. If @var{option} contains
4998 commas, it is split into multiple options at the commas.
5000 @item -u @var{symbol}
5002 Pretend the symbol @var{symbol} is undefined, to force linking of
5003 library modules to define it. You can use @option{-u} multiple times with
5004 different symbols to force loading of additional library modules.
5007 @node Directory Options
5008 @section Options for Directory Search
5009 @cindex directory options
5010 @cindex options, directory search
5013 These options specify directories to search for header files, for
5014 libraries and for parts of the compiler:
5019 Add the directory @var{dir} to the head of the list of directories to be
5020 searched for header files. This can be used to override a system header
5021 file, substituting your own version, since these directories are
5022 searched before the system header file directories. However, you should
5023 not use this option to add directories that contain vendor-supplied
5024 system header files (use @option{-isystem} for that). If you use more than
5025 one @option{-I} option, the directories are scanned in left-to-right
5026 order; the standard system directories come after.
5028 If a standard system include directory, or a directory specified with
5029 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5030 option will be ignored. The directory will still be searched but as a
5031 system directory at its normal position in the system include chain.
5032 This is to ensure that GCC's procedure to fix buggy system headers and
5033 the ordering for the include_next directive are not inadvertently changed.
5034 If you really need to change the search order for system directories,
5035 use the @option{-nostdinc} and/or @option{-isystem} options.
5039 Any directories you specify with @option{-I} options before the @option{-I-}
5040 option are searched only for the case of @samp{#include "@var{file}"};
5041 they are not searched for @samp{#include <@var{file}>}.
5043 If additional directories are specified with @option{-I} options after
5044 the @option{-I-}, these directories are searched for all @samp{#include}
5045 directives. (Ordinarily @emph{all} @option{-I} directories are used
5048 In addition, the @option{-I-} option inhibits the use of the current
5049 directory (where the current input file came from) as the first search
5050 directory for @samp{#include "@var{file}"}. There is no way to
5051 override this effect of @option{-I-}. With @option{-I.} you can specify
5052 searching the directory which was current when the compiler was
5053 invoked. That is not exactly the same as what the preprocessor does
5054 by default, but it is often satisfactory.
5056 @option{-I-} does not inhibit the use of the standard system directories
5057 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5062 Add directory @var{dir} to the list of directories to be searched
5065 @item -B@var{prefix}
5067 This option specifies where to find the executables, libraries,
5068 include files, and data files of the compiler itself.
5070 The compiler driver program runs one or more of the subprograms
5071 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5072 @var{prefix} as a prefix for each program it tries to run, both with and
5073 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5075 For each subprogram to be run, the compiler driver first tries the
5076 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5077 was not specified, the driver tries two standard prefixes, which are
5078 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
5079 those results in a file name that is found, the unmodified program
5080 name is searched for using the directories specified in your
5081 @env{PATH} environment variable.
5083 The compiler will check to see if the path provided by the @option{-B}
5084 refers to a directory, and if necessary it will add a directory
5085 separator character at the end of the path.
5087 @option{-B} prefixes that effectively specify directory names also apply
5088 to libraries in the linker, because the compiler translates these
5089 options into @option{-L} options for the linker. They also apply to
5090 includes files in the preprocessor, because the compiler translates these
5091 options into @option{-isystem} options for the preprocessor. In this case,
5092 the compiler appends @samp{include} to the prefix.
5094 The run-time support file @file{libgcc.a} can also be searched for using
5095 the @option{-B} prefix, if needed. If it is not found there, the two
5096 standard prefixes above are tried, and that is all. The file is left
5097 out of the link if it is not found by those means.
5099 Another way to specify a prefix much like the @option{-B} prefix is to use
5100 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5103 As a special kludge, if the path provided by @option{-B} is
5104 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5105 9, then it will be replaced by @file{[dir/]include}. This is to help
5106 with boot-strapping the compiler.
5108 @item -specs=@var{file}
5110 Process @var{file} after the compiler reads in the standard @file{specs}
5111 file, in order to override the defaults that the @file{gcc} driver
5112 program uses when determining what switches to pass to @file{cc1},
5113 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5114 @option{-specs=@var{file}} can be specified on the command line, and they
5115 are processed in order, from left to right.
5121 @section Specifying subprocesses and the switches to pass to them
5124 @command{gcc} is a driver program. It performs its job by invoking a
5125 sequence of other programs to do the work of compiling, assembling and
5126 linking. GCC interprets its command-line parameters and uses these to
5127 deduce which programs it should invoke, and which command-line options
5128 it ought to place on their command lines. This behavior is controlled
5129 by @dfn{spec strings}. In most cases there is one spec string for each
5130 program that GCC can invoke, but a few programs have multiple spec
5131 strings to control their behavior. The spec strings built into GCC can
5132 be overridden by using the @option{-specs=} command-line switch to specify
5135 @dfn{Spec files} are plaintext files that are used to construct spec
5136 strings. They consist of a sequence of directives separated by blank
5137 lines. The type of directive is determined by the first non-whitespace
5138 character on the line and it can be one of the following:
5141 @item %@var{command}
5142 Issues a @var{command} to the spec file processor. The commands that can
5146 @item %include <@var{file}>
5148 Search for @var{file} and insert its text at the current point in the
5151 @item %include_noerr <@var{file}>
5152 @cindex %include_noerr
5153 Just like @samp{%include}, but do not generate an error message if the include
5154 file cannot be found.
5156 @item %rename @var{old_name} @var{new_name}
5158 Rename the spec string @var{old_name} to @var{new_name}.
5162 @item *[@var{spec_name}]:
5163 This tells the compiler to create, override or delete the named spec
5164 string. All lines after this directive up to the next directive or
5165 blank line are considered to be the text for the spec string. If this
5166 results in an empty string then the spec will be deleted. (Or, if the
5167 spec did not exist, then nothing will happened.) Otherwise, if the spec
5168 does not currently exist a new spec will be created. If the spec does
5169 exist then its contents will be overridden by the text of this
5170 directive, unless the first character of that text is the @samp{+}
5171 character, in which case the text will be appended to the spec.
5173 @item [@var{suffix}]:
5174 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5175 and up to the next directive or blank line are considered to make up the
5176 spec string for the indicated suffix. When the compiler encounters an
5177 input file with the named suffix, it will processes the spec string in
5178 order to work out how to compile that file. For example:
5185 This says that any input file whose name ends in @samp{.ZZ} should be
5186 passed to the program @samp{z-compile}, which should be invoked with the
5187 command-line switch @option{-input} and with the result of performing the
5188 @samp{%i} substitution. (See below.)
5190 As an alternative to providing a spec string, the text that follows a
5191 suffix directive can be one of the following:
5194 @item @@@var{language}
5195 This says that the suffix is an alias for a known @var{language}. This is
5196 similar to using the @option{-x} command-line switch to GCC to specify a
5197 language explicitly. For example:
5204 Says that .ZZ files are, in fact, C++ source files.
5207 This causes an error messages saying:
5210 @var{name} compiler not installed on this system.
5214 GCC already has an extensive list of suffixes built into it.
5215 This directive will add an entry to the end of the list of suffixes, but
5216 since the list is searched from the end backwards, it is effectively
5217 possible to override earlier entries using this technique.
5221 GCC has the following spec strings built into it. Spec files can
5222 override these strings or create their own. Note that individual
5223 targets can also add their own spec strings to this list.
5226 asm Options to pass to the assembler
5227 asm_final Options to pass to the assembler post-processor
5228 cpp Options to pass to the C preprocessor
5229 cc1 Options to pass to the C compiler
5230 cc1plus Options to pass to the C++ compiler
5231 endfile Object files to include at the end of the link
5232 link Options to pass to the linker
5233 lib Libraries to include on the command line to the linker
5234 libgcc Decides which GCC support library to pass to the linker
5235 linker Sets the name of the linker
5236 predefines Defines to be passed to the C preprocessor
5237 signed_char Defines to pass to CPP to say whether @code{char} is signed
5239 startfile Object files to include at the start of the link
5242 Here is a small example of a spec file:
5248 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5251 This example renames the spec called @samp{lib} to @samp{old_lib} and
5252 then overrides the previous definition of @samp{lib} with a new one.
5253 The new definition adds in some extra command-line options before
5254 including the text of the old definition.
5256 @dfn{Spec strings} are a list of command-line options to be passed to their
5257 corresponding program. In addition, the spec strings can contain
5258 @samp{%}-prefixed sequences to substitute variable text or to
5259 conditionally insert text into the command line. Using these constructs
5260 it is possible to generate quite complex command lines.
5262 Here is a table of all defined @samp{%}-sequences for spec
5263 strings. Note that spaces are not generated automatically around the
5264 results of expanding these sequences. Therefore you can concatenate them
5265 together or combine them with constant text in a single argument.
5269 Substitute one @samp{%} into the program name or argument.
5272 Substitute the name of the input file being processed.
5275 Substitute the basename of the input file being processed.
5276 This is the substring up to (and not including) the last period
5277 and not including the directory.
5280 This is the same as @samp{%b}, but include the file suffix (text after
5284 Marks the argument containing or following the @samp{%d} as a
5285 temporary file name, so that that file will be deleted if GCC exits
5286 successfully. Unlike @samp{%g}, this contributes no text to the
5289 @item %g@var{suffix}
5290 Substitute a file name that has suffix @var{suffix} and is chosen
5291 once per compilation, and mark the argument in the same way as
5292 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5293 name is now chosen in a way that is hard to predict even when previously
5294 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5295 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5296 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5297 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5298 was simply substituted with a file name chosen once per compilation,
5299 without regard to any appended suffix (which was therefore treated
5300 just like ordinary text), making such attacks more likely to succeed.
5302 @item %u@var{suffix}
5303 Like @samp{%g}, but generates a new temporary file name even if
5304 @samp{%u@var{suffix}} was already seen.
5306 @item %U@var{suffix}
5307 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5308 new one if there is no such last file name. In the absence of any
5309 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5310 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5311 would involve the generation of two distinct file names, one
5312 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5313 simply substituted with a file name chosen for the previous @samp{%u},
5314 without regard to any appended suffix.
5316 @item %j@var{suffix}
5317 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5318 writable, and if save-temps is off; otherwise, substitute the name
5319 of a temporary file, just like @samp{%u}. This temporary file is not
5320 meant for communication between processes, but rather as a junk
5323 @item %|@var{suffix}
5324 @itemx %m@var{suffix}
5325 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5326 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5327 all. These are the two most common ways to instruct a program that it
5328 should read from standard input or write to standard output. If you
5329 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5330 construct: see for example @file{f/lang-specs.h}.
5332 @item %.@var{SUFFIX}
5333 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5334 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5335 terminated by the next space or %.
5338 Marks the argument containing or following the @samp{%w} as the
5339 designated output file of this compilation. This puts the argument
5340 into the sequence of arguments that @samp{%o} will substitute later.
5343 Substitutes the names of all the output files, with spaces
5344 automatically placed around them. You should write spaces
5345 around the @samp{%o} as well or the results are undefined.
5346 @samp{%o} is for use in the specs for running the linker.
5347 Input files whose names have no recognized suffix are not compiled
5348 at all, but they are included among the output files, so they will
5352 Substitutes the suffix for object files. Note that this is
5353 handled specially when it immediately follows @samp{%g, %u, or %U},
5354 because of the need for those to form complete file names. The
5355 handling is such that @samp{%O} is treated exactly as if it had already
5356 been substituted, except that @samp{%g, %u, and %U} do not currently
5357 support additional @var{suffix} characters following @samp{%O} as they would
5358 following, for example, @samp{.o}.
5361 Substitutes the standard macro predefinitions for the
5362 current target machine. Use this when running @code{cpp}.
5365 Like @samp{%p}, but puts @samp{__} before and after the name of each
5366 predefined macro, except for macros that start with @samp{__} or with
5367 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5371 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5372 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5373 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5377 Current argument is the name of a library or startup file of some sort.
5378 Search for that file in a standard list of directories and substitute
5379 the full name found.
5382 Print @var{str} as an error message. @var{str} is terminated by a newline.
5383 Use this when inconsistent options are detected.
5386 Substitute the contents of spec string @var{name} at this point.
5389 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5391 @item %x@{@var{option}@}
5392 Accumulate an option for @samp{%X}.
5395 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5399 Output the accumulated assembler options specified by @option{-Wa}.
5402 Output the accumulated preprocessor options specified by @option{-Wp}.
5405 Process the @code{asm} spec. This is used to compute the
5406 switches to be passed to the assembler.
5409 Process the @code{asm_final} spec. This is a spec string for
5410 passing switches to an assembler post-processor, if such a program is
5414 Process the @code{link} spec. This is the spec for computing the
5415 command line passed to the linker. Typically it will make use of the
5416 @samp{%L %G %S %D and %E} sequences.
5419 Dump out a @option{-L} option for each directory that GCC believes might
5420 contain startup files. If the target supports multilibs then the
5421 current multilib directory will be prepended to each of these paths.
5424 Output the multilib directory with directory separators replaced with
5425 @samp{_}. If multilib directories are not set, or the multilib directory is
5426 @file{.} then this option emits nothing.
5429 Process the @code{lib} spec. This is a spec string for deciding which
5430 libraries should be included on the command line to the linker.
5433 Process the @code{libgcc} spec. This is a spec string for deciding
5434 which GCC support library should be included on the command line to the linker.
5437 Process the @code{startfile} spec. This is a spec for deciding which
5438 object files should be the first ones passed to the linker. Typically
5439 this might be a file named @file{crt0.o}.
5442 Process the @code{endfile} spec. This is a spec string that specifies
5443 the last object files that will be passed to the linker.
5446 Process the @code{cpp} spec. This is used to construct the arguments
5447 to be passed to the C preprocessor.
5450 Process the @code{signed_char} spec. This is intended to be used
5451 to tell cpp whether a char is signed. It typically has the definition:
5453 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
5457 Process the @code{cc1} spec. This is used to construct the options to be
5458 passed to the actual C compiler (@samp{cc1}).
5461 Process the @code{cc1plus} spec. This is used to construct the options to be
5462 passed to the actual C++ compiler (@samp{cc1plus}).
5465 Substitute the variable part of a matched option. See below.
5466 Note that each comma in the substituted string is replaced by
5470 Remove all occurrences of @code{-S} from the command line. Note---this
5471 command is position dependent. @samp{%} commands in the spec string
5472 before this one will see @code{-S}, @samp{%} commands in the spec string
5473 after this one will not.
5475 @item %:@var{function}(@var{args})
5476 Call the named function @var{function}, passing it @var{args}.
5477 @var{args} is first processed as a nested spec string, then split
5478 into an argument vector in the usual fashion. The function returns
5479 a string which is processed as if it had appeared literally as part
5480 of the current spec.
5482 The following built-in spec functions are provided:
5485 @item @code{if-exists}
5486 The @code{if-exists} spec function takes one argument, an absolute
5487 pathname to a file. If the file exists, @code{if-exists} returns the
5488 pathname. Here is a small example of its usage:
5492 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5495 @item @code{if-exists-else}
5496 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5497 spec function, except that it takes two arguments. The first argument is
5498 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5499 returns the pathname. If it does not exist, it returns the second argument.
5500 This way, @code{if-exists-else} can be used to select one file or another,
5501 based on the existence of the first. Here is a small example of its usage:
5505 crt0%O%s %:if-exists(crti%O%s) \
5506 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5511 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5512 If that switch was not specified, this substitutes nothing. Note that
5513 the leading dash is omitted when specifying this option, and it is
5514 automatically inserted if the substitution is performed. Thus the spec
5515 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5516 and would output the command line option @option{-foo}.
5518 @item %W@{@code{S}@}
5519 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5522 @item %@{@code{S}*@}
5523 Substitutes all the switches specified to GCC whose names start
5524 with @code{-S}, but which also take an argument. This is used for
5525 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5526 GCC considers @option{-o foo} as being
5527 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5528 text, including the space. Thus two arguments would be generated.
5530 @item %@{@code{S}*&@code{T}*@}
5531 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5532 (the order of @code{S} and @code{T} in the spec is not significant).
5533 There can be any number of ampersand-separated variables; for each the
5534 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5536 @item %@{@code{S}:@code{X}@}
5537 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
5539 @item %@{!@code{S}:@code{X}@}
5540 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
5542 @item %@{@code{S}*:@code{X}@}
5543 Substitutes @code{X} if one or more switches whose names start with
5544 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
5545 once, no matter how many such switches appeared. However, if @code{%*}
5546 appears somewhere in @code{X}, then @code{X} will be substituted once
5547 for each matching switch, with the @code{%*} replaced by the part of
5548 that switch that matched the @code{*}.
5550 @item %@{.@code{S}:@code{X}@}
5551 Substitutes @code{X}, if processing a file with suffix @code{S}.
5553 @item %@{!.@code{S}:@code{X}@}
5554 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
5556 @item %@{@code{S}|@code{P}:@code{X}@}
5557 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
5558 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
5559 although they have a stronger binding than the @samp{|}. If @code{%*}
5560 appears in @code{X}, all of the alternatives must be starred, and only
5561 the first matching alternative is substituted.
5563 For example, a spec string like this:
5566 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5569 will output the following command-line options from the following input
5570 command-line options:
5575 -d fred.c -foo -baz -boggle
5576 -d jim.d -bar -baz -boggle
5579 @item %@{S:X; T:Y; :D@}
5581 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
5582 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
5583 be as many clauses as you need. This may be combined with @code{.},
5584 @code{!}, @code{|}, and @code{*} as needed.
5589 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
5590 construct may contain other nested @samp{%} constructs or spaces, or
5591 even newlines. They are processed as usual, as described above.
5592 Trailing white space in @code{X} is ignored. White space may also
5593 appear anywhere on the left side of the colon in these constructs,
5594 except between @code{.} or @code{*} and the corresponding word.
5596 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
5597 handled specifically in these constructs. If another value of
5598 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5599 @option{-W} switch is found later in the command line, the earlier
5600 switch value is ignored, except with @{@code{S}*@} where @code{S} is
5601 just one letter, which passes all matching options.
5603 The character @samp{|} at the beginning of the predicate text is used to
5604 indicate that a command should be piped to the following command, but
5605 only if @option{-pipe} is specified.
5607 It is built into GCC which switches take arguments and which do not.
5608 (You might think it would be useful to generalize this to allow each
5609 compiler's spec to say which switches take arguments. But this cannot
5610 be done in a consistent fashion. GCC cannot even decide which input
5611 files have been specified without knowing which switches take arguments,
5612 and it must know which input files to compile in order to tell which
5615 GCC also knows implicitly that arguments starting in @option{-l} are to be
5616 treated as compiler output files, and passed to the linker in their
5617 proper position among the other output files.
5619 @c man begin OPTIONS
5621 @node Target Options
5622 @section Specifying Target Machine and Compiler Version
5623 @cindex target options
5624 @cindex cross compiling
5625 @cindex specifying machine version
5626 @cindex specifying compiler version and target machine
5627 @cindex compiler version, specifying
5628 @cindex target machine, specifying
5630 The usual way to run GCC is to run the executable called @file{gcc}, or
5631 @file{<machine>-gcc} when cross-compiling, or
5632 @file{<machine>-gcc-<version>} to run a version other than the one that
5633 was installed last. Sometimes this is inconvenient, so GCC provides
5634 options that will switch to another cross-compiler or version.
5637 @item -b @var{machine}
5639 The argument @var{machine} specifies the target machine for compilation.
5641 The value to use for @var{machine} is the same as was specified as the
5642 machine type when configuring GCC as a cross-compiler. For
5643 example, if a cross-compiler was configured with @samp{configure
5644 i386v}, meaning to compile for an 80386 running System V, then you
5645 would specify @option{-b i386v} to run that cross compiler.
5647 @item -V @var{version}
5649 The argument @var{version} specifies which version of GCC to run.
5650 This is useful when multiple versions are installed. For example,
5651 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5654 The @option{-V} and @option{-b} options work by running the
5655 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5656 use them if you can just run that directly.
5658 @node Submodel Options
5659 @section Hardware Models and Configurations
5660 @cindex submodel options
5661 @cindex specifying hardware config
5662 @cindex hardware models and configurations, specifying
5663 @cindex machine dependent options
5665 Earlier we discussed the standard option @option{-b} which chooses among
5666 different installed compilers for completely different target
5667 machines, such as VAX vs.@: 68000 vs.@: 80386.
5669 In addition, each of these target machine types can have its own
5670 special options, starting with @samp{-m}, to choose among various
5671 hardware models or configurations---for example, 68010 vs 68020,
5672 floating coprocessor or none. A single installed version of the
5673 compiler can compile for any model or configuration, according to the
5676 Some configurations of the compiler also support additional special
5677 options, usually for compatibility with other compilers on the same
5680 These options are defined by the macro @code{TARGET_SWITCHES} in the
5681 machine description. The default for the options is also defined by
5682 that macro, which enables you to change the defaults.
5694 * RS/6000 and PowerPC Options::
5698 * i386 and x86-64 Options::
5700 * Intel 960 Options::
5701 * DEC Alpha Options::
5702 * DEC Alpha/VMS Options::
5705 * System V Options::
5706 * TMS320C3x/C4x Options::
5714 * S/390 and zSeries Options::
5718 * Xstormy16 Options::
5723 @node M680x0 Options
5724 @subsection M680x0 Options
5725 @cindex M680x0 options
5727 These are the @samp{-m} options defined for the 68000 series. The default
5728 values for these options depends on which style of 68000 was selected when
5729 the compiler was configured; the defaults for the most common choices are
5737 Generate output for a 68000. This is the default
5738 when the compiler is configured for 68000-based systems.
5740 Use this option for microcontrollers with a 68000 or EC000 core,
5741 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5747 Generate output for a 68020. This is the default
5748 when the compiler is configured for 68020-based systems.
5752 Generate output containing 68881 instructions for floating point.
5753 This is the default for most 68020 systems unless @option{--nfp} was
5754 specified when the compiler was configured.
5758 Generate output for a 68030. This is the default when the compiler is
5759 configured for 68030-based systems.
5763 Generate output for a 68040. This is the default when the compiler is
5764 configured for 68040-based systems.
5766 This option inhibits the use of 68881/68882 instructions that have to be
5767 emulated by software on the 68040. Use this option if your 68040 does not
5768 have code to emulate those instructions.
5772 Generate output for a 68060. This is the default when the compiler is
5773 configured for 68060-based systems.
5775 This option inhibits the use of 68020 and 68881/68882 instructions that
5776 have to be emulated by software on the 68060. Use this option if your 68060
5777 does not have code to emulate those instructions.
5781 Generate output for a CPU32. This is the default
5782 when the compiler is configured for CPU32-based systems.
5784 Use this option for microcontrollers with a
5785 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5786 68336, 68340, 68341, 68349 and 68360.
5790 Generate output for a 520X ``coldfire'' family cpu. This is the default
5791 when the compiler is configured for 520X-based systems.
5793 Use this option for microcontroller with a 5200 core, including
5794 the MCF5202, MCF5203, MCF5204 and MCF5202.
5799 Generate output for a 68040, without using any of the new instructions.
5800 This results in code which can run relatively efficiently on either a
5801 68020/68881 or a 68030 or a 68040. The generated code does use the
5802 68881 instructions that are emulated on the 68040.
5806 Generate output for a 68060, without using any of the new instructions.
5807 This results in code which can run relatively efficiently on either a
5808 68020/68881 or a 68030 or a 68040. The generated code does use the
5809 68881 instructions that are emulated on the 68060.
5812 @opindex msoft-float
5813 Generate output containing library calls for floating point.
5814 @strong{Warning:} the requisite libraries are not available for all m68k
5815 targets. Normally the facilities of the machine's usual C compiler are
5816 used, but this can't be done directly in cross-compilation. You must
5817 make your own arrangements to provide suitable library functions for
5818 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5819 @samp{m68k-*-coff} do provide software floating point support.
5823 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5826 @opindex mnobitfield
5827 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5828 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5832 Do use the bit-field instructions. The @option{-m68020} option implies
5833 @option{-mbitfield}. This is the default if you use a configuration
5834 designed for a 68020.
5838 Use a different function-calling convention, in which functions
5839 that take a fixed number of arguments return with the @code{rtd}
5840 instruction, which pops their arguments while returning. This
5841 saves one instruction in the caller since there is no need to pop
5842 the arguments there.
5844 This calling convention is incompatible with the one normally
5845 used on Unix, so you cannot use it if you need to call libraries
5846 compiled with the Unix compiler.
5848 Also, you must provide function prototypes for all functions that
5849 take variable numbers of arguments (including @code{printf});
5850 otherwise incorrect code will be generated for calls to those
5853 In addition, seriously incorrect code will result if you call a
5854 function with too many arguments. (Normally, extra arguments are
5855 harmlessly ignored.)
