1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001 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 Free Software Foundation, Inc.
11 Permission is granted to make and distribute verbatim copies of this
12 manual provided the copyright notice and this permission notice are
13 preserved on all copies.
15 Permission is granted to copy and distribute modified versions of this
16 manual under the conditions for verbatim copying, provided also that the
17 entire resulting derived work is distributed under the terms of a
18 permission notice identical to this one.
20 Permission is granted to copy and distribute translations of this manual
21 into another language, under the above conditions for modified versions,
22 except that this permission notice may be included in translations
23 approved by the Free Software Foundation instead of in the original
26 @c Set file name and title for the man page.
28 @settitle GNU project C and C++ compiler
30 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
31 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
32 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
33 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
34 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
35 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
36 [@option{-o} @var{outfile}] @var{infile}@dots{}
38 Only the most useful options are listed here; see below for the
39 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
42 cpp(1), gcov(1), g77(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
43 and the Info entries for @file{gcc}, @file{cpp}, @file{g77}, @file{as},
44 @file{ld}, @file{binutils} and @file{gdb}.
47 For instructions on reporting bugs, see
48 @w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug}
49 script to report bugs is recommended.
52 See the Info entry for @file{gcc}, or
53 @w{@uref{http://gcc.gnu.org/thanks.html}}, for contributors to GCC@.
58 @chapter GCC Command Options
59 @cindex GCC command options
60 @cindex command options
61 @cindex options, GCC command
63 @c man begin DESCRIPTION
65 When you invoke GCC, it normally does preprocessing, compilation,
66 assembly and linking. The ``overall options'' allow you to stop this
67 process at an intermediate stage. For example, the @option{-c} option
68 says not to run the linker. Then the output consists of object files
69 output by the assembler.
71 Other options are passed on to one stage of processing. Some options
72 control the preprocessor and others the compiler itself. Yet other
73 options control the assembler and linker; most of these are not
74 documented here, since you rarely need to use any of them.
76 @cindex C compilation options
77 Most of the command line options that you can use with GCC are useful
78 for C programs; when an option is only useful with another language
79 (usually C++), the explanation says so explicitly. If the description
80 for a particular option does not mention a source language, you can use
81 that option with all supported languages.
83 @cindex C++ compilation options
84 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
85 options for compiling C++ programs.
87 @cindex grouping options
88 @cindex options, grouping
89 The @command{gcc} program accepts options and file names as operands. Many
90 options have multi-letter names; therefore multiple single-letter options
91 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
94 @cindex order of options
95 @cindex options, order
96 You can mix options and other arguments. For the most part, the order
97 you use doesn't matter. Order does matter when you use several options
98 of the same kind; for example, if you specify @option{-L} more than once,
99 the directories are searched in the order specified.
101 Many options have long names starting with @samp{-f} or with
102 @samp{-W}---for example, @option{-fforce-mem},
103 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
104 these have both positive and negative forms; the negative form of
105 @option{-ffoo} would be @option{-fno-foo}. This manual documents
106 only one of these two forms, whichever one is not the default.
110 @xref{Option Index}, for an index to GCC's options.
113 * Option Summary:: Brief list of all options, without explanations.
114 * Overall Options:: Controlling the kind of output:
115 an executable, object files, assembler files,
116 or preprocessed source.
117 * Invoking G++:: Compiling C++ programs.
118 * C Dialect Options:: Controlling the variant of C language compiled.
119 * C++ Dialect Options:: Variations on C++.
120 * Objective-C Dialect Options:: Variations on Objective-C.
121 * Language Independent Options:: Controlling how diagnostics should be
123 * Warning Options:: How picky should the compiler be?
124 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
125 * Optimize Options:: How much optimization?
126 * Preprocessor Options:: Controlling header files and macro definitions.
127 Also, getting dependency information for Make.
128 * Assembler Options:: Passing options to the assembler.
129 * Link Options:: Specifying libraries and so on.
130 * Directory Options:: Where to find header files and libraries.
131 Where to find the compiler executable files.
132 * Spec Files:: How to pass switches to sub-processes.
133 * Target Options:: Running a cross-compiler, or an old version of GCC.
134 * Submodel Options:: Specifying minor hardware or convention variations,
135 such as 68010 vs 68020.
136 * Code Gen Options:: Specifying conventions for function calls, data layout
138 * Environment Variables:: Env vars that affect GCC.
139 * Running Protoize:: Automatically adding or removing function prototypes.
145 @section Option Summary
147 Here is a summary of all the options, grouped by type. Explanations are
148 in the following sections.
151 @item Overall Options
152 @xref{Overall Options,,Options Controlling the Kind of Output}.
154 -c -S -E -o @var{file} -pipe -pass-exit-codes -x @var{language} @gol
155 -v --target-help --help}
157 @item C Language Options
158 @xref{C Dialect Options,,Options Controlling C Dialect}.
160 -ansi -std=@var{standard} -aux-info @var{filename} @gol
161 -fno-asm -fno-builtin @gol
162 -fhosted -ffreestanding @gol
163 -trigraphs -traditional -traditional-cpp @gol
164 -fallow-single-precision -fcond-mismatch @gol
165 -fsigned-bitfields -fsigned-char @gol
166 -funsigned-bitfields -funsigned-char @gol
167 -fwritable-strings -fshort-wchar}
169 @item C++ Language Options
170 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
172 -fno-access-control -fcheck-new -fconserve-space @gol
173 -fno-const-strings -fdollars-in-identifiers @gol
174 -fno-elide-constructors @gol
175 -fno-enforce-eh-specs -fexternal-templates @gol
176 -falt-external-templates @gol
177 -ffor-scope -fno-for-scope -fno-gnu-keywords -fno-honor-std @gol
178 -fno-implicit-templates @gol
179 -fno-implicit-inline-templates @gol
180 -fno-implement-inlines -fms-extensions @gol
181 -fno-nonansi-builtins -fno-operator-names @gol
182 -fno-optional-diags -fpermissive @gol
183 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
184 -fuse-cxa-atexit -fvtable-gc -fno-weak -nostdinc++ @gol
185 -fno-default-inline -Wctor-dtor-privacy @gol
186 -Wnon-virtual-dtor -Wreorder @gol
187 -Weffc++ -Wno-deprecated @gol
188 -Wno-non-template-friend -Wold-style-cast @gol
189 -Woverloaded-virtual -Wno-pmf-conversions @gol
190 -Wsign-promo -Wsynth}
192 @item Objective-C Language Options
193 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
195 -fconstant-string-class=@var{class-name} @gol
196 -fgnu-runtime -fnext-runtime -gen-decls @gol
197 -Wno-protocol -Wselector}
199 @item Language Independent Options
200 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
202 -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}.
208 -fsyntax-only -pedantic -pedantic-errors @gol
209 -w -W -Wall -Waggregate-return @gol
210 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
211 -Wconversion -Wdisabled-optimization -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 @gol
218 -Wlarger-than-@var{len} -Wlong-long @gol
219 -Wmain -Wmissing-braces -Wmissing-declarations @gol
220 -Wmissing-format-attribute -Wmissing-noreturn @gol
221 -Wmultichar -Wno-format-extra-args -Wno-format-y2k @gol
222 -Wno-import -Wpacked -Wpadded @gol
223 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
224 -Wreturn-type -Wsequence-point -Wshadow @gol
225 -Wsign-compare -Wswitch -Wsystem-headers @gol
226 -Wtrigraphs -Wundef -Wuninitialized @gol
227 -Wunknown-pragmas -Wunreachable-code @gol
228 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
229 -Wunused-value -Wunused-variable -Wwrite-strings}
231 @item C-only Warning Options
233 -Wbad-function-cast -Wmissing-prototypes -Wnested-externs @gol
234 -Wstrict-prototypes -Wtraditional}
236 @item Debugging Options
237 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
239 -a -ax -d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
240 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
241 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
242 -fdump-ast-original@r{[}-@var{n}@r{]} -fdump-ast-optimized@r{[}-@var{n}@r{]} @gol
243 -fdump-ast-inlined@r{[}-@var{n}@r{]} @gol
244 -fmem-report -fpretend-float @gol
245 -fprofile-arcs -ftest-coverage -ftime-report @gol
246 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
247 -ggdb -gstabs -gstabs+ -gxcoff -gxcoff+ @gol
248 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
249 -print-multi-directory -print-multi-lib @gol
250 -print-prog-name=@var{program} -print-search-dirs -Q @gol
253 @item Optimization Options
254 @xref{Optimize Options,,Options that Control Optimization}.
256 -falign-functions=@var{n} -falign-jumps=@var{n} @gol
257 -falign-labels=@var{n} -falign-loops=@var{n} @gol
258 -fbranch-probabilities -fcaller-saves @gol
259 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
260 -fdelayed-branch -fdelete-null-pointer-checks @gol
261 -fexpensive-optimizations -ffast-math -ffloat-store @gol
262 -fforce-addr -fforce-mem -ffunction-sections @gol
263 -fgcse -fgcse-lm -fgcse-sm @gol
264 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
265 -fkeep-static-consts -fmove-all-movables @gol
266 -fno-default-inline -fno-defer-pop @gol
267 -fno-function-cse -fno-guess-branch-probability @gol
268 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
269 -funsafe-math-optimizations -fno-trapping-math @gol
270 -fomit-frame-pointer -foptimize-register-move @gol
271 -foptimize-sibling-calls -freduce-all-givs @gol
272 -fregmove -frename-registers @gol
273 -frerun-cse-after-loop -frerun-loop-opt @gol
274 -fschedule-insns -fschedule-insns2 @gol
275 -fsingle-precision-constant -fssa -fssa-dce @gol
276 -fstrength-reduce -fstrict-aliasing -fthread-jumps -ftrapv @gol
277 -funroll-all-loops -funroll-loops @gol
278 --param @var{name}=@var{value}
279 -O -O0 -O1 -O2 -O3 -Os}
281 @item Preprocessor Options
282 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
284 -$ -A@var{question}=@var{answer} -A-@var{question}@r{[}=@var{answer}@r{]} @gol
285 -C -dD -dI -dM -dN @gol
286 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
287 -idirafter @var{dir} @gol
288 -include @var{file} -imacros @var{file} @gol
289 -iprefix @var{file} -iwithprefix @var{dir} @gol
290 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
291 -M -MM -MF -MG -MP -MQ -MT -nostdinc -P -remap @gol
292 -trigraphs -undef -U@var{macro} -Wp,@var{option}}
294 @item Assembler Option
295 @xref{Assembler Options,,Passing Options to the Assembler}.
300 @xref{Link Options,,Options for Linking}.
302 @var{object-file-name} -l@var{library} @gol
303 -nostartfiles -nodefaultlibs -nostdlib @gol
304 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
305 -Wl,@var{option} -Xlinker @var{option} @gol
308 @item Directory Options
309 @xref{Directory Options,,Options for Directory Search}.
311 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
314 @c I wrote this xref this way to avoid overfull hbox. -- rms
315 @xref{Target Options}.
317 -b @var{machine} -V @var{version}}
319 @item Machine Dependent Options
320 @xref{Submodel Options,,Hardware Models and Configurations}.
321 @emph{M680x0 Options}
323 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
324 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
325 -mfpa -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
326 -malign-int -mstrict-align}
328 @emph{M68hc1x Options}
330 -m6811 -m6812 -m68hc11 -m68hc12 @gol
331 -mauto-incdec -mshort -msoft-reg-count=@var{count}}
339 -mcpu=@var{cpu-type} @gol
340 -mtune=@var{cpu-type} @gol
341 -mcmodel=@var{code-model} @gol
343 -mapp-regs -mbroken-saverestore -mcypress @gol
344 -mepilogue -mfaster-structs -mflat @gol
345 -mfpu -mhard-float -mhard-quad-float @gol
346 -mimpure-text -mlive-g0 -mno-app-regs @gol
347 -mno-epilogue -mno-faster-structs -mno-flat -mno-fpu @gol
348 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
349 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
350 -msupersparc -munaligned-doubles -mv8}
352 @emph{Convex Options}
354 -mc1 -mc2 -mc32 -mc34 -mc38 @gol
355 -margcount -mnoargcount @gol
356 -mlong32 -mlong64 @gol
357 -mvolatile-cache -mvolatile-nocache}
359 @emph{AMD29K Options}
361 -m29000 -m29050 -mbw -mnbw -mdw -mndw @gol
362 -mlarge -mnormal -msmall @gol
363 -mkernel-registers -mno-reuse-arg-regs @gol
364 -mno-stack-check -mno-storem-bug @gol
365 -mreuse-arg-regs -msoft-float -mstack-check @gol
366 -mstorem-bug -muser-registers}
370 -mapcs-frame -mno-apcs-frame @gol
371 -mapcs-26 -mapcs-32 @gol
372 -mapcs-stack-check -mno-apcs-stack-check @gol
373 -mapcs-float -mno-apcs-float @gol
374 -mapcs-reentrant -mno-apcs-reentrant @gol
375 -msched-prolog -mno-sched-prolog @gol
376 -mlittle-endian -mbig-endian -mwords-little-endian @gol
377 -malignment-traps -mno-alignment-traps @gol
378 -msoft-float -mhard-float -mfpe @gol
379 -mthumb-interwork -mno-thumb-interwork @gol
380 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
381 -mstructure-size-boundary=@var{n} @gol
382 -mbsd -mxopen -mno-symrename @gol
383 -mabort-on-noreturn @gol
384 -mlong-calls -mno-long-calls @gol
385 -msingle-pic-base -mno-single-pic-base @gol
386 -mpic-register=@var{reg} @gol
387 -mnop-fun-dllimport @gol
388 -mpoke-function-name @gol
390 -mtpcs-frame -mtpcs-leaf-frame @gol
391 -mcaller-super-interworking -mcallee-super-interworking }
393 @emph{MN10200 Options}
397 @emph{MN10300 Options}
399 -mmult-bug -mno-mult-bug @gol
400 -mam33 -mno-am33 @gol
403 @emph{M32R/D Options}
405 -mcode-model=@var{model-type} -msdata=@var{sdata-type} @gol
410 -m88000 -m88100 -m88110 -mbig-pic @gol
411 -mcheck-zero-division -mhandle-large-shift @gol
412 -midentify-revision -mno-check-zero-division @gol
413 -mno-ocs-debug-info -mno-ocs-frame-position @gol
414 -mno-optimize-arg-area -mno-serialize-volatile @gol
415 -mno-underscores -mocs-debug-info @gol
416 -mocs-frame-position -moptimize-arg-area @gol
417 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
418 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
419 -mversion-03.00 -mwarn-passed-structs}
421 @emph{RS/6000 and PowerPC Options}
423 -mcpu=@var{cpu-type} @gol
424 -mtune=@var{cpu-type} @gol
425 -mpower -mno-power -mpower2 -mno-power2 @gol
426 -mpowerpc -mpowerpc64 -mno-powerpc @gol
427 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
428 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
429 -mnew-mnemonics -mold-mnemonics @gol
430 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc @gol
431 -m64 -m32 -mxl-call -mno-xl-call -mthreads -mpe @gol
432 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
433 -mstring -mno-string -mupdate -mno-update @gol
434 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
435 -mstrict-align -mno-strict-align -mrelocatable @gol
436 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
437 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
438 -mcall-aix -mcall-sysv -mprototype -mno-prototype @gol
439 -msim -mmvme -mads -myellowknife -memb -msdata @gol
440 -msdata=@var{opt} -mvxworks -G @var{num}}
444 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
445 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
446 -mminimum-fp-blocks -mnohc-struct-return}
450 -mabicalls -mcpu=@var{cpu-type} @gol
451 -membedded-data -muninit-const-in-rodata @gol
452 -membedded-pic -mfp32 -mfp64 -mgas -mgp32 -mgp64 @gol
453 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1 @gol
454 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
455 -mmips-as -mmips-tfile -mno-abicalls @gol
456 -mno-embedded-data -mno-uninit-const-in-rodata @gol
457 -mno-embedded-pic -mno-gpopt -mno-long-calls @gol
458 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats @gol
459 -mrnames -msoft-float @gol
460 -m4650 -msingle-float -mmad @gol
461 -mstats -EL -EB -G @var{num} -nocpp @gol
462 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi @gol
467 -mcpu=@var{cpu-type} -march=@var{cpu-type} @gol
468 -mintel-syntax -mieee-fp -mno-fancy-math-387 @gol
469 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
470 -mno-wide-multiply -mrtd -malign-double @gol
471 -mpreferred-stack-boundary=@var{num} @gol
472 -mthreads -mno-align-stringops -minline-all-stringops @gol
473 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
474 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer}
478 -march=@var{architecture-type} @gol
479 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
480 -mfast-indirect-calls -mgas -mjump-in-delay @gol
481 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
482 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
483 -mno-jump-in-delay -mno-long-load-store @gol
484 -mno-portable-runtime -mno-soft-float @gol
485 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
486 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
487 -mschedule=@var{cpu-type} -mspace-regs}
489 @emph{Intel 960 Options}
491 -m@var{cpu-type} -masm-compat -mclean-linkage @gol
492 -mcode-align -mcomplex-addr -mleaf-procedures @gol
493 -mic-compat -mic2.0-compat -mic3.0-compat @gol
494 -mintel-asm -mno-clean-linkage -mno-code-align @gol
495 -mno-complex-addr -mno-leaf-procedures @gol
496 -mno-old-align -mno-strict-align -mno-tail-call @gol
497 -mnumerics -mold-align -msoft-float -mstrict-align @gol
500 @emph{DEC Alpha Options}
502 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float @gol
503 -malpha-as -mgas @gol
504 -mieee -mieee-with-inexact -mieee-conformant @gol
505 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
506 -mtrap-precision=@var{mode} -mbuild-constants @gol
507 -mcpu=@var{cpu-type} @gol
508 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max @gol
509 -mmemory-latency=@var{time}}
511 @emph{Clipper Options}
515 @emph{H8/300 Options}
517 -mrelax -mh -ms -mint32 -malign-300}
521 -m1 -m2 -m3 -m3e @gol
522 -m4-nofpu -m4-single-only -m4-single -m4 @gol
523 -mb -ml -mdalign -mrelax @gol
524 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
525 -mieee -misize -mpadstruct -mspace @gol
526 -mprefergot -musermode}
528 @emph{System V Options}
530 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
535 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
536 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
538 @emph{TMS320C3x/C4x Options}
540 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
541 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
542 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
543 -mparallel-insns -mparallel-mpy -mpreserve-float}
547 -mlong-calls -mno-long-calls -mep -mno-ep @gol
548 -mprolog-function -mno-prolog-function -mspace @gol
549 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
554 -m32032 -m32332 -m32532 -m32081 -m32381 @gol
555 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
556 -mregparam -mnoregparam -msb -mnosb @gol
557 -mbitfield -mnobitfield -mhimem -mnohimem}
561 -mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
562 -mcall-prologues -mno-tablejump -mtiny-stack}
566 -mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
567 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
568 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
569 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
570 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
574 -mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
575 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
576 -mconstant-gp -mauto-pic -minline-divide-min-latency @gol
577 -minline-divide-max-throughput -mno-dwarf2-asm @gol
578 -mfixed-range=@var{register-range}}
580 @item Code Generation Options
581 @xref{Code Gen Options,,Options for Code Generation Conventions}.
583 -fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
584 -ffixed-@var{reg} -fexceptions @gol
585 -fnon-call-exceptions -funwind-tables @gol
586 -finhibit-size-directive -finstrument-functions @gol
587 -fcheck-memory-usage -fprefix-function-name @gol
588 -fno-common -fno-ident -fno-gnu-linker @gol
589 -fpcc-struct-return -fpic -fPIC @gol
590 -freg-struct-return -fshared-data -fshort-enums @gol
591 -fshort-double -fvolatile @gol
592 -fvolatile-global -fvolatile-static @gol
593 -fverbose-asm -fpack-struct -fstack-check @gol
594 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
595 -fargument-alias -fargument-noalias @gol
596 -fargument-noalias-global -fleading-underscore}
600 * Overall Options:: Controlling the kind of output:
601 an executable, object files, assembler files,
602 or preprocessed source.
603 * C Dialect Options:: Controlling the variant of C language compiled.
604 * C++ Dialect Options:: Variations on C++.
605 * Objective-C Dialect Options:: Variations on Objective-C.
606 * Language Independent Options:: Controlling how diagnostics should be
608 * Warning Options:: How picky should the compiler be?
609 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
610 * Optimize Options:: How much optimization?
611 * Preprocessor Options:: Controlling header files and macro definitions.
612 Also, getting dependency information for Make.
613 * Assembler Options:: Passing options to the assembler.
614 * Link Options:: Specifying libraries and so on.
615 * Directory Options:: Where to find header files and libraries.
616 Where to find the compiler executable files.
617 * Spec Files:: How to pass switches to sub-processes.
618 * Target Options:: Running a cross-compiler, or an old version of GCC.
621 @node Overall Options
622 @section Options Controlling the Kind of Output
624 Compilation can involve up to four stages: preprocessing, compilation
625 proper, assembly and linking, always in that order. The first three
626 stages apply to an individual source file, and end by producing an
627 object file; linking combines all the object files (those newly
628 compiled, and those specified as input) into an executable file.
630 @cindex file name suffix
631 For any given input file, the file name suffix determines what kind of
636 C source code which must be preprocessed.
639 C source code which should not be preprocessed.
642 C++ source code which should not be preprocessed.
645 Objective-C source code. Note that you must link with the library
646 @file{libobjc.a} to make an Objective-C program work.
649 Objective-C source code which should not be preprocessed.
652 C header file (not to be compiled or linked).
656 @itemx @var{file}.cxx
657 @itemx @var{file}.cpp
658 @itemx @var{file}.c++
660 C++ source code which must be preprocessed. Note that in @samp{.cxx},
661 the last two letters must both be literally @samp{x}. Likewise,
662 @samp{.C} refers to a literal capital C@.
665 @itemx @var{file}.for
666 @itemx @var{file}.FOR
667 Fortran source code which should not be preprocessed.
670 @itemx @var{file}.fpp
671 @itemx @var{file}.FPP
672 Fortran source code which must be preprocessed (with the traditional
676 Fortran source code which must be preprocessed with a RATFOR
677 preprocessor (not included with GCC)@.
679 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
680 Using and Porting GNU Fortran}, for more details of the handling of
683 @c FIXME: Descriptions of Java file types.
689 @c GCC also knows about some suffixes for languages not yet included:
699 @itemx @var{file}.chi
700 CHILL source code (preprocessed with the traditional preprocessor).
706 Assembler code which must be preprocessed.
709 An object file to be fed straight into linking.
710 Any file name with no recognized suffix is treated this way.
714 You can specify the input language explicitly with the @option{-x} option:
717 @item -x @var{language}
718 Specify explicitly the @var{language} for the following input files
719 (rather than letting the compiler choose a default based on the file
720 name suffix). This option applies to all following input files until
721 the next @option{-x} option. Possible values for @var{language} are:
723 c c-header cpp-output
725 objective-c objc-cpp-output
726 assembler assembler-with-cpp
727 f77 f77-cpp-input ratfor
730 @c Also f77-version, for internal use only.
733 Turn off any specification of a language, so that subsequent files are
734 handled according to their file name suffixes (as they are if @option{-x}
735 has not been used at all).
737 @item -pass-exit-codes
738 @opindex pass-exit-codes
739 Normally the @command{gcc} program will exit with the code of 1 if any
740 phase of the compiler returns a non-success return code. If you specify
741 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
742 numerically highest error produced by any phase that returned an error
746 If you only want some of the stages of compilation, you can use
747 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
748 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
749 @command{gcc} is to stop. Note that some combinations (for example,
750 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
755 Compile or assemble the source files, but do not link. The linking
756 stage simply is not done. The ultimate output is in the form of an
757 object file for each source file.
759 By default, the object file name for a source file is made by replacing
760 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
762 Unrecognized input files, not requiring compilation or assembly, are
767 Stop after the stage of compilation proper; do not assemble. The output
768 is in the form of an assembler code file for each non-assembler input
771 By default, the assembler file name for a source file is made by
772 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
774 Input files that don't require compilation are ignored.
778 Stop after the preprocessing stage; do not run the compiler proper. The
779 output is in the form of preprocessed source code, which is sent to the
782 Input files which don't require preprocessing are ignored.
784 @cindex output file option
787 Place output in file @var{file}. This applies regardless to whatever
788 sort of output is being produced, whether it be an executable file,
789 an object file, an assembler file or preprocessed C code.
791 Since only one output file can be specified, it does not make sense to
792 use @option{-o} when compiling more than one input file, unless you are
793 producing an executable file as output.
795 If @option{-o} is not specified, the default is to put an executable file
796 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
797 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
798 all preprocessed C source on standard output.
802 Print (on standard error output) the commands executed to run the stages
803 of compilation. Also print the version number of the compiler driver
804 program and of the preprocessor and the compiler proper.
808 Use pipes rather than temporary files for communication between the
809 various stages of compilation. This fails to work on some systems where
810 the assembler is unable to read from a pipe; but the GNU assembler has
815 Print (on the standard output) a description of the command line options
816 understood by @command{gcc}. If the @option{-v} option is also specified
817 then @option{--help} will also be passed on to the various processes
818 invoked by @command{gcc}, so that they can display the command line options
819 they accept. If the @option{-W} option is also specified then command
820 line options which have no documentation associated with them will also
825 Print (on the standard output) a description of target specific command
826 line options for each tool.
830 @section Compiling C++ Programs
832 @cindex suffixes for C++ source
833 @cindex C++ source file suffixes
834 C++ source files conventionally use one of the suffixes @samp{.C},
835 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
836 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
837 files with these names and compiles them as C++ programs even if you
838 call the compiler the same way as for compiling C programs (usually with
839 the name @command{gcc}).
843 However, C++ programs often require class libraries as well as a
844 compiler that understands the C++ language---and under some
845 circumstances, you might want to compile programs from standard input,
846 or otherwise without a suffix that flags them as C++ programs.
847 @command{g++} is a program that calls GCC with the default language
848 set to C++, and automatically specifies linking against the C++
849 library. On many systems, @command{g++} is also
850 installed with the name @command{c++}.
852 @cindex invoking @command{g++}
853 When you compile C++ programs, you may specify many of the same
854 command-line options that you use for compiling programs in any
855 language; or command-line options meaningful for C and related
856 languages; or options that are meaningful only for C++ programs.
857 @xref{C Dialect Options,,Options Controlling C Dialect}, for
858 explanations of options for languages related to C@.
859 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
860 explanations of options that are meaningful only for C++ programs.
862 @node C Dialect Options
863 @section Options Controlling C Dialect
864 @cindex dialect options
865 @cindex language dialect options
866 @cindex options, dialect
868 The following options control the dialect of C (or languages derived
869 from C, such as C++ and Objective-C) that the compiler accepts:
876 In C mode, support all ISO C89 programs. In C++ mode,
877 remove GNU extensions that conflict with ISO C++.
879 This turns off certain features of GCC that are incompatible with ISO
880 C89 (when compiling C code), or of standard C++ (when compiling C++ code),
881 such as the @code{asm} and @code{typeof} keywords, and
882 predefined macros such as @code{unix} and @code{vax} that identify the
883 type of system you are using. It also enables the undesirable and
884 rarely used ISO trigraph feature. For the C compiler,
885 it disables recognition of C++ style @samp{//} comments as well as
886 the @code{inline} keyword.
888 The alternate keywords @code{__asm__}, @code{__extension__},
889 @code{__inline__} and @code{__typeof__} continue to work despite
890 @option{-ansi}. You would not want to use them in an ISO C program, of
891 course, but it is useful to put them in header files that might be included
892 in compilations done with @option{-ansi}. Alternate predefined macros
893 such as @code{__unix__} and @code{__vax__} are also available, with or
894 without @option{-ansi}.
896 The @option{-ansi} option does not cause non-ISO programs to be
897 rejected gratuitously. For that, @option{-pedantic} is required in
898 addition to @option{-ansi}. @xref{Warning Options}.
900 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
901 option is used. Some header files may notice this macro and refrain
902 from declaring certain functions or defining certain macros that the
903 ISO standard doesn't call for; this is to avoid interfering with any
904 programs that might use these names for other things.
906 Functions which would normally be built in but do not have semantics
907 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
908 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
909 built-in functions provided by GCC}, for details of the functions
914 Determine the language standard. This option is currently only
915 supported when compiling C@. A value for this option must be provided;
921 ISO C89 (same as @option{-ansi}).
924 ISO C89 as modified in amendment 1.
930 ISO C99. Note that this standard is not yet fully supported; see
931 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
932 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
935 Default, ISO C89 plus GNU extensions (including some C99 features).
939 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
940 this will become the default. The name @samp{gnu9x} is deprecated.
944 Even when this option is not specified, you can still use some of the
945 features of newer standards in so far as they do not conflict with
946 previous C standards. For example, you may use @code{__restrict__} even
947 when @option{-std=c99} is not specified.
949 The @option{-std} options specifying some version of ISO C have the same
950 effects as @option{-ansi}, except that features that were not in ISO C89
951 but are in the specified version (for example, @samp{//} comments and
952 the @code{inline} keyword in ISO C99) are not disabled.
954 @xref{Standards,,Language Standards Supported by GCC}, for details of
955 these standard versions.
957 @item -aux-info @var{filename}
959 Output to the given filename prototyped declarations for all functions
960 declared and/or defined in a translation unit, including those in header
961 files. This option is silently ignored in any language other than C@.
963 Besides declarations, the file indicates, in comments, the origin of
964 each declaration (source file and line), whether the declaration was
965 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
966 @samp{O} for old, respectively, in the first character after the line
967 number and the colon), and whether it came from a declaration or a
968 definition (@samp{C} or @samp{F}, respectively, in the following
969 character). In the case of function definitions, a K&R-style list of
970 arguments followed by their declarations is also provided, inside
971 comments, after the declaration.
975 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
976 keyword, so that code can use these words as identifiers. You can use
977 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
978 instead. @option{-ansi} implies @option{-fno-asm}.
980 In C++, this switch only affects the @code{typeof} keyword, since
981 @code{asm} and @code{inline} are standard keywords. You may want to
982 use the @option{-fno-gnu-keywords} flag instead, which has the same
983 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
984 switch only affects the @code{asm} and @code{typeof} keywords, since
985 @code{inline} is a standard keyword in ISO C99.
989 @cindex built-in functions
990 Don't recognize built-in functions that do not begin with
991 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
992 functions provided by GCC}, for details of the functions affected,
993 including those which are not built-in functions when @option{-ansi} or
994 @option{-std} options for strict ISO C conformance are used because they
995 do not have an ISO standard meaning.
997 GCC normally generates special code to handle certain built-in functions
998 more efficiently; for instance, calls to @code{alloca} may become single
999 instructions that adjust the stack directly, and calls to @code{memcpy}
1000 may become inline copy loops. The resulting code is often both smaller
1001 and faster, but since the function calls no longer appear as such, you
1002 cannot set a breakpoint on those calls, nor can you change the behavior
1003 of the functions by linking with a different library.
1005 In C++, @option{-fno-builtin} is always in effect. The @option{-fbuiltin}
1006 option has no effect. Therefore, in C++, the only way to get the
1007 optimization benefits of built-in functions is to call the function
1008 using the @samp{__builtin_} prefix. The GNU C++ Standard Library uses
1009 built-in functions to implement many functions (like
1010 @code{std::strchr}), so that you automatically get efficient code.
1014 @cindex hosted environment
1016 Assert that compilation takes place in a hosted environment. This implies
1017 @option{-fbuiltin}. A hosted environment is one in which the
1018 entire standard library is available, and in which @code{main} has a return
1019 type of @code{int}. Examples are nearly everything except a kernel.
1020 This is equivalent to @option{-fno-freestanding}.
1022 @item -ffreestanding
1023 @opindex ffreestanding
1024 @cindex hosted environment
1026 Assert that compilation takes place in a freestanding environment. This
1027 implies @option{-fno-builtin}. A freestanding environment
1028 is one in which the standard library may not exist, and program startup may
1029 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1030 This is equivalent to @option{-fno-hosted}.
1032 @xref{Standards,,Language Standards Supported by GCC}, for details of
1033 freestanding and hosted environments.
1037 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1038 options for strict ISO C conformance) implies @option{-trigraphs}.
1040 @cindex traditional C language
1041 @cindex C language, traditional
1043 @opindex traditional
1044 Attempt to support some aspects of traditional C compilers.
1049 All @code{extern} declarations take effect globally even if they
1050 are written inside of a function definition. This includes implicit
1051 declarations of functions.
1054 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
1055 and @code{volatile} are not recognized. (You can still use the
1056 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
1060 Comparisons between pointers and integers are always allowed.
1063 Integer types @code{unsigned short} and @code{unsigned char} promote
1064 to @code{unsigned int}.
1067 Out-of-range floating point literals are not an error.
1070 Certain constructs which ISO regards as a single invalid preprocessing
1071 number, such as @samp{0xe-0xd}, are treated as expressions instead.
1074 String ``constants'' are not necessarily constant; they are stored in
1075 writable space, and identical looking constants are allocated
1076 separately. (This is the same as the effect of
1077 @option{-fwritable-strings}.)
1079 @cindex @code{longjmp} and automatic variables
1081 All automatic variables not declared @code{register} are preserved by
1082 @code{longjmp}. Ordinarily, GNU C follows ISO C: automatic variables
1083 not declared @code{volatile} may be clobbered.
1088 @cindex escape sequences, traditional
1089 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
1090 literal characters @samp{x} and @samp{a} respectively. Without
1091 @w{@option{-traditional}}, @samp{\x} is a prefix for the hexadecimal
1092 representation of a character, and @samp{\a} produces a bell.
1095 You may wish to use @option{-fno-builtin} as well as @option{-traditional}
1096 if your program uses names that are normally GNU C built-in functions for
1097 other purposes of its own.
1099 You cannot use @option{-traditional} if you include any header files that
1100 rely on ISO C features. Some vendors are starting to ship systems with
1101 ISO C header files and you cannot use @option{-traditional} on such
1102 systems to compile files that include any system headers.
1104 The @option{-traditional} option also enables @option{-traditional-cpp},
1105 which is described next.
1107 @item -traditional-cpp
1108 @opindex traditional-cpp
1109 Attempt to support some aspects of traditional C preprocessors.
1114 Comments convert to nothing at all, rather than to a space. This allows
1115 traditional token concatenation.
1118 In a preprocessing directive, the @samp{#} symbol must appear as the first
1119 character of a line.
1122 Macro arguments are recognized within string constants in a macro
1123 definition (and their values are stringified, though without additional
1124 quote marks, when they appear in such a context). The preprocessor
1125 always considers a string constant to end at a newline.
1128 @cindex detecting @w{@option{-traditional}}
1129 The predefined macro @code{__STDC__} is not defined when you use
1130 @option{-traditional}, but @code{__GNUC__} is (since the GNU extensions
1131 which @code{__GNUC__} indicates are not affected by
1132 @option{-traditional}). If you need to write header files that work
1133 differently depending on whether @option{-traditional} is in use, by
1134 testing both of these predefined macros you can distinguish four
1135 situations: GNU C, traditional GNU C, other ISO C compilers, and other
1136 old C compilers. The predefined macro @code{__STDC_VERSION__} is also
1137 not defined when you use @option{-traditional}. @xref{Standard
1138 Predefined,,Standard Predefined Macros,cpp.info,The C Preprocessor},
1139 for more discussion of these and other predefined macros.
1142 @cindex string constants vs newline
1143 @cindex newline vs string constants
1144 The preprocessor considers a string constant to end at a newline (unless
1145 the newline is escaped with @samp{\}). (Without @w{@option{-traditional}},
1146 string constants can contain the newline character as typed.)
1149 @item -fcond-mismatch
1150 @opindex fcond-mismatch
1151 Allow conditional expressions with mismatched types in the second and
1152 third arguments. The value of such an expression is void. This option
1153 is not supported for C++.
1155 @item -funsigned-char
1156 @opindex funsigned-char
1157 Let the type @code{char} be unsigned, like @code{unsigned char}.
1159 Each kind of machine has a default for what @code{char} should
1160 be. It is either like @code{unsigned char} by default or like
1161 @code{signed char} by default.
1163 Ideally, a portable program should always use @code{signed char} or
1164 @code{unsigned char} when it depends on the signedness of an object.
1165 But many programs have been written to use plain @code{char} and
1166 expect it to be signed, or expect it to be unsigned, depending on the
1167 machines they were written for. This option, and its inverse, let you
1168 make such a program work with the opposite default.
1170 The type @code{char} is always a distinct type from each of
1171 @code{signed char} or @code{unsigned char}, even though its behavior
1172 is always just like one of those two.
1175 @opindex fsigned-char
1176 Let the type @code{char} be signed, like @code{signed char}.
1178 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1179 the negative form of @option{-funsigned-char}. Likewise, the option
1180 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1182 @item -fsigned-bitfields
1183 @itemx -funsigned-bitfields
1184 @itemx -fno-signed-bitfields
1185 @itemx -fno-unsigned-bitfields
1186 @opindex fsigned-bitfields
1187 @opindex funsigned-bitfields
1188 @opindex fno-signed-bitfields
1189 @opindex fno-unsigned-bitfields
1190 These options control whether a bit-field is signed or unsigned, when the
1191 declaration does not use either @code{signed} or @code{unsigned}. By
1192 default, such a bit-field is signed, because this is consistent: the
1193 basic integer types such as @code{int} are signed types.
1195 However, when @option{-traditional} is used, bit-fields are all unsigned
1198 @item -fwritable-strings
1199 @opindex fwritable-strings
1200 Store string constants in the writable data segment and don't uniquize
1201 them. This is for compatibility with old programs which assume they can
1202 write into string constants. The option @option{-traditional} also has
1205 Writing into string constants is a very bad idea; ``constants'' should
1208 @item -fallow-single-precision
1209 @opindex fallow-single-precision
1210 Do not promote single precision math operations to double precision,
1211 even when compiling with @option{-traditional}.
