1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000 Free Software Foundation, Inc.
2 @c This is part of the GCC manual.
3 @c For copying conditions, see the file gcc.texi.
6 @chapter GCC Command Options
7 @cindex GCC command options
8 @cindex command options
9 @cindex options, GCC command
11 When you invoke GCC, it normally does preprocessing, compilation,
12 assembly and linking. The ``overall options'' allow you to stop this
13 process at an intermediate stage. For example, the @samp{-c} option
14 says not to run the linker. Then the output consists of object files
15 output by the assembler.
17 Other options are passed on to one stage of processing. Some options
18 control the preprocessor and others the compiler itself. Yet other
19 options control the assembler and linker; most of these are not
20 documented here, since you rarely need to use any of them.
22 @cindex C compilation options
23 Most of the command line options that you can use with GCC are useful
24 for C programs; when an option is only useful with another language
25 (usually C++), the explanation says so explicitly. If the description
26 for a particular option does not mention a source language, you can use
27 that option with all supported languages.
29 @cindex C++ compilation options
30 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
31 options for compiling C++ programs.
33 @cindex grouping options
34 @cindex options, grouping
35 The @code{gcc} program accepts options and file names as operands. Many
36 options have multiletter names; therefore multiple single-letter options
37 may @emph{not} be grouped: @samp{-dr} is very different from @w{@samp{-d
40 @cindex order of options
41 @cindex options, order
42 You can mix options and other arguments. For the most part, the order
43 you use doesn't matter. Order does matter when you use several options
44 of the same kind; for example, if you specify @samp{-L} more than once,
45 the directories are searched in the order specified.
47 Many options have long names starting with @samp{-f} or with
48 @samp{-W}---for example, @samp{-fforce-mem},
49 @samp{-fstrength-reduce}, @samp{-Wformat} and so on. Most of
50 these have both positive and negative forms; the negative form of
51 @samp{-ffoo} would be @samp{-fno-foo}. This manual documents
52 only one of these two forms, whichever one is not the default.
55 * Option Summary:: Brief list of all options, without explanations.
56 * Overall Options:: Controlling the kind of output:
57 an executable, object files, assembler files,
58 or preprocessed source.
59 * Invoking G++:: Compiling C++ programs.
60 * C Dialect Options:: Controlling the variant of C language compiled.
61 * C++ Dialect Options:: Variations on C++.
62 * Warning Options:: How picky should the compiler be?
63 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
64 * Optimize Options:: How much optimization?
65 * Preprocessor Options:: Controlling header files and macro definitions.
66 Also, getting dependency information for Make.
67 * Assembler Options:: Passing options to the assembler.
68 * Link Options:: Specifying libraries and so on.
69 * Directory Options:: Where to find header files and libraries.
70 Where to find the compiler executable files.
71 * Spec Files:: How to pass switches to sub-processes.
72 * Target Options:: Running a cross-compiler, or an old version of GCC.
73 * Submodel Options:: Specifying minor hardware or convention variations,
74 such as 68010 vs 68020.
75 * Code Gen Options:: Specifying conventions for function calls, data layout
77 * Environment Variables:: Env vars that affect GCC.
78 * Running Protoize:: Automatically adding or removing function prototypes.
82 @section Option Summary
84 Here is a summary of all the options, grouped by type. Explanations are
85 in the following sections.
89 @xref{Overall Options,,Options Controlling the Kind of Output}.
91 -c -S -E -o @var{file} -pipe -pass-exit-codes -v --help -x @var{language}
94 @item C Language Options
95 @xref{C Dialect Options,,Options Controlling C Dialect}.
97 -ansi -fstd -fallow-single-precision -fcond-mismatch -fno-asm
98 -fno-builtin -ffreestanding -fhosted -fsigned-bitfields -fsigned-char
99 -funsigned-bitfields -funsigned-char -fwritable-strings
100 -traditional -traditional-cpp -trigraphs
103 @item C++ Language Options
104 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
106 -fno-access-control -fcheck-new -fconserve-space
107 -fdollars-in-identifiers -fno-elide-constructors -fexternal-templates
108 -ffor-scope -fno-for-scope -fno-gnu-keywords -fguiding-decls -fhonor-std
109 -fhuge-objects -fno-implicit-templates -finit-priority
110 -fno-implement-inlines -fname-mangling-version-@var{n}
111 -fno-default-inline -fno-operator-names -fno-optional-diags -fpermissive
112 -frepo -fstrict-prototype -fsquangle -ftemplate-depth-@var{n}
113 -fuse-cxa-atexit -fvtable-thunks -nostdinc++ -Wctor-dtor-privacy
114 -Wno-deprecated -Weffc++ -Wno-non-template-friend -Wnon-virtual-dtor
115 -Wold-style-cast -Woverloaded-virtual -Wno-pmf-conversions -Wreorder
119 @item Warning Options
120 @xref{Warning Options,,Options to Request or Suppress Warnings}.
122 -fsyntax-only -pedantic -pedantic-errors
123 -w -W -Wall -Waggregate-return
124 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment
125 -Wconversion -Werror -Wformat
126 -Wid-clash-@var{len} -Wimplicit -Wimplicit-int
127 -Wimplicit-function-declaration -Wimport
128 -Werror-implicit-function-declaration -Wfloat-equal -Winline
129 -Wlarger-than-@var{len} -Wlong-long
130 -Wmain -Wmissing-declarations -Wmissing-noreturn
131 -Wmultichar -Wno-import -Wpacked -Wpadded
132 -Wparentheses -Wpointer-arith -Wredundant-decls
133 -Wreturn-type -Wshadow -Wsign-compare -Wswitch
134 -Wtrigraphs -Wundef -Wuninitialized -Wunknown-pragmas -Wunreachable-code
135 -Wunused -Wwrite-strings
138 @item C-only Warning Options
140 -Wbad-function-cast -Wmissing-prototypes -Wnested-externs
141 -Wstrict-prototypes -Wtraditional
144 @item Debugging Options
145 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
147 -a -ax -d@var{letters} -fdump-unnumbered -fdump-translation-unit-@var{file}
148 -fpretend-float -fprofile-arcs -ftest-coverage
149 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2
150 -ggdb -gstabs -gstabs+ -gxcoff -gxcoff+
151 -p -pg -print-file-name=@var{library} -print-libgcc-file-name
152 -print-prog-name=@var{program} -print-search-dirs -save-temps -time
155 @item Optimization Options
156 @xref{Optimize Options,,Options that Control Optimization}.
158 -falign-functions=@var{n} -falign-labels=@var{n} -falign-loops=@var{n}
159 -falign-jumps=@var{n} -fbranch-probabilities
160 -fcaller-saves -fcse-follow-jumps -fcse-skip-blocks
161 -fdelayed-branch -fdelete-null-pointer-checks -fexpensive-optimizations
162 -ffast-math -ffloat-store -fforce-addr -fforce-mem -fno-math-errno
163 -fdata-sections -ffunction-sections -fgcse
164 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions
165 -fmove-all-movables -fno-default-inline -fno-defer-pop
166 -fno-function-cse -fno-inline -fno-peephole
167 -fomit-frame-pointer -foptimize-register-moves -foptimize-sibling-calls
168 -fregmove -frerun-cse-after-loop -frerun-loop-opt -freduce-all-givs
169 -fschedule-insns -fschedule-insns2 -fstrength-reduce
170 -fstrict-aliasing -fthread-jumps -funroll-all-loops
172 -O -O0 -O1 -O2 -O3 -Os
175 @item Preprocessor Options
176 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
178 -A@var{question}(@var{answer}) -C -dD -dM -dN
179 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H
181 -include @var{file} -imacros @var{file}
182 -iprefix @var{file} -iwithprefix @var{dir}
183 -iwithprefixbefore @var{dir} -isystem @var{dir} -isystem-c++ @var{dir}
184 -M -MD -MM -MMD -MG -nostdinc -P -trigraphs
185 -undef -U@var{macro} -Wp,@var{option}
188 @item Assembler Option
189 @xref{Assembler Options,,Passing Options to the Assembler}.
195 @xref{Link Options,,Options for Linking}.
197 @var{object-file-name} -l@var{library}
198 -nostartfiles -nodefaultlibs -nostdlib
199 -s -static -shared -symbolic
200 -Wl,@var{option} -Xlinker @var{option}
204 @item Directory Options
205 @xref{Directory Options,,Options for Directory Search}.
207 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}
211 @c I wrote this xref this way to avoid overfull hbox. -- rms
212 @xref{Target Options}.
214 -b @var{machine} -V @var{version}
217 @item Machine Dependent Options
218 @xref{Submodel Options,,Hardware Models and Configurations}.
220 @emph{M680x0 Options}
221 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040
222 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020
223 -mfpa -mnobitfield -mrtd -mshort -msoft-float -mpcrel
224 -malign-int -mstrict-align
231 -mtune=@var{cpu type}
232 -mcmodel=@var{code model}
234 -mapp-regs -mbroken-saverestore -mcypress
235 -mepilogue -mfaster-structs -mflat
236 -mfpu -mhard-float -mhard-quad-float
237 -mimpure-text -mlive-g0 -mno-app-regs
238 -mno-epilogue -mno-faster-structs -mno-flat -mno-fpu
239 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles
240 -msoft-float -msoft-quad-float -msparclite -mstack-bias
241 -msupersparc -munaligned-doubles -mv8
243 @emph{Convex Options}
244 -mc1 -mc2 -mc32 -mc34 -mc38
245 -margcount -mnoargcount
247 -mvolatile-cache -mvolatile-nocache
249 @emph{AMD29K Options}
250 -m29000 -m29050 -mbw -mnbw -mdw -mndw
251 -mlarge -mnormal -msmall
252 -mkernel-registers -mno-reuse-arg-regs
253 -mno-stack-check -mno-storem-bug
254 -mreuse-arg-regs -msoft-float -mstack-check
255 -mstorem-bug -muser-registers
258 -mapcs-frame -mno-apcs-frame
260 -mapcs-stack-check -mno-apcs-stack-check
261 -mapcs-float -mno-apcs-float
262 -mapcs-reentrant -mno-apcs-reentrant
263 -msched-prolog -mno-sched-prolog
264 -mlittle-endian -mbig-endian -mwords-little-endian
265 -malignment-traps -mno-alignment-traps
266 -msoft-float -mhard-float -mfpe
267 -mthumb-interwork -mno-thumb-interwork
268 -mcpu= -march= -mfpe=
269 -mstructure-size-boundary=
270 -mbsd -mxopen -mno-symrename
272 -mlong-calls -mno-long-calls
273 -mnop-fun-dllimport -mno-nop-fun-dllimport
274 -msingle-pic-base -mno-single-pic-base
278 -mtpcs-frame -mno-tpcs-frame
279 -mtpcs-leaf-frame -mno-tpcs-leaf-frame
280 -mlittle-endian -mbig-endian
281 -mthumb-interwork -mno-thumb-interwork
282 -mstructure-size-boundary=
283 -mnop-fun-dllimport -mno-nop-fun-dllimport
284 -mcallee-super-interworking -mno-callee-super-interworking
285 -mcaller-super-interworking -mno-caller-super-interworking
286 -msingle-pic-base -mno-single-pic-base
289 @emph{MN10200 Options}
292 @emph{MN10300 Options}
299 @emph{M32R/D Options}
300 -mcode-model=@var{model type} -msdata=@var{sdata type}
304 -m88000 -m88100 -m88110 -mbig-pic
305 -mcheck-zero-division -mhandle-large-shift
306 -midentify-revision -mno-check-zero-division
307 -mno-ocs-debug-info -mno-ocs-frame-position
308 -mno-optimize-arg-area -mno-serialize-volatile
309 -mno-underscores -mocs-debug-info
310 -mocs-frame-position -moptimize-arg-area
311 -mserialize-volatile -mshort-data-@var{num} -msvr3
312 -msvr4 -mtrap-large-shift -muse-div-instruction
313 -mversion-03.00 -mwarn-passed-structs
315 @emph{RS/6000 and PowerPC Options}
317 -mtune=@var{cpu type}
318 -mpower -mno-power -mpower2 -mno-power2
319 -mpowerpc -mpowerpc64 -mno-powerpc
320 -mpowerpc-gpopt -mno-powerpc-gpopt
321 -mpowerpc-gfxopt -mno-powerpc-gfxopt
322 -mnew-mnemonics -mold-mnemonics
323 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc
324 -m64 -m32 -mxl-call -mno-xl-call -mthreads -mpe
325 -msoft-float -mhard-float -mmultiple -mno-multiple
326 -mstring -mno-string -mupdate -mno-update
327 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align
328 -mstrict-align -mno-strict-align -mrelocatable
329 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib
330 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian
331 -mcall-aix -mcall-sysv -mprototype -mno-prototype
332 -msim -mmvme -mads -myellowknife -memb -msdata
333 -msdata=@var{opt} -G @var{num}
336 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs
337 -mfull-fp-blocks -mhc-struct-return -min-line-mul
338 -mminimum-fp-blocks -mnohc-struct-return
341 -mabicalls -mcpu=@var{cpu type} -membedded-data -muninit-const-in-rodata
342 -membedded-pic -mfp32 -mfp64 -mgas -mgp32 -mgp64
343 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1
344 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy
345 -mmips-as -mmips-tfile -mno-abicalls
346 -mno-embedded-data -mno-uninit-const-in-rodata -mno-embedded-pic
347 -mno-gpopt -mno-long-calls
348 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats
349 -mrnames -msoft-float
350 -m4650 -msingle-float -mmad
351 -mstats -EL -EB -G @var{num} -nocpp
352 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi
357 -march=@var{cpu type}
358 -mieee-fp -mno-fancy-math-387
359 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib
360 -mno-wide-multiply -mrtd -malign-double
361 -mreg-alloc=@var{list} -mregparm=@var{num}
362 -malign-jumps=@var{num} -malign-loops=@var{num}
363 -malign-functions=@var{num} -mpreferred-stack-boundary=@var{num}
364 -mthreads -mno-align-stringops -minline-all-stringops
365 -mpush-args -maccumulate-outgoing-args
368 -march=@var{architecture type}
369 -mbig-switch -mdisable-fpregs -mdisable-indexing
370 -mfast-indirect-calls -mgas -mjump-in-delay
371 -mlong-load-store -mno-big-switch -mno-disable-fpregs
372 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas
373 -mno-jump-in-delay -mno-long-load-store
374 -mno-portable-runtime -mno-soft-float
375 -mno-space-regs -msoft-float -mpa-risc-1-0
376 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime
377 -mschedule=@var{cpu type} -mspace-regs
379 @emph{Intel 960 Options}
380 -m@var{cpu type} -masm-compat -mclean-linkage
381 -mcode-align -mcomplex-addr -mleaf-procedures
382 -mic-compat -mic2.0-compat -mic3.0-compat
383 -mintel-asm -mno-clean-linkage -mno-code-align
384 -mno-complex-addr -mno-leaf-procedures
385 -mno-old-align -mno-strict-align -mno-tail-call
386 -mnumerics -mold-align -msoft-float -mstrict-align
389 @emph{DEC Alpha Options}
390 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float
392 -mieee -mieee-with-inexact -mieee-conformant
393 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode}
394 -mtrap-precision=@var{mode} -mbuild-constants
396 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max
397 -mmemory-latency=@var{time}
399 @emph{Clipper Options}
402 @emph{H8/300 Options}
403 -mrelax -mh -ms -mint32 -malign-300
406 -m1 -m2 -m3 -m3e -mb -ml -mdalign -mrelax
408 @emph{System V Options}
409 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}
413 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text section}
414 -mdata=@var{data section} -mrodata=@var{readonly data section}
416 @emph{TMS320C3x/C4x Options}
417 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm
418 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload
419 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned
420 -mparallel-insns -mparallel-mpy -mpreserve-float
423 -mlong-calls -mno-long-calls -mep -mno-ep
424 -mprolog-function -mno-prolog-function -mspace
425 -mtda=@var{n} -msda=@var{n} -mzda=@var{n}
429 -m32032 -m32332 -m32532 -m32081 -m32381 -mmult-add -mnomult-add
430 -msoft-float -mrtd -mnortd -mregparam -mnoregparam -msb -mnosb
431 -mbitfield -mnobitfield -mhimem -mnohimem
434 -mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts
438 -mhardlit, -mno-hardlit -mdiv -mno-div -mrelax-immediates
439 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields
440 -m4byte-functions -mno-4byte-functions -mcallgraph-data
441 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim
442 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment
445 @item Code Generation Options
446 @xref{Code Gen Options,,Options for Code Generation Conventions}.
448 -fcall-saved-@var{reg} -fcall-used-@var{reg}
449 -fexceptions -funwind-tables -ffixed-@var{reg} -finhibit-size-directive
450 -fcheck-memory-usage -fprefix-function-name
451 -fno-common -fno-ident -fno-gnu-linker
452 -fpcc-struct-return -fpic -fPIC
453 -freg-struct-return -fshared-data -fshort-enums
454 -fshort-double -fvolatile -fvolatile-global -fvolatile-static
455 -fverbose-asm -fpack-struct -fstack-check
456 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym}
457 -fargument-alias -fargument-noalias
458 -fargument-noalias-global
464 * Overall Options:: Controlling the kind of output:
465 an executable, object files, assembler files,
466 or preprocessed source.
467 * C Dialect Options:: Controlling the variant of C language compiled.
468 * C++ Dialect Options:: Variations on C++.
469 * Warning Options:: How picky should the compiler be?
470 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
471 * Optimize Options:: How much optimization?
472 * Preprocessor Options:: Controlling header files and macro definitions.
473 Also, getting dependency information for Make.
474 * Assembler Options:: Passing options to the assembler.
475 * Link Options:: Specifying libraries and so on.
476 * Directory Options:: Where to find header files and libraries.
477 Where to find the compiler executable files.
478 * Spec Files:: How to pass switches to sub-processes.
479 * Target Options:: Running a cross-compiler, or an old version of GCC.
482 @node Overall Options
483 @section Options Controlling the Kind of Output
485 Compilation can involve up to four stages: preprocessing, compilation
486 proper, assembly and linking, always in that order. The first three
487 stages apply to an individual source file, and end by producing an
488 object file; linking combines all the object files (those newly
489 compiled, and those specified as input) into an executable file.
491 @cindex file name suffix
492 For any given input file, the file name suffix determines what kind of
497 C source code which must be preprocessed.
500 C source code which should not be preprocessed.
503 C++ source code which should not be preprocessed.
506 Objective-C source code. Note that you must link with the library
507 @file{libobjc.a} to make an Objective-C program work.
510 C header file (not to be compiled or linked).
513 @itemx @var{file}.cxx
514 @itemx @var{file}.cpp
516 C++ source code which must be preprocessed. Note that in @samp{.cxx},
517 the last two letters must both be literally @samp{x}. Likewise,
518 @samp{.C} refers to a literal capital C.
524 Assembler code which must be preprocessed.
527 An object file to be fed straight into linking.
528 Any file name with no recognized suffix is treated this way.
531 You can specify the input language explicitly with the @samp{-x} option:
534 @item -x @var{language}
535 Specify explicitly the @var{language} for the following input files
536 (rather than letting the compiler choose a default based on the file
537 name suffix). This option applies to all following input files until
538 the next @samp{-x} option. Possible values for @var{language} are:
541 c-header cpp-output c++-cpp-output
542 assembler assembler-with-cpp
546 Turn off any specification of a language, so that subsequent files are
547 handled according to their file name suffixes (as they are if @samp{-x}
548 has not been used at all).
550 @item -pass-exit-codes
551 Normally the @code{gcc} program will exit with the code of 1 if any
552 phase of the compiler returns a non-success return code. If you specify
553 @samp{-pass-exit-codes}, the @code{gcc} program will instead return with
554 numerically highest error produced by any phase that returned an error
558 If you only want some of the stages of compilation, you can use
559 @samp{-x} (or filename suffixes) to tell @code{gcc} where to start, and
560 one of the options @samp{-c}, @samp{-S}, or @samp{-E} to say where
561 @code{gcc} is to stop. Note that some combinations (for example,
562 @samp{-x cpp-output -E} instruct @code{gcc} to do nothing at all.
566 Compile or assemble the source files, but do not link. The linking
567 stage simply is not done. The ultimate output is in the form of an
568 object file for each source file.
570 By default, the object file name for a source file is made by replacing
571 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
573 Unrecognized input files, not requiring compilation or assembly, are
577 Stop after the stage of compilation proper; do not assemble. The output
578 is in the form of an assembler code file for each non-assembler input
581 By default, the assembler file name for a source file is made by
582 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
584 Input files that don't require compilation are ignored.
587 Stop after the preprocessing stage; do not run the compiler proper. The
588 output is in the form of preprocessed source code, which is sent to the
591 Input files which don't require preprocessing are ignored.
593 @cindex output file option
595 Place output in file @var{file}. This applies regardless to whatever
596 sort of output is being produced, whether it be an executable file,
597 an object file, an assembler file or preprocessed C code.
599 Since only one output file can be specified, it does not make sense to
600 use @samp{-o} when compiling more than one input file, unless you are
601 producing an executable file as output.
603 If @samp{-o} is not specified, the default is to put an executable file
604 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
605 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
606 all preprocessed C source on standard output.@refill
609 Print (on standard error output) the commands executed to run the stages
610 of compilation. Also print the version number of the compiler driver
611 program and of the preprocessor and the compiler proper.
614 Use pipes rather than temporary files for communication between the
615 various stages of compilation. This fails to work on some systems where
616 the assembler is unable to read from a pipe; but the GNU assembler has
620 Print (on the standard output) a description of the command line options
621 understood by @code{gcc}. If the @code{-v} option is also specified
622 then @code{--help} will also be passed on to the various processes
623 invoked by @code{gcc}, so that they can display the command line options
624 they accept. If the @code{-W} option is also specified then command
625 line options which have no documentation associated with them will also
630 @section Compiling C++ Programs
632 @cindex suffixes for C++ source
633 @cindex C++ source file suffixes
634 C++ source files conventionally use one of the suffixes @samp{.C},
635 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
636 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
637 files with these names and compiles them as C++ programs even if you
638 call the compiler the same way as for compiling C programs (usually with
639 the name @code{gcc}).
643 However, C++ programs often require class libraries as well as a
644 compiler that understands the C++ language---and under some
645 circumstances, you might want to compile programs from standard input,
646 or otherwise without a suffix that flags them as C++ programs.
647 @code{g++} is a program that calls GCC with the default language
648 set to C++, and automatically specifies linking against the C++
649 library. On many systems, the script @code{g++} is also
650 installed with the name @code{c++}.
652 @cindex invoking @code{g++}
653 When you compile C++ programs, you may specify many of the same
654 command-line options that you use for compiling programs in any
655 language; or command-line options meaningful for C and related
656 languages; or options that are meaningful only for C++ programs.
657 @xref{C Dialect Options,,Options Controlling C Dialect}, for
658 explanations of options for languages related to C.
659 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
660 explanations of options that are meaningful only for C++ programs.
662 @node C Dialect Options
663 @section Options Controlling C Dialect
664 @cindex dialect options
665 @cindex language dialect options
666 @cindex options, dialect
668 The following options control the dialect of C (or languages derived
669 from C, such as C++ and Objective C) that the compiler accepts:
674 In C mode, support all ANSI standard C programs. In C++ mode,
675 remove GNU extensions that conflict with ISO C++.
676 @c shouldn't we be saying "ISO"?
678 This turns off certain features of GCC that are incompatible with ANSI
679 C (when compiling C code), or of standard C++ (when compiling C++ code),
680 such as the @code{asm} and @code{typeof} keywords, and
681 predefined macros such as @code{unix} and @code{vax} that identify the
682 type of system you are using. It also enables the undesirable and
683 rarely used ANSI trigraph feature. For the C compiler,
684 it disables recognition of C++ style @samp{//} comments as well as
685 the @code{inline} keyword.
687 The alternate keywords @code{__asm__}, @code{__extension__},
688 @code{__inline__} and @code{__typeof__} continue to work despite
689 @samp{-ansi}. You would not want to use them in an ANSI C program, of
690 course, but it is useful to put them in header files that might be included
691 in compilations done with @samp{-ansi}. Alternate predefined macros
692 such as @code{__unix__} and @code{__vax__} are also available, with or
693 without @samp{-ansi}.
695 The @samp{-ansi} option does not cause non-ANSI programs to be
696 rejected gratuitously. For that, @samp{-pedantic} is required in
697 addition to @samp{-ansi}. @xref{Warning Options}.
699 The macro @code{__STRICT_ANSI__} is predefined when the @samp{-ansi}
700 option is used. Some header files may notice this macro and refrain
701 from declaring certain functions or defining certain macros that the
702 ANSI standard doesn't call for; this is to avoid interfering with any
703 programs that might use these names for other things.
705 The functions @code{alloca}, @code{abort}, @code{exit}, and
706 @code{_exit} are not builtin functions when @samp{-ansi} is used.
709 Determine the language standard. A value for this option must be provided;
717 ISO C as modified in amend. 1
723 same as -std=iso9899:1990
726 same as -std=iso9899:199x
729 default, iso9899:1990 + gnu extensions
732 iso9899:199x + gnu extensions
735 Even when this option is not specified, you can still use some of the
736 features of newer standards in so far as they do not conflict with
737 previous C standards. For example, you may use @code{__restrict__} even
738 when -fstd=c9x is not specified.
741 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
742 keyword, so that code can use these words as identifiers. You can use
743 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
744 instead. @samp{-ansi} implies @samp{-fno-asm}.
746 In C++, this switch only affects the @code{typeof} keyword, since
747 @code{asm} and @code{inline} are standard keywords. You may want to
748 use the @samp{-fno-gnu-keywords} flag instead, as it also disables the
749 other, C++-specific, extension keywords such as @code{headof}.
752 @cindex builtin functions
778 Don't recognize builtin functions that do not begin with @samp{__builtin_}
779 as prefix. Currently, the functions affected include @code{abort},
780 @code{abs}, @code{alloca}, @code{cos}, @code{cosf}, @code{cosl},
781 @code{exit}, @code{_exit}, @code{fabs}, @code{fabsf}, @code{fabsl},
782 @code{ffs}, @code{labs}, @code{memcmp}, @code{memcpy}, @code{memset},
783 @code{sin}, @code{sinf}, @code{sinl}, @code{sqrt}, @code{sqrtf},
784 @code{sqrtl}, @code{strcmp}, @code{strcpy}, and @code{strlen}.
786 GCC normally generates special code to handle certain builtin functions
787 more efficiently; for instance, calls to @code{alloca} may become single
788 instructions that adjust the stack directly, and calls to @code{memcpy}
789 may become inline copy loops. The resulting code is often both smaller
790 and faster, but since the function calls no longer appear as such, you
791 cannot set a breakpoint on those calls, nor can you change the behavior
792 of the functions by linking with a different library.
794 The @samp{-ansi} option prevents @code{alloca}, @code{ffs} and @code{_exit}
795 from being builtin functions, since these functions do not have an ANSI
799 @cindex hosted environment
801 Assert that compilation takes place in a hosted environment. This implies
802 @samp{-fbuiltin}. A hosted environment is one in which the
803 entire standard library is available, and in which @code{main} has a return
804 type of @code{int}. Examples are nearly everything except a kernel.
805 This is equivalent to @samp{-fno-freestanding}.
808 @cindex hosted environment
810 Assert that compilation takes place in a freestanding environment. This
811 implies @samp{-fno-builtin}. A freestanding environment
812 is one in which the standard library may not exist, and program startup may
813 not necessarily be at @code{main}. The most obvious example is an OS kernel.
814 This is equivalent to @samp{-fno-hosted}.
817 Support ANSI C trigraphs. You don't want to know about this
818 brain-damage. The @samp{-ansi} option implies @samp{-trigraphs}.
820 @cindex traditional C language
821 @cindex C language, traditional
823 Attempt to support some aspects of traditional C compilers.
828 All @code{extern} declarations take effect globally even if they
829 are written inside of a function definition. This includes implicit
830 declarations of functions.
833 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
834 and @code{volatile} are not recognized. (You can still use the
835 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
839 Comparisons between pointers and integers are always allowed.
842 Integer types @code{unsigned short} and @code{unsigned char} promote
843 to @code{unsigned int}.
846 Out-of-range floating point literals are not an error.
849 Certain constructs which ANSI regards as a single invalid preprocessing
850 number, such as @samp{0xe-0xd}, are treated as expressions instead.
853 String ``constants'' are not necessarily constant; they are stored in
854 writable space, and identical looking constants are allocated
855 separately. (This is the same as the effect of
856 @samp{-fwritable-strings}.)
858 @cindex @code{longjmp} and automatic variables
860 All automatic variables not declared @code{register} are preserved by
861 @code{longjmp}. Ordinarily, GNU C follows ANSI C: automatic variables
862 not declared @code{volatile} may be clobbered.
867 @cindex escape sequences, traditional
868 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
869 literal characters @samp{x} and @samp{a} respectively. Without
870 @w{@samp{-traditional}}, @samp{\x} is a prefix for the hexadecimal
871 representation of a character, and @samp{\a} produces a bell.
874 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
875 if your program uses names that are normally GNU C builtin functions for
876 other purposes of its own.
878 You cannot use @samp{-traditional} if you include any header files that
879 rely on ANSI C features. Some vendors are starting to ship systems with
880 ANSI C header files and you cannot use @samp{-traditional} on such
881 systems to compile files that include any system headers.
883 The @samp{-traditional} option also enables @samp{-traditional-cpp},
884 which is described next.
886 @item -traditional-cpp
887 Attempt to support some aspects of traditional C preprocessors.
892 Comments convert to nothing at all, rather than to a space. This allows
893 traditional token concatenation.
896 In a preprocessing directive, the @samp{#} symbol must appear as the first
900 Macro arguments are recognized within string constants in a macro
901 definition (and their values are stringified, though without additional
902 quote marks, when they appear in such a context). The preprocessor
903 always considers a string constant to end at a newline.
906 @cindex detecting @w{@samp{-traditional}}
907 The predefined macro @code{__STDC__} is not defined when you use
908 @samp{-traditional}, but @code{__GNUC__} is (since the GNU extensions
909 which @code{__GNUC__} indicates are not affected by
910 @samp{-traditional}). If you need to write header files that work
911 differently depending on whether @samp{-traditional} is in use, by
912 testing both of these predefined macros you can distinguish four
913 situations: GNU C, traditional GNU C, other ANSI C compilers, and other
914 old C compilers. The predefined macro @code{__STDC_VERSION__} is also
915 not defined when you use @samp{-traditional}. @xref{Standard
916 Predefined,,Standard Predefined Macros,cpp.info,The C Preprocessor},
917 for more discussion of these and other predefined macros.
920 @cindex string constants vs newline
921 @cindex newline vs string constants
922 The preprocessor considers a string constant to end at a newline (unless
923 the newline is escaped with @samp{\}). (Without @w{@samp{-traditional}},
924 string constants can contain the newline character as typed.)
927 @item -fcond-mismatch
928 Allow conditional expressions with mismatched types in the second and
929 third arguments. The value of such an expression is void.
931 @item -funsigned-char
932 Let the type @code{char} be unsigned, like @code{unsigned char}.
934 Each kind of machine has a default for what @code{char} should
935 be. It is either like @code{unsigned char} by default or like
936 @code{signed char} by default.
938 Ideally, a portable program should always use @code{signed char} or
939 @code{unsigned char} when it depends on the signedness of an object.
940 But many programs have been written to use plain @code{char} and
941 expect it to be signed, or expect it to be unsigned, depending on the
942 machines they were written for. This option, and its inverse, let you
943 make such a program work with the opposite default.
945 The type @code{char} is always a distinct type from each of
946 @code{signed char} or @code{unsigned char}, even though its behavior
947 is always just like one of those two.
950 Let the type @code{char} be signed, like @code{signed char}.
952 Note that this is equivalent to @samp{-fno-unsigned-char}, which is
953 the negative form of @samp{-funsigned-char}. Likewise, the option
954 @samp{-fno-signed-char} is equivalent to @samp{-funsigned-char}.
956 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
957 if your program uses names that are normally GNU C builtin functions for
958 other purposes of its own.
960 You cannot use @samp{-traditional} if you include any header files that
961 rely on ANSI C features. Some vendors are starting to ship systems with
962 ANSI C header files and you cannot use @samp{-traditional} on such
963 systems to compile files that include any system headers.
965 @item -fsigned-bitfields
966 @itemx -funsigned-bitfields
967 @itemx -fno-signed-bitfields
968 @itemx -fno-unsigned-bitfields
969 These options control whether a bitfield is signed or unsigned, when the
970 declaration does not use either @code{signed} or @code{unsigned}. By
971 default, such a bitfield is signed, because this is consistent: the
972 basic integer types such as @code{int} are signed types.
974 However, when @samp{-traditional} is used, bitfields are all unsigned
977 @item -fwritable-strings
978 Store string constants in the writable data segment and don't uniquize
979 them. This is for compatibility with old programs which assume they can
980 write into string constants. The option @samp{-traditional} also has
983 Writing into string constants is a very bad idea; ``constants'' should
986 @item -fallow-single-precision
987 Do not promote single precision math operations to double precision,
988 even when compiling with @samp{-traditional}.
