1 @c Copyright (C) 1988, 89, 92-99, 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 -fdollars-in-identifiers
107 -fno-elide-constructors -fexternal-templates -ffor-scope
108 -fno-for-scope -fno-gnu-keywords -fguiding-decls
109 -fhonor-std -fhuge-objects -fno-implicit-templates -finit-priority
110 -fno-implement-inlines -fname-mangling-version-@var{n} -fno-default-inline
111 -foperator-names -fno-optional-diags -fpermissive -frepo -fstrict-prototype
112 -fsquangle -ftemplate-depth-@var{n} -fuse-cxa-atexit -fvtable-thunks
113 -nostdinc++ -Wctor-dtor-privacy -Wno-deprecated -Weffc++
114 -Wno-non-template-friend
115 -Wnon-virtual-dtor -Wold-style-cast -Woverloaded-virtual
116 -Wno-pmf-conversions -Wreorder -Wsign-promo -Wsynth
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 -fregmove
168 -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 -mnop-fun-dllimport -mno-nop-fun-dllimport
273 -msingle-pic-base -mno-single-pic-base
277 -mtpcs-frame -mno-tpcs-frame
278 -mtpcs-leaf-frame -mno-tpcs-leaf-frame
279 -mlittle-endian -mbig-endian
280 -mthumb-interwork -mno-thumb-interwork
281 -mstructure-size-boundary=
282 -mnop-fun-dllimport -mno-nop-fun-dllimport
283 -mcallee-super-interworking -mno-callee-super-interworking
284 -mcaller-super-interworking -mno-caller-super-interworking
285 -msingle-pic-base -mno-single-pic-base
288 @emph{MN10200 Options}
291 @emph{MN10300 Options}
298 @emph{M32R/D Options}
299 -mcode-model=@var{model type} -msdata=@var{sdata type}
303 -m88000 -m88100 -m88110 -mbig-pic
304 -mcheck-zero-division -mhandle-large-shift
305 -midentify-revision -mno-check-zero-division
306 -mno-ocs-debug-info -mno-ocs-frame-position
307 -mno-optimize-arg-area -mno-serialize-volatile
308 -mno-underscores -mocs-debug-info
309 -mocs-frame-position -moptimize-arg-area
310 -mserialize-volatile -mshort-data-@var{num} -msvr3
311 -msvr4 -mtrap-large-shift -muse-div-instruction
312 -mversion-03.00 -mwarn-passed-structs
314 @emph{RS/6000 and PowerPC Options}
316 -mtune=@var{cpu type}
317 -mpower -mno-power -mpower2 -mno-power2
318 -mpowerpc -mpowerpc64 -mno-powerpc
319 -mpowerpc-gpopt -mno-powerpc-gpopt
320 -mpowerpc-gfxopt -mno-powerpc-gfxopt
321 -mnew-mnemonics -mold-mnemonics
322 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc
323 -m64 -m32 -mxl-call -mno-xl-call -mthreads -mpe
324 -msoft-float -mhard-float -mmultiple -mno-multiple
325 -mstring -mno-string -mupdate -mno-update
326 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align
327 -mstrict-align -mno-strict-align -mrelocatable
328 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib
329 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian
330 -mcall-aix -mcall-sysv -mprototype -mno-prototype
331 -msim -mmvme -mads -myellowknife -memb -msdata
332 -msdata=@var{opt} -G @var{num}
335 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs
336 -mfull-fp-blocks -mhc-struct-return -min-line-mul
337 -mminimum-fp-blocks -mnohc-struct-return
340 -mabicalls -mcpu=@var{cpu type} -membedded-data -muninit-const-in-rodata
341 -membedded-pic -mfp32 -mfp64 -mgas -mgp32 -mgp64
342 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1
343 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy
344 -mmips-as -mmips-tfile -mno-abicalls
345 -mno-embedded-data -mno-uninit-const-in-rodata -mno-embedded-pic
346 -mno-gpopt -mno-long-calls
347 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats
348 -mrnames -msoft-float
349 -m4650 -msingle-float -mmad
350 -mstats -EL -EB -G @var{num} -nocpp
351 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi
356 -march=@var{cpu type}
357 -mieee-fp -mno-fancy-math-387
358 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib
359 -mno-wide-multiply -mrtd -malign-double
360 -mreg-alloc=@var{list} -mregparm=@var{num}
361 -malign-jumps=@var{num} -malign-loops=@var{num}
362 -malign-functions=@var{num} -mpreferred-stack-boundary=@var{num}
363 -mthreads -mno-align-stringops -minline-all-stringops
366 -march=@var{architecture type}
367 -mbig-switch -mdisable-fpregs -mdisable-indexing
368 -mfast-indirect-calls -mgas -mjump-in-delay
369 -mlong-load-store -mno-big-switch -mno-disable-fpregs
370 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas
371 -mno-jump-in-delay -mno-long-load-store
372 -mno-portable-runtime -mno-soft-float
373 -mno-space-regs -msoft-float -mpa-risc-1-0
374 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime
375 -mschedule=@var{cpu type} -mspace-regs
377 @emph{Intel 960 Options}
378 -m@var{cpu type} -masm-compat -mclean-linkage
379 -mcode-align -mcomplex-addr -mleaf-procedures
380 -mic-compat -mic2.0-compat -mic3.0-compat
381 -mintel-asm -mno-clean-linkage -mno-code-align
382 -mno-complex-addr -mno-leaf-procedures
383 -mno-old-align -mno-strict-align -mno-tail-call
384 -mnumerics -mold-align -msoft-float -mstrict-align
387 @emph{DEC Alpha Options}
388 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float
390 -mieee -mieee-with-inexact -mieee-conformant
391 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode}
392 -mtrap-precision=@var{mode} -mbuild-constants
394 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max
395 -mmemory-latency=@var{time}
397 @emph{Clipper Options}
400 @emph{H8/300 Options}
401 -mrelax -mh -ms -mint32 -malign-300
404 -m1 -m2 -m3 -m3e -mb -ml -mdalign -mrelax
406 @emph{System V Options}
407 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}
411 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text section}
412 -mdata=@var{data section} -mrodata=@var{readonly data section}
414 @emph{TMS320C3x/C4x Options}
415 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm
416 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload
417 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned
418 -mparallel-insns -mparallel-mpy -mpreserve-float
421 -mlong-calls -mno-long-calls -mep -mno-ep
422 -mprolog-function -mno-prolog-function -mspace
423 -mtda=@var{n} -msda=@var{n} -mzda=@var{n}
427 -m32032 -m32332 -m32532 -m32081 -m32381 -mmult-add -mnomult-add
428 -msoft-float -mrtd -mnortd -mregparam -mnoregparam -msb -mnosb
429 -mbitfield -mnobitfield -mhimem -mnohimem
432 -mhardlit, -mno-hardlit -mdiv -mno-div -mrelax-immediates
433 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields
434 -m4byte-functions -mno-4byte-functions -mcallgraph-data
435 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim
436 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment
439 @item Code Generation Options
440 @xref{Code Gen Options,,Options for Code Generation Conventions}.