1213 Traditional K&R C promotes all floating point operations to double
1214 precision, regardless of the sizes of the operands. On the
1215 architecture for which you are compiling, single precision may be faster
1216 than double precision. If you must use @option{-traditional}, but want
1217 to use single precision operations when the operands are single
1218 precision, use this option. This option has no effect when compiling
1219 with ISO or GNU C conventions (the default).
1222 @opindex fshort-wchar
1223 Override the underlying type for @samp{wchar_t} to be @samp{short
1224 unsigned int} instead of the default for the target. This option is
1225 useful for building programs to run under WINE@.
1228 @node C++ Dialect Options
1229 @section Options Controlling C++ Dialect
1231 @cindex compiler options, C++
1232 @cindex C++ options, command line
1233 @cindex options, C++
1234 This section describes the command-line options that are only meaningful
1235 for C++ programs; but you can also use most of the GNU compiler options
1236 regardless of what language your program is in. For example, you
1237 might compile a file @code{firstClass.C} like this:
1240 g++ -g -frepo -O -c firstClass.C
1244 In this example, only @option{-frepo} is an option meant
1245 only for C++ programs; you can use the other options with any
1246 language supported by GCC@.
1248 Here is a list of options that are @emph{only} for compiling C++ programs:
1251 @item -fno-access-control
1252 @opindex fno-access-control
1253 Turn off all access checking. This switch is mainly useful for working
1254 around bugs in the access control code.
1258 Check that the pointer returned by @code{operator new} is non-null
1259 before attempting to modify the storage allocated. The current Working
1260 Paper requires that @code{operator new} never return a null pointer, so
1261 this check is normally unnecessary.
1263 An alternative to using this option is to specify that your
1264 @code{operator new} does not throw any exceptions; if you declare it
1265 @samp{throw()}, G++ will check the return value. See also @samp{new
1268 @item -fconserve-space
1269 @opindex fconserve-space
1270 Put uninitialized or runtime-initialized global variables into the
1271 common segment, as C does. This saves space in the executable at the
1272 cost of not diagnosing duplicate definitions. If you compile with this
1273 flag and your program mysteriously crashes after @code{main()} has
1274 completed, you may have an object that is being destroyed twice because
1275 two definitions were merged.
1277 This option is no longer useful on most targets, now that support has
1278 been added for putting variables into BSS without making them common.
1280 @item -fno-const-strings
1281 @opindex fno-const-strings
1282 Give string constants type @code{char *} instead of type @code{const
1283 char *}. By default, G++ uses type @code{const char *} as required by
1284 the standard. Even if you use @option{-fno-const-strings}, you cannot
1285 actually modify the value of a string constant, unless you also use
1286 @option{-fwritable-strings}.
1288 This option might be removed in a future release of G++. For maximum
1289 portability, you should structure your code so that it works with
1290 string constants that have type @code{const char *}.
1292 @item -fdollars-in-identifiers
1293 @opindex fdollars-in-identifiers
1294 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1295 @samp{$} with the option @option{-fno-dollars-in-identifiers}. (GNU C allows
1296 @samp{$} by default on most target systems, but there are a few exceptions.)
1297 Traditional C allowed the character @samp{$} to form part of
1298 identifiers. However, ISO C and C++ forbid @samp{$} in identifiers.
1300 @item -fno-elide-constructors
1301 @opindex fno-elide-constructors
1302 The C++ standard allows an implementation to omit creating a temporary
1303 which is only used to initialize another object of the same type.
1304 Specifying this option disables that optimization, and forces G++ to
1305 call the copy constructor in all cases.
1307 @item -fno-enforce-eh-specs
1308 @opindex fno-enforce-eh-specs
1309 Don't check for violation of exception specifications at runtime. This
1310 option violates the C++ standard, but may be useful for reducing code
1311 size in production builds, much like defining @samp{NDEBUG}. The compiler
1312 will still optimize based on the exception specifications.
1314 @item -fexternal-templates
1315 @opindex fexternal-templates
1316 Cause template instantiations to obey @samp{#pragma interface} and
1317 @samp{implementation}; template instances are emitted or not according
1318 to the location of the template definition. @xref{Template
1319 Instantiation}, for more information.
1321 This option is deprecated.
1323 @item -falt-external-templates
1324 @opindex falt-external-templates
1325 Similar to @option{-fexternal-templates}, but template instances are emitted or
1326 not according to the place where they are first instantiated.
1327 @xref{Template Instantiation}, for more information.
1329 This option is deprecated.
1332 @itemx -fno-for-scope
1334 @opindex fno-for-scope
1335 If @option{-ffor-scope} is specified, the scope of variables declared in
1336 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1337 as specified by the C++ standard.
1338 If @option{-fno-for-scope} is specified, the scope of variables declared in
1339 a @i{for-init-statement} extends to the end of the enclosing scope,
1340 as was the case in old versions of G++, and other (traditional)
1341 implementations of C++.
1343 The default if neither flag is given to follow the standard,
1344 but to allow and give a warning for old-style code that would
1345 otherwise be invalid, or have different behavior.
1347 @item -fno-gnu-keywords
1348 @opindex fno-gnu-keywords
1349 Do not recognize @code{typeof} as a keyword, so that code can use this
1350 word as an identifier. You can use the keyword @code{__typeof__} instead.
1351 @option{-ansi} implies @option{-fno-gnu-keywords}.
1353 @item -fno-honor-std
1354 @opindex fno-honor-std
1355 Ignore @code{namespace std}, instead of treating it as a real namespace.
1356 With this switch, the compiler will ignore
1357 @code{namespace-declarations}, @code{using-declarations},
1358 @code{using-directives}, and @code{namespace-names}, if they involve
1361 This option is only useful if you have manually compiled the C++
1362 run-time library with the same switch. Otherwise, your programs will
1363 not link. The use of this option is not recommended, and the option may
1364 be removed from a future version of G++.
1366 @item -fno-implicit-templates
1367 @opindex fno-implicit-templates
1368 Never emit code for non-inline templates which are instantiated
1369 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1370 @xref{Template Instantiation}, for more information.
1372 @item -fno-implicit-inline-templates
1373 @opindex fno-implicit-inline-templates
1374 Don't emit code for implicit instantiations of inline templates, either.
1375 The default is to handle inlines differently so that compiles with and
1376 without optimization will need the same set of explicit instantiations.
1378 @item -fno-implement-inlines
1379 @opindex fno-implement-inlines
1380 To save space, do not emit out-of-line copies of inline functions
1381 controlled by @samp{#pragma implementation}. This will cause linker
1382 errors if these functions are not inlined everywhere they are called.
1384 @item -fms-extensions
1385 @opindex fms-extensions
1386 Disable pedantic warnings about constructs used in MFC, such as implicit
1387 int and getting a pointer to member function via non-standard syntax.
1389 @item -fno-nonansi-builtins
1390 @opindex fno-nonansi-builtins
1391 Disable built-in declarations of functions that are not mandated by
1392 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1393 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1395 @item -fno-operator-names
1396 @opindex fno-operator-names
1397 Do not treat the operator name keywords @code{and}, @code{bitand},
1398 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1399 synonyms as keywords.
1401 @item -fno-optional-diags
1402 @opindex fno-optional-diags
1403 Disable diagnostics that the standard says a compiler does not need to
1404 issue. Currently, the only such diagnostic issued by G++ is the one for
1405 a name having multiple meanings within a class.
1408 @opindex fpermissive
1409 Downgrade messages about nonconformant code from errors to warnings. By
1410 default, G++ effectively sets @option{-pedantic-errors} without
1411 @option{-pedantic}; this option reverses that. This behavior and this
1412 option are superseded by @option{-pedantic}, which works as it does for GNU C@.
1416 Enable automatic template instantiation. This option also implies
1417 @option{-fno-implicit-templates}. @xref{Template Instantiation}, for more
1422 Disable generation of information about every class with virtual
1423 functions for use by the C++ runtime type identification features
1424 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1425 of the language, you can save some space by using this flag. Note that
1426 exception handling uses the same information, but it will generate it as
1431 Emit statistics about front-end processing at the end of the compilation.
1432 This information is generally only useful to the G++ development team.
1434 @item -ftemplate-depth-@var{n}
1435 @opindex ftemplate-depth
1436 Set the maximum instantiation depth for template classes to @var{n}.
1437 A limit on the template instantiation depth is needed to detect
1438 endless recursions during template class instantiation. ANSI/ISO C++
1439 conforming programs must not rely on a maximum depth greater than 17.
1441 @item -fuse-cxa-atexit
1442 @opindex fuse-cxa-atexit
1443 Register destructors for objects with static storage duration with the
1444 @code{__cxa_atexit} function rather than the @code{atexit} function.
1445 This option is required for fully standards-compliant handling of static
1446 destructors, but will only work if your C library supports
1447 @code{__cxa_atexit}.
1451 Emit special relocations for vtables and virtual function references
1452 so that the linker can identify unused virtual functions and zero out
1453 vtable slots that refer to them. This is most useful with
1454 @option{-ffunction-sections} and @option{-Wl,--gc-sections}, in order to
1455 also discard the functions themselves.
1457 This optimization requires GNU as and GNU ld. Not all systems support
1458 this option. @option{-Wl,--gc-sections} is ignored without @option{-static}.
1462 Do not use weak symbol support, even if it is provided by the linker.
1463 By default, G++ will use weak symbols if they are available. This
1464 option exists only for testing, and should not be used by end-users;
1465 it will result in inferior code and has no benefits. This option may
1466 be removed in a future release of G++.
1470 Do not search for header files in the standard directories specific to
1471 C++, but do still search the other standard directories. (This option
1472 is used when building the C++ library.)
1475 In addition, these optimization, warning, and code generation options
1476 have meanings only for C++ programs:
1479 @item -fno-default-inline
1480 @opindex fno-default-inline
1481 Do not assume @samp{inline} for functions defined inside a class scope.
1482 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1483 functions will have linkage like inline functions; they just won't be
1486 @item -Wctor-dtor-privacy @r{(C++ only)}
1487 @opindex Wctor-dtor-privacy
1488 Warn when a class seems unusable, because all the constructors or
1489 destructors in a class are private and the class has no friends or
1490 public static member functions.
1492 @item -Wnon-virtual-dtor @r{(C++ only)}
1493 @opindex Wnon-virtual-dtor
1494 Warn when a class declares a non-virtual destructor that should probably
1495 be virtual, because it looks like the class will be used polymorphically.
1497 @item -Wreorder @r{(C++ only)}
1499 @cindex reordering, warning
1500 @cindex warning for reordering of member initializers
1501 Warn when the order of member initializers given in the code does not
1502 match the order in which they must be executed. For instance:
1508 A(): j (0), i (1) @{ @}
1512 Here the compiler will warn that the member initializers for @samp{i}
1513 and @samp{j} will be rearranged to match the declaration order of the
1517 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1520 @item -Weffc++ @r{(C++ only)}
1522 Warn about violations of various style guidelines from Scott Meyers'
1523 @cite{Effective C++} books. If you use this option, you should be aware
1524 that the standard library headers do not obey all of these guidelines;
1525 you can use @samp{grep -v} to filter out those warnings.
1527 @item -Wno-deprecated @r{(C++ only)}
1528 @opindex Wno-deprecated
1529 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1531 @item -Wno-non-template-friend @r{(C++ only)}
1532 @opindex Wno-non-template-friend
1533 Disable warnings when non-templatized friend functions are declared
1534 within a template. With the advent of explicit template specification
1535 support in G++, if the name of the friend is an unqualified-id (i.e.,
1536 @samp{friend foo(int)}), the C++ language specification demands that the
1537 friend declare or define an ordinary, nontemplate function. (Section
1538 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1539 could be interpreted as a particular specialization of a templatized
1540 function. Because this non-conforming behavior is no longer the default
1541 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1542 check existing code for potential trouble spots, and is on by default.
1543 This new compiler behavior can be turned off with
1544 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1545 but disables the helpful warning.
1547 @item -Wold-style-cast @r{(C++ only)}
1548 @opindex Wold-style-cast
1549 Warn if an old-style (C-style) cast is used within a C++ program. The
1550 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1551 @samp{const_cast}) are less vulnerable to unintended effects, and much
1554 @item -Woverloaded-virtual @r{(C++ only)}
1555 @opindex Woverloaded-virtual
1556 @cindex overloaded virtual fn, warning
1557 @cindex warning for overloaded virtual fn
1558 Warn when a function declaration hides virtual functions from a
1559 base class. For example, in:
1566 struct B: public A @{
1571 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1579 will fail to compile.
1581 @item -Wno-pmf-conversions @r{(C++ only)}
1582 @opindex Wno-pmf-conversions
1583 Disable the diagnostic for converting a bound pointer to member function
1586 @item -Wsign-promo @r{(C++ only)}
1587 @opindex Wsign-promo
1588 Warn when overload resolution chooses a promotion from unsigned or
1589 enumeral type to a signed type over a conversion to an unsigned type of
1590 the same size. Previous versions of G++ would try to preserve
1591 unsignedness, but the standard mandates the current behavior.
1593 @item -Wsynth @r{(C++ only)}
1595 @cindex warning for synthesized methods
1596 @cindex synthesized methods, warning
1597 Warn when G++'s synthesis behavior does not match that of cfront. For
1603 A& operator = (int);
1613 In this example, G++ will synthesize a default @samp{A& operator =
1614 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1617 @node Objective-C Dialect Options
1618 @section Options Controlling Objective-C Dialect
1620 @cindex compiler options, Objective-C
1621 @cindex Objective-C options, command line
1622 @cindex options, Objective-C
1623 This section describes the command-line options that are only meaningful
1624 for Objective-C programs; but you can also use most of the GNU compiler
1625 options regardless of what language your program is in. For example,
1626 you might compile a file @code{some_class.m} like this:
1629 gcc -g -fgnu-runtime -O -c some_class.m
1633 In this example, only @option{-fgnu-runtime} is an option meant only for
1634 Objective-C programs; you can use the other options with any language
1637 Here is a list of options that are @emph{only} for compiling Objective-C
1641 @item -fconstant-string-class=@var{class-name}
1642 @opindex fconstant-string-class
1643 Use @var{class-name} as the name of the class to instantiate for each
1644 literal string specified with the syntax @code{@@"@dots{}"}. The default
1645 class name is @code{NXConstantString}.
1648 @opindex fgnu-runtime
1649 Generate object code compatible with the standard GNU Objective-C
1650 runtime. This is the default for most types of systems.
1652 @item -fnext-runtime
1653 @opindex fnext-runtime
1654 Generate output compatible with the NeXT runtime. This is the default
1655 for NeXT-based systems, including Darwin and Mac OS X@.
1659 Dump interface declarations for all classes seen in the source file to a
1660 file named @file{@var{sourcename}.decl}.
1663 @opindex Wno-protocol
1664 Do not warn if methods required by a protocol are not implemented
1665 in the class adopting it.
1669 Warn if a selector has multiple methods of different types defined.
1671 @c not documented because only avail via -Wp
1672 @c @item -print-objc-runtime-info
1676 @node Language Independent Options
1677 @section Options to Control Diagnostic Messages Formatting
1678 @cindex options to control diagnostics formatting
1679 @cindex diagnostic messages
1680 @cindex message formatting
1682 Traditionally, diagnostic messages have been formatted irrespective of
1683 the output device's aspect (e.g.@: its width, @dots{}). The options described
1684 below can be used to control the diagnostic messages formatting
1685 algorithm, e.g.@: how many characters per line, how often source location
1686 information should be reported. Right now, only the C++ front end can
1687 honor these options. However it is expected, in the near future, that
1688 the remaining front ends would be able to digest them correctly.
1691 @item -fmessage-length=@var{n}
1692 @opindex fmessage-length
1693 Try to format error messages so that they fit on lines of about @var{n}
1694 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1695 the front ends supported by GCC@. If @var{n} is zero, then no
1696 line-wrapping will be done; each error message will appear on a single
1699 @opindex fdiagnostics-show-location
1700 @item -fdiagnostics-show-location=once
1701 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1702 reporter to emit @emph{once} source location information; that is, in
1703 case the message is too long to fit on a single physical line and has to
1704 be wrapped, the source location won't be emitted (as prefix) again,
1705 over and over, in subsequent continuation lines. This is the default
1708 @item -fdiagnostics-show-location=every-line
1709 Only meaningful in line-wrapping mode. Instructs the diagnostic
1710 messages reporter to emit the same source location information (as
1711 prefix) for physical lines that result from the process of breaking a
1712 a message which is too long to fit on a single line.
1716 @node Warning Options
1717 @section Options to Request or Suppress Warnings
1718 @cindex options to control warnings
1719 @cindex warning messages
1720 @cindex messages, warning
1721 @cindex suppressing warnings
1723 Warnings are diagnostic messages that report constructions which
1724 are not inherently erroneous but which are risky or suggest there
1725 may have been an error.
1727 You can request many specific warnings with options beginning @samp{-W},
1728 for example @option{-Wimplicit} to request warnings on implicit
1729 declarations. Each of these specific warning options also has a
1730 negative form beginning @samp{-Wno-} to turn off warnings;
1731 for example, @option{-Wno-implicit}. This manual lists only one of the
1732 two forms, whichever is not the default.
1734 These options control the amount and kinds of warnings produced by GCC:
1737 @cindex syntax checking
1739 @opindex fsyntax-only
1740 Check the code for syntax errors, but don't do anything beyond that.
1744 Issue all the warnings demanded by strict ISO C and ISO C++;
1745 reject all programs that use forbidden extensions, and some other
1746 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1747 version of the ISO C standard specified by any @option{-std} option used.
1749 Valid ISO C and ISO C++ programs should compile properly with or without
1750 this option (though a rare few will require @option{-ansi} or a
1751 @option{-std} option specifying the required version of ISO C)@. However,
1752 without this option, certain GNU extensions and traditional C and C++
1753 features are supported as well. With this option, they are rejected.
1755 @option{-pedantic} does not cause warning messages for use of the
1756 alternate keywords whose names begin and end with @samp{__}. Pedantic
1757 warnings are also disabled in the expression that follows
1758 @code{__extension__}. However, only system header files should use
1759 these escape routes; application programs should avoid them.
1760 @xref{Alternate Keywords}.
1762 Some users try to use @option{-pedantic} to check programs for strict ISO
1763 C conformance. They soon find that it does not do quite what they want:
1764 it finds some non-ISO practices, but not all---only those for which
1765 ISO C @emph{requires} a diagnostic, and some others for which
1766 diagnostics have been added.
1768 A feature to report any failure to conform to ISO C might be useful in
1769 some instances, but would require considerable additional work and would
1770 be quite different from @option{-pedantic}. We don't have plans to
1771 support such a feature in the near future.
1773 Where the standard specified with @option{-std} represents a GNU
1774 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1775 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1776 extended dialect is based. Warnings from @option{-pedantic} are given
1777 where they are required by the base standard. (It would not make sense
1778 for such warnings to be given only for features not in the specified GNU
1779 C dialect, since by definition the GNU dialects of C include all
1780 features the compiler supports with the given option, and there would be
1781 nothing to warn about.)
1783 @item -pedantic-errors
1784 @opindex pedantic-errors
1785 Like @option{-pedantic}, except that errors are produced rather than
1790 Inhibit all warning messages.
1794 Inhibit warning messages about the use of @samp{#import}.
1796 @item -Wchar-subscripts
1797 @opindex Wchar-subscripts
1798 Warn if an array subscript has type @code{char}. This is a common cause
1799 of error, as programmers often forget that this type is signed on some
1804 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1805 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1809 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1810 the arguments supplied have types appropriate to the format string
1811 specified, and that the conversions specified in the format string make
1812 sense. This includes standard functions, and others specified by format
1813 attributes (@pxref{Function Attributes}), in the @code{printf},
1814 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1815 not in the C standard) families.
1817 The formats are checked against the format features supported by GNU
1818 libc version 2.2. These include all ISO C89 and C99 features, as well
1819 as features from the Single Unix Specification and some BSD and GNU
1820 extensions. Other library implementations may not support all these
1821 features; GCC does not support warning about features that go beyond a
1822 particular library's limitations. However, if @option{-pedantic} is used
1823 with @option{-Wformat}, warnings will be given about format features not
1824 in the selected standard version (but not for @code{strfmon} formats,
1825 since those are not in any version of the C standard). @xref{C Dialect
1826 Options,,Options Controlling C Dialect}.
1828 @option{-Wformat} is included in @option{-Wall}. For more control over some
1829 aspects of format checking, the options @option{-Wno-format-y2k},
1830 @option{-Wno-format-extra-args}, @option{-Wformat-nonliteral},
1831 @option{-Wformat-security} and @option{-Wformat=2} are available, but are
1832 not included in @option{-Wall}.
1834 @item -Wno-format-y2k
1835 @opindex Wno-format-y2k
1836 If @option{-Wformat} is specified, do not warn about @code{strftime}
1837 formats which may yield only a two-digit year.
1839 @item -Wno-format-extra-args
1840 @opindex Wno-format-extra-args
1841 If @option{-Wformat} is specified, do not warn about excess arguments to a
1842 @code{printf} or @code{scanf} format function. The C standard specifies
1843 that such arguments are ignored.
1845 @item -Wformat-nonliteral
1846 @opindex Wformat-nonliteral
1847 If @option{-Wformat} is specified, also warn if the format string is not a
1848 string literal and so cannot be checked, unless the format function
1849 takes its format arguments as a @code{va_list}.
1851 @item -Wformat-security
1852 @opindex Wformat-security
1853 If @option{-Wformat} is specified, also warn about uses of format
1854 functions that represent possible security problems. At present, this
1855 warns about calls to @code{printf} and @code{scanf} functions where the
1856 format string is not a string literal and there are no format arguments,
1857 as in @code{printf (foo);}. This may be a security hole if the format
1858 string came from untrusted input and contains @samp{%n}. (This is
1859 currently a subset of what @option{-Wformat-nonliteral} warns about, but
1860 in future warnings may be added to @option{-Wformat-security} that are not
1861 included in @option{-Wformat-nonliteral}.)
1865 Enable @option{-Wformat} plus format checks not included in
1866 @option{-Wformat}. Currently equivalent to @samp{-Wformat
1867 -Wformat-nonliteral -Wformat-security}.
1869 @item -Wimplicit-int
1870 @opindex Wimplicit-int
1871 Warn when a declaration does not specify a type.
1873 @item -Wimplicit-function-declaration
1874 @itemx -Werror-implicit-function-declaration
1875 @opindex Wimplicit-function-declaration
1876 @opindex Werror-implicit-function-declaration
1877 Give a warning (or error) whenever a function is used before being
1882 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
1886 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1887 function with external linkage, returning int, taking either zero
1888 arguments, two, or three arguments of appropriate types.
1890 @item -Wmissing-braces
1891 @opindex Wmissing-braces
1892 Warn if an aggregate or union initializer is not fully bracketed. In
1893 the following example, the initializer for @samp{a} is not fully
1894 bracketed, but that for @samp{b} is fully bracketed.
1897 int a[2][2] = @{ 0, 1, 2, 3 @};
1898 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
1903 Warn if a multicharacter constant (@samp{'FOOF'}) is used. Usually they
1904 indicate a typo in the user's code, as they have implementation-defined
1905 values, and should not be used in portable code.
1908 @opindex Wparentheses
1909 Warn if parentheses are omitted in certain contexts, such
1910 as when there is an assignment in a context where a truth value
1911 is expected, or when operators are nested whose precedence people
1912 often get confused about.
1914 Also warn about constructions where there may be confusion to which
1915 @code{if} statement an @code{else} branch belongs. Here is an example of
1930 In C, every @code{else} branch belongs to the innermost possible @code{if}
1931 statement, which in this example is @code{if (b)}. This is often not
1932 what the programmer expected, as illustrated in the above example by
1933 indentation the programmer chose. When there is the potential for this
1934 confusion, GCC will issue a warning when this flag is specified.
1935 To eliminate the warning, add explicit braces around the innermost
1936 @code{if} statement so there is no way the @code{else} could belong to
1937 the enclosing @code{if}. The resulting code would look like this:
1953 @item -Wsequence-point
1954 @opindex Wsequence-point
1955 Warn about code that may have undefined semantics because of violations
1956 of sequence point rules in the C standard.
1958 The C standard defines the order in which expressions in a C program are
1959 evaluated in terms of @dfn{sequence points}, which represent a partial
1960 ordering between the execution of parts of the program: those executed
1961 before the sequence point, and those executed after it. These occur
1962 after the evaluation of a full expression (one which is not part of a
1963 larger expression), after the evaluation of the first operand of a
1964 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
1965 function is called (but after the evaluation of its arguments and the
1966 expression denoting the called function), and in certain other places.
1967 Other than as expressed by the sequence point rules, the order of
1968 evaluation of subexpressions of an expression is not specified. All
1969 these rules describe only a partial order rather than a total order,
1970 since, for example, if two functions are called within one expression
1971 with no sequence point between them, the order in which the functions
1972 are called is not specified. However, the standards committee have
1973 ruled that function calls do not overlap.
1975 It is not specified when between sequence points modifications to the
1976 values of objects take effect. Programs whose behavior depends on this
1977 have undefined behavior; the C standard specifies that ``Between the
1978 previous and next sequence point an object shall have its stored value
1979 modified at most once by the evaluation of an expression. Furthermore,
1980 the prior value shall be read only to determine the value to be
1981 stored.''. If a program breaks these rules, the results on any
1982 particular implementation are entirely unpredictable.
1984 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
1985 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
1986 diagnosed by this option, and it may give an occasional false positive
1987 result, but in general it has been found fairly effective at detecting
1988 this sort of problem in programs.
1990 The present implementation of this option only works for C programs. A
1991 future implementation may also work for C++ programs.
1993 There is some controversy over the precise meaning of the sequence point
1994 rules in subtle cases. Links to papers with alternative formal definitions
1995 and other related discussions may be found on our readings page
1996 @w{@uref{http://gcc.gnu.org/readings.html}}.
1999 @opindex Wreturn-type
2000 Warn whenever a function is defined with a return-type that defaults to
2001 @code{int}. Also warn about any @code{return} statement with no
2002 return-value in a function whose return-type is not @code{void}.
2004 For C++, a function without return type always produces a diagnostic
2005 message, even when @option{-Wno-return-type} is specified. The only
2006 exceptions are @samp{main} and functions defined in system headers.
2010 Warn whenever a @code{switch} statement has an index of enumeral type
2011 and lacks a @code{case} for one or more of the named codes of that
2012 enumeration. (The presence of a @code{default} label prevents this
2013 warning.) @code{case} labels outside the enumeration range also
2014 provoke warnings when this option is used.
2018 Warn if any trigraphs are encountered that might change the meaning of
2019 the program (trigraphs within comments are not warned about).
2021 @item -Wunused-function
2022 @opindex Wunused-function
2023 Warn whenever a static function is declared but not defined or a
2024 non\-inline static function is unused.
2026 @item -Wunused-label
2027 @opindex Wunused-label
2028 Warn whenever a label is declared but not used.
2030 To suppress this warning use the @samp{unused} attribute
2031 (@pxref{Variable Attributes}).
2033 @item -Wunused-parameter
2034 @opindex Wunused-parameter
2035 Warn whenever a function parameter is unused aside from its declaration.
2037 To suppress this warning use the @samp{unused} attribute
2038 (@pxref{Variable Attributes}).
2040 @item -Wunused-variable
2041 @opindex Wunused-variable
2042 Warn whenever a local variable or non-constant static variable is unused
2043 aside from its declaration
2045 To suppress this warning use the @samp{unused} attribute
2046 (@pxref{Variable Attributes}).
2048 @item -Wunused-value
2049 @opindex Wunused-value
2050 Warn whenever a statement computes a result that is explicitly not used.
2052 To suppress this warning cast the expression to @samp{void}.
2056 All all the above @option{-Wunused} options combined.
2058 In order to get a warning about an unused function parameter, you must
2059 either specify @samp{-W -Wunused} or separately specify
2060 @option{-Wunused-parameter}.
2062 @item -Wuninitialized
2063 @opindex Wuninitialized
2064 Warn if an automatic variable is used without first being initialized or
2065 if a variable may be clobbered by a @code{setjmp} call.
2067 These warnings are possible only in optimizing compilation,
2068 because they require data flow information that is computed only
2069 when optimizing. If you don't specify @option{-O}, you simply won't
2072 These warnings occur only for variables that are candidates for
2073 register allocation. Therefore, they do not occur for a variable that
2074 is declared @code{volatile}, or whose address is taken, or whose size
2075 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2076 structures, unions or arrays, even when they are in registers.
2078 Note that there may be no warning about a variable that is used only
2079 to compute a value that itself is never used, because such
2080 computations may be deleted by data flow analysis before the warnings
2083 These warnings are made optional because GCC is not smart
2084 enough to see all the reasons why the code might be correct
2085 despite appearing to have an error. Here is one example of how
2106 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2107 always initialized, but GCC doesn't know this. Here is
2108 another common case:
2113 if (change_y) save_y = y, y = new_y;
2115 if (change_y) y = save_y;
2120 This has no bug because @code{save_y} is used only if it is set.
2122 @cindex @code{longjmp} warnings
2123 This option also warns when a non-volatile automatic variable might be
2124 changed by a call to @code{longjmp}. These warnings as well are possible
2125 only in optimizing compilation.
2127 The compiler sees only the calls to @code{setjmp}. It cannot know
2128 where @code{longjmp} will be called; in fact, a signal handler could
2129 call it at any point in the code. As a result, you may get a warning
2130 even when there is in fact no problem because @code{longjmp} cannot
2131 in fact be called at the place which would cause a problem.
2133 Some spurious warnings can be avoided if you declare all the functions
2134 you use that never return as @code{noreturn}. @xref{Function
2137 @item -Wreorder @r{(C++ only)}
2139 @cindex reordering, warning
2140 @cindex warning for reordering of member initializers
2141 Warn when the order of member initializers given in the code does not
2142 match the order in which they must be executed. For instance:
2144 @item -Wunknown-pragmas
2145 @opindex Wunknown-pragmas
2146 @cindex warning for unknown pragmas
2147 @cindex unknown pragmas, warning
2148 @cindex pragmas, warning of unknown
2149 Warn when a #pragma directive is encountered which is not understood by
2150 GCC@. If this command line option is used, warnings will even be issued
2151 for unknown pragmas in system header files. This is not the case if
2152 the warnings were only enabled by the @option{-Wall} command line option.
2156 All of the above @samp{-W} options combined. This enables all the
2157 warnings about constructions that some users consider questionable, and
2158 that are easy to avoid (or modify to prevent the warning), even in
2159 conjunction with macros.
2161 @item -Wsystem-headers
2162 @opindex Wsystem-headers
2163 @cindex warnings from system headers
2164 @cindex system headers, warnings from
2165 Print warning messages for constructs found in system header files.
2166 Warnings from system headers are normally suppressed, on the assumption
2167 that they usually do not indicate real problems and would only make the
2168 compiler output harder to read. Using this command line option tells
2169 GCC to emit warnings from system headers as if they occurred in user
2170 code. However, note that using @option{-Wall} in conjunction with this
2171 option will @emph{not} warn about unknown pragmas in system
2172 headers---for that, @option{-Wunknown-pragmas} must also be used.
2175 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2176 Some of them warn about constructions that users generally do not
2177 consider questionable, but which occasionally you might wish to check
2178 for; others warn about constructions that are necessary or hard to avoid
2179 in some cases, and there is no simple way to modify the code to suppress
2185 Print extra warning messages for these events:
2189 A function can return either with or without a value. (Falling
2190 off the end of the function body is considered returning without
2191 a value.) For example, this function would evoke such a
2205 An expression-statement or the left-hand side of a comma expression
2206 contains no side effects.
2207 To suppress the warning, cast the unused expression to void.
2208 For example, an expression such as @samp{x[i,j]} will cause a warning,
2209 but @samp{x[(void)i,j]} will not.
2212 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
2215 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2216 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2217 that of ordinary mathematical notation.
2220 Storage-class specifiers like @code{static} are not the first things in
2221 a declaration. According to the C Standard, this usage is obsolescent.
2224 The return type of a function has a type qualifier such as @code{const}.
2225 Such a type qualifier has no effect, since the value returned by a
2226 function is not an lvalue. (But don't warn about the GNU extension of
2227 @code{volatile void} return types. That extension will be warned about
2228 if @option{-pedantic} is specified.)
2231 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2235 A comparison between signed and unsigned values could produce an
2236 incorrect result when the signed value is converted to unsigned.
2237 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2240 An aggregate has a partly bracketed initializer.
2241 For example, the following code would evoke such a warning,
2242 because braces are missing around the initializer for @code{x.h}:
2245 struct s @{ int f, g; @};
2246 struct t @{ struct s h; int i; @};
2247 struct t x = @{ 1, 2, 3 @};
2251 An aggregate has an initializer which does not initialize all members.
2252 For example, the following code would cause such a warning, because
2253 @code{x.h} would be implicitly initialized to zero:
2256 struct s @{ int f, g, h; @};
2257 struct s x = @{ 3, 4 @};
2262 @opindex Wfloat-equal
2263 Warn if floating point values are used in equality comparisons.
2265 The idea behind this is that sometimes it is convenient (for the
2266 programmer) to consider floating-point values as approximations to
2267 infinitely precise real numbers. If you are doing this, then you need
2268 to compute (by analysing the code, or in some other way) the maximum or
2269 likely maximum error that the computation introduces, and allow for it
2270 when performing comparisons (and when producing output, but that's a
2271 different problem). In particular, instead of testing for equality, you
2272 would check to see whether the two values have ranges that overlap; and
2273 this is done with the relational operators, so equality comparisons are
2276 @item -Wtraditional @r{(C only)}
2277 @opindex Wtraditional
2278 Warn about certain constructs that behave differently in traditional and
2279 ISO C@. Also warn about ISO C constructs that have no traditional C
2280 equivalent, and/or problematic constructs which should be avoided.
2284 Macro parameters that appear within string literals in the macro body.
2285 In traditional C macro replacement takes place within string literals,
2286 but does not in ISO C@.
2289 In traditional C, some preprocessor directives did not exist.
2290 Traditional preprocessors would only consider a line to be a directive
2291 if the @samp{#} appeared in column 1 on the line. Therefore
2292 @option{-Wtraditional} warns about directives that traditional C
2293 understands but would ignore because the @samp{#} does not appear as the
2294 first character on the line. It also suggests you hide directives like
2295 @samp{#pragma} not understood by traditional C by indenting them. Some
2296 traditional implementations would not recognise @samp{#elif}, so it
2297 suggests avoiding it altogether.
2300 A function-like macro that appears without arguments.
2303 The unary plus operator.
2306 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2307 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2308 constants.) Note, these suffixes appear in macros defined in the system
2309 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2310 Use of these macros in user code might normally lead to spurious
2311 warnings, however gcc's integrated preprocessor has enough context to
2312 avoid warning in these cases.
2315 A function declared external in one block and then used after the end of
2319 A @code{switch} statement has an operand of type @code{long}.
2322 A non-@code{static} function declaration follows a @code{static} one.
2323 This construct is not accepted by some traditional C compilers.
2326 The ISO type of an integer constant has a different width or
2327 signedness from its traditional type. This warning is only issued if
2328 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2329 typically represent bit patterns, are not warned about.
2332 Usage of ISO string concatenation is detected.
2335 Initialization of automatic aggregates.
2338 Identifier conflicts with labels. Traditional C lacks a separate
2339 namespace for labels.
2342 Initialization of unions. If the initializer is zero, the warning is
2343 omitted. This is done under the assumption that the zero initializer in
2344 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2345 initializer warnings and relies on default initialization to zero in the
2349 Conversions by prototypes between fixed/floating point values and vice
2350 versa. The absence of these prototypes when compiling with traditional
2351 C would cause serious problems. This is a subset of the possible
2352 conversion warnings, for the full set use @option{-Wconversion}.
2357 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2361 Warn whenever a local variable shadows another local variable, parameter or
2362 global variable or whenever a built-in function is shadowed.
2364 @item -Wlarger-than-@var{len}
2365 @opindex Wlarger-than
2366 Warn whenever an object of larger than @var{len} bytes is defined.
2368 @item -Wpointer-arith
2369 @opindex Wpointer-arith
2370 Warn about anything that depends on the ``size of'' a function type or
2371 of @code{void}. GNU C assigns these types a size of 1, for
2372 convenience in calculations with @code{void *} pointers and pointers
2375 @item -Wbad-function-cast @r{(C only)}
2376 @opindex Wbad-function-cast
2377 Warn whenever a function call is cast to a non-matching type.
2378 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2382 Warn whenever a pointer is cast so as to remove a type qualifier from
2383 the target type. For example, warn if a @code{const char *} is cast
2384 to an ordinary @code{char *}.
2387 @opindex Wcast-align
2388 Warn whenever a pointer is cast such that the required alignment of the
2389 target is increased. For example, warn if a @code{char *} is cast to
2390 an @code{int *} on machines where integers can only be accessed at
2391 two- or four-byte boundaries.
2393 @item -Wwrite-strings
2394 @opindex Wwrite-strings
2395 When compiling C, give string constants the type @code{const
2396 char[@var{length}]} so that
2397 copying the address of one into a non-@code{const} @code{char *}
2398 pointer will get a warning; when compiling C++, warn about the
2399 deprecated conversion from string constants to @code{char *}.
2400 These warnings will help you find at
2401 compile time code that can try to write into a string constant, but
2402 only if you have been very careful about using @code{const} in
2403 declarations and prototypes. Otherwise, it will just be a nuisance;
2404 this is why we did not make @option{-Wall} request these warnings.
2407 @opindex Wconversion
2408 Warn if a prototype causes a type conversion that is different from what
2409 would happen to the same argument in the absence of a prototype. This
2410 includes conversions of fixed point to floating and vice versa, and
2411 conversions changing the width or signedness of a fixed point argument
2412 except when the same as the default promotion.