990 Traditional K&R C promotes all floating point operations to double
991 precision, regardless of the sizes of the operands. On the
992 architecture for which you are compiling, single precision may be faster
993 than double precision. If you must use @samp{-traditional}, but want
994 to use single precision operations when the operands are single
995 precision, use this option. This option has no effect when compiling
996 with ANSI or GNU C conventions (the default).
999 Override the underlying type for @samp{wchar_t} to be @samp{short
1000 unsigned int} instead of the default for the target. This option is
1001 useful for building programs to run under WINE.
1004 @node C++ Dialect Options
1005 @section Options Controlling C++ Dialect
1007 @cindex compiler options, C++
1008 @cindex C++ options, command line
1009 @cindex options, C++
1010 This section describes the command-line options that are only meaningful
1011 for C++ programs; but you can also use most of the GNU compiler options
1012 regardless of what language your program is in. For example, you
1013 might compile a file @code{firstClass.C} like this:
1016 g++ -g -frepo -O -c firstClass.C
1020 In this example, only @samp{-frepo} is an option meant
1021 only for C++ programs; you can use the other options with any
1022 language supported by GCC.
1024 Here is a list of options that are @emph{only} for compiling C++ programs:
1027 @item -fno-access-control
1028 Turn off all access checking. This switch is mainly useful for working
1029 around bugs in the access control code.
1032 Check that the pointer returned by @code{operator new} is non-null
1033 before attempting to modify the storage allocated. The current Working
1034 Paper requires that @code{operator new} never return a null pointer, so
1035 this check is normally unnecessary.
1037 An alternative to using this option is to specify that your
1038 @code{operator new} does not throw any exceptions; if you declare it
1039 @samp{throw()}, g++ will check the return value. See also @samp{new
1042 @item -fconserve-space
1043 Put uninitialized or runtime-initialized global variables into the
1044 common segment, as C does. This saves space in the executable at the
1045 cost of not diagnosing duplicate definitions. If you compile with this
1046 flag and your program mysteriously crashes after @code{main()} has
1047 completed, you may have an object that is being destroyed twice because
1048 two definitions were merged.
1050 This option is no longer useful on most targets, now that support has
1051 been added for putting variables into BSS without making them common.
1053 @item -fdollars-in-identifiers
1054 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1055 @samp{$} with the option @samp{-fno-dollars-in-identifiers}. (GNU C allows
1056 @samp{$} by default on most target systems, but there are a few exceptions.)
1057 Traditional C allowed the character @samp{$} to form part of
1058 identifiers. However, ANSI C and C++ forbid @samp{$} in identifiers.
1060 @item -fno-elide-constructors
1061 The C++ standard allows an implementation to omit creating a temporary
1062 which is only used to initialize another object of the same type.
1063 Specifying this option disables that optimization, and forces g++ to
1064 call the copy constructor in all cases.
1066 @item -fno-enforce-eh-specs
1067 Don't check for violation of exception specifications at runtime. This
1068 option violates the C++ standard, but may be useful for reducing code
1069 size in production builds, much like defining @samp{NDEBUG}. The compiler
1070 will still optimize based on the exception specifications.
1072 @item -fexternal-templates
1073 Cause template instantiations to obey @samp{#pragma interface} and
1074 @samp{implementation}; template instances are emitted or not according
1075 to the location of the template definition. @xref{Template
1076 Instantiation}, for more information.
1078 This option is deprecated.
1080 @item -falt-external-templates
1081 Similar to -fexternal-templates, but template instances are emitted or
1082 not according to the place where they are first instantiated.
1083 @xref{Template Instantiation}, for more information.
1085 This option is deprecated.
1088 @itemx -fno-for-scope
1089 If -ffor-scope is specified, the scope of variables declared in
1090 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1091 as specified by the C++ standard.
1092 If -fno-for-scope is specified, the scope of variables declared in
1093 a @i{for-init-statement} extends to the end of the enclosing scope,
1094 as was the case in old versions of gcc, and other (traditional)
1095 implementations of C++.
1097 The default if neither flag is given to follow the standard,
1098 but to allow and give a warning for old-style code that would
1099 otherwise be invalid, or have different behavior.
1101 @item -fno-gnu-keywords
1102 Do not recognize @code{typeof} as a keyword, so that code can use these
1103 words as identifiers. You can use the keywords @code{__typeof__}
1104 instead. @samp{-ansi} implies @samp{-fno-gnu-keywords}.
1106 @item -fguiding-decls
1107 Treat a function declaration with the same type as a potential function
1108 template instantiation as though it declares that instantiation, not a
1109 normal function. If a definition is given for the function later in the
1110 translation unit (or another translation unit if the target supports
1111 weak symbols), that definition will be used; otherwise the template will
1112 be instantiated. This behavior reflects the C++ language prior to
1113 September 1996, when guiding declarations were removed.
1115 This option implies @samp{-fname-mangling-version-0}, and will not work
1116 with other name mangling versions. Like all options that change the
1117 ABI, all C++ code, @emph{including libgcc.a} must be built with the same
1118 setting of this option.
1121 Treat the @code{namespace std} as a namespace, instead of ignoring
1122 it. For compatibility with earlier versions of g++, the compiler will,
1123 by default, ignore @code{namespace-declarations},
1124 @code{using-declarations}, @code{using-directives}, and
1125 @code{namespace-names}, if they involve @code{std}.
1127 @item -fhuge-objects
1128 Support virtual function calls for objects that exceed the size
1129 representable by a @samp{short int}. Users should not use this flag by
1130 default; if you need to use it, the compiler will tell you so.
1132 This flag is not useful when compiling with -fvtable-thunks.
1134 Like all options that change the ABI, all C++ code, @emph{including
1135 libgcc} must be built with the same setting of this option.
1137 @item -fmessage-length=@var{n}
1138 Try to format error messages so that they fit on lines of about @var{n}
1139 characters. The default is 72 characters. If @var{n} is zero, then no
1140 line-wrapping will be done; each error message will appear on a single
1143 @item -fno-implicit-templates
1144 Never emit code for non-inline templates which are instantiated
1145 implicitly (i.e. by use); only emit code for explicit instantiations.
1146 @xref{Template Instantiation}, for more information.
1148 @item -fno-implicit-inline-templates
1149 Don't emit code for implicit instantiations of inline templates, either.
1150 The default is to handle inlines differently so that compiles with and
1151 without optimization will need the same set of explicit instantiations.
1153 @item -finit-priority
1154 Support @samp{__attribute__ ((init_priority (n)))} for controlling the
1155 order of initialization of file-scope objects. On ELF targets, this
1156 requires GNU ld 2.10 or later.
1158 @item -fno-implement-inlines
1159 To save space, do not emit out-of-line copies of inline functions
1160 controlled by @samp{#pragma implementation}. This will cause linker
1161 errors if these functions are not inlined everywhere they are called.
1163 @item -fms-extensions
1164 Disable pedwarns about constructs used in MFC, such as implicit int and
1165 getting a pointer to member function via non-standard syntax.
1167 @item -fname-mangling-version-@var{n}
1168 Control the way in which names are mangled. Version 0 is compatible
1169 with versions of g++ before 2.8. Version 1 is the default. Version 1
1170 will allow correct mangling of function templates. For example,
1171 version 0 mangling does not mangle foo<int, double> and foo<int, char>
1172 given this declaration:
1175 template <class T, class U> void foo(T t);
1178 Like all options that change the ABI, all C++ code, @emph{including
1179 libgcc} must be built with the same setting of this option.
1181 @item -fno-operator-names
1182 Do not treat the operator name keywords @code{and}, @code{bitand},
1183 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1184 synonyms as keywords.
1186 @item -fno-optional-diags
1187 Disable diagnostics that the standard says a compiler does not need to
1188 issue. Currently, the only such diagnostic issued by g++ is the one for
1189 a name having multiple meanings within a class.
1192 Downgrade messages about nonconformant code from errors to warnings. By
1193 default, g++ effectively sets @samp{-pedantic-errors} without
1194 @samp{-pedantic}; this option reverses that. This behavior and this
1195 option are superseded by @samp{-pedantic}, which works as it does for GNU C.
1198 Enable automatic template instantiation. This option also implies
1199 @samp{-fno-implicit-templates}. @xref{Template Instantiation}, for more
1203 Disable generation of information about every class with virtual
1204 functions for use by the C++ runtime type identification features
1205 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1206 of the language, you can save some space by using this flag. Note that
1207 exception handling uses the same information, but it will generate it as
1210 @item -fstrict-prototype
1211 Within an @samp{extern "C"} linkage specification, treat a function
1212 declaration with no arguments, such as @samp{int foo ();}, as declaring
1213 the function to take no arguments. Normally, such a declaration means
1214 that the function @code{foo} can take any combination of arguments, as
1215 in C. @samp{-pedantic} implies @samp{-fstrict-prototype} unless
1216 overridden with @samp{-fno-strict-prototype}.
1218 Specifying this option will also suppress implicit declarations of
1221 This flag no longer affects declarations with C++ linkage.
1224 @itemx -fno-squangle
1225 @samp{-fsquangle} will enable a compressed form of name mangling for
1226 identifiers. In particular, it helps to shorten very long names by recognizing
1227 types and class names which occur more than once, replacing them with special
1228 short ID codes. This option also requires any C++ libraries being used to
1229 be compiled with this option as well. The compiler has this disabled (the
1230 equivalent of @samp{-fno-squangle}) by default.
1232 Like all options that change the ABI, all C++ code, @emph{including
1233 libgcc.a} must be built with the same setting of this option.
1235 @item -ftemplate-depth-@var{n}
1236 Set the maximum instantiation depth for template classes to @var{n}.
1237 A limit on the template instantiation depth is needed to detect
1238 endless recursions during template class instantiation. ANSI/ISO C++
1239 conforming programs must not rely on a maximum depth greater than 17.
1241 @item -fuse-cxa-atexit
1242 Register destructors for objects with static storage duration with the
1243 @code{__cxa_atexit} function rather than the @code{atexit} function.
1244 This option is required for fully standards-compliant handling of static
1245 destructors, but will only work if your C library supports
1246 @code{__cxa_atexit}.
1248 @item -fvtable-thunks
1249 Use @samp{thunks} to implement the virtual function dispatch table
1250 (@samp{vtable}). The traditional (cfront-style) approach to
1251 implementing vtables was to store a pointer to the function and two
1252 offsets for adjusting the @samp{this} pointer at the call site. Newer
1253 implementations store a single pointer to a @samp{thunk} function which
1254 does any necessary adjustment and then calls the target function.
1256 This option also enables a heuristic for controlling emission of
1257 vtables; if a class has any non-inline virtual functions, the vtable
1258 will be emitted in the translation unit containing the first one of
1261 Like all options that change the ABI, all C++ code, @emph{including
1262 libgcc.a} must be built with the same setting of this option.
1265 Do not search for header files in the standard directories specific to
1266 C++, but do still search the other standard directories. (This option
1267 is used when building the C++ library.)
1270 In addition, these optimization, warning, and code generation options
1271 have meanings only for C++ programs:
1274 @item -fno-default-inline
1275 Do not assume @samp{inline} for functions defined inside a class scope.
1276 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1277 functions will have linkage like inline functions; they just won't be
1280 @item -Wctor-dtor-privacy (C++ only)
1281 Warn when a class seems unusable, because all the constructors or
1282 destructors in a class are private and the class has no friends or
1283 public static member functions.
1285 @item -Wnon-virtual-dtor (C++ only)
1286 Warn when a class declares a non-virtual destructor that should probably
1287 be virtual, because it looks like the class will be used polymorphically.
1289 @item -Wreorder (C++ only)
1290 @cindex reordering, warning
1291 @cindex warning for reordering of member initializers
1292 Warn when the order of member initializers given in the code does not
1293 match the order in which they must be executed. For instance:
1299 A(): j (0), i (1) @{ @}
1303 Here the compiler will warn that the member initializers for @samp{i}
1304 and @samp{j} will be rearranged to match the declaration order of the
1308 The following @samp{-W@dots{}} options are not affected by @samp{-Wall}.
1311 @item -Weffc++ (C++ only)
1312 Warn about violations of various style guidelines from Scott Meyers'
1313 @cite{Effective C++} books. If you use this option, you should be aware
1314 that the standard library headers do not obey all of these guidelines;
1315 you can use @samp{grep -v} to filter out those warnings.
1317 @item -Wno-deprecated (C++ only)
1318 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1320 @item -Wno-non-template-friend (C++ only)
1321 Disable warnings when non-templatized friend functions are declared
1322 within a template. With the advent of explicit template specification
1323 support in g++, if the name of the friend is an unqualified-id (ie,
1324 @samp{friend foo(int)}), the C++ language specification demands that the
1325 friend declare or define an ordinary, nontemplate function. (Section
1326 14.5.3). Before g++ implemented explicit specification, unqualified-ids
1327 could be interpreted as a particular specialization of a templatized
1328 function. Because this non-conforming behavior is no longer the default
1329 behavior for g++, @samp{-Wnon-template-friend} allows the compiler to
1330 check existing code for potential trouble spots, and is on by default.
1331 This new compiler behavior can also be turned off with the flag
1332 @samp{-fguiding-decls}, which activates the older, non-specification
1333 compiler code, or with @samp{-Wno-non-template-friend} which keeps the
1334 conformant compiler code but disables the helpful warning.
1336 @item -Wold-style-cast (C++ only)
1337 Warn if an old-style (C-style) cast is used within a C++ program. The
1338 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1339 @samp{const_cast}) are less vulnerable to unintended effects.
1341 @item -Woverloaded-virtual (C++ only)
1342 @cindex overloaded virtual fn, warning
1343 @cindex warning for overloaded virtual fn
1344 Warn when a derived class function declaration may be an error in
1345 defining a virtual function. In a derived class, the
1346 definitions of virtual functions must match the type signature of a
1347 virtual function declared in the base class. With this option, the
1348 compiler warns when you define a function with the same name as a
1349 virtual function, but with a type signature that does not match any
1350 declarations from the base class.
1352 @item -Wno-pmf-conversions (C++ only)
1353 Disable the diagnostic for converting a bound pointer to member function
1356 @item -Wsign-promo (C++ only)
1357 Warn when overload resolution chooses a promotion from unsigned or
1358 enumeral type to a signed type over a conversion to an unsigned type of
1359 the same size. Previous versions of g++ would try to preserve
1360 unsignedness, but the standard mandates the current behavior.
1362 @item -Wsynth (C++ only)
1363 @cindex warning for synthesized methods
1364 @cindex synthesized methods, warning
1365 Warn when g++'s synthesis behavior does not match that of cfront. For
1371 A& operator = (int);
1381 In this example, g++ will synthesize a default @samp{A& operator =
1382 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1385 @node Warning Options
1386 @section Options to Request or Suppress Warnings
1387 @cindex options to control warnings
1388 @cindex warning messages
1389 @cindex messages, warning
1390 @cindex suppressing warnings
1392 Warnings are diagnostic messages that report constructions which
1393 are not inherently erroneous but which are risky or suggest there
1394 may have been an error.
1396 You can request many specific warnings with options beginning @samp{-W},
1397 for example @samp{-Wimplicit} to request warnings on implicit
1398 declarations. Each of these specific warning options also has a
1399 negative form beginning @samp{-Wno-} to turn off warnings;
1400 for example, @samp{-Wno-implicit}. This manual lists only one of the
1401 two forms, whichever is not the default.
1403 These options control the amount and kinds of warnings produced by GCC:
1406 @cindex syntax checking
1408 Check the code for syntax errors, but don't do anything beyond that.
1411 Issue all the warnings demanded by strict ANSI C and ISO C++;
1412 reject all programs that use forbidden extensions.
1414 Valid ANSI C and ISO C++ programs should compile properly with or without
1415 this option (though a rare few will require @samp{-ansi}). However,
1416 without this option, certain GNU extensions and traditional C and C++
1417 features are supported as well. With this option, they are rejected.
1419 @samp{-pedantic} does not cause warning messages for use of the
1420 alternate keywords whose names begin and end with @samp{__}. Pedantic
1421 warnings are also disabled in the expression that follows
1422 @code{__extension__}. However, only system header files should use
1423 these escape routes; application programs should avoid them.
1424 @xref{Alternate Keywords}.
1426 This option is not intended to be @i{useful}; it exists only to satisfy
1427 pedants who would otherwise claim that GCC fails to support the ANSI
1430 Some users try to use @samp{-pedantic} to check programs for strict ANSI
1431 C conformance. They soon find that it does not do quite what they want:
1432 it finds some non-ANSI practices, but not all---only those for which
1433 ANSI C @emph{requires} a diagnostic.
1435 A feature to report any failure to conform to ANSI C might be useful in
1436 some instances, but would require considerable additional work and would
1437 be quite different from @samp{-pedantic}. We don't have plans to
1438 support such a feature in the near future.
1440 @item -pedantic-errors
1441 Like @samp{-pedantic}, except that errors are produced rather than
1445 Inhibit all warning messages.
1448 Inhibit warning messages about the use of @samp{#import}.
1450 @item -Wchar-subscripts
1451 Warn if an array subscript has type @code{char}. This is a common cause
1452 of error, as programmers often forget that this type is signed on some
1456 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1457 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1460 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1461 the arguments supplied have types appropriate to the format string
1464 @item -Wimplicit-int
1465 Warn when a declaration does not specify a type.
1467 @item -Wimplicit-function-declaration
1468 @itemx -Werror-implicit-function-declaration
1469 Give a warning (or error) whenever a function is used before being
1473 Same as @samp{-Wimplicit-int} and @samp{-Wimplicit-function-}@*
1477 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1478 function with external linkage, returning int, taking either zero
1479 arguments, two, or three arguments of appropriate types.
1482 Warn if a multicharacter constant (@samp{'FOOF'}) is used. Usually they
1483 indicate a typo in the user's code, as they have implementation-defined
1484 values, and should not be used in portable code.
1487 Warn if parentheses are omitted in certain contexts, such
1488 as when there is an assignment in a context where a truth value
1489 is expected, or when operators are nested whose precedence people
1490 often get confused about.
1492 Also warn about constructions where there may be confusion to which
1493 @code{if} statement an @code{else} branch belongs. Here is an example of
1506 In C, every @code{else} branch belongs to the innermost possible @code{if}
1507 statement, which in this example is @code{if (b)}. This is often not
1508 what the programmer expected, as illustrated in the above example by
1509 indentation the programmer chose. When there is the potential for this
1510 confusion, GNU C will issue a warning when this flag is specified.
1511 To eliminate the warning, add explicit braces around the innermost
1512 @code{if} statement so there is no way the @code{else} could belong to
1513 the enclosing @code{if}. The resulting code would look like this:
1528 Warn whenever a function is defined with a return-type that defaults
1529 to @code{int}. Also warn about any @code{return} statement with no
1530 return-value in a function whose return-type is not @code{void}.
1533 Warn whenever a @code{switch} statement has an index of enumeral type
1534 and lacks a @code{case} for one or more of the named codes of that
1535 enumeration. (The presence of a @code{default} label prevents this
1536 warning.) @code{case} labels outside the enumeration range also
1537 provoke warnings when this option is used.
1540 Warn if any trigraphs are encountered (assuming they are enabled).
1543 Warn whenever a variable is unused aside from its declaration,
1544 whenever a function is declared static but never defined, whenever a
1545 label is declared but not used, and whenever a statement computes a
1546 result that is explicitly not used.
1548 In order to get a warning about an unused function parameter, you must
1549 specify both @samp{-W} and @samp{-Wunused}.
1551 To suppress this warning for an expression, simply cast it to void. For
1552 unused variables, parameters and labels, use the @samp{unused} attribute
1553 (@pxref{Variable Attributes}).
1555 @item -Wuninitialized
1556 Warn if an automatic variable is used without first being initialized or
1557 if a variable may be clobbered by a @code{setjmp} call.
1559 These warnings are possible only in optimizing compilation,
1560 because they require data flow information that is computed only
1561 when optimizing. If you don't specify @samp{-O}, you simply won't
1564 These warnings occur only for variables that are candidates for
1565 register allocation. Therefore, they do not occur for a variable that
1566 is declared @code{volatile}, or whose address is taken, or whose size
1567 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
1568 structures, unions or arrays, even when they are in registers.
1570 Note that there may be no warning about a variable that is used only
1571 to compute a value that itself is never used, because such
1572 computations may be deleted by data flow analysis before the warnings
1575 These warnings are made optional because GCC is not smart
1576 enough to see all the reasons why the code might be correct
1577 despite appearing to have an error. Here is one example of how
1596 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
1597 always initialized, but GCC doesn't know this. Here is
1598 another common case:
1603 if (change_y) save_y = y, y = new_y;
1605 if (change_y) y = save_y;
1610 This has no bug because @code{save_y} is used only if it is set.
1612 @cindex @code{longjmp} warnings
1613 This option also warns when a nonvolatile automatic variable might be
1614 changed by a call to @code{longjmp}. These warnings as well are possible
1615 only in optimizing compilation.
1617 The compiler sees only the calls to @code{setjmp}. It cannot know
1618 where @code{longjmp} will be called; in fact, a signal handler could
1619 call it at any point in the code. As a result, you may get a warning
1620 even when there is in fact no problem because @code{longjmp} cannot
1621 in fact be called at the place which would cause a problem.
1623 Some spurious warnings can be avoided if you declare all the functions
1624 you use that never return as @code{noreturn}. @xref{Function
1627 @item -Wreorder (C++ only)
1628 @cindex reordering, warning
1629 @cindex warning for reordering of member initializers
1630 Warn when the order of member initializers given in the code does not
1631 match the order in which they must be executed. For instance:
1633 @item -Wunknown-pragmas
1634 @cindex warning for unknown pragmas
1635 @cindex unknown pragmas, warning
1636 @cindex pragmas, warning of unknown
1637 Warn when a #pragma directive is encountered which is not understood by
1638 GCC. If this command line option is used, warnings will even be issued
1639 for unknown pragmas in system header files. This is not the case if
1640 the warnings were only enabled by the @samp{-Wall} command line option.
1643 All of the above @samp{-W} options combined. This enables all the
1644 warnings about constructions that some users consider questionable, and
1645 that are easy to avoid (or modify to prevent the warning), even in
1646 conjunction with macros.
1649 The following @samp{-W@dots{}} options are not implied by @samp{-Wall}.
1650 Some of them warn about constructions that users generally do not
1651 consider questionable, but which occasionally you might wish to check
1652 for; others warn about constructions that are necessary or hard to avoid
1653 in some cases, and there is no simple way to modify the code to suppress
1658 Print extra warning messages for these events:
1662 A function can return either with or without a value. (Falling
1663 off the end of the function body is considered returning without
1664 a value.) For example, this function would evoke such a
1678 An expression-statement or the left-hand side of a comma expression
1679 contains no side effects.
1680 To suppress the warning, cast the unused expression to void.
1681 For example, an expression such as @samp{x[i,j]} will cause a warning,
1682 but @samp{x[(void)i,j]} will not.
1685 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
1688 A comparison like @samp{x<=y<=z} appears; this is equivalent to
1689 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
1690 that of ordinary mathematical notation.
1693 Storage-class specifiers like @code{static} are not the first things in
1694 a declaration. According to the C Standard, this usage is obsolescent.
1697 If @samp{-Wall} or @samp{-Wunused} is also specified, warn about unused
1701 A comparison between signed and unsigned values could produce an
1702 incorrect result when the signed value is converted to unsigned.
1703 (But don't warn if @samp{-Wno-sign-compare} is also specified.)
1706 An aggregate has a partly bracketed initializer.
1707 For example, the following code would evoke such a warning,
1708 because braces are missing around the initializer for @code{x.h}:
1711 struct s @{ int f, g; @};
1712 struct t @{ struct s h; int i; @};
1713 struct t x = @{ 1, 2, 3 @};
1717 An aggregate has an initializer which does not initialize all members.
1718 For example, the following code would cause such a warning, because
1719 @code{x.h} would be implicitly initialized to zero:
1722 struct s @{ int f, g, h; @};
1723 struct s x = @{ 3, 4 @};
1728 Warn if floating point values are used in equality comparisons.
1730 The idea behind this is that sometimes it is convenient (for the
1731 programmer) to consider floating-point values as approximations to
1732 infinitely precise real numbers. If you are doing this, then you need
1733 to compute (by analysing the code, or in some other way) the maximum or
1734 likely maximum error that the computation introduces, and allow for it
1735 when performing comparisons (and when producing output, but that's a
1736 different problem). In particular, instead of testing for equality, you
1737 would check to see whether the two values have ranges that overlap; and
1738 this is done with the relational operators, so equality comparisons are
1741 @item -Wtraditional (C only)
1742 Warn about certain constructs that behave differently in traditional and
1747 Macro arguments occurring within string constants in the macro body.
1748 These would substitute the argument in traditional C, but are part of
1749 the constant in ANSI C.
1752 A function declared external in one block and then used after the end of
1756 A @code{switch} statement has an operand of type @code{long}.
1759 A non-@code{static} function declaration follows a @code{static} one.
1760 This construct is not accepted by some traditional C compilers.
1763 The ANSI type of an integer constant has a different width or
1764 signedness from its traditional type. This warning is only issued if
1765 the base of the constant is ten. I.e. hexadecimal or octal values, which
1766 typically represent bit patterns, are not warned about.
1769 Usage of ANSI string concatenation is detected.
1773 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
1776 Warn whenever a local variable shadows another local variable.
1778 @item -Wid-clash-@var{len}
1779 Warn whenever two distinct identifiers match in the first @var{len}
1780 characters. This may help you prepare a program that will compile
1781 with certain obsolete, brain-damaged compilers.
1783 @item -Wlarger-than-@var{len}
1784 Warn whenever an object of larger than @var{len} bytes is defined.
1786 @item -Wpointer-arith
1787 Warn about anything that depends on the ``size of'' a function type or
1788 of @code{void}. GNU C assigns these types a size of 1, for
1789 convenience in calculations with @code{void *} pointers and pointers
1792 @item -Wbad-function-cast (C only)
1793 Warn whenever a function call is cast to a non-matching type.
1794 For example, warn if @code{int malloc()} is cast to @code{anything *}.
1797 Warn whenever a pointer is cast so as to remove a type qualifier from
1798 the target type. For example, warn if a @code{const char *} is cast
1799 to an ordinary @code{char *}.
1802 Warn whenever a pointer is cast such that the required alignment of the
1803 target is increased. For example, warn if a @code{char *} is cast to
1804 an @code{int *} on machines where integers can only be accessed at
1805 two- or four-byte boundaries.
1807 @item -Wwrite-strings
1808 Give string constants the type @code{const char[@var{length}]} so that
1809 copying the address of one into a non-@code{const} @code{char *}
1810 pointer will get a warning. These warnings will help you find at
1811 compile time code that can try to write into a string constant, but
1812 only if you have been very careful about using @code{const} in
1813 declarations and prototypes. Otherwise, it will just be a nuisance;
1814 this is why we did not make @samp{-Wall} request these warnings.
1817 Warn if a prototype causes a type conversion that is different from what
1818 would happen to the same argument in the absence of a prototype. This
1819 includes conversions of fixed point to floating and vice versa, and
1820 conversions changing the width or signedness of a fixed point argument
1821 except when the same as the default promotion.
1823 Also, warn if a negative integer constant expression is implicitly
1824 converted to an unsigned type. For example, warn about the assignment
1825 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
1826 casts like @code{(unsigned) -1}.
1828 @item -Wsign-compare
1829 @cindex warning for comparison of signed and unsigned values
1830 @cindex comparison of signed and unsigned values, warning
1831 @cindex signed and unsigned values, comparison warning
1832 Warn when a comparison between signed and unsigned values could produce
1833 an incorrect result when the signed value is converted to unsigned.
1834 This warning is also enabled by @samp{-W}; to get the other warnings
1835 of @samp{-W} without this warning, use @samp{-W -Wno-sign-compare}.
1837 @item -Waggregate-return
1838 Warn if any functions that return structures or unions are defined or
1839 called. (In languages where you can return an array, this also elicits
1842 @item -Wstrict-prototypes (C only)
1843 Warn if a function is declared or defined without specifying the
1844 argument types. (An old-style function definition is permitted without
1845 a warning if preceded by a declaration which specifies the argument
1848 @item -Wmissing-prototypes (C only)
1849 Warn if a global function is defined without a previous prototype
1850 declaration. This warning is issued even if the definition itself
1851 provides a prototype. The aim is to detect global functions that fail
1852 to be declared in header files.
1854 @item -Wmissing-declarations
1855 Warn if a global function is defined without a previous declaration.
1856 Do so even if the definition itself provides a prototype.
1857 Use this option to detect global functions that are not declared in
1860 @item -Wmissing-noreturn
1861 Warn about functions which might be candidates for attribute @code{noreturn}.
1862 Note these are only possible candidates, not absolute ones. Care should
1863 be taken to manually verify functions actually do not ever return before
1864 adding the @code{noreturn} attribute, otherwise subtle code generation
1865 bugs could be introduced.
1868 Warn if a structure is given the packed attribute, but the packed
1869 attribute has no effect on the layout or size of the structure.
1870 Such structures may be mis-aligned for little benefit. For
1871 instance, in this code, the variable @code{f.x} in @code{struct bar}
1872 will be misaligned even though @code{struct bar} does not itself
1873 have the packed attribute:
1880 @} __attribute__((packed));
1889 Warn if padding is included in a structure, either to align an element
1890 of the structure or to align the whole structure. Sometimes when this
1891 happens it is possible to rearrange the fields of the structure to
1892 reduce the padding and so make the structure smaller.
1894 @item -Wredundant-decls
1895 Warn if anything is declared more than once in the same scope, even in
1896 cases where multiple declaration is valid and changes nothing.
1898 @item -Wnested-externs (C only)
1899 Warn if an @code{extern} declaration is encountered within a function.
1901 @item -Wunreachable-code
1902 Warn if the compiler detects that code will never be executed.
1904 This option is intended to warn when the compiler detects that at
1905 least a whole line of source code will never be executed, because
1906 some condition is never satisfied or because it is after a
1907 procedure that never returns.
1909 It is possible for this option to produce a warning even though there
1910 are circumstances under which part of the affected line can be executed,
1911 so care should be taken when removing apparently-unreachable code.
1913 For instance, when a function is inlined, a warning may mean that the
1914 line is unreachable in only one inlined copy of the function.
1916 This option is not made part of @samp{-Wall} because in a debugging
1917 version of a program there is often substantial code which checks
1918 correct functioning of the program and is, hopefully, unreachable
1919 because the program does work. Another common use of unreachable
1920 code is to provide behaviour which is selectable at compile-time.
1923 Warn if a function can not be inlined and it was declared as inline.
1926 Warn if @samp{long long} type is used. This is default. To inhibit
1927 the warning messages, use @samp{-Wno-long-long}. Flags
1928 @samp{-Wlong-long} and @samp{-Wno-long-long} are taken into account
1929 only when @samp{-pedantic} flag is used.
1932 Make all warnings into errors.
1935 @node Debugging Options
1936 @section Options for Debugging Your Program or GCC
1937 @cindex options, debugging
1938 @cindex debugging information options
1940 GCC has various special options that are used for debugging
1941 either your program or GCC:
1945 Produce debugging information in the operating system's native format
1946 (stabs, COFF, XCOFF, or DWARF). GDB can work with this debugging
1949 On most systems that use stabs format, @samp{-g} enables use of extra
1950 debugging information that only GDB can use; this extra information
1951 makes debugging work better in GDB but will probably make other debuggers
1953 refuse to read the program. If you want to control for certain whether
1954 to generate the extra information, use @samp{-gstabs+}, @samp{-gstabs},
1955 @samp{-gxcoff+}, @samp{-gxcoff}, @samp{-gdwarf-1+}, or @samp{-gdwarf-1}
1958 Unlike most other C compilers, GCC allows you to use @samp{-g} with
1959 @samp{-O}. The shortcuts taken by optimized code may occasionally
1960 produce surprising results: some variables you declared may not exist
1961 at all; flow of control may briefly move where you did not expect it;
1962 some statements may not be executed because they compute constant
1963 results or their values were already at hand; some statements may
1964 execute in different places because they were moved out of loops.
1966 Nevertheless it proves possible to debug optimized output. This makes
1967 it reasonable to use the optimizer for programs that might have bugs.
1969 The following options are useful when GCC is generated with the
1970 capability for more than one debugging format.
1973 Produce debugging information for use by GDB. This means to use the
1974 most expressive format available (DWARF 2, stabs, or the native format
1975 if neither of those are supported), including GDB extensions if at all
1979 Produce debugging information in stabs format (if that is supported),
1980 without GDB extensions. This is the format used by DBX on most BSD
1981 systems. On MIPS, Alpha and System V Release 4 systems this option
1982 produces stabs debugging output which is not understood by DBX or SDB.