442 -fcall-saved-@var{reg} -fcall-used-@var{reg}
443 -fexceptions -funwind-tables -ffixed-@var{reg} -finhibit-size-directive
444 -fcheck-memory-usage -fprefix-function-name
445 -fno-common -fno-ident -fno-gnu-linker
446 -fpcc-struct-return -fpic -fPIC
447 -freg-struct-return -fshared-data -fshort-enums
448 -fshort-double -fvolatile -fvolatile-global -fvolatile-static
449 -fverbose-asm -fpack-struct -fstack-check
450 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym}
451 -fargument-alias -fargument-noalias
452 -fargument-noalias-global
458 * Overall Options:: Controlling the kind of output:
459 an executable, object files, assembler files,
460 or preprocessed source.
461 * C Dialect Options:: Controlling the variant of C language compiled.
462 * C++ Dialect Options:: Variations on C++.
463 * Warning Options:: How picky should the compiler be?
464 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
465 * Optimize Options:: How much optimization?
466 * Preprocessor Options:: Controlling header files and macro definitions.
467 Also, getting dependency information for Make.
468 * Assembler Options:: Passing options to the assembler.
469 * Link Options:: Specifying libraries and so on.
470 * Directory Options:: Where to find header files and libraries.
471 Where to find the compiler executable files.
472 * Spec Files:: How to pass switches to sub-processes.
473 * Target Options:: Running a cross-compiler, or an old version of GCC.
476 @node Overall Options
477 @section Options Controlling the Kind of Output
479 Compilation can involve up to four stages: preprocessing, compilation
480 proper, assembly and linking, always in that order. The first three
481 stages apply to an individual source file, and end by producing an
482 object file; linking combines all the object files (those newly
483 compiled, and those specified as input) into an executable file.
485 @cindex file name suffix
486 For any given input file, the file name suffix determines what kind of
491 C source code which must be preprocessed.
494 C source code which should not be preprocessed.
497 C++ source code which should not be preprocessed.
500 Objective-C source code. Note that you must link with the library
501 @file{libobjc.a} to make an Objective-C program work.
504 C header file (not to be compiled or linked).
507 @itemx @var{file}.cxx
508 @itemx @var{file}.cpp
510 C++ source code which must be preprocessed. Note that in @samp{.cxx},
511 the last two letters must both be literally @samp{x}. Likewise,
512 @samp{.C} refers to a literal capital C.
518 Assembler code which must be preprocessed.
521 An object file to be fed straight into linking.
522 Any file name with no recognized suffix is treated this way.
525 You can specify the input language explicitly with the @samp{-x} option:
528 @item -x @var{language}
529 Specify explicitly the @var{language} for the following input files
530 (rather than letting the compiler choose a default based on the file
531 name suffix). This option applies to all following input files until
532 the next @samp{-x} option. Possible values for @var{language} are:
535 c-header cpp-output c++-cpp-output
536 assembler assembler-with-cpp
540 Turn off any specification of a language, so that subsequent files are
541 handled according to their file name suffixes (as they are if @samp{-x}
542 has not been used at all).
544 @item -pass-exit-codes
545 Normally the @code{gcc} program will exit with the code of 1 if any
546 phase of the compiler returns a non-success return code. If you specify
547 @samp{-pass-exit-codes}, the @code{gcc} program will instead return with
548 numerically highest error produced by any phase that returned an error
552 If you only want some of the stages of compilation, you can use
553 @samp{-x} (or filename suffixes) to tell @code{gcc} where to start, and
554 one of the options @samp{-c}, @samp{-S}, or @samp{-E} to say where
555 @code{gcc} is to stop. Note that some combinations (for example,
556 @samp{-x cpp-output -E} instruct @code{gcc} to do nothing at all.
560 Compile or assemble the source files, but do not link. The linking
561 stage simply is not done. The ultimate output is in the form of an
562 object file for each source file.
564 By default, the object file name for a source file is made by replacing
565 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
567 Unrecognized input files, not requiring compilation or assembly, are
571 Stop after the stage of compilation proper; do not assemble. The output
572 is in the form of an assembler code file for each non-assembler input
575 By default, the assembler file name for a source file is made by
576 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
578 Input files that don't require compilation are ignored.
581 Stop after the preprocessing stage; do not run the compiler proper. The
582 output is in the form of preprocessed source code, which is sent to the
585 Input files which don't require preprocessing are ignored.
587 @cindex output file option
589 Place output in file @var{file}. This applies regardless to whatever
590 sort of output is being produced, whether it be an executable file,
591 an object file, an assembler file or preprocessed C code.