2414 Also, warn if a negative integer constant expression is implicitly
2415 converted to an unsigned type. For example, warn about the assignment
2416 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2417 casts like @code{(unsigned) -1}.
2419 @item -Wsign-compare
2420 @opindex Wsign-compare
2421 @cindex warning for comparison of signed and unsigned values
2422 @cindex comparison of signed and unsigned values, warning
2423 @cindex signed and unsigned values, comparison warning
2424 Warn when a comparison between signed and unsigned values could produce
2425 an incorrect result when the signed value is converted to unsigned.
2426 This warning is also enabled by @option{-W}; to get the other warnings
2427 of @option{-W} without this warning, use @samp{-W -Wno-sign-compare}.
2429 @item -Waggregate-return
2430 @opindex Waggregate-return
2431 Warn if any functions that return structures or unions are defined or
2432 called. (In languages where you can return an array, this also elicits
2435 @item -Wstrict-prototypes @r{(C only)}
2436 @opindex Wstrict-prototypes
2437 Warn if a function is declared or defined without specifying the
2438 argument types. (An old-style function definition is permitted without
2439 a warning if preceded by a declaration which specifies the argument
2442 @item -Wmissing-prototypes @r{(C only)}
2443 @opindex Wmissing-prototypes
2444 Warn if a global function is defined without a previous prototype
2445 declaration. This warning is issued even if the definition itself
2446 provides a prototype. The aim is to detect global functions that fail
2447 to be declared in header files.
2449 @item -Wmissing-declarations
2450 @opindex Wmissing-declarations
2451 Warn if a global function is defined without a previous declaration.
2452 Do so even if the definition itself provides a prototype.
2453 Use this option to detect global functions that are not declared in
2456 @item -Wmissing-noreturn
2457 @opindex Wmissing-noreturn
2458 Warn about functions which might be candidates for attribute @code{noreturn}.
2459 Note these are only possible candidates, not absolute ones. Care should
2460 be taken to manually verify functions actually do not ever return before
2461 adding the @code{noreturn} attribute, otherwise subtle code generation
2462 bugs could be introduced. You will not get a warning for @code{main} in
2463 hosted C environments.
2465 @item -Wmissing-format-attribute
2466 @opindex Wmissing-format-attribute
2468 If @option{-Wformat} is enabled, also warn about functions which might be
2469 candidates for @code{format} attributes. Note these are only possible
2470 candidates, not absolute ones. GCC will guess that @code{format}
2471 attributes might be appropriate for any function that calls a function
2472 like @code{vprintf} or @code{vscanf}, but this might not always be the
2473 case, and some functions for which @code{format} attributes are
2474 appropriate may not be detected. This option has no effect unless
2475 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2479 Warn if a structure is given the packed attribute, but the packed
2480 attribute has no effect on the layout or size of the structure.
2481 Such structures may be mis-aligned for little benefit. For
2482 instance, in this code, the variable @code{f.x} in @code{struct bar}
2483 will be misaligned even though @code{struct bar} does not itself
2484 have the packed attribute:
2491 @} __attribute__((packed));
2501 Warn if padding is included in a structure, either to align an element
2502 of the structure or to align the whole structure. Sometimes when this
2503 happens it is possible to rearrange the fields of the structure to
2504 reduce the padding and so make the structure smaller.
2506 @item -Wredundant-decls
2507 @opindex Wredundant-decls
2508 Warn if anything is declared more than once in the same scope, even in
2509 cases where multiple declaration is valid and changes nothing.
2511 @item -Wnested-externs @r{(C only)}
2512 @opindex Wnested-externs
2513 Warn if an @code{extern} declaration is encountered within a function.
2515 @item -Wunreachable-code
2516 @opindex Wunreachable-code
2517 Warn if the compiler detects that code will never be executed.
2519 This option is intended to warn when the compiler detects that at
2520 least a whole line of source code will never be executed, because
2521 some condition is never satisfied or because it is after a
2522 procedure that never returns.
2524 It is possible for this option to produce a warning even though there
2525 are circumstances under which part of the affected line can be executed,
2526 so care should be taken when removing apparently-unreachable code.
2528 For instance, when a function is inlined, a warning may mean that the
2529 line is unreachable in only one inlined copy of the function.
2531 This option is not made part of @option{-Wall} because in a debugging
2532 version of a program there is often substantial code which checks
2533 correct functioning of the program and is, hopefully, unreachable
2534 because the program does work. Another common use of unreachable
2535 code is to provide behaviour which is selectable at compile-time.
2539 Warn if a function can not be inlined and it was declared as inline.
2543 @opindex Wno-long-long
2544 Warn if @samp{long long} type is used. This is default. To inhibit
2545 the warning messages, use @option{-Wno-long-long}. Flags
2546 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2547 only when @option{-pedantic} flag is used.
2549 @item -Wdisabled-optimization
2550 @opindex Wdisabled-optimization
2551 Warn if a requested optimization pass is disabled. This warning does
2552 not generally indicate that there is anything wrong with your code; it
2553 merely indicates that GCC's optimizers were unable to handle the code
2554 effectively. Often, the problem is that your code is too big or too
2555 complex; GCC will refuse to optimize programs when the optimization
2556 itself is likely to take inordinate amounts of time.
2560 Make all warnings into errors.
2563 @node Debugging Options
2564 @section Options for Debugging Your Program or GCC
2565 @cindex options, debugging
2566 @cindex debugging information options
2568 GCC has various special options that are used for debugging
2569 either your program or GCC:
2574 Produce debugging information in the operating system's native format
2575 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2578 On most systems that use stabs format, @option{-g} enables use of extra
2579 debugging information that only GDB can use; this extra information
2580 makes debugging work better in GDB but will probably make other debuggers
2582 refuse to read the program. If you want to control for certain whether
2583 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
2584 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, or @option{-gdwarf-1}
2587 Unlike most other C compilers, GCC allows you to use @option{-g} with
2588 @option{-O}. The shortcuts taken by optimized code may occasionally
2589 produce surprising results: some variables you declared may not exist
2590 at all; flow of control may briefly move where you did not expect it;
2591 some statements may not be executed because they compute constant
2592 results or their values were already at hand; some statements may
2593 execute in different places because they were moved out of loops.
2595 Nevertheless it proves possible to debug optimized output. This makes
2596 it reasonable to use the optimizer for programs that might have bugs.
2598 The following options are useful when GCC is generated with the
2599 capability for more than one debugging format.
2603 Produce debugging information for use by GDB@. This means to use the
2604 most expressive format available (DWARF 2, stabs, or the native format
2605 if neither of those are supported), including GDB extensions if at all
2610 Produce debugging information in stabs format (if that is supported),
2611 without GDB extensions. This is the format used by DBX on most BSD
2612 systems. On MIPS, Alpha and System V Release 4 systems this option
2613 produces stabs debugging output which is not understood by DBX or SDB@.
2614 On System V Release 4 systems this option requires the GNU assembler.
2618 Produce debugging information in stabs format (if that is supported),
2619 using GNU extensions understood only by the GNU debugger (GDB)@. The
2620 use of these extensions is likely to make other debuggers crash or
2621 refuse to read the program.
2625 Produce debugging information in COFF format (if that is supported).
2626 This is the format used by SDB on most System V systems prior to
2631 Produce debugging information in XCOFF format (if that is supported).
2632 This is the format used by the DBX debugger on IBM RS/6000 systems.
2636 Produce debugging information in XCOFF format (if that is supported),
2637 using GNU extensions understood only by the GNU debugger (GDB)@. The
2638 use of these extensions is likely to make other debuggers crash or
2639 refuse to read the program, and may cause assemblers other than the GNU
2640 assembler (GAS) to fail with an error.
2644 Produce debugging information in DWARF version 1 format (if that is
2645 supported). This is the format used by SDB on most System V Release 4
2650 Produce debugging information in DWARF version 1 format (if that is
2651 supported), using GNU extensions understood only by the GNU debugger
2652 (GDB)@. The use of these extensions is likely to make other debuggers
2653 crash or refuse to read the program.
2657 Produce debugging information in DWARF version 2 format (if that is
2658 supported). This is the format used by DBX on IRIX 6.
2661 @itemx -ggdb@var{level}
2662 @itemx -gstabs@var{level}
2663 @itemx -gcoff@var{level}
2664 @itemx -gxcoff@var{level}
2665 @itemx -gdwarf@var{level}
2666 @itemx -gdwarf-2@var{level}
2667 Request debugging information and also use @var{level} to specify how
2668 much information. The default level is 2.
2670 Level 1 produces minimal information, enough for making backtraces in
2671 parts of the program that you don't plan to debug. This includes
2672 descriptions of functions and external variables, but no information
2673 about local variables and no line numbers.
2675 Level 3 includes extra information, such as all the macro definitions
2676 present in the program. Some debuggers support macro expansion when
2677 you use @option{-g3}.
2682 Generate extra code to write profile information suitable for the
2683 analysis program @code{prof}. You must use this option when compiling
2684 the source files you want data about, and you must also use it when
2687 @cindex @code{gprof}
2690 Generate extra code to write profile information suitable for the
2691 analysis program @code{gprof}. You must use this option when compiling
2692 the source files you want data about, and you must also use it when
2698 Generate extra code to write profile information for basic blocks, which will
2699 record the number of times each basic block is executed, the basic block start
2700 address, and the function name containing the basic block. If @option{-g} is
2701 used, the line number and filename of the start of the basic block will also be
2702 recorded. If not overridden by the machine description, the default action is
2703 to append to the text file @file{bb.out}.
2705 This data could be analyzed by a program like @code{tcov}. Note,
2706 however, that the format of the data is not what @code{tcov} expects.
2707 Eventually GNU @code{gprof} should be extended to process this data.
2711 Makes the compiler print out each function name as it is compiled, and
2712 print some statistics about each pass when it finishes.
2715 @opindex ftime-report
2716 Makes the compiler print some statistics about the time consumed by each
2717 pass when it finishes.
2720 @opindex fmem-report
2721 Makes the compiler print some statistics about permanent memory
2722 allocation when it finishes.
2726 Generate extra code to profile basic blocks. Your executable will
2727 produce output that is a superset of that produced when @option{-a} is
2728 used. Additional output is the source and target address of the basic
2729 blocks where a jump takes place, the number of times a jump is executed,
2730 and (optionally) the complete sequence of basic blocks being executed.
2731 The output is appended to file @file{bb.out}.
2733 You can examine different profiling aspects without recompilation. Your
2734 executable will read a list of function names from file @file{bb.in}.
2735 Profiling starts when a function on the list is entered and stops when
2736 that invocation is exited. To exclude a function from profiling, prefix
2737 its name with @samp{-}. If a function name is not unique, you can
2738 disambiguate it by writing it in the form
2739 @samp{/path/filename.d:functionname}. Your executable will write the
2740 available paths and filenames in file @file{bb.out}.
2742 Several function names have a special meaning:
2745 Write source, target and frequency of jumps to file @file{bb.out}.
2746 @item __bb_hidecall__
2747 Exclude function calls from frequency count.
2748 @item __bb_showret__
2749 Include function returns in frequency count.
2751 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
2752 The file will be compressed using the program @samp{gzip}, which must
2753 exist in your @env{PATH}. On systems without the @samp{popen}
2754 function, the file will be named @file{bbtrace} and will not be
2755 compressed. @strong{Profiling for even a few seconds on these systems
2756 will produce a very large file.} Note: @code{__bb_hidecall__} and
2757 @code{__bb_showret__} will not affect the sequence written to
2761 Here's a short example using different profiling parameters
2762 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
2763 1 and 2 and is called twice from block 3 of function @code{main}. After
2764 the calls, block 3 transfers control to block 4 of @code{main}.
2766 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
2767 the following sequence of blocks is written to file @file{bbtrace.gz}:
2768 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
2769 the return is to a point inside the block and not to the top. The
2770 block address 0 always indicates, that control is transferred
2771 to the trace from somewhere outside the observed functions. With
2772 @samp{-foo} added to @file{bb.in}, the blocks of function
2773 @code{foo} are removed from the trace, so only 0 3 4 remains.
2775 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
2776 jump frequencies will be written to file @file{bb.out}. The
2777 frequencies are obtained by constructing a trace of blocks
2778 and incrementing a counter for every neighbouring pair of blocks
2779 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2783 Jump from block 0x0 to block 0x3 executed 1 time(s)
2784 Jump from block 0x3 to block 0x1 executed 1 time(s)
2785 Jump from block 0x1 to block 0x2 executed 2 time(s)
2786 Jump from block 0x2 to block 0x1 executed 1 time(s)
2787 Jump from block 0x2 to block 0x4 executed 1 time(s)
2790 With @code{__bb_hidecall__}, control transfer due to call instructions
2791 is removed from the trace, that is the trace is cut into three parts: 0
2792 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2793 to return instructions is added to the trace. The trace becomes: 0 3 1
2794 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2795 written to @file{bbtrace.gz}. It is solely used for counting jump
2798 @item -fprofile-arcs
2799 @opindex fprofile-arcs
2800 Instrument @dfn{arcs} during compilation. For each function of your
2801 program, GCC creates a program flow graph, then finds a spanning tree
2802 for the graph. Only arcs that are not on the spanning tree have to be
2803 instrumented: the compiler adds code to count the number of times that these
2804 arcs are executed. When an arc is the only exit or only entrance to a
2805 block, the instrumentation code can be added to the block; otherwise, a
2806 new basic block must be created to hold the instrumentation code.
2808 Since not every arc in the program must be instrumented, programs
2809 compiled with this option run faster than programs compiled with
2810 @option{-a}, which adds instrumentation code to every basic block in the
2811 program. The tradeoff: since @code{gcov} does not have
2812 execution counts for all branches, it must start with the execution
2813 counts for the instrumented branches, and then iterate over the program
2814 flow graph until the entire graph has been solved. Hence, @code{gcov}
2815 runs a little more slowly than a program which uses information from
2818 @option{-fprofile-arcs} also makes it possible to estimate branch
2819 probabilities, and to calculate basic block execution counts. In
2820 general, basic block execution counts do not give enough information to
2821 estimate all branch probabilities. When the compiled program exits, it
2822 saves the arc execution counts to a file called
2823 @file{@var{sourcename}.da}. Use the compiler option
2824 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2825 Control Optimization}) when recompiling, to optimize using estimated
2826 branch probabilities.
2829 @item -ftest-coverage
2830 @opindex ftest-coverage
2831 Create data files for the @code{gcov} code-coverage utility
2832 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2833 The data file names begin with the name of your source file:
2836 @item @var{sourcename}.bb
2837 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2838 associate basic block execution counts with line numbers.
2840 @item @var{sourcename}.bbg
2841 A list of all arcs in the program flow graph. This allows @code{gcov}
2842 to reconstruct the program flow graph, so that it can compute all basic
2843 block and arc execution counts from the information in the
2844 @code{@var{sourcename}.da} file (this last file is the output from
2845 @option{-fprofile-arcs}).
2848 @item -d@var{letters}
2850 Says to make debugging dumps during compilation at times specified by
2851 @var{letters}. This is used for debugging the compiler. The file names
2852 for most of the dumps are made by appending a pass number and a word to
2853 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.sibling}).
2854 Here are the possible letters for use in @var{letters}, and their meanings:
2859 Annotate the assembler output with miscellaneous debugging information.
2862 Dump after computing branch probabilities, to @file{@var{file}.13.bp}.
2865 Dump after block reordering, to @file{@var{file}.27.bbro}.
2868 Dump after instruction combination, to the file @file{@var{file}.15.combine}.
2871 Dump after the first if conversion, to the file @file{@var{file}.16.ce}.
2874 Dump after delayed branch scheduling, to @file{@var{file}.30.dbr}.
2877 Dump all macro definitions, at the end of preprocessing, in addition to
2881 Dump after SSA optimizations, to @file{@var{file}.04.ssa} and
2882 @file{@var{file}.06.ussa}.
2885 Dump after the second if conversion, to @file{@var{file}.25.ce2}.
2888 Dump after life analysis, to @file{@var{file}.14.life}.
2891 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.08.addressof}.
2894 Dump after global register allocation, to @file{@var{file}.20.greg}.
2897 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
2901 Dump after post-reload CSE and other optimizations, to @file{@var{file}.21.postreload}.
2904 Dump after GCSE, to @file{@var{file}.09.gcse}.
2907 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
2910 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
2913 Dump after the last jump optimization, to @file{@var{file}.28.jump2}.
2916 Dump after conversion from registers to stack, to @file{@var{file}.31.stack}.
2919 Dump after local register allocation, to @file{@var{file}.19.lreg}.
2922 Dump after loop optimization, to @file{@var{file}.10.loop}.
2925 Dump after performing the machine dependent reorganisation pass, to
2926 @file{@var{file}.29.mach}.
2929 Dump after register renumbering, to @file{@var{file}.24.rnreg}.
2932 Dump after the register move pass, to @file{@var{file}.17.regmove}.
2935 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2938 Dump after the second instruction scheduling pass, to
2939 @file{@var{file}.26.sched2}.
2942 Dump after CSE (including the jump optimization that sometimes follows
2943 CSE), to @file{@var{file}.07.cse}.
2946 Dump after the first instruction scheduling pass, to
2947 @file{@var{file}.18.sched}.
2950 Dump after the second CSE pass (including the jump optimization that
2951 sometimes follows CSE), to @file{@var{file}.11.cse2}.
2954 Dump after the second flow pass, to @file{@var{file}.22.flow2}.
2957 Dump after SSA aggressive dead code elimination, to @file{@var{file}.05.ssadce}.
2960 Dump after the peephole pass, to @file{@var{file}.23.peephole2}.
2963 Produce all the dumps listed above.
2966 Print statistics on memory usage, at the end of the run, to
2970 Annotate the assembler output with a comment indicating which
2971 pattern and alternative was used. The length of each instruction is
2975 Dump the RTL in the assembler output as a comment before each instruction.
2976 Also turns on @option{-dp} annotation.
2979 For each of the other indicated dump files (except for
2980 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
2981 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
2984 Just generate RTL for a function instead of compiling it. Usually used
2988 Dump debugging information during parsing, to standard error.
2991 @item -fdump-unnumbered
2992 @opindex fdump-unnumbered
2993 When doing debugging dumps (see @option{-d} option above), suppress instruction
2994 numbers and line number note output. This makes it more feasible to
2995 use diff on debugging dumps for compiler invocations with different
2996 options, in particular with and without @option{-g}.
2998 @item -fdump-translation-unit @r{(C and C++ only)}
2999 @itemx -fdump-translation-unit-@var{number} @r{(C and C++ only)}
3000 @opindex fdump-translation-unit
3001 Dump a representation of the tree structure for the entire translation
3002 unit to a file. The file name is made by appending @file{.tu} to the
3003 source file name. If the @samp{-@var{number}} form is used, @var{number}
3004 controls the details of the dump as described for the @option{-fdump-tree} options.
3006 @item -fdump-class-hierarchy @r{(C++ only)}
3007 @itemx -fdump-class-hierarchy-@var{number} @r{(C++ only)}
3008 @opindex fdump-class-hierarchy
3009 Dump a representation of each class's hierarchy and virtual function
3010 table layout to a file. The file name is made by appending @file{.class}
3011 to the source file name. If the @samp{-@var{number}} form is used, @var{number}
3012 controls the details of the dump as described for the @option{-fdump-tree}
3015 @item -fdump-ast-@var{switch} @r{(C++ only)}
3016 @itemx -fdump-ast-@var{switch}-@var{number} @r{(C++ only)}
3018 Control the dumping at various stages of processing the abstract syntax
3019 tree to a file. The file name is generated by appending a switch
3020 specific suffix to the source file name. If the @samp{-@var{number}} form is
3021 used, @var{number} is a bit mask which controls the details of the
3022 dump. The following bits are meaningful (these are not set symbolically,
3023 as the primary function of these dumps is for debugging gcc itself):
3027 Print the address of each node. Usually this is not meaningful as it
3028 changes according to the environment and source file.
3030 Inhibit dumping of members of a scope or body of a function, unless they
3031 are reachable by some other path.
3034 The following tree dumps are possible:
3037 Dump before any tree based optimization, to @file{@var{file}.original}.
3039 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3041 Dump after inlining within the body of the function, to
3042 @file{@var{file}.inlined}.
3045 @item -fpretend-float
3046 @opindex fpretend-float
3047 When running a cross-compiler, pretend that the target machine uses the
3048 same floating point format as the host machine. This causes incorrect
3049 output of the actual floating constants, but the actual instruction
3050 sequence will probably be the same as GCC would make when running on
3055 Store the usual ``temporary'' intermediate files permanently; place them
3056 in the current directory and name them based on the source file. Thus,
3057 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3058 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3059 preprocessed @file{foo.i} output file even though the compiler now
3060 normally uses an integrated preprocessor.
3064 Report the CPU time taken by each subprocess in the compilation
3065 sequence. For C source files, this is the compiler proper and assembler
3066 (plus the linker if linking is done). The output looks like this:
3073 The first number on each line is the ``user time,'' that is time spent
3074 executing the program itself. The second number is ``system time,''
3075 time spent executing operating system routines on behalf of the program.
3076 Both numbers are in seconds.
3078 @item -print-file-name=@var{library}
3079 @opindex print-file-name
3080 Print the full absolute name of the library file @var{library} that
3081 would be used when linking---and don't do anything else. With this
3082 option, GCC does not compile or link anything; it just prints the
3085 @item -print-multi-directory
3086 @opindex print-multi-directory
3087 Print the directory name corresponding to the multilib selected by any
3088 other switches present in the command line. This directory is supposed
3089 to exist in @env{GCC_EXEC_PREFIX}.
3091 @item -print-multi-lib
3092 @opindex print-multi-lib
3093 Print the mapping from multilib directory names to compiler switches
3094 that enable them. The directory name is separated from the switches by
3095 @samp{;}, and each switch starts with an @samp{@@} instead of the
3096 @samp{-}, without spaces between multiple switches. This is supposed to
3097 ease shell-processing.
3099 @item -print-prog-name=@var{program}
3100 @opindex print-prog-name
3101 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3103 @item -print-libgcc-file-name
3104 @opindex print-libgcc-file-name
3105 Same as @option{-print-file-name=libgcc.a}.
3107 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3108 but you do want to link with @file{libgcc.a}. You can do
3111 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3114 @item -print-search-dirs
3115 @opindex print-search-dirs
3116 Print the name of the configured installation directory and a list of
3117 program and library directories gcc will search---and don't do anything else.
3119 This is useful when gcc prints the error message
3120 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3121 To resolve this you either need to put @file{cpp0} and the other compiler
3122 components where gcc expects to find them, or you can set the environment
3123 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3124 Don't forget the trailing '/'.
3125 @xref{Environment Variables}.
3128 @opindex dumpmachine
3129 Print the compiler's target machine (for example,
3130 @samp{i686-pc-linux-gnu})---and don't do anything else.
3133 @opindex dumpversion
3134 Print the compiler version (for example, @samp{3.0})---and don't do
3139 Print the compiler's built-in specs---and don't do anything else. (This
3140 is used when GCC itself is being built.) @xref{Spec Files}.
3143 @node Optimize Options
3144 @section Options That Control Optimization
3145 @cindex optimize options
3146 @cindex options, optimization
3148 These options control various sorts of optimizations:
3155 Optimize. Optimizing compilation takes somewhat more time, and a lot
3156 more memory for a large function.
3158 Without @option{-O}, the compiler's goal is to reduce the cost of
3159 compilation and to make debugging produce the expected results.
3160 Statements are independent: if you stop the program with a breakpoint
3161 between statements, you can then assign a new value to any variable or
3162 change the program counter to any other statement in the function and
3163 get exactly the results you would expect from the source code.
3165 Without @option{-O}, the compiler only allocates variables declared
3166 @code{register} in registers. The resulting compiled code is a little
3167 worse than produced by PCC without @option{-O}.
3169 With @option{-O}, the compiler tries to reduce code size and execution
3172 When you specify @option{-O}, the compiler turns on @option{-fthread-jumps}
3173 and @option{-fdefer-pop} on all machines. The compiler turns on
3174 @option{-fdelayed-branch} on machines that have delay slots, and
3175 @option{-fomit-frame-pointer} on machines that can support debugging even
3176 without a frame pointer. On some machines the compiler also turns
3181 Optimize even more. GCC performs nearly all supported optimizations
3182 that do not involve a space-speed tradeoff. The compiler does not
3183 perform loop unrolling or function inlining when you specify @option{-O2}.
3184 As compared to @option{-O}, this option increases both compilation time
3185 and the performance of the generated code.
3187 @option{-O2} turns on all optional optimizations except for loop unrolling,
3188 function inlining, and register renaming. It also turns on the
3189 @option{-fforce-mem} option on all machines and frame pointer elimination
3190 on machines where doing so does not interfere with debugging.
3194 Optimize yet more. @option{-O3} turns on all optimizations specified by
3195 @option{-O2} and also turns on the @option{-finline-functions} and
3196 @option{-frename-registers} options.
3204 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3205 do not typically increase code size. It also performs further
3206 optimizations designed to reduce code size.
3208 If you use multiple @option{-O} options, with or without level numbers,
3209 the last such option is the one that is effective.
3212 Options of the form @option{-f@var{flag}} specify machine-independent
3213 flags. Most flags have both positive and negative forms; the negative
3214 form of @option{-ffoo} would be @option{-fno-foo}. In the table below,
3215 only one of the forms is listed---the one which is not the default.
3216 You can figure out the other form by either removing @samp{no-} or
3221 @opindex ffloat-store
3222 Do not store floating point variables in registers, and inhibit other
3223 options that might change whether a floating point value is taken from a
3226 @cindex floating point precision
3227 This option prevents undesirable excess precision on machines such as
3228 the 68000 where the floating registers (of the 68881) keep more
3229 precision than a @code{double} is supposed to have. Similarly for the
3230 x86 architecture. For most programs, the excess precision does only
3231 good, but a few programs rely on the precise definition of IEEE floating
3232 point. Use @option{-ffloat-store} for such programs, after modifying
3233 them to store all pertinent intermediate computations into variables.
3235 @item -fno-default-inline
3236 @opindex fno-default-inline
3237 Do not make member functions inline by default merely because they are
3238 defined inside the class scope (C++ only). Otherwise, when you specify
3239 @w{@option{-O}}, member functions defined inside class scope are compiled
3240 inline by default; i.e., you don't need to add @samp{inline} in front of
3241 the member function name.
3243 @item -fno-defer-pop
3244 @opindex fno-defer-pop
3245 Always pop the arguments to each function call as soon as that function
3246 returns. For machines which must pop arguments after a function call,
3247 the compiler normally lets arguments accumulate on the stack for several
3248 function calls and pops them all at once.
3252 Force memory operands to be copied into registers before doing
3253 arithmetic on them. This produces better code by making all memory
3254 references potential common subexpressions. When they are not common
3255 subexpressions, instruction combination should eliminate the separate
3256 register-load. The @option{-O2} option turns on this option.
3259 @opindex fforce-addr
3260 Force memory address constants to be copied into registers before
3261 doing arithmetic on them. This may produce better code just as
3262 @option{-fforce-mem} may.
3264 @item -fomit-frame-pointer
3265 @opindex fomit-frame-pointer
3266 Don't keep the frame pointer in a register for functions that
3267 don't need one. This avoids the instructions to save, set up and
3268 restore frame pointers; it also makes an extra register available
3269 in many functions. @strong{It also makes debugging impossible on
3273 On some machines, such as the Vax, this flag has no effect, because
3274 the standard calling sequence automatically handles the frame pointer
3275 and nothing is saved by pretending it doesn't exist. The
3276 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3277 whether a target machine supports this flag. @xref{Registers}.
3280 On some machines, such as the Vax, this flag has no effect, because
3281 the standard calling sequence automatically handles the frame pointer
3282 and nothing is saved by pretending it doesn't exist. The
3283 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3284 whether a target machine supports this flag. @xref{Registers,,Register
3285 Usage, gcc.info, Using and Porting GCC}.
3288 @item -foptimize-sibling-calls
3289 @opindex foptimize-sibling-calls
3290 Optimize sibling and tail recursive calls.
3294 This option generates traps for signed overflow on addition, subtraction,
3295 multiplication operations.
3299 Don't pay attention to the @code{inline} keyword. Normally this option
3300 is used to keep the compiler from expanding any functions inline.
3301 Note that if you are not optimizing, no functions can be expanded inline.
3303 @item -finline-functions
3304 @opindex finline-functions
3305 Integrate all simple functions into their callers. The compiler
3306 heuristically decides which functions are simple enough to be worth
3307 integrating in this way.
3309 If all calls to a given function are integrated, and the function is
3310 declared @code{static}, then the function is normally not output as
3311 assembler code in its own right.
3313 @item -finline-limit=@var{n}
3314 @opindex finline-limit
3315 By default, gcc limits the size of functions that can be inlined. This flag
3316 allows the control of this limit for functions that are explicitly marked as
3317 inline (ie marked with the inline keyword or defined within the class
3318 definition in c++). @var{n} is the size of functions that can be inlined in
3319 number of pseudo instructions (not counting parameter handling). The default
3320 value of @var{n} is 10000. Increasing this value can result in more inlined code at
3321 the cost of compilation time and memory consumption. Decreasing usually makes
3322 the compilation faster and less code will be inlined (which presumably
3323 means slower programs). This option is particularly useful for programs that
3324 use inlining heavily such as those based on recursive templates with C++.
3326 @emph{Note:} pseudo instruction represents, in this particular context, an
3327 abstract measurement of function's size. In no way, it represents a count
3328 of assembly instructions and as such its exact meaning might change from one
3329 release to an another.
3331 @item -fkeep-inline-functions
3332 @opindex fkeep-inline-functions
3333 Even if all calls to a given function are integrated, and the function
3334 is declared @code{static}, nevertheless output a separate run-time
3335 callable version of the function. This switch does not affect
3336 @code{extern inline} functions.
3338 @item -fkeep-static-consts
3339 @opindex fkeep-static-consts
3340 Emit variables declared @code{static const} when optimization isn't turned
3341 on, even if the variables aren't referenced.
3343 GCC enables this option by default. If you want to force the compiler to
3344 check if the variable was referenced, regardless of whether or not
3345 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3347 @item -fno-function-cse
3348 @opindex fno-function-cse
3349 Do not put function addresses in registers; make each instruction that
3350 calls a constant function contain the function's address explicitly.
3352 This option results in less efficient code, but some strange hacks
3353 that alter the assembler output may be confused by the optimizations
3354 performed when this option is not used.
3358 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
3359 and @option{-fno-trapping-math}.
3361 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
3363 This option should never be turned on by any @option{-O} option since
3364 it can result in incorrect output for programs which depend on
3365 an exact implementation of IEEE or ISO rules/specifications for
3368 @item -fno-math-errno
3369 @opindex fno-math-errno
3370 Do not set ERRNO after calling math functions that are executed
3371 with a single instruction, e.g., sqrt. A program that relies on
3372 IEEE exceptions for math error handling may want to use this flag
3373 for speed while maintaining IEEE arithmetic compatibility.
3375 This option should never be turned on by any @option{-O} option since
3376 it can result in incorrect output for programs which depend on
3377 an exact implementation of IEEE or ISO rules/specifications for
3380 The default is @option{-fmath-errno}. The @option{-ffast-math} option
3381 sets @option{-fno-math-errno}.
3383 @item -funsafe-math-optimizations
3384 @opindex funsafe-math-optimizations
3385 Allow optimizations for floating-point arithmetic that (a) assume
3386 that arguments and results are valid and (b) may violate IEEE or
3387 ANSI standards. When used at link-time, it may include libraries
3388 or startup files that change the default FPU control word or other
3389 similar optimizations.
3391 This option should never be turned on by any @option{-O} option since
3392 it can result in incorrect output for programs which depend on
3393 an exact implementation of IEEE or ISO rules/specifications for
3396 The default is @option{-fno-unsafe-math-optimizations}. The
3397 @option{-ffast-math} option sets @option{-funsafe-math-optimizations}.
3399 @item -fno-trapping-math
3400 @opindex fno-trapping-math
3401 Compile code assuming that floating-point operations cannot generate
3402 user-visible traps. Setting this option may allow faster code
3403 if one relies on ``non-stop'' IEEE arithmetic, for example.
3405 This option should never be turned on by any @option{-O} option since
3406 it can result in incorrect output for programs which depend on
3407 an exact implementation of IEEE or ISO rules/specifications for
3410 The default is @option{-ftrapping-math}. The @option{-ffast-math}
3411 option sets @option{-fno-trapping-math}.
3414 @c following causes underfulls.. they don't look great, but we deal.
3416 The following options control specific optimizations. The @option{-O2}
3417 option turns on all of these optimizations except @option{-funroll-loops}
3418 and @option{-funroll-all-loops}. On most machines, the @option{-O} option
3419 turns on the @option{-fthread-jumps} and @option{-fdelayed-branch} options,
3420 but specific machines may handle it differently.
3422 You can use the following flags in the rare cases when ``fine-tuning''
3423 of optimizations to be performed is desired.
3426 @item -fstrength-reduce
3427 @opindex fstrength-reduce
3428 Perform the optimizations of loop strength reduction and
3429 elimination of iteration variables.
3431 @item -fthread-jumps
3432 @opindex fthread-jumps
3433 Perform optimizations where we check to see if a jump branches to a
3434 location where another comparison subsumed by the first is found. If
3435 so, the first branch is redirected to either the destination of the
3436 second branch or a point immediately following it, depending on whether
3437 the condition is known to be true or false.
3439 @item -fcse-follow-jumps
3440 @opindex fcse-follow-jumps
3441 In common subexpression elimination, scan through jump instructions
3442 when the target of the jump is not reached by any other path. For
3443 example, when CSE encounters an @code{if} statement with an
3444 @code{else} clause, CSE will follow the jump when the condition
3447 @item -fcse-skip-blocks
3448 @opindex fcse-skip-blocks
3449 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3450 follow jumps which conditionally skip over blocks. When CSE
3451 encounters a simple @code{if} statement with no else clause,
3452 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3453 body of the @code{if}.
3455 @item -frerun-cse-after-loop
3456 @opindex frerun-cse-after-loop
3457 Re-run common subexpression elimination after loop optimizations has been
3460 @item -frerun-loop-opt
3461 @opindex frerun-loop-opt
3462 Run the loop optimizer twice.
3466 Perform a global common subexpression elimination pass.
3467 This pass also performs global constant and copy propagation.
3471 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3472 attempt to move loads which are only killed by stores into themselves. This
3473 allows a loop containing a load/store sequence to be changed to a load outside
3474 the loop, and a copy/store within the loop.
3478 When @option{-fgcse-sm} is enabled, A store motion pass is run after global common
3479 subexpression elimination. This pass will attempt to move stores out of loops.
3480 When used in conjunction with @option{-fgcse-lm}, loops containing a load/store sequence
3481 can be changed to a load before the loop and a store after the loop.
3483 @item -fdelete-null-pointer-checks
3484 @opindex fdelete-null-pointer-checks
3485 Use global dataflow analysis to identify and eliminate useless null
3486 pointer checks. Programs which rely on NULL pointer dereferences @emph{not}
3487 halting the program may not work properly with this option. Use
3488 @option{-fno-delete-null-pointer-checks} to disable this optimizing for programs
3489 which depend on that behavior.
3491 @item -fexpensive-optimizations
3492 @opindex fexpensive-optimizations
3493 Perform a number of minor optimizations that are relatively expensive.
3495 @item -foptimize-register-move
3497 @opindex foptimize-register-move
3499 Attempt to reassign register numbers in move instructions and as
3500 operands of other simple instructions in order to maximize the amount of
3501 register tying. This is especially helpful on machines with two-operand
3502 instructions. GCC enables this optimization by default with @option{-O2}
3505 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3508 @item -fdelayed-branch
3509 @opindex fdelayed-branch
3510 If supported for the target machine, attempt to reorder instructions
3511 to exploit instruction slots available after delayed branch
3514 @item -fschedule-insns
3515 @opindex fschedule-insns
3516 If supported for the target machine, attempt to reorder instructions to
3517 eliminate execution stalls due to required data being unavailable. This
3518 helps machines that have slow floating point or memory load instructions
3519 by allowing other instructions to be issued until the result of the load
3520 or floating point instruction is required.
3522 @item -fschedule-insns2
3523 @opindex fschedule-insns2
3524 Similar to @option{-fschedule-insns}, but requests an additional pass of
3525 instruction scheduling after register allocation has been done. This is
3526 especially useful on machines with a relatively small number of
3527 registers and where memory load instructions take more than one cycle.
3529 @item -ffunction-sections
3530 @itemx -fdata-sections
3531 @opindex ffunction-sections
3532 @opindex fdata-sections
3533 Place each function or data item into its own section in the output
3534 file if the target supports arbitrary sections. The name of the
3535 function or the name of the data item determines the section's name
3538 Use these options on systems where the linker can perform optimizations
3539 to improve locality of reference in the instruction space. HPPA
3540 processors running HP-UX and Sparc processors running Solaris 2 have
3541 linkers with such optimizations. Other systems using the ELF object format
3542 as well as AIX may have these optimizations in the future.
3544 Only use these options when there are significant benefits from doing
3545 so. When you specify these options, the assembler and linker will
3546 create larger object and executable files and will also be slower.
3547 You will not be able to use @code{gprof} on all systems if you
3548 specify this option and you may have problems with debugging if
3549 you specify both this option and @option{-g}.
3551 @item -fcaller-saves
3552 @opindex fcaller-saves
3553 Enable values to be allocated in registers that will be clobbered by
3554 function calls, by emitting extra instructions to save and restore the
3555 registers around such calls. Such allocation is done only when it
3556 seems to result in better code than would otherwise be produced.
3558 This option is always enabled by default on certain machines, usually
3559 those which have no call-preserved registers to use instead.
3561 For all machines, optimization level 2 and higher enables this flag by
3564 @item -funroll-loops
3565 @opindex funroll-loops
3566 Perform the optimization of loop unrolling. This is only done for loops
3567 whose number of iterations can be determined at compile time or run time.