1983 On System V Release 4 systems this option requires the GNU assembler.
1986 Produce debugging information in stabs format (if that is supported),
1987 using GNU extensions understood only by the GNU debugger (GDB). The
1988 use of these extensions is likely to make other debuggers crash or
1989 refuse to read the program.
1992 Produce debugging information in COFF format (if that is supported).
1993 This is the format used by SDB on most System V systems prior to
1997 Produce debugging information in XCOFF format (if that is supported).
1998 This is the format used by the DBX debugger on IBM RS/6000 systems.
2001 Produce debugging information in XCOFF format (if that is supported),
2002 using GNU extensions understood only by the GNU debugger (GDB). The
2003 use of these extensions is likely to make other debuggers crash or
2004 refuse to read the program, and may cause assemblers other than the GNU
2005 assembler (GAS) to fail with an error.
2008 Produce debugging information in DWARF version 1 format (if that is
2009 supported). This is the format used by SDB on most System V Release 4
2013 Produce debugging information in DWARF version 1 format (if that is
2014 supported), using GNU extensions understood only by the GNU debugger
2015 (GDB). The use of these extensions is likely to make other debuggers
2016 crash or refuse to read the program.
2019 Produce debugging information in DWARF version 2 format (if that is
2020 supported). This is the format used by DBX on IRIX 6.
2023 @itemx -ggdb@var{level}
2024 @itemx -gstabs@var{level}
2025 @itemx -gcoff@var{level}
2026 @itemx -gxcoff@var{level}
2027 @itemx -gdwarf@var{level}
2028 @itemx -gdwarf-2@var{level}
2029 Request debugging information and also use @var{level} to specify how
2030 much information. The default level is 2.
2032 Level 1 produces minimal information, enough for making backtraces in
2033 parts of the program that you don't plan to debug. This includes
2034 descriptions of functions and external variables, but no information
2035 about local variables and no line numbers.
2037 Level 3 includes extra information, such as all the macro definitions
2038 present in the program. Some debuggers support macro expansion when
2043 Generate extra code to write profile information suitable for the
2044 analysis program @code{prof}. You must use this option when compiling
2045 the source files you want data about, and you must also use it when
2048 @cindex @code{gprof}
2050 Generate extra code to write profile information suitable for the
2051 analysis program @code{gprof}. You must use this option when compiling
2052 the source files you want data about, and you must also use it when
2057 Generate extra code to write profile information for basic blocks, which will
2058 record the number of times each basic block is executed, the basic block start
2059 address, and the function name containing the basic block. If @samp{-g} is
2060 used, the line number and filename of the start of the basic block will also be
2061 recorded. If not overridden by the machine description, the default action is
2062 to append to the text file @file{bb.out}.
2064 This data could be analyzed by a program like @code{tcov}. Note,
2065 however, that the format of the data is not what @code{tcov} expects.
2066 Eventually GNU @code{gprof} should be extended to process this data.
2069 Makes the compiler print out each function name as it is compiled, and
2070 print some statistics about each pass when it finishes.
2073 Generate extra code to profile basic blocks. Your executable will
2074 produce output that is a superset of that produced when @samp{-a} is
2075 used. Additional output is the source and target address of the basic
2076 blocks where a jump takes place, the number of times a jump is executed,
2077 and (optionally) the complete sequence of basic blocks being executed.
2078 The output is appended to file @file{bb.out}.
2080 You can examine different profiling aspects without recompilation. Your
2081 executable will read a list of function names from file @file{bb.in}.
2082 Profiling starts when a function on the list is entered and stops when
2083 that invocation is exited. To exclude a function from profiling, prefix
2084 its name with `-'. If a function name is not unique, you can
2085 disambiguate it by writing it in the form
2086 @samp{/path/filename.d:functionname}. Your executable will write the
2087 available paths and filenames in file @file{bb.out}.
2089 Several function names have a special meaning:
2092 Write source, target and frequency of jumps to file @file{bb.out}.
2093 @item __bb_hidecall__
2094 Exclude function calls from frequency count.
2095 @item __bb_showret__
2096 Include function returns in frequency count.
2098 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
2099 The file will be compressed using the program @samp{gzip}, which must
2100 exist in your @code{PATH}. On systems without the @samp{popen}
2101 function, the file will be named @file{bbtrace} and will not be
2102 compressed. @strong{Profiling for even a few seconds on these systems
2103 will produce a very large file.} Note: @code{__bb_hidecall__} and
2104 @code{__bb_showret__} will not affect the sequence written to
2108 Here's a short example using different profiling parameters
2109 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
2110 1 and 2 and is called twice from block 3 of function @code{main}. After
2111 the calls, block 3 transfers control to block 4 of @code{main}.
2113 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
2114 the following sequence of blocks is written to file @file{bbtrace.gz}:
2115 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
2116 the return is to a point inside the block and not to the top. The
2117 block address 0 always indicates, that control is transferred
2118 to the trace from somewhere outside the observed functions. With
2119 @samp{-foo} added to @file{bb.in}, the blocks of function
2120 @code{foo} are removed from the trace, so only 0 3 4 remains.
2122 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
2123 jump frequencies will be written to file @file{bb.out}. The
2124 frequencies are obtained by constructing a trace of blocks
2125 and incrementing a counter for every neighbouring pair of blocks
2126 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2130 Jump from block 0x0 to block 0x3 executed 1 time(s)
2131 Jump from block 0x3 to block 0x1 executed 1 time(s)
2132 Jump from block 0x1 to block 0x2 executed 2 time(s)
2133 Jump from block 0x2 to block 0x1 executed 1 time(s)
2134 Jump from block 0x2 to block 0x4 executed 1 time(s)
2137 With @code{__bb_hidecall__}, control transfer due to call instructions
2138 is removed from the trace, that is the trace is cut into three parts: 0
2139 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2140 to return instructions is added to the trace. The trace becomes: 0 3 1
2141 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2142 written to @file{bbtrace.gz}. It is solely used for counting jump
2145 @item -fprofile-arcs
2146 Instrument @dfn{arcs} during compilation. For each function of your
2147 program, GCC creates a program flow graph, then finds a spanning tree
2148 for the graph. Only arcs that are not on the spanning tree have to be
2149 instrumented: the compiler adds code to count the number of times that these
2150 arcs are executed. When an arc is the only exit or only entrance to a
2151 block, the instrumentation code can be added to the block; otherwise, a
2152 new basic block must be created to hold the instrumentation code.
2154 Since not every arc in the program must be instrumented, programs
2155 compiled with this option run faster than programs compiled with
2156 @samp{-a}, which adds instrumentation code to every basic block in the
2157 program. The tradeoff: since @code{gcov} does not have
2158 execution counts for all branches, it must start with the execution
2159 counts for the instrumented branches, and then iterate over the program
2160 flow graph until the entire graph has been solved. Hence, @code{gcov}
2161 runs a little more slowly than a program which uses information from
2164 @samp{-fprofile-arcs} also makes it possible to estimate branch
2165 probabilities, and to calculate basic block execution counts. In
2166 general, basic block execution counts do not give enough information to
2167 estimate all branch probabilities. When the compiled program exits, it
2168 saves the arc execution counts to a file called
2169 @file{@var{sourcename}.da}. Use the compiler option
2170 @samp{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2171 Control Optimization}) when recompiling, to optimize using estimated
2172 branch probabilities.
2175 @item -ftest-coverage
2176 Create data files for the @code{gcov} code-coverage utility
2177 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2178 The data file names begin with the name of your source file:
2181 @item @var{sourcename}.bb
2182 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2183 associate basic block execution counts with line numbers.
2185 @item @var{sourcename}.bbg
2186 A list of all arcs in the program flow graph. This allows @code{gcov}
2187 to reconstruct the program flow graph, so that it can compute all basic
2188 block and arc execution counts from the information in the
2189 @code{@var{sourcename}.da} file (this last file is the output from
2190 @samp{-fprofile-arcs}).
2193 @item -d@var{letters}
2194 Says to make debugging dumps during compilation at times specified by
2195 @var{letters}. This is used for debugging the compiler. The file names
2196 for most of the dumps are made by appending a pass number and a word to
2197 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.jump}).
2198 Here are the possible letters for use in @var{letters}, and their meanings:
2202 Annotate the assembler output with miscellaneous debugging information.
2204 Dump after computing branch probabilities, to @file{@var{file}.10.bp}.
2206 Dump after block reordering, to @file{@var{file}.20.bbro}.
2208 Dump after instruction combination, to the file @file{@var{file}.12.combine}.
2210 Dump after delayed branch scheduling, to @file{@var{file}.24.dbr}.
2212 Dump all macro definitions, at the end of preprocessing, in addition to
2215 Dump after SSA optimizations, to @file{@var{file}.05.ssa} and
2216 @file{@var{file}.06.ussa}.
2218 Dump after flow analysis, to @file{@var{file}.11.flow}.
2220 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.04.addressof}.
2222 Dump after global register allocation, to @file{@var{file}.16.greg}.
2224 Dump after GCSE, to @file{@var{file}.07.gcse}.
2226 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
2228 Dump after first jump optimization, to @file{@var{file}.02.jump}.
2230 Dump after last jump optimization, to @file{@var{file}.22.jump2}.
2232 Dump after conversion from registers to stack, to @file{@var{file}.25.stack}.
2234 Dump after local register allocation, to @file{@var{file}.15.lreg}.
2236 Dump after loop optimization, to @file{@var{file}.08.loop}.
2238 Dump after performing the machine dependent reorganisation pass, to
2239 @file{@var{file}.23.mach}.
2241 Dump after register renumbering, to @file{@var{file}.21.rnreg}.
2243 Dump after the register move pass, to @file{@var{file}.13.regmove}.
2245 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2247 Dump after the second instruction scheduling pass, to
2248 @file{@var{file}.19.sched2}.
2250 Dump after CSE (including the jump optimization that sometimes follows
2251 CSE), to @file{@var{file}.03.cse}.
2253 Dump after the first instruction scheduling pass, to
2254 @file{@var{file}.14.sched}.
2256 Dump after the second CSE pass (including the jump optimization that
2257 sometimes follows CSE), to @file{@var{file}.09.cse2}.
2259 Dump after the second flow pass, to @file{@var{file}.17.flow2}.
2261 Dump after the peephold pass, to @file{@var{file}.18.peephole2}.
2263 Produce all the dumps listed above.
2265 Print statistics on memory usage, at the end of the run, to
2268 Annotate the assembler output with a comment indicating which
2269 pattern and alternative was used. The length of each instruction is
2272 For each of the other indicated dump files (except for
2273 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
2274 suitible for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
2276 Dump after the second flow pass to @file{@var{file}.14.flow2}.
2278 Just generate RTL for a function instead of compiling it. Usually used
2281 Dump debugging information during parsing, to standard error.
2283 Dump after the peephole2 pass to @file{@var{file}.15.peephole2}.
2286 @item -fdump-unnumbered
2287 When doing debugging dumps (see -d option above), suppress instruction
2288 numbers and line number note output. This makes it more feasible to
2289 use diff on debugging dumps for compiler invokations with different
2290 options, in particular with and without -g.
2292 @item -fdump-translation-unit-@var{file} (C++ only)
2293 Dump a representation of the tree structure for the entire translation
2296 @item -fpretend-float
2297 When running a cross-compiler, pretend that the target machine uses the
2298 same floating point format as the host machine. This causes incorrect
2299 output of the actual floating constants, but the actual instruction
2300 sequence will probably be the same as GCC would make when running on
2304 Store the usual ``temporary'' intermediate files permanently; place them
2305 in the current directory and name them based on the source file. Thus,
2306 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
2307 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}.
2310 Report the CPU time taken by each subprocess in the compilation
2311 sequence. For C source files, this is the preprocessor, compiler
2312 proper, and assembler. The output looks like this:
2320 The first number on each line is the ``user time,'' that is time spent
2321 executing the program itself. The second number is ``system time,''
2322 time spent executing operating system routines on behalf of the program.
2323 Both numbers are in seconds.
2325 @item -print-file-name=@var{library}
2326 Print the full absolute name of the library file @var{library} that
2327 would be used when linking---and don't do anything else. With this
2328 option, GCC does not compile or link anything; it just prints the
2331 @item -print-prog-name=@var{program}
2332 Like @samp{-print-file-name}, but searches for a program such as @samp{cpp}.
2334 @item -print-libgcc-file-name
2335 Same as @samp{-print-file-name=libgcc.a}.
2337 This is useful when you use @samp{-nostdlib} or @samp{-nodefaultlibs}
2338 but you do want to link with @file{libgcc.a}. You can do
2341 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
2344 @item -print-search-dirs
2345 Print the name of the configured installation directory and a list of
2346 program and library directories gcc will search---and don't do anything else.
2348 This is useful when gcc prints the error message
2349 @samp{installation problem, cannot exec cpp: No such file or directory}.
2350 To resolve this you either need to put @file{cpp} and the other compiler
2351 components where gcc expects to find them, or you can set the environment
2352 variable @code{GCC_EXEC_PREFIX} to the directory where you installed them.
2353 Don't forget the trailing '/'.
2354 @xref{Environment Variables}.
2357 @node Optimize Options
2358 @section Options That Control Optimization
2359 @cindex optimize options
2360 @cindex options, optimization
2362 These options control various sorts of optimizations:
2367 Optimize. Optimizing compilation takes somewhat more time, and a lot
2368 more memory for a large function.
2370 Without @samp{-O}, the compiler's goal is to reduce the cost of
2371 compilation and to make debugging produce the expected results.
2372 Statements are independent: if you stop the program with a breakpoint
2373 between statements, you can then assign a new value to any variable or
2374 change the program counter to any other statement in the function and
2375 get exactly the results you would expect from the source code.
2377 Without @samp{-O}, the compiler only allocates variables declared
2378 @code{register} in registers. The resulting compiled code is a little
2379 worse than produced by PCC without @samp{-O}.
2381 With @samp{-O}, the compiler tries to reduce code size and execution
2384 When you specify @samp{-O}, the compiler turns on @samp{-fthread-jumps}
2385 and @samp{-fdefer-pop} on all machines. The compiler turns on
2386 @samp{-fdelayed-branch} on machines that have delay slots, and
2387 @samp{-fomit-frame-pointer} on machines that can support debugging even
2388 without a frame pointer. On some machines the compiler also turns
2389 on other flags.@refill
2392 Optimize even more. GCC performs nearly all supported optimizations
2393 that do not involve a space-speed tradeoff. The compiler does not
2394 perform loop unrolling or function inlining when you specify @samp{-O2}.
2395 As compared to @samp{-O}, this option increases both compilation time
2396 and the performance of the generated code.
2398 @samp{-O2} turns on all optional optimizations except for loop unrolling
2399 and function inlining. It also turns on the @samp{-fforce-mem} option
2400 on all machines and frame pointer elimination on machines where doing so
2401 does not interfere with debugging.
2404 Optimize yet more. @samp{-O3} turns on all optimizations specified by
2405 @samp{-O2} and also turns on the @samp{inline-functions} option.
2411 Optimize for size. @samp{-Os} enables all @samp{-O2} optimizations that
2412 do not typically increase code size. It also performs further
2413 optimizations designed to reduce code size.
2415 If you use multiple @samp{-O} options, with or without level numbers,
2416 the last such option is the one that is effective.
2419 Options of the form @samp{-f@var{flag}} specify machine-independent
2420 flags. Most flags have both positive and negative forms; the negative
2421 form of @samp{-ffoo} would be @samp{-fno-foo}. In the table below,
2422 only one of the forms is listed---the one which is not the default.
2423 You can figure out the other form by either removing @samp{no-} or
2428 Do not store floating point variables in registers, and inhibit other
2429 options that might change whether a floating point value is taken from a
2432 @cindex floating point precision
2433 This option prevents undesirable excess precision on machines such as
2434 the 68000 where the floating registers (of the 68881) keep more
2435 precision than a @code{double} is supposed to have. Similarly for the
2436 x86 architecture. For most programs, the excess precision does only
2437 good, but a few programs rely on the precise definition of IEEE floating
2438 point. Use @samp{-ffloat-store} for such programs, after modifying
2439 them to store all pertinent intermediate computations into variables.
2441 @item -fno-default-inline
2442 Do not make member functions inline by default merely because they are
2443 defined inside the class scope (C++ only). Otherwise, when you specify
2444 @w{@samp{-O}}, member functions defined inside class scope are compiled
2445 inline by default; i.e., you don't need to add @samp{inline} in front of
2446 the member function name.
2448 @item -fno-defer-pop
2449 Always pop the arguments to each function call as soon as that function
2450 returns. For machines which must pop arguments after a function call,
2451 the compiler normally lets arguments accumulate on the stack for several
2452 function calls and pops them all at once.
2455 Force memory operands to be copied into registers before doing
2456 arithmetic on them. This produces better code by making all memory
2457 references potential common subexpressions. When they are not common
2458 subexpressions, instruction combination should eliminate the separate
2459 register-load. The @samp{-O2} option turns on this option.
2462 Force memory address constants to be copied into registers before
2463 doing arithmetic on them. This may produce better code just as
2464 @samp{-fforce-mem} may.
2466 @item -fomit-frame-pointer
2467 Don't keep the frame pointer in a register for functions that
2468 don't need one. This avoids the instructions to save, set up and
2469 restore frame pointers; it also makes an extra register available
2470 in many functions. @strong{It also makes debugging impossible on
2473 @item -foptimize-sibling-calls
2474 Optimize sibling and tail recursive calls.
2477 On some machines, such as the Vax, this flag has no effect, because
2478 the standard calling sequence automatically handles the frame pointer
2479 and nothing is saved by pretending it doesn't exist. The
2480 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2481 whether a target machine supports this flag. @xref{Registers}.@refill
2484 On some machines, such as the Vax, this flag has no effect, because
2485 the standard calling sequence automatically handles the frame pointer
2486 and nothing is saved by pretending it doesn't exist. The
2487 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2488 whether a target machine supports this flag. @xref{Registers,,Register
2489 Usage, gcc.info, Using and Porting GCC}.@refill
2493 Don't pay attention to the @code{inline} keyword. Normally this option
2494 is used to keep the compiler from expanding any functions inline.
2495 Note that if you are not optimizing, no functions can be expanded inline.
2497 @item -finline-functions
2498 Integrate all simple functions into their callers. The compiler
2499 heuristically decides which functions are simple enough to be worth
2500 integrating in this way.
2502 If all calls to a given function are integrated, and the function is
2503 declared @code{static}, then the function is normally not output as
2504 assembler code in its own right.
2506 @item -finline-limit=@var{n}
2507 By default, gcc limits the size of functions that can be inlined. This flag
2508 allows the control of this limit for functions that are explicitly marked as
2509 inline (ie marked with the inline keyword or defined within the class
2510 definition in c++). @var{n} is the size of functions that can be inlined in
2511 number of pseudo instructions (not counting parameter handling). The default
2512 value of n is 10000. Increasing this value can result in more inlined code at
2513 the cost of compilation time and memory consumption. Decreasing usually makes
2514 the compilation faster and less code will be inlined (which presumably
2515 means slower programs). This option is particularly useful for programs that
2516 use inlining heavily such as those based on recursive templates with c++.
2518 @emph{Note:} pseudo instruction represents, in this particular context, an
2519 abstract measurement of function's size. In no way, it represents a count
2520 of assembly instructions and as such its exact meaning might change from one
2521 release to an another.
2523 @item -fkeep-inline-functions
2524 Even if all calls to a given function are integrated, and the function
2525 is declared @code{static}, nevertheless output a separate run-time
2526 callable version of the function. This switch does not affect
2527 @code{extern inline} functions.
2529 @item -fkeep-static-consts
2530 Emit variables declared @code{static const} when optimization isn't turned
2531 on, even if the variables aren't referenced.
2533 GCC enables this option by default. If you want to force the compiler to
2534 check if the variable was referenced, regardless of whether or not
2535 optimization is turned on, use the @samp{-fno-keep-static-consts} option.
2537 @item -fno-function-cse
2538 Do not put function addresses in registers; make each instruction that
2539 calls a constant function contain the function's address explicitly.
2541 This option results in less efficient code, but some strange hacks
2542 that alter the assembler output may be confused by the optimizations
2543 performed when this option is not used.
2546 This option allows GCC to violate some ANSI or IEEE rules and/or
2547 specifications in the interest of optimizing code for speed. For
2548 example, it allows the compiler to assume arguments to the @code{sqrt}
2549 function are non-negative numbers and that no floating-point values
2552 This option should never be turned on by any @samp{-O} option since
2553 it can result in incorrect output for programs which depend on
2554 an exact implementation of IEEE or ANSI rules/specifications for
2557 @item -fno-math-errno
2558 Do not set ERRNO after calling math functions that are executed
2559 with a single instruction, e.g., sqrt. A program that relies on
2560 IEEE exceptions for math error handling may want to use this flag
2561 for speed while maintaining IEEE arithmetic compatibility.
2563 The default is @samp{-fmath-errno}. The @samp{-ffast-math} option
2564 sets @samp{-fno-math-errno}.
2567 @c following causes underfulls.. they don't look great, but we deal.
2569 The following options control specific optimizations. The @samp{-O2}
2570 option turns on all of these optimizations except @samp{-funroll-loops}
2571 and @samp{-funroll-all-loops}. On most machines, the @samp{-O} option
2572 turns on the @samp{-fthread-jumps} and @samp{-fdelayed-branch} options,
2573 but specific machines may handle it differently.
2575 You can use the following flags in the rare cases when ``fine-tuning''
2576 of optimizations to be performed is desired.
2579 @item -fstrength-reduce
2580 Perform the optimizations of loop strength reduction and
2581 elimination of iteration variables.
2583 @item -fthread-jumps
2584 Perform optimizations where we check to see if a jump branches to a
2585 location where another comparison subsumed by the first is found. If
2586 so, the first branch is redirected to either the destination of the
2587 second branch or a point immediately following it, depending on whether
2588 the condition is known to be true or false.
2590 @item -fcse-follow-jumps
2591 In common subexpression elimination, scan through jump instructions
2592 when the target of the jump is not reached by any other path. For
2593 example, when CSE encounters an @code{if} statement with an
2594 @code{else} clause, CSE will follow the jump when the condition
2597 @item -fcse-skip-blocks
2598 This is similar to @samp{-fcse-follow-jumps}, but causes CSE to
2599 follow jumps which conditionally skip over blocks. When CSE
2600 encounters a simple @code{if} statement with no else clause,
2601 @samp{-fcse-skip-blocks} causes CSE to follow the jump around the
2602 body of the @code{if}.
2604 @item -frerun-cse-after-loop
2605 Re-run common subexpression elimination after loop optimizations has been
2608 @item -frerun-loop-opt
2609 Run the loop optimizer twice.
2612 Perform a global common subexpression elimination pass.
2613 This pass also performs global constant and copy propagation.
2615 @item -fdelete-null-pointer-checks
2616 Use global dataflow analysis to identify and eliminate useless null
2617 pointer checks. Programs which rely on NULL pointer dereferences @emph{not}
2618 halting the program may not work properly with this option. Use
2619 -fno-delete-null-pointer-checks to disable this optimizing for programs
2620 which depend on that behavior.
2623 @item -fexpensive-optimizations
2624 Perform a number of minor optimizations that are relatively expensive.
2626 @item -foptimize-register-moves
2628 Attempt to reassign register numbers in move instructions and as
2629 operands of other simple instructions in order to maximize the amount of
2630 register tying. This is especially helpful on machines with two-operand
2631 instructions. GCC enables this optimization by default with @samp{-O2}
2634 Note @code{-fregmove} and @code{-foptimize-register-moves} are the same
2637 @item -fdelayed-branch
2638 If supported for the target machine, attempt to reorder instructions
2639 to exploit instruction slots available after delayed branch
2642 @item -fschedule-insns
2643 If supported for the target machine, attempt to reorder instructions to
2644 eliminate execution stalls due to required data being unavailable. This
2645 helps machines that have slow floating point or memory load instructions
2646 by allowing other instructions to be issued until the result of the load
2647 or floating point instruction is required.
2649 @item -fschedule-insns2
2650 Similar to @samp{-fschedule-insns}, but requests an additional pass of
2651 instruction scheduling after register allocation has been done. This is
2652 especially useful on machines with a relatively small number of
2653 registers and where memory load instructions take more than one cycle.
2655 @item -ffunction-sections
2656 @itemx -fdata-sections
2657 Place each function or data item into its own section in the output
2658 file if the target supports arbitrary sections. The name of the
2659 function or the name of the data item determines the section's name
2662 Use these options on systems where the linker can perform optimizations
2663 to improve locality of reference in the instruction space. HPPA
2664 processors running HP-UX and Sparc processors running Solaris 2 have
2665 linkers with such optimizations. Other systems using the ELF object format
2666 as well as AIX may have these optimizations in the future.
2668 Only use these options when there are significant benefits from doing
2669 so. When you specify these options, the assembler and linker will
2670 create larger object and executable files and will also be slower.
2671 You will not be able to use @code{gprof} on all systems if you
2672 specify this option and you may have problems with debugging if
2673 you specify both this option and @samp{-g}.
2675 @item -fcaller-saves
2676 Enable values to be allocated in registers that will be clobbered by
2677 function calls, by emitting extra instructions to save and restore the
2678 registers around such calls. Such allocation is done only when it
2679 seems to result in better code than would otherwise be produced.
2681 This option is always enabled by default on certain machines, usually
2682 those which have no call-preserved registers to use instead.
2684 For all machines, optimization level 2 and higher enables this flag by
2687 @item -funroll-loops
2688 Perform the optimization of loop unrolling. This is only done for loops
2689 whose number of iterations can be determined at compile time or run time.
2690 @samp{-funroll-loops} implies both @samp{-fstrength-reduce} and
2691 @samp{-frerun-cse-after-loop}.
2693 @item -funroll-all-loops
2694 Perform the optimization of loop unrolling. This is done for all loops
2695 and usually makes programs run more slowly. @samp{-funroll-all-loops}
2696 implies @samp{-fstrength-reduce} as well as @samp{-frerun-cse-after-loop}.
2698 @item -fmove-all-movables
2699 Forces all invariant computations in loops to be moved
2702 @item -freduce-all-givs
2703 Forces all general-induction variables in loops to be
2706 @emph{Note:} When compiling programs written in Fortran,
2707 @samp{-fmove-all-movables} and @samp{-freduce-all-givs} are enabled
2708 by default when you use the optimizer.
2710 These options may generate better or worse code; results are highly
2711 dependent on the structure of loops within the source code.
2713 These two options are intended to be removed someday, once
2714 they have helped determine the efficacy of various
2715 approaches to improving loop optimizations.
2717 Please let us (@code{gcc@@gcc.gnu.org} and @code{fortran@@gnu.org})
2718 know how use of these options affects
2719 the performance of your production code.
2720 We're very interested in code that runs @emph{slower}
2721 when these options are @emph{enabled}.
2724 Disable any machine-specific peephole optimizations.
2726 @item -fbranch-probabilities
2727 After running a program compiled with @samp{-fprofile-arcs}
2728 (@pxref{Debugging Options,, Options for Debugging Your Program or
2729 @code{gcc}}), you can compile it a second time using
2730 @samp{-fbranch-probabilities}, to improve optimizations based on
2731 guessing the path a branch might take.
2734 With @samp{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
2735 note on the first instruction of each basic block, and a
2736 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
2737 These can be used to improve optimization. Currently, they are only
2738 used in one place: in @file{reorg.c}, instead of guessing which path a
2739 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
2740 exactly determine which path is taken more often.
2743 @item -fstrict-aliasing
2744 Allows the compiler to assume the strictest aliasing rules applicable to
2745 the language being compiled. For C (and C++), this activates
2746 optimizations based on the type of expressions. In particular, an
2747 object of one type is assumed never to reside at the same address as an
2748 object of a different type, unless the types are almost the same. For
2749 example, an @code{unsigned int} can alias an @code{int}, but not a
2750 @code{void*} or a @code{double}. A character type may alias any other
2753 Pay special attention to code like this:
2766 The practice of reading from a different union member than the one most
2767 recently written to (called ``type-punning'') is common. Even with
2768 @samp{-fstrict-aliasing}, type-punning is allowed, provided the memory
2769 is accessed through the union type. So, the code above will work as
2770 expected. However, this code might not:
2782 Every language that wishes to perform language-specific alias analysis
2783 should define a function that computes, given an @code{tree}
2784 node, an alias set for the node. Nodes in different alias sets are not
2785 allowed to alias. For an example, see the C front-end function
2786 @code{c_get_alias_set}.
2789 @item -falign-functions
2790 @itemx -falign-functions=@var{n}
2791 Align the start of functions to the next power-of-two greater than
2792 @var{n}, skipping up to @var{n} bytes. For instance,
2793 @samp{-falign-functions=32} aligns functions to the next 32-byte
2794 boundary, but @samp{-falign-functions=24} would align to the next
2795 32-byte boundary only if this can be done by skipping 23 bytes or less.
2797 @samp{-fno-align-functions} and @samp{-falign-functions=1} are
2798 equivalent and mean that functions will not be aligned.
2800 Some assemblers only support this flag when @var{n} is a power of two;
2801 in that case, it is rounded up.
2803 If @var{n} is not specified, use a machine-dependent default.
2805 @item -falign-labels
2806 @itemx -falign-labels=@var{n}
2807 Align all branch targets to a power-of-two boundary, skipping up to
2808 @var{n} bytes like @samp{-falign-functions}. This option can easily
2809 make code slower, because it must insert dummy operations for when the
2810 branch target is reached in the usual flow of the code.
2812 If @samp{-falign-loops} or @samp{-falign-jumps} are applicable and
2813 are greater than this value, then their values are used instead.
2815 If @var{n} is not specified, use a machine-dependent default which is
2816 very likely to be @samp{1}, meaning no alignment.
2819 @itemx -falign-loops=@var{n}
2820 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
2821 like @samp{-falign-functions}. The hope is that the loop will be
2822 executed many times, which will make up for any execution of the dummy
2825 If @var{n} is not specified, use a machine-dependent default.
2828 @itemx -falign-jumps=@var{n}
2829 Align branch targets to a power-of-two boundary, for branch targets
2830 where the targets can only be reached by jumping, skipping up to @var{n}
2831 bytes like @samp{-falign-functions}. In this case, no dummy operations
2834 If @var{n} is not specified, use a machine-dependent default.
2838 @node Preprocessor Options
2839 @section Options Controlling the Preprocessor
2840 @cindex preprocessor options
2841 @cindex options, preprocessor
2843 These options control the C preprocessor, which is run on each C source
2844 file before actual compilation.
2846 If you use the @samp{-E} option, nothing is done except preprocessing.
2847 Some of these options make sense only together with @samp{-E} because
2848 they cause the preprocessor output to be unsuitable for actual
2852 @item -include @var{file}
2853 Process @var{file} as input before processing the regular input file.
2854 In effect, the contents of @var{file} are compiled first. Any @samp{-D}
2855 and @samp{-U} options on the command line are always processed before
2856 @samp{-include @var{file}}, regardless of the order in which they are
2857 written. All the @samp{-include} and @samp{-imacros} options are
2858 processed in the order in which they are written.
2860 @item -imacros @var{file}
2861 Process @var{file} as input, discarding the resulting output, before
2862 processing the regular input file. Because the output generated from
2863 @var{file} is discarded, the only effect of @samp{-imacros @var{file}}
2864 is to make the macros defined in @var{file} available for use in the
2867 Any @samp{-D} and @samp{-U} options on the command line are always
2868 processed before @samp{-imacros @var{file}}, regardless of the order in
2869 which they are written. All the @samp{-include} and @samp{-imacros}
2870 options are processed in the order in which they are written.
2872 @item -idirafter @var{dir}
2873 @cindex second include path
2874 Add the directory @var{dir} to the second include path. The directories
2875 on the second include path are searched when a header file is not found
2876 in any of the directories in the main include path (the one that
2879 @item -iprefix @var{prefix}
2880 Specify @var{prefix} as the prefix for subsequent @samp{-iwithprefix}
2883 @item -iwithprefix @var{dir}
2884 Add a directory to the second include path. The directory's name is
2885 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
2886 specified previously with @samp{-iprefix}. If you have not specified a
2887 prefix yet, the directory containing the installed passes of the
2888 compiler is used as the default.
2890 @item -iwithprefixbefore @var{dir}
2891 Add a directory to the main include path. The directory's name is made
2892 by concatenating @var{prefix} and @var{dir}, as in the case of
2893 @samp{-iwithprefix}.
2895 @item -isystem @var{dir}
2896 Add a directory to the beginning of the second include path, marking it
2897 as a system directory, so that it gets the same special treatment as
2898 is applied to the standard system directories.
2901 Do not search the standard system directories for header files. Only
2902 the directories you have specified with @samp{-I} options (and the
2903 current directory, if appropriate) are searched. @xref{Directory
2904 Options}, for information on @samp{-I}.
2906 By using both @samp{-nostdinc} and @samp{-I-}, you can limit the include-file
2907 search path to only those directories you specify explicitly.