593 Since only one output file can be specified, it does not make sense to
594 use @samp{-o} when compiling more than one input file, unless you are
595 producing an executable file as output.
597 If @samp{-o} is not specified, the default is to put an executable file
598 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
599 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
600 all preprocessed C source on standard output.@refill
603 Print (on standard error output) the commands executed to run the stages
604 of compilation. Also print the version number of the compiler driver
605 program and of the preprocessor and the compiler proper.
608 Use pipes rather than temporary files for communication between the
609 various stages of compilation. This fails to work on some systems where
610 the assembler is unable to read from a pipe; but the GNU assembler has
614 Print (on the standard output) a description of the command line options
615 understood by @code{gcc}. If the @code{-v} option is also specified
616 then @code{--help} will also be passed on to the various processes
617 invoked by @code{gcc}, so that they can display the command line options
618 they accept. If the @code{-W} option is also specified then command
619 line options which have no documentation associated with them will also
624 @section Compiling C++ Programs
626 @cindex suffixes for C++ source
627 @cindex C++ source file suffixes
628 C++ source files conventionally use one of the suffixes @samp{.C},
629 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
630 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
631 files with these names and compiles them as C++ programs even if you
632 call the compiler the same way as for compiling C programs (usually with
633 the name @code{gcc}).
637 However, C++ programs often require class libraries as well as a
638 compiler that understands the C++ language---and under some
639 circumstances, you might want to compile programs from standard input,
640 or otherwise without a suffix that flags them as C++ programs.
641 @code{g++} is a program that calls GCC with the default language
642 set to C++, and automatically specifies linking against the C++
643 library. On many systems, the script @code{g++} is also
644 installed with the name @code{c++}.
646 @cindex invoking @code{g++}
647 When you compile C++ programs, you may specify many of the same
648 command-line options that you use for compiling programs in any
649 language; or command-line options meaningful for C and related
650 languages; or options that are meaningful only for C++ programs.
651 @xref{C Dialect Options,,Options Controlling C Dialect}, for
652 explanations of options for languages related to C.
653 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
654 explanations of options that are meaningful only for C++ programs.
656 @node C Dialect Options
657 @section Options Controlling C Dialect
658 @cindex dialect options
659 @cindex language dialect options
660 @cindex options, dialect
662 The following options control the dialect of C (or languages derived
663 from C, such as C++ and Objective C) that the compiler accepts:
668 In C mode, support all ANSI standard C programs. In C++ mode,
669 remove GNU extensions that conflict with ANSI C++.
670 @c shouldn't we be saying "ISO"?
672 This turns off certain features of GCC that are incompatible with ANSI
673 C (when compiling C code), or of ANSI standard C++ (when compiling C++ code),
674 such as the @code{asm} and @code{typeof} keywords, and
675 predefined macros such as @code{unix} and @code{vax} that identify the
676 type of system you are using. It also enables the undesirable and
677 rarely used ANSI trigraph feature. For the C compiler,
678 it disables recognition of C++ style @samp{//} comments as well as
679 the @code{inline} keyword. For the C++ compiler,
680 @samp{-foperator-names} is enabled as well.
683 The alternate keywords @code{__asm__}, @code{__extension__},
684 @code{__inline__} and @code{__typeof__} continue to work despite
685 @samp{-ansi}. You would not want to use them in an ANSI C program, of
686 course, but it is useful to put them in header files that might be included
687 in compilations done with @samp{-ansi}. Alternate predefined macros
688 such as @code{__unix__} and @code{__vax__} are also available, with or
689 without @samp{-ansi}.
691 The @samp{-ansi} option does not cause non-ANSI programs to be
692 rejected gratuitously. For that, @samp{-pedantic} is required in
693 addition to @samp{-ansi}. @xref{Warning Options}.
695 The macro @code{__STRICT_ANSI__} is predefined when the @samp{-ansi}
696 option is used. Some header files may notice this macro and refrain
697 from declaring certain functions or defining certain macros that the
698 ANSI standard doesn't call for; this is to avoid interfering with any
699 programs that might use these names for other things.
701 The functions @code{alloca}, @code{abort}, @code{exit}, and
702 @code{_exit} are not builtin functions when @samp{-ansi} is used.
705 Determine the language standard. A value for this option must be provided;
713 ISO C as modified in amend. 1
719 same as -std=iso9899:1990
722 same as -std=iso9899:199x
725 default, iso9899:1990 + gnu extensions
728 iso9899:199x + gnu extensions
731 Even when this option is not specified, you can still use some of the
732 features of newer standards in so far as they do not conflict with
733 previous C standards. For example, you may use @code{__restrict__} even
734 when -fstd=c9x is not specified.
737 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
738 keyword, so that code can use these words as identifiers. You can use
739 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
740 instead. @samp{-ansi} implies @samp{-fno-asm}.
742 In C++, this switch only affects the @code{typeof} keyword, since
743 @code{asm} and @code{inline} are standard keywords. You may want to
744 use the @samp{-fno-gnu-keywords} flag instead, as it also disables the
745 other, C++-specific, extension keywords such as @code{headof}.
748 @cindex builtin functions
774 Don't recognize builtin functions that do not begin with @samp{__builtin_}
775 as prefix. Currently, the functions affected include @code{abort},
776 @code{abs}, @code{alloca}, @code{cos}, @code{cosf}, @code{cosl},
777 @code{exit}, @code{_exit}, @code{fabs}, @code{fabsf}, @code{fabsl},
778 @code{ffs}, @code{labs}, @code{memcmp}, @code{memcpy}, @code{memset},
779 @code{sin}, @code{sinf}, @code{sinl}, @code{sqrt}, @code{sqrtf},
780 @code{sqrtl}, @code{strcmp}, @code{strcpy}, and @code{strlen}.