5857 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5858 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5861 @itemx -mno-align-int
5863 @opindex mno-align-int
5864 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5865 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5866 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
5867 Aligning variables on 32-bit boundaries produces code that runs somewhat
5868 faster on processors with 32-bit busses at the expense of more memory.
5870 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
5871 align structures containing the above types differently than
5872 most published application binary interface specifications for the m68k.
5876 Use the pc-relative addressing mode of the 68000 directly, instead of
5877 using a global offset table. At present, this option implies @option{-fpic},
5878 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
5879 not presently supported with @option{-mpcrel}, though this could be supported for
5880 68020 and higher processors.
5882 @item -mno-strict-align
5883 @itemx -mstrict-align
5884 @opindex mno-strict-align
5885 @opindex mstrict-align
5886 Do not (do) assume that unaligned memory references will be handled by
5891 @node M68hc1x Options
5892 @subsection M68hc1x Options
5893 @cindex M68hc1x options
5895 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5896 microcontrollers. The default values for these options depends on
5897 which style of microcontroller was selected when the compiler was configured;
5898 the defaults for the most common choices are given below.
5905 Generate output for a 68HC11. This is the default
5906 when the compiler is configured for 68HC11-based systems.
5912 Generate output for a 68HC12. This is the default
5913 when the compiler is configured for 68HC12-based systems.
5919 Generate output for a 68HCS12.
5922 @opindex mauto-incdec
5923 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5930 Enable the use of 68HC12 min and max instructions.
5933 @itemx -mno-long-calls
5934 @opindex mlong-calls
5935 @opindex mno-long-calls
5936 Treat all calls as being far away (near). If calls are assumed to be
5937 far away, the compiler will use the @code{call} instruction to
5938 call a function and the @code{rtc} instruction for returning.
5942 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5944 @item -msoft-reg-count=@var{count}
5945 @opindex msoft-reg-count
5946 Specify the number of pseudo-soft registers which are used for the
5947 code generation. The maximum number is 32. Using more pseudo-soft
5948 register may or may not result in better code depending on the program.
5949 The default is 4 for 68HC11 and 2 for 68HC12.
5954 @subsection VAX Options
5957 These @samp{-m} options are defined for the VAX:
5962 Do not output certain jump instructions (@code{aobleq} and so on)
5963 that the Unix assembler for the VAX cannot handle across long
5968 Do output those jump instructions, on the assumption that you
5969 will assemble with the GNU assembler.
5973 Output code for g-format floating point numbers instead of d-format.
5977 @subsection SPARC Options
5978 @cindex SPARC options
5980 These @samp{-m} switches are supported on the SPARC:
5985 @opindex mno-app-regs
5987 Specify @option{-mapp-regs} to generate output using the global registers
5988 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
5991 To be fully SVR4 ABI compliant at the cost of some performance loss,
5992 specify @option{-mno-app-regs}. You should compile libraries and system
5993 software with this option.
5998 @opindex mhard-float
5999 Generate output containing floating point instructions. This is the
6005 @opindex msoft-float
6006 Generate output containing library calls for floating point.
6007 @strong{Warning:} the requisite libraries are not available for all SPARC
6008 targets. Normally the facilities of the machine's usual C compiler are
6009 used, but this cannot be done directly in cross-compilation. You must make
6010 your own arrangements to provide suitable library functions for
6011 cross-compilation. The embedded targets @samp{sparc-*-aout} and
6012 @samp{sparclite-*-*} do provide software floating point support.
6014 @option{-msoft-float} changes the calling convention in the output file;
6015 therefore, it is only useful if you compile @emph{all} of a program with
6016 this option. In particular, you need to compile @file{libgcc.a}, the
6017 library that comes with GCC, with @option{-msoft-float} in order for
6020 @item -mhard-quad-float
6021 @opindex mhard-quad-float
6022 Generate output containing quad-word (long double) floating point
6026 @opindex mimpure-text
6027 @option{-mimpure-text}, used in addition to @option{-shared}, tells
6028 the compiler to not pass @option{-z text} to the linker when linking a
6029 shared object. Using this option, you can link position-dependent
6030 code into a shared object.
6032 @option{-mimpure-text} suppresses the ``relocations remain against
6033 allocatable but non-writable sections'' linker error message.
6034 However, the necessary relocations will trigger copy-on-write, and the
6035 shared object is not actually shared across processes. Instead of
6036 using @option{-mimpure-text}, you should compile all source code with
6037 @option{-fpic} or @option{-fPIC}.
6039 This option is only available on SunOS and Solaris.
6041 @item -msoft-quad-float
6042 @opindex msoft-quad-float
6043 Generate output containing library calls for quad-word (long double)
6044 floating point instructions. The functions called are those specified
6045 in the SPARC ABI@. This is the default.
6047 As of this writing, there are no sparc implementations that have hardware
6048 support for the quad-word floating point instructions. They all invoke
6049 a trap handler for one of these instructions, and then the trap handler
6050 emulates the effect of the instruction. Because of the trap handler overhead,
6051 this is much slower than calling the ABI library routines. Thus the
6052 @option{-msoft-quad-float} option is the default.
6058 With @option{-mflat}, the compiler does not generate save/restore instructions
6059 and will use a ``flat'' or single register window calling convention.
6060 This model uses %i7 as the frame pointer and is compatible with the normal
6061 register window model. Code from either may be intermixed.
6062 The local registers and the input registers (0--5) are still treated as
6063 ``call saved'' registers and will be saved on the stack as necessary.
6065 With @option{-mno-flat} (the default), the compiler emits save/restore
6066 instructions (except for leaf functions) and is the normal mode of operation.
6068 @item -mno-unaligned-doubles
6069 @itemx -munaligned-doubles
6070 @opindex mno-unaligned-doubles
6071 @opindex munaligned-doubles
6072 Assume that doubles have 8 byte alignment. This is the default.
6074 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
6075 alignment only if they are contained in another type, or if they have an
6076 absolute address. Otherwise, it assumes they have 4 byte alignment.
6077 Specifying this option avoids some rare compatibility problems with code
6078 generated by other compilers. It is not the default because it results
6079 in a performance loss, especially for floating point code.
6081 @item -mno-faster-structs
6082 @itemx -mfaster-structs
6083 @opindex mno-faster-structs
6084 @opindex mfaster-structs
6085 With @option{-mfaster-structs}, the compiler assumes that structures
6086 should have 8 byte alignment. This enables the use of pairs of
6087 @code{ldd} and @code{std} instructions for copies in structure
6088 assignment, in place of twice as many @code{ld} and @code{st} pairs.
6089 However, the use of this changed alignment directly violates the SPARC
6090 ABI@. Thus, it's intended only for use on targets where the developer
6091 acknowledges that their resulting code will not be directly in line with
6092 the rules of the ABI@.
6098 These two options select variations on the SPARC architecture.
6100 By default (unless specifically configured for the Fujitsu SPARClite),
6101 GCC generates code for the v7 variant of the SPARC architecture.
6103 @option{-mv8} will give you SPARC v8 code. The only difference from v7
6104 code is that the compiler emits the integer multiply and integer
6105 divide instructions which exist in SPARC v8 but not in SPARC v7.
6107 @option{-msparclite} will give you SPARClite code. This adds the integer
6108 multiply, integer divide step and scan (@code{ffs}) instructions which
6109 exist in SPARClite but not in SPARC v7.
6111 These options are deprecated and will be deleted in a future GCC release.
6112 They have been replaced with @option{-mcpu=xxx}.
6117 @opindex msupersparc
6118 These two options select the processor for which the code is optimized.
6120 With @option{-mcypress} (the default), the compiler optimizes code for the
6121 Cypress CY7C602 chip, as used in the SPARCStation/SPARCServer 3xx series.
6122 This is also appropriate for the older SPARCStation 1, 2, IPX etc.
6124 With @option{-msupersparc} the compiler optimizes code for the SuperSPARC cpu, as
6125 used in the SPARCStation 10, 1000 and 2000 series. This flag also enables use
6126 of the full SPARC v8 instruction set.
6128 These options are deprecated and will be deleted in a future GCC release.
6129 They have been replaced with @option{-mcpu=xxx}.
6131 @item -mcpu=@var{cpu_type}
6133 Set the instruction set, register set, and instruction scheduling parameters
6134 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
6135 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
6136 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
6137 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
6140 Default instruction scheduling parameters are used for values that select
6141 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
6142 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
6144 Here is a list of each supported architecture and their supported
6149 v8: supersparc, hypersparc
6150 sparclite: f930, f934, sparclite86x
6152 v9: ultrasparc, ultrasparc3
6155 @item -mtune=@var{cpu_type}
6157 Set the instruction scheduling parameters for machine type
6158 @var{cpu_type}, but do not set the instruction set or register set that the
6159 option @option{-mcpu=@var{cpu_type}} would.
6161 The same values for @option{-mcpu=@var{cpu_type}} can be used for
6162 @option{-mtune=@var{cpu_type}}, but the only useful values are those
6163 that select a particular cpu implementation. Those are @samp{cypress},
6164 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
6165 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
6170 These @samp{-m} switches are supported in addition to the above
6171 on the SPARCLET processor.
6174 @item -mlittle-endian
6175 @opindex mlittle-endian
6176 Generate code for a processor running in little-endian mode.
6180 Treat register @code{%g0} as a normal register.
6181 GCC will continue to clobber it as necessary but will not assume
6182 it always reads as 0.
6184 @item -mbroken-saverestore
6185 @opindex mbroken-saverestore
6186 Generate code that does not use non-trivial forms of the @code{save} and
6187 @code{restore} instructions. Early versions of the SPARCLET processor do
6188 not correctly handle @code{save} and @code{restore} instructions used with
6189 arguments. They correctly handle them used without arguments. A @code{save}
6190 instruction used without arguments increments the current window pointer
6191 but does not allocate a new stack frame. It is assumed that the window
6192 overflow trap handler will properly handle this case as will interrupt
6196 These @samp{-m} switches are supported in addition to the above
6197 on SPARC V9 processors in 64-bit environments.
6200 @item -mlittle-endian
6201 @opindex mlittle-endian
6202 Generate code for a processor running in little-endian mode.
6208 Generate code for a 32-bit or 64-bit environment.
6209 The 32-bit environment sets int, long and pointer to 32 bits.
6210 The 64-bit environment sets int to 32 bits and long and pointer
6213 @item -mcmodel=medlow
6214 @opindex mcmodel=medlow
6215 Generate code for the Medium/Low code model: the program must be linked
6216 in the low 32 bits of the address space. Pointers are 64 bits.
6217 Programs can be statically or dynamically linked.
6219 @item -mcmodel=medmid
6220 @opindex mcmodel=medmid
6221 Generate code for the Medium/Middle code model: the program must be linked
6222 in the low 44 bits of the address space, the text segment must be less than
6223 2G bytes, and data segment must be within 2G of the text segment.
6224 Pointers are 64 bits.
6226 @item -mcmodel=medany
6227 @opindex mcmodel=medany
6228 Generate code for the Medium/Anywhere code model: the program may be linked
6229 anywhere in the address space, the text segment must be less than
6230 2G bytes, and data segment must be within 2G of the text segment.
6231 Pointers are 64 bits.
6233 @item -mcmodel=embmedany
6234 @opindex mcmodel=embmedany
6235 Generate code for the Medium/Anywhere code model for embedded systems:
6236 assume a 32-bit text and a 32-bit data segment, both starting anywhere
6237 (determined at link time). Register %g4 points to the base of the
6238 data segment. Pointers are still 64 bits.
6239 Programs are statically linked, PIC is not supported.
6242 @itemx -mno-stack-bias
6243 @opindex mstack-bias
6244 @opindex mno-stack-bias
6245 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
6246 frame pointer if present, are offset by @minus{}2047 which must be added back
6247 when making stack frame references.
6248 Otherwise, assume no such offset is present.
6252 @subsection ARM Options
6255 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6260 @opindex mapcs-frame
6261 Generate a stack frame that is compliant with the ARM Procedure Call
6262 Standard for all functions, even if this is not strictly necessary for
6263 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6264 with this option will cause the stack frames not to be generated for
6265 leaf functions. The default is @option{-mno-apcs-frame}.
6269 This is a synonym for @option{-mapcs-frame}.
6273 Generate code for a processor running with a 26-bit program counter,
6274 and conforming to the function calling standards for the APCS 26-bit
6275 option. This option replaces the @option{-m2} and @option{-m3} options
6276 of previous releases of the compiler.
6280 Generate code for a processor running with a 32-bit program counter,
6281 and conforming to the function calling standards for the APCS 32-bit
6282 option. This option replaces the @option{-m6} option of previous releases
6286 @c not currently implemented
6287 @item -mapcs-stack-check
6288 @opindex mapcs-stack-check
6289 Generate code to check the amount of stack space available upon entry to
6290 every function (that actually uses some stack space). If there is
6291 insufficient space available then either the function
6292 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6293 called, depending upon the amount of stack space required. The run time
6294 system is required to provide these functions. The default is
6295 @option{-mno-apcs-stack-check}, since this produces smaller code.
6297 @c not currently implemented
6299 @opindex mapcs-float
6300 Pass floating point arguments using the float point registers. This is
6301 one of the variants of the APCS@. This option is recommended if the
6302 target hardware has a floating point unit or if a lot of floating point
6303 arithmetic is going to be performed by the code. The default is
6304 @option{-mno-apcs-float}, since integer only code is slightly increased in
6305 size if @option{-mapcs-float} is used.
6307 @c not currently implemented
6308 @item -mapcs-reentrant
6309 @opindex mapcs-reentrant
6310 Generate reentrant, position independent code. The default is
6311 @option{-mno-apcs-reentrant}.
6314 @item -mthumb-interwork
6315 @opindex mthumb-interwork
6316 Generate code which supports calling between the ARM and Thumb
6317 instruction sets. Without this option the two instruction sets cannot
6318 be reliably used inside one program. The default is
6319 @option{-mno-thumb-interwork}, since slightly larger code is generated
6320 when @option{-mthumb-interwork} is specified.
6322 @item -mno-sched-prolog
6323 @opindex mno-sched-prolog
6324 Prevent the reordering of instructions in the function prolog, or the
6325 merging of those instruction with the instructions in the function's
6326 body. This means that all functions will start with a recognizable set
6327 of instructions (or in fact one of a choice from a small set of
6328 different function prologues), and this information can be used to
6329 locate the start if functions inside an executable piece of code. The
6330 default is @option{-msched-prolog}.
6333 @opindex mhard-float
6334 Generate output containing floating point instructions. This is the
6338 @opindex msoft-float
6339 Generate output containing library calls for floating point.
6340 @strong{Warning:} the requisite libraries are not available for all ARM
6341 targets. Normally the facilities of the machine's usual C compiler are
6342 used, but this cannot be done directly in cross-compilation. You must make
6343 your own arrangements to provide suitable library functions for
6346 @option{-msoft-float} changes the calling convention in the output file;
6347 therefore, it is only useful if you compile @emph{all} of a program with
6348 this option. In particular, you need to compile @file{libgcc.a}, the
6349 library that comes with GCC, with @option{-msoft-float} in order for
6352 @item -mlittle-endian
6353 @opindex mlittle-endian
6354 Generate code for a processor running in little-endian mode. This is
6355 the default for all standard configurations.
6358 @opindex mbig-endian
6359 Generate code for a processor running in big-endian mode; the default is
6360 to compile code for a little-endian processor.
6362 @item -mwords-little-endian
6363 @opindex mwords-little-endian
6364 This option only applies when generating code for big-endian processors.
6365 Generate code for a little-endian word order but a big-endian byte
6366 order. That is, a byte order of the form @samp{32107654}. Note: this
6367 option should only be used if you require compatibility with code for
6368 big-endian ARM processors generated by versions of the compiler prior to
6371 @item -malignment-traps
6372 @opindex malignment-traps
6373 Generate code that will not trap if the MMU has alignment traps enabled.
6374 On ARM architectures prior to ARMv4, there were no instructions to
6375 access half-word objects stored in memory. However, when reading from
6376 memory a feature of the ARM architecture allows a word load to be used,
6377 even if the address is unaligned, and the processor core will rotate the
6378 data as it is being loaded. This option tells the compiler that such
6379 misaligned accesses will cause a MMU trap and that it should instead
6380 synthesize the access as a series of byte accesses. The compiler can
6381 still use word accesses to load half-word data if it knows that the
6382 address is aligned to a word boundary.
6384 This option is ignored when compiling for ARM architecture 4 or later,
6385 since these processors have instructions to directly access half-word
6388 @item -mno-alignment-traps
6389 @opindex mno-alignment-traps
6390 Generate code that assumes that the MMU will not trap unaligned
6391 accesses. This produces better code when the target instruction set
6392 does not have half-word memory operations (i.e.@: implementations prior to
6395 Note that you cannot use this option to access unaligned word objects,
6396 since the processor will only fetch one 32-bit aligned object from
6399 The default setting for most targets is @option{-mno-alignment-traps}, since
6400 this produces better code when there are no half-word memory
6401 instructions available.
6403 @item -mshort-load-bytes
6404 @itemx -mno-short-load-words
6405 @opindex mshort-load-bytes
6406 @opindex mno-short-load-words
6407 These are deprecated aliases for @option{-malignment-traps}.
6409 @item -mno-short-load-bytes
6410 @itemx -mshort-load-words
6411 @opindex mno-short-load-bytes
6412 @opindex mshort-load-words
6413 This are deprecated aliases for @option{-mno-alignment-traps}.
6415 @item -mcpu=@var{name}
6417 This specifies the name of the target ARM processor. GCC uses this name
6418 to determine what kind of instructions it can emit when generating
6419 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6420 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6421 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6422 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6423 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6424 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6425 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6426 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6427 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
6428 @samp{arm1020t}, @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
6430 @itemx -mtune=@var{name}
6432 This option is very similar to the @option{-mcpu=} option, except that
6433 instead of specifying the actual target processor type, and hence
6434 restricting which instructions can be used, it specifies that GCC should
6435 tune the performance of the code as if the target were of the type
6436 specified in this option, but still choosing the instructions that it
6437 will generate based on the cpu specified by a @option{-mcpu=} option.
6438 For some ARM implementations better performance can be obtained by using
6441 @item -march=@var{name}
6443 This specifies the name of the target ARM architecture. GCC uses this
6444 name to determine what kind of instructions it can emit when generating
6445 assembly code. This option can be used in conjunction with or instead
6446 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6447 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6448 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{iwmmxt}, @samp{ep9312}.
6450 @item -mfpe=@var{number}
6451 @itemx -mfp=@var{number}
6454 This specifies the version of the floating point emulation available on
6455 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6456 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6458 @item -mstructure-size-boundary=@var{n}
6459 @opindex mstructure-size-boundary
6460 The size of all structures and unions will be rounded up to a multiple
6461 of the number of bits set by this option. Permissible values are 8 and
6462 32. The default value varies for different toolchains. For the COFF
6463 targeted toolchain the default value is 8. Specifying the larger number
6464 can produce faster, more efficient code, but can also increase the size
6465 of the program. The two values are potentially incompatible. Code
6466 compiled with one value cannot necessarily expect to work with code or
6467 libraries compiled with the other value, if they exchange information
6468 using structures or unions.
6470 @item -mabort-on-noreturn
6471 @opindex mabort-on-noreturn
6472 Generate a call to the function @code{abort} at the end of a
6473 @code{noreturn} function. It will be executed if the function tries to
6477 @itemx -mno-long-calls
6478 @opindex mlong-calls
6479 @opindex mno-long-calls
6480 Tells the compiler to perform function calls by first loading the
6481 address of the function into a register and then performing a subroutine
6482 call on this register. This switch is needed if the target function
6483 will lie outside of the 64 megabyte addressing range of the offset based
6484 version of subroutine call instruction.
6486 Even if this switch is enabled, not all function calls will be turned
6487 into long calls. The heuristic is that static functions, functions
6488 which have the @samp{short-call} attribute, functions that are inside
6489 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6490 definitions have already been compiled within the current compilation
6491 unit, will not be turned into long calls. The exception to this rule is
6492 that weak function definitions, functions with the @samp{long-call}
6493 attribute or the @samp{section} attribute, and functions that are within
6494 the scope of a @samp{#pragma long_calls} directive, will always be
6495 turned into long calls.
6497 This feature is not enabled by default. Specifying
6498 @option{-mno-long-calls} will restore the default behavior, as will
6499 placing the function calls within the scope of a @samp{#pragma
6500 long_calls_off} directive. Note these switches have no effect on how
6501 the compiler generates code to handle function calls via function
6504 @item -mnop-fun-dllimport
6505 @opindex mnop-fun-dllimport
6506 Disable support for the @code{dllimport} attribute.
6508 @item -msingle-pic-base
6509 @opindex msingle-pic-base
6510 Treat the register used for PIC addressing as read-only, rather than
6511 loading it in the prologue for each function. The run-time system is
6512 responsible for initializing this register with an appropriate value
6513 before execution begins.
6515 @item -mpic-register=@var{reg}
6516 @opindex mpic-register
6517 Specify the register to be used for PIC addressing. The default is R10
6518 unless stack-checking is enabled, when R9 is used.
6520 @item -mcirrus-fix-invalid-insns
6521 @opindex mcirrus-fix-invalid-insns
6522 @opindex mno-cirrus-fix-invalid-insns
6523 Insert NOPs into the instruction stream to in order to work around
6524 problems with invalid Maverick instruction combinations. This option
6525 is only valid if the @option{-mcpu=ep9312} option has been used to
6526 enable generation of instructions for the Cirrus Maverick floating
6527 point co-processor. This option is not enabled by default, since the
6528 problem is only present in older Maverick implementations. The default
6529 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6532 @item -mpoke-function-name
6533 @opindex mpoke-function-name
6534 Write the name of each function into the text section, directly
6535 preceding the function prologue. The generated code is similar to this:
6539 .ascii "arm_poke_function_name", 0
6542 .word 0xff000000 + (t1 - t0)
6543 arm_poke_function_name
6545 stmfd sp!, @{fp, ip, lr, pc@}
6549 When performing a stack backtrace, code can inspect the value of
6550 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6551 location @code{pc - 12} and the top 8 bits are set, then we know that
6552 there is a function name embedded immediately preceding this location
6553 and has length @code{((pc[-3]) & 0xff000000)}.
6557 Generate code for the 16-bit Thumb instruction set. The default is to
6558 use the 32-bit ARM instruction set.
6561 @opindex mtpcs-frame
6562 Generate a stack frame that is compliant with the Thumb Procedure Call
6563 Standard for all non-leaf functions. (A leaf function is one that does
6564 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6566 @item -mtpcs-leaf-frame
6567 @opindex mtpcs-leaf-frame
6568 Generate a stack frame that is compliant with the Thumb Procedure Call
6569 Standard for all leaf functions. (A leaf function is one that does
6570 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6572 @item -mcallee-super-interworking
6573 @opindex mcallee-super-interworking
6574 Gives all externally visible functions in the file being compiled an ARM
6575 instruction set header which switches to Thumb mode before executing the
6576 rest of the function. This allows these functions to be called from
6577 non-interworking code.
6579 @item -mcaller-super-interworking
6580 @opindex mcaller-super-interworking
6581 Allows calls via function pointers (including virtual functions) to
6582 execute correctly regardless of whether the target code has been
6583 compiled for interworking or not. There is a small overhead in the cost
6584 of executing a function pointer if this option is enabled.
6588 @node MN10200 Options
6589 @subsection MN10200 Options
6590 @cindex MN10200 options
6592 These @option{-m} options are defined for Matsushita MN10200 architectures:
6597 Indicate to the linker that it should perform a relaxation optimization pass
6598 to shorten branches, calls and absolute memory addresses. This option only
6599 has an effect when used on the command line for the final link step.
6601 This option makes symbolic debugging impossible.
6604 @node MN10300 Options
6605 @subsection MN10300 Options
6606 @cindex MN10300 options
6608 These @option{-m} options are defined for Matsushita MN10300 architectures:
6613 Generate code to avoid bugs in the multiply instructions for the MN10300
6614 processors. This is the default.
6617 @opindex mno-mult-bug
6618 Do not generate code to avoid bugs in the multiply instructions for the
6623 Generate code which uses features specific to the AM33 processor.
6627 Do not generate code which uses features specific to the AM33 processor. This
6632 Do not link in the C run-time initialization object file.
6636 Indicate to the linker that it should perform a relaxation optimization pass
6637 to shorten branches, calls and absolute memory addresses. This option only
6638 has an effect when used on the command line for the final link step.
6640 This option makes symbolic debugging impossible.
6644 @node M32R/D Options
6645 @subsection M32R/D Options
6646 @cindex M32R/D options
6648 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6653 Generate code for the M32R/X@.