2910 Do not predefine any nonstandard macros. (Including architecture flags).
2913 Run only the C preprocessor. Preprocess all the C source files
2914 specified and output the results to standard output or to the
2915 specified output file.
2918 Tell the preprocessor not to discard comments. Used with the
2922 Tell the preprocessor not to generate @samp{#line} directives.
2923 Used with the @samp{-E} option.
2926 @cindex dependencies, make
2928 Tell the preprocessor to output a rule suitable for @code{make}
2929 describing the dependencies of each object file. For each source file,
2930 the preprocessor outputs one @code{make}-rule whose target is the object
2931 file name for that source file and whose dependencies are all the
2932 @code{#include} header files it uses. This rule may be a single line or
2933 may be continued with @samp{\}-newline if it is long. The list of rules
2934 is printed on standard output instead of the preprocessed C program.
2936 @samp{-M} implies @samp{-E}.
2938 Another way to specify output of a @code{make} rule is by setting
2939 the environment variable @code{DEPENDENCIES_OUTPUT} (@pxref{Environment
2943 Like @samp{-M} but the output mentions only the user header files
2944 included with @samp{#include "@var{file}"}. System header files
2945 included with @samp{#include <@var{file}>} are omitted.
2948 Like @samp{-M} but the dependency information is written to a file made by
2949 replacing ".c" with ".d" at the end of the input file names.
2950 This is in addition to compiling the file as specified---@samp{-MD} does
2951 not inhibit ordinary compilation the way @samp{-M} does.
2953 In Mach, you can use the utility @code{md} to merge multiple dependency
2954 files into a single dependency file suitable for using with the @samp{make}
2958 Like @samp{-MD} except mention only user header files, not system
2962 Treat missing header files as generated files and assume they live in the
2963 same directory as the source file. If you specify @samp{-MG}, you
2964 must also specify either @samp{-M} or @samp{-MM}. @samp{-MG} is not
2965 supported with @samp{-MD} or @samp{-MMD}.
2968 Print the name of each header file used, in addition to other normal
2971 @item -A@var{question}(@var{answer})
2972 Assert the answer @var{answer} for @var{question}, in case it is tested
2973 with a preprocessing conditional such as @samp{#if
2974 #@var{question}(@var{answer})}. @samp{-A-} disables the standard
2975 assertions that normally describe the target machine.
2978 Define macro @var{macro} with the string @samp{1} as its definition.
2980 @item -D@var{macro}=@var{defn}
2981 Define macro @var{macro} as @var{defn}. All instances of @samp{-D} on
2982 the command line are processed before any @samp{-U} options.
2985 Undefine macro @var{macro}. @samp{-U} options are evaluated after all
2986 @samp{-D} options, but before any @samp{-include} and @samp{-imacros}
2990 Tell the preprocessor to output only a list of the macro definitions
2991 that are in effect at the end of preprocessing. Used with the @samp{-E}
2995 Tell the preprocessing to pass all macro definitions into the output, in
2996 their proper sequence in the rest of the output.
2999 Like @samp{-dD} except that the macro arguments and contents are omitted.
3000 Only @samp{#define @var{name}} is included in the output.
3003 Support ANSI C trigraphs. The @samp{-ansi} option also has this effect.
3005 @item -Wp,@var{option}
3006 Pass @var{option} as an option to the preprocessor. If @var{option}
3007 contains commas, it is split into multiple options at the commas.
3010 @node Assembler Options
3011 @section Passing Options to the Assembler
3013 @c prevent bad page break with this line
3014 You can pass options to the assembler.
3017 @item -Wa,@var{option}
3018 Pass @var{option} as an option to the assembler. If @var{option}
3019 contains commas, it is split into multiple options at the commas.
3023 @section Options for Linking
3024 @cindex link options
3025 @cindex options, linking
3027 These options come into play when the compiler links object files into
3028 an executable output file. They are meaningless if the compiler is
3029 not doing a link step.
3033 @item @var{object-file-name}
3034 A file name that does not end in a special recognized suffix is
3035 considered to name an object file or library. (Object files are
3036 distinguished from libraries by the linker according to the file
3037 contents.) If linking is done, these object files are used as input
3043 If any of these options is used, then the linker is not run, and
3044 object file names should not be used as arguments. @xref{Overall
3048 @item -l@var{library}
3049 Search the library named @var{library} when linking.
3051 It makes a difference where in the command you write this option; the
3052 linker searches processes libraries and object files in the order they
3053 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
3054 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
3055 to functions in @samp{z}, those functions may not be loaded.
3057 The linker searches a standard list of directories for the library,
3058 which is actually a file named @file{lib@var{library}.a}. The linker
3059 then uses this file as if it had been specified precisely by name.
3061 The directories searched include several standard system directories
3062 plus any that you specify with @samp{-L}.
3064 Normally the files found this way are library files---archive files
3065 whose members are object files. The linker handles an archive file by
3066 scanning through it for members which define symbols that have so far
3067 been referenced but not defined. But if the file that is found is an
3068 ordinary object file, it is linked in the usual fashion. The only
3069 difference between using an @samp{-l} option and specifying a file name
3070 is that @samp{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
3071 and searches several directories.
3074 You need this special case of the @samp{-l} option in order to
3075 link an Objective C program.
3078 Do not use the standard system startup files when linking.
3079 The standard system libraries are used normally, unless @code{-nostdlib}
3080 or @code{-nodefaultlibs} is used.
3082 @item -nodefaultlibs
3083 Do not use the standard system libraries when linking.
3084 Only the libraries you specify will be passed to the linker.
3085 The standard startup files are used normally, unless @code{-nostartfiles}
3086 is used. The compiler may generate calls to memcmp, memset, and memcpy
3087 for System V (and ANSI C) environments or to bcopy and bzero for
3088 BSD environments. These entries are usually resolved by entries in
3089 libc. These entry points should be supplied through some other
3090 mechanism when this option is specified.
3093 Do not use the standard system startup files or libraries when linking.
3094 No startup files and only the libraries you specify will be passed to
3095 the linker. The compiler may generate calls to memcmp, memset, and memcpy
3096 for System V (and ANSI C) environments or to bcopy and bzero for
3097 BSD environments. These entries are usually resolved by entries in
3098 libc. These entry points should be supplied through some other
3099 mechanism when this option is specified.
3101 @cindex @code{-lgcc}, use with @code{-nostdlib}
3102 @cindex @code{-nostdlib} and unresolved references
3103 @cindex unresolved references and @code{-nostdlib}
3104 @cindex @code{-lgcc}, use with @code{-nodefaultlibs}
3105 @cindex @code{-nodefaultlibs} and unresolved references
3106 @cindex unresolved references and @code{-nodefaultlibs}
3107 One of the standard libraries bypassed by @samp{-nostdlib} and
3108 @samp{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
3109 that GCC uses to overcome shortcomings of particular machines, or special
3110 needs for some languages.
3112 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
3116 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
3117 for more discussion of @file{libgcc.a}.)
3119 In most cases, you need @file{libgcc.a} even when you want to avoid
3120 other standard libraries. In other words, when you specify @samp{-nostdlib}
3121 or @samp{-nodefaultlibs} you should usually specify @samp{-lgcc} as well.
3122 This ensures that you have no unresolved references to internal GCC
3123 library subroutines. (For example, @samp{__main}, used to ensure C++
3124 constructors will be called; @pxref{Collect2,,@code{collect2}}.)
3127 Remove all symbol table and relocation information from the executable.
3130 On systems that support dynamic linking, this prevents linking with the shared
3131 libraries. On other systems, this option has no effect.
3134 Produce a shared object which can then be linked with other objects to
3135 form an executable. Not all systems support this option. You must
3136 also specify @samp{-fpic} or @samp{-fPIC} on some systems when
3137 you specify this option.
3140 Bind references to global symbols when building a shared object. Warn
3141 about any unresolved references (unless overridden by the link editor
3142 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
3145 @item -Xlinker @var{option}
3146 Pass @var{option} as an option to the linker. You can use this to
3147 supply system-specific linker options which GCC does not know how to
3150 If you want to pass an option that takes an argument, you must use
3151 @samp{-Xlinker} twice, once for the option and once for the argument.
3152 For example, to pass @samp{-assert definitions}, you must write
3153 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
3154 @samp{-Xlinker "-assert definitions"}, because this passes the entire
3155 string as a single argument, which is not what the linker expects.
3157 @item -Wl,@var{option}
3158 Pass @var{option} as an option to the linker. If @var{option} contains
3159 commas, it is split into multiple options at the commas.
3161 @item -u @var{symbol}
3162 Pretend the symbol @var{symbol} is undefined, to force linking of
3163 library modules to define it. You can use @samp{-u} multiple times with
3164 different symbols to force loading of additional library modules.
3167 @node Directory Options
3168 @section Options for Directory Search
3169 @cindex directory options
3170 @cindex options, directory search
3173 These options specify directories to search for header files, for
3174 libraries and for parts of the compiler:
3178 Add the directory @var{dir} to the head of the list of directories to be
3179 searched for header files. This can be used to override a system header
3180 file, substituting your own version, since these directories are
3181 searched before the system header file directories. If you use more
3182 than one @samp{-I} option, the directories are scanned in left-to-right
3183 order; the standard system directories come after.
3186 Any directories you specify with @samp{-I} options before the @samp{-I-}
3187 option are searched only for the case of @samp{#include "@var{file}"};
3188 they are not searched for @samp{#include <@var{file}>}.
3190 If additional directories are specified with @samp{-I} options after
3191 the @samp{-I-}, these directories are searched for all @samp{#include}
3192 directives. (Ordinarily @emph{all} @samp{-I} directories are used
3195 In addition, the @samp{-I-} option inhibits the use of the current
3196 directory (where the current input file came from) as the first search
3197 directory for @samp{#include "@var{file}"}. There is no way to
3198 override this effect of @samp{-I-}. With @samp{-I.} you can specify
3199 searching the directory which was current when the compiler was
3200 invoked. That is not exactly the same as what the preprocessor does
3201 by default, but it is often satisfactory.
3203 @samp{-I-} does not inhibit the use of the standard system directories
3204 for header files. Thus, @samp{-I-} and @samp{-nostdinc} are
3208 Add directory @var{dir} to the list of directories to be searched
3211 @item -B@var{prefix}
3212 This option specifies where to find the executables, libraries,
3213 include files, and data files of the compiler itself.
3215 The compiler driver program runs one or more of the subprograms
3216 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
3217 @var{prefix} as a prefix for each program it tries to run, both with and
3218 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
3220 For each subprogram to be run, the compiler driver first tries the
3221 @samp{-B} prefix, if any. If that name is not found, or if @samp{-B}
3222 was not specified, the driver tries two standard prefixes, which are
3223 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
3224 those results in a file name that is found, the unmodified program
3225 name is searched for using the directories specified in your
3226 @samp{PATH} environment variable.
3228 @samp{-B} prefixes that effectively specify directory names also apply
3229 to libraries in the linker, because the compiler translates these
3230 options into @samp{-L} options for the linker. They also apply to
3231 includes files in the preprocessor, because the compiler translates these
3232 options into @samp{-isystem} options for the preprocessor. In this case,
3233 the compiler appends @samp{include} to the prefix.
3235 The run-time support file @file{libgcc.a} can also be searched for using
3236 the @samp{-B} prefix, if needed. If it is not found there, the two
3237 standard prefixes above are tried, and that is all. The file is left
3238 out of the link if it is not found by those means.
3240 Another way to specify a prefix much like the @samp{-B} prefix is to use
3241 the environment variable @code{GCC_EXEC_PREFIX}. @xref{Environment
3244 @item -specs=@var{file}
3245 Process @var{file} after the compiler reads in the standard @file{specs}
3246 file, in order to override the defaults that the @file{gcc} driver
3247 program uses when determining what switches to pass to @file{cc1},
3248 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
3249 @samp{-specs=}@var{file} can be specified on the command line, and they
3250 are processed in order, from left to right.
3254 @section Specifying subprocesses and the switches to pass to them
3256 @code{GCC} is a driver program. It performs its job by invoking a
3257 sequence of other programs to do the work of compiling, assembling and
3258 linking. GCC interprets its command-line parameters and uses these to
3259 deduce which programs it should invoke, and which command-line options
3260 it ought to place on their command lines. This behaviour is controlled
3261 by @dfn{spec strings}. In most cases there is one spec string for each
3262 program that GCC can invoke, but a few programs have multiple spec
3263 strings to control their behaviour. The spec strings built into GCC can
3264 be overridden by using the @samp{-specs=} command-line switch to specify
3267 @dfn{Spec files} are plaintext files that are used to construct spec
3268 strings. They consist of a sequence of directives separated by blank
3269 lines. The type of directive is determined by the first non-whitespace
3270 character on the line and it can be one of the following:
3273 @item %@var{command}
3274 Issues a @var{command} to the spec file processor. The commands that can
3278 @item %include <@var{file}>
3280 Search for @var{file} and insert its text at the current point in the
3283 @item %include_noerr <@var{file}>
3284 @cindex %include_noerr
3285 Just like @samp{%include}, but do not generate an error message if the include
3286 file cannot be found.
3288 @item %rename @var{old_name} @var{new_name}
3290 Rename the spec string @var{old_name} to @var{new_name}.
3294 @item *[@var{spec_name}]:
3295 This tells the compiler to create, override or delete the named spec
3296 string. All lines after this directive up to the next directive or
3297 blank line are considered to be the text for the spec string. If this
3298 results in an empty string then the spec will be deleted. (Or, if the
3299 spec did not exist, then nothing will happened.) Otherwise, if the spec
3300 does not currently exist a new spec will be created. If the spec does
3301 exist then its contents will be overridden by the text of this
3302 directive, unless the first character of that text is the @samp{+}
3303 character, in which case the text will be appended to the spec.
3305 @item [@var{suffix}]:
3306 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
3307 and up to the next directive or blank line are considered to make up the
3308 spec string for the indicated suffix. When the compiler encounters an
3309 input file with the named suffix, it will processes the spec string in
3310 order to work out how to compile that file. For example:
3317 This says that any input file whose name ends in @samp{.ZZ} should be
3318 passed to the program @samp{z-compile}, which should be invoked with the
3319 command-line switch @samp{-input} and with the result of performing the
3320 @samp{%i} substitution. (See below.)
3322 As an alternative to providing a spec string, the text that follows a
3323 suffix directive can be one of the following:
3326 @item @@@var{language}
3327 This says that the suffix is an alias for a known @var{language}. This is
3328 similar to using the @code{-x} command-line switch to GCC to specify a
3329 language explicitly. For example:
3336 Says that .ZZ files are, in fact, C++ source files.
3339 This causes an error messages saying:
3342 @var{name} compiler not installed on this system.
3346 GCC already has an extensive list of suffixes built into it.
3347 This directive will add an entry to the end of the list of suffixes, but
3348 since the list is searched from the end backwards, it is effectively
3349 possible to override earlier entries using this technique.
3353 GCC has the following spec strings built into it. Spec files can
3354 override these strings or create their own. Note that individual
3355 targets can also add their own spec strings to this list.
3358 asm Options to pass to the assembler
3359 asm_final Options to pass to the assembler post-processor
3360 cpp Options to pass to the C preprocessor
3361 cc1 Options to pass to the C compiler
3362 cc1plus Options to pass to the C++ compiler
3363 endfile Object files to include at the end of the link
3364 link Options to pass to the linker
3365 lib Libraries to include on the command line to the linker
3366 libgcc Decides which GCC support library to pass to the linker
3367 linker Sets the name of the linker
3368 predefines Defines to be passed to the C preprocessor
3369 signed_char Defines to pass to CPP to say whether @code{char} is signed by default
3370 startfile Object files to include at the start of the link
3373 Here is a small example of a spec file:
3379 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
3382 This example renames the spec called @samp{lib} to @samp{old_lib} and
3383 then overrides the previous definition of @samp{lib} with a new one.
3384 The new definition adds in some extra command-line options before
3385 including the text of the old definition.
3387 @dfn{Spec strings} are a list of command-line options to be passed to their
3388 corresponding program. In addition, the spec strings can contain
3389 @samp{%}-prefixed sequences to substitute variable text or to
3390 conditionally insert text into the command line. Using these constructs
3391 it is possible to generate quite complex command lines.
3393 Here is a table of all defined @samp{%}-sequences for spec
3394 strings. Note that spaces are not generated automatically around the
3395 results of expanding these sequences. Therefore you can concatenate them
3396 together or combine them with constant text in a single argument.
3400 Substitute one @samp{%} into the program name or argument.
3403 Substitute the name of the input file being processed.
3406 Substitute the basename of the input file being processed.
3407 This is the substring up to (and not including) the last period
3408 and not including the directory.
3411 Marks the argument containing or following the @samp{%d} as a
3412 temporary file name, so that that file will be deleted if GCC exits
3413 successfully. Unlike @samp{%g}, this contributes no text to the
3416 @item %g@var{suffix}
3417 Substitute a file name that has suffix @var{suffix} and is chosen
3418 once per compilation, and mark the argument in the same way as
3419 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
3420 name is now chosen in a way that is hard to predict even when previously
3421 chosen file names are known. For example, @samp{%g.s ... %g.o ... %g.s}
3422 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
3423 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
3424 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
3425 was simply substituted with a file name chosen once per compilation,
3426 without regard to any appended suffix (which was therefore treated
3427 just like ordinary text), making such attacks more likely to succeed.
3429 @item %u@var{suffix}
3430 Like @samp{%g}, but generates a new temporary file name even if
3431 @samp{%u@var{suffix}} was already seen.
3433 @item %U@var{suffix}
3434 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
3435 new one if there is no such last file name. In the absence of any
3436 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
3437 the same suffix @emph{space}, so @samp{%g.s ... %U.s ... %g.s ... %U.s}
3438 would involve the generation of two distinct file names, one
3439 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
3440 simply substituted with a file name chosen for the previous @samp{%u},
3441 without regard to any appended suffix.
3444 Marks the argument containing or following the @samp{%w} as the
3445 designated output file of this compilation. This puts the argument
3446 into the sequence of arguments that @samp{%o} will substitute later.
3449 Substitutes the names of all the output files, with spaces
3450 automatically placed around them. You should write spaces
3451 around the @samp{%o} as well or the results are undefined.
3452 @samp{%o} is for use in the specs for running the linker.
3453 Input files whose names have no recognized suffix are not compiled
3454 at all, but they are included among the output files, so they will
3458 Substitutes the suffix for object files. Note that this is
3459 handled specially when it immediately follows @samp{%g, %u, or %U},
3460 because of the need for those to form complete file names. The
3461 handling is such that @samp{%O} is treated exactly as if it had already
3462 been substituted, except that @samp{%g, %u, and %U} do not currently
3463 support additional @var{suffix} characters following @samp{%O} as they would
3464 following, for example, @samp{.o}.
3467 Substitutes the standard macro predefinitions for the
3468 current target machine. Use this when running @code{cpp}.
3471 Like @samp{%p}, but puts @samp{__} before and after the name of each
3472 predefined macro, except for macros that start with @samp{__} or with
3473 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ANSI
3477 Substitute a @samp{-iprefix} option made from GCC_EXEC_PREFIX.
3480 Current argument is the name of a library or startup file of some sort.
3481 Search for that file in a standard list of directories and substitute
3482 the full name found.
3485 Print @var{str} as an error message. @var{str} is terminated by a newline.
3486 Use this when inconsistent options are detected.
3489 Output @samp{-} if the input for the current command is coming from a pipe.
3492 Substitute the contents of spec string @var{name} at this point.
3495 Like @samp{%(...)} but put @samp{__} around @samp{-D} arguments.
3497 @item %x@{@var{option}@}
3498 Accumulate an option for @samp{%X}.
3501 Output the accumulated linker options specified by @samp{-Wl} or a @samp{%x}
3505 Output the accumulated assembler options specified by @samp{-Wa}.
3508 Output the accumulated preprocessor options specified by @samp{-Wp}.
3511 Substitute the major version number of GCC.
3512 (For version 2.9.5, this is 2.)
3515 Substitute the minor version number of GCC.
3516 (For version 2.9.5, this is 9.)
3519 Process the @code{asm} spec. This is used to compute the
3520 switches to be passed to the assembler.
3523 Process the @code{asm_final} spec. This is a spec string for
3524 passing switches to an assembler post-processor, if such a program is
3528 Process the @code{link} spec. This is the spec for computing the
3529 command line passed to the linker. Typically it will make use of the
3530 @samp{%L %G %S %D and %E} sequences.
3533 Dump out a @samp{-L} option for each directory that GCC believes might
3534 contain startup files. If the target supports multilibs then the
3535 current multilib directory will be prepended to each of these paths.
3538 Process the @code{lib} spec. This is a spec string for deciding which
3539 libraries should be included on the command line to the linker.
3542 Process the @code{libgcc} spec. This is a spec string for deciding
3543 which GCC support library should be included on the command line to the linker.
3546 Process the @code{startfile} spec. This is a spec for deciding which
3547 object files should be the first ones passed to the linker. Typically
3548 this might be a file named @file{crt0.o}.
3551 Process the @code{endfile} spec. This is a spec string that specifies
3552 the last object files that will be passed to the linker.
3555 Process the @code{cpp} spec. This is used to construct the arguments
3556 to be passed to the C preprocessor.
3559 Process the @code{signed_char} spec. This is intended to be used
3560 to tell cpp whether a char is signed. It typically has the definition:
3562 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
3566 Process the @code{cc1} spec. This is used to construct the options to be
3567 passed to the actual C compiler (@samp{cc1}).
3570 Process the @code{cc1plus} spec. This is used to construct the options to be
3571 passed to the actual C++ compiler (@samp{cc1plus}).
3574 Substitute the variable part of a matched option. See below.
3575 Note that each comma in the substituted string is replaced by
3579 Substitutes the @code{-S} switch, if that switch was given to GCC.
3580 If that switch was not specified, this substitutes nothing. Note that
3581 the leading dash is omitted when specifying this option, and it is
3582 automatically inserted if the substitution is performed. Thus the spec
3583 string @samp{%@{foo@}} would match the command-line option @samp{-foo}
3584 and would output the command line option @samp{-foo}.
3586 @item %W@{@code{S}@}
3587 Like %@{@code{S}@} but mark last argument supplied within as a file to be
3590 @item %@{@code{S}*@}
3591 Substitutes all the switches specified to GCC whose names start
3592 with @code{-S}, but which also take an argument. This is used for
3593 switches like @samp{-o, -D, -I}, etc. GCC considers @samp{-o foo} as being
3594 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
3595 text, including the space. Thus two arguments would be generated.
3597 @item %@{^@code{S}*@}
3598 Like %@{@code{S}*@}, but don't put a blank between a switch and its
3599 argument. Thus %@{^o*@} would only generate one argument, not two.
3601 @item %@{<@code{S}@}
3602 Remove all occurences of @code{S} from the command line. Note - this
3603 command is position dependent. @samp{%} commands in the spec string
3604 before this option will see @code{S}, @samp{%} commands in the spec
3605 string after this option will not.
3607 @item %@{@code{S}*:@code{X}@}
3608 Substitutes @code{X} if one or more switches whose names start with
3609 @code{-S} are specified to GCC. Note that the tail part of the
3610 @code{-S} option (i.e. the part matched by the @samp{*}) will be substituted
3611 for each occurrence of @samp{%*} within @code{X}.
3613 @item %@{@code{S}:@code{X}@}
3614 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC.
3616 @item %@{!@code{S}:@code{X}@}
3617 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC.
3619 @item %@{|@code{S}:@code{X}@}
3620 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
3622 @item %@{|!@code{S}:@code{X}@}
3623 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
3625 @item %@{.@code{S}:@code{X}@}
3626 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
3628 @item %@{!.@code{S}:@code{X}@}
3629 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
3631 @item %@{@code{S}|@code{P}:@code{X}@}
3632 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC. This may be
3633 combined with @samp{!} and @samp{.} sequences as well, although they
3634 have a stronger binding than the @samp{|}. For example a spec string
3638 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
3641 will output the following command-line options from the following input
3642 command-line options:
3647 -d fred.c -foo -baz -boggle
3648 -d jim.d -bar -baz -boggle
3653 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
3654 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
3655 or spaces, or even newlines. They are processed as usual, as described
3658 The @samp{-O, -f, -m, and -W} switches are handled specifically in these
3659 constructs. If another value of @samp{-O} or the negated form of a @samp{-f, -m, or
3660 -W} switch is found later in the command line, the earlier switch
3661 value is ignored, except with @{@code{S}*@} where @code{S} is just one
3662 letter, which passes all matching options.
3664 The character @samp{|} at the beginning of the predicate text is used to indicate
3665 that a command should be piped to the following command, but only if @samp{-pipe}
3668 It is built into GCC which switches take arguments and which do not.
3669 (You might think it would be useful to generalize this to allow each
3670 compiler's spec to say which switches take arguments. But this cannot
3671 be done in a consistent fashion. GCC cannot even decide which input
3672 files have been specified without knowing which switches take arguments,
3673 and it must know which input files to compile in order to tell which
3676 GCC also knows implicitly that arguments starting in @samp{-l} are to be
3677 treated as compiler output files, and passed to the linker in their
3678 proper position among the other output files.
3680 @node Target Options
3681 @section Specifying Target Machine and Compiler Version
3682 @cindex target options
3683 @cindex cross compiling
3684 @cindex specifying machine version
3685 @cindex specifying compiler version and target machine
3686 @cindex compiler version, specifying
3687 @cindex target machine, specifying
3689 By default, GCC compiles code for the same type of machine that you
3690 are using. However, it can also be installed as a cross-compiler, to
3691 compile for some other type of machine. In fact, several different
3692 configurations of GCC, for different target machines, can be
3693 installed side by side. Then you specify which one to use with the
3696 In addition, older and newer versions of GCC can be installed side
3697 by side. One of them (probably the newest) will be the default, but
3698 you may sometimes wish to use another.
3701 @item -b @var{machine}
3702 The argument @var{machine} specifies the target machine for compilation.
3703 This is useful when you have installed GCC as a cross-compiler.
3705 The value to use for @var{machine} is the same as was specified as the
3706 machine type when configuring GCC as a cross-compiler. For
3707 example, if a cross-compiler was configured with @samp{configure
3708 i386v}, meaning to compile for an 80386 running System V, then you
3709 would specify @samp{-b i386v} to run that cross compiler.
3711 When you do not specify @samp{-b}, it normally means to compile for
3712 the same type of machine that you are using.
3714 @item -V @var{version}
3715 The argument @var{version} specifies which version of GCC to run.
3716 This is useful when multiple versions are installed. For example,
3717 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
3719 The default version, when you do not specify @samp{-V}, is the last
3720 version of GCC that you installed.
3723 The @samp{-b} and @samp{-V} options actually work by controlling part of
3724 the file name used for the executable files and libraries used for
3725 compilation. A given version of GCC, for a given target machine, is
3726 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.@refill
3728 Thus, sites can customize the effect of @samp{-b} or @samp{-V} either by
3729 changing the names of these directories or adding alternate names (or
3730 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
3731 file @file{80386} is a link to the file @file{i386v}, then @samp{-b
3732 80386} becomes an alias for @samp{-b i386v}.
3734 In one respect, the @samp{-b} or @samp{-V} do not completely change
3735 to a different compiler: the top-level driver program @code{gcc}
3736 that you originally invoked continues to run and invoke the other
3737 executables (preprocessor, compiler per se, assembler and linker)
3738 that do the real work. However, since no real work is done in the
3739 driver program, it usually does not matter that the driver program
3740 in use is not the one for the specified target and version.
3742 The only way that the driver program depends on the target machine is
3743 in the parsing and handling of special machine-specific options.
3744 However, this is controlled by a file which is found, along with the
3745 other executables, in the directory for the specified version and
3746 target machine. As a result, a single installed driver program adapts
3747 to any specified target machine and compiler version.
3749 The driver program executable does control one significant thing,
3750 however: the default version and target machine. Therefore, you can
3751 install different instances of the driver program, compiled for
3752 different targets or versions, under different names.
3754 For example, if the driver for version 2.0 is installed as @code{ogcc}
3755 and that for version 2.1 is installed as @code{gcc}, then the command
3756 @code{gcc} will use version 2.1 by default, while @code{ogcc} will use
3757 2.0 by default. However, you can choose either version with either
3758 command with the @samp{-V} option.
3760 @node Submodel Options
3761 @section Hardware Models and Configurations
3762 @cindex submodel options
3763 @cindex specifying hardware config
3764 @cindex hardware models and configurations, specifying
3765 @cindex machine dependent options
3767 Earlier we discussed the standard option @samp{-b} which chooses among
3768 different installed compilers for completely different target
3769 machines, such as Vax vs. 68000 vs. 80386.
3771 In addition, each of these target machine types can have its own
3772 special options, starting with @samp{-m}, to choose among various
3773 hardware models or configurations---for example, 68010 vs 68020,
3774 floating coprocessor or none. A single installed version of the
3775 compiler can compile for any model or configuration, according to the
3778 Some configurations of the compiler also support additional special
3779 options, usually for compatibility with other compilers on the same
3783 These options are defined by the macro @code{TARGET_SWITCHES} in the
3784 machine description. The default for the options is also defined by
3785 that macro, which enables you to change the defaults.
3800 * RS/6000 and PowerPC Options::
3805 * Intel 960 Options::
3806 * DEC Alpha Options::
3810 * System V Options::
3811 * TMS320C3x/C4x Options::
3819 @node M680x0 Options
3820 @subsection M680x0 Options
3821 @cindex M680x0 options
3823 These are the @samp{-m} options defined for the 68000 series. The default
3824 values for these options depends on which style of 68000 was selected when
3825 the compiler was configured; the defaults for the most common choices are
3831 Generate output for a 68000. This is the default
3832 when the compiler is configured for 68000-based systems.
3834 Use this option for microcontrollers with a 68000 or EC000 core,
3835 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
3839 Generate output for a 68020. This is the default
3840 when the compiler is configured for 68020-based systems.
3843 Generate output containing 68881 instructions for floating point.
3844 This is the default for most 68020 systems unless @samp{-nfp} was
3845 specified when the compiler was configured.
3848 Generate output for a 68030. This is the default when the compiler is
3849 configured for 68030-based systems.
3852 Generate output for a 68040. This is the default when the compiler is
3853 configured for 68040-based systems.
3855 This option inhibits the use of 68881/68882 instructions that have to be
3856 emulated by software on the 68040. Use this option if your 68040 does not
3857 have code to emulate those instructions.
3860 Generate output for a 68060. This is the default when the compiler is
3861 configured for 68060-based systems.
3863 This option inhibits the use of 68020 and 68881/68882 instructions that
3864 have to be emulated by software on the 68060. Use this option if your 68060
3865 does not have code to emulate those instructions.
3868 Generate output for a CPU32. This is the default
3869 when the compiler is configured for CPU32-based systems.
3871 Use this option for microcontrollers with a
3872 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
3873 68336, 68340, 68341, 68349 and 68360.
3876 Generate output for a 520X "coldfire" family cpu. This is the default
3877 when the compiler is configured for 520X-based systems.
3879 Use this option for microcontroller with a 5200 core, including
3880 the MCF5202, MCF5203, MCF5204 and MCF5202.
3884 Generate output for a 68040, without using any of the new instructions.
3885 This results in code which can run relatively efficiently on either a
3886 68020/68881 or a 68030 or a 68040. The generated code does use the
3887 68881 instructions that are emulated on the 68040.
3890 Generate output for a 68060, without using any of the new instructions.
3891 This results in code which can run relatively efficiently on either a
3892 68020/68881 or a 68030 or a 68040. The generated code does use the
3893 68881 instructions that are emulated on the 68060.
3896 Generate output containing Sun FPA instructions for floating point.
3899 Generate output containing library calls for floating point.
3900 @strong{Warning:} the requisite libraries are not available for all m68k
3901 targets. Normally the facilities of the machine's usual C compiler are
3902 used, but this can't be done directly in cross-compilation. You must
3903 make your own arrangements to provide suitable library functions for
3904 cross-compilation. The embedded targets @samp{m68k-*-aout} and
3905 @samp{m68k-*-coff} do provide software floating point support.
3908 Consider type @code{int} to be 16 bits wide, like @code{short int}.
3911 Do not use the bit-field instructions. The @samp{-m68000}, @samp{-mcpu32}
3912 and @samp{-m5200} options imply @w{@samp{-mnobitfield}}.
3915 Do use the bit-field instructions. The @samp{-m68020} option implies
3916 @samp{-mbitfield}. This is the default if you use a configuration
3917 designed for a 68020.
3920 Use a different function-calling convention, in which functions
3921 that take a fixed number of arguments return with the @code{rtd}
3922 instruction, which pops their arguments while returning. This
3923 saves one instruction in the caller since there is no need to pop
3924 the arguments there.
3926 This calling convention is incompatible with the one normally
3927 used on Unix, so you cannot use it if you need to call libraries
3928 compiled with the Unix compiler.