3568 @option{-funroll-loops} implies both @option{-fstrength-reduce} and
3569 @option{-frerun-cse-after-loop}.
3571 @item -funroll-all-loops
3572 @opindex funroll-all-loops
3573 Perform the optimization of loop unrolling. This is done for all loops
3574 and usually makes programs run more slowly. @option{-funroll-all-loops}
3575 implies @option{-fstrength-reduce} as well as @option{-frerun-cse-after-loop}.
3577 @item -fmove-all-movables
3578 @opindex fmove-all-movables
3579 Forces all invariant computations in loops to be moved
3582 @item -freduce-all-givs
3583 @opindex freduce-all-givs
3584 Forces all general-induction variables in loops to be
3587 @emph{Note:} When compiling programs written in Fortran,
3588 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
3589 by default when you use the optimizer.
3591 These options may generate better or worse code; results are highly
3592 dependent on the structure of loops within the source code.
3594 These two options are intended to be removed someday, once
3595 they have helped determine the efficacy of various
3596 approaches to improving loop optimizations.
3598 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
3599 know how use of these options affects
3600 the performance of your production code.
3601 We're very interested in code that runs @emph{slower}
3602 when these options are @emph{enabled}.
3605 @itemx -fno-peephole2
3606 @opindex fno-peephole
3607 @opindex fno-peephole2
3608 Disable any machine-specific peephole optimizations. The difference
3609 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
3610 are implemented in the compiler; some targets use one, some use the
3611 other, a few use both.
3613 @item -fbranch-probabilities
3614 @opindex fbranch-probabilities
3615 After running a program compiled with @option{-fprofile-arcs}
3616 (@pxref{Debugging Options,, Options for Debugging Your Program or
3617 @command{gcc}}), you can compile it a second time using
3618 @option{-fbranch-probabilities}, to improve optimizations based on
3619 guessing the path a branch might take.
3622 With @option{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
3623 note on the first instruction of each basic block, and a
3624 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
3625 These can be used to improve optimization. Currently, they are only
3626 used in one place: in @file{reorg.c}, instead of guessing which path a
3627 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
3628 exactly determine which path is taken more often.
3631 @item -fno-guess-branch-probability
3632 @opindex fno-guess-branch-probability
3633 Sometimes gcc will opt to guess branch probabilities when none are
3634 available from either profile directed feedback (@option{-fprofile-arcs})
3635 or @samp{__builtin_expect}. In a hard real-time system, people don't
3636 want different runs of the compiler to produce code that has different
3637 behavior; minimizing non-determinism is of paramount import. This
3638 switch allows users to reduce non-determinism, possibly at the expense
3639 of inferior optimization.
3641 @item -fstrict-aliasing
3642 @opindex fstrict-aliasing
3643 Allows the compiler to assume the strictest aliasing rules applicable to
3644 the language being compiled. For C (and C++), this activates
3645 optimizations based on the type of expressions. In particular, an
3646 object of one type is assumed never to reside at the same address as an
3647 object of a different type, unless the types are almost the same. For
3648 example, an @code{unsigned int} can alias an @code{int}, but not a
3649 @code{void*} or a @code{double}. A character type may alias any other
3652 Pay special attention to code like this:
3665 The practice of reading from a different union member than the one most
3666 recently written to (called ``type-punning'') is common. Even with
3667 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
3668 is accessed through the union type. So, the code above will work as
3669 expected. However, this code might not:
3681 Every language that wishes to perform language-specific alias analysis
3682 should define a function that computes, given an @code{tree}
3683 node, an alias set for the node. Nodes in different alias sets are not
3684 allowed to alias. For an example, see the C front-end function
3685 @code{c_get_alias_set}.
3688 @item -falign-functions
3689 @itemx -falign-functions=@var{n}
3690 @opindex falign-functions
3691 Align the start of functions to the next power-of-two greater than
3692 @var{n}, skipping up to @var{n} bytes. For instance,
3693 @option{-falign-functions=32} aligns functions to the next 32-byte
3694 boundary, but @option{-falign-functions=24} would align to the next
3695 32-byte boundary only if this can be done by skipping 23 bytes or less.
3697 @option{-fno-align-functions} and @option{-falign-functions=1} are
3698 equivalent and mean that functions will not be aligned.
3700 Some assemblers only support this flag when @var{n} is a power of two;
3701 in that case, it is rounded up.
3703 If @var{n} is not specified, use a machine-dependent default.
3705 @item -falign-labels
3706 @itemx -falign-labels=@var{n}
3707 @opindex falign-labels
3708 Align all branch targets to a power-of-two boundary, skipping up to
3709 @var{n} bytes like @option{-falign-functions}. This option can easily
3710 make code slower, because it must insert dummy operations for when the
3711 branch target is reached in the usual flow of the code.
3713 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
3714 are greater than this value, then their values are used instead.
3716 If @var{n} is not specified, use a machine-dependent default which is
3717 very likely to be @samp{1}, meaning no alignment.
3720 @itemx -falign-loops=@var{n}
3721 @opindex falign-loops
3722 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
3723 like @option{-falign-functions}. The hope is that the loop will be
3724 executed many times, which will make up for any execution of the dummy
3727 If @var{n} is not specified, use a machine-dependent default.
3730 @itemx -falign-jumps=@var{n}
3731 @opindex falign-jumps
3732 Align branch targets to a power-of-two boundary, for branch targets
3733 where the targets can only be reached by jumping, skipping up to @var{n}
3734 bytes like @option{-falign-functions}. In this case, no dummy operations
3737 If @var{n} is not specified, use a machine-dependent default.
3741 Perform optimizations in static single assignment form. Each function's
3742 flow graph is translated into SSA form, optimizations are performed, and
3743 the flow graph is translated back from SSA form. Users should not
3744 specify this option, since it is not yet ready for production use.
3748 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
3749 Like @option{-fssa}, this is an experimental feature.
3751 @item -fsingle-precision-constant
3752 @opindex fsingle-precision-constant
3753 Treat floating point constant as single precision constant instead of
3754 implicitly converting it to double precision constant.
3756 @item -frename-registers
3757 @opindex frename-registers
3758 Attempt to avoid false dependencies in scheduled code by making use
3759 of registers left over after register allocation. This optimization
3760 will most benefit processors with lots of registers. It can, however,
3761 make debugging impossible, since variables will no longer stay in
3762 a ``home register''.
3764 @item --param @var{name}=@var{value}
3766 In some places, GCC uses various constants to control the amount of
3767 optimization that is done. For example, GCC will not inline functions
3768 that contain more that a certain number of instructions. You can
3769 control some of these constants on the command-line using the
3770 @option{--param} option.
3772 In each case, the @var{value} is a integer. The allowable choices for
3773 @var{name} are given in the following table:
3776 @item max-delay-slot-insn-search
3777 The maximum number of instructions to consider when looking for an
3778 instruction to fill a delay slot. If more than this arbitrary number of
3779 instructions is searched, the time savings from filling the delay slot
3780 will be minimal so stop searching. Increasing values mean more
3781 aggressive optimization, making the compile time increase with probably
3782 small improvement in executable run time.
3784 @item max-delay-slot-live-search
3785 When trying to fill delay slots, the maximum number of instructions to
3786 consider when searching for a block with valid live register
3787 information. Increasing this arbitrarily chosen value means more
3788 aggressive optimization, increasing the compile time. This parameter
3789 should be removed when the delay slot code is rewritten to maintain the
3792 @item max-gcse-memory
3793 The approximate maximum amount of memory that will be allocated in
3794 order to perform the global common subexpression elimination
3795 optimization. If more memory than specified is required, the
3796 optimization will not be done.
3798 @item max-gcse-passes
3799 The maximum number of passes of GCSE to run.
3801 @item max-inline-insns
3802 If an function contains more than this many instructions, it
3803 will not be inlined. This option is precisely equivalent to
3804 @option{-finline-limit}.
3809 @node Preprocessor Options
3810 @section Options Controlling the Preprocessor
3811 @cindex preprocessor options
3812 @cindex options, preprocessor
3814 These options control the C preprocessor, which is run on each C source
3815 file before actual compilation.
3817 If you use the @option{-E} option, nothing is done except preprocessing.
3818 Some of these options make sense only together with @option{-E} because
3819 they cause the preprocessor output to be unsuitable for actual
3823 @item -include @var{file}
3825 Process @var{file} as input before processing the regular input file.
3826 In effect, the contents of @var{file} are compiled first. Any @option{-D}
3827 and @option{-U} options on the command line are always processed before
3828 @option{-include @var{file}}, regardless of the order in which they are
3829 written. All the @option{-include} and @option{-imacros} options are
3830 processed in the order in which they are written.
3832 @item -imacros @var{file}
3834 Process @var{file} as input, discarding the resulting output, before
3835 processing the regular input file. Because the output generated from
3836 @var{file} is discarded, the only effect of @option{-imacros @var{file}}
3837 is to make the macros defined in @var{file} available for use in the
3838 main input. All the @option{-include} and @option{-imacros} options are
3839 processed in the order in which they are written.
3841 @item -idirafter @var{dir}
3843 @cindex second include path
3844 Add the directory @var{dir} to the second include path. The directories
3845 on the second include path are searched when a header file is not found
3846 in any of the directories in the main include path (the one that
3847 @option{-I} adds to).
3849 @item -iprefix @var{prefix}
3851 Specify @var{prefix} as the prefix for subsequent @option{-iwithprefix}
3854 @item -iwithprefix @var{dir}
3855 @opindex iwithprefix
3856 Add a directory to the second include path. The directory's name is
3857 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
3858 specified previously with @option{-iprefix}. If you have not specified a
3859 prefix yet, the directory containing the installed passes of the
3860 compiler is used as the default.
3862 @item -iwithprefixbefore @var{dir}
3863 @opindex iwithprefixbefore
3864 Add a directory to the main include path. The directory's name is made
3865 by concatenating @var{prefix} and @var{dir}, as in the case of
3866 @option{-iwithprefix}.
3868 @item -isystem @var{dir}
3870 Add a directory to the beginning of the second include path, marking it
3871 as a system directory, so that it gets the same special treatment as
3872 is applied to the standard system directories.
3876 Do not search the standard system directories for header files. Only
3877 the directories you have specified with @option{-I} options (and the
3878 current directory, if appropriate) are searched. @xref{Directory
3879 Options}, for information on @option{-I}.
3881 By using both @option{-nostdinc} and @option{-I-}, you can limit the include-file
3882 search path to only those directories you specify explicitly.
3886 When searching for a header file in a directory, remap file names if a
3887 file named @file{header.gcc} exists in that directory. This can be used
3888 to work around limitations of file systems with file name restrictions.
3889 The @file{header.gcc} file should contain a series of lines with two
3890 tokens on each line: the first token is the name to map, and the second
3891 token is the actual name to use.
3895 Do not predefine any nonstandard macros. (Including architecture flags).
3899 Run only the C preprocessor. Preprocess all the C source files
3900 specified and output the results to standard output or to the
3901 specified output file.
3905 Tell the preprocessor not to discard comments. Used with the
3910 Tell the preprocessor not to generate @samp{#line} directives.
3911 Used with the @option{-E} option.
3914 @cindex dependencies, make
3917 Instead of outputting the result of preprocessing, output a rule
3918 suitable for @code{make} describing the dependencies of the main source
3919 file. The preprocessor outputs one @code{make} rule containing the
3920 object file name for that source file, a colon, and the names of all the
3921 included files. Unless overridden explicitly, the object file name
3922 consists of the basename of the source file with any suffix replaced with
3923 object file suffix. If there are many included files then the
3924 rule is split into several lines using @samp{\}-newline.
3926 @option{-M} implies @option{-E}.
3930 Like @option{-M}, but mention only the files included with @samp{#include
3931 "@var{file}"}. System header files included with @samp{#include
3932 <@var{file}>} are omitted.
3936 Like @option{-M} but the dependency information is written to a file
3937 rather than stdout. @code{gcc} will use the same file name and
3938 directory as the object file, but with the suffix @file{.d} instead.
3940 This is in addition to compiling the main file as specified---@option{-MD}
3941 does not inhibit ordinary compilation the way @option{-M} does,
3942 unless you also specify @option{-MG}.
3944 With Mach, you can use the utility @code{md} to merge multiple
3945 dependency files into a single dependency file suitable for using with
3946 the @samp{make} command.
3950 Like @option{-MD} except mention only user header files, not system
3953 @item -MF @var{file}
3955 When used with @option{-M} or @option{-MM}, specifies a file to write the
3956 dependencies to. This allows the preprocessor to write the preprocessed
3957 file to stdout normally. If no @option{-MF} switch is given, CPP sends
3958 the rules to stdout and suppresses normal preprocessed output.
3960 Another way to specify output of a @code{make} rule is by setting
3961 the environment variable @env{DEPENDENCIES_OUTPUT} (@pxref{Environment
3966 When used with @option{-M} or @option{-MM}, @option{-MG} says to treat missing
3967 header files as generated files and assume they live in the same
3968 directory as the source file. It suppresses preprocessed output, as a
3969 missing header file is ordinarily an error.
3971 This feature is used in automatic updating of makefiles.
3975 This option instructs CPP to add a phony target for each dependency
3976 other than the main file, causing each to depend on nothing. These
3977 dummy rules work around errors @code{make} gives if you remove header
3978 files without updating the @code{Makefile} to match.
3980 This is typical output:-
3983 /tmp/test.o: /tmp/test.c /tmp/test.h
3988 @item -MQ @var{target}
3989 @item -MT @var{target}
3992 By default CPP uses the main file name, including any path, and appends
3993 the object suffix, normally ``.o'', to it to obtain the name of the
3994 target for dependency generation. With @option{-MT} you can specify a
3995 target yourself, overriding the default one.
3997 If you want multiple targets, you can specify them as a single argument
3998 to @option{-MT}, or use multiple @option{-MT} options.
4000 The targets you specify are output in the order they appear on the
4001 command line. @option{-MQ} is identical to @option{-MT}, except that the
4002 target name is quoted for Make, but with @option{-MT} it isn't. For
4003 example, @option{-MT '$(objpfx)foo.o'} gives
4006 $(objpfx)foo.o: /tmp/foo.c
4009 but @option{-MQ '$(objpfx)foo.o'} gives
4012 $$(objpfx)foo.o: /tmp/foo.c
4015 The default target is automatically quoted, as if it were given with
4020 Print the name of each header file used, in addition to other normal
4023 @item -A@var{question}(@var{answer})
4025 Assert the answer @var{answer} for @var{question}, in case it is tested
4026 with a preprocessing conditional such as @samp{#if
4027 #@var{question}(@var{answer})}. @option{-A-} disables the standard
4028 assertions that normally describe the target machine.
4032 Define macro @var{macro} with the string @samp{1} as its definition.
4034 @item -D@var{macro}=@var{defn}
4035 Define macro @var{macro} as @var{defn}. All instances of @option{-D} on
4036 the command line are processed before any @option{-U} options.
4038 Any @option{-D} and @option{-U} options on the command line are processed in
4039 order, and always before @option{-imacros @var{file}}, regardless of the
4040 order in which they are written.
4044 Undefine macro @var{macro}. @option{-U} options are evaluated after all
4045 @option{-D} options, but before any @option{-include} and @option{-imacros}
4048 Any @option{-D} and @option{-U} options on the command line are processed in
4049 order, and always before @option{-imacros @var{file}}, regardless of the
4050 order in which they are written.
4054 Tell the preprocessor to output only a list of the macro definitions
4055 that are in effect at the end of preprocessing. Used with the @option{-E}
4060 Tell the preprocessing to pass all macro definitions into the output, in
4061 their proper sequence in the rest of the output.
4065 Like @option{-dD} except that the macro arguments and contents are omitted.
4066 Only @samp{#define @var{name}} is included in the output.
4070 Output @samp{#include} directives in addition to the result of
4073 @item -fpreprocessed
4074 @opindex fpreprocessed
4075 Indicate to the preprocessor that the input file has already been
4076 preprocessed. This suppresses things like macro expansion, trigraph
4077 conversion, escaped newline splicing, and processing of most directives.
4078 The preprocessor still recognizes and removes comments, so that you can
4079 pass a file preprocessed with @option{-C} to the compiler without
4080 problems. In this mode the integrated preprocessor is little more than
4081 a tokenizer for the front ends.
4083 @option{-fpreprocessed} is implicit if the input file has one of the
4084 extensions @samp{i}, @samp{ii} or @samp{mi}. These are the extensions
4085 that GCC uses for preprocessed files created by @option{-save-temps}.
4089 Process ISO standard trigraph sequences. These are three-character
4090 sequences, all starting with @samp{??}, that are defined by ISO C to
4091 stand for single characters. For example, @samp{??/} stands for
4092 @samp{\}, so @samp{'??/n'} is a character constant for a newline. By
4093 default, GCC ignores trigraphs, but in standard-conforming modes it
4094 converts them. See the @option{-std} and @option{-ansi} options.
4096 The nine trigraph sequences are
4099 @expansion{} @samp{[}
4102 @expansion{} @samp{]}
4105 @expansion{} @samp{@{}
4108 @expansion{} @samp{@}}
4111 @expansion{} @samp{#}
4114 @expansion{} @samp{\}
4117 @expansion{} @samp{^}
4120 @expansion{} @samp{|}
4123 @expansion{} @samp{~}
4127 Trigraph support is not popular, so many compilers do not implement it
4128 properly. Portable code should not rely on trigraphs being either
4129 converted or ignored.
4131 @item -Wp,@var{option}
4133 Pass @var{option} as an option to the preprocessor. If @var{option}
4134 contains commas, it is split into multiple options at the commas.
4137 @node Assembler Options
4138 @section Passing Options to the Assembler
4140 @c prevent bad page break with this line
4141 You can pass options to the assembler.
4144 @item -Wa,@var{option}
4146 Pass @var{option} as an option to the assembler. If @var{option}
4147 contains commas, it is split into multiple options at the commas.
4151 @section Options for Linking
4152 @cindex link options
4153 @cindex options, linking
4155 These options come into play when the compiler links object files into
4156 an executable output file. They are meaningless if the compiler is
4157 not doing a link step.
4161 @item @var{object-file-name}
4162 A file name that does not end in a special recognized suffix is
4163 considered to name an object file or library. (Object files are
4164 distinguished from libraries by the linker according to the file
4165 contents.) If linking is done, these object files are used as input
4174 If any of these options is used, then the linker is not run, and
4175 object file names should not be used as arguments. @xref{Overall
4179 @item -l@var{library}
4180 @itemx -l @var{library}
4182 Search the library named @var{library} when linking. (The second
4183 alternative with the library as a separate argument is only for
4184 POSIX compliance and is not recommended.)
4186 It makes a difference where in the command you write this option; the
4187 linker searches and processes libraries and object files in the order they
4188 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4189 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4190 to functions in @samp{z}, those functions may not be loaded.
4192 The linker searches a standard list of directories for the library,
4193 which is actually a file named @file{lib@var{library}.a}. The linker
4194 then uses this file as if it had been specified precisely by name.
4196 The directories searched include several standard system directories
4197 plus any that you specify with @option{-L}.
4199 Normally the files found this way are library files---archive files
4200 whose members are object files. The linker handles an archive file by
4201 scanning through it for members which define symbols that have so far
4202 been referenced but not defined. But if the file that is found is an
4203 ordinary object file, it is linked in the usual fashion. The only
4204 difference between using an @option{-l} option and specifying a file name
4205 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4206 and searches several directories.
4210 You need this special case of the @option{-l} option in order to
4211 link an Objective-C program.
4214 @opindex nostartfiles
4215 Do not use the standard system startup files when linking.
4216 The standard system libraries are used normally, unless @option{-nostdlib}
4217 or @option{-nodefaultlibs} is used.
4219 @item -nodefaultlibs
4220 @opindex nodefaultlibs
4221 Do not use the standard system libraries when linking.
4222 Only the libraries you specify will be passed to the linker.
4223 The standard startup files are used normally, unless @option{-nostartfiles}
4224 is used. The compiler may generate calls to memcmp, memset, and memcpy
4225 for System V (and ISO C) environments or to bcopy and bzero for
4226 BSD environments. These entries are usually resolved by entries in
4227 libc. These entry points should be supplied through some other
4228 mechanism when this option is specified.
4232 Do not use the standard system startup files or libraries when linking.
4233 No startup files and only the libraries you specify will be passed to
4234 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4235 for System V (and ISO C) environments or to bcopy and bzero for
4236 BSD environments. These entries are usually resolved by entries in
4237 libc. These entry points should be supplied through some other
4238 mechanism when this option is specified.
4240 @cindex @option{-lgcc}, use with @option{-nostdlib}
4241 @cindex @option{-nostdlib} and unresolved references
4242 @cindex unresolved references and @option{-nostdlib}
4243 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
4244 @cindex @option{-nodefaultlibs} and unresolved references
4245 @cindex unresolved references and @option{-nodefaultlibs}
4246 One of the standard libraries bypassed by @option{-nostdlib} and
4247 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4248 that GCC uses to overcome shortcomings of particular machines, or special
4249 needs for some languages.
4251 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
4255 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
4256 for more discussion of @file{libgcc.a}.)
4258 In most cases, you need @file{libgcc.a} even when you want to avoid
4259 other standard libraries. In other words, when you specify @option{-nostdlib}
4260 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
4261 This ensures that you have no unresolved references to internal GCC
4262 library subroutines. (For example, @samp{__main}, used to ensure C++
4263 constructors will be called; @pxref{Collect2,,@command{collect2}}.)
4267 Remove all symbol table and relocation information from the executable.
4271 On systems that support dynamic linking, this prevents linking with the shared
4272 libraries. On other systems, this option has no effect.
4276 Produce a shared object which can then be linked with other objects to
4277 form an executable. Not all systems support this option. For predictable
4278 results, you must also specify the same set of options that were used to
4279 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
4280 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4281 needs to build supplementary stub code for constructors to work. On
4282 multi-libbed systems, @samp{gcc -shared} must select the correct support
4283 libraries to link against. Failing to supply the correct flags may lead
4284 to subtle defects. Supplying them in cases where they are not necessary
4287 @item -shared-libgcc
4288 @itemx -static-libgcc
4289 @opindex shared-libgcc
4290 @opindex static-libgcc
4291 On systems that provide @file{libgcc} as a shared library, these options
4292 force the use of either the shared or static version respectively.
4293 If no shared version of @file{libgcc} was built when the compiler was
4294 configured, these options have no effect.
4296 There are several situations in which an application should use the
4297 shared @file{libgcc} instead of the static version. The most common
4298 of these is when the application wishes to throw and catch exceptions
4299 across different shared libraries. In that case, each of the libraries
4300 as well as the application itself should use the shared @file{libgcc}.
4302 Therefore, whenever you specify the @option{-shared} option, the GCC
4303 driver automatically adds @option{-shared-libgcc}, unless you explicitly
4304 specify @option{-static-libgcc}. The G++ driver automatically adds
4305 @option{-shared-libgcc} when you build a main executable as well because
4306 for C++ programs that is typically the right thing to do.
4307 (Exception-handling will not work reliably otherwise.)
4309 However, when linking a main executable written in C, you must
4310 explicitly say @option{-shared-libgcc} if you want to use the shared
4315 Bind references to global symbols when building a shared object. Warn
4316 about any unresolved references (unless overridden by the link editor
4317 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
4320 @item -Xlinker @var{option}
4322 Pass @var{option} as an option to the linker. You can use this to
4323 supply system-specific linker options which GCC does not know how to
4326 If you want to pass an option that takes an argument, you must use
4327 @option{-Xlinker} twice, once for the option and once for the argument.
4328 For example, to pass @option{-assert definitions}, you must write
4329 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
4330 @option{-Xlinker "-assert definitions"}, because this passes the entire
4331 string as a single argument, which is not what the linker expects.
4333 @item -Wl,@var{option}
4335 Pass @var{option} as an option to the linker. If @var{option} contains
4336 commas, it is split into multiple options at the commas.
4338 @item -u @var{symbol}
4340 Pretend the symbol @var{symbol} is undefined, to force linking of
4341 library modules to define it. You can use @option{-u} multiple times with
4342 different symbols to force loading of additional library modules.
4345 @node Directory Options
4346 @section Options for Directory Search
4347 @cindex directory options
4348 @cindex options, directory search
4351 These options specify directories to search for header files, for
4352 libraries and for parts of the compiler:
4357 Add the directory @var{dir} to the head of the list of directories to be
4358 searched for header files. This can be used to override a system header
4359 file, substituting your own version, since these directories are
4360 searched before the system header file directories. However, you should
4361 not use this option to add directories that contain vendor-supplied
4362 system header files (use @option{-isystem} for that). If you use more than
4363 one @option{-I} option, the directories are scanned in left-to-right
4364 order; the standard system directories come after.
4366 If a standard system include directory, or a directory specified with
4367 @option{-isystem}, is also specified with @option{-I}, it will be
4368 searched only in the position requested by @option{-I}. Also, it will
4369 not be considered a system include directory. If that directory really
4370 does contain system headers, there is a good chance that they will
4371 break. For instance, if GCC's installation procedure edited the headers
4372 in @file{/usr/include} to fix bugs, @samp{-I/usr/include} will cause the
4373 original, buggy headers to be found instead of the corrected ones. GCC
4374 will issue a warning when a system include directory is hidden in this
4379 Any directories you specify with @option{-I} options before the @option{-I-}
4380 option are searched only for the case of @samp{#include "@var{file}"};
4381 they are not searched for @samp{#include <@var{file}>}.
4383 If additional directories are specified with @option{-I} options after
4384 the @option{-I-}, these directories are searched for all @samp{#include}
4385 directives. (Ordinarily @emph{all} @option{-I} directories are used
4388 In addition, the @option{-I-} option inhibits the use of the current
4389 directory (where the current input file came from) as the first search
4390 directory for @samp{#include "@var{file}"}. There is no way to
4391 override this effect of @option{-I-}. With @option{-I.} you can specify
4392 searching the directory which was current when the compiler was
4393 invoked. That is not exactly the same as what the preprocessor does
4394 by default, but it is often satisfactory.
4396 @option{-I-} does not inhibit the use of the standard system directories
4397 for header files. Thus, @option{-I-} and @option{-nostdinc} are
4402 Add directory @var{dir} to the list of directories to be searched
4405 @item -B@var{prefix}
4407 This option specifies where to find the executables, libraries,
4408 include files, and data files of the compiler itself.
4410 The compiler driver program runs one or more of the subprograms
4411 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
4412 @var{prefix} as a prefix for each program it tries to run, both with and
4413 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
4415 For each subprogram to be run, the compiler driver first tries the
4416 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
4417 was not specified, the driver tries two standard prefixes, which are
4418 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
4419 those results in a file name that is found, the unmodified program
4420 name is searched for using the directories specified in your
4421 @env{PATH} environment variable.
4423 The compiler will check to see if the path provided by the @option{-B}
4424 refers to a directory, and if necessary it will add a directory
4425 separator character at the end of the path.
4427 @option{-B} prefixes that effectively specify directory names also apply
4428 to libraries in the linker, because the compiler translates these
4429 options into @option{-L} options for the linker. They also apply to
4430 includes files in the preprocessor, because the compiler translates these
4431 options into @option{-isystem} options for the preprocessor. In this case,
4432 the compiler appends @samp{include} to the prefix.
4434 The run-time support file @file{libgcc.a} can also be searched for using
4435 the @option{-B} prefix, if needed. If it is not found there, the two
4436 standard prefixes above are tried, and that is all. The file is left
4437 out of the link if it is not found by those means.
4439 Another way to specify a prefix much like the @option{-B} prefix is to use
4440 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
4443 As a special kludge, if the path provided by @option{-B} is
4444 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
4445 9, then it will be replaced by @file{[dir/]include}. This is to help
4446 with boot-strapping the compiler.
4448 @item -specs=@var{file}
4450 Process @var{file} after the compiler reads in the standard @file{specs}
4451 file, in order to override the defaults that the @file{gcc} driver
4452 program uses when determining what switches to pass to @file{cc1},
4453 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
4454 @option{-specs=@var{file}} can be specified on the command line, and they
4455 are processed in order, from left to right.
4461 @section Specifying subprocesses and the switches to pass to them
4463 @command{gcc} is a driver program. It performs its job by invoking a
4464 sequence of other programs to do the work of compiling, assembling and
4465 linking. GCC interprets its command-line parameters and uses these to
4466 deduce which programs it should invoke, and which command-line options
4467 it ought to place on their command lines. This behaviour is controlled
4468 by @dfn{spec strings}. In most cases there is one spec string for each
4469 program that GCC can invoke, but a few programs have multiple spec
4470 strings to control their behaviour. The spec strings built into GCC can
4471 be overridden by using the @option{-specs=} command-line switch to specify
4474 @dfn{Spec files} are plaintext files that are used to construct spec
4475 strings. They consist of a sequence of directives separated by blank
4476 lines. The type of directive is determined by the first non-whitespace
4477 character on the line and it can be one of the following:
4480 @item %@var{command}
4481 Issues a @var{command} to the spec file processor. The commands that can
4485 @item %include <@var{file}>
4487 Search for @var{file} and insert its text at the current point in the
4490 @item %include_noerr <@var{file}>
4491 @cindex %include_noerr
4492 Just like @samp{%include}, but do not generate an error message if the include
4493 file cannot be found.
4495 @item %rename @var{old_name} @var{new_name}
4497 Rename the spec string @var{old_name} to @var{new_name}.
4501 @item *[@var{spec_name}]:
4502 This tells the compiler to create, override or delete the named spec
4503 string. All lines after this directive up to the next directive or
4504 blank line are considered to be the text for the spec string. If this
4505 results in an empty string then the spec will be deleted. (Or, if the
4506 spec did not exist, then nothing will happened.) Otherwise, if the spec
4507 does not currently exist a new spec will be created. If the spec does
4508 exist then its contents will be overridden by the text of this
4509 directive, unless the first character of that text is the @samp{+}
4510 character, in which case the text will be appended to the spec.
4512 @item [@var{suffix}]:
4513 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
4514 and up to the next directive or blank line are considered to make up the
4515 spec string for the indicated suffix. When the compiler encounters an
4516 input file with the named suffix, it will processes the spec string in
4517 order to work out how to compile that file. For example:
4524 This says that any input file whose name ends in @samp{.ZZ} should be
4525 passed to the program @samp{z-compile}, which should be invoked with the
4526 command-line switch @option{-input} and with the result of performing the
4527 @samp{%i} substitution. (See below.)
4529 As an alternative to providing a spec string, the text that follows a
4530 suffix directive can be one of the following:
4533 @item @@@var{language}
4534 This says that the suffix is an alias for a known @var{language}. This is
4535 similar to using the @option{-x} command-line switch to GCC to specify a
4536 language explicitly. For example:
4543 Says that .ZZ files are, in fact, C++ source files.
4546 This causes an error messages saying:
4549 @var{name} compiler not installed on this system.
4553 GCC already has an extensive list of suffixes built into it.
4554 This directive will add an entry to the end of the list of suffixes, but
4555 since the list is searched from the end backwards, it is effectively
4556 possible to override earlier entries using this technique.
4560 GCC has the following spec strings built into it. Spec files can
4561 override these strings or create their own. Note that individual
4562 targets can also add their own spec strings to this list.
4565 asm Options to pass to the assembler
4566 asm_final Options to pass to the assembler post-processor
4567 cpp Options to pass to the C preprocessor
4568 cc1 Options to pass to the C compiler
4569 cc1plus Options to pass to the C++ compiler
4570 endfile Object files to include at the end of the link
4571 link Options to pass to the linker
4572 lib Libraries to include on the command line to the linker
4573 libgcc Decides which GCC support library to pass to the linker
4574 linker Sets the name of the linker
4575 predefines Defines to be passed to the C preprocessor
4576 signed_char Defines to pass to CPP to say whether @code{char} is signed
4578 startfile Object files to include at the start of the link
4581 Here is a small example of a spec file:
4587 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
4590 This example renames the spec called @samp{lib} to @samp{old_lib} and
4591 then overrides the previous definition of @samp{lib} with a new one.
4592 The new definition adds in some extra command-line options before
4593 including the text of the old definition.
4595 @dfn{Spec strings} are a list of command-line options to be passed to their
4596 corresponding program. In addition, the spec strings can contain
4597 @samp{%}-prefixed sequences to substitute variable text or to
4598 conditionally insert text into the command line. Using these constructs
4599 it is possible to generate quite complex command lines.
4601 Here is a table of all defined @samp{%}-sequences for spec
4602 strings. Note that spaces are not generated automatically around the
4603 results of expanding these sequences. Therefore you can concatenate them
4604 together or combine them with constant text in a single argument.
4608 Substitute one @samp{%} into the program name or argument.
4611 Substitute the name of the input file being processed.
4614 Substitute the basename of the input file being processed.
4615 This is the substring up to (and not including) the last period
4616 and not including the directory.
4619 This is the same as @samp{%b}, but include the file suffix (text after
4623 Marks the argument containing or following the @samp{%d} as a
4624 temporary file name, so that that file will be deleted if GCC exits
4625 successfully. Unlike @samp{%g}, this contributes no text to the
4628 @item %g@var{suffix}
4629 Substitute a file name that has suffix @var{suffix} and is chosen
4630 once per compilation, and mark the argument in the same way as
4631 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
4632 name is now chosen in a way that is hard to predict even when previously
4633 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
4634 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
4635 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
4636 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
4637 was simply substituted with a file name chosen once per compilation,
4638 without regard to any appended suffix (which was therefore treated
4639 just like ordinary text), making such attacks more likely to succeed.
4641 @item %u@var{suffix}
4642 Like @samp{%g}, but generates a new temporary file name even if
4643 @samp{%u@var{suffix}} was already seen.
4645 @item %U@var{suffix}
4646 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
4647 new one if there is no such last file name. In the absence of any
4648 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
4649 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
4650 would involve the generation of two distinct file names, one
4651 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
4652 simply substituted with a file name chosen for the previous @samp{%u},
4653 without regard to any appended suffix.
4655 @item %j@var{SUFFIX}
4656 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
4657 writable, and if save-temps is off; otherwise, substitute the name
4658 of a temporary file, just like @samp{%u}. This temporary file is not
4659 meant for communication between processes, but rather as a junk
4662 @item %.@var{SUFFIX}
4663 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
4664 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
4665 terminated by the next space or %.
4668 Marks the argument containing or following the @samp{%w} as the
4669 designated output file of this compilation. This puts the argument
4670 into the sequence of arguments that @samp{%o} will substitute later.
4673 Substitutes the names of all the output files, with spaces
4674 automatically placed around them. You should write spaces
4675 around the @samp{%o} as well or the results are undefined.
4676 @samp{%o} is for use in the specs for running the linker.
4677 Input files whose names have no recognized suffix are not compiled
4678 at all, but they are included among the output files, so they will
4682 Substitutes the suffix for object files. Note that this is
4683 handled specially when it immediately follows @samp{%g, %u, or %U},
4684 because of the need for those to form complete file names. The
4685 handling is such that @samp{%O} is treated exactly as if it had already
4686 been substituted, except that @samp{%g, %u, and %U} do not currently
4687 support additional @var{suffix} characters following @samp{%O} as they would
4688 following, for example, @samp{.o}.
4691 Substitutes the standard macro predefinitions for the
4692 current target machine. Use this when running @code{cpp}.
4695 Like @samp{%p}, but puts @samp{__} before and after the name of each
4696 predefined macro, except for macros that start with @samp{__} or with
4697 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
4701 Substitute a @option{-iprefix} option made from @env{GCC_EXEC_PREFIX}.
4704 Current argument is the name of a library or startup file of some sort.
4705 Search for that file in a standard list of directories and substitute
4706 the full name found.
4709 Print @var{str} as an error message. @var{str} is terminated by a newline.
4710 Use this when inconsistent options are detected.
4713 Output @samp{-} if the input for the current command is coming from a pipe.
4716 Substitute the contents of spec string @var{name} at this point.
4719 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
4721 @item %x@{@var{option}@}
4722 Accumulate an option for @samp{%X}.
4725 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
4729 Output the accumulated assembler options specified by @option{-Wa}.
4732 Output the accumulated preprocessor options specified by @option{-Wp}.
4735 Substitute the major version number of GCC@.
4736 (For version 2.9.5, this is 2.)
4739 Substitute the minor version number of GCC@.
4740 (For version 2.9.5, this is 9.)
4743 Substitute the patch level number of GCC@.
4744 (For version 2.9.5, this is 5.)
4747 Process the @code{asm} spec. This is used to compute the
4748 switches to be passed to the assembler.
4751 Process the @code{asm_final} spec. This is a spec string for
4752 passing switches to an assembler post-processor, if such a program is
4756 Process the @code{link} spec. This is the spec for computing the
4757 command line passed to the linker. Typically it will make use of the
4758 @samp{%L %G %S %D and %E} sequences.
4761 Dump out a @option{-L} option for each directory that GCC believes might
4762 contain startup files. If the target supports multilibs then the
4763 current multilib directory will be prepended to each of these paths.
4766 Output the multilib directory with directory separators replaced with
4767 @samp{_}. If multilib directories are not set, or the multilib directory is
4768 @file{.} then this option emits nothing.
4771 Process the @code{lib} spec. This is a spec string for deciding which
4772 libraries should be included on the command line to the linker.
4775 Process the @code{libgcc} spec. This is a spec string for deciding
4776 which GCC support library should be included on the command line to the linker.
4779 Process the @code{startfile} spec. This is a spec for deciding which
4780 object files should be the first ones passed to the linker. Typically
4781 this might be a file named @file{crt0.o}.
4784 Process the @code{endfile} spec. This is a spec string that specifies
4785 the last object files that will be passed to the linker.
4788 Process the @code{cpp} spec. This is used to construct the arguments
4789 to be passed to the C preprocessor.