6657 Generate code for the M32R@. This is the default.
6659 @item -mcode-model=small
6660 @opindex mcode-model=small
6661 Assume all objects live in the lower 16MB of memory (so that their addresses
6662 can be loaded with the @code{ld24} instruction), and assume all subroutines
6663 are reachable with the @code{bl} instruction.
6664 This is the default.
6666 The addressability of a particular object can be set with the
6667 @code{model} attribute.
6669 @item -mcode-model=medium
6670 @opindex mcode-model=medium
6671 Assume objects may be anywhere in the 32-bit address space (the compiler
6672 will generate @code{seth/add3} instructions to load their addresses), and
6673 assume all subroutines are reachable with the @code{bl} instruction.
6675 @item -mcode-model=large
6676 @opindex mcode-model=large
6677 Assume objects may be anywhere in the 32-bit address space (the compiler
6678 will generate @code{seth/add3} instructions to load their addresses), and
6679 assume subroutines may not be reachable with the @code{bl} instruction
6680 (the compiler will generate the much slower @code{seth/add3/jl}
6681 instruction sequence).
6684 @opindex msdata=none
6685 Disable use of the small data area. Variables will be put into
6686 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6687 @code{section} attribute has been specified).
6688 This is the default.
6690 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6691 Objects may be explicitly put in the small data area with the
6692 @code{section} attribute using one of these sections.
6695 @opindex msdata=sdata
6696 Put small global and static data in the small data area, but do not
6697 generate special code to reference them.
6701 Put small global and static data in the small data area, and generate
6702 special instructions to reference them.
6706 @cindex smaller data references
6707 Put global and static objects less than or equal to @var{num} bytes
6708 into the small data or bss sections instead of the normal data or bss
6709 sections. The default value of @var{num} is 8.
6710 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6711 for this option to have any effect.
6713 All modules should be compiled with the same @option{-G @var{num}} value.
6714 Compiling with different values of @var{num} may or may not work; if it
6715 doesn't the linker will give an error message---incorrect code will not be
6721 @subsection M88K Options
6722 @cindex M88k options
6724 These @samp{-m} options are defined for Motorola 88k architectures:
6729 Generate code that works well on both the m88100 and the
6734 Generate code that works best for the m88100, but that also
6739 Generate code that works best for the m88110, and may not run
6744 Obsolete option to be removed from the next revision.
6747 @item -midentify-revision
6748 @opindex midentify-revision
6749 @cindex identifying source, compiler (88k)
6750 Include an @code{ident} directive in the assembler output recording the
6751 source file name, compiler name and version, timestamp, and compilation
6754 @item -mno-underscores
6755 @opindex mno-underscores
6756 @cindex underscores, avoiding (88k)
6757 In assembler output, emit symbol names without adding an underscore
6758 character at the beginning of each name. The default is to use an
6759 underscore as prefix on each name.
6761 @item -mocs-debug-info
6762 @itemx -mno-ocs-debug-info
6763 @opindex mocs-debug-info
6764 @opindex mno-ocs-debug-info
6766 @cindex debugging, 88k OCS
6767 Include (or omit) additional debugging information (about registers used
6768 in each stack frame) as specified in the 88open Object Compatibility
6769 Standard, ``OCS''@. This extra information allows debugging of code that
6770 has had the frame pointer eliminated. The default for SVr4 and Delta 88
6771 SVr3.2 is to include this information; other 88k configurations omit this
6772 information by default.
6774 @item -mocs-frame-position
6775 @opindex mocs-frame-position
6776 @cindex register positions in frame (88k)
6777 When emitting COFF debugging information for automatic variables and
6778 parameters stored on the stack, use the offset from the canonical frame
6779 address, which is the stack pointer (register 31) on entry to the
6780 function. The SVr4 and Delta88 SVr3.2, and BCS configurations use
6781 @option{-mocs-frame-position}; other 88k configurations have the default
6782 @option{-mno-ocs-frame-position}.
6784 @item -mno-ocs-frame-position
6785 @opindex mno-ocs-frame-position
6786 @cindex register positions in frame (88k)
6787 When emitting COFF debugging information for automatic variables and
6788 parameters stored on the stack, use the offset from the frame pointer
6789 register (register 30). When this option is in effect, the frame
6790 pointer is not eliminated when debugging information is selected by the
6793 @item -moptimize-arg-area
6794 @opindex moptimize-arg-area
6795 @cindex arguments in frame (88k)
6796 Save space by reorganizing the stack frame. This option generates code
6797 that does not agree with the 88open specifications, but uses less
6800 @itemx -mno-optimize-arg-area
6801 @opindex mno-optimize-arg-area
6802 Do not reorganize the stack frame to save space. This is the default.
6803 The generated conforms to the specification, but uses more memory.
6805 @item -mshort-data-@var{num}
6806 @opindex mshort-data
6807 @cindex smaller data references (88k)
6808 @cindex r0-relative references (88k)
6809 Generate smaller data references by making them relative to @code{r0},
6810 which allows loading a value using a single instruction (rather than the
6811 usual two). You control which data references are affected by
6812 specifying @var{num} with this option. For example, if you specify
6813 @option{-mshort-data-512}, then the data references affected are those
6814 involving displacements of less than 512 bytes.
6815 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6818 @item -mserialize-volatile
6819 @opindex mserialize-volatile
6820 @itemx -mno-serialize-volatile
6821 @opindex mno-serialize-volatile
6822 @cindex sequential consistency on 88k
6823 Do, or don't, generate code to guarantee sequential consistency
6824 of volatile memory references. By default, consistency is
6827 The order of memory references made by the MC88110 processor does
6828 not always match the order of the instructions requesting those
6829 references. In particular, a load instruction may execute before
6830 a preceding store instruction. Such reordering violates
6831 sequential consistency of volatile memory references, when there
6832 are multiple processors. When consistency must be guaranteed,
6833 GCC generates special instructions, as needed, to force
6834 execution in the proper order.
6836 The MC88100 processor does not reorder memory references and so
6837 always provides sequential consistency. However, by default, GCC
6838 generates the special instructions to guarantee consistency
6839 even when you use @option{-m88100}, so that the code may be run on an
6840 MC88110 processor. If you intend to run your code only on the
6841 MC88100 processor, you may use @option{-mno-serialize-volatile}.
6843 The extra code generated to guarantee consistency may affect the
6844 performance of your application. If you know that you can safely
6845 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
6851 @cindex assembler syntax, 88k
6853 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
6854 related to System V release 4 (SVr4). This controls the following:
6858 Which variant of the assembler syntax to emit.
6860 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6861 that is used on System V release 4.
6863 @option{-msvr4} makes GCC issue additional declaration directives used in
6867 @option{-msvr4} is the default for the m88k-motorola-sysv4 configuration.
6868 @option{-msvr3} is the default for all other m88k configurations.
6870 @item -mversion-03.00
6871 @opindex mversion-03.00
6872 This option is obsolete, and is ignored.
6873 @c ??? which asm syntax better for GAS? option there too?
6875 @item -mno-check-zero-division
6876 @itemx -mcheck-zero-division
6877 @opindex mno-check-zero-division
6878 @opindex mcheck-zero-division
6879 @cindex zero division on 88k
6880 Do, or don't, generate code to guarantee that integer division by
6881 zero will be detected. By default, detection is guaranteed.
6883 Some models of the MC88100 processor fail to trap upon integer
6884 division by zero under certain conditions. By default, when
6885 compiling code that might be run on such a processor, GCC
6886 generates code that explicitly checks for zero-valued divisors
6887 and traps with exception number 503 when one is detected. Use of
6888 @option{-mno-check-zero-division} suppresses such checking for code
6889 generated to run on an MC88100 processor.
6891 GCC assumes that the MC88110 processor correctly detects all instances
6892 of integer division by zero. When @option{-m88110} is specified, no
6893 explicit checks for zero-valued divisors are generated, and both
6894 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
6897 @item -muse-div-instruction
6898 @opindex muse-div-instruction
6899 @cindex divide instruction, 88k
6900 Use the div instruction for signed integer division on the
6901 MC88100 processor. By default, the div instruction is not used.
6903 On the MC88100 processor the signed integer division instruction
6904 div) traps to the operating system on a negative operand. The
6905 operating system transparently completes the operation, but at a
6906 large cost in execution time. By default, when compiling code
6907 that might be run on an MC88100 processor, GCC emulates signed
6908 integer division using the unsigned integer division instruction
6909 divu), thereby avoiding the large penalty of a trap to the
6910 operating system. Such emulation has its own, smaller, execution
6911 cost in both time and space. To the extent that your code's
6912 important signed integer division operations are performed on two
6913 nonnegative operands, it may be desirable to use the div
6914 instruction directly.
6916 On the MC88110 processor the div instruction (also known as the
6917 divs instruction) processes negative operands without trapping to
6918 the operating system. When @option{-m88110} is specified,
6919 @option{-muse-div-instruction} is ignored, and the div instruction is used
6920 for signed integer division.
6922 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
6923 particular, the behavior of such a division with and without
6924 @option{-muse-div-instruction} may differ.
6926 @item -mtrap-large-shift
6927 @itemx -mhandle-large-shift
6928 @opindex mtrap-large-shift
6929 @opindex mhandle-large-shift
6930 @cindex bit shift overflow (88k)
6931 @cindex large bit shifts (88k)
6932 Include code to detect bit-shifts of more than 31 bits; respectively,
6933 trap such shifts or emit code to handle them properly. By default GCC
6934 makes no special provision for large bit shifts.
6936 @item -mwarn-passed-structs
6937 @opindex mwarn-passed-structs
6938 @cindex structure passing (88k)
6939 Warn when a function passes a struct as an argument or result.
6940 Structure-passing conventions have changed during the evolution of the C
6941 language, and are often the source of portability problems. By default,
6942 GCC issues no such warning.
6945 @c break page here to avoid unsightly interparagraph stretch.
6949 @node RS/6000 and PowerPC Options
6950 @subsection IBM RS/6000 and PowerPC Options
6951 @cindex RS/6000 and PowerPC Options
6952 @cindex IBM RS/6000 and PowerPC Options
6954 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6962 @itemx -mpowerpc-gpopt
6963 @itemx -mno-powerpc-gpopt
6964 @itemx -mpowerpc-gfxopt
6965 @itemx -mno-powerpc-gfxopt
6967 @itemx -mno-powerpc64
6973 @opindex mno-powerpc
6974 @opindex mpowerpc-gpopt
6975 @opindex mno-powerpc-gpopt
6976 @opindex mpowerpc-gfxopt
6977 @opindex mno-powerpc-gfxopt
6979 @opindex mno-powerpc64
6980 GCC supports two related instruction set architectures for the
6981 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6982 instructions supported by the @samp{rios} chip set used in the original
6983 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6984 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6985 the IBM 4xx microprocessors.
6987 Neither architecture is a subset of the other. However there is a
6988 large common subset of instructions supported by both. An MQ
6989 register is included in processors supporting the POWER architecture.
6991 You use these options to specify which instructions are available on the
6992 processor you are using. The default value of these options is
6993 determined when configuring GCC@. Specifying the
6994 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6995 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6996 rather than the options listed above.
6998 The @option{-mpower} option allows GCC to generate instructions that
6999 are found only in the POWER architecture and to use the MQ register.
7000 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
7001 to generate instructions that are present in the POWER2 architecture but
7002 not the original POWER architecture.
7004 The @option{-mpowerpc} option allows GCC to generate instructions that
7005 are found only in the 32-bit subset of the PowerPC architecture.
7006 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
7007 GCC to use the optional PowerPC architecture instructions in the
7008 General Purpose group, including floating-point square root. Specifying
7009 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
7010 use the optional PowerPC architecture instructions in the Graphics
7011 group, including floating-point select.
7013 The @option{-mpowerpc64} option allows GCC to generate the additional
7014 64-bit instructions that are found in the full PowerPC64 architecture
7015 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
7016 @option{-mno-powerpc64}.
7018 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
7019 will use only the instructions in the common subset of both
7020 architectures plus some special AIX common-mode calls, and will not use
7021 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
7022 permits GCC to use any instruction from either architecture and to
7023 allow use of the MQ register; specify this for the Motorola MPC601.
7025 @item -mnew-mnemonics
7026 @itemx -mold-mnemonics
7027 @opindex mnew-mnemonics
7028 @opindex mold-mnemonics
7029 Select which mnemonics to use in the generated assembler code. With
7030 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
7031 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
7032 assembler mnemonics defined for the POWER architecture. Instructions
7033 defined in only one architecture have only one mnemonic; GCC uses that
7034 mnemonic irrespective of which of these options is specified.
7036 GCC defaults to the mnemonics appropriate for the architecture in
7037 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
7038 value of these option. Unless you are building a cross-compiler, you
7039 should normally not specify either @option{-mnew-mnemonics} or
7040 @option{-mold-mnemonics}, but should instead accept the default.
7042 @item -mcpu=@var{cpu_type}
7044 Set architecture type, register usage, choice of mnemonics, and
7045 instruction scheduling parameters for machine type @var{cpu_type}.
7046 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
7047 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
7048 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
7049 @samp{630}, @samp{740}, @samp{7400}, @samp{7450}, @samp{750},
7050 @samp{power}, @samp{power2}, @samp{powerpc}, @samp{403}, @samp{505},
7051 @samp{801}, @samp{821}, @samp{823}, and @samp{860} and @samp{common}.
7053 @option{-mcpu=common} selects a completely generic processor. Code
7054 generated under this option will run on any POWER or PowerPC processor.
7055 GCC will use only the instructions in the common subset of both
7056 architectures, and will not use the MQ register. GCC assumes a generic
7057 processor model for scheduling purposes.
7059 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
7060 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
7061 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
7062 types, with an appropriate, generic processor model assumed for
7063 scheduling purposes.
7065 The other options specify a specific processor. Code generated under
7066 those options will run best on that processor, and may not run at all on
7069 The @option{-mcpu} options automatically enable or disable other
7070 @option{-m} options as follows:
7074 @option{-mno-power}, @option{-mno-powerpc}
7081 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
7096 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
7099 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
7104 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
7107 @item -mtune=@var{cpu_type}
7109 Set the instruction scheduling parameters for machine type
7110 @var{cpu_type}, but do not set the architecture type, register usage, or
7111 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
7112 values for @var{cpu_type} are used for @option{-mtune} as for
7113 @option{-mcpu}. If both are specified, the code generated will use the
7114 architecture, registers, and mnemonics set by @option{-mcpu}, but the
7115 scheduling parameters set by @option{-mtune}.
7120 @opindex mno-altivec
7121 These switches enable or disable the use of built-in functions that
7122 allow access to the AltiVec instruction set. You may also need to set
7123 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
7128 Extend the current ABI with SPE ABI extensions. This does not change
7129 the default ABI, instead it adds the SPE ABI extensions to the current
7133 @opindex mabi=no-spe
7134 Disable Booke SPE ABI extensions for the current ABI.
7136 @item -misel=@var{yes/no}
7139 This switch enables or disables the generation of ISEL instructions.
7141 @item -mspe=@var{yes/no}
7144 This switch enables or disables the generation of SPE simd
7147 @item -mfloat-gprs=@var{yes/no}
7149 @opindex mfloat-gprs
7150 This switch enables or disables the generation of floating point
7151 operations on the general purpose registers for architectures that
7152 support it. This option is currently only available on the MPC8540.
7155 @itemx -mno-fp-in-toc
7156 @itemx -mno-sum-in-toc
7157 @itemx -mminimal-toc
7159 @opindex mno-fp-in-toc
7160 @opindex mno-sum-in-toc
7161 @opindex mminimal-toc
7162 Modify generation of the TOC (Table Of Contents), which is created for
7163 every executable file. The @option{-mfull-toc} option is selected by
7164 default. In that case, GCC will allocate at least one TOC entry for
7165 each unique non-automatic variable reference in your program. GCC
7166 will also place floating-point constants in the TOC@. However, only
7167 16,384 entries are available in the TOC@.
7169 If you receive a linker error message that saying you have overflowed
7170 the available TOC space, you can reduce the amount of TOC space used
7171 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
7172 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
7173 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
7174 generate code to calculate the sum of an address and a constant at
7175 run-time instead of putting that sum into the TOC@. You may specify one
7176 or both of these options. Each causes GCC to produce very slightly
7177 slower and larger code at the expense of conserving TOC space.
7179 If you still run out of space in the TOC even when you specify both of
7180 these options, specify @option{-mminimal-toc} instead. This option causes
7181 GCC to make only one TOC entry for every file. When you specify this
7182 option, GCC will produce code that is slower and larger but which
7183 uses extremely little TOC space. You may wish to use this option
7184 only on files that contain less frequently executed code.
7190 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
7191 @code{long} type, and the infrastructure needed to support them.
7192 Specifying @option{-maix64} implies @option{-mpowerpc64} and
7193 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
7194 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
7199 @opindex mno-xl-call
7200 On AIX, pass floating-point arguments to prototyped functions beyond the
7201 register save area (RSA) on the stack in addition to argument FPRs. The
7202 AIX calling convention was extended but not initially documented to
7203 handle an obscure K&R C case of calling a function that takes the
7204 address of its arguments with fewer arguments than declared. AIX XL
7205 compilers access floating point arguments which do not fit in the
7206 RSA from the stack when a subroutine is compiled without
7207 optimization. Because always storing floating-point arguments on the
7208 stack is inefficient and rarely needed, this option is not enabled by
7209 default and only is necessary when calling subroutines compiled by AIX
7210 XL compilers without optimization.
7214 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
7215 application written to use message passing with special startup code to
7216 enable the application to run. The system must have PE installed in the
7217 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
7218 must be overridden with the @option{-specs=} option to specify the
7219 appropriate directory location. The Parallel Environment does not
7220 support threads, so the @option{-mpe} option and the @option{-pthread}
7221 option are incompatible.
7223 @item -malign-natural
7224 @itemx -malign-power
7225 @opindex malign-natural
7226 @opindex malign-power
7227 On AIX, Darwin, and 64-bit PowerPC Linux, the option
7228 @option{-malign-natural} overrides the ABI-defined alignment of larger
7229 types, such as floating-point doubles, on their natural size-based boundary.
7230 The option @option{-malign-power} instructs GCC to follow the ABI-specified
7231 alignment rules. GCC defaults to the standard alignment defined in the ABI.
7235 @opindex msoft-float
7236 @opindex mhard-float
7237 Generate code that does not use (uses) the floating-point register set.
7238 Software floating point emulation is provided if you use the
7239 @option{-msoft-float} option, and pass the option to GCC when linking.
7242 @itemx -mno-multiple
7244 @opindex mno-multiple
7245 Generate code that uses (does not use) the load multiple word
7246 instructions and the store multiple word instructions. These
7247 instructions are generated by default on POWER systems, and not
7248 generated on PowerPC systems. Do not use @option{-mmultiple} on little
7249 endian PowerPC systems, since those instructions do not work when the
7250 processor is in little endian mode. The exceptions are PPC740 and
7251 PPC750 which permit the instructions usage in little endian mode.
7257 Generate code that uses (does not use) the load string instructions
7258 and the store string word instructions to save multiple registers and
7259 do small block moves. These instructions are generated by default on
7260 POWER systems, and not generated on PowerPC systems. Do not use
7261 @option{-mstring} on little endian PowerPC systems, since those
7262 instructions do not work when the processor is in little endian mode.
7263 The exceptions are PPC740 and PPC750 which permit the instructions
7264 usage in little endian mode.
7270 Generate code that uses (does not use) the load or store instructions
7271 that update the base register to the address of the calculated memory
7272 location. These instructions are generated by default. If you use
7273 @option{-mno-update}, there is a small window between the time that the
7274 stack pointer is updated and the address of the previous frame is
7275 stored, which means code that walks the stack frame across interrupts or
7276 signals may get corrupted data.
7279 @itemx -mno-fused-madd
7280 @opindex mfused-madd
7281 @opindex mno-fused-madd
7282 Generate code that uses (does not use) the floating point multiply and
7283 accumulate instructions. These instructions are generated by default if
7284 hardware floating is used.
7286 @item -mno-bit-align
7288 @opindex mno-bit-align
7290 On System V.4 and embedded PowerPC systems do not (do) force structures
7291 and unions that contain bit-fields to be aligned to the base type of the
7294 For example, by default a structure containing nothing but 8
7295 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7296 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7297 the structure would be aligned to a 1 byte boundary and be one byte in
7300 @item -mno-strict-align
7301 @itemx -mstrict-align
7302 @opindex mno-strict-align
7303 @opindex mstrict-align
7304 On System V.4 and embedded PowerPC systems do not (do) assume that
7305 unaligned memory references will be handled by the system.
7308 @itemx -mno-relocatable
7309 @opindex mrelocatable
7310 @opindex mno-relocatable
7311 On embedded PowerPC systems generate code that allows (does not allow)
7312 the program to be relocated to a different address at runtime. If you
7313 use @option{-mrelocatable} on any module, all objects linked together must
7314 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7316 @item -mrelocatable-lib
7317 @itemx -mno-relocatable-lib
7318 @opindex mrelocatable-lib
7319 @opindex mno-relocatable-lib
7320 On embedded PowerPC systems generate code that allows (does not allow)
7321 the program to be relocated to a different address at runtime. Modules
7322 compiled with @option{-mrelocatable-lib} can be linked with either modules
7323 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7324 with modules compiled with the @option{-mrelocatable} options.
7330 On System V.4 and embedded PowerPC systems do not (do) assume that
7331 register 2 contains a pointer to a global area pointing to the addresses
7332 used in the program.
7335 @itemx -mlittle-endian
7337 @opindex mlittle-endian
7338 On System V.4 and embedded PowerPC systems compile code for the
7339 processor in little endian mode. The @option{-mlittle-endian} option is
7340 the same as @option{-mlittle}.
7345 @opindex mbig-endian
7346 On System V.4 and embedded PowerPC systems compile code for the
7347 processor in big endian mode. The @option{-mbig-endian} option is
7348 the same as @option{-mbig}.
7350 @item -mdynamic-no-pic
7351 @opindex mdynamic-no-pic
7352 On Darwin and Mac OS X systems, compile code so that it is not
7353 relocatable, but that its external references are relocatable. The
7354 resulting code is suitable for applications, but not shared
7359 On System V.4 and embedded PowerPC systems compile code using calling
7360 conventions that adheres to the March 1995 draft of the System V
7361 Application Binary Interface, PowerPC processor supplement. This is the
7362 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7364 @item -mcall-sysv-eabi
7365 @opindex mcall-sysv-eabi
7366 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7368 @item -mcall-sysv-noeabi
7369 @opindex mcall-sysv-noeabi
7370 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7372 @item -mcall-solaris
7373 @opindex mcall-solaris
7374 On System V.4 and embedded PowerPC systems compile code for the Solaris
7378 @opindex mcall-linux
7379 On System V.4 and embedded PowerPC systems compile code for the
7380 Linux-based GNU system.
7384 On System V.4 and embedded PowerPC systems compile code for the
7385 Hurd-based GNU system.
7388 @opindex mcall-netbsd
7389 On System V.4 and embedded PowerPC systems compile code for the
7390 NetBSD operating system.
7392 @item -maix-struct-return
7393 @opindex maix-struct-return
7394 Return all structures in memory (as specified by the AIX ABI)@.
7396 @item -msvr4-struct-return
7397 @opindex msvr4-struct-return
7398 Return structures smaller than 8 bytes in registers (as specified by the
7402 @opindex mabi=altivec
7403 Extend the current ABI with AltiVec ABI extensions. This does not
7404 change the default ABI, instead it adds the AltiVec ABI extensions to
7407 @item -mabi=no-altivec
7408 @opindex mabi=no-altivec
7409 Disable AltiVec ABI extensions for the current ABI.
7412 @itemx -mno-prototype
7414 @opindex mno-prototype
7415 On System V.4 and embedded PowerPC systems assume that all calls to
7416 variable argument functions are properly prototyped. Otherwise, the
7417 compiler must insert an instruction before every non prototyped call to
7418 set or clear bit 6 of the condition code register (@var{CR}) to
7419 indicate whether floating point values were passed in the floating point
7420 registers in case the function takes a variable arguments. With
7421 @option{-mprototype}, only calls to prototyped variable argument functions
7422 will set or clear the bit.
7426 On embedded PowerPC systems, assume that the startup module is called
7427 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7428 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7433 On embedded PowerPC systems, assume that the startup module is called
7434 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7439 On embedded PowerPC systems, assume that the startup module is called
7440 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7444 @opindex myellowknife
7445 On embedded PowerPC systems, assume that the startup module is called
7446 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7451 On System V.4 and embedded PowerPC systems, specify that you are
7452 compiling for a VxWorks system.