3930 Also, you must provide function prototypes for all functions that
3931 take variable numbers of arguments (including @code{printf});
3932 otherwise incorrect code will be generated for calls to those
3935 In addition, seriously incorrect code will result if you call a
3936 function with too many arguments. (Normally, extra arguments are
3937 harmlessly ignored.)
3939 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
3940 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
3943 @itemx -mno-align-int
3944 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
3945 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
3946 boundary (@samp{-malign-int}) or a 16-bit boundary (@samp{-mno-align-int}).
3947 Aligning variables on 32-bit boundaries produces code that runs somewhat
3948 faster on processors with 32-bit busses at the expense of more memory.
3950 @strong{Warning:} if you use the @samp{-malign-int} switch, GCC will
3951 align structures containing the above types differently than
3952 most published application binary interface specifications for the m68k.
3955 Use the pc-relative addressing mode of the 68000 directly, instead of
3956 using a global offset table. At present, this option implies -fpic,
3957 allowing at most a 16-bit offset for pc-relative addressing. -fPIC is
3958 not presently supported with -mpcrel, though this could be supported for
3959 68020 and higher processors.
3961 @item -mno-strict-align
3962 @itemx -mstrict-align
3963 @kindex -mstrict-align
3964 Do not (do) assume that unaligned memory references will be handled by
3970 @subsection VAX Options
3973 These @samp{-m} options are defined for the Vax:
3977 Do not output certain jump instructions (@code{aobleq} and so on)
3978 that the Unix assembler for the Vax cannot handle across long
3982 Do output those jump instructions, on the assumption that you
3983 will assemble with the GNU assembler.
3986 Output code for g-format floating point numbers instead of d-format.
3990 @subsection SPARC Options
3991 @cindex SPARC options
3993 These @samp{-m} switches are supported on the SPARC:
3998 Specify @samp{-mapp-regs} to generate output using the global registers
3999 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
4002 To be fully SVR4 ABI compliant at the cost of some performance loss,
4003 specify @samp{-mno-app-regs}. You should compile libraries and system
4004 software with this option.
4008 Generate output containing floating point instructions. This is the
4013 Generate output containing library calls for floating point.
4014 @strong{Warning:} the requisite libraries are not available for all SPARC
4015 targets. Normally the facilities of the machine's usual C compiler are
4016 used, but this cannot be done directly in cross-compilation. You must make
4017 your own arrangements to provide suitable library functions for
4018 cross-compilation. The embedded targets @samp{sparc-*-aout} and
4019 @samp{sparclite-*-*} do provide software floating point support.
4021 @samp{-msoft-float} changes the calling convention in the output file;
4022 therefore, it is only useful if you compile @emph{all} of a program with
4023 this option. In particular, you need to compile @file{libgcc.a}, the
4024 library that comes with GCC, with @samp{-msoft-float} in order for
4027 @item -mhard-quad-float
4028 Generate output containing quad-word (long double) floating point
4031 @item -msoft-quad-float
4032 Generate output containing library calls for quad-word (long double)
4033 floating point instructions. The functions called are those specified
4034 in the SPARC ABI. This is the default.
4036 As of this writing, there are no sparc implementations that have hardware
4037 support for the quad-word floating point instructions. They all invoke
4038 a trap handler for one of these instructions, and then the trap handler
4039 emulates the effect of the instruction. Because of the trap handler overhead,
4040 this is much slower than calling the ABI library routines. Thus the
4041 @samp{-msoft-quad-float} option is the default.
4045 With @samp{-mepilogue} (the default), the compiler always emits code for
4046 function exit at the end of each function. Any function exit in
4047 the middle of the function (such as a return statement in C) will
4048 generate a jump to the exit code at the end of the function.
4050 With @samp{-mno-epilogue}, the compiler tries to emit exit code inline
4051 at every function exit.
4055 With @samp{-mflat}, the compiler does not generate save/restore instructions
4056 and will use a "flat" or single register window calling convention.
4057 This model uses %i7 as the frame pointer and is compatible with the normal
4058 register window model. Code from either may be intermixed.
4059 The local registers and the input registers (0-5) are still treated as
4060 "call saved" registers and will be saved on the stack as necessary.
4062 With @samp{-mno-flat} (the default), the compiler emits save/restore
4063 instructions (except for leaf functions) and is the normal mode of operation.
4065 @item -mno-unaligned-doubles
4066 @itemx -munaligned-doubles
4067 Assume that doubles have 8 byte alignment. This is the default.
4069 With @samp{-munaligned-doubles}, GCC assumes that doubles have 8 byte
4070 alignment only if they are contained in another type, or if they have an
4071 absolute address. Otherwise, it assumes they have 4 byte alignment.
4072 Specifying this option avoids some rare compatibility problems with code
4073 generated by other compilers. It is not the default because it results
4074 in a performance loss, especially for floating point code.
4076 @item -mno-faster-structs
4077 @itemx -mfaster-structs
4078 With @samp{-mfaster-structs}, the compiler assumes that structures
4079 should have 8 byte alignment. This enables the use of pairs of
4080 @code{ldd} and @code{std} instructions for copies in structure
4081 assignment, in place of twice as many @code{ld} and @code{st} pairs.
4082 However, the use of this changed alignment directly violates the Sparc
4083 ABI. Thus, it's intended only for use on targets where the developer
4084 acknowledges that their resulting code will not be directly in line with
4085 the rules of the ABI.
4089 These two options select variations on the SPARC architecture.
4091 By default (unless specifically configured for the Fujitsu SPARClite),
4092 GCC generates code for the v7 variant of the SPARC architecture.
4094 @samp{-mv8} will give you SPARC v8 code. The only difference from v7
4095 code is that the compiler emits the integer multiply and integer
4096 divide instructions which exist in SPARC v8 but not in SPARC v7.
4098 @samp{-msparclite} will give you SPARClite code. This adds the integer
4099 multiply, integer divide step and scan (@code{ffs}) instructions which
4100 exist in SPARClite but not in SPARC v7.
4102 These options are deprecated and will be deleted in a future GCC release.
4103 They have been replaced with @samp{-mcpu=xxx}.
4107 These two options select the processor for which the code is optimised.
4109 With @samp{-mcypress} (the default), the compiler optimizes code for the
4110 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
4111 This is also appropriate for the older SparcStation 1, 2, IPX etc.
4113 With @samp{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
4114 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
4115 of the full SPARC v8 instruction set.
4117 These options are deprecated and will be deleted in a future GCC release.
4118 They have been replaced with @samp{-mcpu=xxx}.
4120 @item -mcpu=@var{cpu_type}
4121 Set the instruction set, register set, and instruction scheduling parameters
4122 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
4123 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
4124 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
4125 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
4127 Default instruction scheduling parameters are used for values that select
4128 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
4129 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
4131 Here is a list of each supported architecture and their supported
4136 v8: supersparc, hypersparc
4137 sparclite: f930, f934, sparclite86x
4142 @item -mtune=@var{cpu_type}
4143 Set the instruction scheduling parameters for machine type
4144 @var{cpu_type}, but do not set the instruction set or register set that the
4145 option @samp{-mcpu=}@var{cpu_type} would.
4147 The same values for @samp{-mcpu=}@var{cpu_type} are used for
4148 @samp{-mtune=}@*@var{cpu_type}, though the only useful values are those that
4149 select a particular cpu implementation: @samp{cypress}, @samp{supersparc},
4150 @samp{hypersparc}, @samp{f930}, @samp{f934}, @samp{sparclite86x},
4151 @samp{tsc701}, @samp{ultrasparc}.
4155 These @samp{-m} switches are supported in addition to the above
4156 on the SPARCLET processor.
4159 @item -mlittle-endian
4160 Generate code for a processor running in little-endian mode.
4163 Treat register @code{%g0} as a normal register.
4164 GCC will continue to clobber it as necessary but will not assume
4165 it always reads as 0.
4167 @item -mbroken-saverestore
4168 Generate code that does not use non-trivial forms of the @code{save} and
4169 @code{restore} instructions. Early versions of the SPARCLET processor do
4170 not correctly handle @code{save} and @code{restore} instructions used with
4171 arguments. They correctly handle them used without arguments. A @code{save}
4172 instruction used without arguments increments the current window pointer
4173 but does not allocate a new stack frame. It is assumed that the window
4174 overflow trap handler will properly handle this case as will interrupt
4178 These @samp{-m} switches are supported in addition to the above
4179 on SPARC V9 processors in 64 bit environments.
4182 @item -mlittle-endian
4183 Generate code for a processor running in little-endian mode.
4187 Generate code for a 32 bit or 64 bit environment.
4188 The 32 bit environment sets int, long and pointer to 32 bits.
4189 The 64 bit environment sets int to 32 bits and long and pointer
4192 @item -mcmodel=medlow
4193 Generate code for the Medium/Low code model: the program must be linked
4194 in the low 32 bits of the address space. Pointers are 64 bits.
4195 Programs can be statically or dynamically linked.
4197 @item -mcmodel=medmid
4198 Generate code for the Medium/Middle code model: the program must be linked
4199 in the low 44 bits of the address space, the text segment must be less than
4200 2G bytes, and data segment must be within 2G of the text segment.
4201 Pointers are 64 bits.
4203 @item -mcmodel=medany
4204 Generate code for the Medium/Anywhere code model: the program may be linked
4205 anywhere in the address space, the text segment must be less than
4206 2G bytes, and data segment must be within 2G of the text segment.
4207 Pointers are 64 bits.
4209 @item -mcmodel=embmedany
4210 Generate code for the Medium/Anywhere code model for embedded systems:
4211 assume a 32 bit text and a 32 bit data segment, both starting anywhere
4212 (determined at link time). Register %g4 points to the base of the
4213 data segment. Pointers still 64 bits.
4214 Programs are statically linked, PIC is not supported.
4217 @itemx -mno-stack-bias
4218 With @samp{-mstack-bias}, GCC assumes that the stack pointer, and
4219 frame pointer if present, are offset by -2047 which must be added back
4220 when making stack frame references.
4221 Otherwise, assume no such offset is present.
4224 @node Convex Options
4225 @subsection Convex Options
4226 @cindex Convex options
4228 These @samp{-m} options are defined for Convex:
4232 Generate output for C1. The code will run on any Convex machine.
4233 The preprocessor symbol @code{__convex__c1__} is defined.
4236 Generate output for C2. Uses instructions not available on C1.
4237 Scheduling and other optimizations are chosen for max performance on C2.
4238 The preprocessor symbol @code{__convex_c2__} is defined.
4241 Generate output for C32xx. Uses instructions not available on C1.
4242 Scheduling and other optimizations are chosen for max performance on C32.
4243 The preprocessor symbol @code{__convex_c32__} is defined.
4246 Generate output for C34xx. Uses instructions not available on C1.
4247 Scheduling and other optimizations are chosen for max performance on C34.
4248 The preprocessor symbol @code{__convex_c34__} is defined.
4251 Generate output for C38xx. Uses instructions not available on C1.
4252 Scheduling and other optimizations are chosen for max performance on C38.
4253 The preprocessor symbol @code{__convex_c38__} is defined.
4256 Generate code which puts an argument count in the word preceding each
4257 argument list. This is compatible with regular CC, and a few programs
4258 may need the argument count word. GDB and other source-level debuggers
4259 do not need it; this info is in the symbol table.
4262 Omit the argument count word. This is the default.
4264 @item -mvolatile-cache
4265 Allow volatile references to be cached. This is the default.
4267 @item -mvolatile-nocache
4268 Volatile references bypass the data cache, going all the way to memory.
4269 This is only needed for multi-processor code that does not use standard
4270 synchronization instructions. Making non-volatile references to volatile
4271 locations will not necessarily work.
4274 Type long is 32 bits, the same as type int. This is the default.
4277 Type long is 64 bits, the same as type long long. This option is useless,
4278 because no library support exists for it.
4281 @node AMD29K Options
4282 @subsection AMD29K Options
4283 @cindex AMD29K options
4285 These @samp{-m} options are defined for the AMD Am29000:
4290 @cindex DW bit (29k)
4291 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
4292 halfword operations are directly supported by the hardware. This is the
4297 Generate code that assumes the @code{DW} bit is not set.
4301 @cindex byte writes (29k)
4302 Generate code that assumes the system supports byte and halfword write
4303 operations. This is the default.
4307 Generate code that assumes the systems does not support byte and
4308 halfword write operations. @samp{-mnbw} implies @samp{-mndw}.
4312 @cindex memory model (29k)
4313 Use a small memory model that assumes that all function addresses are
4314 either within a single 256 KB segment or at an absolute address of less
4315 than 256k. This allows the @code{call} instruction to be used instead
4316 of a @code{const}, @code{consth}, @code{calli} sequence.
4320 Use the normal memory model: Generate @code{call} instructions only when
4321 calling functions in the same file and @code{calli} instructions
4322 otherwise. This works if each file occupies less than 256 KB but allows
4323 the entire executable to be larger than 256 KB. This is the default.
4326 Always use @code{calli} instructions. Specify this option if you expect
4327 a single file to compile into more than 256 KB of code.
4331 @cindex processor selection (29k)
4332 Generate code for the Am29050.
4336 Generate code for the Am29000. This is the default.
4338 @item -mkernel-registers
4339 @kindex -mkernel-registers
4340 @cindex kernel and user registers (29k)
4341 Generate references to registers @code{gr64-gr95} instead of to
4342 registers @code{gr96-gr127}. This option can be used when compiling
4343 kernel code that wants a set of global registers disjoint from that used
4346 Note that when this option is used, register names in @samp{-f} flags
4347 must use the normal, user-mode, names.
4349 @item -muser-registers
4350 @kindex -muser-registers
4351 Use the normal set of global registers, @code{gr96-gr127}. This is the
4355 @itemx -mno-stack-check
4356 @kindex -mstack-check
4357 @cindex stack checks (29k)
4358 Insert (or do not insert) a call to @code{__msp_check} after each stack
4359 adjustment. This is often used for kernel code.
4362 @itemx -mno-storem-bug
4363 @kindex -mstorem-bug
4364 @cindex storem bug (29k)
4365 @samp{-mstorem-bug} handles 29k processors which cannot handle the
4366 separation of a mtsrim insn and a storem instruction (most 29000 chips
4367 to date, but not the 29050).
4369 @item -mno-reuse-arg-regs
4370 @itemx -mreuse-arg-regs
4371 @kindex -mreuse-arg-regs
4372 @samp{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
4373 registers for copying out arguments. This helps detect calling a function
4374 with fewer arguments than it was declared with.
4376 @item -mno-impure-text
4377 @itemx -mimpure-text
4378 @kindex -mimpure-text
4379 @samp{-mimpure-text}, used in addition to @samp{-shared}, tells the compiler to
4380 not pass @samp{-assert pure-text} to the linker when linking a shared object.
4383 @kindex -msoft-float
4384 Generate output containing library calls for floating point.
4385 @strong{Warning:} the requisite libraries are not part of GCC.
4386 Normally the facilities of the machine's usual C compiler are used, but
4387 this can't be done directly in cross-compilation. You must make your
4388 own arrangements to provide suitable library functions for
4393 Do not generate multm or multmu instructions. This is useful for some embedded
4394 systems which do not have trap handlers for these instructions.
4398 @subsection ARM Options
4401 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
4406 @kindex -mapcs-frame
4407 Generate a stack frame that is compliant with the ARM Procedure Call
4408 Standard for all functions, even if this is not strictly necessary for
4409 correct execution of the code. Specifying @samp{-fomit-frame-pointer}
4410 with this option will cause the stack frames not to be generated for
4411 leaf functions. The default is @samp{-mno-apcs-frame}.
4415 This is a synonym for @samp{-mapcs-frame}.
4419 Generate code for a processor running with a 26-bit program counter,
4420 and conforming to the function calling standards for the APCS 26-bit
4421 option. This option replaces the @samp{-m2} and @samp{-m3} options
4422 of previous releases of the compiler.
4426 Generate code for a processor running with a 32-bit program counter,
4427 and conforming to the function calling standards for the APCS 32-bit
4428 option. This option replaces the @samp{-m6} option of previous releases
4431 @item -mapcs-stack-check
4432 @kindex -mapcs-stack-check
4433 @kindex -mno-apcs-stack-check
4434 Generate code to check the amount of stack space available upon entry to
4435 every function (that actually uses some stack space). If there is
4436 insufficient space available then either the function
4437 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
4438 called, depending upon the amount of stack space required. The run time
4439 system is required to provide these functions. The default is
4440 @samp{-mno-apcs-stack-check}, since this produces smaller code.
4443 @kindex -mapcs-float
4444 @kindex -mno-apcs-float
4445 Pass floating point arguments using the float point registers. This is
4446 one of the variants of the APCS. This option is recommended if the
4447 target hardware has a floating point unit or if a lot of floating point
4448 arithmetic is going to be performed by the code. The default is
4449 @samp{-mno-apcs-float}, since integer only code is slightly increased in
4450 size if @samp{-mapcs-float} is used.
4452 @item -mapcs-reentrant
4453 @kindex -mapcs-reentrant
4454 @kindex -mno-apcs-reentrant
4455 Generate reentrant, position independent code. This is the equivalent
4456 to specifying the @samp{-fpic} option. The default is
4457 @samp{-mno-apcs-reentrant}.
4459 @item -mthumb-interwork
4460 @kindex -mthumb-interwork
4461 @kindex -mno-thumb-interwork
4462 Generate code which supports calling between the ARM and THUMB
4463 instruction sets. Without this option the two instruction sets cannot
4464 be reliably used inside one program. The default is
4465 @samp{-mno-thumb-interwork}, since slightly larger code is generated
4466 when @samp{-mthumb-interwork} is specified.
4468 @item -mno-sched-prolog
4469 @kindex -mno-sched-prolog
4470 @kindex -msched-prolog
4471 Prevent the reordering of instructions in the function prolog, or the
4472 merging of those instruction with the instructions in the function's
4473 body. This means that all functions will start with a recognizable set
4474 of instructions (or in fact one of a choice from a small set of
4475 different function prologues), and this information can be used to
4476 locate the start if functions inside an executable piece of code. The
4477 default is @samp{-msched-prolog}.
4480 Generate output containing floating point instructions. This is the
4484 Generate output containing library calls for floating point.
4485 @strong{Warning:} the requisite libraries are not available for all ARM
4486 targets. Normally the facilities of the machine's usual C compiler are
4487 used, but this cannot be done directly in cross-compilation. You must make
4488 your own arrangements to provide suitable library functions for
4491 @samp{-msoft-float} changes the calling convention in the output file;
4492 therefore, it is only useful if you compile @emph{all} of a program with
4493 this option. In particular, you need to compile @file{libgcc.a}, the
4494 library that comes with GCC, with @samp{-msoft-float} in order for
4497 @item -mlittle-endian
4498 Generate code for a processor running in little-endian mode. This is
4499 the default for all standard configurations.
4502 Generate code for a processor running in big-endian mode; the default is
4503 to compile code for a little-endian processor.
4505 @item -mwords-little-endian
4506 This option only applies when generating code for big-endian processors.
4507 Generate code for a little-endian word order but a big-endian byte
4508 order. That is, a byte order of the form @samp{32107654}. Note: this
4509 option should only be used if you require compatibility with code for
4510 big-endian ARM processors generated by versions of the compiler prior to
4513 @item -malignment-traps
4514 @kindex -malignment-traps
4515 Generate code that will not trap if the MMU has alignment traps enabled.
4516 On ARM architectures prior to ARMv4, there were no instructions to
4517 access half-word objects stored in memory. However, when reading from
4518 memory a feature of the ARM architecture allows a word load to be used,
4519 even if the address is unaligned, and the processor core will rotate the
4520 data as it is being loaded. This option tells the compiler that such
4521 misaligned accesses will cause a MMU trap and that it should instead
4522 synthesise the access as a series of byte accesses. The compiler can
4523 still use word accesses to load half-word data if it knows that the
4524 address is aligned to a word boundary.
4526 This option is ignored when compiling for ARM architecture 4 or later,
4527 since these processors have instructions to directly access half-word
4530 @item -mno-alignment-traps
4531 @kindex -mno-alignment-traps
4532 Generate code that assumes that the MMU will not trap unaligned
4533 accesses. This produces better code when the target instruction set
4534 does not have half-word memory operations (implementations prior to
4537 Note that you cannot use this option to access unaligned word objects,
4538 since the processor will only fetch one 32-bit aligned object from
4541 The default setting for most targets is -mno-alignment-traps, since
4542 this produces better code when there are no half-word memory
4543 instructions available.
4545 @item -mshort-load-bytes
4546 @kindex -mshort-load-bytes
4547 This is a depreciated alias for @samp{-malignment-traps}.
4549 @item -mno-short-load-bytes
4550 @kindex -mno-short-load-bytes
4551 This is a depreciated alias for @samp{-mno-alignment-traps}.
4553 @item -mshort-load-words
4554 @kindex -mshort-load-words
4555 This is a depreciated alias for @samp{-mno-alignment-traps}.
4557 @item -mno-short-load-words
4558 @kindex -mno-short-load-words
4559 This is a depreciated alias for @samp{-malignment-traps}.
4563 This option only applies to RISC iX. Emulate the native BSD-mode
4564 compiler. This is the default if @samp{-ansi} is not specified.
4568 This option only applies to RISC iX. Emulate the native X/Open-mode
4571 @item -mno-symrename
4572 @kindex -mno-symrename
4573 This option only applies to RISC iX. Do not run the assembler
4574 post-processor, @samp{symrename}, after code has been assembled.
4575 Normally it is necessary to modify some of the standard symbols in
4576 preparation for linking with the RISC iX C library; this option
4577 suppresses this pass. The post-processor is never run when the
4578 compiler is built for cross-compilation.
4582 This specifies the name of the target ARM processor. GCC uses this name
4583 to determine what kind of instructions it can use when generating
4584 assembly code. Permissible names are: arm2, arm250, arm3, arm6, arm60,
4585 arm600, arm610, arm620, arm7, arm7m, arm7d, arm7dm, arm7di, arm7dmi,
4586 arm70, arm700, arm700i, arm710, arm710c, arm7100, arm7500, arm7500fe,
4587 arm7tdmi, arm8, strongarm, strongarm110, strongarm1100, arm8, arm810,
4588 arm9, arm920, arm920t, arm9tdmi.
4590 @itemx -mtune=<name>
4592 This option is very similar to the @samp{-mcpu=} option, except that
4593 instead of specifying the actual target processor type, and hence
4594 restricting which instructions can be used, it specifies that GCC should
4595 tune the performance of the code as if the target were of the type
4596 specified in this option, but still choosing the instructions that it
4597 will generate based on the cpu specified by a @samp{-mcpu=} option.
4598 For some arm implementations better performance can be obtained by using
4603 This specifies the name of the target ARM architecture. GCC uses this
4604 name to determine what kind of instructions it can use when generating
4605 assembly code. This option can be used in conjunction with or instead
4606 of the @samp{-mcpu=} option. Permissible names are: armv2, armv2a,
4607 armv3, armv3m, armv4, armv4t, armv5.
4609 @item -mfpe=<number>
4610 @itemx -mfp=<number>
4613 This specifes the version of the floating point emulation available on
4614 the target. Permissible values are 2 and 3. @samp{-mfp=} is a synonym
4615 for @samp{-mfpe=} to support older versions of GCC.
4617 @item -mstructure-size-boundary=<n>
4618 @kindex -mstructure-size-boundary
4619 The size of all structures and unions will be rounded up to a multiple
4620 of the number of bits set by this option. Permissible values are 8 and
4621 32. The default value varies for different toolchains. For the COFF
4622 targeted toolchain the default value is 8. Specifying the larger number
4623 can produce faster, more efficient code, but can also increase the size
4624 of the program. The two values are potentially incompatible. Code
4625 compiled with one value cannot necessarily expect to work with code or
4626 libraries compiled with the other value, if they exchange information
4627 using structures or unions. Programmers are encouraged to use the 32
4628 value as future versions of the toolchain may default to this value.
4630 @item -mabort-on-noreturn
4631 @kindex -mabort-on-noreturn
4632 @kindex -mnoabort-on-noreturn
4633 Generate a call to the function abort at the end of a noreturn function.
4634 It will be executed if the function tries to return.
4637 @itemx -mno-long-calls
4638 Tells the compiler to perform function calls by first loading the
4639 address of the function into a register and then performing a subroutine
4640 call on this register. This switch is needed if the target function
4641 will lie outside of the 64 megabyte addressing range of the offset based
4642 version of subroutine call instruction.
4644 Even if this switch is enabled, not all function calls will be turned
4645 into long calls. The heuristic is that static functions, functions
4646 which have the @samp{short-call} attribute, functions that are inside
4647 the scope of a @samp{#pragma no_long_calls} directive and functions whose
4648 definitions have already been compiled within the current compilation
4649 unit, will not be turned into long calls. The exception to this rule is
4650 that weak function defintions, functions with the @samp{long-call}
4651 attribute or the @samp{section} attribute, and functions that are within
4652 the scope of a @samp{#pragma long_calls} directive, will always be
4653 turned into long calls.
4655 This feature is not enabled by default. Specifying
4656 @samp{--no-long-calls} will restore the default behaviour, as will
4657 placing the function calls within the scope of a @samp{#pragma
4658 long_calls_off} directive. Note these switches have no effect on how
4659 the compiler generates code to handle function calls via function
4662 @item -mnop-fun-dllimport
4663 @kindex -mnop-fun-dllimport
4664 Disable the support for the @emph{dllimport} attribute.
4666 @item -msingle-pic-base
4667 @kindex -msingle-pic-base
4668 Treat the register used for PIC addressing as read-only, rather than
4669 loading it in the prologue for each function. The run-time system is
4670 responsible for initialising this register with an appropriate value
4671 before execution begins.
4673 @item -mpic-register=<reg>
4674 @kindex -mpic-register=
4675 Specify the register to be used for PIC addressing. The default is R10
4676 unless stack-checking is enabled, when R9 is used.
4681 @subsection Thumb Options
4682 @cindex Thumb Options
4686 @item -mthumb-interwork
4687 @kindex -mthumb-interwork
4688 @kindex -mno-thumb-interwork
4689 Generate code which supports calling between the THUMB and ARM
4690 instruction sets. Without this option the two instruction sets cannot
4691 be reliably used inside one program. The default is
4692 @samp{-mno-thumb-interwork}, since slightly smaller code is generated
4696 @kindex -mtpcs-frame
4697 @kindex -mno-tpcs-frame
4698 Generate a stack frame that is compliant with the Thumb Procedure Call
4699 Standard for all non-leaf functions. (A leaf function is one that does
4700 not call any other functions). The default is @samp{-mno-apcs-frame}.
4702 @item -mtpcs-leaf-frame
4703 @kindex -mtpcs-leaf-frame
4704 @kindex -mno-tpcs-leaf-frame
4705 Generate a stack frame that is compliant with the Thumb Procedure Call
4706 Standard for all leaf functions. (A leaf function is one that does
4707 not call any other functions). The default is @samp{-mno-apcs-leaf-frame}.
4709 @item -mlittle-endian
4710 @kindex -mlittle-endian
4711 Generate code for a processor running in little-endian mode. This is
4712 the default for all standard configurations.
4715 @kindex -mbig-endian
4716 Generate code for a processor running in big-endian mode.
4718 @item -mstructure-size-boundary=<n>
4719 @kindex -mstructure-size-boundary
4720 The size of all structures and unions will be rounded up to a multiple
4721 of the number of bits set by this option. Permissible values are 8 and
4722 32. The default value varies for different toolchains. For the COFF
4723 targeted toolchain the default value is 8. Specifying the larger number
4724 can produced faster, more efficient code, but can also increase the size
4725 of the program. The two values are potentially incompatible. Code
4726 compiled with one value cannot necessarily expect to work with code or
4727 libraries compiled with the other value, if they exchange information
4728 using structures or unions. Programmers are encouraged to use the 32
4729 value as future versions of the toolchain may default to this value.
4731 @item -mnop-fun-dllimport
4732 @kindex -mnop-fun-dllimport
4733 Disable the support for the @emph{dllimport} attribute.
4735 @item -mcallee-super-interworking
4736 @kindex -mcallee-super-interworking
4737 Gives all externally visible functions in the file being compiled an ARM
4738 instruction set header which switches to Thumb mode before executing the
4739 rest of the function. This allows these functions to be called from
4740 non-interworking code.
4742 @item -mcaller-super-interworking
4743 @kindex -mcaller-super-interworking
4744 Allows calls via function pointers (including virtual functions) to
4745 execute correctly regardless of whether the target code has been
4746 compiled for interworking or not. There is a small overhead in the cost
4747 of executing a function pointer if this option is enabled.
4749 @item -msingle-pic-base
4750 @kindex -msingle-pic-base
4751 Treat the register used for PIC addressing as read-only, rather than
4752 loading it in the prologue for each function. The run-time system is
4753 responsible for initialising this register with an appropriate value
4754 before execution begins.
4756 @item -mpic-register=<reg>
4757 @kindex -mpic-register=
4758 Specify the register to be used for PIC addressing. The default is R10.
4762 @node MN10200 Options
4763 @subsection MN10200 Options
4764 @cindex MN10200 options
4765 These @samp{-m} options are defined for Matsushita MN10200 architectures:
4769 Indicate to the linker that it should perform a relaxation optimization pass
4770 to shorten branches, calls and absolute memory addresses. This option only
4771 has an effect when used on the command line for the final link step.
4773 This option makes symbolic debugging impossible.
4776 @node MN10300 Options
4777 @subsection MN10300 Options
4778 @cindex MN10300 options
4779 These @samp{-m} options are defined for Matsushita MN10300 architectures:
4783 Generate code to avoid bugs in the multiply instructions for the MN10300
4784 processors. This is the default.
4787 Do not generate code to avoid bugs in the multiply instructions for the
4791 Generate code which uses features specific to the AM33 processor.
4794 Do not generate code which uses features specific to the AM33 processor. This
4798 Indicate to the linker that it should perform a relaxation optimization pass
4799 to shorten branches, calls and absolute memory addresses. This option only
4800 has an effect when used on the command line for the final link step.
4802 This option makes symbolic debugging impossible.
4806 @node M32R/D Options
4807 @subsection M32R/D Options
4808 @cindex M32R/D options
4810 These @samp{-m} options are defined for Mitsubishi M32R/D architectures:
4813 @item -mcode-model=small
4814 Assume all objects live in the lower 16MB of memory (so that their addresses
4815 can be loaded with the @code{ld24} instruction), and assume all subroutines
4816 are reachable with the @code{bl} instruction.
4817 This is the default.
4819 The addressability of a particular object can be set with the
4820 @code{model} attribute.
4822 @item -mcode-model=medium
4823 Assume objects may be anywhere in the 32 bit address space (the compiler
4824 will generate @code{seth/add3} instructions to load their addresses), and
4825 assume all subroutines are reachable with the @code{bl} instruction.
4827 @item -mcode-model=large
4828 Assume objects may be anywhere in the 32 bit address space (the compiler
4829 will generate @code{seth/add3} instructions to load their addresses), and
4830 assume subroutines may not be reachable with the @code{bl} instruction
4831 (the compiler will generate the much slower @code{seth/add3/jl}
4832 instruction sequence).
4835 Disable use of the small data area. Variables will be put into
4836 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
4837 @code{section} attribute has been specified).
4838 This is the default.
4840 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
4841 Objects may be explicitly put in the small data area with the
4842 @code{section} attribute using one of these sections.
4845 Put small global and static data in the small data area, but do not
4846 generate special code to reference them.
4849 Put small global and static data in the small data area, and generate
4850 special instructions to reference them.
4853 @cindex smaller data references
4854 Put global and static objects less than or equal to @var{num} bytes
4855 into the small data or bss sections instead of the normal data or bss
4856 sections. The default value of @var{num} is 8.
4857 The @samp{-msdata} option must be set to one of @samp{sdata} or @samp{use}
4858 for this option to have any effect.
4860 All modules should be compiled with the same @samp{-G @var{num}} value.
4861 Compiling with different values of @var{num} may or may not work; if it
4862 doesn't the linker will give an error message - incorrect code will not be
4868 @subsection M88K Options
4869 @cindex M88k options
4871 These @samp{-m} options are defined for Motorola 88k architectures:
4876 Generate code that works well on both the m88100 and the
4881 Generate code that works best for the m88100, but that also
4886 Generate code that works best for the m88110, and may not run
4891 Obsolete option to be removed from the next revision.
4894 @item -midentify-revision
4895 @kindex -midentify-revision
4897 @cindex identifying source, compiler (88k)
4898 Include an @code{ident} directive in the assembler output recording the
4899 source file name, compiler name and version, timestamp, and compilation
4902 @item -mno-underscores
4903 @kindex -mno-underscores
4904 @cindex underscores, avoiding (88k)
4905 In assembler output, emit symbol names without adding an underscore
4906 character at the beginning of each name. The default is to use an
4907 underscore as prefix on each name.