4792 Process the @code{signed_char} spec. This is intended to be used
4793 to tell cpp whether a char is signed. It typically has the definition:
4795 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
4799 Process the @code{cc1} spec. This is used to construct the options to be
4800 passed to the actual C compiler (@samp{cc1}).
4803 Process the @code{cc1plus} spec. This is used to construct the options to be
4804 passed to the actual C++ compiler (@samp{cc1plus}).
4807 Substitute the variable part of a matched option. See below.
4808 Note that each comma in the substituted string is replaced by
4812 Substitutes the @code{-S} switch, if that switch was given to GCC@.
4813 If that switch was not specified, this substitutes nothing. Note that
4814 the leading dash is omitted when specifying this option, and it is
4815 automatically inserted if the substitution is performed. Thus the spec
4816 string @samp{%@{foo@}} would match the command-line option @option{-foo}
4817 and would output the command line option @option{-foo}.
4819 @item %W@{@code{S}@}
4820 Like %@{@code{S}@} but mark last argument supplied within as a file to be
4823 @item %@{@code{S}*@}
4824 Substitutes all the switches specified to GCC whose names start
4825 with @code{-S}, but which also take an argument. This is used for
4826 switches like @option{-o}, @option{-D}, @option{-I}, etc.
4827 GCC considers @option{-o foo} as being
4828 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
4829 text, including the space. Thus two arguments would be generated.
4831 @item %@{^@code{S}*@}
4832 Like %@{@code{S}*@}, but don't put a blank between a switch and its
4833 argument. Thus %@{^o*@} would only generate one argument, not two.
4835 @item %@{@code{S}*&@code{T}*@}
4836 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
4837 (the order of @code{S} and @code{T} in the spec is not significant).
4838 There can be any number of ampersand-separated variables; for each the
4839 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
4841 @item %@{<@code{S}@}
4842 Remove all occurrences of @code{-S} from the command line. Note---this
4843 command is position dependent. @samp{%} commands in the spec string
4844 before this option will see @code{-S}, @samp{%} commands in the spec
4845 string after this option will not.
4847 @item %@{@code{S}*:@code{X}@}
4848 Substitutes @code{X} if one or more switches whose names start with
4849 @code{-S} are specified to GCC@. Note that the tail part of the
4850 @code{-S} option (i.e.@: the part matched by the @samp{*}) will be substituted
4851 for each occurrence of @samp{%*} within @code{X}.
4853 @item %@{@code{S}:@code{X}@}
4854 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC@.
4856 @item %@{!@code{S}:@code{X}@}
4857 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC@.
4859 @item %@{|@code{S}:@code{X}@}
4860 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
4862 @item %@{|!@code{S}:@code{X}@}
4863 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
4865 @item %@{.@code{S}:@code{X}@}
4866 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
4868 @item %@{!.@code{S}:@code{X}@}
4869 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
4871 @item %@{@code{S}|@code{P}:@code{X}@}
4872 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@. This may be
4873 combined with @samp{!} and @samp{.} sequences as well, although they
4874 have a stronger binding than the @samp{|}. For example a spec string
4878 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
4881 will output the following command-line options from the following input
4882 command-line options:
4887 -d fred.c -foo -baz -boggle
4888 -d jim.d -bar -baz -boggle
4893 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
4894 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
4895 or spaces, or even newlines. They are processed as usual, as described
4898 The @option{-O}, @option{-f}, @option{-m}, and @option{-W}
4899 switches are handled specifically in these
4900 constructs. If another value of @option{-O} or the negated form of a @option{-f}, @option{-m}, or
4901 @option{-W} switch is found later in the command line, the earlier switch
4902 value is ignored, except with @{@code{S}*@} where @code{S} is just one
4903 letter, which passes all matching options.
4905 The character @samp{|} at the beginning of the predicate text is used to indicate
4906 that a command should be piped to the following command, but only if @option{-pipe}
4909 It is built into GCC which switches take arguments and which do not.
4910 (You might think it would be useful to generalize this to allow each
4911 compiler's spec to say which switches take arguments. But this cannot
4912 be done in a consistent fashion. GCC cannot even decide which input
4913 files have been specified without knowing which switches take arguments,
4914 and it must know which input files to compile in order to tell which
4917 GCC also knows implicitly that arguments starting in @option{-l} are to be
4918 treated as compiler output files, and passed to the linker in their
4919 proper position among the other output files.
4921 @c man begin OPTIONS
4923 @node Target Options
4924 @section Specifying Target Machine and Compiler Version
4925 @cindex target options
4926 @cindex cross compiling
4927 @cindex specifying machine version
4928 @cindex specifying compiler version and target machine
4929 @cindex compiler version, specifying
4930 @cindex target machine, specifying
4932 By default, GCC compiles code for the same type of machine that you
4933 are using. However, it can also be installed as a cross-compiler, to
4934 compile for some other type of machine. In fact, several different
4935 configurations of GCC, for different target machines, can be
4936 installed side by side. Then you specify which one to use with the
4939 In addition, older and newer versions of GCC can be installed side
4940 by side. One of them (probably the newest) will be the default, but
4941 you may sometimes wish to use another.
4944 @item -b @var{machine}
4946 The argument @var{machine} specifies the target machine for compilation.
4947 This is useful when you have installed GCC as a cross-compiler.
4949 The value to use for @var{machine} is the same as was specified as the
4950 machine type when configuring GCC as a cross-compiler. For
4951 example, if a cross-compiler was configured with @samp{configure
4952 i386v}, meaning to compile for an 80386 running System V, then you
4953 would specify @option{-b i386v} to run that cross compiler.
4955 When you do not specify @option{-b}, it normally means to compile for
4956 the same type of machine that you are using.
4958 @item -V @var{version}
4960 The argument @var{version} specifies which version of GCC to run.
4961 This is useful when multiple versions are installed. For example,
4962 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
4964 The default version, when you do not specify @option{-V}, is the last
4965 version of GCC that you installed.
4968 The @option{-b} and @option{-V} options actually work by controlling part of
4969 the file name used for the executable files and libraries used for
4970 compilation. A given version of GCC, for a given target machine, is
4971 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.
4973 Thus, sites can customize the effect of @option{-b} or @option{-V} either by
4974 changing the names of these directories or adding alternate names (or
4975 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
4976 file @file{80386} is a link to the file @file{i386v}, then @option{-b
4977 80386} becomes an alias for @option{-b i386v}.
4979 In one respect, the @option{-b} or @option{-V} do not completely change
4980 to a different compiler: the top-level driver program @command{gcc}
4981 that you originally invoked continues to run and invoke the other
4982 executables (preprocessor, compiler per se, assembler and linker)
4983 that do the real work. However, since no real work is done in the
4984 driver program, it usually does not matter that the driver program
4985 in use is not the one for the specified target. It is common for the
4986 interface to the other executables to change incompatibly between
4987 compiler versions, so unless the version specified is very close to that
4988 of the driver (for example, @option{-V 3.0} with a driver program from GCC
4989 version 3.0.1), use of @option{-V} may not work; for example, using
4990 @option{-V 2.95.2} will not work with a driver program from GCC 3.0.
4992 The only way that the driver program depends on the target machine is
4993 in the parsing and handling of special machine-specific options.
4994 However, this is controlled by a file which is found, along with the
4995 other executables, in the directory for the specified version and
4996 target machine. As a result, a single installed driver program adapts
4997 to any specified target machine, and sufficiently similar compiler
5000 The driver program executable does control one significant thing,
5001 however: the default version and target machine. Therefore, you can
5002 install different instances of the driver program, compiled for
5003 different targets or versions, under different names.
5005 For example, if the driver for version 2.0 is installed as @command{ogcc}
5006 and that for version 2.1 is installed as @command{gcc}, then the command
5007 @command{gcc} will use version 2.1 by default, while @command{ogcc} will use
5008 2.0 by default. However, you can choose either version with either
5009 command with the @option{-V} option.
5011 @node Submodel Options
5012 @section Hardware Models and Configurations
5013 @cindex submodel options
5014 @cindex specifying hardware config
5015 @cindex hardware models and configurations, specifying
5016 @cindex machine dependent options
5018 Earlier we discussed the standard option @option{-b} which chooses among
5019 different installed compilers for completely different target
5020 machines, such as Vax vs.@: 68000 vs.@: 80386.
5022 In addition, each of these target machine types can have its own
5023 special options, starting with @samp{-m}, to choose among various
5024 hardware models or configurations---for example, 68010 vs 68020,
5025 floating coprocessor or none. A single installed version of the
5026 compiler can compile for any model or configuration, according to the
5029 Some configurations of the compiler also support additional special
5030 options, usually for compatibility with other compilers on the same
5034 These options are defined by the macro @code{TARGET_SWITCHES} in the
5035 machine description. The default for the options is also defined by
5036 that macro, which enables you to change the defaults.
5051 * RS/6000 and PowerPC Options::
5056 * Intel 960 Options::
5057 * DEC Alpha Options::
5061 * System V Options::
5062 * TMS320C3x/C4x Options::
5072 @node M680x0 Options
5073 @subsection M680x0 Options
5074 @cindex M680x0 options
5076 These are the @samp{-m} options defined for the 68000 series. The default
5077 values for these options depends on which style of 68000 was selected when
5078 the compiler was configured; the defaults for the most common choices are
5086 Generate output for a 68000. This is the default
5087 when the compiler is configured for 68000-based systems.
5089 Use this option for microcontrollers with a 68000 or EC000 core,
5090 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5096 Generate output for a 68020. This is the default
5097 when the compiler is configured for 68020-based systems.
5101 Generate output containing 68881 instructions for floating point.
5102 This is the default for most 68020 systems unless @option{--nfp} was
5103 specified when the compiler was configured.
5107 Generate output for a 68030. This is the default when the compiler is
5108 configured for 68030-based systems.
5112 Generate output for a 68040. This is the default when the compiler is
5113 configured for 68040-based systems.
5115 This option inhibits the use of 68881/68882 instructions that have to be
5116 emulated by software on the 68040. Use this option if your 68040 does not
5117 have code to emulate those instructions.
5121 Generate output for a 68060. This is the default when the compiler is
5122 configured for 68060-based systems.
5124 This option inhibits the use of 68020 and 68881/68882 instructions that
5125 have to be emulated by software on the 68060. Use this option if your 68060
5126 does not have code to emulate those instructions.
5130 Generate output for a CPU32. This is the default
5131 when the compiler is configured for CPU32-based systems.
5133 Use this option for microcontrollers with a
5134 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5135 68336, 68340, 68341, 68349 and 68360.
5139 Generate output for a 520X ``coldfire'' family cpu. This is the default
5140 when the compiler is configured for 520X-based systems.
5142 Use this option for microcontroller with a 5200 core, including
5143 the MCF5202, MCF5203, MCF5204 and MCF5202.
5148 Generate output for a 68040, without using any of the new instructions.
5149 This results in code which can run relatively efficiently on either a
5150 68020/68881 or a 68030 or a 68040. The generated code does use the
5151 68881 instructions that are emulated on the 68040.
5155 Generate output for a 68060, without using any of the new instructions.
5156 This results in code which can run relatively efficiently on either a
5157 68020/68881 or a 68030 or a 68040. The generated code does use the
5158 68881 instructions that are emulated on the 68060.
5162 Generate output containing Sun FPA instructions for floating point.
5165 @opindex msoft-float
5166 Generate output containing library calls for floating point.
5167 @strong{Warning:} the requisite libraries are not available for all m68k
5168 targets. Normally the facilities of the machine's usual C compiler are
5169 used, but this can't be done directly in cross-compilation. You must
5170 make your own arrangements to provide suitable library functions for
5171 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5172 @samp{m68k-*-coff} do provide software floating point support.
5176 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5179 @opindex mnobitfield
5180 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5181 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5185 Do use the bit-field instructions. The @option{-m68020} option implies
5186 @option{-mbitfield}. This is the default if you use a configuration
5187 designed for a 68020.
5191 Use a different function-calling convention, in which functions
5192 that take a fixed number of arguments return with the @code{rtd}
5193 instruction, which pops their arguments while returning. This
5194 saves one instruction in the caller since there is no need to pop
5195 the arguments there.
5197 This calling convention is incompatible with the one normally
5198 used on Unix, so you cannot use it if you need to call libraries
5199 compiled with the Unix compiler.
5201 Also, you must provide function prototypes for all functions that
5202 take variable numbers of arguments (including @code{printf});
5203 otherwise incorrect code will be generated for calls to those
5206 In addition, seriously incorrect code will result if you call a
5207 function with too many arguments. (Normally, extra arguments are
5208 harmlessly ignored.)
5210 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5211 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5214 @itemx -mno-align-int
5216 @opindex mno-align-int
5217 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5218 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5219 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
5220 Aligning variables on 32-bit boundaries produces code that runs somewhat
5221 faster on processors with 32-bit busses at the expense of more memory.
5223 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
5224 align structures containing the above types differently than
5225 most published application binary interface specifications for the m68k.
5229 Use the pc-relative addressing mode of the 68000 directly, instead of
5230 using a global offset table. At present, this option implies @option{-fpic},
5231 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
5232 not presently supported with @option{-mpcrel}, though this could be supported for
5233 68020 and higher processors.
5235 @item -mno-strict-align
5236 @itemx -mstrict-align
5237 @opindex mno-strict-align
5238 @opindex mstrict-align
5239 Do not (do) assume that unaligned memory references will be handled by
5244 @node M68hc1x Options
5245 @subsection M68hc1x Options
5246 @cindex M68hc1x options
5248 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5249 microcontrollers. The default values for these options depends on
5250 which style of microcontroller was selected when the compiler was configured;
5251 the defaults for the most common choices are given below.
5258 Generate output for a 68HC11. This is the default
5259 when the compiler is configured for 68HC11-based systems.
5265 Generate output for a 68HC12. This is the default
5266 when the compiler is configured for 68HC12-based systems.
5269 @opindex mauto-incdec
5270 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5275 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5277 @item -msoft-reg-count=@var{count}
5278 @opindex msoft-reg-count
5279 Specify the number of pseudo-soft registers which are used for the
5280 code generation. The maximum number is 32. Using more pseudo-soft
5281 register may or may not result in better code depending on the program.
5282 The default is 4 for 68HC11 and 2 for 68HC12.
5287 @subsection VAX Options
5290 These @samp{-m} options are defined for the Vax:
5295 Do not output certain jump instructions (@code{aobleq} and so on)
5296 that the Unix assembler for the Vax cannot handle across long
5301 Do output those jump instructions, on the assumption that you
5302 will assemble with the GNU assembler.
5306 Output code for g-format floating point numbers instead of d-format.
5310 @subsection SPARC Options
5311 @cindex SPARC options
5313 These @samp{-m} switches are supported on the SPARC:
5318 @opindex mno-app-regs
5320 Specify @option{-mapp-regs} to generate output using the global registers
5321 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
5324 To be fully SVR4 ABI compliant at the cost of some performance loss,
5325 specify @option{-mno-app-regs}. You should compile libraries and system
5326 software with this option.
5331 @opindex mhard-float
5332 Generate output containing floating point instructions. This is the
5338 @opindex msoft-float
5339 Generate output containing library calls for floating point.
5340 @strong{Warning:} the requisite libraries are not available for all SPARC
5341 targets. Normally the facilities of the machine's usual C compiler are
5342 used, but this cannot be done directly in cross-compilation. You must make
5343 your own arrangements to provide suitable library functions for
5344 cross-compilation. The embedded targets @samp{sparc-*-aout} and
5345 @samp{sparclite-*-*} do provide software floating point support.
5347 @option{-msoft-float} changes the calling convention in the output file;
5348 therefore, it is only useful if you compile @emph{all} of a program with
5349 this option. In particular, you need to compile @file{libgcc.a}, the
5350 library that comes with GCC, with @option{-msoft-float} in order for
5353 @item -mhard-quad-float
5354 @opindex mhard-quad-float
5355 Generate output containing quad-word (long double) floating point
5358 @item -msoft-quad-float
5359 @opindex msoft-quad-float
5360 Generate output containing library calls for quad-word (long double)
5361 floating point instructions. The functions called are those specified
5362 in the SPARC ABI@. This is the default.
5364 As of this writing, there are no sparc implementations that have hardware
5365 support for the quad-word floating point instructions. They all invoke
5366 a trap handler for one of these instructions, and then the trap handler
5367 emulates the effect of the instruction. Because of the trap handler overhead,
5368 this is much slower than calling the ABI library routines. Thus the
5369 @option{-msoft-quad-float} option is the default.
5373 @opindex mno-epilogue
5375 With @option{-mepilogue} (the default), the compiler always emits code for
5376 function exit at the end of each function. Any function exit in
5377 the middle of the function (such as a return statement in C) will
5378 generate a jump to the exit code at the end of the function.
5380 With @option{-mno-epilogue}, the compiler tries to emit exit code inline
5381 at every function exit.
5387 With @option{-mflat}, the compiler does not generate save/restore instructions
5388 and will use a ``flat'' or single register window calling convention.
5389 This model uses %i7 as the frame pointer and is compatible with the normal
5390 register window model. Code from either may be intermixed.
5391 The local registers and the input registers (0--5) are still treated as
5392 ``call saved'' registers and will be saved on the stack as necessary.
5394 With @option{-mno-flat} (the default), the compiler emits save/restore
5395 instructions (except for leaf functions) and is the normal mode of operation.
5397 @item -mno-unaligned-doubles
5398 @itemx -munaligned-doubles
5399 @opindex mno-unaligned-doubles
5400 @opindex munaligned-doubles
5401 Assume that doubles have 8 byte alignment. This is the default.
5403 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
5404 alignment only if they are contained in another type, or if they have an
5405 absolute address. Otherwise, it assumes they have 4 byte alignment.
5406 Specifying this option avoids some rare compatibility problems with code
5407 generated by other compilers. It is not the default because it results
5408 in a performance loss, especially for floating point code.
5410 @item -mno-faster-structs
5411 @itemx -mfaster-structs
5412 @opindex mno-faster-structs
5413 @opindex mfaster-structs
5414 With @option{-mfaster-structs}, the compiler assumes that structures
5415 should have 8 byte alignment. This enables the use of pairs of
5416 @code{ldd} and @code{std} instructions for copies in structure
5417 assignment, in place of twice as many @code{ld} and @code{st} pairs.
5418 However, the use of this changed alignment directly violates the Sparc
5419 ABI@. Thus, it's intended only for use on targets where the developer
5420 acknowledges that their resulting code will not be directly in line with
5421 the rules of the ABI@.
5427 These two options select variations on the SPARC architecture.
5429 By default (unless specifically configured for the Fujitsu SPARClite),
5430 GCC generates code for the v7 variant of the SPARC architecture.
5432 @option{-mv8} will give you SPARC v8 code. The only difference from v7
5433 code is that the compiler emits the integer multiply and integer
5434 divide instructions which exist in SPARC v8 but not in SPARC v7.
5436 @option{-msparclite} will give you SPARClite code. This adds the integer
5437 multiply, integer divide step and scan (@code{ffs}) instructions which
5438 exist in SPARClite but not in SPARC v7.
5440 These options are deprecated and will be deleted in a future GCC release.
5441 They have been replaced with @option{-mcpu=xxx}.
5446 @opindex msupersparc
5447 These two options select the processor for which the code is optimised.
5449 With @option{-mcypress} (the default), the compiler optimizes code for the
5450 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
5451 This is also appropriate for the older SparcStation 1, 2, IPX etc.
5453 With @option{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
5454 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
5455 of the full SPARC v8 instruction set.
5457 These options are deprecated and will be deleted in a future GCC release.
5458 They have been replaced with @option{-mcpu=xxx}.
5460 @item -mcpu=@var{cpu_type}
5462 Set the instruction set, register set, and instruction scheduling parameters
5463 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
5464 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
5465 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
5466 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
5468 Default instruction scheduling parameters are used for values that select
5469 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
5470 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
5472 Here is a list of each supported architecture and their supported
5477 v8: supersparc, hypersparc
5478 sparclite: f930, f934, sparclite86x
5483 @item -mtune=@var{cpu_type}
5485 Set the instruction scheduling parameters for machine type
5486 @var{cpu_type}, but do not set the instruction set or register set that the
5487 option @option{-mcpu=@var{cpu_type}} would.
5489 The same values for @option{-mcpu=@var{cpu_type}} are used for
5490 @option{-mtune=@var{cpu_type}}, though the only useful values are those that
5491 select a particular cpu implementation: @samp{cypress}, @samp{supersparc},
5492 @samp{hypersparc}, @samp{f930}, @samp{f934}, @samp{sparclite86x},
5493 @samp{tsc701}, @samp{ultrasparc}.
5497 These @samp{-m} switches are supported in addition to the above
5498 on the SPARCLET processor.
5501 @item -mlittle-endian
5502 @opindex mlittle-endian
5503 Generate code for a processor running in little-endian mode.
5507 Treat register @code{%g0} as a normal register.
5508 GCC will continue to clobber it as necessary but will not assume
5509 it always reads as 0.
5511 @item -mbroken-saverestore
5512 @opindex mbroken-saverestore
5513 Generate code that does not use non-trivial forms of the @code{save} and
5514 @code{restore} instructions. Early versions of the SPARCLET processor do
5515 not correctly handle @code{save} and @code{restore} instructions used with
5516 arguments. They correctly handle them used without arguments. A @code{save}
5517 instruction used without arguments increments the current window pointer
5518 but does not allocate a new stack frame. It is assumed that the window
5519 overflow trap handler will properly handle this case as will interrupt
5523 These @samp{-m} switches are supported in addition to the above
5524 on SPARC V9 processors in 64-bit environments.
5527 @item -mlittle-endian
5528 @opindex mlittle-endian
5529 Generate code for a processor running in little-endian mode.
5535 Generate code for a 32-bit or 64-bit environment.
5536 The 32-bit environment sets int, long and pointer to 32 bits.
5537 The 64-bit environment sets int to 32 bits and long and pointer
5540 @item -mcmodel=medlow
5541 @opindex mcmodel=medlow
5542 Generate code for the Medium/Low code model: the program must be linked
5543 in the low 32 bits of the address space. Pointers are 64 bits.
5544 Programs can be statically or dynamically linked.
5546 @item -mcmodel=medmid
5547 @opindex mcmodel=medmid
5548 Generate code for the Medium/Middle code model: the program must be linked
5549 in the low 44 bits of the address space, the text segment must be less than
5550 2G bytes, and data segment must be within 2G of the text segment.
5551 Pointers are 64 bits.
5553 @item -mcmodel=medany
5554 @opindex mcmodel=medany
5555 Generate code for the Medium/Anywhere code model: the program may be linked
5556 anywhere in the address space, the text segment must be less than
5557 2G bytes, and data segment must be within 2G of the text segment.
5558 Pointers are 64 bits.
5560 @item -mcmodel=embmedany
5561 @opindex mcmodel=embmedany
5562 Generate code for the Medium/Anywhere code model for embedded systems:
5563 assume a 32-bit text and a 32-bit data segment, both starting anywhere
5564 (determined at link time). Register %g4 points to the base of the
5565 data segment. Pointers are still 64 bits.
5566 Programs are statically linked, PIC is not supported.
5569 @itemx -mno-stack-bias
5570 @opindex mstack-bias
5571 @opindex mno-stack-bias
5572 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
5573 frame pointer if present, are offset by @minus{}2047 which must be added back
5574 when making stack frame references.
5575 Otherwise, assume no such offset is present.
5578 @node Convex Options
5579 @subsection Convex Options
5580 @cindex Convex options
5582 These @samp{-m} options are defined for Convex:
5587 Generate output for C1. The code will run on any Convex machine.
5588 The preprocessor symbol @code{__convex__c1__} is defined.
5592 Generate output for C2. Uses instructions not available on C1.
5593 Scheduling and other optimizations are chosen for max performance on C2.
5594 The preprocessor symbol @code{__convex_c2__} is defined.
5598 Generate output for C32xx. Uses instructions not available on C1.
5599 Scheduling and other optimizations are chosen for max performance on C32.
5600 The preprocessor symbol @code{__convex_c32__} is defined.
5604 Generate output for C34xx. Uses instructions not available on C1.
5605 Scheduling and other optimizations are chosen for max performance on C34.
5606 The preprocessor symbol @code{__convex_c34__} is defined.
5610 Generate output for C38xx. Uses instructions not available on C1.
5611 Scheduling and other optimizations are chosen for max performance on C38.
5612 The preprocessor symbol @code{__convex_c38__} is defined.
5616 Generate code which puts an argument count in the word preceding each
5617 argument list. This is compatible with regular CC, and a few programs
5618 may need the argument count word. GDB and other source-level debuggers
5619 do not need it; this info is in the symbol table.
5622 @opindex mnoargcount
5623 Omit the argument count word. This is the default.
5625 @item -mvolatile-cache
5626 @opindex mvolatile-cache
5627 Allow volatile references to be cached. This is the default.
5629 @item -mvolatile-nocache
5630 @opindex mvolatile-nocache
5631 Volatile references bypass the data cache, going all the way to memory.
5632 This is only needed for multi-processor code that does not use standard
5633 synchronization instructions. Making non-volatile references to volatile
5634 locations will not necessarily work.
5638 Type long is 32 bits, the same as type int. This is the default.
5642 Type long is 64 bits, the same as type long long. This option is useless,
5643 because no library support exists for it.
5646 @node AMD29K Options
5647 @subsection AMD29K Options
5648 @cindex AMD29K options
5650 These @samp{-m} options are defined for the AMD Am29000:
5655 @cindex DW bit (29k)
5656 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
5657 halfword operations are directly supported by the hardware. This is the
5662 Generate code that assumes the @code{DW} bit is not set.
5666 @cindex byte writes (29k)
5667 Generate code that assumes the system supports byte and halfword write
5668 operations. This is the default.
5672 Generate code that assumes the systems does not support byte and
5673 halfword write operations. @option{-mnbw} implies @option{-mndw}.
5677 @cindex memory model (29k)
5678 Use a small memory model that assumes that all function addresses are
5679 either within a single 256 KB segment or at an absolute address of less
5680 than 256k. This allows the @code{call} instruction to be used instead
5681 of a @code{const}, @code{consth}, @code{calli} sequence.
5685 Use the normal memory model: Generate @code{call} instructions only when
5686 calling functions in the same file and @code{calli} instructions
5687 otherwise. This works if each file occupies less than 256 KB but allows
5688 the entire executable to be larger than 256 KB@. This is the default.
5692 Always use @code{calli} instructions. Specify this option if you expect
5693 a single file to compile into more than 256 KB of code.
5697 @cindex processor selection (29k)
5698 Generate code for the Am29050.
5702 Generate code for the Am29000. This is the default.
5704 @item -mkernel-registers
5705 @opindex mkernel-registers
5706 @cindex kernel and user registers (29k)
5707 Generate references to registers @code{gr64-gr95} instead of to
5708 registers @code{gr96-gr127}. This option can be used when compiling
5709 kernel code that wants a set of global registers disjoint from that used
5712 Note that when this option is used, register names in @samp{-f} flags
5713 must use the normal, user-mode, names.
5715 @item -muser-registers
5716 @opindex muser-registers
5717 Use the normal set of global registers, @code{gr96-gr127}. This is the
5721 @itemx -mno-stack-check
5722 @opindex mstack-check
5723 @opindex mno-stack-check
5724 @cindex stack checks (29k)
5725 Insert (or do not insert) a call to @code{__msp_check} after each stack
5726 adjustment. This is often used for kernel code.
5729 @itemx -mno-storem-bug
5730 @opindex mstorem-bug
5731 @opindex mno-storem-bug
5732 @cindex storem bug (29k)
5733 @option{-mstorem-bug} handles 29k processors which cannot handle the
5734 separation of a mtsrim insn and a storem instruction (most 29000 chips
5735 to date, but not the 29050).
5737 @item -mno-reuse-arg-regs
5738 @itemx -mreuse-arg-regs
5739 @opindex mno-reuse-arg-regs
5740 @opindex mreuse-arg-regs
5741 @option{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
5742 registers for copying out arguments. This helps detect calling a function
5743 with fewer arguments than it was declared with.
5745 @item -mno-impure-text
5746 @itemx -mimpure-text
5747 @opindex mno-impure-text
5748 @opindex mimpure-text
5749 @option{-mimpure-text}, used in addition to @option{-shared}, tells the compiler to
5750 not pass @option{-assert pure-text} to the linker when linking a shared object.
5753 @opindex msoft-float
5754 Generate output containing library calls for floating point.
5755 @strong{Warning:} the requisite libraries are not part of GCC@.
5756 Normally the facilities of the machine's usual C compiler are used, but
5757 this can't be done directly in cross-compilation. You must make your
5758 own arrangements to provide suitable library functions for
5763 Do not generate multm or multmu instructions. This is useful for some embedded
5764 systems which do not have trap handlers for these instructions.
5768 @subsection ARM Options
5771 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
5776 @opindex mapcs-frame
5777 Generate a stack frame that is compliant with the ARM Procedure Call
5778 Standard for all functions, even if this is not strictly necessary for
5779 correct execution of the code. Specifying @option{-fomit-frame-pointer}
5780 with this option will cause the stack frames not to be generated for
5781 leaf functions. The default is @option{-mno-apcs-frame}.
5785 This is a synonym for @option{-mapcs-frame}.
5789 Generate code for a processor running with a 26-bit program counter,
5790 and conforming to the function calling standards for the APCS 26-bit
5791 option. This option replaces the @option{-m2} and @option{-m3} options
5792 of previous releases of the compiler.
5796 Generate code for a processor running with a 32-bit program counter,
5797 and conforming to the function calling standards for the APCS 32-bit
5798 option. This option replaces the @option{-m6} option of previous releases
5802 @c not currently implemented
5803 @item -mapcs-stack-check
5804 @opindex mapcs-stack-check
5805 Generate code to check the amount of stack space available upon entry to
5806 every function (that actually uses some stack space). If there is
5807 insufficient space available then either the function
5808 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
5809 called, depending upon the amount of stack space required. The run time
5810 system is required to provide these functions. The default is
5811 @option{-mno-apcs-stack-check}, since this produces smaller code.
5813 @c not currently implemented
5815 @opindex mapcs-float
5816 Pass floating point arguments using the float point registers. This is
5817 one of the variants of the APCS@. This option is recommended if the
5818 target hardware has a floating point unit or if a lot of floating point
5819 arithmetic is going to be performed by the code. The default is
5820 @option{-mno-apcs-float}, since integer only code is slightly increased in
5821 size if @option{-mapcs-float} is used.
5823 @c not currently implemented
5824 @item -mapcs-reentrant
5825 @opindex mapcs-reentrant
5826 Generate reentrant, position independent code. The default is
5827 @option{-mno-apcs-reentrant}.
5830 @item -mthumb-interwork
5831 @opindex mthumb-interwork
5832 Generate code which supports calling between the ARM and Thumb
5833 instruction sets. Without this option the two instruction sets cannot
5834 be reliably used inside one program. The default is
5835 @option{-mno-thumb-interwork}, since slightly larger code is generated
5836 when @option{-mthumb-interwork} is specified.
5838 @item -mno-sched-prolog
5839 @opindex mno-sched-prolog
5840 Prevent the reordering of instructions in the function prolog, or the
5841 merging of those instruction with the instructions in the function's
5842 body. This means that all functions will start with a recognizable set
5843 of instructions (or in fact one of a choice from a small set of
5844 different function prologues), and this information can be used to
5845 locate the start if functions inside an executable piece of code. The
5846 default is @option{-msched-prolog}.
5849 @opindex mhard-float
5850 Generate output containing floating point instructions. This is the
5854 @opindex msoft-float
5855 Generate output containing library calls for floating point.
5856 @strong{Warning:} the requisite libraries are not available for all ARM
5857 targets. Normally the facilities of the machine's usual C compiler are
5858 used, but this cannot be done directly in cross-compilation. You must make
5859 your own arrangements to provide suitable library functions for
5862 @option{-msoft-float} changes the calling convention in the output file;
5863 therefore, it is only useful if you compile @emph{all} of a program with
5864 this option. In particular, you need to compile @file{libgcc.a}, the
5865 library that comes with GCC, with @option{-msoft-float} in order for
5868 @item -mlittle-endian
5869 @opindex mlittle-endian
5870 Generate code for a processor running in little-endian mode. This is
5871 the default for all standard configurations.
5874 @opindex mbig-endian
5875 Generate code for a processor running in big-endian mode; the default is
5876 to compile code for a little-endian processor.
5878 @item -mwords-little-endian
5879 @opindex mwords-little-endian
5880 This option only applies when generating code for big-endian processors.
5881 Generate code for a little-endian word order but a big-endian byte
5882 order. That is, a byte order of the form @samp{32107654}. Note: this
5883 option should only be used if you require compatibility with code for
5884 big-endian ARM processors generated by versions of the compiler prior to
5887 @item -malignment-traps
5888 @opindex malignment-traps
5889 Generate code that will not trap if the MMU has alignment traps enabled.
5890 On ARM architectures prior to ARMv4, there were no instructions to
5891 access half-word objects stored in memory. However, when reading from
5892 memory a feature of the ARM architecture allows a word load to be used,
5893 even if the address is unaligned, and the processor core will rotate the
5894 data as it is being loaded. This option tells the compiler that such
5895 misaligned accesses will cause a MMU trap and that it should instead
5896 synthesise the access as a series of byte accesses. The compiler can
5897 still use word accesses to load half-word data if it knows that the
5898 address is aligned to a word boundary.
5900 This option is ignored when compiling for ARM architecture 4 or later,
5901 since these processors have instructions to directly access half-word
5904 @item -mno-alignment-traps
5905 @opindex mno-alignment-traps
5906 Generate code that assumes that the MMU will not trap unaligned
5907 accesses. This produces better code when the target instruction set
5908 does not have half-word memory operations (i.e.@: implementations prior to
5911 Note that you cannot use this option to access unaligned word objects,
5912 since the processor will only fetch one 32-bit aligned object from
5915 The default setting for most targets is @option{-mno-alignment-traps}, since
5916 this produces better code when there are no half-word memory
5917 instructions available.
5919 @item -mshort-load-bytes
5920 @itemx -mno-short-load-words
5921 @opindex mshort-load-bytes
5922 @opindex mno-short-load-words
5923 These are deprecated aliases for @option{-malignment-traps}.
5925 @item -mno-short-load-bytes
5926 @itemx -mshort-load-words
5927 @opindex mno-short-load-bytes
5928 @opindex mshort-load-words
5929 This are deprecated aliases for @option{-mno-alignment-traps}.
5933 This option only applies to RISC iX@. Emulate the native BSD-mode
5934 compiler. This is the default if @option{-ansi} is not specified.
5938 This option only applies to RISC iX@. Emulate the native X/Open-mode
5941 @item -mno-symrename
5942 @opindex mno-symrename
5943 This option only applies to RISC iX@. Do not run the assembler
5944 post-processor, @samp{symrename}, after code has been assembled.
5945 Normally it is necessary to modify some of the standard symbols in
5946 preparation for linking with the RISC iX C library; this option
5947 suppresses this pass. The post-processor is never run when the
5948 compiler is built for cross-compilation.
5950 @item -mcpu=@var{name}
5952 This specifies the name of the target ARM processor. GCC uses this name
5953 to determine what kind of instructions it can emit when generating
5954 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
5955 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
5956 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
5957 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
5958 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
5959 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
5960 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
5961 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
5962 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
5963 @samp{arm1020t}, @samp{xscale}.
5965 @itemx -mtune=@var{name}
5967 This option is very similar to the @option{-mcpu=} option, except that
5968 instead of specifying the actual target processor type, and hence
5969 restricting which instructions can be used, it specifies that GCC should
5970 tune the performance of the code as if the target were of the type
5971 specified in this option, but still choosing the instructions that it
5972 will generate based on the cpu specified by a @option{-mcpu=} option.
5973 For some ARM implementations better performance can be obtained by using
5976 @item -march=@var{name}
5978 This specifies the name of the target ARM architecture. GCC uses this
5979 name to determine what kind of instructions it can emit when generating
5980 assembly code. This option can be used in conjunction with or instead
5981 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
5982 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
5983 @samp{armv5}, @samp{armv5t}, @samp{armv5te}.
5985 @item -mfpe=@var{number}
5986 @itemx -mfp=@var{number}
5989 This specifies the version of the floating point emulation available on
5990 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
5991 for @option{-mfpe=}, for compatibility with older versions of GCC@.
5993 @item -mstructure-size-boundary=@var{n}
5994 @opindex mstructure-size-boundary
5995 The size of all structures and unions will be rounded up to a multiple
5996 of the number of bits set by this option. Permissible values are 8 and
5997 32. The default value varies for different toolchains. For the COFF
5998 targeted toolchain the default value is 8. Specifying the larger number
5999 can produce faster, more efficient code, but can also increase the size
6000 of the program. The two values are potentially incompatible. Code
6001 compiled with one value cannot necessarily expect to work with code or
6002 libraries compiled with the other value, if they exchange information
6003 using structures or unions.
6005 @item -mabort-on-noreturn
6006 @opindex mabort-on-noreturn
6007 Generate a call to the function @code{abort} at the end of a
6008 @code{noreturn} function. It will be executed if the function tries to
6012 @itemx -mno-long-calls
6013 @opindex mlong-calls
6014 @opindex mno-long-calls
6015 Tells the compiler to perform function calls by first loading the
6016 address of the function into a register and then performing a subroutine
6017 call on this register. This switch is needed if the target function
6018 will lie outside of the 64 megabyte addressing range of the offset based
6019 version of subroutine call instruction.
6021 Even if this switch is enabled, not all function calls will be turned
6022 into long calls. The heuristic is that static functions, functions
6023 which have the @samp{short-call} attribute, functions that are inside
6024 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6025 definitions have already been compiled within the current compilation
6026 unit, will not be turned into long calls. The exception to this rule is
6027 that weak function definitions, functions with the @samp{long-call}
6028 attribute or the @samp{section} attribute, and functions that are within
6029 the scope of a @samp{#pragma long_calls} directive, will always be
6030 turned into long calls.