7456 Specify that you are compiling for the WindISS simulation environment.
7460 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7461 header to indicate that @samp{eabi} extended relocations are used.
7467 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7468 Embedded Applications Binary Interface (eabi) which is a set of
7469 modifications to the System V.4 specifications. Selecting @option{-meabi}
7470 means that the stack is aligned to an 8 byte boundary, a function
7471 @code{__eabi} is called to from @code{main} to set up the eabi
7472 environment, and the @option{-msdata} option can use both @code{r2} and
7473 @code{r13} to point to two separate small data areas. Selecting
7474 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7475 do not call an initialization function from @code{main}, and the
7476 @option{-msdata} option will only use @code{r13} to point to a single
7477 small data area. The @option{-meabi} option is on by default if you
7478 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7481 @opindex msdata=eabi
7482 On System V.4 and embedded PowerPC systems, put small initialized
7483 @code{const} global and static data in the @samp{.sdata2} section, which
7484 is pointed to by register @code{r2}. Put small initialized
7485 non-@code{const} global and static data in the @samp{.sdata} section,
7486 which is pointed to by register @code{r13}. Put small uninitialized
7487 global and static data in the @samp{.sbss} section, which is adjacent to
7488 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7489 incompatible with the @option{-mrelocatable} option. The
7490 @option{-msdata=eabi} option also sets the @option{-memb} option.
7493 @opindex msdata=sysv
7494 On System V.4 and embedded PowerPC systems, put small global and static
7495 data in the @samp{.sdata} section, which is pointed to by register
7496 @code{r13}. Put small uninitialized global and static data in the
7497 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7498 The @option{-msdata=sysv} option is incompatible with the
7499 @option{-mrelocatable} option.
7501 @item -msdata=default
7503 @opindex msdata=default
7505 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7506 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7507 same as @option{-msdata=sysv}.
7510 @opindex msdata-data
7511 On System V.4 and embedded PowerPC systems, put small global and static
7512 data in the @samp{.sdata} section. Put small uninitialized global and
7513 static data in the @samp{.sbss} section. Do not use register @code{r13}
7514 to address small data however. This is the default behavior unless
7515 other @option{-msdata} options are used.
7519 @opindex msdata=none
7521 On embedded PowerPC systems, put all initialized global and static data
7522 in the @samp{.data} section, and all uninitialized data in the
7523 @samp{.bss} section.
7527 @cindex smaller data references (PowerPC)
7528 @cindex .sdata/.sdata2 references (PowerPC)
7529 On embedded PowerPC systems, put global and static items less than or
7530 equal to @var{num} bytes into the small data or bss sections instead of
7531 the normal data or bss section. By default, @var{num} is 8. The
7532 @option{-G @var{num}} switch is also passed to the linker.
7533 All modules should be compiled with the same @option{-G @var{num}} value.
7536 @itemx -mno-regnames
7538 @opindex mno-regnames
7539 On System V.4 and embedded PowerPC systems do (do not) emit register
7540 names in the assembly language output using symbolic forms.
7543 @itemx -mno-longcall
7545 @opindex mno-longcall
7546 Default to making all function calls via pointers, so that functions
7547 which reside further than 64 megabytes (67,108,864 bytes) from the
7548 current location can be called. This setting can be overridden by the
7549 @code{shortcall} function attribute, or by @code{#pragma longcall(0)}.
7551 Some linkers are capable of detecting out-of-range calls and generating
7552 glue code on the fly. On these systems, long calls are unnecessary and
7553 generate slower code. As of this writing, the AIX linker can do this,
7554 as can the GNU linker for PowerPC/64. It is planned to add this feature
7555 to the GNU linker for 32-bit PowerPC systems as well.
7557 In the future, we may cause GCC to ignore all longcall specifications
7558 when the linker is known to generate glue.
7562 Adds support for multithreading with the @dfn{pthreads} library.
7563 This option sets flags for both the preprocessor and linker.
7567 @node Darwin Options
7568 @subsection Darwin Options
7569 @cindex Darwin options
7571 These options are defined for all architectures running the Darwin operating
7572 system. They are useful for compatibility with other Mac OS compilers.
7577 Loads all members of static archive libraries.
7578 See man ld(1) for more information.
7580 @item -arch_errors_fatal
7581 @opindex arch_errors_fatal
7582 Cause the errors having to do with files that have the wrong architecture
7586 @opindex bind_at_load
7587 Causes the output file to be marked such that the dynamic linker will
7588 bind all undefined references when the file is loaded or launched.
7592 Produce a Mach-o bundle format file.
7593 See man ld(1) for more information.
7595 @item -bundle_loader @var{executable}
7596 @opindex bundle_loader
7597 This specifies the @var{executable} that will be loading the build
7598 output file being linked. See man ld(1) for more information.
7600 @item -allowable_client @var{client_name}
7604 @item -compatibility_version
7605 @item -current_version
7606 @item -dependency-file
7608 @item -dylinker_install_name
7611 @item -exported_symbols_list
7613 @item -flat_namespace
7614 @item -force_cpusubtype_ALL
7615 @item -force_flat_namespace
7616 @item -headerpad_max_install_names
7620 @item -keep_private_externs
7622 @item -multiply_defined
7623 @item -multiply_defined_unused
7627 @item -noseglinkedit
7628 @item -pagezero_size
7630 @item -prebind_all_twolevel_modules
7631 @item -private_bundle
7632 @item -read_only_relocs
7634 @item -sectobjectsymbols
7638 @item -sectobjectsymbols
7640 @item -seg_addr_table
7641 @item -seg_addr_table_filename
7644 @item -segs_read_only_addr
7645 @item -segs_read_write_addr
7646 @item -single_module
7650 @item -twolevel_namespace
7653 @item -unexported_symbols_list
7654 @item -weak_reference_mismatches
7657 @opindex allowable_client
7659 @opindex client_name
7660 @opindex compatibility_version
7661 @opindex current_version
7662 @opindex dependency-file
7664 @opindex dylinker_install_name
7667 @opindex exported_symbols_list
7669 @opindex flat_namespace
7670 @opindex force_cpusubtype_ALL
7671 @opindex force_flat_namespace
7672 @opindex headerpad_max_install_names
7675 @opindex install_name
7676 @opindex keep_private_externs
7677 @opindex multi_module
7678 @opindex multiply_defined
7679 @opindex multiply_defined_unused
7681 @opindex nomultidefs
7683 @opindex noseglinkedit
7684 @opindex pagezero_size
7686 @opindex prebind_all_twolevel_modules
7687 @opindex private_bundle
7688 @opindex read_only_relocs
7690 @opindex sectobjectsymbols
7694 @opindex sectobjectsymbols
7696 @opindex seg_addr_table
7697 @opindex seg_addr_table_filename
7698 @opindex seglinkedit
7700 @opindex segs_read_only_addr
7701 @opindex segs_read_write_addr
7702 @opindex single_module
7704 @opindex sub_library
7705 @opindex sub_umbrella
7706 @opindex twolevel_namespace
7709 @opindex unexported_symbols_list
7710 @opindex weak_reference_mismatches
7711 @opindex whatsloaded
7713 This options are available for Darwin linker. Darwin linker man page
7714 describes them in detail.
7719 @subsection IBM RT Options
7721 @cindex IBM RT options
7723 These @samp{-m} options are defined for the IBM RT PC:
7727 @opindex min-line-mul
7728 Use an in-line code sequence for integer multiplies. This is the
7731 @item -mcall-lib-mul
7732 @opindex mcall-lib-mul
7733 Call @code{lmul$$} for integer multiples.
7735 @item -mfull-fp-blocks
7736 @opindex mfull-fp-blocks
7737 Generate full-size floating point data blocks, including the minimum
7738 amount of scratch space recommended by IBM@. This is the default.
7740 @item -mminimum-fp-blocks
7741 @opindex mminimum-fp-blocks
7742 Do not include extra scratch space in floating point data blocks. This
7743 results in smaller code, but slower execution, since scratch space must
7744 be allocated dynamically.
7746 @cindex @file{stdarg.h} and RT PC
7747 @item -mfp-arg-in-fpregs
7748 @opindex mfp-arg-in-fpregs
7749 Use a calling sequence incompatible with the IBM calling convention in
7750 which floating point arguments are passed in floating point registers.
7751 Note that @code{stdarg.h} will not work with floating point operands
7752 if this option is specified.
7754 @item -mfp-arg-in-gregs
7755 @opindex mfp-arg-in-gregs
7756 Use the normal calling convention for floating point arguments. This is
7759 @item -mhc-struct-return
7760 @opindex mhc-struct-return
7761 Return structures of more than one word in memory, rather than in a
7762 register. This provides compatibility with the MetaWare HighC (hc)
7763 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7764 with the Portable C Compiler (pcc).
7766 @item -mnohc-struct-return
7767 @opindex mnohc-struct-return
7768 Return some structures of more than one word in registers, when
7769 convenient. This is the default. For compatibility with the
7770 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7771 option @option{-mhc-struct-return}.
7775 @subsection MIPS Options
7776 @cindex MIPS options
7778 These @samp{-m} options are defined for the MIPS family of computers:
7782 @item -march=@var{arch}
7784 Generate code that will run on @var{arch}, which can be the name of a
7785 generic MIPS ISA, or the name of a particular processor.
7787 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
7788 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
7789 The processor names are:
7790 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
7792 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
7793 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
7796 @samp{vr4100}, @samp{vr4300}, and @samp{vr5000}.
7797 The special value @samp{from-abi} selects the
7798 most compatible architecture for the selected ABI (that is,
7799 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
7801 In processor names, a final @samp{000} can be abbreviated as @samp{k}
7802 (for example, @samp{-march=r2k}). Prefixes are optional, and
7803 @samp{vr} may be written @samp{r}.
7805 GCC defines two macros based on the value of this option. The first
7806 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
7807 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
7808 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
7809 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
7810 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
7812 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
7813 above. In other words, it will have the full prefix and will not
7814 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
7815 the macro names the resolved architecture (either @samp{"mips1"} or
7816 @samp{"mips3"}). It names the default architecture when no
7817 @option{-march} option is given.
7819 @item -mtune=@var{arch}
7821 Optimize for @var{arch}. Among other things, this option controls
7822 the way instructions are scheduled, and the perceived cost of arithmetic
7823 operations. The list of @var{arch} values is the same as for
7826 When this option is not used, GCC will optimize for the processor
7827 specified by @option{-march}. By using @option{-march} and
7828 @option{-mtune} together, it is possible to generate code that will
7829 run on a family of processors, but optimize the code for one
7830 particular member of that family.
7832 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
7833 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
7834 @samp{-march} ones described above.
7838 Equivalent to @samp{-march=mips1}.
7842 Equivalent to @samp{-march=mips2}.
7846 Equivalent to @samp{-march=mips3}.
7850 Equivalent to @samp{-march=mips4}.
7854 Equivalent to @samp{-march=mips32}.
7858 Equivalent to @samp{-march=mips32r2}.
7862 Equivalent to @samp{-march=mips64}.
7865 @itemx -mno-fused-madd
7866 @opindex mfused-madd
7867 @opindex mno-fused-madd
7868 Generate code that uses (does not use) the floating point multiply and
7869 accumulate instructions, when they are available. These instructions
7870 are generated by default if they are available, but this may be
7871 undesirable if the extra precision causes problems or on certain chips
7872 in the mode where denormals are rounded to zero where denormals
7873 generated by multiply and accumulate instructions cause exceptions
7878 Assume that floating point registers are 32 bits wide.
7882 Assume that floating point registers are 64 bits wide.
7886 Assume that general purpose registers are 32 bits wide.
7890 Assume that general purpose registers are 64 bits wide.
7894 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
7895 explanation of the default, and the width of pointers.
7899 Force long types to be 64 bits wide. See @option{-mlong32} for an
7900 explanation of the default, and the width of pointers.
7904 Force long, int, and pointer types to be 32 bits wide.
7906 The default size of ints, longs and pointers depends on the ABI@. All
7907 the supported ABIs use 32-bit ints. The n64 ABI uses 64-bit longs, as
7908 does the 64-bit Cygnus EABI; the others use 32-bit longs. Pointers
7909 are the same size as longs, or the same size as integer registers,
7910 whichever is smaller.
7924 Generate code for the given ABI@.
7926 Note that there are two embedded ABIs: @option{-mabi=eabi}
7927 selects the one defined by Cygnus while @option{-meabi=meabi}
7928 selects the one defined by MIPS@. Both these ABIs have
7929 32-bit and 64-bit variants. Normally, GCC will generate
7930 64-bit code when you select a 64-bit architecture, but you
7931 can use @option{-mgp32} to get 32-bit code instead.
7933 @item -mabi-fake-default
7934 @opindex mabi-fake-default
7935 You don't want to know what this option does. No, really. I mean
7936 it. Move on to the next option.
7938 What? You're still here? Oh, well@enddots{} Ok, here's the deal. GCC
7939 wants the default set of options to get the root of the multilib tree,
7940 and the shared library SONAMEs without any multilib-indicating
7941 suffixes. This is not convenience for @samp{mips64-linux-gnu}, since
7942 we want to default to the N32 ABI, while still being binary-compatible
7943 with @samp{mips-linux-gnu} if you stick to the O32 ABI@. Being
7944 binary-compatible means shared libraries should have the same SONAMEs,
7945 and libraries should live in the same location. Having O32 libraries
7946 in a sub-directory named say @file{o32} is not acceptable.
7948 So we trick GCC into believing that O32 is the default ABI, except
7949 that we override the default with some internal command-line
7950 processing magic. Problem is, if we stopped at that, and you then
7951 created a multilib-aware package that used the output of @command{gcc
7952 -print-multi-lib} to decide which multilibs to build, and how, and
7953 you'd find yourself in an awkward situation when you found out that
7954 some of the options listed ended up mapping to the same multilib, and
7955 none of your libraries was actually built for the multilib that
7956 @option{-print-multi-lib} claims to be the default. So we added this
7957 option that disables the default switcher, falling back to GCC's
7958 original notion of the default library. Confused yet?
7960 For short: don't ever use this option, unless you find it in the list
7961 of additional options to be used when building for multilibs, in the
7962 output of @option{gcc -print-multi-lib}.
7966 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
7967 add normal debug information. This is the default for all
7968 platforms except for the OSF/1 reference platform, using the OSF/rose
7969 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
7970 switches are used, the @file{mips-tfile} program will encapsulate the
7971 stabs within MIPS ECOFF@.
7975 Generate code for the GNU assembler. This is the default on the OSF/1
7976 reference platform, using the OSF/rose object format. Also, this is
7977 the default if the configure option @option{--with-gnu-as} is used.
7979 @item -msplit-addresses
7980 @itemx -mno-split-addresses
7981 @opindex msplit-addresses
7982 @opindex mno-split-addresses
7983 Generate code to load the high and low parts of address constants separately.
7984 This allows GCC to optimize away redundant loads of the high order
7985 bits of addresses. This optimization requires GNU as and GNU ld.
7986 This optimization is enabled by default for some embedded targets where
7987 GNU as and GNU ld are standard.
7993 The @option{-mrnames} switch says to output code using the MIPS software
7994 names for the registers, instead of the hardware names (ie, @var{a0}
7995 instead of @var{$4}). The only known assembler that supports this option
7996 is the Algorithmics assembler.
8002 The @option{-mmemcpy} switch makes all block moves call the appropriate
8003 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
8004 generating inline code.
8007 @itemx -mno-mips-tfile
8008 @opindex mmips-tfile
8009 @opindex mno-mips-tfile
8010 The @option{-mno-mips-tfile} switch causes the compiler not
8011 postprocess the object file with the @file{mips-tfile} program,
8012 after the MIPS assembler has generated it to add debug support. If
8013 @file{mips-tfile} is not run, then no local variables will be
8014 available to the debugger. In addition, @file{stage2} and
8015 @file{stage3} objects will have the temporary file names passed to the
8016 assembler embedded in the object file, which means the objects will
8017 not compare the same. The @option{-mno-mips-tfile} switch should only
8018 be used when there are bugs in the @file{mips-tfile} program that
8019 prevents compilation.
8022 @opindex msoft-float
8023 Generate output containing library calls for floating point.
8024 @strong{Warning:} the requisite libraries are not part of GCC@.
8025 Normally the facilities of the machine's usual C compiler are used, but
8026 this can't be done directly in cross-compilation. You must make your
8027 own arrangements to provide suitable library functions for
8031 @opindex mhard-float
8032 Generate output containing floating point instructions. This is the
8033 default if you use the unmodified sources.
8036 @itemx -mno-abicalls
8038 @opindex mno-abicalls
8039 Emit (or do not emit) the pseudo operations @samp{.abicalls},
8040 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
8041 position independent code.
8044 @itemx -mno-long-calls
8045 @opindex mlong-calls
8046 @opindex mno-long-calls
8047 Do all calls with the @samp{JALR} instruction, which requires
8048 loading up a function's address into a register before the call.
8049 You need to use this switch, if you call outside of the current
8050 512 megabyte segment to functions that are not through pointers.
8052 @item -membedded-pic
8053 @itemx -mno-embedded-pic
8054 @opindex membedded-pic
8055 @opindex mno-embedded-pic
8056 Generate PIC code suitable for some embedded systems. All calls are
8057 made using PC relative address, and all data is addressed using the $gp
8058 register. No more than 65536 bytes of global data may be used. This
8059 requires GNU as and GNU ld which do most of the work. This currently
8060 only works on targets which use ECOFF; it does not work with ELF@.
8062 @item -membedded-data
8063 @itemx -mno-embedded-data
8064 @opindex membedded-data
8065 @opindex mno-embedded-data
8066 Allocate variables to the read-only data section first if possible, then
8067 next in the small data section if possible, otherwise in data. This gives
8068 slightly slower code than the default, but reduces the amount of RAM required
8069 when executing, and thus may be preferred for some embedded systems.
8071 @item -muninit-const-in-rodata
8072 @itemx -mno-uninit-const-in-rodata
8073 @opindex muninit-const-in-rodata
8074 @opindex mno-uninit-const-in-rodata
8075 When used together with @option{-membedded-data}, it will always store uninitialized
8076 const variables in the read-only data section.
8078 @item -msingle-float
8079 @itemx -mdouble-float
8080 @opindex msingle-float
8081 @opindex mdouble-float
8082 The @option{-msingle-float} switch tells gcc to assume that the floating
8083 point coprocessor only supports single precision operations, as on the
8084 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
8085 double precision operations. This is the default.
8091 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
8092 as on the @samp{r4650} chip.
8096 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
8097 @option{-mcpu=r4650}.
8103 Enable 16-bit instructions.
8107 Use the entry and exit pseudo ops. This option can only be used with
8112 Compile code for the processor in little endian mode.
8113 The requisite libraries are assumed to exist.
8117 Compile code for the processor in big endian mode.
8118 The requisite libraries are assumed to exist.
8122 @cindex smaller data references (MIPS)
8123 @cindex gp-relative references (MIPS)
8124 Put global and static items less than or equal to @var{num} bytes into
8125 the small data or bss sections instead of the normal data or bss
8126 section. This allows the assembler to emit one word memory reference
8127 instructions based on the global pointer (@var{gp} or @var{$28}),
8128 instead of the normal two words used. By default, @var{num} is 8 when
8129 the MIPS assembler is used, and 0 when the GNU assembler is used. The
8130 @option{-G @var{num}} switch is also passed to the assembler and linker.
8131 All modules should be compiled with the same @option{-G @var{num}}
8136 Tell the MIPS assembler to not run its preprocessor over user
8137 assembler files (with a @samp{.s} suffix) when assembling them.
8141 Pass an option to gas which will cause nops to be inserted if
8142 the read of the destination register of an mfhi or mflo instruction
8143 occurs in the following two instructions.
8147 Do not include the default crt0.
8149 @item -mflush-func=@var{func}
8150 @itemx -mno-flush-func
8151 @opindex mflush-func
8152 Specifies the function to call to flush the I and D caches, or to not
8153 call any such function. If called, the function must take the same
8154 arguments as the common @code{_flush_func()}, that is, the address of the
8155 memory range for which the cache is being flushed, the size of the
8156 memory range, and the number 3 (to flush both caches). The default
8157 depends on the target gcc was configured for, but commonly is either
8158 @samp{_flush_func} or @samp{__cpu_flush}.
8160 @item -mbranch-likely
8161 @itemx -mno-branch-likely
8162 @opindex mbranch-likely
8163 @opindex mno-branch-likely
8164 Enable or disable use of Branch Likely instructions, regardless of the
8165 default for the selected architecture. By default, Branch Likely
8166 instructions may be generated if they are supported by the selected
8167 architecture. An exception is for the MIPS32 and MIPS64 architectures
8168 and processors which implement those architectures; for those, Branch
8169 Likely instructions will not be generated by default because the MIPS32
8170 and MIPS64 architectures specifically deprecate their use.
8173 @node i386 and x86-64 Options
8174 @subsection Intel 386 and AMD x86-64 Options
8175 @cindex i386 Options
8176 @cindex x86-64 Options
8177 @cindex Intel 386 Options
8178 @cindex AMD x86-64 Options
8180 These @samp{-m} options are defined for the i386 and x86-64 family of
8184 @item -mtune=@var{cpu-type}
8186 Tune to @var{cpu-type} everything applicable about the generated code, except
8187 for the ABI and the set of available instructions. The choices for
8188 @var{cpu-type} are @samp{i386}, @samp{i486}, @samp{i586}, @samp{i686},
8189 @samp{pentium}, @samp{pentium-mmx}, @samp{pentiumpro}, @samp{pentium2},
8190 @samp{pentium3}, @samp{pentium4}, @samp{k6}, @samp{k6-2}, @samp{k6-3},
8191 @samp{athlon}, @samp{athlon-tbird}, @samp{athlon-4}, @samp{athlon-xp},
8192 @samp{athlon-mp}, @samp{winchip-c6}, @samp{winchip2}, @samp{k8}, @samp{c3}
8195 While picking a specific @var{cpu-type} will schedule things appropriately
8196 for that particular chip, the compiler will not generate any code that
8197 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8198 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
8199 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
8200 AMD chips as opposed to the Intel ones.
8202 @item -march=@var{cpu-type}
8204 Generate instructions for the machine type @var{cpu-type}. The choices
8205 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8206 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8208 @item -mcpu=@var{cpu-type}
8210 A deprecated synonym for @option{-mtune}.
8219 @opindex mpentiumpro
8220 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8221 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8222 These synonyms are deprecated.
8224 @item -mfpmath=@var{unit}
8226 generate floating point arithmetics for selected unit @var{unit}. the choices
8231 Use the standard 387 floating point coprocessor present majority of chips and
8232 emulated otherwise. Code compiled with this option will run almost everywhere.
8233 The temporary results are computed in 80bit precision instead of precision
8234 specified by the type resulting in slightly different results compared to most
8235 of other chips. See @option{-ffloat-store} for more detailed description.
8237 This is the default choice for i386 compiler.
8240 Use scalar floating point instructions present in the SSE instruction set.
8241 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8242 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8243 instruction set supports only single precision arithmetics, thus the double and
8244 extended precision arithmetics is still done using 387. Later version, present
8245 only in Pentium4 and the future AMD x86-64 chips supports double precision
8248 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8249 @option{-msse2} switches to enable SSE extensions and make this option
8250 effective. For x86-64 compiler, these extensions are enabled by default.
8252 The resulting code should be considerably faster in majority of cases and avoid
8253 the numerical instability problems of 387 code, but may break some existing
8254 code that expects temporaries to be 80bit.
8256 This is the default choice for x86-64 compiler.
8259 Use all SSE extensions enabled by @option{-msse2} as well as the new
8260 SSE extensions in Prescott New Instructions. @option{-mpni} also
8261 enables 2 builtin functions, @code{__builtin_ia32_monitor} and
8262 @code{__builtin_ia32_mwait}, for new instructions @code{monitor} and
8266 Attempt to utilize both instruction sets at once. This effectively double the
8267 amount of available registers and on chips with separate execution units for
8268 387 and SSE the execution resources too. Use this option with care, as it is
8269 still experimental, because gcc register allocator does not model separate
8270 functional units well resulting in instable performance.
8273 @item -masm=@var{dialect}
8274 @opindex masm=@var{dialect}
8275 Output asm instructions using selected @var{dialect}. Supported choices are
8276 @samp{intel} or @samp{att} (the default one).
8281 @opindex mno-ieee-fp
8282 Control whether or not the compiler uses IEEE floating point
8283 comparisons. These handle correctly the case where the result of a
8284 comparison is unordered.
8287 @opindex msoft-float
8288 Generate output containing library calls for floating point.
8289 @strong{Warning:} the requisite libraries are not part of GCC@.