782 GCC normally generates special code to handle certain builtin functions
783 more efficiently; for instance, calls to @code{alloca} may become single
784 instructions that adjust the stack directly, and calls to @code{memcpy}
785 may become inline copy loops. The resulting code is often both smaller
786 and faster, but since the function calls no longer appear as such, you
787 cannot set a breakpoint on those calls, nor can you change the behavior
788 of the functions by linking with a different library.
790 The @samp{-ansi} option prevents @code{alloca}, @code{ffs} and @code{_exit}
791 from being builtin functions, since these functions do not have an ANSI
795 @cindex hosted environment
797 Assert that compilation takes place in a hosted environment. This implies
798 @samp{-fbuiltin}. A hosted environment is one in which the
799 entire standard library is available, and in which @code{main} has a return
800 type of @code{int}. Examples are nearly everything except a kernel.
801 This is equivalent to @samp{-fno-freestanding}.
804 @cindex hosted environment
806 Assert that compilation takes place in a freestanding environment. This
807 implies @samp{-fno-builtin}. A freestanding environment
808 is one in which the standard library may not exist, and program startup may
809 not necessarily be at @code{main}. The most obvious example is an OS kernel.
810 This is equivalent to @samp{-fno-hosted}.
813 Support ANSI C trigraphs. You don't want to know about this
814 brain-damage. The @samp{-ansi} option implies @samp{-trigraphs}.
816 @cindex traditional C language
817 @cindex C language, traditional
819 Attempt to support some aspects of traditional C compilers.
824 All @code{extern} declarations take effect globally even if they
825 are written inside of a function definition. This includes implicit
826 declarations of functions.
829 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
830 and @code{volatile} are not recognized. (You can still use the
831 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
835 Comparisons between pointers and integers are always allowed.
838 Integer types @code{unsigned short} and @code{unsigned char} promote
839 to @code{unsigned int}.
842 Out-of-range floating point literals are not an error.
845 Certain constructs which ANSI regards as a single invalid preprocessing
846 number, such as @samp{0xe-0xd}, are treated as expressions instead.
849 String ``constants'' are not necessarily constant; they are stored in
850 writable space, and identical looking constants are allocated
851 separately. (This is the same as the effect of
852 @samp{-fwritable-strings}.)
854 @cindex @code{longjmp} and automatic variables
856 All automatic variables not declared @code{register} are preserved by
857 @code{longjmp}. Ordinarily, GNU C follows ANSI C: automatic variables
858 not declared @code{volatile} may be clobbered.
863 @cindex escape sequences, traditional
864 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
865 literal characters @samp{x} and @samp{a} respectively. Without
866 @w{@samp{-traditional}}, @samp{\x} is a prefix for the hexadecimal
867 representation of a character, and @samp{\a} produces a bell.
870 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
871 if your program uses names that are normally GNU C builtin functions for
872 other purposes of its own.
874 You cannot use @samp{-traditional} if you include any header files that
875 rely on ANSI C features. Some vendors are starting to ship systems with
876 ANSI C header files and you cannot use @samp{-traditional} on such
877 systems to compile files that include any system headers.
879 The @samp{-traditional} option also enables @samp{-traditional-cpp},
880 which is described next.
882 @item -traditional-cpp
883 Attempt to support some aspects of traditional C preprocessors.
888 Comments convert to nothing at all, rather than to a space. This allows
889 traditional token concatenation.
892 In a preprocessing directive, the @samp{#} symbol must appear as the first
896 Macro arguments are recognized within string constants in a macro
897 definition (and their values are stringified, though without additional
898 quote marks, when they appear in such a context). The preprocessor
899 always considers a string constant to end at a newline.
902 @cindex detecting @w{@samp{-traditional}}
903 The predefined macro @code{__STDC__} is not defined when you use
904 @samp{-traditional}, but @code{__GNUC__} is (since the GNU extensions
905 which @code{__GNUC__} indicates are not affected by
906 @samp{-traditional}). If you need to write header files that work
907 differently depending on whether @samp{-traditional} is in use, by
908 testing both of these predefined macros you can distinguish four
909 situations: GNU C, traditional GNU C, other ANSI C compilers, and other
910 old C compilers. The predefined macro @code{__STDC_VERSION__} is also
911 not defined when you use @samp{-traditional}. @xref{Standard
912 Predefined,,Standard Predefined Macros,cpp.info,The C Preprocessor},
913 for more discussion of these and other predefined macros.
916 @cindex string constants vs newline
917 @cindex newline vs string constants
918 The preprocessor considers a string constant to end at a newline (unless
919 the newline is escaped with @samp{\}). (Without @w{@samp{-traditional}},
920 string constants can contain the newline character as typed.)
923 @item -fcond-mismatch
924 Allow conditional expressions with mismatched types in the second and
925 third arguments. The value of such an expression is void.
927 @item -funsigned-char
928 Let the type @code{char} be unsigned, like @code{unsigned char}.
930 Each kind of machine has a default for what @code{char} should
931 be. It is either like @code{unsigned char} by default or like
932 @code{signed char} by default.
934 Ideally, a portable program should always use @code{signed char} or
935 @code{unsigned char} when it depends on the signedness of an object.
936 But many programs have been written to use plain @code{char} and
937 expect it to be signed, or expect it to be unsigned, depending on the
938 machines they were written for. This option, and its inverse, let you
939 make such a program work with the opposite default.
941 The type @code{char} is always a distinct type from each of
942 @code{signed char} or @code{unsigned char}, even though its behavior
943 is always just like one of those two.
946 Let the type @code{char} be signed, like @code{signed char}.
948 Note that this is equivalent to @samp{-fno-unsigned-char}, which is
949 the negative form of @samp{-funsigned-char}. Likewise, the option
950 @samp{-fno-signed-char} is equivalent to @samp{-funsigned-char}.