4909 @item -mocs-debug-info
4910 @itemx -mno-ocs-debug-info
4911 @kindex -mocs-debug-info
4912 @kindex -mno-ocs-debug-info
4914 @cindex debugging, 88k OCS
4915 Include (or omit) additional debugging information (about registers used
4916 in each stack frame) as specified in the 88open Object Compatibility
4917 Standard, ``OCS''. This extra information allows debugging of code that
4918 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
4919 Delta 88 SVr3.2 is to include this information; other 88k configurations
4920 omit this information by default.
4922 @item -mocs-frame-position
4923 @kindex -mocs-frame-position
4924 @cindex register positions in frame (88k)
4925 When emitting COFF debugging information for automatic variables and
4926 parameters stored on the stack, use the offset from the canonical frame
4927 address, which is the stack pointer (register 31) on entry to the
4928 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
4929 @samp{-mocs-frame-position}; other 88k configurations have the default
4930 @samp{-mno-ocs-frame-position}.
4932 @item -mno-ocs-frame-position
4933 @kindex -mno-ocs-frame-position
4934 @cindex register positions in frame (88k)
4935 When emitting COFF debugging information for automatic variables and
4936 parameters stored on the stack, use the offset from the frame pointer
4937 register (register 30). When this option is in effect, the frame
4938 pointer is not eliminated when debugging information is selected by the
4941 @item -moptimize-arg-area
4942 @itemx -mno-optimize-arg-area
4943 @kindex -moptimize-arg-area
4944 @kindex -mno-optimize-arg-area
4945 @cindex arguments in frame (88k)
4946 Control how function arguments are stored in stack frames.
4947 @samp{-moptimize-arg-area} saves space by optimizing them, but this
4948 conflicts with the 88open specifications. The opposite alternative,
4949 @samp{-mno-optimize-arg-area}, agrees with 88open standards. By default
4950 GCC does not optimize the argument area.
4952 @item -mshort-data-@var{num}
4953 @kindex -mshort-data-@var{num}
4954 @cindex smaller data references (88k)
4955 @cindex r0-relative references (88k)
4956 Generate smaller data references by making them relative to @code{r0},
4957 which allows loading a value using a single instruction (rather than the
4958 usual two). You control which data references are affected by
4959 specifying @var{num} with this option. For example, if you specify
4960 @samp{-mshort-data-512}, then the data references affected are those
4961 involving displacements of less than 512 bytes.
4962 @samp{-mshort-data-@var{num}} is not effective for @var{num} greater
4965 @item -mserialize-volatile
4966 @kindex -mserialize-volatile
4967 @itemx -mno-serialize-volatile
4968 @kindex -mno-serialize-volatile
4969 @cindex sequential consistency on 88k
4970 Do, or don't, generate code to guarantee sequential consistency
4971 of volatile memory references. By default, consistency is
4974 The order of memory references made by the MC88110 processor does
4975 not always match the order of the instructions requesting those
4976 references. In particular, a load instruction may execute before
4977 a preceding store instruction. Such reordering violates
4978 sequential consistency of volatile memory references, when there
4979 are multiple processors. When consistency must be guaranteed,
4980 GNU C generates special instructions, as needed, to force
4981 execution in the proper order.
4983 The MC88100 processor does not reorder memory references and so
4984 always provides sequential consistency. However, by default, GNU
4985 C generates the special instructions to guarantee consistency
4986 even when you use @samp{-m88100}, so that the code may be run on an
4987 MC88110 processor. If you intend to run your code only on the
4988 MC88100 processor, you may use @samp{-mno-serialize-volatile}.
4990 The extra code generated to guarantee consistency may affect the
4991 performance of your application. If you know that you can safely
4992 forgo this guarantee, you may use @samp{-mno-serialize-volatile}.
4998 @cindex assembler syntax, 88k
5000 Turn on (@samp{-msvr4}) or off (@samp{-msvr3}) compiler extensions
5001 related to System V release 4 (SVr4). This controls the following:
5005 Which variant of the assembler syntax to emit.
5007 @samp{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
5008 that is used on System V release 4.
5010 @samp{-msvr4} makes GCC issue additional declaration directives used in
5014 @samp{-msvr4} is the default for the m88k-motorola-sysv4 and
5015 m88k-dg-dgux m88k configurations. @samp{-msvr3} is the default for all
5016 other m88k configurations.
5018 @item -mversion-03.00
5019 @kindex -mversion-03.00
5020 This option is obsolete, and is ignored.
5021 @c ??? which asm syntax better for GAS? option there too?
5023 @item -mno-check-zero-division
5024 @itemx -mcheck-zero-division
5025 @kindex -mno-check-zero-division
5026 @kindex -mcheck-zero-division
5027 @cindex zero division on 88k
5028 Do, or don't, generate code to guarantee that integer division by
5029 zero will be detected. By default, detection is guaranteed.
5031 Some models of the MC88100 processor fail to trap upon integer
5032 division by zero under certain conditions. By default, when
5033 compiling code that might be run on such a processor, GNU C
5034 generates code that explicitly checks for zero-valued divisors
5035 and traps with exception number 503 when one is detected. Use of
5036 mno-check-zero-division suppresses such checking for code
5037 generated to run on an MC88100 processor.
5039 GNU C assumes that the MC88110 processor correctly detects all
5040 instances of integer division by zero. When @samp{-m88110} is
5041 specified, both @samp{-mcheck-zero-division} and
5042 @samp{-mno-check-zero-division} are ignored, and no explicit checks for
5043 zero-valued divisors are generated.
5045 @item -muse-div-instruction
5046 @kindex -muse-div-instruction
5047 @cindex divide instruction, 88k
5048 Use the div instruction for signed integer division on the
5049 MC88100 processor. By default, the div instruction is not used.
5051 On the MC88100 processor the signed integer division instruction
5052 div) traps to the operating system on a negative operand. The
5053 operating system transparently completes the operation, but at a
5054 large cost in execution time. By default, when compiling code
5055 that might be run on an MC88100 processor, GNU C emulates signed
5056 integer division using the unsigned integer division instruction
5057 divu), thereby avoiding the large penalty of a trap to the
5058 operating system. Such emulation has its own, smaller, execution
5059 cost in both time and space. To the extent that your code's
5060 important signed integer division operations are performed on two
5061 nonnegative operands, it may be desirable to use the div
5062 instruction directly.
5064 On the MC88110 processor the div instruction (also known as the
5065 divs instruction) processes negative operands without trapping to
5066 the operating system. When @samp{-m88110} is specified,
5067 @samp{-muse-div-instruction} is ignored, and the div instruction is used
5068 for signed integer division.
5070 Note that the result of dividing INT_MIN by -1 is undefined. In
5071 particular, the behavior of such a division with and without
5072 @samp{-muse-div-instruction} may differ.
5074 @item -mtrap-large-shift
5075 @itemx -mhandle-large-shift
5076 @kindex -mtrap-large-shift
5077 @kindex -mhandle-large-shift
5078 @cindex bit shift overflow (88k)
5079 @cindex large bit shifts (88k)
5080 Include code to detect bit-shifts of more than 31 bits; respectively,
5081 trap such shifts or emit code to handle them properly. By default GCC
5082 makes no special provision for large bit shifts.
5084 @item -mwarn-passed-structs
5085 @kindex -mwarn-passed-structs
5086 @cindex structure passing (88k)
5087 Warn when a function passes a struct as an argument or result.
5088 Structure-passing conventions have changed during the evolution of the C
5089 language, and are often the source of portability problems. By default,
5090 GCC issues no such warning.
5093 @node RS/6000 and PowerPC Options
5094 @subsection IBM RS/6000 and PowerPC Options
5095 @cindex RS/6000 and PowerPC Options
5096 @cindex IBM RS/6000 and PowerPC Options
5098 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
5106 @itemx -mpowerpc-gpopt
5107 @itemx -mno-powerpc-gpopt
5108 @itemx -mpowerpc-gfxopt
5109 @itemx -mno-powerpc-gfxopt
5111 @itemx -mno-powerpc64
5115 @kindex -mpowerpc-gpopt
5116 @kindex -mpowerpc-gfxopt
5118 GCC supports two related instruction set architectures for the
5119 RS/6000 and PowerPC. The @dfn{POWER} instruction set are those
5120 instructions supported by the @samp{rios} chip set used in the original
5121 RS/6000 systems and the @dfn{PowerPC} instruction set is the
5122 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
5123 the IBM 4xx microprocessors.
5125 Neither architecture is a subset of the other. However there is a
5126 large common subset of instructions supported by both. An MQ
5127 register is included in processors supporting the POWER architecture.
5129 You use these options to specify which instructions are available on the
5130 processor you are using. The default value of these options is
5131 determined when configuring GCC. Specifying the
5132 @samp{-mcpu=@var{cpu_type}} overrides the specification of these
5133 options. We recommend you use the @samp{-mcpu=@var{cpu_type}} option
5134 rather than the options listed above.
5136 The @samp{-mpower} option allows GCC to generate instructions that
5137 are found only in the POWER architecture and to use the MQ register.
5138 Specifying @samp{-mpower2} implies @samp{-power} and also allows GCC
5139 to generate instructions that are present in the POWER2 architecture but
5140 not the original POWER architecture.
5142 The @samp{-mpowerpc} option allows GCC to generate instructions that
5143 are found only in the 32-bit subset of the PowerPC architecture.
5144 Specifying @samp{-mpowerpc-gpopt} implies @samp{-mpowerpc} and also allows
5145 GCC to use the optional PowerPC architecture instructions in the
5146 General Purpose group, including floating-point square root. Specifying
5147 @samp{-mpowerpc-gfxopt} implies @samp{-mpowerpc} and also allows GCC to
5148 use the optional PowerPC architecture instructions in the Graphics
5149 group, including floating-point select.
5151 The @samp{-mpowerpc64} option allows GCC to generate the additional
5152 64-bit instructions that are found in the full PowerPC64 architecture
5153 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
5154 @samp{-mno-powerpc64}.
5156 If you specify both @samp{-mno-power} and @samp{-mno-powerpc}, GCC
5157 will use only the instructions in the common subset of both
5158 architectures plus some special AIX common-mode calls, and will not use
5159 the MQ register. Specifying both @samp{-mpower} and @samp{-mpowerpc}
5160 permits GCC to use any instruction from either architecture and to
5161 allow use of the MQ register; specify this for the Motorola MPC601.
5163 @item -mnew-mnemonics
5164 @itemx -mold-mnemonics
5165 @kindex -mnew-mnemonics
5166 @kindex -mold-mnemonics
5167 Select which mnemonics to use in the generated assembler code.
5168 @samp{-mnew-mnemonics} requests output that uses the assembler mnemonics
5169 defined for the PowerPC architecture, while @samp{-mold-mnemonics}
5170 requests the assembler mnemonics defined for the POWER architecture.
5171 Instructions defined in only one architecture have only one mnemonic;
5172 GCC uses that mnemonic irrespective of which of these options is
5175 GCC defaults to the mnemonics appropriate for the architecture in
5176 use. Specifying @samp{-mcpu=@var{cpu_type}} sometimes overrides the
5177 value of these option. Unless you are building a cross-compiler, you
5178 should normally not specify either @samp{-mnew-mnemonics} or
5179 @samp{-mold-mnemonics}, but should instead accept the default.
5181 @item -mcpu=@var{cpu_type}
5183 Set architecture type, register usage, choice of mnemonics, and
5184 instruction scheduling parameters for machine type @var{cpu_type}.
5185 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
5186 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
5187 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
5188 @samp{630}, @samp{740}, @samp{750}, @samp{power}, @samp{power2},
5189 @samp{powerpc}, @samp{403}, @samp{505}, @samp{801}, @samp{821},
5190 @samp{823}, and @samp{860} and @samp{common}. @samp{-mcpu=power},
5191 @samp{-mcpu=power2}, @samp{-mcpu=powerpc}, and @samp{-mcpu=powerpc64}
5192 specify generic POWER, POWER2, pure 32-bit PowerPC (i.e., not MPC601),
5193 and 64-bit PowerPC architecture machine types, with an appropriate,
5194 generic processor model assumed for scheduling purposes.@refill
5196 Specifying any of the following options:
5197 @samp{-mcpu=rios1}, @samp{-mcpu=rios2}, @samp{-mcpu=rsc},
5198 @samp{-mcpu=power}, or @samp{-mcpu=power2}
5199 enables the @samp{-mpower} option and disables the @samp{-mpowerpc} option;
5200 @samp{-mcpu=601} enables both the @samp{-mpower} and @samp{-mpowerpc} options.
5201 All of @samp{-mcpu=rs64a}, @samp{-mcpu=602}, @samp{-mcpu=603},
5202 @samp{-mcpu=603e}, @samp{-mcpu=604}, @samp{-mcpu=620}, @samp{-mcpu=630},
5203 @samp{-mcpu=740}, and @samp{-mcpu=750}
5204 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5205 Exactly similarly, all of @samp{-mcpu=403},
5206 @samp{-mcpu=505}, @samp{-mcpu=821}, @samp{-mcpu=860} and @samp{-mcpu=powerpc}
5207 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5208 @samp{-mcpu=common} disables both the
5209 @samp{-mpower} and @samp{-mpowerpc} options.@refill
5211 AIX versions 4 or greater selects @samp{-mcpu=common} by default, so
5212 that code will operate on all members of the RS/6000 POWER and PowerPC
5213 families. In that case, GCC will use only the instructions in the
5214 common subset of both architectures plus some special AIX common-mode
5215 calls, and will not use the MQ register. GCC assumes a generic
5216 processor model for scheduling purposes.
5218 Specifying any of the options @samp{-mcpu=rios1}, @samp{-mcpu=rios2},
5219 @samp{-mcpu=rsc}, @samp{-mcpu=power}, or @samp{-mcpu=power2} also
5220 disables the @samp{new-mnemonics} option. Specifying @samp{-mcpu=601},
5221 @samp{-mcpu=602}, @samp{-mcpu=603}, @samp{-mcpu=603e}, @samp{-mcpu=604},
5222 @samp{-mcpu=620}, @samp{-mcpu=630}, @samp{-mcpu=403}, @samp{-mcpu=505},
5223 @samp{-mcpu=821}, @samp{-mcpu=860} or @samp{-mcpu=powerpc} also enables
5224 the @samp{new-mnemonics} option.@refill
5226 Specifying @samp{-mcpu=403}, @samp{-mcpu=821}, or @samp{-mcpu=860} also
5227 enables the @samp{-msoft-float} option.
5229 @item -mtune=@var{cpu_type}
5230 Set the instruction scheduling parameters for machine type
5231 @var{cpu_type}, but do not set the architecture type, register usage,
5232 choice of mnemonics like @samp{-mcpu=}@var{cpu_type} would. The same
5233 values for @var{cpu_type} are used for @samp{-mtune=}@var{cpu_type} as
5234 for @samp{-mcpu=}@var{cpu_type}. The @samp{-mtune=}@var{cpu_type}
5235 option overrides the @samp{-mcpu=}@var{cpu_type} option in terms of
5236 instruction scheduling parameters.
5239 @itemx -mno-fp-in-toc
5240 @itemx -mno-sum-in-toc
5241 @itemx -mminimal-toc
5242 @kindex -mminimal-toc
5243 Modify generation of the TOC (Table Of Contents), which is created for
5244 every executable file. The @samp{-mfull-toc} option is selected by
5245 default. In that case, GCC will allocate at least one TOC entry for
5246 each unique non-automatic variable reference in your program. GCC
5247 will also place floating-point constants in the TOC. However, only
5248 16,384 entries are available in the TOC.
5250 If you receive a linker error message that saying you have overflowed
5251 the available TOC space, you can reduce the amount of TOC space used
5252 with the @samp{-mno-fp-in-toc} and @samp{-mno-sum-in-toc} options.
5253 @samp{-mno-fp-in-toc} prevents GCC from putting floating-point
5254 constants in the TOC and @samp{-mno-sum-in-toc} forces GCC to
5255 generate code to calculate the sum of an address and a constant at
5256 run-time instead of putting that sum into the TOC. You may specify one
5257 or both of these options. Each causes GCC to produce very slightly
5258 slower and larger code at the expense of conserving TOC space.
5260 If you still run out of space in the TOC even when you specify both of
5261 these options, specify @samp{-mminimal-toc} instead. This option causes
5262 GCC to make only one TOC entry for every file. When you specify this
5263 option, GCC will produce code that is slower and larger but which
5264 uses extremely little TOC space. You may wish to use this option
5265 only on files that contain less frequently executed code. @refill
5271 Enable 64-bit PowerPC ABI and calling convention: 64-bit pointers, 64-bit
5272 @code{long} type, and the infrastructure needed to support them.
5273 Specifying @samp{-m64} implies @samp{-mpowerpc64} and
5274 @samp{-mpowerpc}, while @samp{-m32} disables the 64-bit ABI and
5275 implies @samp{-mno-powerpc64}. GCC defaults to @samp{-m32}.
5280 On AIX, pass floating-point arguments to prototyped functions beyond the
5281 register save area (RSA) on the stack in addition to argument FPRs. The
5282 AIX calling convention was extended but not initially documented to
5283 handle an obscure K&R C case of calling a function that takes the
5284 address of its arguments with fewer arguments than declared. AIX XL
5285 compilers access floating point arguments which do not fit in the
5286 RSA from the stack when a subroutine is compiled without
5287 optimization. Because always storing floating-point arguments on the
5288 stack is inefficient and rarely needed, this option is not enabled by
5289 default and only is necessary when calling subroutines compiled by AIX
5290 XL compilers without optimization.
5294 Support @dfn{AIX Threads}. Link an application written to use
5295 @dfn{pthreads} with special libraries and startup code to enable the
5300 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE). Link an
5301 application written to use message passing with special startup code to
5302 enable the application to run. The system must have PE installed in the
5303 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
5304 must be overridden with the @samp{-specs=} option to specify the
5305 appropriate directory location. The Parallel Environment does not
5306 support threads, so the @samp{-mpe} option and the @samp{-mthreads}
5307 option are incompatible.
5311 @kindex -msoft-float
5312 Generate code that does not use (uses) the floating-point register set.
5313 Software floating point emulation is provided if you use the
5314 @samp{-msoft-float} option, and pass the option to GCC when linking.
5317 @itemx -mno-multiple
5318 Generate code that uses (does not use) the load multiple word
5319 instructions and the store multiple word instructions. These
5320 instructions are generated by default on POWER systems, and not
5321 generated on PowerPC systems. Do not use @samp{-mmultiple} on little
5322 endian PowerPC systems, since those instructions do not work when the
5323 processor is in little endian mode. The exceptions are PPC740 and
5324 PPC750 which permit the instructions usage in little endian mode.
5329 Generate code that uses (does not use) the load string instructions
5330 and the store string word instructions to save multiple registers and
5331 do small block moves. These instructions are generated by default on
5332 POWER systems, and not generated on PowerPC systems. Do not use
5333 @samp{-mstring} on little endian PowerPC systems, since those
5334 instructions do not work when the processor is in little endian mode.
5335 The exceptions are PPC740 and PPC750 which permit the instructions
5336 usage in little endian mode.
5341 Generate code that uses (does not use) the load or store instructions
5342 that update the base register to the address of the calculated memory
5343 location. These instructions are generated by default. If you use
5344 @samp{-mno-update}, there is a small window between the time that the
5345 stack pointer is updated and the address of the previous frame is
5346 stored, which means code that walks the stack frame across interrupts or
5347 signals may get corrupted data.
5350 @itemx -mno-fused-madd
5351 @kindex -mfused-madd
5352 Generate code that uses (does not use) the floating point multiply and
5353 accumulate instructions. These instructions are generated by default if
5354 hardware floating is used.
5356 @item -mno-bit-align
5359 On System V.4 and embedded PowerPC systems do not (do) force structures
5360 and unions that contain bit fields to be aligned to the base type of the
5363 For example, by default a structure containing nothing but 8
5364 @code{unsigned} bitfields of length 1 would be aligned to a 4 byte
5365 boundary and have a size of 4 bytes. By using @samp{-mno-bit-align},
5366 the structure would be aligned to a 1 byte boundary and be one byte in
5369 @item -mno-strict-align
5370 @itemx -mstrict-align
5371 @kindex -mstrict-align
5372 On System V.4 and embedded PowerPC systems do not (do) assume that
5373 unaligned memory references will be handled by the system.
5376 @itemx -mno-relocatable
5377 @kindex -mrelocatable
5378 On embedded PowerPC systems generate code that allows (does not allow)
5379 the program to be relocated to a different address at runtime. If you
5380 use @samp{-mrelocatable} on any module, all objects linked together must
5381 be compiled with @samp{-mrelocatable} or @samp{-mrelocatable-lib}.
5383 @item -mrelocatable-lib
5384 @itemx -mno-relocatable-lib
5385 On embedded PowerPC systems generate code that allows (does not allow)
5386 the program to be relocated to a different address at runtime. Modules
5387 compiled with @samp{-mrelocatable-lib} can be linked with either modules
5388 compiled without @samp{-mrelocatable} and @samp{-mrelocatable-lib} or
5389 with modules compiled with the @samp{-mrelocatable} options.
5393 On System V.4 and embedded PowerPC systems do not (do) assume that
5394 register 2 contains a pointer to a global area pointing to the addresses
5395 used in the program.
5398 @itemx -mlittle-endian
5399 On System V.4 and embedded PowerPC systems compile code for the
5400 processor in little endian mode. The @samp{-mlittle-endian} option is
5401 the same as @samp{-mlittle}.
5405 On System V.4 and embedded PowerPC systems compile code for the
5406 processor in big endian mode. The @samp{-mbig-endian} option is
5407 the same as @samp{-mbig}.
5410 On System V.4 and embedded PowerPC systems compile code using calling
5411 conventions that adheres to the March 1995 draft of the System V
5412 Application Binary Interface, PowerPC processor supplement. This is the
5413 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
5415 @item -mcall-sysv-eabi
5416 Specify both @samp{-mcall-sysv} and @samp{-meabi} options.
5418 @item -mcall-sysv-noeabi
5419 Specify both @samp{-mcall-sysv} and @samp{-mno-eabi} options.
5422 On System V.4 and embedded PowerPC systems compile code using calling
5423 conventions that are similar to those used on AIX. This is the
5424 default if you configured GCC using @samp{powerpc-*-eabiaix}.
5426 @item -mcall-solaris
5427 On System V.4 and embedded PowerPC systems compile code for the Solaris
5431 On System V.4 and embedded PowerPC systems compile code for the
5432 Linux-based GNU system.
5435 @itemx -mno-prototype
5436 On System V.4 and embedded PowerPC systems assume that all calls to
5437 variable argument functions are properly prototyped. Otherwise, the
5438 compiler must insert an instruction before every non prototyped call to
5439 set or clear bit 6 of the condition code register (@var{CR}) to
5440 indicate whether floating point values were passed in the floating point
5441 registers in case the function takes a variable arguments. With
5442 @samp{-mprototype}, only calls to prototyped variable argument functions
5443 will set or clear the bit.
5446 On embedded PowerPC systems, assume that the startup module is called
5447 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
5448 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
5452 On embedded PowerPC systems, assume that the startup module is called
5453 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
5457 On embedded PowerPC systems, assume that the startup module is called
5458 @file{crt0.o} and the standard C libraries are @file{libads.a} and
5462 On embedded PowerPC systems, assume that the startup module is called
5463 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
5467 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
5468 header to indicate that @samp{eabi} extended relocations are used.
5472 On System V.4 and embedded PowerPC systems do (do not) adhere to the
5473 Embedded Applications Binary Interface (eabi) which is a set of
5474 modifications to the System V.4 specifications. Selecting @code{-meabi}
5475 means that the stack is aligned to an 8 byte boundary, a function
5476 @code{__eabi} is called to from @code{main} to set up the eabi
5477 environment, and the @samp{-msdata} option can use both @code{r2} and
5478 @code{r13} to point to two separate small data areas. Selecting
5479 @code{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
5480 do not call an initialization function from @code{main}, and the
5481 @samp{-msdata} option will only use @code{r13} to point to a single
5482 small data area. The @samp{-meabi} option is on by default if you
5483 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
5486 On System V.4 and embedded PowerPC systems, put small initialized
5487 @code{const} global and static data in the @samp{.sdata2} section, which
5488 is pointed to by register @code{r2}. Put small initialized
5489 non-@code{const} global and static data in the @samp{.sdata} section,
5490 which is pointed to by register @code{r13}. Put small uninitialized
5491 global and static data in the @samp{.sbss} section, which is adjacent to
5492 the @samp{.sdata} section. The @samp{-msdata=eabi} option is
5493 incompatible with the @samp{-mrelocatable} option. The
5494 @samp{-msdata=eabi} option also sets the @samp{-memb} option.
5497 On System V.4 and embedded PowerPC systems, put small global and static
5498 data in the @samp{.sdata} section, which is pointed to by register
5499 @code{r13}. Put small uninitialized global and static data in the
5500 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
5501 The @samp{-msdata=sysv} option is incompatible with the
5502 @samp{-mrelocatable} option.
5504 @item -msdata=default
5506 On System V.4 and embedded PowerPC systems, if @samp{-meabi} is used,
5507 compile code the same as @samp{-msdata=eabi}, otherwise compile code the
5508 same as @samp{-msdata=sysv}.
5511 On System V.4 and embedded PowerPC systems, put small global and static
5512 data in the @samp{.sdata} section. Put small uninitialized global and
5513 static data in the @samp{.sbss} section. Do not use register @code{r13}
5514 to address small data however. This is the default behavior unless
5515 other @samp{-msdata} options are used.
5519 On embedded PowerPC systems, put all initialized global and static data
5520 in the @samp{.data} section, and all uninitialized data in the
5521 @samp{.bss} section.
5524 @cindex smaller data references (PowerPC)
5525 @cindex .sdata/.sdata2 references (PowerPC)
5526 On embedded PowerPC systems, put global and static items less than or
5527 equal to @var{num} bytes into the small data or bss sections instead of
5528 the normal data or bss section. By default, @var{num} is 8. The
5529 @samp{-G @var{num}} switch is also passed to the linker.
5530 All modules should be compiled with the same @samp{-G @var{num}} value.
5533 @itemx -mno-regnames
5534 On System V.4 and embedded PowerPC systems do (do not) emit register
5535 names in the assembly language output using symbolic forms.
5540 @subsection IBM RT Options
5542 @cindex IBM RT options
5544 These @samp{-m} options are defined for the IBM RT PC:
5548 Use an in-line code sequence for integer multiplies. This is the
5551 @item -mcall-lib-mul
5552 Call @code{lmul$$} for integer multiples.
5554 @item -mfull-fp-blocks
5555 Generate full-size floating point data blocks, including the minimum
5556 amount of scratch space recommended by IBM. This is the default.
5558 @item -mminimum-fp-blocks
5559 Do not include extra scratch space in floating point data blocks. This
5560 results in smaller code, but slower execution, since scratch space must
5561 be allocated dynamically.
5563 @cindex @file{varargs.h} and RT PC
5564 @cindex @file{stdarg.h} and RT PC
5565 @item -mfp-arg-in-fpregs
5566 Use a calling sequence incompatible with the IBM calling convention in
5567 which floating point arguments are passed in floating point registers.
5568 Note that @code{varargs.h} and @code{stdargs.h} will not work with
5569 floating point operands if this option is specified.
5571 @item -mfp-arg-in-gregs
5572 Use the normal calling convention for floating point arguments. This is
5575 @item -mhc-struct-return
5576 Return structures of more than one word in memory, rather than in a
5577 register. This provides compatibility with the MetaWare HighC (hc)
5578 compiler. Use the option @samp{-fpcc-struct-return} for compatibility
5579 with the Portable C Compiler (pcc).
5581 @item -mnohc-struct-return
5582 Return some structures of more than one word in registers, when
5583 convenient. This is the default. For compatibility with the
5584 IBM-supplied compilers, use the option @samp{-fpcc-struct-return} or the
5585 option @samp{-mhc-struct-return}.
5589 @subsection MIPS Options
5590 @cindex MIPS options
5592 These @samp{-m} options are defined for the MIPS family of computers:
5595 @item -mcpu=@var{cpu type}
5596 Assume the defaults for the machine type @var{cpu type} when scheduling
5597 instructions. The choices for @var{cpu type} are @samp{r2000}, @samp{r3000},
5598 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
5599 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
5600 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
5601 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
5602 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
5603 @var{cpu type} will schedule things appropriately for that particular
5604 chip, the compiler will not generate any code that does not meet level 1
5605 of the MIPS ISA (instruction set architecture) without a @samp{-mipsX}
5606 or @samp{-mabi} switch being used.
5609 Issue instructions from level 1 of the MIPS ISA. This is the default.
5610 @samp{r3000} is the default @var{cpu type} at this ISA level.
5613 Issue instructions from level 2 of the MIPS ISA (branch likely, square
5614 root instructions). @samp{r6000} is the default @var{cpu type} at this
5618 Issue instructions from level 3 of the MIPS ISA (64 bit instructions).
5619 @samp{r4000} is the default @var{cpu type} at this ISA level.
5622 Issue instructions from level 4 of the MIPS ISA (conditional move,
5623 prefetch, enhanced FPU instructions). @samp{r8000} is the default
5624 @var{cpu type} at this ISA level.
5627 Assume that 32 32-bit floating point registers are available. This is
5631 Assume that 32 64-bit floating point registers are available. This is
5632 the default when the @samp{-mips3} option is used.
5635 Assume that 32 32-bit general purpose registers are available. This is
5639 Assume that 32 64-bit general purpose registers are available. This is
5640 the default when the @samp{-mips3} option is used.
5643 Force int and long types to be 64 bits wide. See @samp{-mlong32} for an
5644 explanation of the default, and the width of pointers.
5647 Force long types to be 64 bits wide. See @samp{-mlong32} for an
5648 explanation of the default, and the width of pointers.
5651 Force long, int, and pointer types to be 32 bits wide.
5653 If none of @samp{-mlong32}, @samp{-mlong64}, or @samp{-mint64} are set,
5654 the size of ints, longs, and pointers depends on the ABI and ISA choosen.
5655 For @samp{-mabi=32}, and @samp{-mabi=n32}, ints and longs are 32 bits
5656 wide. For @samp{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
5657 For @samp{-mabi=eabi} and either @samp{-mips1} or @samp{-mips2}, ints
5658 and longs are 32 bits wide. For @samp{-mabi=eabi} and higher ISAs, ints
5659 are 32 bits, and longs are 64 bits wide. The width of pointer types is
5660 the smaller of the width of longs or the width of general purpose
5661 registers (which in turn depends on the ISA).
5668 Generate code for the indicated ABI. The default instruction level is
5669 @samp{-mips1} for @samp{32}, @samp{-mips3} for @samp{n32}, and
5670 @samp{-mips4} otherwise. Conversely, with @samp{-mips1} or
5671 @samp{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
5675 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
5676 add normal debug information. This is the default for all
5677 platforms except for the OSF/1 reference platform, using the OSF/rose
5678 object format. If the either of the @samp{-gstabs} or @samp{-gstabs+}
5679 switches are used, the @file{mips-tfile} program will encapsulate the
5680 stabs within MIPS ECOFF.
5683 Generate code for the GNU assembler. This is the default on the OSF/1
5684 reference platform, using the OSF/rose object format. Also, this is
5685 the default if the configure option @samp{--with-gnu-as} is used.
5687 @item -msplit-addresses
5688 @itemx -mno-split-addresses
5689 Generate code to load the high and low parts of address constants separately.
5690 This allows @code{gcc} to optimize away redundant loads of the high order
5691 bits of addresses. This optimization requires GNU as and GNU ld.
5692 This optimization is enabled by default for some embedded targets where
5693 GNU as and GNU ld are standard.
5697 The @samp{-mrnames} switch says to output code using the MIPS software
5698 names for the registers, instead of the hardware names (ie, @var{a0}
5699 instead of @var{$4}). The only known assembler that supports this option
5700 is the Algorithmics assembler.
5704 The @samp{-mgpopt} switch says to write all of the data declarations
5705 before the instructions in the text section, this allows the MIPS
5706 assembler to generate one word memory references instead of using two
5707 words for short global or static data items. This is on by default if
5708 optimization is selected.
5712 For each non-inline function processed, the @samp{-mstats} switch
5713 causes the compiler to emit one line to the standard error file to
5714 print statistics about the program (number of registers saved, stack
5719 The @samp{-mmemcpy} switch makes all block moves call the appropriate
5720 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
5721 generating inline code.
5724 @itemx -mno-mips-tfile
5725 The @samp{-mno-mips-tfile} switch causes the compiler not
5726 postprocess the object file with the @file{mips-tfile} program,
5727 after the MIPS assembler has generated it to add debug support. If
5728 @file{mips-tfile} is not run, then no local variables will be
5729 available to the debugger. In addition, @file{stage2} and
5730 @file{stage3} objects will have the temporary file names passed to the
5731 assembler embedded in the object file, which means the objects will
5732 not compare the same. The @samp{-mno-mips-tfile} switch should only
5733 be used when there are bugs in the @file{mips-tfile} program that
5734 prevents compilation.
5737 Generate output containing library calls for floating point.
5738 @strong{Warning:} the requisite libraries are not part of GCC.