6032 This feature is not enabled by default. Specifying
6033 @option{-mno-long-calls} will restore the default behaviour, as will
6034 placing the function calls within the scope of a @samp{#pragma
6035 long_calls_off} directive. Note these switches have no effect on how
6036 the compiler generates code to handle function calls via function
6039 @item -mnop-fun-dllimport
6040 @opindex mnop-fun-dllimport
6041 Disable support for the @code{dllimport} attribute.
6043 @item -msingle-pic-base
6044 @opindex msingle-pic-base
6045 Treat the register used for PIC addressing as read-only, rather than
6046 loading it in the prologue for each function. The run-time system is
6047 responsible for initialising this register with an appropriate value
6048 before execution begins.
6050 @item -mpic-register=@var{reg}
6051 @opindex mpic-register
6052 Specify the register to be used for PIC addressing. The default is R10
6053 unless stack-checking is enabled, when R9 is used.
6055 @item -mpoke-function-name
6056 @opindex mpoke-function-name
6057 Write the name of each function into the text section, directly
6058 preceding the function prologue. The generated code is similar to this:
6062 .ascii "arm_poke_function_name", 0
6065 .word 0xff000000 + (t1 - t0)
6066 arm_poke_function_name
6068 stmfd sp!, @{fp, ip, lr, pc@}
6072 When performing a stack backtrace, code can inspect the value of
6073 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6074 location @code{pc - 12} and the top 8 bits are set, then we know that
6075 there is a function name embedded immediately preceding this location
6076 and has length @code{((pc[-3]) & 0xff000000)}.
6080 Generate code for the 16-bit Thumb instruction set. The default is to
6081 use the 32-bit ARM instruction set.
6084 @opindex mtpcs-frame
6085 Generate a stack frame that is compliant with the Thumb Procedure Call
6086 Standard for all non-leaf functions. (A leaf function is one that does
6087 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6089 @item -mtpcs-leaf-frame
6090 @opindex mtpcs-leaf-frame
6091 Generate a stack frame that is compliant with the Thumb Procedure Call
6092 Standard for all leaf functions. (A leaf function is one that does
6093 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6095 @item -mcallee-super-interworking
6096 @opindex mcallee-super-interworking
6097 Gives all externally visible functions in the file being compiled an ARM
6098 instruction set header which switches to Thumb mode before executing the
6099 rest of the function. This allows these functions to be called from
6100 non-interworking code.
6102 @item -mcaller-super-interworking
6103 @opindex mcaller-super-interworking
6104 Allows calls via function pointers (including virtual functions) to
6105 execute correctly regardless of whether the target code has been
6106 compiled for interworking or not. There is a small overhead in the cost
6107 of executing a function pointer if this option is enabled.
6111 @node MN10200 Options
6112 @subsection MN10200 Options
6113 @cindex MN10200 options
6114 These @option{-m} options are defined for Matsushita MN10200 architectures:
6119 Indicate to the linker that it should perform a relaxation optimization pass
6120 to shorten branches, calls and absolute memory addresses. This option only
6121 has an effect when used on the command line for the final link step.
6123 This option makes symbolic debugging impossible.
6126 @node MN10300 Options
6127 @subsection MN10300 Options
6128 @cindex MN10300 options
6129 These @option{-m} options are defined for Matsushita MN10300 architectures:
6134 Generate code to avoid bugs in the multiply instructions for the MN10300
6135 processors. This is the default.
6138 @opindex mno-mult-bug
6139 Do not generate code to avoid bugs in the multiply instructions for the
6144 Generate code which uses features specific to the AM33 processor.
6148 Do not generate code which uses features specific to the AM33 processor. This
6153 Do not link in the C run-time initialization object file.
6157 Indicate to the linker that it should perform a relaxation optimization pass
6158 to shorten branches, calls and absolute memory addresses. This option only
6159 has an effect when used on the command line for the final link step.
6161 This option makes symbolic debugging impossible.
6165 @node M32R/D Options
6166 @subsection M32R/D Options
6167 @cindex M32R/D options
6169 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6172 @item -mcode-model=small
6173 @opindex mcode-model=small
6174 Assume all objects live in the lower 16MB of memory (so that their addresses
6175 can be loaded with the @code{ld24} instruction), and assume all subroutines
6176 are reachable with the @code{bl} instruction.
6177 This is the default.
6179 The addressability of a particular object can be set with the
6180 @code{model} attribute.
6182 @item -mcode-model=medium
6183 @opindex mcode-model=medium
6184 Assume objects may be anywhere in the 32-bit address space (the compiler
6185 will generate @code{seth/add3} instructions to load their addresses), and
6186 assume all subroutines are reachable with the @code{bl} instruction.
6188 @item -mcode-model=large
6189 @opindex mcode-model=large
6190 Assume objects may be anywhere in the 32-bit address space (the compiler
6191 will generate @code{seth/add3} instructions to load their addresses), and
6192 assume subroutines may not be reachable with the @code{bl} instruction
6193 (the compiler will generate the much slower @code{seth/add3/jl}
6194 instruction sequence).
6197 @opindex msdata=none
6198 Disable use of the small data area. Variables will be put into
6199 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6200 @code{section} attribute has been specified).
6201 This is the default.
6203 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6204 Objects may be explicitly put in the small data area with the
6205 @code{section} attribute using one of these sections.
6208 @opindex msdata=sdata
6209 Put small global and static data in the small data area, but do not
6210 generate special code to reference them.
6214 Put small global and static data in the small data area, and generate
6215 special instructions to reference them.
6219 @cindex smaller data references
6220 Put global and static objects less than or equal to @var{num} bytes
6221 into the small data or bss sections instead of the normal data or bss
6222 sections. The default value of @var{num} is 8.
6223 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6224 for this option to have any effect.
6226 All modules should be compiled with the same @option{-G @var{num}} value.
6227 Compiling with different values of @var{num} may or may not work; if it
6228 doesn't the linker will give an error message---incorrect code will not be
6234 @subsection M88K Options
6235 @cindex M88k options
6237 These @samp{-m} options are defined for Motorola 88k architectures:
6242 Generate code that works well on both the m88100 and the
6247 Generate code that works best for the m88100, but that also
6252 Generate code that works best for the m88110, and may not run
6257 Obsolete option to be removed from the next revision.
6260 @item -midentify-revision
6261 @opindex midentify-revision
6262 @cindex identifying source, compiler (88k)
6263 Include an @code{ident} directive in the assembler output recording the
6264 source file name, compiler name and version, timestamp, and compilation
6267 @item -mno-underscores
6268 @opindex mno-underscores
6269 @cindex underscores, avoiding (88k)
6270 In assembler output, emit symbol names without adding an underscore
6271 character at the beginning of each name. The default is to use an
6272 underscore as prefix on each name.
6274 @item -mocs-debug-info
6275 @itemx -mno-ocs-debug-info
6276 @opindex mocs-debug-info
6277 @opindex mno-ocs-debug-info
6279 @cindex debugging, 88k OCS
6280 Include (or omit) additional debugging information (about registers used
6281 in each stack frame) as specified in the 88open Object Compatibility
6282 Standard, ``OCS''@. This extra information allows debugging of code that
6283 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
6284 Delta 88 SVr3.2 is to include this information; other 88k configurations
6285 omit this information by default.
6287 @item -mocs-frame-position
6288 @opindex mocs-frame-position
6289 @cindex register positions in frame (88k)
6290 When emitting COFF debugging information for automatic variables and
6291 parameters stored on the stack, use the offset from the canonical frame
6292 address, which is the stack pointer (register 31) on entry to the
6293 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
6294 @option{-mocs-frame-position}; other 88k configurations have the default
6295 @option{-mno-ocs-frame-position}.
6297 @item -mno-ocs-frame-position
6298 @opindex mno-ocs-frame-position
6299 @cindex register positions in frame (88k)
6300 When emitting COFF debugging information for automatic variables and
6301 parameters stored on the stack, use the offset from the frame pointer
6302 register (register 30). When this option is in effect, the frame
6303 pointer is not eliminated when debugging information is selected by the
6306 @item -moptimize-arg-area
6307 @itemx -mno-optimize-arg-area
6308 @opindex moptimize-arg-area
6309 @opindex mno-optimize-arg-area
6310 @cindex arguments in frame (88k)
6311 Control how function arguments are stored in stack frames.
6312 @option{-moptimize-arg-area} saves space by optimizing them, but this
6313 conflicts with the 88open specifications. The opposite alternative,
6314 @option{-mno-optimize-arg-area}, agrees with 88open standards. By default
6315 GCC does not optimize the argument area.
6317 @item -mshort-data-@var{num}
6318 @opindex mshort-data
6319 @cindex smaller data references (88k)
6320 @cindex r0-relative references (88k)
6321 Generate smaller data references by making them relative to @code{r0},
6322 which allows loading a value using a single instruction (rather than the
6323 usual two). You control which data references are affected by
6324 specifying @var{num} with this option. For example, if you specify
6325 @option{-mshort-data-512}, then the data references affected are those
6326 involving displacements of less than 512 bytes.
6327 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6330 @item -mserialize-volatile
6331 @opindex mserialize-volatile
6332 @itemx -mno-serialize-volatile
6333 @opindex mno-serialize-volatile
6334 @cindex sequential consistency on 88k
6335 Do, or don't, generate code to guarantee sequential consistency
6336 of volatile memory references. By default, consistency is
6339 The order of memory references made by the MC88110 processor does
6340 not always match the order of the instructions requesting those
6341 references. In particular, a load instruction may execute before
6342 a preceding store instruction. Such reordering violates
6343 sequential consistency of volatile memory references, when there
6344 are multiple processors. When consistency must be guaranteed,
6345 GCC generates special instructions, as needed, to force
6346 execution in the proper order.
6348 The MC88100 processor does not reorder memory references and so
6349 always provides sequential consistency. However, by default, GCC
6350 generates the special instructions to guarantee consistency
6351 even when you use @option{-m88100}, so that the code may be run on an
6352 MC88110 processor. If you intend to run your code only on the
6353 MC88100 processor, you may use @option{-mno-serialize-volatile}.
6355 The extra code generated to guarantee consistency may affect the
6356 performance of your application. If you know that you can safely
6357 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
6363 @cindex assembler syntax, 88k
6365 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
6366 related to System V release 4 (SVr4). This controls the following:
6370 Which variant of the assembler syntax to emit.
6372 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6373 that is used on System V release 4.
6375 @option{-msvr4} makes GCC issue additional declaration directives used in
6379 @option{-msvr4} is the default for the m88k-motorola-sysv4 and
6380 m88k-dg-dgux m88k configurations. @option{-msvr3} is the default for all
6381 other m88k configurations.
6383 @item -mversion-03.00
6384 @opindex mversion-03.00
6385 This option is obsolete, and is ignored.
6386 @c ??? which asm syntax better for GAS? option there too?
6388 @item -mno-check-zero-division
6389 @itemx -mcheck-zero-division
6390 @opindex mno-check-zero-division
6391 @opindex mcheck-zero-division
6392 @cindex zero division on 88k
6393 Do, or don't, generate code to guarantee that integer division by
6394 zero will be detected. By default, detection is guaranteed.
6396 Some models of the MC88100 processor fail to trap upon integer
6397 division by zero under certain conditions. By default, when
6398 compiling code that might be run on such a processor, GCC
6399 generates code that explicitly checks for zero-valued divisors
6400 and traps with exception number 503 when one is detected. Use of
6401 mno-check-zero-division suppresses such checking for code
6402 generated to run on an MC88100 processor.
6404 GCC assumes that the MC88110 processor correctly detects all
6405 instances of integer division by zero. When @option{-m88110} is
6406 specified, both @option{-mcheck-zero-division} and
6407 @option{-mno-check-zero-division} are ignored, and no explicit checks for
6408 zero-valued divisors are generated.
6410 @item -muse-div-instruction
6411 @opindex muse-div-instruction
6412 @cindex divide instruction, 88k
6413 Use the div instruction for signed integer division on the
6414 MC88100 processor. By default, the div instruction is not used.
6416 On the MC88100 processor the signed integer division instruction
6417 div) traps to the operating system on a negative operand. The
6418 operating system transparently completes the operation, but at a
6419 large cost in execution time. By default, when compiling code
6420 that might be run on an MC88100 processor, GCC emulates signed
6421 integer division using the unsigned integer division instruction
6422 divu), thereby avoiding the large penalty of a trap to the
6423 operating system. Such emulation has its own, smaller, execution
6424 cost in both time and space. To the extent that your code's
6425 important signed integer division operations are performed on two
6426 nonnegative operands, it may be desirable to use the div
6427 instruction directly.
6429 On the MC88110 processor the div instruction (also known as the
6430 divs instruction) processes negative operands without trapping to
6431 the operating system. When @option{-m88110} is specified,
6432 @option{-muse-div-instruction} is ignored, and the div instruction is used
6433 for signed integer division.
6435 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
6436 particular, the behavior of such a division with and without
6437 @option{-muse-div-instruction} may differ.
6439 @item -mtrap-large-shift
6440 @itemx -mhandle-large-shift
6441 @opindex mtrap-large-shift
6442 @opindex mhandle-large-shift
6443 @cindex bit shift overflow (88k)
6444 @cindex large bit shifts (88k)
6445 Include code to detect bit-shifts of more than 31 bits; respectively,
6446 trap such shifts or emit code to handle them properly. By default GCC
6447 makes no special provision for large bit shifts.
6449 @item -mwarn-passed-structs
6450 @opindex mwarn-passed-structs
6451 @cindex structure passing (88k)
6452 Warn when a function passes a struct as an argument or result.
6453 Structure-passing conventions have changed during the evolution of the C
6454 language, and are often the source of portability problems. By default,
6455 GCC issues no such warning.
6458 @node RS/6000 and PowerPC Options
6459 @subsection IBM RS/6000 and PowerPC Options
6460 @cindex RS/6000 and PowerPC Options
6461 @cindex IBM RS/6000 and PowerPC Options
6463 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6471 @itemx -mpowerpc-gpopt
6472 @itemx -mno-powerpc-gpopt
6473 @itemx -mpowerpc-gfxopt
6474 @itemx -mno-powerpc-gfxopt
6476 @itemx -mno-powerpc64
6482 @opindex mno-powerpc
6483 @opindex mpowerpc-gpopt
6484 @opindex mno-powerpc-gpopt
6485 @opindex mpowerpc-gfxopt
6486 @opindex mno-powerpc-gfxopt
6488 @opindex mno-powerpc64
6489 GCC supports two related instruction set architectures for the
6490 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6491 instructions supported by the @samp{rios} chip set used in the original
6492 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6493 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6494 the IBM 4xx microprocessors.
6496 Neither architecture is a subset of the other. However there is a
6497 large common subset of instructions supported by both. An MQ
6498 register is included in processors supporting the POWER architecture.
6500 You use these options to specify which instructions are available on the
6501 processor you are using. The default value of these options is
6502 determined when configuring GCC@. Specifying the
6503 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6504 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6505 rather than the options listed above.
6507 The @option{-mpower} option allows GCC to generate instructions that
6508 are found only in the POWER architecture and to use the MQ register.
6509 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
6510 to generate instructions that are present in the POWER2 architecture but
6511 not the original POWER architecture.
6513 The @option{-mpowerpc} option allows GCC to generate instructions that
6514 are found only in the 32-bit subset of the PowerPC architecture.
6515 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
6516 GCC to use the optional PowerPC architecture instructions in the
6517 General Purpose group, including floating-point square root. Specifying
6518 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
6519 use the optional PowerPC architecture instructions in the Graphics
6520 group, including floating-point select.
6522 The @option{-mpowerpc64} option allows GCC to generate the additional
6523 64-bit instructions that are found in the full PowerPC64 architecture
6524 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6525 @option{-mno-powerpc64}.
6527 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
6528 will use only the instructions in the common subset of both
6529 architectures plus some special AIX common-mode calls, and will not use
6530 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
6531 permits GCC to use any instruction from either architecture and to
6532 allow use of the MQ register; specify this for the Motorola MPC601.
6534 @item -mnew-mnemonics
6535 @itemx -mold-mnemonics
6536 @opindex mnew-mnemonics
6537 @opindex mold-mnemonics
6538 Select which mnemonics to use in the generated assembler code.
6539 @option{-mnew-mnemonics} requests output that uses the assembler mnemonics
6540 defined for the PowerPC architecture, while @option{-mold-mnemonics}
6541 requests the assembler mnemonics defined for the POWER architecture.
6542 Instructions defined in only one architecture have only one mnemonic;
6543 GCC uses that mnemonic irrespective of which of these options is
6546 GCC defaults to the mnemonics appropriate for the architecture in
6547 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
6548 value of these option. Unless you are building a cross-compiler, you
6549 should normally not specify either @option{-mnew-mnemonics} or
6550 @option{-mold-mnemonics}, but should instead accept the default.
6552 @item -mcpu=@var{cpu_type}
6554 Set architecture type, register usage, choice of mnemonics, and
6555 instruction scheduling parameters for machine type @var{cpu_type}.
6556 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
6557 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
6558 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
6559 @samp{630}, @samp{740}, @samp{750}, @samp{power}, @samp{power2},
6560 @samp{powerpc}, @samp{403}, @samp{505}, @samp{801}, @samp{821},
6561 @samp{823}, and @samp{860} and @samp{common}. @option{-mcpu=power},
6562 @option{-mcpu=power2}, @option{-mcpu=powerpc}, and @option{-mcpu=powerpc64}
6563 specify generic POWER, POWER2, pure 32-bit PowerPC (i.e., not MPC601),
6564 and 64-bit PowerPC architecture machine types, with an appropriate,
6565 generic processor model assumed for scheduling purposes.
6567 Specifying any of the following options:
6568 @option{-mcpu=rios1}, @option{-mcpu=rios2}, @option{-mcpu=rsc},
6569 @option{-mcpu=power}, or @option{-mcpu=power2}
6570 enables the @option{-mpower} option and disables the @option{-mpowerpc} option;
6571 @option{-mcpu=601} enables both the @option{-mpower} and @option{-mpowerpc} options.
6572 All of @option{-mcpu=rs64a}, @option{-mcpu=602}, @option{-mcpu=603},
6573 @option{-mcpu=603e}, @option{-mcpu=604}, @option{-mcpu=620}, @option{-mcpu=630},
6574 @option{-mcpu=740}, and @option{-mcpu=750}
6575 enable the @option{-mpowerpc} option and disable the @option{-mpower} option.
6576 Exactly similarly, all of @option{-mcpu=403},
6577 @option{-mcpu=505}, @option{-mcpu=821}, @option{-mcpu=860} and @option{-mcpu=powerpc}
6578 enable the @option{-mpowerpc} option and disable the @option{-mpower} option.
6579 @option{-mcpu=common} disables both the
6580 @option{-mpower} and @option{-mpowerpc} options.
6582 AIX versions 4 or greater selects @option{-mcpu=common} by default, so
6583 that code will operate on all members of the RS/6000 POWER and PowerPC
6584 families. In that case, GCC will use only the instructions in the
6585 common subset of both architectures plus some special AIX common-mode
6586 calls, and will not use the MQ register. GCC assumes a generic
6587 processor model for scheduling purposes.
6589 Specifying any of the options @option{-mcpu=rios1}, @option{-mcpu=rios2},
6590 @option{-mcpu=rsc}, @option{-mcpu=power}, or @option{-mcpu=power2} also
6591 disables the @samp{new-mnemonics} option. Specifying @option{-mcpu=601},
6592 @option{-mcpu=602}, @option{-mcpu=603}, @option{-mcpu=603e}, @option{-mcpu=604},
6593 @option{-mcpu=620}, @option{-mcpu=630}, @option{-mcpu=403}, @option{-mcpu=505},
6594 @option{-mcpu=821}, @option{-mcpu=860} or @option{-mcpu=powerpc} also enables
6595 the @samp{new-mnemonics} option.
6597 Specifying @option{-mcpu=403}, @option{-mcpu=821}, or @option{-mcpu=860} also
6598 enables the @option{-msoft-float} option.
6600 @item -mtune=@var{cpu_type}
6602 Set the instruction scheduling parameters for machine type
6603 @var{cpu_type}, but do not set the architecture type, register usage,
6604 choice of mnemonics like @option{-mcpu=@var{cpu_type}} would. The same
6605 values for @var{cpu_type} are used for @option{-mtune=@var{cpu_type}} as
6606 for @option{-mcpu=@var{cpu_type}}. The @option{-mtune=@var{cpu_type}}
6607 option overrides the @option{-mcpu=@var{cpu_type}} option in terms of
6608 instruction scheduling parameters.
6611 @itemx -mno-fp-in-toc
6612 @itemx -mno-sum-in-toc
6613 @itemx -mminimal-toc
6615 @opindex mno-fp-in-toc
6616 @opindex mno-sum-in-toc
6617 @opindex mminimal-toc
6618 Modify generation of the TOC (Table Of Contents), which is created for
6619 every executable file. The @option{-mfull-toc} option is selected by
6620 default. In that case, GCC will allocate at least one TOC entry for
6621 each unique non-automatic variable reference in your program. GCC
6622 will also place floating-point constants in the TOC@. However, only
6623 16,384 entries are available in the TOC@.
6625 If you receive a linker error message that saying you have overflowed
6626 the available TOC space, you can reduce the amount of TOC space used
6627 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
6628 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
6629 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
6630 generate code to calculate the sum of an address and a constant at
6631 run-time instead of putting that sum into the TOC@. You may specify one
6632 or both of these options. Each causes GCC to produce very slightly
6633 slower and larger code at the expense of conserving TOC space.
6635 If you still run out of space in the TOC even when you specify both of
6636 these options, specify @option{-mminimal-toc} instead. This option causes
6637 GCC to make only one TOC entry for every file. When you specify this
6638 option, GCC will produce code that is slower and larger but which
6639 uses extremely little TOC space. You may wish to use this option
6640 only on files that contain less frequently executed code.
6646 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
6647 @code{long} type, and the infrastructure needed to support them.
6648 Specifying @option{-maix64} implies @option{-mpowerpc64} and
6649 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
6650 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
6655 @opindex mno-xl-call
6656 On AIX, pass floating-point arguments to prototyped functions beyond the
6657 register save area (RSA) on the stack in addition to argument FPRs. The
6658 AIX calling convention was extended but not initially documented to
6659 handle an obscure K&R C case of calling a function that takes the
6660 address of its arguments with fewer arguments than declared. AIX XL
6661 compilers access floating point arguments which do not fit in the
6662 RSA from the stack when a subroutine is compiled without
6663 optimization. Because always storing floating-point arguments on the
6664 stack is inefficient and rarely needed, this option is not enabled by
6665 default and only is necessary when calling subroutines compiled by AIX
6666 XL compilers without optimization.
6670 Support @dfn{AIX Threads}. Link an application written to use
6671 @dfn{pthreads} with special libraries and startup code to enable the
6676 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
6677 application written to use message passing with special startup code to
6678 enable the application to run. The system must have PE installed in the
6679 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
6680 must be overridden with the @option{-specs=} option to specify the
6681 appropriate directory location. The Parallel Environment does not
6682 support threads, so the @option{-mpe} option and the @option{-mthreads}
6683 option are incompatible.
6687 @opindex msoft-float
6688 @opindex mhard-float
6689 Generate code that does not use (uses) the floating-point register set.
6690 Software floating point emulation is provided if you use the
6691 @option{-msoft-float} option, and pass the option to GCC when linking.
6694 @itemx -mno-multiple
6696 @opindex mno-multiple
6697 Generate code that uses (does not use) the load multiple word
6698 instructions and the store multiple word instructions. These
6699 instructions are generated by default on POWER systems, and not
6700 generated on PowerPC systems. Do not use @option{-mmultiple} on little
6701 endian PowerPC systems, since those instructions do not work when the
6702 processor is in little endian mode. The exceptions are PPC740 and
6703 PPC750 which permit the instructions usage in little endian mode.
6709 Generate code that uses (does not use) the load string instructions
6710 and the store string word instructions to save multiple registers and
6711 do small block moves. These instructions are generated by default on
6712 POWER systems, and not generated on PowerPC systems. Do not use
6713 @option{-mstring} on little endian PowerPC systems, since those
6714 instructions do not work when the processor is in little endian mode.
6715 The exceptions are PPC740 and PPC750 which permit the instructions
6716 usage in little endian mode.
6722 Generate code that uses (does not use) the load or store instructions
6723 that update the base register to the address of the calculated memory
6724 location. These instructions are generated by default. If you use
6725 @option{-mno-update}, there is a small window between the time that the
6726 stack pointer is updated and the address of the previous frame is
6727 stored, which means code that walks the stack frame across interrupts or
6728 signals may get corrupted data.
6731 @itemx -mno-fused-madd
6732 @opindex mfused-madd
6733 @opindex mno-fused-madd
6734 Generate code that uses (does not use) the floating point multiply and
6735 accumulate instructions. These instructions are generated by default if
6736 hardware floating is used.
6738 @item -mno-bit-align
6740 @opindex mno-bit-align
6742 On System V.4 and embedded PowerPC systems do not (do) force structures
6743 and unions that contain bit-fields to be aligned to the base type of the
6746 For example, by default a structure containing nothing but 8
6747 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
6748 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
6749 the structure would be aligned to a 1 byte boundary and be one byte in
6752 @item -mno-strict-align
6753 @itemx -mstrict-align
6754 @opindex mno-strict-align
6755 @opindex mstrict-align
6756 On System V.4 and embedded PowerPC systems do not (do) assume that
6757 unaligned memory references will be handled by the system.
6760 @itemx -mno-relocatable
6761 @opindex mrelocatable
6762 @opindex mno-relocatable
6763 On embedded PowerPC systems generate code that allows (does not allow)
6764 the program to be relocated to a different address at runtime. If you
6765 use @option{-mrelocatable} on any module, all objects linked together must
6766 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
6768 @item -mrelocatable-lib
6769 @itemx -mno-relocatable-lib
6770 @opindex mrelocatable-lib
6771 @opindex mno-relocatable-lib
6772 On embedded PowerPC systems generate code that allows (does not allow)
6773 the program to be relocated to a different address at runtime. Modules
6774 compiled with @option{-mrelocatable-lib} can be linked with either modules
6775 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
6776 with modules compiled with the @option{-mrelocatable} options.
6782 On System V.4 and embedded PowerPC systems do not (do) assume that
6783 register 2 contains a pointer to a global area pointing to the addresses
6784 used in the program.
6787 @itemx -mlittle-endian
6789 @opindex mlittle-endian
6790 On System V.4 and embedded PowerPC systems compile code for the
6791 processor in little endian mode. The @option{-mlittle-endian} option is
6792 the same as @option{-mlittle}.
6797 @opindex mbig-endian
6798 On System V.4 and embedded PowerPC systems compile code for the
6799 processor in big endian mode. The @option{-mbig-endian} option is
6800 the same as @option{-mbig}.
6804 On System V.4 and embedded PowerPC systems compile code using calling
6805 conventions that adheres to the March 1995 draft of the System V
6806 Application Binary Interface, PowerPC processor supplement. This is the
6807 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
6809 @item -mcall-sysv-eabi
6810 @opindex mcall-sysv-eabi
6811 Specify both @option{-mcall-sysv} and @option{-meabi} options.
6813 @item -mcall-sysv-noeabi
6814 @opindex mcall-sysv-noeabi
6815 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
6819 On System V.4 and embedded PowerPC systems compile code using calling
6820 conventions that are similar to those used on AIX@. This is the
6821 default if you configured GCC using @samp{powerpc-*-eabiaix}.
6823 @item -mcall-solaris
6824 @opindex mcall-solaris
6825 On System V.4 and embedded PowerPC systems compile code for the Solaris
6829 @opindex mcall-linux
6830 On System V.4 and embedded PowerPC systems compile code for the
6831 Linux-based GNU system.
6834 @itemx -mno-prototype
6836 @opindex mno-prototype
6837 On System V.4 and embedded PowerPC systems assume that all calls to
6838 variable argument functions are properly prototyped. Otherwise, the
6839 compiler must insert an instruction before every non prototyped call to
6840 set or clear bit 6 of the condition code register (@var{CR}) to
6841 indicate whether floating point values were passed in the floating point
6842 registers in case the function takes a variable arguments. With
6843 @option{-mprototype}, only calls to prototyped variable argument functions
6844 will set or clear the bit.
6848 On embedded PowerPC systems, assume that the startup module is called
6849 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
6850 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
6855 On embedded PowerPC systems, assume that the startup module is called
6856 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
6861 On embedded PowerPC systems, assume that the startup module is called
6862 @file{crt0.o} and the standard C libraries are @file{libads.a} and
6866 @opindex myellowknife
6867 On embedded PowerPC systems, assume that the startup module is called
6868 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
6873 On System V.4 and embedded PowerPC systems, specify that you are
6874 compiling for a VxWorks system.
6878 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
6879 header to indicate that @samp{eabi} extended relocations are used.
6885 On System V.4 and embedded PowerPC systems do (do not) adhere to the
6886 Embedded Applications Binary Interface (eabi) which is a set of
6887 modifications to the System V.4 specifications. Selecting @option{-meabi}
6888 means that the stack is aligned to an 8 byte boundary, a function
6889 @code{__eabi} is called to from @code{main} to set up the eabi
6890 environment, and the @option{-msdata} option can use both @code{r2} and
6891 @code{r13} to point to two separate small data areas. Selecting
6892 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
6893 do not call an initialization function from @code{main}, and the
6894 @option{-msdata} option will only use @code{r13} to point to a single
6895 small data area. The @option{-meabi} option is on by default if you
6896 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
6899 @opindex msdata=eabi
6900 On System V.4 and embedded PowerPC systems, put small initialized
6901 @code{const} global and static data in the @samp{.sdata2} section, which
6902 is pointed to by register @code{r2}. Put small initialized
6903 non-@code{const} global and static data in the @samp{.sdata} section,
6904 which is pointed to by register @code{r13}. Put small uninitialized
6905 global and static data in the @samp{.sbss} section, which is adjacent to
6906 the @samp{.sdata} section. The @option{-msdata=eabi} option is
6907 incompatible with the @option{-mrelocatable} option. The
6908 @option{-msdata=eabi} option also sets the @option{-memb} option.
6911 @opindex msdata=sysv
6912 On System V.4 and embedded PowerPC systems, put small global and static
6913 data in the @samp{.sdata} section, which is pointed to by register
6914 @code{r13}. Put small uninitialized global and static data in the
6915 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
6916 The @option{-msdata=sysv} option is incompatible with the
6917 @option{-mrelocatable} option.
6919 @item -msdata=default
6921 @opindex msdata=default
6923 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
6924 compile code the same as @option{-msdata=eabi}, otherwise compile code the
6925 same as @option{-msdata=sysv}.
6928 @opindex msdata-data
6929 On System V.4 and embedded PowerPC systems, put small global and static
6930 data in the @samp{.sdata} section. Put small uninitialized global and
6931 static data in the @samp{.sbss} section. Do not use register @code{r13}
6932 to address small data however. This is the default behavior unless
6933 other @option{-msdata} options are used.
6937 @opindex msdata=none
6939 On embedded PowerPC systems, put all initialized global and static data
6940 in the @samp{.data} section, and all uninitialized data in the
6941 @samp{.bss} section.
6945 @cindex smaller data references (PowerPC)
6946 @cindex .sdata/.sdata2 references (PowerPC)
6947 On embedded PowerPC systems, put global and static items less than or
6948 equal to @var{num} bytes into the small data or bss sections instead of
6949 the normal data or bss section. By default, @var{num} is 8. The
6950 @option{-G @var{num}} switch is also passed to the linker.
6951 All modules should be compiled with the same @option{-G @var{num}} value.
6954 @itemx -mno-regnames
6956 @opindex mno-regnames
6957 On System V.4 and embedded PowerPC systems do (do not) emit register
6958 names in the assembly language output using symbolic forms.
6963 @subsection IBM RT Options
6965 @cindex IBM RT options
6967 These @samp{-m} options are defined for the IBM RT PC:
6971 @opindex min-line-mul
6972 Use an in-line code sequence for integer multiplies. This is the
6975 @item -mcall-lib-mul
6976 @opindex mcall-lib-mul
6977 Call @code{lmul$$} for integer multiples.
6979 @item -mfull-fp-blocks
6980 @opindex mfull-fp-blocks
6981 Generate full-size floating point data blocks, including the minimum
6982 amount of scratch space recommended by IBM@. This is the default.
6984 @item -mminimum-fp-blocks
6985 @opindex mminimum-fp-blocks
6986 Do not include extra scratch space in floating point data blocks. This
6987 results in smaller code, but slower execution, since scratch space must
6988 be allocated dynamically.
6990 @cindex @file{varargs.h} and RT PC
6991 @cindex @file{stdarg.h} and RT PC
6992 @item -mfp-arg-in-fpregs
6993 @opindex mfp-arg-in-fpregs
6994 Use a calling sequence incompatible with the IBM calling convention in
6995 which floating point arguments are passed in floating point registers.
6996 Note that @code{varargs.h} and @code{stdarg.h} will not work with
6997 floating point operands if this option is specified.
6999 @item -mfp-arg-in-gregs
7000 @opindex mfp-arg-in-gregs
7001 Use the normal calling convention for floating point arguments. This is
7004 @item -mhc-struct-return
7005 @opindex mhc-struct-return
7006 Return structures of more than one word in memory, rather than in a
7007 register. This provides compatibility with the MetaWare HighC (hc)
7008 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7009 with the Portable C Compiler (pcc).
7011 @item -mnohc-struct-return
7012 @opindex mnohc-struct-return
7013 Return some structures of more than one word in registers, when
7014 convenient. This is the default. For compatibility with the
7015 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7016 option @option{-mhc-struct-return}.
7020 @subsection MIPS Options
7021 @cindex MIPS options
7023 These @samp{-m} options are defined for the MIPS family of computers:
7026 @item -mcpu=@var{cpu-type}
7028 Assume the defaults for the machine type @var{cpu-type} when scheduling
7029 instructions. The choices for @var{cpu-type} are @samp{r2000}, @samp{r3000},
7030 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
7031 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
7032 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
7033 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
7034 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
7035 @var{cpu-type} will schedule things appropriately for that particular
7036 chip, the compiler will not generate any code that does not meet level 1
7037 of the MIPS ISA (instruction set architecture) without a @option{-mipsX}
7038 or @option{-mabi} switch being used.
7042 Issue instructions from level 1 of the MIPS ISA@. This is the default.
7043 @samp{r3000} is the default @var{cpu-type} at this ISA level.
7047 Issue instructions from level 2 of the MIPS ISA (branch likely, square
7048 root instructions). @samp{r6000} is the default @var{cpu-type} at this
7053 Issue instructions from level 3 of the MIPS ISA (64-bit instructions).
7054 @samp{r4000} is the default @var{cpu-type} at this ISA level.
7058 Issue instructions from level 4 of the MIPS ISA (conditional move,
7059 prefetch, enhanced FPU instructions). @samp{r8000} is the default
7060 @var{cpu-type} at this ISA level.
7064 Assume that 32 32-bit floating point registers are available. This is
7069 Assume that 32 64-bit floating point registers are available. This is
7070 the default when the @option{-mips3} option is used.
7074 Assume that 32 32-bit general purpose registers are available. This is
7079 Assume that 32 64-bit general purpose registers are available. This is
7080 the default when the @option{-mips3} option is used.
7084 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
7085 explanation of the default, and the width of pointers.
7089 Force long types to be 64 bits wide. See @option{-mlong32} for an
7090 explanation of the default, and the width of pointers.
7094 Force long, int, and pointer types to be 32 bits wide.
7096 If none of @option{-mlong32}, @option{-mlong64}, or @option{-mint64} are set,
7097 the size of ints, longs, and pointers depends on the ABI and ISA chosen.
7098 For @option{-mabi=32}, and @option{-mabi=n32}, ints and longs are 32 bits
7099 wide. For @option{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
7100 For @option{-mabi=eabi} and either @option{-mips1} or @option{-mips2}, ints
7101 and longs are 32 bits wide. For @option{-mabi=eabi} and higher ISAs, ints
7102 are 32 bits, and longs are 64 bits wide. The width of pointer types is
7103 the smaller of the width of longs or the width of general purpose
7104 registers (which in turn depends on the ISA)@.
7116 Generate code for the indicated ABI@. The default instruction level is
7117 @option{-mips1} for @samp{32}, @option{-mips3} for @samp{n32}, and
7118 @option{-mips4} otherwise. Conversely, with @option{-mips1} or
7119 @option{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
7124 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
7125 add normal debug information. This is the default for all
7126 platforms except for the OSF/1 reference platform, using the OSF/rose
7127 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
7128 switches are used, the @file{mips-tfile} program will encapsulate the
7129 stabs within MIPS ECOFF@.
7133 Generate code for the GNU assembler. This is the default on the OSF/1
7134 reference platform, using the OSF/rose object format. Also, this is
7135 the default if the configure option @option{--with-gnu-as} is used.
7137 @item -msplit-addresses
7138 @itemx -mno-split-addresses
7139 @opindex msplit-addresses
7140 @opindex mno-split-addresses
7141 Generate code to load the high and low parts of address constants separately.
7142 This allows GCC to optimize away redundant loads of the high order
7143 bits of addresses. This optimization requires GNU as and GNU ld.
7144 This optimization is enabled by default for some embedded targets where
7145 GNU as and GNU ld are standard.
7151 The @option{-mrnames} switch says to output code using the MIPS software
7152 names for the registers, instead of the hardware names (ie, @var{a0}
7153 instead of @var{$4}). The only known assembler that supports this option
7154 is the Algorithmics assembler.
7160 The @option{-mgpopt} switch says to write all of the data declarations
7161 before the instructions in the text section, this allows the MIPS
7162 assembler to generate one word memory references instead of using two
7163 words for short global or static data items. This is on by default if
7164 optimization is selected.