952 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
953 if your program uses names that are normally GNU C builtin functions for
954 other purposes of its own.
956 You cannot use @samp{-traditional} if you include any header files that
957 rely on ANSI C features. Some vendors are starting to ship systems with
958 ANSI C header files and you cannot use @samp{-traditional} on such
959 systems to compile files that include any system headers.
961 @item -fsigned-bitfields
962 @itemx -funsigned-bitfields
963 @itemx -fno-signed-bitfields
964 @itemx -fno-unsigned-bitfields
965 These options control whether a bitfield is signed or unsigned, when the
966 declaration does not use either @code{signed} or @code{unsigned}. By
967 default, such a bitfield is signed, because this is consistent: the
968 basic integer types such as @code{int} are signed types.
970 However, when @samp{-traditional} is used, bitfields are all unsigned
973 @item -fwritable-strings
974 Store string constants in the writable data segment and don't uniquize
975 them. This is for compatibility with old programs which assume they can
976 write into string constants. The option @samp{-traditional} also has
979 Writing into string constants is a very bad idea; ``constants'' should
982 @item -fallow-single-precision
983 Do not promote single precision math operations to double precision,
984 even when compiling with @samp{-traditional}.
986 Traditional K&R C promotes all floating point operations to double
987 precision, regardless of the sizes of the operands. On the
988 architecture for which you are compiling, single precision may be faster
989 than double precision. If you must use @samp{-traditional}, but want
990 to use single precision operations when the operands are single
991 precision, use this option. This option has no effect when compiling
992 with ANSI or GNU C conventions (the default).
995 Override the underlying type for @samp{wchar_t} to be @samp{short
996 unsigned int} instead of the default for the target. This option is
997 useful for building programs to run under WINE.
1000 @node C++ Dialect Options
1001 @section Options Controlling C++ Dialect
1003 @cindex compiler options, C++
1004 @cindex C++ options, command line
1005 @cindex options, C++
1006 This section describes the command-line options that are only meaningful
1007 for C++ programs; but you can also use most of the GNU compiler options
1008 regardless of what language your program is in. For example, you
1009 might compile a file @code{firstClass.C} like this:
1012 g++ -g -frepo -O -c firstClass.C
1016 In this example, only @samp{-frepo} is an option meant
1017 only for C++ programs; you can use the other options with any
1018 language supported by GCC.
1020 Here is a list of options that are @emph{only} for compiling C++ programs:
1023 @item -fno-access-control
1024 Turn off all access checking. This switch is mainly useful for working
1025 around bugs in the access control code.
1028 Check that the pointer returned by @code{operator new} is non-null
1029 before attempting to modify the storage allocated. The current Working
1030 Paper requires that @code{operator new} never return a null pointer, so
1031 this check is normally unnecessary.
1033 An alternative to using this option is to specify that your
1034 @code{operator new} does not throw any exceptions; if you declare it
1035 @samp{throw()}, g++ will check the return value. See also @samp{new
1038 @item -fconserve-space
1039 Put uninitialized or runtime-initialized global variables into the
1040 common segment, as C does. This saves space in the executable at the
1041 cost of not diagnosing duplicate definitions. If you compile with this
1042 flag and your program mysteriously crashes after @code{main()} has
1043 completed, you may have an object that is being destroyed twice because
1044 two definitions were merged.
1046 This option is no longer useful on most targets, now that support has
1047 been added for putting variables into BSS without making them common.
1049 @item -fdollars-in-identifiers
1050 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1051 @samp{$} with the option @samp{-fno-dollars-in-identifiers}. (GNU C allows
1052 @samp{$} by default on most target systems, but there are a few exceptions.)
1053 Traditional C allowed the character @samp{$} to form part of
1054 identifiers. However, ANSI C and C++ forbid @samp{$} in identifiers.
1056 @item -fno-elide-constructors
1057 The C++ standard allows an implementation to omit creating a temporary
1058 which is only used to initialize another object of the same type.
1059 Specifying this option disables that optimization, and forces g++ to
1060 call the copy constructor in all cases.
1062 @item -fexternal-templates
1063 Cause template instantiations to obey @samp{#pragma interface} and
1064 @samp{implementation}; template instances are emitted or not according
1065 to the location of the template definition. @xref{Template
1066 Instantiation}, for more information.
1068 This option is deprecated.
1070 @item -falt-external-templates
1071 Similar to -fexternal-templates, but template instances are emitted or
1072 not according to the place where they are first instantiated.
1073 @xref{Template Instantiation}, for more information.
1075 This option is deprecated.
1078 @itemx -fno-for-scope
1079 If -ffor-scope is specified, the scope of variables declared in
1080 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1081 as specified by the draft C++ standard.
1082 If -fno-for-scope is specified, the scope of variables declared in
1083 a @i{for-init-statement} extends to the end of the enclosing scope,
1084 as was the case in old versions of gcc, and other (traditional)
1085 implementations of C++.
1087 The default if neither flag is given to follow the standard,
1088 but to allow and give a warning for old-style code that would
1089 otherwise be invalid, or have different behavior.
1091 @item -fno-gnu-keywords
1092 Do not recognize @code{classof}, @code{headof}, or @code{typeof} as a
1093 keyword, so that code can use these words as identifiers. You can use
1094 the keywords @code{__classof__}, @code{__headof__}, and
1095 @code{__typeof__} instead. @samp{-ansi} implies
1096 @samp{-fno-gnu-keywords}.
1098 @item -fguiding-decls
1099 Treat a function declaration with the same type as a potential function
1100 template instantiation as though it declares that instantiation, not a
1101 normal function. If a definition is given for the function later in the
1102 translation unit (or another translation unit if the target supports
1103 weak symbols), that definition will be used; otherwise the template will
1104 be instantiated. This behavior reflects the C++ language prior to
1105 September 1996, when guiding declarations were removed.
1107 This option implies @samp{-fname-mangling-version-0}, and will not work
1108 with other name mangling versions. Like all options that change the
1109 ABI, all C++ code, @emph{including libgcc.a} must be built with the same
1110 setting of this option.