8290 Normally the facilities of the machine's usual C compiler are used, but
8291 this can't be done directly in cross-compilation. You must make your
8292 own arrangements to provide suitable library functions for
8295 On machines where a function returns floating point results in the 80387
8296 register stack, some floating point opcodes may be emitted even if
8297 @option{-msoft-float} is used.
8299 @item -mno-fp-ret-in-387
8300 @opindex mno-fp-ret-in-387
8301 Do not use the FPU registers for return values of functions.
8303 The usual calling convention has functions return values of types
8304 @code{float} and @code{double} in an FPU register, even if there
8305 is no FPU@. The idea is that the operating system should emulate
8308 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8309 in ordinary CPU registers instead.
8311 @item -mno-fancy-math-387
8312 @opindex mno-fancy-math-387
8313 Some 387 emulators do not support the @code{sin}, @code{cos} and
8314 @code{sqrt} instructions for the 387. Specify this option to avoid
8315 generating those instructions. This option is the default on FreeBSD,
8316 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8317 indicates that the target cpu will always have an FPU and so the
8318 instruction will not need emulation. As of revision 2.6.1, these
8319 instructions are not generated unless you also use the
8320 @option{-funsafe-math-optimizations} switch.
8322 @item -malign-double
8323 @itemx -mno-align-double
8324 @opindex malign-double
8325 @opindex mno-align-double
8326 Control whether GCC aligns @code{double}, @code{long double}, and
8327 @code{long long} variables on a two word boundary or a one word
8328 boundary. Aligning @code{double} variables on a two word boundary will
8329 produce code that runs somewhat faster on a @samp{Pentium} at the
8330 expense of more memory.
8332 @strong{Warning:} if you use the @option{-malign-double} switch,
8333 structures containing the above types will be aligned differently than
8334 the published application binary interface specifications for the 386
8335 and will not be binary compatible with structures in code compiled
8336 without that switch.
8338 @item -m96bit-long-double
8339 @item -m128bit-long-double
8340 @opindex m96bit-long-double
8341 @opindex m128bit-long-double
8342 These switches control the size of @code{long double} type. The i386
8343 application binary interface specifies the size to be 96 bits,
8344 so @option{-m96bit-long-double} is the default in 32 bit mode.
8346 Modern architectures (Pentium and newer) would prefer @code{long double}
8347 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8348 conforming to the ABI, this would not be possible. So specifying a
8349 @option{-m128bit-long-double} will align @code{long double}
8350 to a 16 byte boundary by padding the @code{long double} with an additional
8353 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8354 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8356 Notice that neither of these options enable any extra precision over the x87
8357 standard of 80 bits for a @code{long double}.
8359 @strong{Warning:} if you override the default value for your target ABI, the
8360 structures and arrays containing @code{long double} will change their size as
8361 well as function calling convention for function taking @code{long double}
8362 will be modified. Hence they will not be binary compatible with arrays or
8363 structures in code compiled without that switch.
8367 @itemx -mno-svr3-shlib
8368 @opindex msvr3-shlib
8369 @opindex mno-svr3-shlib
8370 Control whether GCC places uninitialized local variables into the
8371 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8372 into @code{bss}. These options are meaningful only on System V Release 3.
8376 Use a different function-calling convention, in which functions that
8377 take a fixed number of arguments return with the @code{ret} @var{num}
8378 instruction, which pops their arguments while returning. This saves one
8379 instruction in the caller since there is no need to pop the arguments
8382 You can specify that an individual function is called with this calling
8383 sequence with the function attribute @samp{stdcall}. You can also
8384 override the @option{-mrtd} option by using the function attribute
8385 @samp{cdecl}. @xref{Function Attributes}.
8387 @strong{Warning:} this calling convention is incompatible with the one
8388 normally used on Unix, so you cannot use it if you need to call
8389 libraries compiled with the Unix compiler.
8391 Also, you must provide function prototypes for all functions that
8392 take variable numbers of arguments (including @code{printf});
8393 otherwise incorrect code will be generated for calls to those
8396 In addition, seriously incorrect code will result if you call a
8397 function with too many arguments. (Normally, extra arguments are
8398 harmlessly ignored.)
8400 @item -mregparm=@var{num}
8402 Control how many registers are used to pass integer arguments. By
8403 default, no registers are used to pass arguments, and at most 3
8404 registers can be used. You can control this behavior for a specific
8405 function by using the function attribute @samp{regparm}.
8406 @xref{Function Attributes}.
8408 @strong{Warning:} if you use this switch, and
8409 @var{num} is nonzero, then you must build all modules with the same
8410 value, including any libraries. This includes the system libraries and
8413 @item -mpreferred-stack-boundary=@var{num}
8414 @opindex mpreferred-stack-boundary
8415 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8416 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8417 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8418 size (@option{-Os}), in which case the default is the minimum correct
8419 alignment (4 bytes for x86, and 8 bytes for x86-64).
8421 On Pentium and PentiumPro, @code{double} and @code{long double} values
8422 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8423 suffer significant run time performance penalties. On Pentium III, the
8424 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8425 penalties if it is not 16 byte aligned.
8427 To ensure proper alignment of this values on the stack, the stack boundary
8428 must be as aligned as that required by any value stored on the stack.
8429 Further, every function must be generated such that it keeps the stack
8430 aligned. Thus calling a function compiled with a higher preferred
8431 stack boundary from a function compiled with a lower preferred stack
8432 boundary will most likely misalign the stack. It is recommended that
8433 libraries that use callbacks always use the default setting.
8435 This extra alignment does consume extra stack space, and generally
8436 increases code size. Code that is sensitive to stack space usage, such
8437 as embedded systems and operating system kernels, may want to reduce the
8438 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8456 These switches enable or disable the use of built-in functions that allow
8457 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
8459 @xref{X86 Built-in Functions}, for details of the functions enabled
8460 and disabled by these switches.
8462 To have SSE/SSE2 instructions generated automatically from floating-point
8463 code, see @option{-mfpmath=sse}.
8466 @itemx -mno-push-args
8468 @opindex mno-push-args
8469 Use PUSH operations to store outgoing parameters. This method is shorter
8470 and usually equally fast as method using SUB/MOV operations and is enabled
8471 by default. In some cases disabling it may improve performance because of
8472 improved scheduling and reduced dependencies.
8474 @item -maccumulate-outgoing-args
8475 @opindex maccumulate-outgoing-args
8476 If enabled, the maximum amount of space required for outgoing arguments will be
8477 computed in the function prologue. This is faster on most modern CPUs
8478 because of reduced dependencies, improved scheduling and reduced stack usage
8479 when preferred stack boundary is not equal to 2. The drawback is a notable
8480 increase in code size. This switch implies @option{-mno-push-args}.
8484 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8485 on thread-safe exception handling must compile and link all code with the
8486 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8487 @option{-D_MT}; when linking, it links in a special thread helper library
8488 @option{-lmingwthrd} which cleans up per thread exception handling data.
8490 @item -mno-align-stringops
8491 @opindex mno-align-stringops
8492 Do not align destination of inlined string operations. This switch reduces
8493 code size and improves performance in case the destination is already aligned,
8494 but gcc don't know about it.
8496 @item -minline-all-stringops
8497 @opindex minline-all-stringops
8498 By default GCC inlines string operations only when destination is known to be
8499 aligned at least to 4 byte boundary. This enables more inlining, increase code
8500 size, but may improve performance of code that depends on fast memcpy, strlen
8501 and memset for short lengths.
8503 @item -momit-leaf-frame-pointer
8504 @opindex momit-leaf-frame-pointer
8505 Don't keep the frame pointer in a register for leaf functions. This
8506 avoids the instructions to save, set up and restore frame pointers and
8507 makes an extra register available in leaf functions. The option
8508 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8509 which might make debugging harder.
8511 @item -mtls-direct-seg-refs
8512 @itemx -mno-tls-direct-seg-refs
8513 @opindex mtls-direct-seg-refs
8514 Controls whether TLS variables may be accessed with offsets from the
8515 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8516 or whether the thread base pointer must be added. Whether or not this
8517 is legal depends on the operating system, and whether it maps the
8518 segment to cover the entire TLS area.
8520 For systems that use GNU libc, the default is on.
8523 These @samp{-m} switches are supported in addition to the above
8524 on AMD x86-64 processors in 64-bit environments.
8531 Generate code for a 32-bit or 64-bit environment.
8532 The 32-bit environment sets int, long and pointer to 32 bits and
8533 generates code that runs on any i386 system.
8534 The 64-bit environment sets int to 32 bits and long and pointer
8535 to 64 bits and generates code for AMD's x86-64 architecture.
8538 @opindex no-red-zone
8539 Do not use a so called red zone for x86-64 code. The red zone is mandated
8540 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8541 stack pointer that will not be modified by signal or interrupt handlers
8542 and therefore can be used for temporary data without adjusting the stack
8543 pointer. The flag @option{-mno-red-zone} disables this red zone.
8545 @item -mcmodel=small
8546 @opindex mcmodel=small
8547 Generate code for the small code model: the program and its symbols must
8548 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8549 Programs can be statically or dynamically linked. This is the default
8552 @item -mcmodel=kernel
8553 @opindex mcmodel=kernel
8554 Generate code for the kernel code model. The kernel runs in the
8555 negative 2 GB of the address space.
8556 This model has to be used for Linux kernel code.
8558 @item -mcmodel=medium
8559 @opindex mcmodel=medium
8560 Generate code for the medium model: The program is linked in the lower 2
8561 GB of the address space but symbols can be located anywhere in the
8562 address space. Programs can be statically or dynamically linked, but
8563 building of shared libraries are not supported with the medium model.
8565 @item -mcmodel=large
8566 @opindex mcmodel=large
8567 Generate code for the large model: This model makes no assumptions
8568 about addresses and sizes of sections. Currently GCC does not implement
8573 @subsection HPPA Options
8574 @cindex HPPA Options
8576 These @samp{-m} options are defined for the HPPA family of computers:
8579 @item -march=@var{architecture-type}
8581 Generate code for the specified architecture. The choices for
8582 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8583 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8584 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8585 architecture option for your machine. Code compiled for lower numbered
8586 architectures will run on higher numbered architectures, but not the
8589 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8590 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8594 @itemx -mpa-risc-1-1
8595 @itemx -mpa-risc-2-0
8596 @opindex mpa-risc-1-0
8597 @opindex mpa-risc-1-1
8598 @opindex mpa-risc-2-0
8599 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8602 @opindex mbig-switch
8603 Generate code suitable for big switch tables. Use this option only if
8604 the assembler/linker complain about out of range branches within a switch
8607 @item -mjump-in-delay
8608 @opindex mjump-in-delay
8609 Fill delay slots of function calls with unconditional jump instructions
8610 by modifying the return pointer for the function call to be the target
8611 of the conditional jump.
8613 @item -mdisable-fpregs
8614 @opindex mdisable-fpregs
8615 Prevent floating point registers from being used in any manner. This is
8616 necessary for compiling kernels which perform lazy context switching of
8617 floating point registers. If you use this option and attempt to perform
8618 floating point operations, the compiler will abort.
8620 @item -mdisable-indexing
8621 @opindex mdisable-indexing
8622 Prevent the compiler from using indexing address modes. This avoids some
8623 rather obscure problems when compiling MIG generated code under MACH@.
8625 @item -mno-space-regs
8626 @opindex mno-space-regs
8627 Generate code that assumes the target has no space registers. This allows
8628 GCC to generate faster indirect calls and use unscaled index address modes.
8630 Such code is suitable for level 0 PA systems and kernels.
8632 @item -mfast-indirect-calls
8633 @opindex mfast-indirect-calls
8634 Generate code that assumes calls never cross space boundaries. This
8635 allows GCC to emit code which performs faster indirect calls.
8637 This option will not work in the presence of shared libraries or nested
8640 @item -mlong-load-store
8641 @opindex mlong-load-store
8642 Generate 3-instruction load and store sequences as sometimes required by
8643 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8646 @item -mportable-runtime
8647 @opindex mportable-runtime
8648 Use the portable calling conventions proposed by HP for ELF systems.
8652 Enable the use of assembler directives only GAS understands.
8654 @item -mschedule=@var{cpu-type}
8656 Schedule code according to the constraints for the machine type
8657 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8658 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8659 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8660 proper scheduling option for your machine. The default scheduling is
8664 @opindex mlinker-opt
8665 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8666 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8667 linkers in which they give bogus error messages when linking some programs.
8670 @opindex msoft-float
8671 Generate output containing library calls for floating point.
8672 @strong{Warning:} the requisite libraries are not available for all HPPA
8673 targets. Normally the facilities of the machine's usual C compiler are
8674 used, but this cannot be done directly in cross-compilation. You must make
8675 your own arrangements to provide suitable library functions for
8676 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8677 does provide software floating point support.
8679 @option{-msoft-float} changes the calling convention in the output file;
8680 therefore, it is only useful if you compile @emph{all} of a program with
8681 this option. In particular, you need to compile @file{libgcc.a}, the
8682 library that comes with GCC, with @option{-msoft-float} in order for
8687 Generate the predefine, @code{_SIO}, for server IO. The default is
8688 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8689 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8690 options are available under HP-UX and HI-UX.
8694 Use GNU ld specific options. This passes @option{-shared} to ld when
8695 building a shared library. It is the default when GCC is configured,
8696 explicitly or implicitly, with the GNU linker. This option does not
8697 have any affect on which ld is called, it only changes what parameters
8698 are passed to that ld. The ld that is called is determined by the
8699 @option{--with-ld} configure option, gcc's program search path, and
8700 finally by the user's @env{PATH}. The linker used by GCC can be printed
8701 using @samp{which `gcc -print-prog-name=ld`}.
8705 Use HP ld specific options. This passes @option{-b} to ld when building
8706 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8707 links. It is the default when GCC is configured, explicitly or
8708 implicitly, with the HP linker. This option does not have any affect on
8709 which ld is called, it only changes what parameters are passed to that
8710 ld. The ld that is called is determined by the @option{--with-ld}
8711 configure option, gcc's program search path, and finally by the user's
8712 @env{PATH}. The linker used by GCC can be printed using @samp{which
8713 `gcc -print-prog-name=ld`}.
8716 @opindex mno-long-calls
8717 Generate code that uses long call sequences. This ensures that a call
8718 is always able to reach linker generated stubs. The default is to generate
8719 long calls only when the distance from the call site to the beginning
8720 of the function or translation unit, as the case may be, exceeds a
8721 predefined limit set by the branch type being used. The limits for
8722 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8723 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8726 Distances are measured from the beginning of functions when using the
8727 @option{-ffunction-sections} option, or when using the @option{-mgas}
8728 and @option{-mno-portable-runtime} options together under HP-UX with
8731 It is normally not desirable to use this option as it will degrade
8732 performance. However, it may be useful in large applications,
8733 particularly when partial linking is used to build the application.
8735 The types of long calls used depends on the capabilities of the
8736 assembler and linker, and the type of code being generated. The
8737 impact on systems that support long absolute calls, and long pic
8738 symbol-difference or pc-relative calls should be relatively small.
8739 However, an indirect call is used on 32-bit ELF systems in pic code
8740 and it is quite long.
8744 Suppress the generation of link options to search libdld.sl when the
8745 @option{-static} option is specified on HP-UX 10 and later.
8749 The HP-UX implementation of setlocale in libc has a dependency on
8750 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8751 when the @option{-static} option is specified, special link options
8752 are needed to resolve this dependency.
8754 On HP-UX 10 and later, the GCC driver adds the necessary options to
8755 link with libdld.sl when the @option{-static} option is specified.
8756 This causes the resulting binary to be dynamic. On the 64-bit port,
8757 the linkers generate dynamic binaries by default in any case. The
8758 @option{-nolibdld} option can be used to prevent the GCC driver from
8759 adding these link options.
8763 Add support for multithreading with the @dfn{dce thread} library
8764 under HP-UX. This option sets flags for both the preprocessor and
8768 @node Intel 960 Options
8769 @subsection Intel 960 Options
8771 These @samp{-m} options are defined for the Intel 960 implementations:
8774 @item -m@var{cpu-type}
8782 Assume the defaults for the machine type @var{cpu-type} for some of
8783 the other options, including instruction scheduling, floating point
8784 support, and addressing modes. The choices for @var{cpu-type} are
8785 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
8786 @samp{sa}, and @samp{sb}.
8793 @opindex msoft-float
8794 The @option{-mnumerics} option indicates that the processor does support
8795 floating-point instructions. The @option{-msoft-float} option indicates
8796 that floating-point support should not be assumed.
8798 @item -mleaf-procedures
8799 @itemx -mno-leaf-procedures
8800 @opindex mleaf-procedures
8801 @opindex mno-leaf-procedures
8802 Do (or do not) attempt to alter leaf procedures to be callable with the
8803 @code{bal} instruction as well as @code{call}. This will result in more
8804 efficient code for explicit calls when the @code{bal} instruction can be
8805 substituted by the assembler or linker, but less efficient code in other
8806 cases, such as calls via function pointers, or using a linker that doesn't
8807 support this optimization.
8810 @itemx -mno-tail-call
8812 @opindex mno-tail-call
8813 Do (or do not) make additional attempts (beyond those of the
8814 machine-independent portions of the compiler) to optimize tail-recursive
8815 calls into branches. You may not want to do this because the detection of
8816 cases where this is not valid is not totally complete. The default is
8817 @option{-mno-tail-call}.
8819 @item -mcomplex-addr
8820 @itemx -mno-complex-addr
8821 @opindex mcomplex-addr
8822 @opindex mno-complex-addr
8823 Assume (or do not assume) that the use of a complex addressing mode is a
8824 win on this implementation of the i960. Complex addressing modes may not
8825 be worthwhile on the K-series, but they definitely are on the C-series.
8826 The default is currently @option{-mcomplex-addr} for all processors except
8830 @itemx -mno-code-align
8831 @opindex mcode-align
8832 @opindex mno-code-align
8833 Align code to 8-byte boundaries for faster fetching (or don't bother).
8834 Currently turned on by default for C-series implementations only.
8837 @item -mclean-linkage
8838 @itemx -mno-clean-linkage
8839 @opindex mclean-linkage
8840 @opindex mno-clean-linkage
8841 These options are not fully implemented.
8845 @itemx -mic2.0-compat
8846 @itemx -mic3.0-compat
8848 @opindex mic2.0-compat
8849 @opindex mic3.0-compat
8850 Enable compatibility with iC960 v2.0 or v3.0.
8854 @opindex masm-compat
8856 Enable compatibility with the iC960 assembler.
8858 @item -mstrict-align
8859 @itemx -mno-strict-align
8860 @opindex mstrict-align
8861 @opindex mno-strict-align
8862 Do not permit (do permit) unaligned accesses.
8866 Enable structure-alignment compatibility with Intel's gcc release version
8867 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
8869 @item -mlong-double-64
8870 @opindex mlong-double-64
8871 Implement type @samp{long double} as 64-bit floating point numbers.
8872 Without the option @samp{long double} is implemented by 80-bit
8873 floating point numbers. The only reason we have it because there is
8874 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
8875 is only useful for people using soft-float targets. Otherwise, we
8876 should recommend against use of it.
8880 @node DEC Alpha Options
8881 @subsection DEC Alpha Options
8883 These @samp{-m} options are defined for the DEC Alpha implementations:
8886 @item -mno-soft-float
8888 @opindex mno-soft-float
8889 @opindex msoft-float
8890 Use (do not use) the hardware floating-point instructions for
8891 floating-point operations. When @option{-msoft-float} is specified,
8892 functions in @file{libgcc.a} will be used to perform floating-point
8893 operations. Unless they are replaced by routines that emulate the
8894 floating-point operations, or compiled in such a way as to call such
8895 emulations routines, these routines will issue floating-point
8896 operations. If you are compiling for an Alpha without floating-point
8897 operations, you must ensure that the library is built so as not to call
8900 Note that Alpha implementations without floating-point operations are
8901 required to have floating-point registers.
8906 @opindex mno-fp-regs
8907 Generate code that uses (does not use) the floating-point register set.
8908 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8909 register set is not used, floating point operands are passed in integer
8910 registers as if they were integers and floating-point results are passed
8911 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8912 so any function with a floating-point argument or return value called by code
8913 compiled with @option{-mno-fp-regs} must also be compiled with that
8916 A typical use of this option is building a kernel that does not use,
8917 and hence need not save and restore, any floating-point registers.
8921 The Alpha architecture implements floating-point hardware optimized for
8922 maximum performance. It is mostly compliant with the IEEE floating
8923 point standard. However, for full compliance, software assistance is
8924 required. This option generates code fully IEEE compliant code
8925 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8926 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8927 defined during compilation. The resulting code is less efficient but is
8928 able to correctly support denormalized numbers and exceptional IEEE
8929 values such as not-a-number and plus/minus infinity. Other Alpha
8930 compilers call this option @option{-ieee_with_no_inexact}.
8932 @item -mieee-with-inexact
8933 @opindex mieee-with-inexact
8934 This is like @option{-mieee} except the generated code also maintains
8935 the IEEE @var{inexact-flag}. Turning on this option causes the
8936 generated code to implement fully-compliant IEEE math. In addition to
8937 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8938 macro. On some Alpha implementations the resulting code may execute
8939 significantly slower than the code generated by default. Since there is
8940 very little code that depends on the @var{inexact-flag}, you should
8941 normally not specify this option. Other Alpha compilers call this
8942 option @option{-ieee_with_inexact}.
8944 @item -mfp-trap-mode=@var{trap-mode}
8945 @opindex mfp-trap-mode
8946 This option controls what floating-point related traps are enabled.
8947 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8948 The trap mode can be set to one of four values:
8952 This is the default (normal) setting. The only traps that are enabled
8953 are the ones that cannot be disabled in software (e.g., division by zero
8957 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8961 Like @samp{su}, but the instructions are marked to be safe for software
8962 completion (see Alpha architecture manual for details).
8965 Like @samp{su}, but inexact traps are enabled as well.
8968 @item -mfp-rounding-mode=@var{rounding-mode}
8969 @opindex mfp-rounding-mode
8970 Selects the IEEE rounding mode. Other Alpha compilers call this option
8971 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8976 Normal IEEE rounding mode. Floating point numbers are rounded towards
8977 the nearest machine number or towards the even machine number in case
8981 Round towards minus infinity.
8984 Chopped rounding mode. Floating point numbers are rounded towards zero.
8987 Dynamic rounding mode. A field in the floating point control register
8988 (@var{fpcr}, see Alpha architecture reference manual) controls the
8989 rounding mode in effect. The C library initializes this register for
8990 rounding towards plus infinity. Thus, unless your program modifies the
8991 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8994 @item -mtrap-precision=@var{trap-precision}
8995 @opindex mtrap-precision
8996 In the Alpha architecture, floating point traps are imprecise. This
8997 means without software assistance it is impossible to recover from a
8998 floating trap and program execution normally needs to be terminated.
8999 GCC can generate code that can assist operating system trap handlers
9000 in determining the exact location that caused a floating point trap.
9001 Depending on the requirements of an application, different levels of
9002 precisions can be selected:
9006 Program precision. This option is the default and means a trap handler
9007 can only identify which program caused a floating point exception.
9010 Function precision. The trap handler can determine the function that
9011 caused a floating point exception.
9014 Instruction precision. The trap handler can determine the exact
9015 instruction that caused a floating point exception.
9018 Other Alpha compilers provide the equivalent options called
9019 @option{-scope_safe} and @option{-resumption_safe}.
9021 @item -mieee-conformant
9022 @opindex mieee-conformant
9023 This option marks the generated code as IEEE conformant. You must not
9024 use this option unless you also specify @option{-mtrap-precision=i} and either
9025 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9026 is to emit the line @samp{.eflag 48} in the function prologue of the
9027 generated assembly file. Under DEC Unix, this has the effect that
9028 IEEE-conformant math library routines will be linked in.
9030 @item -mbuild-constants
9031 @opindex mbuild-constants
9032 Normally GCC examines a 32- or 64-bit integer constant to
9033 see if it can construct it from smaller constants in two or three
9034 instructions. If it cannot, it will output the constant as a literal and
9035 generate code to load it from the data segment at runtime.
9037 Use this option to require GCC to construct @emph{all} integer constants
9038 using code, even if it takes more instructions (the maximum is six).
9040 You would typically use this option to build a shared library dynamic
9041 loader. Itself a shared library, it must relocate itself in memory
9042 before it can find the variables and constants in its own data segment.
9048 Select whether to generate code to be assembled by the vendor-supplied
9049 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9067 Indicate whether GCC should generate code to use the optional BWX,
9068 CIX, FIX and MAX instruction sets. The default is to use the instruction
9069 sets supported by the CPU type specified via @option{-mcpu=} option or that
9070 of the CPU on which GCC was built if none was specified.