7170 For each non-inline function processed, the @option{-mstats} switch
7171 causes the compiler to emit one line to the standard error file to
7172 print statistics about the program (number of registers saved, stack
7179 The @option{-mmemcpy} switch makes all block moves call the appropriate
7180 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
7181 generating inline code.
7184 @itemx -mno-mips-tfile
7185 @opindex mmips-tfile
7186 @opindex mno-mips-tfile
7187 The @option{-mno-mips-tfile} switch causes the compiler not
7188 postprocess the object file with the @file{mips-tfile} program,
7189 after the MIPS assembler has generated it to add debug support. If
7190 @file{mips-tfile} is not run, then no local variables will be
7191 available to the debugger. In addition, @file{stage2} and
7192 @file{stage3} objects will have the temporary file names passed to the
7193 assembler embedded in the object file, which means the objects will
7194 not compare the same. The @option{-mno-mips-tfile} switch should only
7195 be used when there are bugs in the @file{mips-tfile} program that
7196 prevents compilation.
7199 @opindex msoft-float
7200 Generate output containing library calls for floating point.
7201 @strong{Warning:} the requisite libraries are not part of GCC@.
7202 Normally the facilities of the machine's usual C compiler are used, but
7203 this can't be done directly in cross-compilation. You must make your
7204 own arrangements to provide suitable library functions for
7208 @opindex mhard-float
7209 Generate output containing floating point instructions. This is the
7210 default if you use the unmodified sources.
7213 @itemx -mno-abicalls
7215 @opindex mno-abicalls
7216 Emit (or do not emit) the pseudo operations @samp{.abicalls},
7217 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
7218 position independent code.
7221 @itemx -mno-long-calls
7222 @opindex mlong-calls
7223 @opindex mno-long-calls
7224 Do all calls with the @samp{JALR} instruction, which requires
7225 loading up a function's address into a register before the call.
7226 You need to use this switch, if you call outside of the current
7227 512 megabyte segment to functions that are not through pointers.
7230 @itemx -mno-half-pic
7232 @opindex mno-half-pic
7233 Put pointers to extern references into the data section and load them
7234 up, rather than put the references in the text section.
7236 @item -membedded-pic
7237 @itemx -mno-embedded-pic
7238 @opindex membedded-pic
7239 @opindex mno-embedded-pic
7240 Generate PIC code suitable for some embedded systems. All calls are
7241 made using PC relative address, and all data is addressed using the $gp
7242 register. No more than 65536 bytes of global data may be used. This
7243 requires GNU as and GNU ld which do most of the work. This currently
7244 only works on targets which use ECOFF; it does not work with ELF@.
7246 @item -membedded-data
7247 @itemx -mno-embedded-data
7248 @opindex membedded-data
7249 @opindex mno-embedded-data
7250 Allocate variables to the read-only data section first if possible, then
7251 next in the small data section if possible, otherwise in data. This gives
7252 slightly slower code than the default, but reduces the amount of RAM required
7253 when executing, and thus may be preferred for some embedded systems.
7255 @item -muninit-const-in-rodata
7256 @itemx -mno-uninit-const-in-rodata
7257 @opindex muninit-const-in-rodata
7258 @opindex mno-uninit-const-in-rodata
7259 When used together with @option{-membedded-data}, it will always store uninitialized
7260 const variables in the read-only data section.
7262 @item -msingle-float
7263 @itemx -mdouble-float
7264 @opindex msingle-float
7265 @opindex mdouble-float
7266 The @option{-msingle-float} switch tells gcc to assume that the floating
7267 point coprocessor only supports single precision operations, as on the
7268 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
7269 double precision operations. This is the default.
7275 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
7276 as on the @samp{r4650} chip.
7280 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
7281 @option{-mcpu=r4650}.
7287 Enable 16-bit instructions.
7291 Use the entry and exit pseudo ops. This option can only be used with
7296 Compile code for the processor in little endian mode.
7297 The requisite libraries are assumed to exist.
7301 Compile code for the processor in big endian mode.
7302 The requisite libraries are assumed to exist.
7306 @cindex smaller data references (MIPS)
7307 @cindex gp-relative references (MIPS)
7308 Put global and static items less than or equal to @var{num} bytes into
7309 the small data or bss sections instead of the normal data or bss
7310 section. This allows the assembler to emit one word memory reference
7311 instructions based on the global pointer (@var{gp} or @var{$28}),
7312 instead of the normal two words used. By default, @var{num} is 8 when
7313 the MIPS assembler is used, and 0 when the GNU assembler is used. The
7314 @option{-G @var{num}} switch is also passed to the assembler and linker.
7315 All modules should be compiled with the same @option{-G @var{num}}
7320 Tell the MIPS assembler to not run its preprocessor over user
7321 assembler files (with a @samp{.s} suffix) when assembling them.
7325 Pass an option to gas which will cause nops to be inserted if
7326 the read of the destination register of an mfhi or mflo instruction
7327 occurs in the following two instructions.
7331 Do not include the default crt0.
7335 These options are defined by the macro
7336 @code{TARGET_SWITCHES} in the machine description. The default for the
7337 options is also defined by that macro, which enables you to change the
7342 @subsection Intel 386 Options
7343 @cindex i386 Options
7344 @cindex Intel 386 Options
7346 These @samp{-m} options are defined for the i386 family of computers:
7349 @item -mcpu=@var{cpu-type}
7351 Assume the defaults for the machine type @var{cpu-type} when scheduling
7352 instructions. The choices for @var{cpu-type} are @samp{i386},
7353 @samp{i486}, @samp{i586}, @samp{i686}, @samp{pentium},
7354 @samp{pentiumpro}, @samp{pentium4}, @samp{k6}, and @samp{athlon}
7356 While picking a specific @var{cpu-type} will schedule things appropriately
7357 for that particular chip, the compiler will not generate any code that
7358 does not run on the i386 without the @option{-march=@var{cpu-type}} option
7359 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
7360 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
7361 AMD chips as opposed to the Intel ones.
7363 @item -march=@var{cpu-type}
7365 Generate instructions for the machine type @var{cpu-type}. The choices
7366 for @var{cpu-type} are the same as for @option{-mcpu}. Moreover,
7367 specifying @option{-march=@var{cpu-type}} implies @option{-mcpu=@var{cpu-type}}.
7376 @opindex mpentiumpro
7377 Synonyms for @option{-mcpu=i386}, @option{-mcpu=i486}, @option{-mcpu=pentium}, and @option{-mcpu=pentiumpro}
7378 respectively. These synonyms are deprecated.
7380 @item -mintel-syntax
7381 @opindex mintel-syntax
7382 Emit assembly using Intel syntax opcodes instead of AT&T syntax.
7387 @opindex mno-ieee-fp
7388 Control whether or not the compiler uses IEEE floating point
7389 comparisons. These handle correctly the case where the result of a
7390 comparison is unordered.
7393 @opindex msoft-float
7394 Generate output containing library calls for floating point.
7395 @strong{Warning:} the requisite libraries are not part of GCC@.
7396 Normally the facilities of the machine's usual C compiler are used, but
7397 this can't be done directly in cross-compilation. You must make your
7398 own arrangements to provide suitable library functions for
7401 On machines where a function returns floating point results in the 80387
7402 register stack, some floating point opcodes may be emitted even if
7403 @option{-msoft-float} is used.
7405 @item -mno-fp-ret-in-387
7406 @opindex mno-fp-ret-in-387
7407 Do not use the FPU registers for return values of functions.
7409 The usual calling convention has functions return values of types
7410 @code{float} and @code{double} in an FPU register, even if there
7411 is no FPU@. The idea is that the operating system should emulate
7414 The option @option{-mno-fp-ret-in-387} causes such values to be returned
7415 in ordinary CPU registers instead.
7417 @item -mno-fancy-math-387
7418 @opindex mno-fancy-math-387
7419 Some 387 emulators do not support the @code{sin}, @code{cos} and
7420 @code{sqrt} instructions for the 387. Specify this option to avoid
7421 generating those instructions. This option is the default on FreeBSD@.
7422 As of revision 2.6.1, these instructions are not generated unless you
7423 also use the @option{-funsafe-math-optimizations} switch.
7425 @item -malign-double
7426 @itemx -mno-align-double
7427 @opindex malign-double
7428 @opindex mno-align-double
7429 Control whether GCC aligns @code{double}, @code{long double}, and
7430 @code{long long} variables on a two word boundary or a one word
7431 boundary. Aligning @code{double} variables on a two word boundary will
7432 produce code that runs somewhat faster on a @samp{Pentium} at the
7433 expense of more memory.
7435 @item -m128bit-long-double
7436 @itemx -m128bit-long-double
7437 @opindex m128bit-long-double
7438 @opindex m128bit-long-double
7439 Control the size of @code{long double} type. i386 application binary interface
7440 specify the size to be 12 bytes, while modern architectures (Pentium and newer)
7441 prefer @code{long double} aligned to 8 or 16 byte boundary. This is
7442 impossible to reach with 12 byte long doubles in the array accesses.
7444 @strong{Warning:} if you use the @option{-m128bit-long-double} switch, the
7445 structures and arrays containing @code{long double} will change their size as
7446 well as function calling convention for function taking @code{long double}
7449 @item -m96bit-long-double
7450 @itemx -m96bit-long-double
7451 @opindex m96bit-long-double
7452 @opindex m96bit-long-double
7453 Set the size of @code{long double} to 96 bits as required by the i386
7454 application binary interface. This is the default.
7457 @itemx -mno-svr3-shlib
7458 @opindex msvr3-shlib
7459 @opindex mno-svr3-shlib
7460 Control whether GCC places uninitialized locals into @code{bss} or
7461 @code{data}. @option{-msvr3-shlib} places these locals into @code{bss}.
7462 These options are meaningful only on System V Release 3.
7464 @item -mno-wide-multiply
7465 @itemx -mwide-multiply
7466 @opindex mno-wide-multiply
7467 @opindex mwide-multiply
7468 Control whether GCC uses the @code{mul} and @code{imul} that produce
7469 64-bit results in @code{eax:edx} from 32-bit operands to do @code{long
7470 long} multiplies and 32-bit division by constants.
7474 Use a different function-calling convention, in which functions that
7475 take a fixed number of arguments return with the @code{ret} @var{num}
7476 instruction, which pops their arguments while returning. This saves one
7477 instruction in the caller since there is no need to pop the arguments
7480 You can specify that an individual function is called with this calling
7481 sequence with the function attribute @samp{stdcall}. You can also
7482 override the @option{-mrtd} option by using the function attribute
7483 @samp{cdecl}. @xref{Function Attributes}.
7485 @strong{Warning:} this calling convention is incompatible with the one
7486 normally used on Unix, so you cannot use it if you need to call
7487 libraries compiled with the Unix compiler.
7489 Also, you must provide function prototypes for all functions that
7490 take variable numbers of arguments (including @code{printf});
7491 otherwise incorrect code will be generated for calls to those
7494 In addition, seriously incorrect code will result if you call a
7495 function with too many arguments. (Normally, extra arguments are
7496 harmlessly ignored.)
7498 @item -mregparm=@var{num}
7500 Control how many registers are used to pass integer arguments. By
7501 default, no registers are used to pass arguments, and at most 3
7502 registers can be used. You can control this behavior for a specific
7503 function by using the function attribute @samp{regparm}.
7504 @xref{Function Attributes}.
7506 @strong{Warning:} if you use this switch, and
7507 @var{num} is nonzero, then you must build all modules with the same
7508 value, including any libraries. This includes the system libraries and
7511 @item -mpreferred-stack-boundary=@var{num}
7512 @opindex mpreferred-stack-boundary
7513 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
7514 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
7515 the default is 4 (16 bytes or 128 bits).
7517 The stack is required to be aligned on a 4 byte boundary. On Pentium
7518 and PentiumPro, @code{double} and @code{long double} values should be
7519 aligned to an 8 byte boundary (see @option{-malign-double}) or suffer
7520 significant run time performance penalties. On Pentium III, the
7521 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
7522 penalties if it is not 16 byte aligned.
7524 To ensure proper alignment of this values on the stack, the stack boundary
7525 must be as aligned as that required by any value stored on the stack.
7526 Further, every function must be generated such that it keeps the stack
7527 aligned. Thus calling a function compiled with a higher preferred
7528 stack boundary from a function compiled with a lower preferred stack
7529 boundary will most likely misalign the stack. It is recommended that
7530 libraries that use callbacks always use the default setting.
7532 This extra alignment does consume extra stack space. Code that is sensitive
7533 to stack space usage, such as embedded systems and operating system kernels,
7534 may want to reduce the preferred alignment to
7535 @option{-mpreferred-stack-boundary=2}.
7539 Use PUSH operations to store outgoing parameters. This method is shorter
7540 and usually equally fast as method using SUB/MOV operations and is enabled
7541 by default. In some cases disabling it may improve performance because of
7542 improved scheduling and reduced dependencies.
7544 @item -maccumulate-outgoing-args
7545 @opindex maccumulate-outgoing-args
7546 If enabled, the maximum amount of space required for outgoing arguments will be
7547 computed in the function prologue. This in faster on most modern CPUs
7548 because of reduced dependencies, improved scheduling and reduced stack usage
7549 when preferred stack boundary is not equal to 2. The drawback is a notable
7550 increase in code size. This switch implies @option{-mno-push-args}.
7554 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
7555 on thread-safe exception handling must compile and link all code with the
7556 @option{-mthreads} option. When compiling, @option{-mthreads} defines
7557 @option{-D_MT}; when linking, it links in a special thread helper library
7558 @option{-lmingwthrd} which cleans up per thread exception handling data.
7560 @item -mno-align-stringops
7561 @opindex mno-align-stringops
7562 Do not align destination of inlined string operations. This switch reduces
7563 code size and improves performance in case the destination is already aligned,
7564 but gcc don't know about it.
7566 @item -minline-all-stringops
7567 @opindex minline-all-stringops
7568 By default GCC inlines string operations only when destination is known to be
7569 aligned at least to 4 byte boundary. This enables more inlining, increase code
7570 size, but may improve performance of code that depends on fast memcpy, strlen
7571 and memset for short lengths.
7573 @item -momit-leaf-frame-pointer
7574 @opindex momit-leaf-frame-pointer
7575 Don't keep the frame pointer in a register for leaf functions. This
7576 avoids the instructions to save, set up and restore frame pointers and
7577 makes an extra register available in leaf functions. The option
7578 @option{-fomit-frame-pointer} removes the frame pointer for all functions
7579 which might make debugging harder.
7583 @subsection HPPA Options
7584 @cindex HPPA Options
7586 These @samp{-m} options are defined for the HPPA family of computers:
7589 @item -march=@var{architecture-type}
7591 Generate code for the specified architecture. The choices for
7592 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
7593 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
7594 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
7595 architecture option for your machine. Code compiled for lower numbered
7596 architectures will run on higher numbered architectures, but not the
7599 PA 2.0 support currently requires gas snapshot 19990413 or later. The
7600 next release of binutils (current is 2.9.1) will probably contain PA 2.0
7604 @itemx -mpa-risc-1-1
7605 @itemx -mpa-risc-2-0
7606 @opindex mpa-risc-1-0
7607 @opindex mpa-risc-1-1
7608 @opindex mpa-risc-2-0
7609 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
7612 @opindex mbig-switch
7613 Generate code suitable for big switch tables. Use this option only if
7614 the assembler/linker complain about out of range branches within a switch
7617 @item -mjump-in-delay
7618 @opindex mjump-in-delay
7619 Fill delay slots of function calls with unconditional jump instructions
7620 by modifying the return pointer for the function call to be the target
7621 of the conditional jump.
7623 @item -mdisable-fpregs
7624 @opindex mdisable-fpregs
7625 Prevent floating point registers from being used in any manner. This is
7626 necessary for compiling kernels which perform lazy context switching of
7627 floating point registers. If you use this option and attempt to perform
7628 floating point operations, the compiler will abort.
7630 @item -mdisable-indexing
7631 @opindex mdisable-indexing
7632 Prevent the compiler from using indexing address modes. This avoids some
7633 rather obscure problems when compiling MIG generated code under MACH@.
7635 @item -mno-space-regs
7636 @opindex mno-space-regs
7637 Generate code that assumes the target has no space registers. This allows
7638 GCC to generate faster indirect calls and use unscaled index address modes.
7640 Such code is suitable for level 0 PA systems and kernels.
7642 @item -mfast-indirect-calls
7643 @opindex mfast-indirect-calls
7644 Generate code that assumes calls never cross space boundaries. This
7645 allows GCC to emit code which performs faster indirect calls.
7647 This option will not work in the presence of shared libraries or nested
7650 @item -mlong-load-store
7651 @opindex mlong-load-store
7652 Generate 3-instruction load and store sequences as sometimes required by
7653 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
7656 @item -mportable-runtime
7657 @opindex mportable-runtime
7658 Use the portable calling conventions proposed by HP for ELF systems.
7662 Enable the use of assembler directives only GAS understands.
7664 @item -mschedule=@var{cpu-type}
7666 Schedule code according to the constraints for the machine type
7667 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
7668 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
7669 @file{/usr/lib/sched.models} on an HP-UX system to determine the
7670 proper scheduling option for your machine.
7673 @opindex mlinker-opt
7674 Enable the optimization pass in the HPUX linker. Note this makes symbolic
7675 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
7676 in which they give bogus error messages when linking some programs.
7679 @opindex msoft-float
7680 Generate output containing library calls for floating point.
7681 @strong{Warning:} the requisite libraries are not available for all HPPA
7682 targets. Normally the facilities of the machine's usual C compiler are
7683 used, but this cannot be done directly in cross-compilation. You must make
7684 your own arrangements to provide suitable library functions for
7685 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
7686 does provide software floating point support.
7688 @option{-msoft-float} changes the calling convention in the output file;
7689 therefore, it is only useful if you compile @emph{all} of a program with
7690 this option. In particular, you need to compile @file{libgcc.a}, the
7691 library that comes with GCC, with @option{-msoft-float} in order for
7695 @node Intel 960 Options
7696 @subsection Intel 960 Options
7698 These @samp{-m} options are defined for the Intel 960 implementations:
7701 @item -m@var{cpu-type}
7709 Assume the defaults for the machine type @var{cpu-type} for some of
7710 the other options, including instruction scheduling, floating point
7711 support, and addressing modes. The choices for @var{cpu-type} are
7712 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
7713 @samp{sa}, and @samp{sb}.
7720 @opindex msoft-float
7721 The @option{-mnumerics} option indicates that the processor does support
7722 floating-point instructions. The @option{-msoft-float} option indicates
7723 that floating-point support should not be assumed.
7725 @item -mleaf-procedures
7726 @itemx -mno-leaf-procedures
7727 @opindex mleaf-procedures
7728 @opindex mno-leaf-procedures
7729 Do (or do not) attempt to alter leaf procedures to be callable with the
7730 @code{bal} instruction as well as @code{call}. This will result in more
7731 efficient code for explicit calls when the @code{bal} instruction can be
7732 substituted by the assembler or linker, but less efficient code in other
7733 cases, such as calls via function pointers, or using a linker that doesn't
7734 support this optimization.
7737 @itemx -mno-tail-call
7739 @opindex mno-tail-call
7740 Do (or do not) make additional attempts (beyond those of the
7741 machine-independent portions of the compiler) to optimize tail-recursive
7742 calls into branches. You may not want to do this because the detection of
7743 cases where this is not valid is not totally complete. The default is
7744 @option{-mno-tail-call}.
7746 @item -mcomplex-addr
7747 @itemx -mno-complex-addr
7748 @opindex mcomplex-addr
7749 @opindex mno-complex-addr
7750 Assume (or do not assume) that the use of a complex addressing mode is a
7751 win on this implementation of the i960. Complex addressing modes may not
7752 be worthwhile on the K-series, but they definitely are on the C-series.
7753 The default is currently @option{-mcomplex-addr} for all processors except
7757 @itemx -mno-code-align
7758 @opindex mcode-align
7759 @opindex mno-code-align
7760 Align code to 8-byte boundaries for faster fetching (or don't bother).
7761 Currently turned on by default for C-series implementations only.
7764 @item -mclean-linkage
7765 @itemx -mno-clean-linkage
7766 @opindex mclean-linkage
7767 @opindex mno-clean-linkage
7768 These options are not fully implemented.
7772 @itemx -mic2.0-compat
7773 @itemx -mic3.0-compat
7775 @opindex mic2.0-compat
7776 @opindex mic3.0-compat
7777 Enable compatibility with iC960 v2.0 or v3.0.
7781 @opindex masm-compat
7783 Enable compatibility with the iC960 assembler.
7785 @item -mstrict-align
7786 @itemx -mno-strict-align
7787 @opindex mstrict-align
7788 @opindex mno-strict-align
7789 Do not permit (do permit) unaligned accesses.
7793 Enable structure-alignment compatibility with Intel's gcc release version
7794 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
7796 @item -mlong-double-64
7797 @opindex mlong-double-64
7798 Implement type @samp{long double} as 64-bit floating point numbers.
7799 Without the option @samp{long double} is implemented by 80-bit
7800 floating point numbers. The only reason we have it because there is
7801 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
7802 is only useful for people using soft-float targets. Otherwise, we
7803 should recommend against use of it.
7807 @node DEC Alpha Options
7808 @subsection DEC Alpha Options
7810 These @samp{-m} options are defined for the DEC Alpha implementations:
7813 @item -mno-soft-float
7815 @opindex mno-soft-float
7816 @opindex msoft-float
7817 Use (do not use) the hardware floating-point instructions for
7818 floating-point operations. When @option{-msoft-float} is specified,
7819 functions in @file{libgcc.a} will be used to perform floating-point
7820 operations. Unless they are replaced by routines that emulate the
7821 floating-point operations, or compiled in such a way as to call such
7822 emulations routines, these routines will issue floating-point
7823 operations. If you are compiling for an Alpha without floating-point
7824 operations, you must ensure that the library is built so as not to call
7827 Note that Alpha implementations without floating-point operations are
7828 required to have floating-point registers.
7833 @opindex mno-fp-regs
7834 Generate code that uses (does not use) the floating-point register set.
7835 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
7836 register set is not used, floating point operands are passed in integer
7837 registers as if they were integers and floating-point results are passed
7838 in $0 instead of $f0. This is a non-standard calling sequence, so any
7839 function with a floating-point argument or return value called by code
7840 compiled with @option{-mno-fp-regs} must also be compiled with that
7843 A typical use of this option is building a kernel that does not use,
7844 and hence need not save and restore, any floating-point registers.
7848 The Alpha architecture implements floating-point hardware optimized for
7849 maximum performance. It is mostly compliant with the IEEE floating
7850 point standard. However, for full compliance, software assistance is
7851 required. This option generates code fully IEEE compliant code
7852 @emph{except} that the @var{inexact-flag} is not maintained (see below).
7853 If this option is turned on, the CPP macro @code{_IEEE_FP} is defined
7854 during compilation. The option is a shorthand for: @option{-D_IEEE_FP
7855 -mfp-trap-mode=su -mtrap-precision=i -mieee-conformant}. The resulting
7856 code is less efficient but is able to correctly support denormalized
7857 numbers and exceptional IEEE values such as not-a-number and plus/minus
7858 infinity. Other Alpha compilers call this option
7859 @option{-ieee_with_no_inexact}.
7861 @item -mieee-with-inexact
7862 @opindex mieee-with-inexact
7863 @c overfull hbox here --bob 22 jul96
7864 @c original text between ignore ... end ignore
7866 This is like @option{-mieee} except the generated code also maintains the
7867 IEEE @var{inexact-flag}. Turning on this option causes the generated
7868 code to implement fully-compliant IEEE math. The option is a shorthand
7869 for @option{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus @option{-mieee-conformant},
7870 @option{-mfp-trap-mode=sui}, and @option{-mtrap-precision=i}. On some Alpha
7871 implementations the resulting code may execute significantly slower than
7872 the code generated by default. Since there is very little code that
7873 depends on the @var{inexact-flag}, you should normally not specify this
7874 option. Other Alpha compilers call this option
7875 @option{-ieee_with_inexact}.
7877 @c changed paragraph
7878 This is like @option{-mieee} except the generated code also maintains the
7879 IEEE @var{inexact-flag}. Turning on this option causes the generated
7880 code to implement fully-compliant IEEE math. The option is a shorthand
7881 for @option{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus the three following:
7882 @option{-mieee-conformant},
7883 @option{-mfp-trap-mode=sui},
7884 and @option{-mtrap-precision=i}.
7885 On some Alpha implementations the resulting code may execute
7886 significantly slower than the code generated by default. Since there
7887 is very little code that depends on the @var{inexact-flag}, you should
7888 normally not specify this option. Other Alpha compilers call this
7889 option @option{-ieee_with_inexact}.
7890 @c end changes to prevent overfull hboxes
7892 @item -mfp-trap-mode=@var{trap-mode}
7893 @opindex mfp-trap-mode
7894 This option controls what floating-point related traps are enabled.
7895 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
7896 The trap mode can be set to one of four values:
7900 This is the default (normal) setting. The only traps that are enabled
7901 are the ones that cannot be disabled in software (e.g., division by zero
7905 In addition to the traps enabled by @samp{n}, underflow traps are enabled
7909 Like @samp{su}, but the instructions are marked to be safe for software
7910 completion (see Alpha architecture manual for details).
7913 Like @samp{su}, but inexact traps are enabled as well.
7916 @item -mfp-rounding-mode=@var{rounding-mode}
7917 @opindex mfp-rounding-mode
7918 Selects the IEEE rounding mode. Other Alpha compilers call this option
7919 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
7924 Normal IEEE rounding mode. Floating point numbers are rounded towards
7925 the nearest machine number or towards the even machine number in case
7929 Round towards minus infinity.
7932 Chopped rounding mode. Floating point numbers are rounded towards zero.
7935 Dynamic rounding mode. A field in the floating point control register
7936 (@var{fpcr}, see Alpha architecture reference manual) controls the
7937 rounding mode in effect. The C library initializes this register for
7938 rounding towards plus infinity. Thus, unless your program modifies the
7939 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
7942 @item -mtrap-precision=@var{trap-precision}
7943 @opindex mtrap-precision
7944 In the Alpha architecture, floating point traps are imprecise. This
7945 means without software assistance it is impossible to recover from a
7946 floating trap and program execution normally needs to be terminated.
7947 GCC can generate code that can assist operating system trap handlers
7948 in determining the exact location that caused a floating point trap.
7949 Depending on the requirements of an application, different levels of
7950 precisions can be selected:
7954 Program precision. This option is the default and means a trap handler
7955 can only identify which program caused a floating point exception.
7958 Function precision. The trap handler can determine the function that
7959 caused a floating point exception.
7962 Instruction precision. The trap handler can determine the exact
7963 instruction that caused a floating point exception.
7966 Other Alpha compilers provide the equivalent options called
7967 @option{-scope_safe} and @option{-resumption_safe}.
7969 @item -mieee-conformant
7970 @opindex mieee-conformant
7971 This option marks the generated code as IEEE conformant. You must not
7972 use this option unless you also specify @option{-mtrap-precision=i} and either
7973 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
7974 is to emit the line @samp{.eflag 48} in the function prologue of the
7975 generated assembly file. Under DEC Unix, this has the effect that
7976 IEEE-conformant math library routines will be linked in.
7978 @item -mbuild-constants
7979 @opindex mbuild-constants
7980 Normally GCC examines a 32- or 64-bit integer constant to
7981 see if it can construct it from smaller constants in two or three
7982 instructions. If it cannot, it will output the constant as a literal and
7983 generate code to load it from the data segment at runtime.
7985 Use this option to require GCC to construct @emph{all} integer constants
7986 using code, even if it takes more instructions (the maximum is six).
7988 You would typically use this option to build a shared library dynamic
7989 loader. Itself a shared library, it must relocate itself in memory
7990 before it can find the variables and constants in its own data segment.
7996 Select whether to generate code to be assembled by the vendor-supplied
7997 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8011 Indicate whether GCC should generate code to use the optional BWX,
8012 CIX, and MAX instruction sets. The default is to use the instruction sets
8013 supported by the CPU type specified via @option{-mcpu=} option or that
8014 of the CPU on which GCC was built if none was specified.
8016 @item -mcpu=@var{cpu_type}
8018 Set the instruction set, register set, and instruction scheduling
8019 parameters for machine type @var{cpu_type}. You can specify either the
8020 @samp{EV} style name or the corresponding chip number. GCC
8021 supports scheduling parameters for the EV4 and EV5 family of processors
8022 and will choose the default values for the instruction set from
8023 the processor you specify. If you do not specify a processor type,
8024 GCC will default to the processor on which the compiler was built.
8026 Supported values for @var{cpu_type} are
8031 Schedules as an EV4 and has no instruction set extensions.
8035 Schedules as an EV5 and has no instruction set extensions.
8039 Schedules as an EV5 and supports the BWX extension.
8044 Schedules as an EV5 and supports the BWX and MAX extensions.
8048 Schedules as an EV5 (until Digital releases the scheduling parameters
8049 for the EV6) and supports the BWX, CIX, and MAX extensions.
8052 @item -mmemory-latency=@var{time}
8053 @opindex mmemory-latency
8054 Sets the latency the scheduler should assume for typical memory
8055 references as seen by the application. This number is highly
8056 dependent on the memory access patterns used by the application
8057 and the size of the external cache on the machine.
8059 Valid options for @var{time} are
8063 A decimal number representing clock cycles.
8069 The compiler contains estimates of the number of clock cycles for
8070 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8071 (also called Dcache, Scache, and Bcache), as well as to main memory.
8072 Note that L3 is only valid for EV5.
8077 @node Clipper Options
8078 @subsection Clipper Options
8080 These @samp{-m} options are defined for the Clipper implementations:
8085 Produce code for a C300 Clipper processor. This is the default.
8089 Produce code for a C400 Clipper processor i.e.@: use floating point
8093 @node H8/300 Options
8094 @subsection H8/300 Options
8096 These @samp{-m} options are defined for the H8/300 implementations:
8101 Shorten some address references at link time, when possible; uses the
8102 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8103 ld.info, Using ld}, for a fuller description.
8107 Generate code for the H8/300H@.
8111 Generate code for the H8/S@.
8115 Generate code for the H8/S2600. This switch must be used with @option{-ms}.
8119 Make @code{int} data 32 bits by default.
8123 On the H8/300H and H8/S, use the same alignment rules as for the H8/300.
8124 The default for the H8/300H and H8/S is to align longs and floats on 4
8126 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8127 This option has no effect on the H8/300.
8131 @subsection SH Options
8133 These @samp{-m} options are defined for the SH implementations:
8138 Generate code for the SH1.
8142 Generate code for the SH2.
8146 Generate code for the SH3.
8150 Generate code for the SH3e.
8154 Generate code for the SH4 without a floating-point unit.
8156 @item -m4-single-only
8157 @opindex m4-single-only
8158 Generate code for the SH4 with a floating-point unit that only
8159 supports single-precision arithmetic.
8163 Generate code for the SH4 assuming the floating-point unit is in
8164 single-precision mode by default.
8168 Generate code for the SH4.
8172 Compile code for the processor in big endian mode.
8176 Compile code for the processor in little endian mode.
8180 Align doubles at 64-bit boundaries. Note that this changes the calling
8181 conventions, and thus some functions from the standard C library will
8182 not work unless you recompile it first with @option{-mdalign}.
8186 Shorten some address references at link time, when possible; uses the
8187 linker option @option{-relax}.
8191 Use 32-bit offsets in @code{switch} tables. The default is to use
8196 Enable the use of the instruction @code{fmovd}.
8200 Comply with the calling conventions defined by Hitachi.
8204 Mark the @code{MAC} register as call-clobbered, even if
8205 @option{-mhitachi} is given.
8209 Increase IEEE-compliance of floating-point code.
8213 Dump instruction size and location in the assembly code.
8217 This option is deprecated. It pads structures to multiple of 4 bytes,
8218 which is incompatible with the SH ABI@.
8222 Optimize for space instead of speed. Implied by @option{-Os}.
8226 When generating position-independent code, emit function calls using
8227 the Global Offset Table instead of the Procedure Linkage Table.
8231 Generate a library function call to invalidate instruction cache
8232 entries, after fixing up a trampoline. This library function call
8233 doesn't assume it can write to the whole memory address space. This
8234 is the default when the target is @code{sh-*-linux*}.
8237 @node System V Options
8238 @subsection Options for System V
8240 These additional options are available on System V Release 4 for
8241 compatibility with other compilers on those systems:
8246 Create a shared object.
8247 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
8251 Identify the versions of each tool used by the compiler, in a
8252 @code{.ident} assembler directive in the output.
8256 Refrain from adding @code{.ident} directives to the output file (this is
8259 @item -YP,@var{dirs}
8261 Search the directories @var{dirs}, and no others, for libraries
8262 specified with @option{-l}.
8266 Look in the directory @var{dir} to find the M4 preprocessor.
8267 The assembler uses this option.
8268 @c This is supposed to go with a -Yd for predefined M4 macro files, but
8269 @c the generic assembler that comes with Solaris takes just -Ym.
8272 @node TMS320C3x/C4x Options
8273 @subsection TMS320C3x/C4x Options
8274 @cindex TMS320C3x/C4x Options
8276 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
8280 @item -mcpu=@var{cpu_type}
8282 Set the instruction set, register set, and instruction scheduling
8283 parameters for machine type @var{cpu_type}. Supported values for
8284 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
8285 @samp{c44}. The default is @samp{c40} to generate code for the
8290 @itemx -msmall-memory
8292 @opindex mbig-memory
8294 @opindex msmall-memory
8296 Generates code for the big or small memory model. The small memory
8297 model assumed that all data fits into one 64K word page. At run-time
8298 the data page (DP) register must be set to point to the 64K page
8299 containing the .bss and .data program sections. The big memory model is
8300 the default and requires reloading of the DP register for every direct
8307 Allow (disallow) allocation of general integer operands into the block
8314 Enable (disable) generation of code using decrement and branch,
8315 DBcond(D), instructions. This is enabled by default for the C4x. To be
8316 on the safe side, this is disabled for the C3x, since the maximum
8317 iteration count on the C3x is @math{2^23 + 1} (but who iterates loops more than
8318 @math{2^23} times on the C3x?). Note that GCC will try to reverse a loop so
8319 that it can utilise the decrement and branch instruction, but will give
8320 up if there is more than one memory reference in the loop. Thus a loop
8321 where the loop counter is decremented can generate slightly more
8322 efficient code, in cases where the RPTB instruction cannot be utilised.
8324 @item -mdp-isr-reload
8326 @opindex mdp-isr-reload
8328 Force the DP register to be saved on entry to an interrupt service
8329 routine (ISR), reloaded to point to the data section, and restored on
8330 exit from the ISR@. This should not be required unless someone has
8331 violated the small memory model by modifying the DP register, say within
8338 For the C3x use the 24-bit MPYI instruction for integer multiplies
8339 instead of a library call to guarantee 32-bit results. Note that if one
8340 of the operands is a constant, then the multiplication will be performed
8341 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
8342 then squaring operations are performed inline instead of a library call.
8345 @itemx -mno-fast-fix
8347 @opindex mno-fast-fix
8348 The C3x/C4x FIX instruction to convert a floating point value to an
8349 integer value chooses the nearest integer less than or equal to the
8350 floating point value rather than to the nearest integer. Thus if the
8351 floating point number is negative, the result will be incorrectly
8352 truncated an additional code is necessary to detect and correct this
8353 case. This option can be used to disable generation of the additional
8354 code required to correct the result.
8360 Enable (disable) generation of repeat block sequences using the RPTB
8361 instruction for zero overhead looping. The RPTB construct is only used
8362 for innermost loops that do not call functions or jump across the loop
8363 boundaries. There is no advantage having nested RPTB loops due to the
8364 overhead required to save and restore the RC, RS, and RE registers.
8365 This is enabled by default with @option{-O2}.
8367 @item -mrpts=@var{count}
8371 Enable (disable) the use of the single instruction repeat instruction
8372 RPTS@. If a repeat block contains a single instruction, and the loop
8373 count can be guaranteed to be less than the value @var{count}, GCC will
8374 emit a RPTS instruction instead of a RPTB@. If no value is specified,
8375 then a RPTS will be emitted even if the loop count cannot be determined
8376 at compile time. Note that the repeated instruction following RPTS does
8377 not have to be reloaded from memory each iteration, thus freeing up the
8378 CPU buses for operands. However, since interrupts are blocked by this
8379 instruction, it is disabled by default.
8381 @item -mloop-unsigned
8382 @itemx -mno-loop-unsigned
8383 @opindex mloop-unsigned
8384 @opindex mno-loop-unsigned
8385 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
8386 is @math{2^31 + 1} since these instructions test if the iteration count is
8387 negative to terminate the loop. If the iteration count is unsigned
8388 there is a possibility than the @math{2^31 + 1} maximum iteration count may be
8389 exceeded. This switch allows an unsigned iteration count.
8393 Try to emit an assembler syntax that the TI assembler (asm30) is happy
8394 with. This also enforces compatibility with the API employed by the TI
8395 C3x C compiler. For example, long doubles are passed as structures
8396 rather than in floating point registers.
8402 Generate code that uses registers (stack) for passing arguments to functions.
8403 By default, arguments are passed in registers where possible rather
8404 than by pushing arguments on to the stack.
8406 @item -mparallel-insns
8407 @itemx -mno-parallel-insns
8408 @opindex mparallel-insns
8409 @opindex mno-parallel-insns
8410 Allow the generation of parallel instructions. This is enabled by
8411 default with @option{-O2}.
8413 @item -mparallel-mpy
8414 @itemx -mno-parallel-mpy
8415 @opindex mparallel-mpy
8416 @opindex mno-parallel-mpy
8417 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
8418 provided @option{-mparallel-insns} is also specified. These instructions have
8419 tight register constraints which can pessimize the code generation
8425 @subsection V850 Options
8426 @cindex V850 Options
8428 These @samp{-m} options are defined for V850 implementations:
8432 @itemx -mno-long-calls
8433 @opindex mlong-calls
8434 @opindex mno-long-calls
8435 Treat all calls as being far away (near). If calls are assumed to be
8436 far away, the compiler will always load the functions address up into a
8437 register, and call indirect through the pointer.