1113 Treat the @code{namespace std} as a namespace, instead of ignoring
1114 it. For compatibility with earlier versions of g++, the compiler will,
1115 by default, ignore @code{namespace-declarations},
1116 @code{using-declarations}, @code{using-directives}, and
1117 @code{namespace-names}, if they involve @code{std}.
1119 @item -fhuge-objects
1120 Support virtual function calls for objects that exceed the size
1121 representable by a @samp{short int}. Users should not use this flag by
1122 default; if you need to use it, the compiler will tell you so.
1124 This flag is not useful when compiling with -fvtable-thunks.
1126 Like all options that change the ABI, all C++ code, @emph{including
1127 libgcc} must be built with the same setting of this option.
1129 @item -fno-implicit-templates
1130 Never emit code for non-inline templates which are instantiated
1131 implicitly (i.e. by use); only emit code for explicit instantiations.
1132 @xref{Template Instantiation}, for more information.
1134 @item -fno-implicit-inline-templates
1135 Don't emit code for implicit instantiations of inline templates, either.
1136 The default is to handle inlines differently so that compiles with and
1137 without optimization will need the same set of explicit instantiations.
1139 @item -finit-priority
1140 Support @samp{__attribute__ ((init_priority (n)))} for controlling the
1141 order of initialization of file-scope objects. On ELF targets, this
1142 requires GNU ld 2.10 or later.
1144 @item -fno-implement-inlines
1145 To save space, do not emit out-of-line copies of inline functions
1146 controlled by @samp{#pragma implementation}. This will cause linker
1147 errors if these functions are not inlined everywhere they are called.
1149 @item -fms-extensions
1150 Disable pedwarns about constructs used in MFC, such as implicit int and
1151 getting a pointer to member function via non-standard syntax.
1153 @item -fname-mangling-version-@var{n}
1154 Control the way in which names are mangled. Version 0 is compatible
1155 with versions of g++ before 2.8. Version 1 is the default. Version 1
1156 will allow correct mangling of function templates. For example,
1157 version 0 mangling does not mangle foo<int, double> and foo<int, char>
1158 given this declaration:
1161 template <class T, class U> void foo(T t);
1164 Like all options that change the ABI, all C++ code, @emph{including
1165 libgcc} must be built with the same setting of this option.
1167 @item -foperator-names
1168 Recognize the operator name keywords @code{and}, @code{bitand},
1169 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1170 synonyms for the symbols they refer to. @samp{-ansi} implies
1171 @samp{-foperator-names}.
1173 @item -fno-optional-diags
1174 Disable diagnostics that the standard says a compiler does not need to
1175 issue. Currently, the only such diagnostic issued by g++ is the one for
1176 a name having multiple meanings within a class.
1179 Downgrade messages about nonconformant code from errors to warnings. By
1180 default, g++ effectively sets @samp{-pedantic-errors} without
1181 @samp{-pedantic}; this option reverses that. This behavior and this
1182 option are superseded by @samp{-pedantic}, which works as it does for GNU C.
1185 Enable automatic template instantiation. This option also implies
1186 @samp{-fno-implicit-templates}. @xref{Template Instantiation}, for more
1190 Disable generation of information about every class with virtual
1191 functions for use by the C++ runtime type identification features
1192 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1193 of the language, you can save some space by using this flag. Note that
1194 exception handling uses the same information, but it will generate it as
1197 @item -fstrict-prototype
1198 Within an @samp{extern "C"} linkage specification, treat a function
1199 declaration with no arguments, such as @samp{int foo ();}, as declaring
1200 the function to take no arguments. Normally, such a declaration means
1201 that the function @code{foo} can take any combination of arguments, as
1202 in C. @samp{-pedantic} implies @samp{-fstrict-prototype} unless
1203 overridden with @samp{-fno-strict-prototype}.
1205 Specifying this option will also suppress implicit declarations of
1208 This flag no longer affects declarations with C++ linkage.
1211 @itemx -fno-squangle
1212 @samp{-fsquangle} will enable a compressed form of name mangling for
1213 identifiers. In particular, it helps to shorten very long names by recognizing
1214 types and class names which occur more than once, replacing them with special
1215 short ID codes. This option also requires any C++ libraries being used to
1216 be compiled with this option as well. The compiler has this disabled (the
1217 equivalent of @samp{-fno-squangle}) by default.
1219 Like all options that change the ABI, all C++ code, @emph{including
1220 libgcc.a} must be built with the same setting of this option.
1222 @item -ftemplate-depth-@var{n}
1223 Set the maximum instantiation depth for template classes to @var{n}.
1224 A limit on the template instantiation depth is needed to detect
1225 endless recursions during template class instantiation. ANSI/ISO C++
1226 conforming programs must not rely on a maximum depth greater than 17.
1228 @item -fuse-cxa-atexit
1229 Register destructors for objects with static storage duration with the
1230 @code{__cxa_atexit} function rather than the @code{atexit} function.
1231 This option is required for fully standards-compliant handling of static
1232 destructors, but will only work if your C library supports
1233 @code{__cxa_atexit}.
1235 @item -fvtable-thunks
1236 Use @samp{thunks} to implement the virtual function dispatch table
1237 (@samp{vtable}). The traditional (cfront-style) approach to
1238 implementing vtables was to store a pointer to the function and two
1239 offsets for adjusting the @samp{this} pointer at the call site. Newer
1240 implementations store a single pointer to a @samp{thunk} function which
1241 does any necessary adjustment and then calls the target function.