9075 @opindex mfloat-ieee
9076 Generate code that uses (does not use) VAX F and G floating point
9077 arithmetic instead of IEEE single and double precision.
9079 @item -mexplicit-relocs
9080 @itemx -mno-explicit-relocs
9081 @opindex mexplicit-relocs
9082 @opindex mno-explicit-relocs
9083 Older Alpha assemblers provided no way to generate symbol relocations
9084 except via assembler macros. Use of these macros does not allow
9085 optimal instruction scheduling. GNU binutils as of version 2.12
9086 supports a new syntax that allows the compiler to explicitly mark
9087 which relocations should apply to which instructions. This option
9088 is mostly useful for debugging, as GCC detects the capabilities of
9089 the assembler when it is built and sets the default accordingly.
9093 @opindex msmall-data
9094 @opindex mlarge-data
9095 When @option{-mexplicit-relocs} is in effect, static data is
9096 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9097 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9098 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9099 16-bit relocations off of the @code{$gp} register. This limits the
9100 size of the small data area to 64KB, but allows the variables to be
9101 directly accessed via a single instruction.
9103 The default is @option{-mlarge-data}. With this option the data area
9104 is limited to just below 2GB. Programs that require more than 2GB of
9105 data must use @code{malloc} or @code{mmap} to allocate the data in the
9106 heap instead of in the program's data segment.
9108 When generating code for shared libraries, @option{-fpic} implies
9109 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9113 @opindex msmall-text
9114 @opindex mlarge-text
9115 When @option{-msmall-text} is used, the compiler assumes that the
9116 code of the entire program (or shared library) fits in 4MB, and is
9117 thus reachable with a branch instruction. When @option{-msmall-data}
9118 is used, the compiler can assume that all local symbols share the
9119 same @code{$gp} value, and thus reduce the number of instructions
9120 required for a function call from 4 to 1.
9122 The default is @option{-mlarge-text}.
9124 @item -mcpu=@var{cpu_type}
9126 Set the instruction set and instruction scheduling parameters for
9127 machine type @var{cpu_type}. You can specify either the @samp{EV}
9128 style name or the corresponding chip number. GCC supports scheduling
9129 parameters for the EV4, EV5 and EV6 family of processors and will
9130 choose the default values for the instruction set from the processor
9131 you specify. If you do not specify a processor type, GCC will default
9132 to the processor on which the compiler was built.
9134 Supported values for @var{cpu_type} are
9140 Schedules as an EV4 and has no instruction set extensions.
9144 Schedules as an EV5 and has no instruction set extensions.
9148 Schedules as an EV5 and supports the BWX extension.
9153 Schedules as an EV5 and supports the BWX and MAX extensions.
9157 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9161 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9164 @item -mtune=@var{cpu_type}
9166 Set only the instruction scheduling parameters for machine type
9167 @var{cpu_type}. The instruction set is not changed.
9169 @item -mmemory-latency=@var{time}
9170 @opindex mmemory-latency
9171 Sets the latency the scheduler should assume for typical memory
9172 references as seen by the application. This number is highly
9173 dependent on the memory access patterns used by the application
9174 and the size of the external cache on the machine.
9176 Valid options for @var{time} are
9180 A decimal number representing clock cycles.
9186 The compiler contains estimates of the number of clock cycles for
9187 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9188 (also called Dcache, Scache, and Bcache), as well as to main memory.
9189 Note that L3 is only valid for EV5.
9194 @node DEC Alpha/VMS Options
9195 @subsection DEC Alpha/VMS Options
9197 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9200 @item -mvms-return-codes
9201 @opindex mvms-return-codes
9202 Return VMS condition codes from main. The default is to return POSIX
9203 style condition (e.g.@ error) codes.
9206 @node H8/300 Options
9207 @subsection H8/300 Options
9209 These @samp{-m} options are defined for the H8/300 implementations:
9214 Shorten some address references at link time, when possible; uses the
9215 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9216 ld.info, Using ld}, for a fuller description.
9220 Generate code for the H8/300H@.
9224 Generate code for the H8S@.
9228 Generate code for the H8S and H8/300H in the normal mode. This switch
9229 must be used either with -mh or -ms.
9233 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9237 Make @code{int} data 32 bits by default.
9241 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9242 The default for the H8/300H and H8S is to align longs and floats on 4
9244 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9245 This option has no effect on the H8/300.
9249 @subsection SH Options
9251 These @samp{-m} options are defined for the SH implementations:
9256 Generate code for the SH1.
9260 Generate code for the SH2.
9263 Generate code for the SH2e.
9267 Generate code for the SH3.
9271 Generate code for the SH3e.
9275 Generate code for the SH4 without a floating-point unit.
9277 @item -m4-single-only
9278 @opindex m4-single-only
9279 Generate code for the SH4 with a floating-point unit that only
9280 supports single-precision arithmetic.
9284 Generate code for the SH4 assuming the floating-point unit is in
9285 single-precision mode by default.
9289 Generate code for the SH4.
9293 Compile code for the processor in big endian mode.
9297 Compile code for the processor in little endian mode.
9301 Align doubles at 64-bit boundaries. Note that this changes the calling
9302 conventions, and thus some functions from the standard C library will
9303 not work unless you recompile it first with @option{-mdalign}.
9307 Shorten some address references at link time, when possible; uses the
9308 linker option @option{-relax}.
9312 Use 32-bit offsets in @code{switch} tables. The default is to use
9317 Enable the use of the instruction @code{fmovd}.
9321 Comply with the calling conventions defined by Renesas.
9325 Mark the @code{MAC} register as call-clobbered, even if
9326 @option{-mhitachi} is given.
9330 Increase IEEE-compliance of floating-point code.
9334 Dump instruction size and location in the assembly code.
9338 This option is deprecated. It pads structures to multiple of 4 bytes,
9339 which is incompatible with the SH ABI@.
9343 Optimize for space instead of speed. Implied by @option{-Os}.
9347 When generating position-independent code, emit function calls using
9348 the Global Offset Table instead of the Procedure Linkage Table.
9352 Generate a library function call to invalidate instruction cache
9353 entries, after fixing up a trampoline. This library function call
9354 doesn't assume it can write to the whole memory address space. This
9355 is the default when the target is @code{sh-*-linux*}.
9358 @node System V Options
9359 @subsection Options for System V
9361 These additional options are available on System V Release 4 for
9362 compatibility with other compilers on those systems:
9367 Create a shared object.
9368 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9372 Identify the versions of each tool used by the compiler, in a
9373 @code{.ident} assembler directive in the output.
9377 Refrain from adding @code{.ident} directives to the output file (this is
9380 @item -YP,@var{dirs}
9382 Search the directories @var{dirs}, and no others, for libraries
9383 specified with @option{-l}.
9387 Look in the directory @var{dir} to find the M4 preprocessor.
9388 The assembler uses this option.
9389 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9390 @c the generic assembler that comes with Solaris takes just -Ym.
9393 @node TMS320C3x/C4x Options
9394 @subsection TMS320C3x/C4x Options
9395 @cindex TMS320C3x/C4x Options
9397 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9401 @item -mcpu=@var{cpu_type}
9403 Set the instruction set, register set, and instruction scheduling
9404 parameters for machine type @var{cpu_type}. Supported values for
9405 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9406 @samp{c44}. The default is @samp{c40} to generate code for the
9411 @itemx -msmall-memory
9413 @opindex mbig-memory
9415 @opindex msmall-memory
9417 Generates code for the big or small memory model. The small memory
9418 model assumed that all data fits into one 64K word page. At run-time
9419 the data page (DP) register must be set to point to the 64K page
9420 containing the .bss and .data program sections. The big memory model is
9421 the default and requires reloading of the DP register for every direct
9428 Allow (disallow) allocation of general integer operands into the block
9435 Enable (disable) generation of code using decrement and branch,
9436 DBcond(D), instructions. This is enabled by default for the C4x. To be
9437 on the safe side, this is disabled for the C3x, since the maximum
9438 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9439 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9440 that it can utilize the decrement and branch instruction, but will give
9441 up if there is more than one memory reference in the loop. Thus a loop
9442 where the loop counter is decremented can generate slightly more
9443 efficient code, in cases where the RPTB instruction cannot be utilized.
9445 @item -mdp-isr-reload
9447 @opindex mdp-isr-reload
9449 Force the DP register to be saved on entry to an interrupt service
9450 routine (ISR), reloaded to point to the data section, and restored on
9451 exit from the ISR@. This should not be required unless someone has
9452 violated the small memory model by modifying the DP register, say within
9459 For the C3x use the 24-bit MPYI instruction for integer multiplies
9460 instead of a library call to guarantee 32-bit results. Note that if one
9461 of the operands is a constant, then the multiplication will be performed
9462 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9463 then squaring operations are performed inline instead of a library call.
9466 @itemx -mno-fast-fix
9468 @opindex mno-fast-fix
9469 The C3x/C4x FIX instruction to convert a floating point value to an
9470 integer value chooses the nearest integer less than or equal to the
9471 floating point value rather than to the nearest integer. Thus if the
9472 floating point number is negative, the result will be incorrectly
9473 truncated an additional code is necessary to detect and correct this
9474 case. This option can be used to disable generation of the additional
9475 code required to correct the result.
9481 Enable (disable) generation of repeat block sequences using the RPTB
9482 instruction for zero overhead looping. The RPTB construct is only used
9483 for innermost loops that do not call functions or jump across the loop
9484 boundaries. There is no advantage having nested RPTB loops due to the
9485 overhead required to save and restore the RC, RS, and RE registers.
9486 This is enabled by default with @option{-O2}.
9488 @item -mrpts=@var{count}
9492 Enable (disable) the use of the single instruction repeat instruction
9493 RPTS@. If a repeat block contains a single instruction, and the loop
9494 count can be guaranteed to be less than the value @var{count}, GCC will
9495 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9496 then a RPTS will be emitted even if the loop count cannot be determined
9497 at compile time. Note that the repeated instruction following RPTS does
9498 not have to be reloaded from memory each iteration, thus freeing up the
9499 CPU buses for operands. However, since interrupts are blocked by this
9500 instruction, it is disabled by default.
9502 @item -mloop-unsigned
9503 @itemx -mno-loop-unsigned
9504 @opindex mloop-unsigned
9505 @opindex mno-loop-unsigned
9506 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9507 is @math{2^{31} + 1} since these instructions test if the iteration count is
9508 negative to terminate the loop. If the iteration count is unsigned
9509 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9510 exceeded. This switch allows an unsigned iteration count.
9514 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9515 with. This also enforces compatibility with the API employed by the TI
9516 C3x C compiler. For example, long doubles are passed as structures
9517 rather than in floating point registers.
9523 Generate code that uses registers (stack) for passing arguments to functions.
9524 By default, arguments are passed in registers where possible rather
9525 than by pushing arguments on to the stack.
9527 @item -mparallel-insns
9528 @itemx -mno-parallel-insns
9529 @opindex mparallel-insns
9530 @opindex mno-parallel-insns
9531 Allow the generation of parallel instructions. This is enabled by
9532 default with @option{-O2}.
9534 @item -mparallel-mpy
9535 @itemx -mno-parallel-mpy
9536 @opindex mparallel-mpy
9537 @opindex mno-parallel-mpy
9538 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9539 provided @option{-mparallel-insns} is also specified. These instructions have
9540 tight register constraints which can pessimize the code generation
9546 @subsection V850 Options
9547 @cindex V850 Options
9549 These @samp{-m} options are defined for V850 implementations:
9553 @itemx -mno-long-calls
9554 @opindex mlong-calls
9555 @opindex mno-long-calls
9556 Treat all calls as being far away (near). If calls are assumed to be
9557 far away, the compiler will always load the functions address up into a
9558 register, and call indirect through the pointer.
9564 Do not optimize (do optimize) basic blocks that use the same index
9565 pointer 4 or more times to copy pointer into the @code{ep} register, and
9566 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9567 option is on by default if you optimize.
9569 @item -mno-prolog-function
9570 @itemx -mprolog-function
9571 @opindex mno-prolog-function
9572 @opindex mprolog-function
9573 Do not use (do use) external functions to save and restore registers at
9574 the prolog and epilog of a function. The external functions are slower,
9575 but use less code space if more than one function saves the same number
9576 of registers. The @option{-mprolog-function} option is on by default if
9581 Try to make the code as small as possible. At present, this just turns
9582 on the @option{-mep} and @option{-mprolog-function} options.
9586 Put static or global variables whose size is @var{n} bytes or less into
9587 the tiny data area that register @code{ep} points to. The tiny data
9588 area can hold up to 256 bytes in total (128 bytes for byte references).
9592 Put static or global variables whose size is @var{n} bytes or less into
9593 the small data area that register @code{gp} points to. The small data
9594 area can hold up to 64 kilobytes.
9598 Put static or global variables whose size is @var{n} bytes or less into
9599 the first 32 kilobytes of memory.
9603 Specify that the target processor is the V850.
9606 @opindex mbig-switch
9607 Generate code suitable for big switch tables. Use this option only if
9608 the assembler/linker complain about out of range branches within a switch
9613 This option will cause r2 and r5 to be used in the code generated by
9614 the compiler. This setting is the default.
9617 @opindex mno-app-regs
9618 This option will cause r2 and r5 to be treated as fixed registers.
9622 Specify that the target processor is the V850E. The preprocessor
9623 constant @samp{__v850e__} will be defined if this option is used.
9625 If neither @option{-mv850} nor @option{-mv850e} are defined
9626 then a default target processor will be chosen and the relevant
9627 @samp{__v850*__} preprocessor constant will be defined.
9629 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9630 defined, regardless of which processor variant is the target.
9632 @item -mdisable-callt
9633 @opindex mdisable-callt
9634 This option will suppress generation of the CALLT instruction for the
9635 v850e flavors of the v850 architecture. The default is
9636 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9641 @subsection ARC Options
9644 These options are defined for ARC implementations:
9649 Compile code for little endian mode. This is the default.
9653 Compile code for big endian mode.
9656 @opindex mmangle-cpu
9657 Prepend the name of the cpu to all public symbol names.
9658 In multiple-processor systems, there are many ARC variants with different
9659 instruction and register set characteristics. This flag prevents code
9660 compiled for one cpu to be linked with code compiled for another.
9661 No facility exists for handling variants that are ``almost identical''.
9662 This is an all or nothing option.
9664 @item -mcpu=@var{cpu}
9666 Compile code for ARC variant @var{cpu}.
9667 Which variants are supported depend on the configuration.
9668 All variants support @option{-mcpu=base}, this is the default.
9670 @item -mtext=@var{text-section}
9671 @itemx -mdata=@var{data-section}
9672 @itemx -mrodata=@var{readonly-data-section}
9676 Put functions, data, and readonly data in @var{text-section},
9677 @var{data-section}, and @var{readonly-data-section} respectively
9678 by default. This can be overridden with the @code{section} attribute.
9679 @xref{Variable Attributes}.
9684 @subsection NS32K Options
9685 @cindex NS32K options
9687 These are the @samp{-m} options defined for the 32000 series. The default
9688 values for these options depends on which style of 32000 was selected when
9689 the compiler was configured; the defaults for the most common choices are
9697 Generate output for a 32032. This is the default
9698 when the compiler is configured for 32032 and 32016 based systems.
9704 Generate output for a 32332. This is the default
9705 when the compiler is configured for 32332-based systems.
9711 Generate output for a 32532. This is the default
9712 when the compiler is configured for 32532-based systems.
9716 Generate output containing 32081 instructions for floating point.
9717 This is the default for all systems.
9721 Generate output containing 32381 instructions for floating point. This
9722 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9723 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9727 Try and generate multiply-add floating point instructions @code{polyF}
9728 and @code{dotF}. This option is only available if the @option{-m32381}
9729 option is in effect. Using these instructions requires changes to
9730 register allocation which generally has a negative impact on
9731 performance. This option should only be enabled when compiling code
9732 particularly likely to make heavy use of multiply-add instructions.
9735 @opindex mnomulti-add
9736 Do not try and generate multiply-add floating point instructions
9737 @code{polyF} and @code{dotF}. This is the default on all platforms.
9740 @opindex msoft-float
9741 Generate output containing library calls for floating point.
9742 @strong{Warning:} the requisite libraries may not be available.
9744 @item -mieee-compare
9745 @itemx -mno-ieee-compare
9746 @opindex mieee-compare
9747 @opindex mno-ieee-compare
9748 Control whether or not the compiler uses IEEE floating point
9749 comparisons. These handle correctly the case where the result of a
9750 comparison is unordered.
9751 @strong{Warning:} the requisite kernel support may not be available.
9754 @opindex mnobitfield
9755 Do not use the bit-field instructions. On some machines it is faster to
9756 use shifting and masking operations. This is the default for the pc532.
9760 Do use the bit-field instructions. This is the default for all platforms
9765 Use a different function-calling convention, in which functions
9766 that take a fixed number of arguments return pop their
9767 arguments on return with the @code{ret} instruction.
9769 This calling convention is incompatible with the one normally
9770 used on Unix, so you cannot use it if you need to call libraries
9771 compiled with the Unix compiler.
9773 Also, you must provide function prototypes for all functions that
9774 take variable numbers of arguments (including @code{printf});
9775 otherwise incorrect code will be generated for calls to those
9778 In addition, seriously incorrect code will result if you call a
9779 function with too many arguments. (Normally, extra arguments are
9780 harmlessly ignored.)
9782 This option takes its name from the 680x0 @code{rtd} instruction.
9787 Use a different function-calling convention where the first two arguments
9788 are passed in registers.
9790 This calling convention is incompatible with the one normally
9791 used on Unix, so you cannot use it if you need to call libraries
9792 compiled with the Unix compiler.
9795 @opindex mnoregparam
9796 Do not pass any arguments in registers. This is the default for all
9801 It is OK to use the sb as an index register which is always loaded with
9802 zero. This is the default for the pc532-netbsd target.
9806 The sb register is not available for use or has not been initialized to
9807 zero by the run time system. This is the default for all targets except
9808 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9809 @option{-fpic} is set.
9813 Many ns32000 series addressing modes use displacements of up to 512MB@.
9814 If an address is above 512MB then displacements from zero can not be used.
9815 This option causes code to be generated which can be loaded above 512MB@.
9816 This may be useful for operating systems or ROM code.
9820 Assume code will be loaded in the first 512MB of virtual address space.
9821 This is the default for all platforms.
9827 @subsection AVR Options
9830 These options are defined for AVR implementations:
9833 @item -mmcu=@var{mcu}
9835 Specify ATMEL AVR instruction set or MCU type.
9837 Instruction set avr1 is for the minimal AVR core, not supported by the C
9838 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9839 attiny11, attiny12, attiny15, attiny28).
9841 Instruction set avr2 (default) is for the classic AVR core with up to
9842 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9843 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9844 at90c8534, at90s8535).
9846 Instruction set avr3 is for the classic AVR core with up to 128K program
9847 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9849 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9850 memory space (MCU types: atmega8, atmega83, atmega85).
9852 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9853 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9854 atmega64, atmega128, at43usb355, at94k).
9858 Output instruction sizes to the asm file.
9860 @item -minit-stack=@var{N}
9861 @opindex minit-stack
9862 Specify the initial stack address, which may be a symbol or numeric value,
9863 @samp{__stack} is the default.
9865 @item -mno-interrupts
9866 @opindex mno-interrupts
9867 Generated code is not compatible with hardware interrupts.
9868 Code size will be smaller.
9870 @item -mcall-prologues
9871 @opindex mcall-prologues
9872 Functions prologues/epilogues expanded as call to appropriate
9873 subroutines. Code size will be smaller.
9875 @item -mno-tablejump
9876 @opindex mno-tablejump
9877 Do not generate tablejump insns which sometimes increase code size.
9880 @opindex mtiny-stack
9881 Change only the low 8 bits of the stack pointer.
9885 @subsection MCore Options
9886 @cindex MCore options
9888 These are the @samp{-m} options defined for the Motorola M*Core
9896 @opindex mno-hardlit
9897 Inline constants into the code stream if it can be done in two
9898 instructions or less.
9904 Use the divide instruction. (Enabled by default).
9906 @item -mrelax-immediate
9907 @itemx -mno-relax-immediate
9908 @opindex mrelax-immediate
9909 @opindex mno-relax-immediate
9910 Allow arbitrary sized immediates in bit operations.
9912 @item -mwide-bitfields
9913 @itemx -mno-wide-bitfields
9914 @opindex mwide-bitfields
9915 @opindex mno-wide-bitfields
9916 Always treat bit-fields as int-sized.
9918 @item -m4byte-functions
9919 @itemx -mno-4byte-functions
9920 @opindex m4byte-functions
9921 @opindex mno-4byte-functions
9922 Force all functions to be aligned to a four byte boundary.
9924 @item -mcallgraph-data
9925 @itemx -mno-callgraph-data
9926 @opindex mcallgraph-data
9927 @opindex mno-callgraph-data
9928 Emit callgraph information.
9931 @itemx -mno-slow-bytes
9932 @opindex mslow-bytes
9933 @opindex mno-slow-bytes
9934 Prefer word access when reading byte quantities.
9936 @item -mlittle-endian
9938 @opindex mlittle-endian
9939 @opindex mbig-endian
9940 Generate code for a little endian target.
9946 Generate code for the 210 processor.
9950 @subsection IA-64 Options
9951 @cindex IA-64 Options
9953 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9957 @opindex mbig-endian
9958 Generate code for a big endian target. This is the default for HP-UX@.
9960 @item -mlittle-endian
9961 @opindex mlittle-endian
9962 Generate code for a little endian target. This is the default for AIX5
9969 Generate (or don't) code for the GNU assembler. This is the default.
9970 @c Also, this is the default if the configure option @option{--with-gnu-as}
9977 Generate (or don't) code for the GNU linker. This is the default.
9978 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9983 Generate code that does not use a global pointer register. The result
9984 is not position independent code, and violates the IA-64 ABI@.
9986 @item -mvolatile-asm-stop
9987 @itemx -mno-volatile-asm-stop
9988 @opindex mvolatile-asm-stop
9989 @opindex mno-volatile-asm-stop
9990 Generate (or don't) a stop bit immediately before and after volatile asm
9995 Generate code that works around Itanium B step errata.
9997 @item -mregister-names
9998 @itemx -mno-register-names
9999 @opindex mregister-names
10000 @opindex mno-register-names
10001 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10002 the stacked registers. This may make assembler output more readable.
10008 Disable (or enable) optimizations that use the small data section. This may
10009 be useful for working around optimizer bugs.
10011 @item -mconstant-gp
10012 @opindex mconstant-gp
10013 Generate code that uses a single constant global pointer value. This is
10014 useful when compiling kernel code.
10018 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10019 This is useful when compiling firmware code.
10021 @item -minline-float-divide-min-latency
10022 @opindex minline-float-divide-min-latency
10023 Generate code for inline divides of floating point values
10024 using the minimum latency algorithm.
10026 @item -minline-float-divide-max-throughput
10027 @opindex minline-float-divide-max-throughput
10028 Generate code for inline divides of floating point values
10029 using the maximum throughput algorithm.
10031 @item -minline-int-divide-min-latency
10032 @opindex minline-int-divide-min-latency
10033 Generate code for inline divides of integer values
10034 using the minimum latency algorithm.
10036 @item -minline-int-divide-max-throughput
10037 @opindex minline-int-divide-max-throughput
10038 Generate code for inline divides of integer values
10039 using the maximum throughput algorithm.
10041 @item -mno-dwarf2-asm
10042 @itemx -mdwarf2-asm
10043 @opindex mno-dwarf2-asm
10044 @opindex mdwarf2-asm
10045 Don't (or do) generate assembler code for the DWARF2 line number debugging
10046 info. This may be useful when not using the GNU assembler.
10048 @item -mfixed-range=@var{register-range}
10049 @opindex mfixed-range
10050 Generate code treating the given register range as fixed registers.
10051 A fixed register is one that the register allocator can not use. This is
10052 useful when compiling kernel code. A register range is specified as
10053 two registers separated by a dash. Multiple register ranges can be
10054 specified separated by a comma.
10056 @item -mearly-stop-bits
10057 @itemx -mno-early-stop-bits
10058 @opindex mearly-stop-bits
10059 @opindex mno-early-stop-bits
10060 Allow stop bits to be placed earlier than immediately preceding the
10061 instruction that triggered the stop bit. This can improve instruction
10062 scheduling, but does not always do so.
10066 @subsection D30V Options
10067 @cindex D30V Options
10069 These @samp{-m} options are defined for D30V implementations:
10074 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
10075 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
10076 memory, which starts at location @code{0x80000000}.
10079 @opindex mextmemory
10080 Same as the @option{-mextmem} switch.
10084 Link the @samp{.text} section into onchip text memory, which starts at
10085 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
10086 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
10087 into onchip data memory, which starts at location @code{0x20000000}.