8443 Do not optimize (do optimize) basic blocks that use the same index
8444 pointer 4 or more times to copy pointer into the @code{ep} register, and
8445 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
8446 option is on by default if you optimize.
8448 @item -mno-prolog-function
8449 @itemx -mprolog-function
8450 @opindex mno-prolog-function
8451 @opindex mprolog-function
8452 Do not use (do use) external functions to save and restore registers at
8453 the prolog and epilog of a function. The external functions are slower,
8454 but use less code space if more than one function saves the same number
8455 of registers. The @option{-mprolog-function} option is on by default if
8460 Try to make the code as small as possible. At present, this just turns
8461 on the @option{-mep} and @option{-mprolog-function} options.
8465 Put static or global variables whose size is @var{n} bytes or less into
8466 the tiny data area that register @code{ep} points to. The tiny data
8467 area can hold up to 256 bytes in total (128 bytes for byte references).
8471 Put static or global variables whose size is @var{n} bytes or less into
8472 the small data area that register @code{gp} points to. The small data
8473 area can hold up to 64 kilobytes.
8477 Put static or global variables whose size is @var{n} bytes or less into
8478 the first 32 kilobytes of memory.
8482 Specify that the target processor is the V850.
8485 @opindex mbig-switch
8486 Generate code suitable for big switch tables. Use this option only if
8487 the assembler/linker complain about out of range branches within a switch
8492 @subsection ARC Options
8495 These options are defined for ARC implementations:
8500 Compile code for little endian mode. This is the default.
8504 Compile code for big endian mode.
8507 @opindex mmangle-cpu
8508 Prepend the name of the cpu to all public symbol names.
8509 In multiple-processor systems, there are many ARC variants with different
8510 instruction and register set characteristics. This flag prevents code
8511 compiled for one cpu to be linked with code compiled for another.
8512 No facility exists for handling variants that are ``almost identical''.
8513 This is an all or nothing option.
8515 @item -mcpu=@var{cpu}
8517 Compile code for ARC variant @var{cpu}.
8518 Which variants are supported depend on the configuration.
8519 All variants support @option{-mcpu=base}, this is the default.
8521 @item -mtext=@var{text-section}
8522 @itemx -mdata=@var{data-section}
8523 @itemx -mrodata=@var{readonly-data-section}
8527 Put functions, data, and readonly data in @var{text-section},
8528 @var{data-section}, and @var{readonly-data-section} respectively
8529 by default. This can be overridden with the @code{section} attribute.
8530 @xref{Variable Attributes}.
8535 @subsection NS32K Options
8536 @cindex NS32K options
8538 These are the @samp{-m} options defined for the 32000 series. The default
8539 values for these options depends on which style of 32000 was selected when
8540 the compiler was configured; the defaults for the most common choices are
8548 Generate output for a 32032. This is the default
8549 when the compiler is configured for 32032 and 32016 based systems.
8555 Generate output for a 32332. This is the default
8556 when the compiler is configured for 32332-based systems.
8562 Generate output for a 32532. This is the default
8563 when the compiler is configured for 32532-based systems.
8567 Generate output containing 32081 instructions for floating point.
8568 This is the default for all systems.
8572 Generate output containing 32381 instructions for floating point. This
8573 also implies @option{-m32081}. The 32381 is only compatible with the 32332
8574 and 32532 cpus. This is the default for the pc532-netbsd configuration.
8578 Try and generate multiply-add floating point instructions @code{polyF}
8579 and @code{dotF}. This option is only available if the @option{-m32381}
8580 option is in effect. Using these instructions requires changes to to
8581 register allocation which generally has a negative impact on
8582 performance. This option should only be enabled when compiling code
8583 particularly likely to make heavy use of multiply-add instructions.
8586 @opindex mnomulti-add
8587 Do not try and generate multiply-add floating point instructions
8588 @code{polyF} and @code{dotF}. This is the default on all platforms.
8591 @opindex msoft-float
8592 Generate output containing library calls for floating point.
8593 @strong{Warning:} the requisite libraries may not be available.
8596 @opindex mnobitfield
8597 Do not use the bit-field instructions. On some machines it is faster to
8598 use shifting and masking operations. This is the default for the pc532.
8602 Do use the bit-field instructions. This is the default for all platforms
8607 Use a different function-calling convention, in which functions
8608 that take a fixed number of arguments return pop their
8609 arguments on return with the @code{ret} instruction.
8611 This calling convention is incompatible with the one normally
8612 used on Unix, so you cannot use it if you need to call libraries
8613 compiled with the Unix compiler.
8615 Also, you must provide function prototypes for all functions that
8616 take variable numbers of arguments (including @code{printf});
8617 otherwise incorrect code will be generated for calls to those
8620 In addition, seriously incorrect code will result if you call a
8621 function with too many arguments. (Normally, extra arguments are
8622 harmlessly ignored.)
8624 This option takes its name from the 680x0 @code{rtd} instruction.
8629 Use a different function-calling convention where the first two arguments
8630 are passed in registers.
8632 This calling convention is incompatible with the one normally
8633 used on Unix, so you cannot use it if you need to call libraries
8634 compiled with the Unix compiler.
8637 @opindex mnoregparam
8638 Do not pass any arguments in registers. This is the default for all
8643 It is OK to use the sb as an index register which is always loaded with
8644 zero. This is the default for the pc532-netbsd target.
8648 The sb register is not available for use or has not been initialized to
8649 zero by the run time system. This is the default for all targets except
8650 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
8651 @option{-fpic} is set.
8655 Many ns32000 series addressing modes use displacements of up to 512MB@.
8656 If an address is above 512MB then displacements from zero can not be used.
8657 This option causes code to be generated which can be loaded above 512MB@.
8658 This may be useful for operating systems or ROM code.
8662 Assume code will be loaded in the first 512MB of virtual address space.
8663 This is the default for all platforms.
8669 @subsection AVR Options
8672 These options are defined for AVR implementations:
8675 @item -mmcu=@var{mcu}
8677 Specify ATMEL AVR instruction set or MCU type.
8679 Instruction set avr1 is for the minimal AVR core, not supported by the C
8680 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
8681 attiny11, attiny12, attiny15, attiny28).
8683 Instruction set avr2 (default) is for the classic AVR core with up to
8684 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
8685 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
8686 at90c8534, at90s8535).
8688 Instruction set avr3 is for the classic AVR core with up to 128K program
8689 memory space (MCU types: atmega103, atmega603).
8691 Instruction set avr4 is for the enhanced AVR core with up to 8K program
8692 memory space (MCU types: atmega83, atmega85).
8694 Instruction set avr5 is for the enhanced AVR core with up to 128K program
8695 memory space (MCU types: atmega161, atmega163, atmega32, at94k).
8699 Output instruction sizes to the asm file.
8701 @item -minit-stack=@var{N}
8702 @opindex minit-stack
8703 Specify the initial stack address, which may be a symbol or numeric value,
8704 @samp{__stack} is the default.
8706 @item -mno-interrupts
8707 @opindex mno-interrupts
8708 Generated code is not compatible with hardware interrupts.
8709 Code size will be smaller.
8711 @item -mcall-prologues
8712 @opindex mcall-prologues
8713 Functions prologues/epilogues expanded as call to appropriate
8714 subroutines. Code size will be smaller.
8716 @item -mno-tablejump
8717 @opindex mno-tablejump
8718 Do not generate tablejump insns which sometimes increase code size.
8721 @opindex mtiny-stack
8722 Change only the low 8 bits of the stack pointer.
8726 @subsection MCore Options
8727 @cindex MCore options
8729 These are the @samp{-m} options defined for the Motorola M*Core
8739 @opindex mno-hardlit
8740 Inline constants into the code stream if it can be done in two
8741 instructions or less.
8749 Use the divide instruction. (Enabled by default).
8751 @item -mrelax-immediate
8752 @itemx -mrelax-immediate
8753 @itemx -mno-relax-immediate
8754 @opindex mrelax-immediate
8755 @opindex mrelax-immediate
8756 @opindex mno-relax-immediate
8757 Allow arbitrary sized immediates in bit operations.
8759 @item -mwide-bitfields
8760 @itemx -mwide-bitfields
8761 @itemx -mno-wide-bitfields
8762 @opindex mwide-bitfields
8763 @opindex mwide-bitfields
8764 @opindex mno-wide-bitfields
8765 Always treat bit-fields as int-sized.
8767 @item -m4byte-functions
8768 @itemx -m4byte-functions
8769 @itemx -mno-4byte-functions
8770 @opindex m4byte-functions
8771 @opindex m4byte-functions
8772 @opindex mno-4byte-functions
8773 Force all functions to be aligned to a four byte boundary.
8775 @item -mcallgraph-data
8776 @itemx -mcallgraph-data
8777 @itemx -mno-callgraph-data
8778 @opindex mcallgraph-data
8779 @opindex mcallgraph-data
8780 @opindex mno-callgraph-data
8781 Emit callgraph information.
8785 @itemx -mno-slow-bytes
8786 @opindex mslow-bytes
8787 @opindex mslow-bytes
8788 @opindex mno-slow-bytes
8789 Prefer word access when reading byte quantities.
8791 @item -mlittle-endian
8792 @itemx -mlittle-endian
8794 @opindex mlittle-endian
8795 @opindex mlittle-endian
8796 @opindex mbig-endian
8797 Generate code for a little endian target.
8805 Generate code for the 210 processor.
8809 @subsection IA-64 Options
8810 @cindex IA-64 Options
8812 These are the @samp{-m} options defined for the Intel IA-64 architecture.
8816 @opindex mbig-endian
8817 Generate code for a big endian target. This is the default for HPUX@.
8819 @item -mlittle-endian
8820 @opindex mlittle-endian
8821 Generate code for a little endian target. This is the default for AIX5
8828 Generate (or don't) code for the GNU assembler. This is the default.
8829 @c Also, this is the default if the configure option @option{--with-gnu-as}
8836 Generate (or don't) code for the GNU linker. This is the default.
8837 @c Also, this is the default if the configure option @option{--with-gnu-ld}
8842 Generate code that does not use a global pointer register. The result
8843 is not position independent code, and violates the IA-64 ABI@.
8845 @item -mvolatile-asm-stop
8846 @itemx -mno-volatile-asm-stop
8847 @opindex mvolatile-asm-stop
8848 @opindex mno-volatile-asm-stop
8849 Generate (or don't) a stop bit immediately before and after volatile asm
8854 Generate code that works around Itanium B step errata.
8856 @item -mregister-names
8857 @itemx -mno-register-names
8858 @opindex mregister-names
8859 @opindex mno-register-names
8860 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
8861 the stacked registers. This may make assembler output more readable.
8867 Disable (or enable) optimizations that use the small data section. This may
8868 be useful for working around optimizer bugs.
8871 @opindex mconstant-gp
8872 Generate code that uses a single constant global pointer value. This is
8873 useful when compiling kernel code.
8877 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
8878 This is useful when compiling firmware code.
8880 @item -minline-divide-min-latency
8881 @opindex minline-divide-min-latency
8882 Generate code for inline divides using the minimum latency algorithm.
8884 @item -minline-divide-max-throughput
8885 @opindex minline-divide-max-throughput
8886 Generate code for inline divides using the maximum throughput algorithm.
8888 @item -mno-dwarf2-asm
8890 @opindex mno-dwarf2-asm
8891 @opindex mdwarf2-asm
8892 Don't (or do) generate assembler code for the DWARF2 line number debugging
8893 info. This may be useful when not using the GNU assembler.
8895 @item -mfixed-range=@var{register-range}
8896 @opindex mfixed-range
8897 Generate code treating the given register range as fixed registers.
8898 A fixed register is one that the register allocator can not use. This is
8899 useful when compiling kernel code. A register range is specified as
8900 two registers separated by a dash. Multiple register ranges can be
8901 specified separated by a comma.
8905 @subsection D30V Options
8906 @cindex D30V Options
8908 These @samp{-m} options are defined for D30V implementations:
8913 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
8914 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
8915 memory, which starts at location @code{0x80000000}.
8919 Same as the @option{-mextmem} switch.
8923 Link the @samp{.text} section into onchip text memory, which starts at
8924 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
8925 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
8926 into onchip data memory, which starts at location @code{0x20000000}.
8928 @item -mno-asm-optimize
8929 @itemx -masm-optimize
8930 @opindex mno-asm-optimize
8931 @opindex masm-optimize
8932 Disable (enable) passing @option{-O} to the assembler when optimizing.
8933 The assembler uses the @option{-O} option to automatically parallelize
8934 adjacent short instructions where possible.
8936 @item -mbranch-cost=@var{n}
8937 @opindex mbranch-cost
8938 Increase the internal costs of branches to @var{n}. Higher costs means
8939 that the compiler will issue more instructions to avoid doing a branch.
8942 @item -mcond-exec=@var{n}
8944 Specify the maximum number of conditionally executed instructions that
8945 replace a branch. The default is 4.
8948 @node Code Gen Options
8949 @section Options for Code Generation Conventions
8950 @cindex code generation conventions
8951 @cindex options, code generation
8952 @cindex run-time options
8954 These machine-independent options control the interface conventions
8955 used in code generation.
8957 Most of them have both positive and negative forms; the negative form
8958 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
8959 one of the forms is listed---the one which is not the default. You
8960 can figure out the other form by either removing @samp{no-} or adding
8965 @opindex fexceptions
8966 Enable exception handling. Generates extra code needed to propagate
8967 exceptions. For some targets, this implies GCC will generate frame
8968 unwind information for all functions, which can produce significant data
8969 size overhead, although it does not affect execution. If you do not
8970 specify this option, GCC will enable it by default for languages like
8971 C++ which normally require exception handling, and disable it for
8972 languages like C that do not normally require it. However, you may need
8973 to enable this option when compiling C code that needs to interoperate
8974 properly with exception handlers written in C++. You may also wish to
8975 disable this option if you are compiling older C++ programs that don't
8976 use exception handling.
8978 @item -fnon-call-exceptions
8979 @opindex fnon-call-exceptions
8980 Generate code that allows trapping instructions to throw exceptions.
8981 Note that this requires platform-specific runtime support that does
8982 not exist everywhere. Moreover, it only allows @emph{trapping}
8983 instructions to throw exceptions, i.e.@: memory references or floating
8984 point instructions. It does not allow exceptions to be thrown from
8985 arbitrary signal handlers such as @code{SIGALRM}.
8987 @item -funwind-tables
8988 @opindex funwind-tables
8989 Similar to @option{-fexceptions}, except that it will just generate any needed
8990 static data, but will not affect the generated code in any other way.
8991 You will normally not enable this option; instead, a language processor
8992 that needs this handling would enable it on your behalf.
8994 @item -fpcc-struct-return
8995 @opindex fpcc-struct-return
8996 Return ``short'' @code{struct} and @code{union} values in memory like
8997 longer ones, rather than in registers. This convention is less
8998 efficient, but it has the advantage of allowing intercallability between
8999 GCC-compiled files and files compiled with other compilers.
9001 The precise convention for returning structures in memory depends
9002 on the target configuration macros.
9004 Short structures and unions are those whose size and alignment match
9005 that of some integer type.
9007 @item -freg-struct-return
9008 @opindex freg-struct-return
9009 Use the convention that @code{struct} and @code{union} values are
9010 returned in registers when possible. This is more efficient for small
9011 structures than @option{-fpcc-struct-return}.
9013 If you specify neither @option{-fpcc-struct-return} nor its contrary
9014 @option{-freg-struct-return}, GCC defaults to whichever convention is
9015 standard for the target. If there is no standard convention, GCC
9016 defaults to @option{-fpcc-struct-return}, except on targets where GCC
9017 is the principal compiler. In those cases, we can choose the standard,
9018 and we chose the more efficient register return alternative.
9021 @opindex fshort-enums
9022 Allocate to an @code{enum} type only as many bytes as it needs for the
9023 declared range of possible values. Specifically, the @code{enum} type
9024 will be equivalent to the smallest integer type which has enough room.
9026 @item -fshort-double
9027 @opindex fshort-double
9028 Use the same size for @code{double} as for @code{float}.
9031 @opindex fshared-data
9032 Requests that the data and non-@code{const} variables of this
9033 compilation be shared data rather than private data. The distinction
9034 makes sense only on certain operating systems, where shared data is
9035 shared between processes running the same program, while private data
9036 exists in one copy per process.
9040 In C, allocate even uninitialized global variables in the data section of the
9041 object file, rather than generating them as common blocks. This has the
9042 effect that if the same variable is declared (without @code{extern}) in
9043 two different compilations, you will get an error when you link them.
9044 The only reason this might be useful is if you wish to verify that the
9045 program will work on other systems which always work this way.
9049 Ignore the @samp{#ident} directive.
9051 @item -fno-gnu-linker
9052 @opindex fno-gnu-linker
9053 Do not output global initializations (such as C++ constructors and
9054 destructors) in the form used by the GNU linker (on systems where the GNU
9055 linker is the standard method of handling them). Use this option when
9056 you want to use a non-GNU linker, which also requires using the
9057 @command{collect2} program to make sure the system linker includes
9058 constructors and destructors. (@command{collect2} is included in the GCC
9059 distribution.) For systems which @emph{must} use @command{collect2}, the
9060 compiler driver @command{gcc} is configured to do this automatically.
9062 @item -finhibit-size-directive
9063 @opindex finhibit-size-directive
9064 Don't output a @code{.size} assembler directive, or anything else that
9065 would cause trouble if the function is split in the middle, and the
9066 two halves are placed at locations far apart in memory. This option is
9067 used when compiling @file{crtstuff.c}; you should not need to use it
9071 @opindex fverbose-asm
9072 Put extra commentary information in the generated assembly code to
9073 make it more readable. This option is generally only of use to those
9074 who actually need to read the generated assembly code (perhaps while
9075 debugging the compiler itself).
9077 @option{-fno-verbose-asm}, the default, causes the
9078 extra information to be omitted and is useful when comparing two assembler
9083 Consider all memory references through pointers to be volatile.
9085 @item -fvolatile-global
9086 @opindex fvolatile-global
9087 Consider all memory references to extern and global data items to
9088 be volatile. GCC does not consider static data items to be volatile
9089 because of this switch.
9091 @item -fvolatile-static
9092 @opindex fvolatile-static
9093 Consider all memory references to static data to be volatile.
9097 @cindex global offset table
9099 Generate position-independent code (PIC) suitable for use in a shared
9100 library, if supported for the target machine. Such code accesses all
9101 constant addresses through a global offset table (GOT)@. The dynamic
9102 loader resolves the GOT entries when the program starts (the dynamic
9103 loader is not part of GCC; it is part of the operating system). If
9104 the GOT size for the linked executable exceeds a machine-specific
9105 maximum size, you get an error message from the linker indicating that
9106 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
9107 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
9108 on the m68k and RS/6000. The 386 has no such limit.)
9110 Position-independent code requires special support, and therefore works
9111 only on certain machines. For the 386, GCC supports PIC for System V
9112 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
9113 position-independent.
9117 If supported for the target machine, emit position-independent code,
9118 suitable for dynamic linking and avoiding any limit on the size of the
9119 global offset table. This option makes a difference on the m68k, m88k,
9122 Position-independent code requires special support, and therefore works
9123 only on certain machines.
9125 @item -ffixed-@var{reg}
9127 Treat the register named @var{reg} as a fixed register; generated code
9128 should never refer to it (except perhaps as a stack pointer, frame
9129 pointer or in some other fixed role).
9131 @var{reg} must be the name of a register. The register names accepted
9132 are machine-specific and are defined in the @code{REGISTER_NAMES}
9133 macro in the machine description macro file.
9135 This flag does not have a negative form, because it specifies a
9138 @item -fcall-used-@var{reg}
9140 Treat the register named @var{reg} as an allocable register that is
9141 clobbered by function calls. It may be allocated for temporaries or
9142 variables that do not live across a call. Functions compiled this way
9143 will not save and restore the register @var{reg}.
9145 It is an error to used this flag with the frame pointer or stack pointer.
9146 Use of this flag for other registers that have fixed pervasive roles in
9147 the machine's execution model will produce disastrous results.
9149 This flag does not have a negative form, because it specifies a
9152 @item -fcall-saved-@var{reg}
9153 @opindex fcall-saved
9154 Treat the register named @var{reg} as an allocable register saved by
9155 functions. It may be allocated even for temporaries or variables that
9156 live across a call. Functions compiled this way will save and restore
9157 the register @var{reg} if they use it.
9159 It is an error to used this flag with the frame pointer or stack pointer.
9160 Use of this flag for other registers that have fixed pervasive roles in
9161 the machine's execution model will produce disastrous results.
9163 A different sort of disaster will result from the use of this flag for
9164 a register in which function values may be returned.
9166 This flag does not have a negative form, because it specifies a
9170 @opindex fpack-struct
9171 Pack all structure members together without holes. Usually you would
9172 not want to use this option, since it makes the code suboptimal, and
9173 the offsets of structure members won't agree with system libraries.
9175 @item -fcheck-memory-usage
9176 @opindex fcheck-memory-usage
9177 Generate extra code to check each memory access. GCC will generate
9178 code that is suitable for a detector of bad memory accesses such as
9181 Normally, you should compile all, or none, of your code with this option.
9183 If you do mix code compiled with and without this option,
9184 you must ensure that all code that has side effects
9185 and that is called by code compiled with this option
9186 is, itself, compiled with this option.
9187 If you do not, you might get erroneous messages from the detector.
9189 If you use functions from a library that have side-effects (such as
9190 @code{read}), you might not be able to recompile the library and
9191 specify this option. In that case, you can enable the
9192 @option{-fprefix-function-name} option, which requests GCC to encapsulate
9193 your code and make other functions look as if they were compiled with
9194 @option{-fcheck-memory-usage}. This is done by calling ``stubs'',
9195 which are provided by the detector. If you cannot find or build
9196 stubs for every function you call, you might have to specify
9197 @option{-fcheck-memory-usage} without @option{-fprefix-function-name}.
9199 If you specify this option, you can not use the @code{asm} or
9200 @code{__asm__} keywords in functions with memory checking enabled. GCC
9201 cannot understand what the @code{asm} statement may do, and therefore
9202 cannot generate the appropriate code, so it will reject it. However, if
9203 you specify the function attribute @code{no_check_memory_usage}
9204 (@pxref{Function Attributes}), GCC will disable memory checking within a
9205 function; you may use @code{asm} statements inside such functions. You
9206 may have an inline expansion of a non-checked function within a checked
9207 function; in that case GCC will not generate checks for the inlined
9208 function's memory accesses.
9210 If you move your @code{asm} statements to non-checked inline functions
9211 and they do access memory, you can add calls to the support code in your
9212 inline function, to indicate any reads, writes, or copies being done.
9213 These calls would be similar to those done in the stubs described above.
9215 @item -fprefix-function-name
9216 @opindex fprefix-function-name
9217 Request GCC to add a prefix to the symbols generated for function names.
9218 GCC adds a prefix to the names of functions defined as well as
9219 functions called. Code compiled with this option and code compiled
9220 without the option can't be linked together, unless stubs are used.
9222 If you compile the following code with @option{-fprefix-function-name}
9224 extern void bar (int);
9233 GCC will compile the code as if it was written:
9235 extern void prefix_bar (int);
9239 return prefix_bar (a + 5);
9242 This option is designed to be used with @option{-fcheck-memory-usage}.
9244 @item -finstrument-functions
9245 @opindex finstrument-functions
9246 Generate instrumentation calls for entry and exit to functions. Just
9247 after function entry and just before function exit, the following
9248 profiling functions will be called with the address of the current
9249 function and its call site. (On some platforms,
9250 @code{__builtin_return_address} does not work beyond the current
9251 function, so the call site information may not be available to the
9252 profiling functions otherwise.)
9255 void __cyg_profile_func_enter (void *this_fn,
9257 void __cyg_profile_func_exit (void *this_fn,
9261 The first argument is the address of the start of the current function,
9262 which may be looked up exactly in the symbol table.
9264 This instrumentation is also done for functions expanded inline in other
9265 functions. The profiling calls will indicate where, conceptually, the
9266 inline function is entered and exited. This means that addressable
9267 versions of such functions must be available. If all your uses of a
9268 function are expanded inline, this may mean an additional expansion of
9269 code size. If you use @samp{extern inline} in your C code, an
9270 addressable version of such functions must be provided. (This is
9271 normally the case anyways, but if you get lucky and the optimizer always
9272 expands the functions inline, you might have gotten away without
9273 providing static copies.)
9275 A function may be given the attribute @code{no_instrument_function}, in
9276 which case this instrumentation will not be done. This can be used, for
9277 example, for the profiling functions listed above, high-priority
9278 interrupt routines, and any functions from which the profiling functions
9279 cannot safely be called (perhaps signal handlers, if the profiling
9280 routines generate output or allocate memory).
9283 @opindex fstack-check
9284 Generate code to verify that you do not go beyond the boundary of the
9285 stack. You should specify this flag if you are running in an
9286 environment with multiple threads, but only rarely need to specify it in
9287 a single-threaded environment since stack overflow is automatically
9288 detected on nearly all systems if there is only one stack.
9290 Note that this switch does not actually cause checking to be done; the
9291 operating system must do that. The switch causes generation of code
9292 to ensure that the operating system sees the stack being extended.
9294 @item -fstack-limit-register=@var{reg}
9295 @itemx -fstack-limit-symbol=@var{sym}
9296 @itemx -fno-stack-limit
9297 @opindex fstack-limit-register
9298 @opindex fstack-limit-symbol
9299 @opindex fno-stack-limit
9300 Generate code to ensure that the stack does not grow beyond a certain value,
9301 either the value of a register or the address of a symbol. If the stack
9302 would grow beyond the value, a signal is raised. For most targets,
9303 the signal is raised before the stack overruns the boundary, so
9304 it is possible to catch the signal without taking special precautions.
9306 For instance, if the stack starts at address @samp{0x80000000} and grows
9307 downwards you can use the flags
9308 @samp{-fstack-limit-symbol=__stack_limit
9309 -Wl,--defsym,__stack_limit=0x7ffe0000} which will enforce a stack
9312 @cindex aliasing of parameters
9313 @cindex parameters, aliased
9314 @item -fargument-alias
9315 @itemx -fargument-noalias
9316 @itemx -fargument-noalias-global
9317 @opindex fargument-alias
9318 @opindex fargument-noalias
9319 @opindex fargument-noalias-global
9320 Specify the possible relationships among parameters and between
9321 parameters and global data.
9323 @option{-fargument-alias} specifies that arguments (parameters) may
9324 alias each other and may alias global storage.
9325 @option{-fargument-noalias} specifies that arguments do not alias
9326 each other, but may alias global storage.
9327 @option{-fargument-noalias-global} specifies that arguments do not
9328 alias each other and do not alias global storage.
9330 Each language will automatically use whatever option is required by
9331 the language standard. You should not need to use these options yourself.
9333 @item -fleading-underscore
9334 @opindex fleading-underscore
9335 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
9336 change the way C symbols are represented in the object file. One use
9337 is to help link with legacy assembly code.
9339 Be warned that you should know what you are doing when invoking this
9340 option, and that not all targets provide complete support for it.
9345 @node Environment Variables
9346 @section Environment Variables Affecting GCC
9347 @cindex environment variables
9349 @c man begin ENVIRONMENT
9351 This section describes several environment variables that affect how GCC
9352 operates. Some of them work by specifying directories or prefixes to use
9353 when searching for various kinds of files. Some are used to specify other
9354 aspects of the compilation environment.
9357 Note that you can also specify places to search using options such as
9358 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
9359 take precedence over places specified using environment variables, which
9360 in turn take precedence over those specified by the configuration of GCC@.
9364 Note that you can also specify places to search using options such as
9365 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
9366 take precedence over places specified using environment variables, which
9367 in turn take precedence over those specified by the configuration of GCC@.
9374 @c @itemx LC_COLLATE
9376 @c @itemx LC_MONETARY
9377 @c @itemx LC_NUMERIC
9382 @c @findex LC_COLLATE
9384 @c @findex LC_MONETARY
9385 @c @findex LC_NUMERIC
9389 These environment variables control the way that GCC uses
9390 localization information that allow GCC to work with different
9391 national conventions. GCC inspects the locale categories
9392 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
9393 so. These locale categories can be set to any value supported by your
9394 installation. A typical value is @samp{en_UK} for English in the United
9397 The @env{LC_CTYPE} environment variable specifies character
9398 classification. GCC uses it to determine the character boundaries in
9399 a string; this is needed for some multibyte encodings that contain quote
9400 and escape characters that would otherwise be interpreted as a string
9403 The @env{LC_MESSAGES} environment variable specifies the language to
9404 use in diagnostic messages.
9406 If the @env{LC_ALL} environment variable is set, it overrides the value
9407 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
9408 and @env{LC_MESSAGES} default to the value of the @env{LANG}
9409 environment variable. If none of these variables are set, GCC
9410 defaults to traditional C English behavior.
9414 If @env{TMPDIR} is set, it specifies the directory to use for temporary
9415 files. GCC uses temporary files to hold the output of one stage of
9416 compilation which is to be used as input to the next stage: for example,
9417 the output of the preprocessor, which is the input to the compiler
9420 @item GCC_EXEC_PREFIX
9421 @findex GCC_EXEC_PREFIX
9422 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
9423 names of the subprograms executed by the compiler. No slash is added
9424 when this prefix is combined with the name of a subprogram, but you can
9425 specify a prefix that ends with a slash if you wish.
9427 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
9428 an appropriate prefix to use based on the pathname it was invoked with.
9430 If GCC cannot find the subprogram using the specified prefix, it
9431 tries looking in the usual places for the subprogram.
9433 The default value of @env{GCC_EXEC_PREFIX} is
9434 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
9435 of @code{prefix} when you ran the @file{configure} script.
9437 Other prefixes specified with @option{-B} take precedence over this prefix.
9439 This prefix is also used for finding files such as @file{crt0.o} that are
9442 In addition, the prefix is used in an unusual way in finding the
9443 directories to search for header files. For each of the standard
9444 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
9445 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
9446 replacing that beginning with the specified prefix to produce an
9447 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
9448 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
9449 These alternate directories are searched first; the standard directories
9453 @findex COMPILER_PATH
9454 The value of @env{COMPILER_PATH} is a colon-separated list of
9455 directories, much like @env{PATH}. GCC tries the directories thus
9456 specified when searching for subprograms, if it can't find the
9457 subprograms using @env{GCC_EXEC_PREFIX}.
9460 @findex LIBRARY_PATH
9461 The value of @env{LIBRARY_PATH} is a colon-separated list of
9462 directories, much like @env{PATH}. When configured as a native compiler,
9463 GCC tries the directories thus specified when searching for special
9464 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
9465 using GCC also uses these directories when searching for ordinary
9466 libraries for the @option{-l} option (but directories specified with
9467 @option{-L} come first).
9469 @item C_INCLUDE_PATH
9470 @itemx CPLUS_INCLUDE_PATH
9471 @itemx OBJC_INCLUDE_PATH
9472 @findex C_INCLUDE_PATH
9473 @findex CPLUS_INCLUDE_PATH
9474 @findex OBJC_INCLUDE_PATH
9475 @c @itemx OBJCPLUS_INCLUDE_PATH
9476 These environment variables pertain to particular languages. Each
9477 variable's value is a colon-separated list of directories, much like
9478 @env{PATH}. When GCC searches for header files, it tries the
9479 directories listed in the variable for the language you are using, after
9480 the directories specified with @option{-I} but before the standard header
9483 @item DEPENDENCIES_OUTPUT
9484 @findex DEPENDENCIES_OUTPUT
9485 @cindex dependencies for make as output
9486 If this variable is set, its value specifies how to output dependencies
9487 for Make based on the header files processed by the compiler. This
9488 output looks much like the output from the @option{-M} option
9489 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
9490 in addition to the usual results of compilation.
9492 The value of @env{DEPENDENCIES_OUTPUT} can be just a file name, in
9493 which case the Make rules are written to that file, guessing the target
9494 name from the source file name. Or the value can have the form
9495 @samp{@var{file} @var{target}}, in which case the rules are written to
9496 file @var{file} using @var{target} as the target name.
9500 @cindex locale definition
9501 This variable is used to pass locale information to the compiler. One way in
9502 which this information is used is to determine the character set to be used
9503 when character literals, string literals and comments are parsed in C and C++.
9504 When the compiler is configured to allow multibyte characters,
9505 the following values for @env{LANG} are recognized:
9509 Recognize JIS characters.
9511 Recognize SJIS characters.
9513 Recognize EUCJP characters.
9516 If @env{LANG} is not defined, or if it has some other value, then the
9517 compiler will use mblen and mbtowc as defined by the default locale to
9518 recognize and translate multibyte characters.
9523 @node Running Protoize
9524 @section Running Protoize
9526 The program @code{protoize} is an optional part of GCC@. You can use
9527 it to add prototypes to a program, thus converting the program to ISO
9528 C in one respect. The companion program @code{unprotoize} does the
9529 reverse: it removes argument types from any prototypes that are found.
9531 When you run these programs, you must specify a set of source files as
9532 command line arguments. The conversion programs start out by compiling
9533 these files to see what functions they define. The information gathered
9534 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
9536 After scanning comes actual conversion. The specified files are all
9537 eligible to be converted; any files they include (whether sources or
9538 just headers) are eligible as well.
9540 But not all the eligible files are converted. By default,
9541 @code{protoize} and @code{unprotoize} convert only source and header
9542 files in the current directory. You can specify additional directories
9543 whose files should be converted with the @option{-d @var{directory}}
9544 option. You can also specify particular files to exclude with the
9545 @option{-x @var{file}} option. A file is converted if it is eligible, its
9546 directory name matches one of the specified directory names, and its
9547 name within the directory has not been excluded.
9549 Basic conversion with @code{protoize} consists of rewriting most
9550 function definitions and function declarations to specify the types of
9551 the arguments. The only ones not rewritten are those for varargs
9554 @code{protoize} optionally inserts prototype declarations at the
9555 beginning of the source file, to make them available for any calls that
9556 precede the function's definition. Or it can insert prototype
9557 declarations with block scope in the blocks where undeclared functions
9560 Basic conversion with @code{unprotoize} consists of rewriting most
9561 function declarations to remove any argument types, and rewriting
9562 function definitions to the old-style pre-ISO form.
9564 Both conversion programs print a warning for any function declaration or
9565 definition that they can't convert. You can suppress these warnings
9568 The output from @code{protoize} or @code{unprotoize} replaces the
9569 original source file. The original file is renamed to a name ending
9570 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
9571 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
9572 for DOS) file already exists, then the source file is simply discarded.
9574 @code{protoize} and @code{unprotoize} both depend on GCC itself to
9575 scan the program and collect information about the functions it uses.
9576 So neither of these programs will work until GCC is installed.
9578 Here is a table of the options you can use with @code{protoize} and
9579 @code{unprotoize}. Each option works with both programs unless
9583 @item -B @var{directory}
9584 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
9585 usual directory (normally @file{/usr/local/lib}). This file contains
9586 prototype information about standard system functions. This option
9587 applies only to @code{protoize}.
9589 @item -c @var{compilation-options}
9590 Use @var{compilation-options} as the options when running @code{gcc} to
9591 produce the @samp{.X} files. The special option @option{-aux-info} is
9592 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
9594 Note that the compilation options must be given as a single argument to
9595 @code{protoize} or @code{unprotoize}. If you want to specify several
9596 @code{gcc} options, you must quote the entire set of compilation options
9597 to make them a single word in the shell.
9599 There are certain @code{gcc} arguments that you cannot use, because they
9600 would produce the wrong kind of output. These include @option{-g},
9601 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
9602 the @var{compilation-options}, they are ignored.
9605 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
9606 systems) instead of @samp{.c}. This is convenient if you are converting
9607 a C program to C++. This option applies only to @code{protoize}.
9610 Add explicit global declarations. This means inserting explicit
9611 declarations at the beginning of each source file for each function
9612 that is called in the file and was not declared. These declarations
9613 precede the first function definition that contains a call to an
9614 undeclared function. This option applies only to @code{protoize}.
9616 @item -i @var{string}
9617 Indent old-style parameter declarations with the string @var{string}.
9618 This option applies only to @code{protoize}.
9620 @code{unprotoize} converts prototyped function definitions to old-style
9621 function definitions, where the arguments are declared between the
9622 argument list and the initial @samp{@{}. By default, @code{unprotoize}
9623 uses five spaces as the indentation. If you want to indent with just
9624 one space instead, use @option{-i " "}.
9627 Keep the @samp{.X} files. Normally, they are deleted after conversion
9631 Add explicit local declarations. @code{protoize} with @option{-l} inserts
9632 a prototype declaration for each function in each block which calls the
9633 function without any declaration. This option applies only to
9637 Make no real changes. This mode just prints information about the conversions
9638 that would have been done without @option{-n}.
9641 Make no @samp{.save} files. The original files are simply deleted.
9642 Use this option with caution.
9644 @item -p @var{program}
9645 Use the program @var{program} as the compiler. Normally, the name
9649 Work quietly. Most warnings are suppressed.
9652 Print the version number, just like @option{-v} for @code{gcc}.
9655 If you need special compiler options to compile one of your program's
9656 source files, then you should generate that file's @samp{.X} file
9657 specially, by running @code{gcc} on that source file with the
9658 appropriate options and the option @option{-aux-info}. Then run
9659 @code{protoize} on the entire set of files. @code{protoize} will use
9660 the existing @samp{.X} file because it is newer than the source file.
9664 gcc -Dfoo=bar file1.c -aux-info file1.X
9669 You need to include the special files along with the rest in the
9670 @code{protoize} command, even though their @samp{.X} files already
9671 exist, because otherwise they won't get converted.
9673 @xref{Protoize Caveats}, for more information on how to use
9674 @code{protoize} successfully.