1243 This option also enables a heuristic for controlling emission of
1244 vtables; if a class has any non-inline virtual functions, the vtable
1245 will be emitted in the translation unit containing the first one of
1248 Like all options that change the ABI, all C++ code, @emph{including
1249 libgcc.a} must be built with the same setting of this option.
1252 Do not search for header files in the standard directories specific to
1253 C++, but do still search the other standard directories. (This option
1254 is used when building the C++ library.)
1257 In addition, these optimization, warning, and code generation options
1258 have meanings only for C++ programs:
1261 @item -fno-default-inline
1262 Do not assume @samp{inline} for functions defined inside a class scope.
1263 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1264 functions will have linkage like inline functions; they just won't be
1267 @item -Wctor-dtor-privacy (C++ only)
1268 Warn when a class seems unusable, because all the constructors or
1269 destructors in a class are private and the class has no friends or
1270 public static member functions.
1272 @item -Wnon-virtual-dtor (C++ only)
1273 Warn when a class declares a non-virtual destructor that should probably
1274 be virtual, because it looks like the class will be used polymorphically.
1276 @item -Wreorder (C++ only)
1277 @cindex reordering, warning
1278 @cindex warning for reordering of member initializers
1279 Warn when the order of member initializers given in the code does not
1280 match the order in which they must be executed. For instance:
1286 A(): j (0), i (1) @{ @}
1290 Here the compiler will warn that the member initializers for @samp{i}
1291 and @samp{j} will be rearranged to match the declaration order of the
1295 The following @samp{-W@dots{}} options are not affected by @samp{-Wall}.
1298 @item -Weffc++ (C++ only)
1299 Warn about violations of various style guidelines from Scott Meyers'
1300 @cite{Effective C++} books. If you use this option, you should be aware
1301 that the standard library headers do not obey all of these guidelines;
1302 you can use @samp{grep -v} to filter out those warnings.
1304 @item -Wno-deprecated (C++ only)
1305 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1307 @item -Wno-non-template-friend (C++ only)
1308 Disable warnings when non-templatized friend functions are declared
1309 within a template. With the advent of explicit template specification
1310 support in g++, if the name of the friend is an unqualified-id (ie,
1311 @samp{friend foo(int)}), the C++ language specification demands that the
1312 friend declare or define an ordinary, nontemplate function. (Section
1313 14.5.3). Before g++ implemented explicit specification, unqualified-ids
1314 could be interpreted as a particular specialization of a templatized
1315 function. Because this non-conforming behavior is no longer the default
1316 behavior for g++, @samp{-Wnon-template-friend} allows the compiler to
1317 check existing code for potential trouble spots, and is on by default.
1318 This new compiler behavior can also be turned off with the flag
1319 @samp{-fguiding-decls}, which activates the older, non-specification
1320 compiler code, or with @samp{-Wno-non-template-friend} which keeps the
1321 conformant compiler code but disables the helpful warning.
1323 @item -Wold-style-cast (C++ only)
1324 Warn if an old-style (C-style) cast is used within a C++ program. The
1325 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1326 @samp{const_cast}) are less vulnerable to unintended effects.
1328 @item -Woverloaded-virtual (C++ only)
1329 @cindex overloaded virtual fn, warning
1330 @cindex warning for overloaded virtual fn
1331 Warn when a derived class function declaration may be an error in
1332 defining a virtual function. In a derived class, the
1333 definitions of virtual functions must match the type signature of a
1334 virtual function declared in the base class. With this option, the
1335 compiler warns when you define a function with the same name as a
1336 virtual function, but with a type signature that does not match any
1337 declarations from the base class.
1339 @item -Wno-pmf-conversions (C++ only)
1340 Disable the diagnostic for converting a bound pointer to member function
1343 @item -Wsign-promo (C++ only)
1344 Warn when overload resolution chooses a promotion from unsigned or
1345 enumeral type to a signed type over a conversion to an unsigned type of
1346 the same size. Previous versions of g++ would try to preserve
1347 unsignedness, but the standard mandates the current behavior.
1349 @item -Wsynth (C++ only)
1350 @cindex warning for synthesized methods
1351 @cindex synthesized methods, warning
1352 Warn when g++'s synthesis behavior does not match that of cfront. For
1358 A& operator = (int);
1368 In this example, g++ will synthesize a default @samp{A& operator =
1369 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1372 @node Warning Options
1373 @section Options to Request or Suppress Warnings
1374 @cindex options to control warnings
1375 @cindex warning messages
1376 @cindex messages, warning
1377 @cindex suppressing warnings
1379 Warnings are diagnostic messages that report constructions which
1380 are not inherently erroneous but which are risky or suggest there
1381 may have been an error.
1383 You can request many specific warnings with options beginning @samp{-W},
1384 for example @samp{-Wimplicit} to request warnings on implicit
1385 declarations. Each of these specific warning options also has a
1386 negative form beginning @samp{-Wno-} to turn off warnings;
1387 for example, @samp{-Wno-implicit}. This manual lists only one of the
1388 two forms, whichever is not the default.
1390 These options control the amount and kinds of warnings produced by GCC:
1393 @cindex syntax checking
1395 Check the code for syntax errors, but don't do anything beyond that.
1398 Issue all the warnings demanded by strict ANSI C and ISO C++;
1399 reject all programs that use forbidden extensions.
1401 Valid ANSI C and ISO C++ programs should compile properly with or without
1402 this option (though a rare few will require @samp{-ansi}). However,
1403 without this option, certain GNU extensions and traditional C and C++
1404 features are supported as well. With this option, they are rejected.
1406 @samp{-pedantic} does not cause warning messages for use of the
1407 alternate keywords whose names begin and end with @samp{__}. Pedantic
1408 warnings are also disabled in the expression that follows
1409 @code{__extension__}. However, only system header files should use
1410 these escape routes; application programs should avoid them.