10089 @item -mno-asm-optimize
10090 @itemx -masm-optimize
10091 @opindex mno-asm-optimize
10092 @opindex masm-optimize
10093 Disable (enable) passing @option{-O} to the assembler when optimizing.
10094 The assembler uses the @option{-O} option to automatically parallelize
10095 adjacent short instructions where possible.
10097 @item -mbranch-cost=@var{n}
10098 @opindex mbranch-cost
10099 Increase the internal costs of branches to @var{n}. Higher costs means
10100 that the compiler will issue more instructions to avoid doing a branch.
10103 @item -mcond-exec=@var{n}
10104 @opindex mcond-exec
10105 Specify the maximum number of conditionally executed instructions that
10106 replace a branch. The default is 4.
10109 @node S/390 and zSeries Options
10110 @subsection S/390 and zSeries Options
10111 @cindex S/390 and zSeries Options
10113 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10117 @itemx -msoft-float
10118 @opindex mhard-float
10119 @opindex msoft-float
10120 Use (do not use) the hardware floating-point instructions and registers
10121 for floating-point operations. When @option{-msoft-float} is specified,
10122 functions in @file{libgcc.a} will be used to perform floating-point
10123 operations. When @option{-mhard-float} is specified, the compiler
10124 generates IEEE floating-point instructions. This is the default.
10127 @itemx -mno-backchain
10128 @opindex mbackchain
10129 @opindex mno-backchain
10130 Generate (or do not generate) code which maintains an explicit
10131 backchain within the stack frame that points to the caller's frame.
10132 This is currently needed to allow debugging. The default is to
10133 generate the backchain.
10136 @itemx -mno-small-exec
10137 @opindex msmall-exec
10138 @opindex mno-small-exec
10139 Generate (or do not generate) code using the @code{bras} instruction
10140 to do subroutine calls.
10141 This only works reliably if the total executable size does not
10142 exceed 64k. The default is to use the @code{basr} instruction instead,
10143 which does not have this limitation.
10149 When @option{-m31} is specified, generate code compliant to the
10150 Linux for S/390 ABI@. When @option{-m64} is specified, generate
10151 code compliant to the Linux for zSeries ABI@. This allows GCC in
10152 particular to generate 64-bit instructions. For the @samp{s390}
10153 targets, the default is @option{-m31}, while the @samp{s390x}
10154 targets default to @option{-m64}.
10160 When @option{-mzarch} is specified, generate code using the
10161 instructions available on z/Architecture.
10162 When @option{-mesa} is specified, generate code using the
10163 instructions available on ESA/390. Note that @option{-mesa} is
10164 not possible with @option{-m64}.
10165 When generating code compliant to the Linux for S/390 ABI,
10166 the default is @option{-mesa}. When generating code compliant
10167 to the Linux for zSeries ABI, the default is @option{-mzarch}.
10173 Generate (or do not generate) code using the @code{mvcle} instruction
10174 to perform block moves. When @option{-mno-mvcle} is specified,
10175 use a @code{mvc} loop instead. This is the default.
10181 Print (or do not print) additional debug information when compiling.
10182 The default is to not print debug information.
10184 @item -march=@var{cpu-type}
10186 Generate code that will run on @var{cpu-type}, which is the name of a system
10187 representing a certain processor type. Possible values for
10188 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10189 When generating code using the instructions available on z/Architecture,
10190 the default is @option{-march=z900}. Otherwise, the default is
10191 @option{-march=g5}.
10193 @item -mtune=@var{cpu-type}
10195 Tune to @var{cpu-type} everything applicable about the generated code,
10196 except for the ABI and the set of available instructions.
10197 The list of @var{cpu-type} values is the same as for @option{-march}.
10198 The default is the value used for @option{-march}.
10203 @subsection CRIS Options
10204 @cindex CRIS Options
10206 These options are defined specifically for the CRIS ports.
10209 @item -march=@var{architecture-type}
10210 @itemx -mcpu=@var{architecture-type}
10213 Generate code for the specified architecture. The choices for
10214 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10215 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
10216 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10219 @item -mtune=@var{architecture-type}
10221 Tune to @var{architecture-type} everything applicable about the generated
10222 code, except for the ABI and the set of available instructions. The
10223 choices for @var{architecture-type} are the same as for
10224 @option{-march=@var{architecture-type}}.
10226 @item -mmax-stack-frame=@var{n}
10227 @opindex mmax-stack-frame
10228 Warn when the stack frame of a function exceeds @var{n} bytes.
10230 @item -melinux-stacksize=@var{n}
10231 @opindex melinux-stacksize
10232 Only available with the @samp{cris-axis-aout} target. Arranges for
10233 indications in the program to the kernel loader that the stack of the
10234 program should be set to @var{n} bytes.
10240 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10241 @option{-march=v3} and @option{-march=v8} respectively.
10245 Enable CRIS-specific verbose debug-related information in the assembly
10246 code. This option also has the effect to turn off the @samp{#NO_APP}
10247 formatted-code indicator to the assembler at the beginning of the
10252 Do not use condition-code results from previous instruction; always emit
10253 compare and test instructions before use of condition codes.
10255 @item -mno-side-effects
10256 @opindex mno-side-effects
10257 Do not emit instructions with side-effects in addressing modes other than
10260 @item -mstack-align
10261 @itemx -mno-stack-align
10262 @itemx -mdata-align
10263 @itemx -mno-data-align
10264 @itemx -mconst-align
10265 @itemx -mno-const-align
10266 @opindex mstack-align
10267 @opindex mno-stack-align
10268 @opindex mdata-align
10269 @opindex mno-data-align
10270 @opindex mconst-align
10271 @opindex mno-const-align
10272 These options (no-options) arranges (eliminate arrangements) for the
10273 stack-frame, individual data and constants to be aligned for the maximum
10274 single data access size for the chosen CPU model. The default is to
10275 arrange for 32-bit alignment. ABI details such as structure layout are
10276 not affected by these options.
10284 Similar to the stack- data- and const-align options above, these options
10285 arrange for stack-frame, writable data and constants to all be 32-bit,
10286 16-bit or 8-bit aligned. The default is 32-bit alignment.
10288 @item -mno-prologue-epilogue
10289 @itemx -mprologue-epilogue
10290 @opindex mno-prologue-epilogue
10291 @opindex mprologue-epilogue
10292 With @option{-mno-prologue-epilogue}, the normal function prologue and
10293 epilogue that sets up the stack-frame are omitted and no return
10294 instructions or return sequences are generated in the code. Use this
10295 option only together with visual inspection of the compiled code: no
10296 warnings or errors are generated when call-saved registers must be saved,
10297 or storage for local variable needs to be allocated.
10301 @opindex mno-gotplt
10303 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10304 instruction sequences that load addresses for functions from the PLT part
10305 of the GOT rather than (traditional on other architectures) calls to the
10306 PLT. The default is @option{-mgotplt}.
10310 Legacy no-op option only recognized with the cris-axis-aout target.
10314 Legacy no-op option only recognized with the cris-axis-elf and
10315 cris-axis-linux-gnu targets.
10319 Only recognized with the cris-axis-aout target, where it selects a
10320 GNU/linux-like multilib, include files and instruction set for
10321 @option{-march=v8}.
10325 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10329 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
10330 to link with input-output functions from a simulator library. Code,
10331 initialized data and zero-initialized data are allocated consecutively.
10335 Like @option{-sim}, but pass linker options to locate initialized data at
10336 0x40000000 and zero-initialized data at 0x80000000.
10340 @subsection MMIX Options
10341 @cindex MMIX Options
10343 These options are defined for the MMIX:
10347 @itemx -mno-libfuncs
10349 @opindex mno-libfuncs
10350 Specify that intrinsic library functions are being compiled, passing all
10351 values in registers, no matter the size.
10354 @itemx -mno-epsilon
10356 @opindex mno-epsilon
10357 Generate floating-point comparison instructions that compare with respect
10358 to the @code{rE} epsilon register.
10360 @item -mabi=mmixware
10362 @opindex mabi-mmixware
10364 Generate code that passes function parameters and return values that (in
10365 the called function) are seen as registers @code{$0} and up, as opposed to
10366 the GNU ABI which uses global registers @code{$231} and up.
10368 @item -mzero-extend
10369 @itemx -mno-zero-extend
10370 @opindex mzero-extend
10371 @opindex mno-zero-extend
10372 When reading data from memory in sizes shorter than 64 bits, use (do not
10373 use) zero-extending load instructions by default, rather than
10374 sign-extending ones.
10377 @itemx -mno-knuthdiv
10379 @opindex mno-knuthdiv
10380 Make the result of a division yielding a remainder have the same sign as
10381 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10382 remainder follows the sign of the dividend. Both methods are
10383 arithmetically valid, the latter being almost exclusively used.
10385 @item -mtoplevel-symbols
10386 @itemx -mno-toplevel-symbols
10387 @opindex mtoplevel-symbols
10388 @opindex mno-toplevel-symbols
10389 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10390 code can be used with the @code{PREFIX} assembly directive.
10394 Generate an executable in the ELF format, rather than the default
10395 @samp{mmo} format used by the @command{mmix} simulator.
10397 @item -mbranch-predict
10398 @itemx -mno-branch-predict
10399 @opindex mbranch-predict
10400 @opindex mno-branch-predict
10401 Use (do not use) the probable-branch instructions, when static branch
10402 prediction indicates a probable branch.
10404 @item -mbase-addresses
10405 @itemx -mno-base-addresses
10406 @opindex mbase-addresses
10407 @opindex mno-base-addresses
10408 Generate (do not generate) code that uses @emph{base addresses}. Using a
10409 base address automatically generates a request (handled by the assembler
10410 and the linker) for a constant to be set up in a global register. The
10411 register is used for one or more base address requests within the range 0
10412 to 255 from the value held in the register. The generally leads to short
10413 and fast code, but the number of different data items that can be
10414 addressed is limited. This means that a program that uses lots of static
10415 data may require @option{-mno-base-addresses}.
10417 @item -msingle-exit
10418 @itemx -mno-single-exit
10419 @opindex msingle-exit
10420 @opindex mno-single-exit
10421 Force (do not force) generated code to have a single exit point in each
10425 @node PDP-11 Options
10426 @subsection PDP-11 Options
10427 @cindex PDP-11 Options
10429 These options are defined for the PDP-11:
10434 Use hardware FPP floating point. This is the default. (FIS floating
10435 point on the PDP-11/40 is not supported.)
10438 @opindex msoft-float
10439 Do not use hardware floating point.
10443 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10447 Return floating-point results in memory. This is the default.
10451 Generate code for a PDP-11/40.
10455 Generate code for a PDP-11/45. This is the default.
10459 Generate code for a PDP-11/10.
10461 @item -mbcopy-builtin
10462 @opindex bcopy-builtin
10463 Use inline @code{movstrhi} patterns for copying memory. This is the
10468 Do not use inline @code{movstrhi} patterns for copying memory.
10474 Use 16-bit @code{int}. This is the default.
10480 Use 32-bit @code{int}.
10483 @itemx -mno-float32
10485 @opindex mno-float32
10486 Use 64-bit @code{float}. This is the default.
10491 @opindex mno-float64
10492 Use 32-bit @code{float}.
10496 Use @code{abshi2} pattern. This is the default.
10500 Do not use @code{abshi2} pattern.
10502 @item -mbranch-expensive
10503 @opindex mbranch-expensive
10504 Pretend that branches are expensive. This is for experimenting with
10505 code generation only.
10507 @item -mbranch-cheap
10508 @opindex mbranch-cheap
10509 Do not pretend that branches are expensive. This is the default.
10513 Generate code for a system with split I&D.
10517 Generate code for a system without split I&D. This is the default.
10521 Use Unix assembler syntax. This is the default when configured for
10522 @samp{pdp11-*-bsd}.
10526 Use DEC assembler syntax. This is the default when configured for any
10527 PDP-11 target other than @samp{pdp11-*-bsd}.
10530 @node Xstormy16 Options
10531 @subsection Xstormy16 Options
10532 @cindex Xstormy16 Options
10534 These options are defined for Xstormy16:
10539 Choose startup files and linker script suitable for the simulator.
10543 @subsection FRV Options
10544 @cindex FRV Options
10550 Only use the first 32 general purpose registers.
10555 Use all 64 general purpose registers.
10560 Use only the first 32 floating point registers.
10565 Use all 64 floating point registers
10568 @opindex mhard-float
10570 Use hardware instructions for floating point operations.
10573 @opindex msoft-float
10575 Use library routines for floating point operations.
10580 Dynamically allocate condition code registers.
10585 Do not try to dynamically allocate condition code registers, only
10586 use @code{icc0} and @code{fcc0}.
10591 Change ABI to use double word insns.
10596 Do not use double word instructions.
10601 Use floating point double instructions.
10604 @opindex mno-double
10606 Do not use floating point double instructions.
10611 Use media instructions.
10616 Do not use media instructions.
10621 Use multiply and add/subtract instructions.
10624 @opindex mno-muladd
10626 Do not use multiply and add/subtract instructions.
10628 @item -mlibrary-pic
10629 @opindex mlibrary-pic
10631 Enable PIC support for building libraries
10636 Use only the first four media accumulator registers.
10641 Use all eight media accumulator registers.
10646 Pack VLIW instructions.
10651 Do not pack VLIW instructions.
10654 @opindex mno-eflags
10656 Do not mark ABI switches in e_flags.
10659 @opindex mcond-move
10661 Enable the use of conditional-move instructions (default).
10663 This switch is mainly for debugging the compiler and will likely be removed
10664 in a future version.
10666 @item -mno-cond-move
10667 @opindex mno-cond-move
10669 Disable the use of conditional-move instructions.
10671 This switch is mainly for debugging the compiler and will likely be removed
10672 in a future version.
10677 Enable the use of conditional set instructions (default).
10679 This switch is mainly for debugging the compiler and will likely be removed
10680 in a future version.
10685 Disable the use of conditional set instructions.
10687 This switch is mainly for debugging the compiler and will likely be removed
10688 in a future version.
10691 @opindex mcond-exec
10693 Enable the use of conditional execution (default).
10695 This switch is mainly for debugging the compiler and will likely be removed
10696 in a future version.
10698 @item -mno-cond-exec
10699 @opindex mno-cond-exec
10701 Disable the use of conditional execution.
10703 This switch is mainly for debugging the compiler and will likely be removed
10704 in a future version.
10706 @item -mvliw-branch
10707 @opindex mvliw-branch
10709 Run a pass to pack branches into VLIW instructions (default).
10711 This switch is mainly for debugging the compiler and will likely be removed
10712 in a future version.
10714 @item -mno-vliw-branch
10715 @opindex mno-vliw-branch
10717 Do not run a pass to pack branches into VLIW instructions.
10719 This switch is mainly for debugging the compiler and will likely be removed
10720 in a future version.
10722 @item -mmulti-cond-exec
10723 @opindex mmulti-cond-exec
10725 Enable optimization of @code{&&} and @code{||} in conditional execution
10728 This switch is mainly for debugging the compiler and will likely be removed
10729 in a future version.
10731 @item -mno-multi-cond-exec
10732 @opindex mno-multi-cond-exec
10734 Disable optimization of @code{&&} and @code{||} in conditional execution.
10736 This switch is mainly for debugging the compiler and will likely be removed
10737 in a future version.
10739 @item -mnested-cond-exec
10740 @opindex mnested-cond-exec
10742 Enable nested conditional execution optimizations (default).
10744 This switch is mainly for debugging the compiler and will likely be removed
10745 in a future version.
10747 @item -mno-nested-cond-exec
10748 @opindex mno-nested-cond-exec
10750 Disable nested conditional execution optimizations.
10752 This switch is mainly for debugging the compiler and will likely be removed
10753 in a future version.
10755 @item -mtomcat-stats
10756 @opindex mtomcat-stats
10758 Cause gas to print out tomcat statistics.
10760 @item -mcpu=@var{cpu}
10763 Select the processor type for which to generate code. Possible values are
10764 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
10769 @node Xtensa Options
10770 @subsection Xtensa Options
10771 @cindex Xtensa Options
10773 The Xtensa architecture is designed to support many different
10774 configurations. The compiler's default options can be set to match a
10775 particular Xtensa configuration by copying a configuration file into the
10776 GCC sources when building GCC@. The options below may be used to
10777 override the default options.
10781 @itemx -mlittle-endian
10782 @opindex mbig-endian
10783 @opindex mlittle-endian
10784 Specify big-endian or little-endian byte ordering for the target Xtensa
10788 @itemx -mno-density
10790 @opindex mno-density
10791 Enable or disable use of the optional Xtensa code density instructions.
10794 @itemx -mno-const16
10796 @opindex mno-const16
10797 Enable or disable use of @code{CONST16} instructions for loading
10798 constant values. The @code{CONST16} instruction is currently not a
10799 standard option from Tensilica. When enabled, @code{CONST16}
10800 instructions are always used in place of the standard @code{L32R}
10801 instructions. The use of @code{CONST16} is enabled by default only if
10802 the @code{L32R} instruction is not available.
10808 Enable or disable use of the Xtensa @code{ABS} instruction for absolute
10815 Enable or disable use of the Xtensa @code{ADDX} and @code{SUBX}
10822 Enable or disable use of the Xtensa MAC16 option. When enabled, GCC
10823 will generate MAC16 instructions from standard C code, with the
10824 limitation that it will use neither the MR register file nor any
10825 instruction that operates on the MR registers. When this option is
10826 disabled, GCC will translate 16-bit multiply/accumulate operations to a
10827 combination of core instructions and library calls, depending on whether
10828 any other multiplier options are enabled.
10834 Enable or disable use of the 16-bit integer multiplier option. When
10835 enabled, the compiler will generate 16-bit multiply instructions for
10836 multiplications of 16 bits or smaller in standard C code. When this
10837 option is disabled, the compiler will either use 32-bit multiply or
10838 MAC16 instructions if they are available or generate library calls to
10839 perform the multiply operations using shifts and adds.
10845 Enable or disable use of the 32-bit integer multiplier option. When
10846 enabled, the compiler will generate 32-bit multiply instructions for
10847 multiplications of 32 bits or smaller in standard C code. When this
10848 option is disabled, the compiler will generate library calls to perform
10849 the multiply operations using either shifts and adds or 16-bit multiply
10850 instructions if they are available.
10856 Enable or disable use of the optional normalization shift amount
10857 (@code{NSA}) instructions to implement the built-in @code{ffs} function.
10862 @opindex mno-minmax
10863 Enable or disable use of the optional minimum and maximum value
10870 Enable or disable use of the optional sign extend (@code{SEXT})
10874 @itemx -mno-booleans
10876 @opindex mno-booleans
10877 Enable or disable support for the boolean register file used by Xtensa
10878 coprocessors. This is not typically useful by itself but may be
10879 required for other options that make use of the boolean registers (e.g.,
10880 the floating-point option).
10883 @itemx -msoft-float
10884 @opindex mhard-float
10885 @opindex msoft-float
10886 Enable or disable use of the floating-point option. When enabled, GCC
10887 generates floating-point instructions for 32-bit @code{float}
10888 operations. When this option is disabled, GCC generates library calls
10889 to emulate 32-bit floating-point operations using integer instructions.
10890 Regardless of this option, 64-bit @code{double} operations are always
10891 emulated with calls to library functions.
10894 @itemx -mno-fused-madd
10895 @opindex mfused-madd
10896 @opindex mno-fused-madd
10897 Enable or disable use of fused multiply/add and multiply/subtract
10898 instructions in the floating-point option. This has no effect if the
10899 floating-point option is not also enabled. Disabling fused multiply/add
10900 and multiply/subtract instructions forces the compiler to use separate
10901 instructions for the multiply and add/subtract operations. This may be
10902 desirable in some cases where strict IEEE 754-compliant results are
10903 required: the fused multiply add/subtract instructions do not round the
10904 intermediate result, thereby producing results with @emph{more} bits of
10905 precision than specified by the IEEE standard. Disabling fused multiply
10906 add/subtract instructions also ensures that the program output is not
10907 sensitive to the compiler's ability to combine multiply and add/subtract
10910 @item -mtext-section-literals
10911 @itemx -mno-text-section-literals
10912 @opindex mtext-section-literals
10913 @opindex mno-text-section-literals
10914 Control the treatment of literal pools. The default is
10915 @option{-mno-text-section-literals}, which places literals in a separate
10916 section in the output file. This allows the literal pool to be placed
10917 in a data RAM/ROM, and it also allows the linker to combine literal
10918 pools from separate object files to remove redundant literals and
10919 improve code size. With @option{-mtext-section-literals}, the literals
10920 are interspersed in the text section in order to keep them as close as
10921 possible to their references. This may be necessary for large assembly
10924 @item -mtarget-align
10925 @itemx -mno-target-align
10926 @opindex mtarget-align
10927 @opindex mno-target-align
10928 When this option is enabled, GCC instructs the assembler to
10929 automatically align instructions to reduce branch penalties at the
10930 expense of some code density. The assembler attempts to widen density
10931 instructions to align branch targets and the instructions following call
10932 instructions. If there are not enough preceding safe density
10933 instructions to align a target, no widening will be performed. The
10934 default is @option{-mtarget-align}. These options do not affect the
10935 treatment of auto-aligned instructions like @code{LOOP}, which the
10936 assembler will always align, either by widening density instructions or
10937 by inserting no-op instructions.
10940 @itemx -mno-longcalls
10941 @opindex mlongcalls
10942 @opindex mno-longcalls
10943 When this option is enabled, GCC instructs the assembler to translate
10944 direct calls to indirect calls unless it can determine that the target
10945 of a direct call is in the range allowed by the call instruction. This
10946 translation typically occurs for calls to functions in other source
10947 files. Specifically, the assembler translates a direct @code{CALL}
10948 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
10949 The default is @option{-mno-longcalls}. This option should be used in
10950 programs where the call target can potentially be out of range. This
10951 option is implemented in the assembler, not the compiler, so the
10952 assembly code generated by GCC will still show direct call
10953 instructions---look at the disassembled object code to see the actual
10954 instructions. Note that the assembler will use an indirect call for
10955 every cross-file call, not just those that really will be out of range.
10958 @node Code Gen Options
10959 @section Options for Code Generation Conventions
10960 @cindex code generation conventions
10961 @cindex options, code generation
10962 @cindex run-time options
10964 These machine-independent options control the interface conventions
10965 used in code generation.
10967 Most of them have both positive and negative forms; the negative form
10968 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
10969 one of the forms is listed---the one which is not the default. You
10970 can figure out the other form by either removing @samp{no-} or adding
10974 @item -fbounds-check
10975 @opindex fbounds-check
10976 For front-ends that support it, generate additional code to check that
10977 indices used to access arrays are within the declared range. This is
10978 currently only supported by the Java and Fortran 77 front-ends, where
10979 this option defaults to true and false respectively.
10983 This option generates traps for signed overflow on addition, subtraction,
10984 multiplication operations.
10988 This option instructs the compiler to assume that signed arithmetic
10989 overflow of addition, subtraction and multiplication wraps around
10990 using twos-complement representation. This flag enables some optimizations
10991 and disables other. This option is enabled by default for the Java
10992 front-end, as required by the Java language specification.
10995 @opindex fexceptions
10996 Enable exception handling. Generates extra code needed to propagate
10997 exceptions. For some targets, this implies GCC will generate frame
10998 unwind information for all functions, which can produce significant data
10999 size overhead, although it does not affect execution. If you do not
11000 specify this option, GCC will enable it by default for languages like
11001 C++ which normally require exception handling, and disable it for
11002 languages like C that do not normally require it. However, you may need
11003 to enable this option when compiling C code that needs to interoperate
11004 properly with exception handlers written in C++. You may also wish to
11005 disable this option if you are compiling older C++ programs that don't
11006 use exception handling.
11008 @item -fnon-call-exceptions
11009 @opindex fnon-call-exceptions
11010 Generate code that allows trapping instructions to throw exceptions.
11011 Note that this requires platform-specific runtime support that does
11012 not exist everywhere. Moreover, it only allows @emph{trapping}
11013 instructions to throw exceptions, i.e.@: memory references or floating
11014 point instructions. It does not allow exceptions to be thrown from
11015 arbitrary signal handlers such as @code{SIGALRM}.
11017 @item -funwind-tables
11018 @opindex funwind-tables
11019 Similar to @option{-fexceptions}, except that it will just generate any needed
11020 static data, but will not affect the generated code in any other way.
11021 You will normally not enable this option; instead, a language processor
11022 that needs this handling would enable it on your behalf.
11024 @item -fasynchronous-unwind-tables
11025 @opindex funwind-tables
11026 Generate unwind table in dwarf2 format, if supported by target machine. The
11027 table is exact at each instruction boundary, so it can be used for stack
11028 unwinding from asynchronous events (such as debugger or garbage collector).