5739 Normally the facilities of the machine's usual C compiler are used, but
5740 this can't be done directly in cross-compilation. You must make your
5741 own arrangements to provide suitable library functions for
5745 Generate output containing floating point instructions. This is the
5746 default if you use the unmodified sources.
5749 @itemx -mno-abicalls
5750 Emit (or do not emit) the pseudo operations @samp{.abicalls},
5751 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
5752 position independent code.
5755 @itemx -mno-long-calls
5756 Do all calls with the @samp{JALR} instruction, which requires
5757 loading up a function's address into a register before the call.
5758 You need to use this switch, if you call outside of the current
5759 512 megabyte segment to functions that are not through pointers.
5762 @itemx -mno-half-pic
5763 Put pointers to extern references into the data section and load them
5764 up, rather than put the references in the text section.
5766 @item -membedded-pic
5767 @itemx -mno-embedded-pic
5768 Generate PIC code suitable for some embedded systems. All calls are
5769 made using PC relative address, and all data is addressed using the $gp
5770 register. No more than 65536 bytes of global data may be used. This
5771 requires GNU as and GNU ld which do most of the work. This currently
5772 only works on targets which use ECOFF; it does not work with ELF.
5774 @item -membedded-data
5775 @itemx -mno-embedded-data
5776 Allocate variables to the read-only data section first if possible, then
5777 next in the small data section if possible, otherwise in data. This gives
5778 slightly slower code than the default, but reduces the amount of RAM required
5779 when executing, and thus may be preferred for some embedded systems.
5781 @item -muninit-const-in-rodata
5782 @itemx -mno-uninit-const-in-rodata
5783 When used together with -membedded-data, it will always store uninitialized
5784 const variables in the read-only data section.
5786 @item -msingle-float
5787 @itemx -mdouble-float
5788 The @samp{-msingle-float} switch tells gcc to assume that the floating
5789 point coprocessor only supports single precision operations, as on the
5790 @samp{r4650} chip. The @samp{-mdouble-float} switch permits gcc to use
5791 double precision operations. This is the default.
5795 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
5796 as on the @samp{r4650} chip.
5799 Turns on @samp{-msingle-float}, @samp{-mmad}, and, at least for now,
5804 Enable 16-bit instructions.
5807 Use the entry and exit pseudo ops. This option can only be used with
5811 Compile code for the processor in little endian mode.
5812 The requisite libraries are assumed to exist.
5815 Compile code for the processor in big endian mode.
5816 The requisite libraries are assumed to exist.
5819 @cindex smaller data references (MIPS)
5820 @cindex gp-relative references (MIPS)
5821 Put global and static items less than or equal to @var{num} bytes into
5822 the small data or bss sections instead of the normal data or bss
5823 section. This allows the assembler to emit one word memory reference
5824 instructions based on the global pointer (@var{gp} or @var{$28}),
5825 instead of the normal two words used. By default, @var{num} is 8 when
5826 the MIPS assembler is used, and 0 when the GNU assembler is used. The
5827 @samp{-G @var{num}} switch is also passed to the assembler and linker.
5828 All modules should be compiled with the same @samp{-G @var{num}}
5832 Tell the MIPS assembler to not run its preprocessor over user
5833 assembler files (with a @samp{.s} suffix) when assembling them.
5836 Pass an option to gas which will cause nops to be inserted if
5837 the read of the destination register of an mfhi or mflo instruction
5838 occurs in the following two instructions.
5841 Do not include the default crt0.
5845 These options are defined by the macro
5846 @code{TARGET_SWITCHES} in the machine description. The default for the
5847 options is also defined by that macro, which enables you to change the
5852 @subsection Intel 386 Options
5853 @cindex i386 Options
5854 @cindex Intel 386 Options
5856 These @samp{-m} options are defined for the i386 family of computers:
5859 @item -mcpu=@var{cpu type}
5860 Assume the defaults for the machine type @var{cpu type} when scheduling
5861 instructions. The choices for @var{cpu type} are:
5863 @multitable @columnfractions .20 .20 .20 .20
5864 @item @samp{i386} @tab @samp{i486} @tab @samp{i586} @tab @samp{i686}
5865 @item @samp{pentium} @tab @samp{pentiumpro} @tab @samp{k6}
5868 While picking a specific @var{cpu type} will schedule things appropriately
5869 for that particular chip, the compiler will not generate any code that
5870 does not run on the i386 without the @samp{-march=@var{cpu type}} option
5871 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
5872 is equivalent to @samp{pentiumpro}. @samp{k6} is the AMD chip as
5873 opposed to the Intel ones.
5875 @item -march=@var{cpu type}
5876 Generate instructions for the machine type @var{cpu type}. The choices
5877 for @var{cpu type} are the same as for @samp{-mcpu}. Moreover,
5878 specifying @samp{-march=@var{cpu type}} implies @samp{-mcpu=@var{cpu type}}.
5884 Synonyms for -mcpu=i386, -mcpu=i486, -mcpu=pentium, and -mcpu=pentiumpro
5885 respectively. These synonyms are deprecated.
5889 Control whether or not the compiler uses IEEE floating point
5890 comparisons. These handle correctly the case where the result of a
5891 comparison is unordered.
5894 Generate output containing library calls for floating point.
5895 @strong{Warning:} the requisite libraries are not part of GCC.
5896 Normally the facilities of the machine's usual C compiler are used, but
5897 this can't be done directly in cross-compilation. You must make your
5898 own arrangements to provide suitable library functions for
5901 On machines where a function returns floating point results in the 80387
5902 register stack, some floating point opcodes may be emitted even if
5903 @samp{-msoft-float} is used.
5905 @item -mno-fp-ret-in-387
5906 Do not use the FPU registers for return values of functions.
5908 The usual calling convention has functions return values of types
5909 @code{float} and @code{double} in an FPU register, even if there
5910 is no FPU. The idea is that the operating system should emulate
5913 The option @samp{-mno-fp-ret-in-387} causes such values to be returned
5914 in ordinary CPU registers instead.
5916 @item -mno-fancy-math-387
5917 Some 387 emulators do not support the @code{sin}, @code{cos} and
5918 @code{sqrt} instructions for the 387. Specify this option to avoid
5919 generating those instructions. This option is the default on FreeBSD.
5920 As of revision 2.6.1, these instructions are not generated unless you
5921 also use the @samp{-ffast-math} switch.
5923 @item -malign-double
5924 @itemx -mno-align-double
5925 Control whether GCC aligns @code{double}, @code{long double}, and
5926 @code{long long} variables on a two word boundary or a one word
5927 boundary. Aligning @code{double} variables on a two word boundary will
5928 produce code that runs somewhat faster on a @samp{Pentium} at the
5929 expense of more memory.
5931 @strong{Warning:} if you use the @samp{-malign-double} switch,
5932 structures containing the above types will be aligned differently than
5933 the published application binary interface specifications for the 386.
5936 @itemx -mno-svr3-shlib
5937 Control whether GCC places uninitialized locals into @code{bss} or
5938 @code{data}. @samp{-msvr3-shlib} places these locals into @code{bss}.
5939 These options are meaningful only on System V Release 3.
5941 @item -mno-wide-multiply
5942 @itemx -mwide-multiply
5943 Control whether GCC uses the @code{mul} and @code{imul} that produce
5944 64 bit results in @code{eax:edx} from 32 bit operands to do @code{long
5945 long} multiplies and 32-bit division by constants.
5948 Use a different function-calling convention, in which functions that
5949 take a fixed number of arguments return with the @code{ret} @var{num}
5950 instruction, which pops their arguments while returning. This saves one
5951 instruction in the caller since there is no need to pop the arguments
5954 You can specify that an individual function is called with this calling
5955 sequence with the function attribute @samp{stdcall}. You can also
5956 override the @samp{-mrtd} option by using the function attribute
5957 @samp{cdecl}. @xref{Function Attributes}.
5959 @strong{Warning:} this calling convention is incompatible with the one
5960 normally used on Unix, so you cannot use it if you need to call
5961 libraries compiled with the Unix compiler.
5963 Also, you must provide function prototypes for all functions that
5964 take variable numbers of arguments (including @code{printf});
5965 otherwise incorrect code will be generated for calls to those
5968 In addition, seriously incorrect code will result if you call a
5969 function with too many arguments. (Normally, extra arguments are
5970 harmlessly ignored.)
5972 @item -mreg-alloc=@var{regs}
5973 Control the default allocation order of integer registers. The
5974 string @var{regs} is a series of letters specifying a register. The
5975 supported letters are: @code{a} allocate EAX; @code{b} allocate EBX;
5976 @code{c} allocate ECX; @code{d} allocate EDX; @code{S} allocate ESI;
5977 @code{D} allocate EDI; @code{B} allocate EBP.
5979 @item -mregparm=@var{num}
5980 Control how many registers are used to pass integer arguments. By
5981 default, no registers are used to pass arguments, and at most 3
5982 registers can be used. You can control this behavior for a specific
5983 function by using the function attribute @samp{regparm}.
5984 @xref{Function Attributes}.
5986 @strong{Warning:} if you use this switch, and
5987 @var{num} is nonzero, then you must build all modules with the same
5988 value, including any libraries. This includes the system libraries and
5991 @item -malign-loops=@var{num}
5992 Align loops to a 2 raised to a @var{num} byte boundary. If
5993 @samp{-malign-loops} is not specified, the default is 2 unless
5994 gas 2.8 (or later) is being used in which case the default is
5995 to align the loop on a 16 byte boundary if it is less than 8
5998 @item -malign-jumps=@var{num}
5999 Align instructions that are only jumped to to a 2 raised to a @var{num}
6000 byte boundary. If @samp{-malign-jumps} is not specified, the default is
6001 2 if optimizing for a 386, and 4 if optimizing for a 486 unless
6002 gas 2.8 (or later) is being used in which case the default is
6003 to align the instruction on a 16 byte boundary if it is less
6006 @item -malign-functions=@var{num}
6007 Align the start of functions to a 2 raised to @var{num} byte boundary.
6008 If @samp{-malign-functions} is not specified, the default is 2 if optimizing
6009 for a 386, and 4 if optimizing for a 486.
6011 @item -mpreferred-stack-boundary=@var{num}
6012 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
6013 byte boundary. If @samp{-mpreferred-stack-boundary} is not specified,
6014 the default is 4 (16 bytes or 128 bits).
6016 The stack is required to be aligned on a 4 byte boundary. On Pentium
6017 and PentiumPro, @code{double} and @code{long double} values should be
6018 aligned to an 8 byte boundary (see @samp{-malign-double}) or suffer
6019 significant run time performance penalties. On Pentium III, the
6020 Streaming SIMD Extention (SSE) data type @code{__m128} suffers similar
6021 penalties if it is not 16 byte aligned.
6023 To ensure proper alignment of this values on the stack, the stack boundary
6024 must be as aligned as that required by any value stored on the stack.
6025 Further, every function must be generated such that it keeps the stack
6026 aligned. Thus calling a function compiled with a higher preferred
6027 stack boundary from a function compiled with a lower preferred stack
6028 boundary will most likely misalign the stack. It is recommended that
6029 libraries that use callbacks always use the default setting.
6031 This extra alignment does consume extra stack space. Code that is sensitive
6032 to stack space usage, such as embedded systems and operating system kernels,
6033 may want to reduce the preferred alignment to
6034 @samp{-mpreferred-stack-boundary=2}.
6038 Use PUSH operations to store outgoing parameters. This method is shorter
6039 and usually equally fast as method using SUB/MOV operations and is enabled
6040 by default. In some cases disabling it may improve performance because of
6041 improved scheduling and reduced dependencies.
6043 @item -maccumulate-outgoing-args
6044 @kindex -maccumulate-outgoing-args
6045 If enabled, the maximum amount of space required for outgoing arguments will be
6046 computed in the function prologue. This in faster on most modern CPUs
6047 because of reduced dependecies, improved scheduling and reduced stack usage
6048 when preferred stack boundary is not equal to 2. The drawback is a notable
6049 increase in code size. This switch implies -mno-push-args.
6053 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
6054 on thread-safe exception handling must compile and link all code with the
6055 @samp{-mthreads} option. When compiling, @samp{-mthreads} defines
6056 @samp{-D_MT}; when linking, it links in a special thread helper library
6057 @samp{-lmingwthrd} which cleans up per thread exception handling data.
6059 @item -mno-align-stringops
6060 @kindex -mno-align-stringops
6061 Do not align destination of inlined string operations. This switch reduces
6062 code size and improves performance in case the destination is already aligned,
6063 but gcc don't know about it.
6065 @item -minline-all-stringops
6066 @kindex -minline-all-stringops
6067 By default GCC inlines string operations only when destination is known to be
6068 aligned at least to 4 byte boundary. This enables more inlining, increase code
6069 size, but may improve performance of code that depends on fast memcpy, strlen
6070 and memset for short lengths.
6074 @subsection HPPA Options
6075 @cindex HPPA Options
6077 These @samp{-m} options are defined for the HPPA family of computers:
6080 @item -march=@var{architecture type}
6081 Generate code for the specified architecture. The choices for
6082 @var{architecture type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
6083 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
6084 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
6085 architecture option for your machine. Code compiled for lower numbered
6086 architectures will run on higher numbered architectures, but not the
6089 PA 2.0 support currently requires gas snapshot 19990413 or later. The
6090 next release of binutils (current is 2.9.1) will probably contain PA 2.0
6094 @itemx -mpa-risc-1-1
6095 @itemx -mpa-risc-2-0
6096 Synonyms for -march=1.0, -march=1.1, and -march=2.0 respectively.
6099 Generate code suitable for big switch tables. Use this option only if
6100 the assembler/linker complain about out of range branches within a switch
6103 @item -mjump-in-delay
6104 Fill delay slots of function calls with unconditional jump instructions
6105 by modifying the return pointer for the function call to be the target
6106 of the conditional jump.
6108 @item -mdisable-fpregs
6109 Prevent floating point registers from being used in any manner. This is
6110 necessary for compiling kernels which perform lazy context switching of
6111 floating point registers. If you use this option and attempt to perform
6112 floating point operations, the compiler will abort.
6114 @item -mdisable-indexing
6115 Prevent the compiler from using indexing address modes. This avoids some
6116 rather obscure problems when compiling MIG generated code under MACH.
6118 @item -mno-space-regs
6119 Generate code that assumes the target has no space registers. This allows
6120 GCC to generate faster indirect calls and use unscaled index address modes.
6122 Such code is suitable for level 0 PA systems and kernels.
6124 @item -mfast-indirect-calls
6125 Generate code that assumes calls never cross space boundaries. This
6126 allows GCC to emit code which performs faster indirect calls.
6128 This option will not work in the presense of shared libraries or nested
6131 @item -mlong-load-store
6132 Generate 3-instruction load and store sequences as sometimes required by
6133 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
6136 @item -mportable-runtime
6137 Use the portable calling conventions proposed by HP for ELF systems.
6140 Enable the use of assembler directives only GAS understands.
6142 @item -mschedule=@var{cpu type}
6143 Schedule code according to the constraints for the machine type
6144 @var{cpu type}. The choices for @var{cpu type} are @samp{700}
6145 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
6146 @file{/usr/lib/sched.models} on an HP-UX system to determine the
6147 proper scheduling option for your machine.
6150 Enable the optimization pass in the HPUX linker. Note this makes symbolic
6151 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
6152 in which they give bogus error messages when linking some programs.
6155 Generate output containing library calls for floating point.
6156 @strong{Warning:} the requisite libraries are not available for all HPPA
6157 targets. Normally the facilities of the machine's usual C compiler are
6158 used, but this cannot be done directly in cross-compilation. You must make
6159 your own arrangements to provide suitable library functions for
6160 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
6161 does provide software floating point support.
6163 @samp{-msoft-float} changes the calling convention in the output file;
6164 therefore, it is only useful if you compile @emph{all} of a program with
6165 this option. In particular, you need to compile @file{libgcc.a}, the
6166 library that comes with GCC, with @samp{-msoft-float} in order for
6170 @node Intel 960 Options
6171 @subsection Intel 960 Options
6173 These @samp{-m} options are defined for the Intel 960 implementations:
6176 @item -m@var{cpu type}
6177 Assume the defaults for the machine type @var{cpu type} for some of
6178 the other options, including instruction scheduling, floating point
6179 support, and addressing modes. The choices for @var{cpu type} are
6180 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
6181 @samp{sa}, and @samp{sb}.
6187 The @samp{-mnumerics} option indicates that the processor does support
6188 floating-point instructions. The @samp{-msoft-float} option indicates
6189 that floating-point support should not be assumed.
6191 @item -mleaf-procedures
6192 @itemx -mno-leaf-procedures
6193 Do (or do not) attempt to alter leaf procedures to be callable with the
6194 @code{bal} instruction as well as @code{call}. This will result in more
6195 efficient code for explicit calls when the @code{bal} instruction can be
6196 substituted by the assembler or linker, but less efficient code in other
6197 cases, such as calls via function pointers, or using a linker that doesn't
6198 support this optimization.
6201 @itemx -mno-tail-call
6202 Do (or do not) make additional attempts (beyond those of the
6203 machine-independent portions of the compiler) to optimize tail-recursive
6204 calls into branches. You may not want to do this because the detection of
6205 cases where this is not valid is not totally complete. The default is
6206 @samp{-mno-tail-call}.
6208 @item -mcomplex-addr
6209 @itemx -mno-complex-addr
6210 Assume (or do not assume) that the use of a complex addressing mode is a
6211 win on this implementation of the i960. Complex addressing modes may not
6212 be worthwhile on the K-series, but they definitely are on the C-series.
6213 The default is currently @samp{-mcomplex-addr} for all processors except
6217 @itemx -mno-code-align
6218 Align code to 8-byte boundaries for faster fetching (or don't bother).
6219 Currently turned on by default for C-series implementations only.
6222 @item -mclean-linkage
6223 @itemx -mno-clean-linkage
6224 These options are not fully implemented.
6228 @itemx -mic2.0-compat
6229 @itemx -mic3.0-compat
6230 Enable compatibility with iC960 v2.0 or v3.0.
6234 Enable compatibility with the iC960 assembler.
6236 @item -mstrict-align
6237 @itemx -mno-strict-align
6238 Do not permit (do permit) unaligned accesses.
6241 Enable structure-alignment compatibility with Intel's gcc release version
6242 1.3 (based on gcc 1.37). This option implies @samp{-mstrict-align}.
6244 @item -mlong-double-64
6245 Implement type @samp{long double} as 64-bit floating point numbers.
6246 Without the option @samp{long double} is implemented by 80-bit
6247 floating point numbers. The only reason we have it because there is
6248 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
6249 is only useful for people using soft-float targets. Otherwise, we
6250 should recommend against use of it.
6254 @node DEC Alpha Options
6255 @subsection DEC Alpha Options
6257 These @samp{-m} options are defined for the DEC Alpha implementations:
6260 @item -mno-soft-float
6262 Use (do not use) the hardware floating-point instructions for
6263 floating-point operations. When @code{-msoft-float} is specified,
6264 functions in @file{libgcc1.c} will be used to perform floating-point
6265 operations. Unless they are replaced by routines that emulate the
6266 floating-point operations, or compiled in such a way as to call such
6267 emulations routines, these routines will issue floating-point
6268 operations. If you are compiling for an Alpha without floating-point
6269 operations, you must ensure that the library is built so as not to call
6272 Note that Alpha implementations without floating-point operations are
6273 required to have floating-point registers.
6277 Generate code that uses (does not use) the floating-point register set.
6278 @code{-mno-fp-regs} implies @code{-msoft-float}. If the floating-point
6279 register set is not used, floating point operands are passed in integer
6280 registers as if they were integers and floating-point results are passed
6281 in $0 instead of $f0. This is a non-standard calling sequence, so any
6282 function with a floating-point argument or return value called by code
6283 compiled with @code{-mno-fp-regs} must also be compiled with that
6286 A typical use of this option is building a kernel that does not use,
6287 and hence need not save and restore, any floating-point registers.
6290 The Alpha architecture implements floating-point hardware optimized for
6291 maximum performance. It is mostly compliant with the IEEE floating
6292 point standard. However, for full compliance, software assistance is
6293 required. This option generates code fully IEEE compliant code
6294 @emph{except} that the @var{inexact flag} is not maintained (see below).
6295 If this option is turned on, the CPP macro @code{_IEEE_FP} is defined
6296 during compilation. The option is a shorthand for: @samp{-D_IEEE_FP
6297 -mfp-trap-mode=su -mtrap-precision=i -mieee-conformant}. The resulting
6298 code is less efficient but is able to correctly support denormalized
6299 numbers and exceptional IEEE values such as not-a-number and plus/minus
6300 infinity. Other Alpha compilers call this option
6301 @code{-ieee_with_no_inexact}.
6303 @item -mieee-with-inexact
6304 @c overfull hbox here --bob 22 jul96
6305 @c original text between ignore ... end ignore
6307 This is like @samp{-mieee} except the generated code also maintains the
6308 IEEE @var{inexact flag}. Turning on this option causes the generated
6309 code to implement fully-compliant IEEE math. The option is a shorthand
6310 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus @samp{-mieee-conformant},
6311 @samp{-mfp-trap-mode=sui}, and @samp{-mtrap-precision=i}. On some Alpha
6312 implementations the resulting code may execute significantly slower than
6313 the code generated by default. Since there is very little code that
6314 depends on the @var{inexact flag}, you should normally not specify this
6315 option. Other Alpha compilers call this option
6316 @samp{-ieee_with_inexact}.
6318 @c changed paragraph
6319 This is like @samp{-mieee} except the generated code also maintains the
6320 IEEE @var{inexact flag}. Turning on this option causes the generated
6321 code to implement fully-compliant IEEE math. The option is a shorthand
6322 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus the three following:
6323 @samp{-mieee-conformant},
6324 @samp{-mfp-trap-mode=sui},
6325 and @samp{-mtrap-precision=i}.
6326 On some Alpha implementations the resulting code may execute
6327 significantly slower than the code generated by default. Since there
6328 is very little code that depends on the @var{inexact flag}, you should
6329 normally not specify this option. Other Alpha compilers call this
6330 option @samp{-ieee_with_inexact}.
6331 @c end changes to prevent overfull hboxes
6333 @item -mfp-trap-mode=@var{trap mode}
6334 This option controls what floating-point related traps are enabled.
6335 Other Alpha compilers call this option @samp{-fptm }@var{trap mode}.
6336 The trap mode can be set to one of four values:
6340 This is the default (normal) setting. The only traps that are enabled
6341 are the ones that cannot be disabled in software (e.g., division by zero
6345 In addition to the traps enabled by @samp{n}, underflow traps are enabled
6349 Like @samp{su}, but the instructions are marked to be safe for software
6350 completion (see Alpha architecture manual for details).
6353 Like @samp{su}, but inexact traps are enabled as well.
6356 @item -mfp-rounding-mode=@var{rounding mode}
6357 Selects the IEEE rounding mode. Other Alpha compilers call this option
6358 @samp{-fprm }@var{rounding mode}. The @var{rounding mode} can be one
6363 Normal IEEE rounding mode. Floating point numbers are rounded towards
6364 the nearest machine number or towards the even machine number in case
6368 Round towards minus infinity.
6371 Chopped rounding mode. Floating point numbers are rounded towards zero.
6374 Dynamic rounding mode. A field in the floating point control register
6375 (@var{fpcr}, see Alpha architecture reference manual) controls the
6376 rounding mode in effect. The C library initializes this register for
6377 rounding towards plus infinity. Thus, unless your program modifies the
6378 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
6381 @item -mtrap-precision=@var{trap precision}
6382 In the Alpha architecture, floating point traps are imprecise. This
6383 means without software assistance it is impossible to recover from a
6384 floating trap and program execution normally needs to be terminated.
6385 GCC can generate code that can assist operating system trap handlers
6386 in determining the exact location that caused a floating point trap.
6387 Depending on the requirements of an application, different levels of
6388 precisions can be selected:
6392 Program precision. This option is the default and means a trap handler
6393 can only identify which program caused a floating point exception.
6396 Function precision. The trap handler can determine the function that
6397 caused a floating point exception.
6400 Instruction precision. The trap handler can determine the exact
6401 instruction that caused a floating point exception.
6404 Other Alpha compilers provide the equivalent options called
6405 @samp{-scope_safe} and @samp{-resumption_safe}.
6407 @item -mieee-conformant
6408 This option marks the generated code as IEEE conformant. You must not
6409 use this option unless you also specify @samp{-mtrap-precision=i} and either
6410 @samp{-mfp-trap-mode=su} or @samp{-mfp-trap-mode=sui}. Its only effect
6411 is to emit the line @samp{.eflag 48} in the function prologue of the
6412 generated assembly file. Under DEC Unix, this has the effect that
6413 IEEE-conformant math library routines will be linked in.
6415 @item -mbuild-constants
6416 Normally GCC examines a 32- or 64-bit integer constant to
6417 see if it can construct it from smaller constants in two or three
6418 instructions. If it cannot, it will output the constant as a literal and
6419 generate code to load it from the data segment at runtime.
6421 Use this option to require GCC to construct @emph{all} integer constants
6422 using code, even if it takes more instructions (the maximum is six).
6424 You would typically use this option to build a shared library dynamic
6425 loader. Itself a shared library, it must relocate itself in memory
6426 before it can find the variables and constants in its own data segment.
6430 Select whether to generate code to be assembled by the vendor-supplied
6431 assembler (@samp{-malpha-as}) or by the GNU assembler @samp{-mgas}.
6439 Indicate whether GCC should generate code to use the optional BWX,
6440 CIX, and MAX instruction sets. The default is to use the instruction sets
6441 supported by the CPU type specified via @samp{-mcpu=} option or that
6442 of the CPU on which GCC was built if none was specified.
6444 @item -mcpu=@var{cpu_type}
6445 Set the instruction set, register set, and instruction scheduling
6446 parameters for machine type @var{cpu_type}. You can specify either the
6447 @samp{EV} style name or the corresponding chip number. GCC
6448 supports scheduling parameters for the EV4 and EV5 family of processors
6449 and will choose the default values for the instruction set from
6450 the processor you specify. If you do not specify a processor type,
6451 GCC will default to the processor on which the compiler was built.
6453 Supported values for @var{cpu_type} are
6458 Schedules as an EV4 and has no instruction set extensions.
6462 Schedules as an EV5 and has no instruction set extensions.
6466 Schedules as an EV5 and supports the BWX extension.
6471 Schedules as an EV5 and supports the BWX and MAX extensions.
6475 Schedules as an EV5 (until Digital releases the scheduling parameters
6476 for the EV6) and supports the BWX, CIX, and MAX extensions.
6479 @item -mmemory-latency=@var{time}
6480 Sets the latency the scheduler should assume for typical memory
6481 references as seen by the application. This number is highly
6482 dependant on the memory access patterns used by the application
6483 and the size of the external cache on the machine.
6485 Valid options for @var{time} are
6489 A decimal number representing clock cycles.
6495 The compiler contains estimates of the number of clock cycles for
6496 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
6497 (also called Dcache, Scache, and Bcache), as well as to main memory.
6498 Note that L3 is only valid for EV5.
6503 @node Clipper Options
6504 @subsection Clipper Options
6506 These @samp{-m} options are defined for the Clipper implementations:
6510 Produce code for a C300 Clipper processor. This is the default.
6513 Produce code for a C400 Clipper processor i.e. use floating point
6517 @node H8/300 Options
6518 @subsection H8/300 Options
6520 These @samp{-m} options are defined for the H8/300 implementations:
6524 Shorten some address references at link time, when possible; uses the
6525 linker option @samp{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
6526 ld.info, Using ld}, for a fuller description.
6529 Generate code for the H8/300H.
6532 Generate code for the H8/S.
6535 Make @code{int} data 32 bits by default.
6538 On the h8/300h, use the same alignment rules as for the h8/300.
6539 The default for the h8/300h is to align longs and floats on 4 byte boundaries.
6540 @samp{-malign-300} causes them to be aligned on 2 byte boundaries.
6541 This option has no effect on the h8/300.
6545 @subsection SH Options
6547 These @samp{-m} options are defined for the SH implementations:
6551 Generate code for the SH1.
6554 Generate code for the SH2.
6557 Generate code for the SH3.
6560 Generate code for the SH3e.
6563 Compile code for the processor in big endian mode.
6566 Compile code for the processor in little endian mode.
6569 Align doubles at 64 bit boundaries. Note that this changes the calling
6570 conventions, and thus some functions from the standard C library will
6571 not work unless you recompile it first with -mdalign.
6574 Shorten some address references at link time, when possible; uses the
6575 linker option @samp{-relax}.
6578 @node System V Options
6579 @subsection Options for System V
6581 These additional options are available on System V Release 4 for
6582 compatibility with other compilers on those systems:
6586 Create a shared object.
6587 It is recommended that @samp{-symbolic} or @samp{-shared} be used instead.
6590 Identify the versions of each tool used by the compiler, in a
6591 @code{.ident} assembler directive in the output.
6594 Refrain from adding @code{.ident} directives to the output file (this is
6597 @item -YP,@var{dirs}
6598 Search the directories @var{dirs}, and no others, for libraries
6599 specified with @samp{-l}.
6602 Look in the directory @var{dir} to find the M4 preprocessor.
6603 The assembler uses this option.
6604 @c This is supposed to go with a -Yd for predefined M4 macro files, but
6605 @c the generic assembler that comes with Solaris takes just -Ym.
6608 @node TMS320C3x/C4x Options
6609 @subsection TMS320C3x/C4x Options
6610 @cindex TMS320C3x/C4x Options
6612 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
6616 @item -mcpu=@var{cpu_type}
6617 Set the instruction set, register set, and instruction scheduling
6618 parameters for machine type @var{cpu_type}. Supported values for
6619 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
6620 @samp{c44}. The default is @samp{c40} to generate code for the
6625 @itemx -msmall-memory
6627 Generates code for the big or small memory model. The small memory
6628 model assumed that all data fits into one 64K word page. At run-time
6629 the data page (DP) register must be set to point to the 64K page
6630 containing the .bss and .data program sections. The big memory model is
6631 the default and requires reloading of the DP register for every direct
6636 Allow (disallow) allocation of general integer operands into the block
6641 Enable (disable) generation of code using decrement and branch,
6642 DBcond(D), instructions. This is enabled by default for the C4x. To be
6643 on the safe side, this is disabled for the C3x, since the maximum
6644 iteration count on the C3x is 2^23 + 1 (but who iterates loops more than
6645 2^23 times on the C3x?). Note that GCC will try to reverse a loop so
6646 that it can utilise the decrement and branch instruction, but will give
6647 up if there is more than one memory reference in the loop. Thus a loop
6648 where the loop counter is decremented can generate slightly more
6649 efficient code, in cases where the RPTB instruction cannot be utilised.
6651 @item -mdp-isr-reload
6653 Force the DP register to be saved on entry to an interrupt service
6654 routine (ISR), reloaded to point to the data section, and restored on
6655 exit from the ISR. This should not be required unless someone has
6656 violated the small memory model by modifying the DP register, say within
6661 For the C3x use the 24-bit MPYI instruction for integer multiplies
6662 instead of a library call to guarantee 32-bit results. Note that if one
6663 of the operands is a constant, then the multiplication will be performed
6664 using shifts and adds. If the -mmpyi option is not specified for the C3x,
6665 then squaring operations are performed inline instead of a library call.
6668 @itemx -mno-fast-fix
6669 The C3x/C4x FIX instruction to convert a floating point value to an
6670 integer value chooses the nearest integer less than or equal to the
6671 floating point value rather than to the nearest integer. Thus if the
6672 floating point number is negative, the result will be incorrectly
6673 truncated an additional code is necessary to detect and correct this
6674 case. This option can be used to disable generation of the additional
6675 code required to correct the result.
6679 Enable (disable) generation of repeat block sequences using the RPTB
6680 instruction for zero overhead looping. The RPTB construct is only used
6681 for innermost loops that do not call functions or jump across the loop
6682 boundaries. There is no advantage having nested RPTB loops due to the
6683 overhead required to save and restore the RC, RS, and RE registers.
6684 This is enabled by default with -O2.
6686 @item -mrpts=@var{count}
6688 Enable (disable) the use of the single instruction repeat instruction
6689 RPTS. If a repeat block contains a single instruction, and the loop
6690 count can be guaranteed to be less than the value @var{count}, GCC will
6691 emit a RPTS instruction instead of a RPTB. If no value is specified,
6692 then a RPTS will be emitted even if the loop count cannot be determined
6693 at compile time. Note that the repeated instruction following RPTS does
6694 not have to be reloaded from memory each iteration, thus freeing up the
6695 CPU buses for oeprands. However, since interrupts are blocked by this
6696 instruction, it is disabled by default.
6698 @item -mloop-unsigned
6699 @itemx -mno-loop-unsigned
6700 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
6701 is 2^31 + 1 since these instructions test if the iteration count is
6702 negative to terminate the loop. If the iteration count is unsigned
6703 there is a possibility than the 2^31 + 1 maximum iteration count may be
6704 exceeded. This switch allows an unsigned iteration count.
6707 Try to emit an assembler syntax that the TI assembler (asm30) is happy
6708 with. This also enforces compatibility with the API employed by the TI
6709 C3x C compiler. For example, long doubles are passed as structures
6710 rather than in floating point registers.
6714 Generate code that uses registers (stack) for passing arguments to functions.
6715 By default, arguments are passed in registers where possible rather
6716 than by pushing arguments on to the stack.
6718 @item -mparallel-insns
6719 @itemx -mno-parallel-insns
6720 Allow the generation of parallel instructions. This is enabled by
6723 @item -mparallel-mpy
6724 @itemx -mno-parallel-mpy
6725 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
6726 provided -mparallel-insns is also specified. These instructions have
6727 tight register constraints which can pessimize the code generation
6733 @subsection V850 Options
6734 @cindex V850 Options
6736 These @samp{-m} options are defined for V850 implementations:
6740 @itemx -mno-long-calls
6741 Treat all calls as being far away (near). If calls are assumed to be
6742 far away, the compiler will always load the functions address up into a
6743 register, and call indirect through the pointer.
6747 Do not optimize (do optimize) basic blocks that use the same index
6748 pointer 4 or more times to copy pointer into the @code{ep} register, and
6749 use the shorter @code{sld} and @code{sst} instructions. The @samp{-mep}
6750 option is on by default if you optimize.
6752 @item -mno-prolog-function
6753 @itemx -mprolog-function
6754 Do not use (do use) external functions to save and restore registers at
6755 the prolog and epilog of a function. The external functions are slower,
6756 but use less code space if more than one function saves the same number
6757 of registers. The @samp{-mprolog-function} option is on by default if
6761 Try to make the code as small as possible. At present, this just turns
6762 on the @samp{-mep} and @samp{-mprolog-function} options.
6765 Put static or global variables whose size is @var{n} bytes or less into
6766 the tiny data area that register @code{ep} points to. The tiny data
6767 area can hold up to 256 bytes in total (128 bytes for byte references).
6770 Put static or global variables whose size is @var{n} bytes or less into
6771 the small data area that register @code{gp} points to. The small data
6772 area can hold up to 64 kilobytes.
6775 Put static or global variables whose size is @var{n} bytes or less into
6776 the first 32 kilobytes of memory.
6779 Specify that the target processor is the V850.
6782 Generate code suitable for big switch tables. Use this option only if
6783 the assembler/linker complain about out of range branches within a switch
6788 @subsection ARC Options
6791 These options are defined for ARC implementations:
6795 Compile code for little endian mode. This is the default.
6798 Compile code for big endian mode.
6801 Prepend the name of the cpu to all public symbol names.
6802 In multiple-processor systems, there are many ARC variants with different
6803 instruction and register set characteristics. This flag prevents code
6804 compiled for one cpu to be linked with code compiled for another.
6805 No facility exists for handling variants that are "almost identical".
6806 This is an all or nothing option.
6808 @item -mcpu=@var{cpu}
6809 Compile code for ARC variant @var{cpu}.
6810 Which variants are supported depend on the configuration.
6811 All variants support @samp{-mcpu=base}, this is the default.
6813 @item -mtext=@var{text section}
6814 @itemx -mdata=@var{data section}
6815 @itemx -mrodata=@var{readonly data section}
6816 Put functions, data, and readonly data in @var{text section},
6817 @var{data section}, and @var{readonly data section} respectively
6818 by default. This can be overridden with the @code{section} attribute.
6819 @xref{Variable Attributes}.
6824 @subsection NS32K Options
6825 @cindex NS32K options
6827 These are the @samp{-m} options defined for the 32000 series. The default
6828 values for these options depends on which style of 32000 was selected when
6829 the compiler was configured; the defaults for the most common choices are
6835 Generate output for a 32032. This is the default
6836 when the compiler is configured for 32032 and 32016 based systems.
6840 Generate output for a 32332. This is the default
6841 when the compiler is configured for 32332-based systems.
6845 Generate output for a 32532. This is the default
6846 when the compiler is configured for 32532-based systems.
6849 Generate output containing 32081 instructions for floating point.
6850 This is the default for all systems.
6853 Generate output containing 32381 instructions for floating point. This
6854 also implies @samp{-m32081}. The 32381 is only compatible with the 32332
6855 and 32532 cpus. This is the default for the pc532-netbsd configuration.
6858 Try and generate multiply-add floating point instructions @code{polyF}
6859 and @code{dotF}. This option is only available if the @samp{-m32381}
6860 option is in effect. Using these instructions requires changes to to
6861 register allocation which generally has a negative impact on
6862 performance. This option should only be enabled when compiling code
6863 particularly likely to make heavy use of multiply-add instructions.
6866 Do not try and generate multiply-add floating point instructions
6867 @code{polyF} and @code{dotF}. This is the default on all platforms.
6870 Generate output containing library calls for floating point.
6871 @strong{Warning:} the requisite libraries may not be available.
6874 Do not use the bit-field instructions. On some machines it is faster to
6875 use shifting and masking operations. This is the default for the pc532.
6878 Do use the bit-field instructions. This is the default for all platforms
6882 Use a different function-calling convention, in which functions
6883 that take a fixed number of arguments return pop their
6884 arguments on return with the @code{ret} instruction.
6886 This calling convention is incompatible with the one normally
6887 used on Unix, so you cannot use it if you need to call libraries
6888 compiled with the Unix compiler.
6890 Also, you must provide function prototypes for all functions that
6891 take variable numbers of arguments (including @code{printf});
6892 otherwise incorrect code will be generated for calls to those
6895 In addition, seriously incorrect code will result if you call a
6896 function with too many arguments. (Normally, extra arguments are
6897 harmlessly ignored.)
6899 This option takes its name from the 680x0 @code{rtd} instruction.
6903 Use a different function-calling convention where the first two arguments
6904 are passed in registers.
6906 This calling convention is incompatible with the one normally
6907 used on Unix, so you cannot use it if you need to call libraries
6908 compiled with the Unix compiler.
6911 Do not pass any arguments in registers. This is the default for all
6915 It is OK to use the sb as an index register which is always loaded with
6916 zero. This is the default for the pc532-netbsd target.
6919 The sb register is not available for use or has not been initialized to
6920 zero by the run time system. This is the default for all targets except
6921 the pc532-netbsd. It is also implied whenever @samp{-mhimem} or
6922 @samp{-fpic} is set.
6925 Many ns32000 series addressing modes use displacements of up to 512MB.
6926 If an address is above 512MB then displacements from zero can not be used.
6927 This option causes code to be generated which can be loaded above 512MB.
6928 This may be useful for operating systems or ROM code.
6931 Assume code will be loaded in the first 512MB of virtual address space.
6932 This is the default for all platforms.
6938 @subsection AVR Options
6941 These options are defined for AVR implementations:
6944 @item -mmcu=@var{mcu}
6945 Specify ATMEL AVR mcu (at90s23xx,attiny22,at90s44xx,at90s85xx,atmega603,
6949 Output instruction size's to the asm file
6951 @item -minit-stack=@var{N}
6952 Specify the initial stack address
6954 @item -mno-interrupts
6955 Generated code is not compatible with hardware interrupts.
6956 Code size will be smaller.
6958 @item -mcall-prologues
6959 Functions prologues/epilogues expanded as call to appropriate
6960 subroutines. Code size will be smaller.
6964 @subsection MCore Options
6965 @cindex MCore options
6967 These are the @samp{-m} options defined for the Motorola M*Core
6975 Inline constants into the code stream if it can be done in two
6976 instructions or less.
6981 Use the divide instruction. (Enabled by default).
6983 @item -mrelax-immediate
6984 @itemx -mrelax-immediate
6985 @itemx -mno-relax-immediate
6986 Allow arbitary sized immediated in bit operations.
6988 @item -mwide-bitfields
6989 @itemx -mwide-bitfields
6990 @itemx -mno-wide-bitfields
6991 Always treat bitfields as int-sized.
6993 @item -m4byte-functions
6994 @itemx -m4byte-functions
6995 @itemx -mno-4byte-functions
6996 Force all functions to be aligfned to a four byte boundary.
6998 @item -mcallgraph-data
6999 @itemx -mcallgraph-data
7000 @itemx -mno-callgraph-data
7001 Emit callgraph information.
7005 @itemx -mno-slow-bytes
7006 Prefer word access when reading byte quantities.
7008 @item -mlittle-endian
7009 @itemx -mlittle-endian
7011 Genreate code for a little endian target.
7016 Generate code for the 210 processor.
7021 @node Code Gen Options
7022 @section Options for Code Generation Conventions
7023 @cindex code generation conventions
7024 @cindex options, code generation
7025 @cindex run-time options
7027 These machine-independent options control the interface conventions
7028 used in code generation.
7030 Most of them have both positive and negative forms; the negative form
7031 of @samp{-ffoo} would be @samp{-fno-foo}. In the table below, only
7032 one of the forms is listed---the one which is not the default. You
7033 can figure out the other form by either removing @samp{no-} or adding
7038 Enable exception handling. Generates extra code needed to propagate
7039 exceptions. For some targets, this implies GNU CC will generate frame
7040 unwind information for all functions, which can produce significant data
7041 size overhead, although it does not affect execution. If you do not
7042 specify this option, GNU CC will enable it by default for languages like
7043 C++ which normally require exception handling, and disable itfor
7044 languages like C that do not normally require it. However, you may need
7045 to enable this option when compiling C code that needs to interoperate
7046 properly with exception handlers written in C++. You may also wish to
7047 disable this option if you are compiling older C++ programs that don't
7048 use exception handling.
7050 @item -funwind-tables
7051 Similar to @code{-fexceptions}, except that it will just generate any needed
7052 static data, but will not affect the generated code in any other way.
7053 You will normally not enable this option; instead, a language processor
7054 that needs this handling would enable it on your behalf.
7056 @item -fpcc-struct-return
7057 Return ``short'' @code{struct} and @code{union} values in memory like
7058 longer ones, rather than in registers. This convention is less
7059 efficient, but it has the advantage of allowing intercallability between
7060 GCC-compiled files and files compiled with other compilers.
7062 The precise convention for returning structures in memory depends
7063 on the target configuration macros.
7065 Short structures and unions are those whose size and alignment match
7066 that of some integer type.
7068 @item -freg-struct-return
7069 Use the convention that @code{struct} and @code{union} values are
7070 returned in registers when possible. This is more efficient for small
7071 structures than @samp{-fpcc-struct-return}.
7073 If you specify neither @samp{-fpcc-struct-return} nor its contrary
7074 @samp{-freg-struct-return}, GCC defaults to whichever convention is
7075 standard for the target. If there is no standard convention, GCC
7076 defaults to @samp{-fpcc-struct-return}, except on targets where GCC
7077 is the principal compiler. In those cases, we can choose the standard,
7078 and we chose the more efficient register return alternative.
7081 Allocate to an @code{enum} type only as many bytes as it needs for the
7082 declared range of possible values. Specifically, the @code{enum} type
7083 will be equivalent to the smallest integer type which has enough room.
7085 @item -fshort-double
7086 Use the same size for @code{double} as for @code{float}.
7089 Requests that the data and non-@code{const} variables of this
7090 compilation be shared data rather than private data. The distinction
7091 makes sense only on certain operating systems, where shared data is
7092 shared between processes running the same program, while private data
7093 exists in one copy per process.
7096 Allocate even uninitialized global variables in the data section of the
7097 object file, rather than generating them as common blocks. This has the
7098 effect that if the same variable is declared (without @code{extern}) in
7099 two different compilations, you will get an error when you link them.
7100 The only reason this might be useful is if you wish to verify that the
7101 program will work on other systems which always work this way.
7104 Ignore the @samp{#ident} directive.
7106 @item -fno-gnu-linker
7107 Do not output global initializations (such as C++ constructors and
7108 destructors) in the form used by the GNU linker (on systems where the GNU
7109 linker is the standard method of handling them). Use this option when
7110 you want to use a non-GNU linker, which also requires using the
7111 @code{collect2} program to make sure the system linker includes
7112 constructors and destructors. (@code{collect2} is included in the GCC
7113 distribution.) For systems which @emph{must} use @code{collect2}, the
7114 compiler driver @code{gcc} is configured to do this automatically.
7116 @item -finhibit-size-directive
7117 Don't output a @code{.size} assembler directive, or anything else that
7118 would cause trouble if the function is split in the middle, and the
7119 two halves are placed at locations far apart in memory. This option is
7120 used when compiling @file{crtstuff.c}; you should not need to use it
7124 Put extra commentary information in the generated assembly code to
7125 make it more readable. This option is generally only of use to those
7126 who actually need to read the generated assembly code (perhaps while
7127 debugging the compiler itself).
7129 @samp{-fno-verbose-asm}, the default, causes the
7130 extra information to be omitted and is useful when comparing two assembler
7134 Consider all memory references through pointers to be volatile.
7136 @item -fvolatile-global
7137 Consider all memory references to extern and global data items to
7138 be volatile. GCC does not consider static data items to be volatile
7139 because of this switch.
7141 @item -fvolatile-static
7142 Consider all memory references to static data to be volatile.
7145 @cindex global offset table
7147 Generate position-independent code (PIC) suitable for use in a shared
7148 library, if supported for the target machine. Such code accesses all
7149 constant addresses through a global offset table (GOT). The dynamic
7150 loader resolves the GOT entries when the program starts (the dynamic
7151 loader is not part of GCC; it is part of the operating system). If
7152 the GOT size for the linked executable exceeds a machine-specific
7153 maximum size, you get an error message from the linker indicating that
7154 @samp{-fpic} does not work; in that case, recompile with @samp{-fPIC}
7155 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
7156 on the m68k and RS/6000. The 386 has no such limit.)
7158 Position-independent code requires special support, and therefore works
7159 only on certain machines. For the 386, GCC supports PIC for System V
7160 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
7161 position-independent.
7164 If supported for the target machine, emit position-independent code,
7165 suitable for dynamic linking and avoiding any limit on the size of the
7166 global offset table. This option makes a difference on the m68k, m88k,
7169 Position-independent code requires special support, and therefore works
7170 only on certain machines.
7172 @item -ffixed-@var{reg}
7173 Treat the register named @var{reg} as a fixed register; generated code
7174 should never refer to it (except perhaps as a stack pointer, frame
7175 pointer or in some other fixed role).
7177 @var{reg} must be the name of a register. The register names accepted
7178 are machine-specific and are defined in the @code{REGISTER_NAMES}
7179 macro in the machine description macro file.
7181 This flag does not have a negative form, because it specifies a
7184 @item -fcall-used-@var{reg}
7185 Treat the register named @var{reg} as an allocable register that is
7186 clobbered by function calls. It may be allocated for temporaries or
7187 variables that do not live across a call. Functions compiled this way
7188 will not save and restore the register @var{reg}.
7190 It is an error to used this flag with the frame pointer or stack pointer.
7191 Use of this flag for other registers that have fixed pervasive roles in
7192 the machine's execution model will produce disastrous results.
7194 This flag does not have a negative form, because it specifies a
7197 @item -fcall-saved-@var{reg}
7198 Treat the register named @var{reg} as an allocable register saved by
7199 functions. It may be allocated even for temporaries or variables that
7200 live across a call. Functions compiled this way will save and restore
7201 the register @var{reg} if they use it.
7203 It is an error to used this flag with the frame pointer or stack pointer.
7204 Use of this flag for other registers that have fixed pervasive roles in
7205 the machine's execution model will produce disastrous results.
7207 A different sort of disaster will result from the use of this flag for
7208 a register in which function values may be returned.
7210 This flag does not have a negative form, because it specifies a
7214 Pack all structure members together without holes. Usually you would
7215 not want to use this option, since it makes the code suboptimal, and
7216 the offsets of structure members won't agree with system libraries.
7218 @item -fcheck-memory-usage
7219 Generate extra code to check each memory access. GCC will generate
7220 code that is suitable for a detector of bad memory accesses such as
7223 Normally, you should compile all, or none, of your code with this option.
7225 If you do mix code compiled with and without this option,
7226 you must ensure that all code that has side effects
7227 and that is called by code compiled with this option
7228 is, itself, compiled with this option.
7229 If you do not, you might get erroneous messages from the detector.
7231 If you use functions from a library that have side-effects (such as
7232 @code{read}), you might not be able to recompile the library and
7233 specify this option. In that case, you can enable the
7234 @samp{-fprefix-function-name} option, which requests GCC to encapsulate
7235 your code and make other functions look as if they were compiled with
7236 @samp{-fcheck-memory-usage}. This is done by calling ``stubs'',
7237 which are provided by the detector. If you cannot find or build
7238 stubs for every function you call, you might have to specify
7239 @samp{-fcheck-memory-usage} without @samp{-fprefix-function-name}.
7241 If you specify this option, you can not use the @code{asm} or
7242 @code{__asm__} keywords in functions with memory checking enabled. GNU
7243 CC cannot understand what the @code{asm} statement may do, and therefore
7244 cannot generate the appropriate code, so it will reject it. However, if
7245 you specify the function attribute @code{no_check_memory_usage} (see
7246 @pxref{Function Attributes}, GNU CC will disable memory checking within a
7247 function; you may use @code{asm} statements inside such functions. You
7248 may have an inline expansion of a non-checked function within a checked
7249 function; in that case GNU CC will not generate checks for the inlined
7250 function's memory accesses.
7252 If you move your @code{asm} statements to non-checked inline functions
7253 and they do access memory, you can add calls to the support code in your
7254 inline function, to indicate any reads, writes, or copies being done.
7255 These calls would be similar to those done in the stubs described above.
7257 @item -fprefix-function-name
7258 Request GCC to add a prefix to the symbols generated for function names.
7259 GCC adds a prefix to the names of functions defined as well as
7260 functions called. Code compiled with this option and code compiled
7261 without the option can't be linked together, unless stubs are used.
7263 If you compile the following code with @samp{-fprefix-function-name}
7265 extern void bar (int);
7274 GCC will compile the code as if it was written:
7276 extern void prefix_bar (int);
7280 return prefix_bar (a + 5);
7283 This option is designed to be used with @samp{-fcheck-memory-usage}.
7285 @item -finstrument-functions
7286 Generate instrumentation calls for entry and exit to functions. Just
7287 after function entry and just before function exit, the following
7288 profiling functions will be called with the address of the current
7289 function and its call site. (On some platforms,
7290 @code{__builtin_return_address} does not work beyond the current
7291 function, so the call site information may not be available to the
7292 profiling functions otherwise.)
7295 void __cyg_profile_func_enter (void *this_fn, void *call_site);
7296 void __cyg_profile_func_exit (void *this_fn, void *call_site);
7299 The first argument is the address of the start of the current function,
7300 which may be looked up exactly in the symbol table.
7302 This instrumentation is also done for functions expanded inline in other
7303 functions. The profiling calls will indicate where, conceptually, the
7304 inline function is entered and exited. This means that addressable
7305 versions of such functions must be available. If all your uses of a
7306 function are expanded inline, this may mean an additional expansion of
7307 code size. If you use @samp{extern inline} in your C code, an
7308 addressable version of such functions must be provided. (This is
7309 normally the case anyways, but if you get lucky and the optimizer always
7310 expands the functions inline, you might have gotten away without
7311 providing static copies.)
7313 A function may be given the attribute @code{no_instrument_function}, in
7314 which case this instrumentation will not be done. This can be used, for
7315 example, for the profiling functions listed above, high-priority
7316 interrupt routines, and any functions from which the profiling functions
7317 cannot safely be called (perhaps signal handlers, if the profiling
7318 routines generate output or allocate memory).
7321 Generate code to verify that you do not go beyond the boundary of the
7322 stack. You should specify this flag if you are running in an
7323 environment with multiple threads, but only rarely need to specify it in
7324 a single-threaded environment since stack overflow is automatically
7325 detected on nearly all systems if there is only one stack.
7327 Note that this switch does not actually cause checking to be done; the
7328 operating system must do that. The switch causes generation of code
7329 to ensure that the operating system sees the stack being extended.
7331 @item -fstack-limit-register=@var{reg}
7332 @itemx -fstack-limit-symbol=@var{sym}
7333 @itemx -fno-stack-limit
7334 Generate code to ensure that the stack does not grow beyond a certain value,
7335 either the value of a register or the address of a symbol. If the stack
7336 would grow beyond the value, a signal is raised. For most targets,
7337 the signal is raised before the stack overruns the boundary, so
7338 it is possible to catch the signal without taking special precautions.
7340 For instance, if the stack starts at address @samp{0x80000000} and grows
7341 downwards you can use the flags
7342 @samp{-fstack-limit-symbol=__stack_limit}
7343 @samp{-Wl,--defsym,__stack_limit=0x7ffe0000} which will enforce a stack
7346 @cindex aliasing of parameters
7347 @cindex parameters, aliased
7348 @item -fargument-alias
7349 @itemx -fargument-noalias
7350 @itemx -fargument-noalias-global
7351 Specify the possible relationships among parameters and between
7352 parameters and global data.
7354 @samp{-fargument-alias} specifies that arguments (parameters) may
7355 alias each other and may alias global storage.
7356 @samp{-fargument-noalias} specifies that arguments do not alias
7357 each other, but may alias global storage.
7358 @samp{-fargument-noalias-global} specifies that arguments do not
7359 alias each other and do not alias global storage.
7361 Each language will automatically use whatever option is required by
7362 the language standard. You should not need to use these options yourself.
7364 @item -fleading-underscore
7365 This option and its counterpart, -fno-leading-underscore, forcibly
7366 change the way C symbols are represented in the object file. One use
7367 is to help link with legacy assembly code.
7369 Be warned that you should know what you are doing when invoking this
7370 option, and that not all targets provide complete support for it.
7373 @node Environment Variables
7374 @section Environment Variables Affecting GCC
7375 @cindex environment variables
7377 This section describes several environment variables that affect how GCC
7378 operates. Some of them work by specifying directories or prefixes to use
7379 when searching for various kinds of files. Some are used to specify other
7380 aspects of the compilation environment.
7383 Note that you can also specify places to search using options such as
7384 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
7385 take precedence over places specified using environment variables, which
7386 in turn take precedence over those specified by the configuration of GCC.
7390 Note that you can also specify places to search using options such as
7391 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
7392 take precedence over places specified using environment variables, which
7393 in turn take precedence over those specified by the configuration of GCC.
7400 @c @itemx LC_COLLATE
7402 @c @itemx LC_MONETARY
7403 @c @itemx LC_NUMERIC
7408 @c @findex LC_COLLATE
7410 @c @findex LC_MONETARY
7411 @c @findex LC_NUMERIC
7415 These environment variables control the way that GCC uses
7416 localization information that allow GCC to work with different
7417 national conventions. GCC inspects the locale categories
7418 @code{LC_CTYPE} and @code{LC_MESSAGES} if it has been configured to do
7419 so. These locale categories can be set to any value supported by your
7420 installation. A typical value is @samp{en_UK} for English in the United
7423 The @code{LC_CTYPE} environment variable specifies character
7424 classification. GCC uses it to determine the character boundaries in
7425 a string; this is needed for some multibyte encodings that contain quote
7426 and escape characters that would otherwise be interpreted as a string
7429 The @code{LC_MESSAGES} environment variable specifies the language to
7430 use in diagnostic messages.
7432 If the @code{LC_ALL} environment variable is set, it overrides the value
7433 of @code{LC_CTYPE} and @code{LC_MESSAGES}; otherwise, @code{LC_CTYPE}
7434 and @code{LC_MESSAGES} default to the value of the @code{LANG}
7435 environment variable. If none of these variables are set, GCC
7436 defaults to traditional C English behavior.
7440 If @code{TMPDIR} is set, it specifies the directory to use for temporary
7441 files. GCC uses temporary files to hold the output of one stage of
7442 compilation which is to be used as input to the next stage: for example,
7443 the output of the preprocessor, which is the input to the compiler
7446 @item GCC_EXEC_PREFIX
7447 @findex GCC_EXEC_PREFIX
7448 If @code{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
7449 names of the subprograms executed by the compiler. No slash is added
7450 when this prefix is combined with the name of a subprogram, but you can
7451 specify a prefix that ends with a slash if you wish.
7453 If @code{GCC_EXEC_PREFIX} is not set, GNU CC will attempt to figure out
7454 an appropriate prefix to use based on the pathname it was invoked with.
7456 If GCC cannot find the subprogram using the specified prefix, it
7457 tries looking in the usual places for the subprogram.
7459 The default value of @code{GCC_EXEC_PREFIX} is
7460 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
7461 of @code{prefix} when you ran the @file{configure} script.
7463 Other prefixes specified with @samp{-B} take precedence over this prefix.
7465 This prefix is also used for finding files such as @file{crt0.o} that are
7468 In addition, the prefix is used in an unusual way in finding the
7469 directories to search for header files. For each of the standard
7470 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
7471 (more precisely, with the value of @code{GCC_INCLUDE_DIR}), GCC tries
7472 replacing that beginning with the specified prefix to produce an
7473 alternate directory name. Thus, with @samp{-Bfoo/}, GCC will search
7474 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
7475 These alternate directories are searched first; the standard directories
7479 @findex COMPILER_PATH
7480 The value of @code{COMPILER_PATH} is a colon-separated list of
7481 directories, much like @code{PATH}. GCC tries the directories thus
7482 specified when searching for subprograms, if it can't find the
7483 subprograms using @code{GCC_EXEC_PREFIX}.
7486 @findex LIBRARY_PATH
7487 The value of @code{LIBRARY_PATH} is a colon-separated list of
7488 directories, much like @code{PATH}. When configured as a native compiler,
7489 GCC tries the directories thus specified when searching for special
7490 linker files, if it can't find them using @code{GCC_EXEC_PREFIX}. Linking
7491 using GCC also uses these directories when searching for ordinary
7492 libraries for the @samp{-l} option (but directories specified with
7493 @samp{-L} come first).
7495 @item C_INCLUDE_PATH
7496 @itemx CPLUS_INCLUDE_PATH
7497 @itemx OBJC_INCLUDE_PATH
7498 @findex C_INCLUDE_PATH
7499 @findex CPLUS_INCLUDE_PATH
7500 @findex OBJC_INCLUDE_PATH
7501 @c @itemx OBJCPLUS_INCLUDE_PATH
7502 These environment variables pertain to particular languages. Each
7503 variable's value is a colon-separated list of directories, much like
7504 @code{PATH}. When GCC searches for header files, it tries the
7505 directories listed in the variable for the language you are using, after
7506 the directories specified with @samp{-I} but before the standard header
7509 @item DEPENDENCIES_OUTPUT
7510 @findex DEPENDENCIES_OUTPUT
7511 @cindex dependencies for make as output
7512 If this variable is set, its value specifies how to output dependencies
7513 for Make based on the header files processed by the compiler. This
7514 output looks much like the output from the @samp{-M} option
7515 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
7516 in addition to the usual results of compilation.
7518 The value of @code{DEPENDENCIES_OUTPUT} can be just a file name, in
7519 which case the Make rules are written to that file, guessing the target
7520 name from the source file name. Or the value can have the form
7521 @samp{@var{file} @var{target}}, in which case the rules are written to
7522 file @var{file} using @var{target} as the target name.
7526 @cindex locale definition
7527 This variable is used to pass locale information to the compiler. One way in
7528 which this information is used is to determine the character set to be used
7529 when character literals, string literals and comments are parsed in C and C++.
7530 When the compiler is configured to allow multibyte characters,
7531 the following values for @code{LANG} are recognized:
7535 Recognize JIS characters.
7537 Recognize SJIS characters.
7539 Recognize EUCJP characters.
7542 If @code{LANG} is not defined, or if it has some other value, then the
7543 compiler will use mblen and mbtowc as defined by the default locale to
7544 recognize and translate multibyte characters.
7547 @node Running Protoize
7548 @section Running Protoize
7550 The program @code{protoize} is an optional part of GNU C. You can use
7551 it to add prototypes to a program, thus converting the program to ANSI
7552 C in one respect. The companion program @code{unprotoize} does the
7553 reverse: it removes argument types from any prototypes that are found.
7555 When you run these programs, you must specify a set of source files as
7556 command line arguments. The conversion programs start out by compiling
7557 these files to see what functions they define. The information gathered
7558 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
7560 After scanning comes actual conversion. The specified files are all
7561 eligible to be converted; any files they include (whether sources or
7562 just headers) are eligible as well.
7564 But not all the eligible files are converted. By default,
7565 @code{protoize} and @code{unprotoize} convert only source and header
7566 files in the current directory. You can specify additional directories
7567 whose files should be converted with the @samp{-d @var{directory}}
7568 option. You can also specify particular files to exclude with the
7569 @samp{-x @var{file}} option. A file is converted if it is eligible, its
7570 directory name matches one of the specified directory names, and its
7571 name within the directory has not been excluded.
7573 Basic conversion with @code{protoize} consists of rewriting most
7574 function definitions and function declarations to specify the types of
7575 the arguments. The only ones not rewritten are those for varargs
7578 @code{protoize} optionally inserts prototype declarations at the
7579 beginning of the source file, to make them available for any calls that
7580 precede the function's definition. Or it can insert prototype
7581 declarations with block scope in the blocks where undeclared functions
7584 Basic conversion with @code{unprotoize} consists of rewriting most
7585 function declarations to remove any argument types, and rewriting
7586 function definitions to the old-style pre-ANSI form.
7588 Both conversion programs print a warning for any function declaration or
7589 definition that they can't convert. You can suppress these warnings
7592 The output from @code{protoize} or @code{unprotoize} replaces the
7593 original source file. The original file is renamed to a name ending
7594 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
7595 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
7596 for DOS) file already exists, then the source file is simply discarded.
7598 @code{protoize} and @code{unprotoize} both depend on GCC itself to
7599 scan the program and collect information about the functions it uses.
7600 So neither of these programs will work until GCC is installed.
7602 Here is a table of the options you can use with @code{protoize} and
7603 @code{unprotoize}. Each option works with both programs unless
7607 @item -B @var{directory}
7608 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
7609 usual directory (normally @file{/usr/local/lib}). This file contains
7610 prototype information about standard system functions. This option
7611 applies only to @code{protoize}.
7613 @item -c @var{compilation-options}
7614 Use @var{compilation-options} as the options when running @code{gcc} to
7615 produce the @samp{.X} files. The special option @samp{-aux-info} is
7616 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
7618 Note that the compilation options must be given as a single argument to
7619 @code{protoize} or @code{unprotoize}. If you want to specify several
7620 @code{gcc} options, you must quote the entire set of compilation options
7621 to make them a single word in the shell.
7623 There are certain @code{gcc} arguments that you cannot use, because they
7624 would produce the wrong kind of output. These include @samp{-g},
7625 @samp{-O}, @samp{-c}, @samp{-S}, and @samp{-o} If you include these in
7626 the @var{compilation-options}, they are ignored.
7629 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
7630 systems) instead of @samp{.c}. This is convenient if you are converting
7631 a C program to C++. This option applies only to @code{protoize}.
7634 Add explicit global declarations. This means inserting explicit
7635 declarations at the beginning of each source file for each function
7636 that is called in the file and was not declared. These declarations
7637 precede the first function definition that contains a call to an
7638 undeclared function. This option applies only to @code{protoize}.
7640 @item -i @var{string}
7641 Indent old-style parameter declarations with the string @var{string}.
7642 This option applies only to @code{protoize}.
7644 @code{unprotoize} converts prototyped function definitions to old-style
7645 function definitions, where the arguments are declared between the
7646 argument list and the initial @samp{@{}. By default, @code{unprotoize}
7647 uses five spaces as the indentation. If you want to indent with just
7648 one space instead, use @samp{-i " "}.
7651 Keep the @samp{.X} files. Normally, they are deleted after conversion
7655 Add explicit local declarations. @code{protoize} with @samp{-l} inserts
7656 a prototype declaration for each function in each block which calls the
7657 function without any declaration. This option applies only to
7661 Make no real changes. This mode just prints information about the conversions
7662 that would have been done without @samp{-n}.
7665 Make no @samp{.save} files. The original files are simply deleted.
7666 Use this option with caution.
7668 @item -p @var{program}
7669 Use the program @var{program} as the compiler. Normally, the name
7673 Work quietly. Most warnings are suppressed.
7676 Print the version number, just like @samp{-v} for @code{gcc}.
7679 If you need special compiler options to compile one of your program's
7680 source files, then you should generate that file's @samp{.X} file
7681 specially, by running @code{gcc} on that source file with the
7682 appropriate options and the option @samp{-aux-info}. Then run
7683 @code{protoize} on the entire set of files. @code{protoize} will use
7684 the existing @samp{.X} file because it is newer than the source file.
7688 gcc -Dfoo=bar file1.c -aux-info
7693 You need to include the special files along with the rest in the
7694 @code{protoize} command, even though their @samp{.X} files already
7695 exist, because otherwise they won't get converted.
7697 @xref{Protoize Caveats}, for more information on how to use
7698 @code{protoize} successfully.