11030 @item -fpcc-struct-return
11031 @opindex fpcc-struct-return
11032 Return ``short'' @code{struct} and @code{union} values in memory like
11033 longer ones, rather than in registers. This convention is less
11034 efficient, but it has the advantage of allowing intercallability between
11035 GCC-compiled files and files compiled with other compilers, particularly
11036 the Portable C Compiler (pcc).
11038 The precise convention for returning structures in memory depends
11039 on the target configuration macros.
11041 Short structures and unions are those whose size and alignment match
11042 that of some integer type.
11044 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11045 switch is not binary compatible with code compiled with the
11046 @option{-freg-struct-return} switch.
11047 Use it to conform to a non-default application binary interface.
11049 @item -freg-struct-return
11050 @opindex freg-struct-return
11051 Return @code{struct} and @code{union} values in registers when possible.
11052 This is more efficient for small structures than
11053 @option{-fpcc-struct-return}.
11055 If you specify neither @option{-fpcc-struct-return} nor
11056 @option{-freg-struct-return}, GCC defaults to whichever convention is
11057 standard for the target. If there is no standard convention, GCC
11058 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11059 the principal compiler. In those cases, we can choose the standard, and
11060 we chose the more efficient register return alternative.
11062 @strong{Warning:} code compiled with the @option{-freg-struct-return}
11063 switch is not binary compatible with code compiled with the
11064 @option{-fpcc-struct-return} switch.
11065 Use it to conform to a non-default application binary interface.
11067 @item -fshort-enums
11068 @opindex fshort-enums
11069 Allocate to an @code{enum} type only as many bytes as it needs for the
11070 declared range of possible values. Specifically, the @code{enum} type
11071 will be equivalent to the smallest integer type which has enough room.
11073 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11074 code that is not binary compatible with code generated without that switch.
11075 Use it to conform to a non-default application binary interface.
11077 @item -fshort-double
11078 @opindex fshort-double
11079 Use the same size for @code{double} as for @code{float}.
11081 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
11082 code that is not binary compatible with code generated without that switch.
11083 Use it to conform to a non-default application binary interface.
11085 @item -fshort-wchar
11086 @opindex fshort-wchar
11087 Override the underlying type for @samp{wchar_t} to be @samp{short
11088 unsigned int} instead of the default for the target. This option is
11089 useful for building programs to run under WINE@.
11091 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11092 code that is not binary compatible with code generated without that switch.
11093 Use it to conform to a non-default application binary interface.
11095 @item -fshared-data
11096 @opindex fshared-data
11097 Requests that the data and non-@code{const} variables of this
11098 compilation be shared data rather than private data. The distinction
11099 makes sense only on certain operating systems, where shared data is
11100 shared between processes running the same program, while private data
11101 exists in one copy per process.
11104 @opindex fno-common
11105 In C, allocate even uninitialized global variables in the data section of the
11106 object file, rather than generating them as common blocks. This has the
11107 effect that if the same variable is declared (without @code{extern}) in
11108 two different compilations, you will get an error when you link them.
11109 The only reason this might be useful is if you wish to verify that the
11110 program will work on other systems which always work this way.
11114 Ignore the @samp{#ident} directive.
11116 @item -fno-gnu-linker
11117 @opindex fno-gnu-linker
11118 Do not output global initializations (such as C++ constructors and
11119 destructors) in the form used by the GNU linker (on systems where the GNU
11120 linker is the standard method of handling them). Use this option when
11121 you want to use a non-GNU linker, which also requires using the
11122 @command{collect2} program to make sure the system linker includes
11123 constructors and destructors. (@command{collect2} is included in the GCC
11124 distribution.) For systems which @emph{must} use @command{collect2}, the
11125 compiler driver @command{gcc} is configured to do this automatically.
11127 @item -finhibit-size-directive
11128 @opindex finhibit-size-directive
11129 Don't output a @code{.size} assembler directive, or anything else that
11130 would cause trouble if the function is split in the middle, and the
11131 two halves are placed at locations far apart in memory. This option is
11132 used when compiling @file{crtstuff.c}; you should not need to use it
11135 @item -fverbose-asm
11136 @opindex fverbose-asm
11137 Put extra commentary information in the generated assembly code to
11138 make it more readable. This option is generally only of use to those
11139 who actually need to read the generated assembly code (perhaps while
11140 debugging the compiler itself).
11142 @option{-fno-verbose-asm}, the default, causes the
11143 extra information to be omitted and is useful when comparing two assembler
11148 @cindex global offset table
11150 Generate position-independent code (PIC) suitable for use in a shared
11151 library, if supported for the target machine. Such code accesses all
11152 constant addresses through a global offset table (GOT)@. The dynamic
11153 loader resolves the GOT entries when the program starts (the dynamic
11154 loader is not part of GCC; it is part of the operating system). If
11155 the GOT size for the linked executable exceeds a machine-specific
11156 maximum size, you get an error message from the linker indicating that
11157 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11158 instead. (These maximums are 16k on the m88k, 8k on the SPARC, and 32k
11159 on the m68k and RS/6000. The 386 has no such limit.)
11161 Position-independent code requires special support, and therefore works
11162 only on certain machines. For the 386, GCC supports PIC for System V
11163 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11164 position-independent.
11168 If supported for the target machine, emit position-independent code,
11169 suitable for dynamic linking and avoiding any limit on the size of the
11170 global offset table. This option makes a difference on the m68k, m88k,
11173 Position-independent code requires special support, and therefore works
11174 only on certain machines.
11180 These options are similar to @option{-fpic} and @option{-fPIC}, but
11181 generated position independent code can be only linked into executables.
11182 Usually these options are used when @option{-pie} GCC option will be
11183 used during linking.
11185 @item -ffixed-@var{reg}
11187 Treat the register named @var{reg} as a fixed register; generated code
11188 should never refer to it (except perhaps as a stack pointer, frame
11189 pointer or in some other fixed role).
11191 @var{reg} must be the name of a register. The register names accepted
11192 are machine-specific and are defined in the @code{REGISTER_NAMES}
11193 macro in the machine description macro file.
11195 This flag does not have a negative form, because it specifies a
11198 @item -fcall-used-@var{reg}
11199 @opindex fcall-used
11200 Treat the register named @var{reg} as an allocable register that is
11201 clobbered by function calls. It may be allocated for temporaries or
11202 variables that do not live across a call. Functions compiled this way
11203 will not save and restore the register @var{reg}.
11205 It is an error to used this flag with the frame pointer or stack pointer.
11206 Use of this flag for other registers that have fixed pervasive roles in
11207 the machine's execution model will produce disastrous results.
11209 This flag does not have a negative form, because it specifies a
11212 @item -fcall-saved-@var{reg}
11213 @opindex fcall-saved
11214 Treat the register named @var{reg} as an allocable register saved by
11215 functions. It may be allocated even for temporaries or variables that
11216 live across a call. Functions compiled this way will save and restore
11217 the register @var{reg} if they use it.
11219 It is an error to used this flag with the frame pointer or stack pointer.
11220 Use of this flag for other registers that have fixed pervasive roles in
11221 the machine's execution model will produce disastrous results.
11223 A different sort of disaster will result from the use of this flag for
11224 a register in which function values may be returned.
11226 This flag does not have a negative form, because it specifies a
11229 @item -fpack-struct
11230 @opindex fpack-struct
11231 Pack all structure members together without holes.
11233 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11234 code that is not binary compatible with code generated without that switch.
11235 Additionally, it makes the code suboptimal.
11236 Use it to conform to a non-default application binary interface.
11238 @item -finstrument-functions
11239 @opindex finstrument-functions
11240 Generate instrumentation calls for entry and exit to functions. Just
11241 after function entry and just before function exit, the following
11242 profiling functions will be called with the address of the current
11243 function and its call site. (On some platforms,
11244 @code{__builtin_return_address} does not work beyond the current
11245 function, so the call site information may not be available to the
11246 profiling functions otherwise.)
11249 void __cyg_profile_func_enter (void *this_fn,
11251 void __cyg_profile_func_exit (void *this_fn,
11255 The first argument is the address of the start of the current function,
11256 which may be looked up exactly in the symbol table.
11258 This instrumentation is also done for functions expanded inline in other
11259 functions. The profiling calls will indicate where, conceptually, the
11260 inline function is entered and exited. This means that addressable
11261 versions of such functions must be available. If all your uses of a
11262 function are expanded inline, this may mean an additional expansion of
11263 code size. If you use @samp{extern inline} in your C code, an
11264 addressable version of such functions must be provided. (This is
11265 normally the case anyways, but if you get lucky and the optimizer always
11266 expands the functions inline, you might have gotten away without
11267 providing static copies.)
11269 A function may be given the attribute @code{no_instrument_function}, in
11270 which case this instrumentation will not be done. This can be used, for
11271 example, for the profiling functions listed above, high-priority
11272 interrupt routines, and any functions from which the profiling functions
11273 cannot safely be called (perhaps signal handlers, if the profiling
11274 routines generate output or allocate memory).
11276 @item -fstack-check
11277 @opindex fstack-check
11278 Generate code to verify that you do not go beyond the boundary of the
11279 stack. You should specify this flag if you are running in an
11280 environment with multiple threads, but only rarely need to specify it in
11281 a single-threaded environment since stack overflow is automatically
11282 detected on nearly all systems if there is only one stack.
11284 Note that this switch does not actually cause checking to be done; the
11285 operating system must do that. The switch causes generation of code
11286 to ensure that the operating system sees the stack being extended.
11288 @item -fstack-limit-register=@var{reg}
11289 @itemx -fstack-limit-symbol=@var{sym}
11290 @itemx -fno-stack-limit
11291 @opindex fstack-limit-register
11292 @opindex fstack-limit-symbol
11293 @opindex fno-stack-limit
11294 Generate code to ensure that the stack does not grow beyond a certain value,
11295 either the value of a register or the address of a symbol. If the stack
11296 would grow beyond the value, a signal is raised. For most targets,
11297 the signal is raised before the stack overruns the boundary, so
11298 it is possible to catch the signal without taking special precautions.
11300 For instance, if the stack starts at absolute address @samp{0x80000000}
11301 and grows downwards, you can use the flags
11302 @option{-fstack-limit-symbol=__stack_limit} and
11303 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11304 of 128KB@. Note that this may only work with the GNU linker.
11306 @cindex aliasing of parameters
11307 @cindex parameters, aliased
11308 @item -fargument-alias
11309 @itemx -fargument-noalias
11310 @itemx -fargument-noalias-global
11311 @opindex fargument-alias
11312 @opindex fargument-noalias
11313 @opindex fargument-noalias-global
11314 Specify the possible relationships among parameters and between
11315 parameters and global data.
11317 @option{-fargument-alias} specifies that arguments (parameters) may
11318 alias each other and may alias global storage.@*
11319 @option{-fargument-noalias} specifies that arguments do not alias
11320 each other, but may alias global storage.@*
11321 @option{-fargument-noalias-global} specifies that arguments do not
11322 alias each other and do not alias global storage.
11324 Each language will automatically use whatever option is required by
11325 the language standard. You should not need to use these options yourself.
11327 @item -fleading-underscore
11328 @opindex fleading-underscore
11329 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11330 change the way C symbols are represented in the object file. One use
11331 is to help link with legacy assembly code.
11333 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11334 generate code that is not binary compatible with code generated without that
11335 switch. Use it to conform to a non-default application binary interface.
11336 Not all targets provide complete support for this switch.
11338 @item -ftls-model=@var{model}
11339 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11340 The @var{model} argument should be one of @code{global-dynamic},
11341 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11343 The default without @option{-fpic} is @code{initial-exec}; with
11344 @option{-fpic} the default is @code{global-dynamic}.
11349 @node Environment Variables
11350 @section Environment Variables Affecting GCC
11351 @cindex environment variables
11353 @c man begin ENVIRONMENT
11354 This section describes several environment variables that affect how GCC
11355 operates. Some of them work by specifying directories or prefixes to use
11356 when searching for various kinds of files. Some are used to specify other
11357 aspects of the compilation environment.
11359 Note that you can also specify places to search using options such as
11360 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11361 take precedence over places specified using environment variables, which
11362 in turn take precedence over those specified by the configuration of GCC@.
11363 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11364 GNU Compiler Collection (GCC) Internals}.
11369 @c @itemx LC_COLLATE
11371 @c @itemx LC_MONETARY
11372 @c @itemx LC_NUMERIC
11377 @c @findex LC_COLLATE
11378 @findex LC_MESSAGES
11379 @c @findex LC_MONETARY
11380 @c @findex LC_NUMERIC
11384 These environment variables control the way that GCC uses
11385 localization information that allow GCC to work with different
11386 national conventions. GCC inspects the locale categories
11387 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11388 so. These locale categories can be set to any value supported by your
11389 installation. A typical value is @samp{en_UK} for English in the United
11392 The @env{LC_CTYPE} environment variable specifies character
11393 classification. GCC uses it to determine the character boundaries in
11394 a string; this is needed for some multibyte encodings that contain quote
11395 and escape characters that would otherwise be interpreted as a string
11398 The @env{LC_MESSAGES} environment variable specifies the language to
11399 use in diagnostic messages.
11401 If the @env{LC_ALL} environment variable is set, it overrides the value
11402 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11403 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11404 environment variable. If none of these variables are set, GCC
11405 defaults to traditional C English behavior.
11409 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11410 files. GCC uses temporary files to hold the output of one stage of
11411 compilation which is to be used as input to the next stage: for example,
11412 the output of the preprocessor, which is the input to the compiler
11415 @item GCC_EXEC_PREFIX
11416 @findex GCC_EXEC_PREFIX
11417 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11418 names of the subprograms executed by the compiler. No slash is added
11419 when this prefix is combined with the name of a subprogram, but you can
11420 specify a prefix that ends with a slash if you wish.
11422 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11423 an appropriate prefix to use based on the pathname it was invoked with.
11425 If GCC cannot find the subprogram using the specified prefix, it
11426 tries looking in the usual places for the subprogram.
11428 The default value of @env{GCC_EXEC_PREFIX} is
11429 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
11430 of @code{prefix} when you ran the @file{configure} script.
11432 Other prefixes specified with @option{-B} take precedence over this prefix.
11434 This prefix is also used for finding files such as @file{crt0.o} that are
11437 In addition, the prefix is used in an unusual way in finding the
11438 directories to search for header files. For each of the standard
11439 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
11440 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11441 replacing that beginning with the specified prefix to produce an
11442 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11443 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11444 These alternate directories are searched first; the standard directories
11447 @item COMPILER_PATH
11448 @findex COMPILER_PATH
11449 The value of @env{COMPILER_PATH} is a colon-separated list of
11450 directories, much like @env{PATH}. GCC tries the directories thus
11451 specified when searching for subprograms, if it can't find the
11452 subprograms using @env{GCC_EXEC_PREFIX}.
11455 @findex LIBRARY_PATH
11456 The value of @env{LIBRARY_PATH} is a colon-separated list of
11457 directories, much like @env{PATH}. When configured as a native compiler,
11458 GCC tries the directories thus specified when searching for special
11459 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11460 using GCC also uses these directories when searching for ordinary
11461 libraries for the @option{-l} option (but directories specified with
11462 @option{-L} come first).
11466 @cindex locale definition
11467 This variable is used to pass locale information to the compiler. One way in
11468 which this information is used is to determine the character set to be used
11469 when character literals, string literals and comments are parsed in C and C++.
11470 When the compiler is configured to allow multibyte characters,
11471 the following values for @env{LANG} are recognized:
11475 Recognize JIS characters.
11477 Recognize SJIS characters.
11479 Recognize EUCJP characters.
11482 If @env{LANG} is not defined, or if it has some other value, then the
11483 compiler will use mblen and mbtowc as defined by the default locale to
11484 recognize and translate multibyte characters.
11488 Some additional environments variables affect the behavior of the
11491 @include cppenv.texi
11495 @node Precompiled Headers
11496 @section Using Precompiled Headers
11497 @cindex precompiled headers
11498 @cindex speed of compilation
11500 Often large projects have many header files that are included in every
11501 source file. The time the compiler takes to process these header files
11502 over and over again can account for nearly all of the time required to
11503 build the project. To make builds faster, GCC allows users to
11504 `precompile' a header file; then, if builds can use the precompiled
11505 header file they will be much faster.
11507 To create a precompiled header file, simply compile it as you would any
11508 other file, if necessary using the @option{-x} option to make the driver
11509 treat it as a C or C++ header file. You will probably want to use a
11510 tool like @command{make} to keep the precompiled header up-to-date when
11511 the headers it contains change.
11513 A precompiled header file will be searched for when @code{#include} is
11514 seen in the compilation. As it searches for the included file
11515 (@pxref{Search Path,,Search Path,cpp.info,The C Preprocessor}) the
11516 compiler looks for a precompiled header in each directory just before it
11517 looks for the include file in that directory. The name searched for is
11518 the name specified in the @code{#include} with @samp{.gch} appended. If
11519 the precompiled header file can't be used, it is ignored.
11521 For instance, if you have @code{#include "all.h"}, and you have
11522 @file{all.h.gch} in the same directory as @file{all.h}, then the
11523 precompiled header file will be used if possible, and the original
11524 header will be used otherwise.
11526 Alternatively, you might decide to put the precompiled header file in a
11527 directory and use @option{-I} to ensure that directory is searched
11528 before (or instead of) the directory containing the original header.
11529 Then, if you want to check that the precompiled header file is always
11530 used, you can put a file of the same name as the original header in this
11531 directory containing an @code{#error} command.
11533 This also works with @option{-include}. So yet another way to use
11534 precompiled headers, good for projects not designed with precompiled
11535 header files in mind, is to simply take most of the header files used by
11536 a project, include them from another header file, precompile that header
11537 file, and @option{-include} the precompiled header. If the header files
11538 have guards against multiple inclusion, they will be skipped because
11539 they've already been included (in the precompiled header).
11541 If you need to precompile the same header file for different
11542 languages, targets, or compiler options, you can instead make a
11543 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11544 header in the directory. (It doesn't matter what you call the files
11545 in the directory, every precompiled header in the directory will be
11546 considered.) The first precompiled header encountered in the
11547 directory that is valid for this compilation will be used; they're
11548 searched in no particular order.
11550 There are many other possibilities, limited only by your imagination,
11551 good sense, and the constraints of your build system.
11553 A precompiled header file can be used only when these conditions apply:
11557 Only one precompiled header can be used in a particular compilation.
11559 A precompiled header can't be used once the first C token is seen. You
11560 can have preprocessor directives before a precompiled header; you can
11561 even include a precompiled header from inside another header, so long as
11562 there are no C tokens before the @code{#include}.
11564 The precompiled header file must be produced for the same language as
11565 the current compilation. You can't use a C precompiled header for a C++
11568 The precompiled header file must be produced by the same compiler
11569 version and configuration as the current compilation is using.
11570 The easiest way to guarantee this is to use the same compiler binary
11571 for creating and using precompiled headers.
11573 Any macros defined before the precompiled header (including with
11574 @option{-D}) must either be defined in the same way as when the
11575 precompiled header was generated, or must not affect the precompiled
11576 header, which usually means that the they don't appear in the
11577 precompiled header at all.
11579 Certain command-line options must be defined in the same way as when the
11580 precompiled header was generated. At present, it's not clear which
11581 options are safe to change and which are not; the safest choice is to
11582 use exactly the same options when generating and using the precompiled
11586 For all of these but the last, the compiler will automatically ignore
11587 the precompiled header if the conditions aren't met. For the last item,
11588 some option changes will cause the precompiled header to be rejected,
11589 but not all incompatible option combinations have yet been found. If
11590 you find a new incompatible combination, please consider filing a bug
11591 report, see @ref{Bugs}.
11593 @node Running Protoize
11594 @section Running Protoize
11596 The program @code{protoize} is an optional part of GCC@. You can use
11597 it to add prototypes to a program, thus converting the program to ISO
11598 C in one respect. The companion program @code{unprotoize} does the
11599 reverse: it removes argument types from any prototypes that are found.
11601 When you run these programs, you must specify a set of source files as
11602 command line arguments. The conversion programs start out by compiling
11603 these files to see what functions they define. The information gathered
11604 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11606 After scanning comes actual conversion. The specified files are all
11607 eligible to be converted; any files they include (whether sources or
11608 just headers) are eligible as well.
11610 But not all the eligible files are converted. By default,
11611 @code{protoize} and @code{unprotoize} convert only source and header
11612 files in the current directory. You can specify additional directories
11613 whose files should be converted with the @option{-d @var{directory}}
11614 option. You can also specify particular files to exclude with the
11615 @option{-x @var{file}} option. A file is converted if it is eligible, its
11616 directory name matches one of the specified directory names, and its
11617 name within the directory has not been excluded.
11619 Basic conversion with @code{protoize} consists of rewriting most
11620 function definitions and function declarations to specify the types of
11621 the arguments. The only ones not rewritten are those for varargs
11624 @code{protoize} optionally inserts prototype declarations at the
11625 beginning of the source file, to make them available for any calls that
11626 precede the function's definition. Or it can insert prototype
11627 declarations with block scope in the blocks where undeclared functions
11630 Basic conversion with @code{unprotoize} consists of rewriting most
11631 function declarations to remove any argument types, and rewriting
11632 function definitions to the old-style pre-ISO form.
11634 Both conversion programs print a warning for any function declaration or
11635 definition that they can't convert. You can suppress these warnings
11638 The output from @code{protoize} or @code{unprotoize} replaces the
11639 original source file. The original file is renamed to a name ending
11640 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11641 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11642 for DOS) file already exists, then the source file is simply discarded.
11644 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11645 scan the program and collect information about the functions it uses.
11646 So neither of these programs will work until GCC is installed.
11648 Here is a table of the options you can use with @code{protoize} and
11649 @code{unprotoize}. Each option works with both programs unless
11653 @item -B @var{directory}
11654 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11655 usual directory (normally @file{/usr/local/lib}). This file contains
11656 prototype information about standard system functions. This option
11657 applies only to @code{protoize}.
11659 @item -c @var{compilation-options}
11660 Use @var{compilation-options} as the options when running @command{gcc} to
11661 produce the @samp{.X} files. The special option @option{-aux-info} is
11662 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11664 Note that the compilation options must be given as a single argument to
11665 @code{protoize} or @code{unprotoize}. If you want to specify several
11666 @command{gcc} options, you must quote the entire set of compilation options
11667 to make them a single word in the shell.
11669 There are certain @command{gcc} arguments that you cannot use, because they
11670 would produce the wrong kind of output. These include @option{-g},
11671 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11672 the @var{compilation-options}, they are ignored.
11675 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11676 systems) instead of @samp{.c}. This is convenient if you are converting
11677 a C program to C++. This option applies only to @code{protoize}.
11680 Add explicit global declarations. This means inserting explicit
11681 declarations at the beginning of each source file for each function
11682 that is called in the file and was not declared. These declarations
11683 precede the first function definition that contains a call to an
11684 undeclared function. This option applies only to @code{protoize}.
11686 @item -i @var{string}
11687 Indent old-style parameter declarations with the string @var{string}.
11688 This option applies only to @code{protoize}.
11690 @code{unprotoize} converts prototyped function definitions to old-style
11691 function definitions, where the arguments are declared between the
11692 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11693 uses five spaces as the indentation. If you want to indent with just
11694 one space instead, use @option{-i " "}.
11697 Keep the @samp{.X} files. Normally, they are deleted after conversion
11701 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11702 a prototype declaration for each function in each block which calls the
11703 function without any declaration. This option applies only to
11707 Make no real changes. This mode just prints information about the conversions
11708 that would have been done without @option{-n}.
11711 Make no @samp{.save} files. The original files are simply deleted.
11712 Use this option with caution.
11714 @item -p @var{program}
11715 Use the program @var{program} as the compiler. Normally, the name
11716 @file{gcc} is used.
11719 Work quietly. Most warnings are suppressed.
11722 Print the version number, just like @option{-v} for @command{gcc}.
11725 If you need special compiler options to compile one of your program's
11726 source files, then you should generate that file's @samp{.X} file
11727 specially, by running @command{gcc} on that source file with the
11728 appropriate options and the option @option{-aux-info}. Then run
11729 @code{protoize} on the entire set of files. @code{protoize} will use
11730 the existing @samp{.X} file because it is newer than the source file.
11734 gcc -Dfoo=bar file1.c -aux-info file1.X
11739 You need to include the special files along with the rest in the
11740 @code{protoize} command, even though their @samp{.X} files already
11741 exist, because otherwise they won't get converted.
11743 @xref{Protoize Caveats}, for more information on how to use
11744 @code{protoize} successfully.