1411 @xref{Alternate Keywords}.
1413 This option is not intended to be @i{useful}; it exists only to satisfy
1414 pedants who would otherwise claim that GCC fails to support the ANSI
1417 Some users try to use @samp{-pedantic} to check programs for strict ANSI
1418 C conformance. They soon find that it does not do quite what they want:
1419 it finds some non-ANSI practices, but not all---only those for which
1420 ANSI C @emph{requires} a diagnostic.
1422 A feature to report any failure to conform to ANSI C might be useful in
1423 some instances, but would require considerable additional work and would
1424 be quite different from @samp{-pedantic}. We don't have plans to
1425 support such a feature in the near future.
1427 @item -pedantic-errors
1428 Like @samp{-pedantic}, except that errors are produced rather than
1432 Inhibit all warning messages.
1435 Inhibit warning messages about the use of @samp{#import}.
1437 @item -Wchar-subscripts
1438 Warn if an array subscript has type @code{char}. This is a common cause
1439 of error, as programmers often forget that this type is signed on some
1443 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1444 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1447 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1448 the arguments supplied have types appropriate to the format string
1451 @item -Wimplicit-int
1452 Warn when a declaration does not specify a type.
1454 @item -Wimplicit-function-declaration
1455 @itemx -Werror-implicit-function-declaration
1456 Give a warning (or error) whenever a function is used before being
1460 Same as @samp{-Wimplicit-int} and @samp{-Wimplicit-function-}@*
1464 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1465 function with external linkage, returning int, taking either zero
1466 arguments, two, or three arguments of appropriate types.
1469 Warn if a multicharacter constant (@samp{'FOOF'}) is used. Usually they
1470 indicate a typo in the user's code, as they have implementation-defined
1471 values, and should not be used in portable code.
1474 Warn if parentheses are omitted in certain contexts, such
1475 as when there is an assignment in a context where a truth value
1476 is expected, or when operators are nested whose precedence people
1477 often get confused about.
1479 Also warn about constructions where there may be confusion to which
1480 @code{if} statement an @code{else} branch belongs. Here is an example of
1493 In C, every @code{else} branch belongs to the innermost possible @code{if}
1494 statement, which in this example is @code{if (b)}. This is often not
1495 what the programmer expected, as illustrated in the above example by
1496 indentation the programmer chose. When there is the potential for this
1497 confusion, GNU C will issue a warning when this flag is specified.
1498 To eliminate the warning, add explicit braces around the innermost
1499 @code{if} statement so there is no way the @code{else} could belong to
1500 the enclosing @code{if}. The resulting code would look like this:
1515 Warn whenever a function is defined with a return-type that defaults
1516 to @code{int}. Also warn about any @code{return} statement with no
1517 return-value in a function whose return-type is not @code{void}.
1520 Warn whenever a @code{switch} statement has an index of enumeral type
1521 and lacks a @code{case} for one or more of the named codes of that
1522 enumeration. (The presence of a @code{default} label prevents this
1523 warning.) @code{case} labels outside the enumeration range also
1524 provoke warnings when this option is used.
1527 Warn if any trigraphs are encountered (assuming they are enabled).
1530 Warn whenever a variable is unused aside from its declaration,
1531 whenever a function is declared static but never defined, whenever a
1532 label is declared but not used, and whenever a statement computes a
1533 result that is explicitly not used.
1535 In order to get a warning about an unused function parameter, you must
1536 specify both @samp{-W} and @samp{-Wunused}.
1538 To suppress this warning for an expression, simply cast it to void. For
1539 unused variables, parameters and labels, use the @samp{unused} attribute
1540 (@pxref{Variable Attributes}).
1542 @item -Wuninitialized
1543 Warn if an automatic variable is used without first being initialized or
1544 if a variable may be clobbered by a @code{setjmp} call.
1546 These warnings are possible only in optimizing compilation,
1547 because they require data flow information that is computed only
1548 when optimizing. If you don't specify @samp{-O}, you simply won't
1551 These warnings occur only for variables that are candidates for
1552 register allocation. Therefore, they do not occur for a variable that
1553 is declared @code{volatile}, or whose address is taken, or whose size
1554 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
1555 structures, unions or arrays, even when they are in registers.
1557 Note that there may be no warning about a variable that is used only
1558 to compute a value that itself is never used, because such
1559 computations may be deleted by data flow analysis before the warnings
1562 These warnings are made optional because GCC is not smart
1563 enough to see all the reasons why the code might be correct
1564 despite appearing to have an error. Here is one example of how
1583 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
1584 always initialized, but GCC doesn't know this. Here is
1585 another common case:
1590 if (change_y) save_y = y, y = new_y;
1592 if (change_y) y = save_y;
1597 This has no bug because @code{save_y} is used only if it is set.
1599 @cindex @code{longjmp} warnings
1600 This option also warns when a nonvolatile automatic variable might be
1601 changed by a call to @code{longjmp}. These warnings as well are possible
1602 only in optimizing compilation.
1604 The compiler sees only the calls to @code{setjmp}. It cannot know
1605 where @code{longjmp} will be called; in fact, a signal handler could
1606 call it at any point in the code. As a result, you may get a warning
1607 even when there is in fact no problem because @code{longjmp} cannot
1608 in fact be called at the place which would cause a problem.
1610 Some spurious warnings can be avoided if you declare all the functions
1611 you use that never return as @code{noreturn}. @xref{Function
1614 @item -Wreorder (C++ only)
1615 @cindex reordering, warning
1616 @cindex warning for reordering of member initializers
1617 Warn when the order of member initializers given in the code does not
1618 match the order in which they must be executed. For instance:
1620 @item -Wunknown-pragmas
1621 @cindex warning for unknown pragmas
1622 @cindex unknown pragmas, warning
1623 @cindex pragmas, warning of unknown
1624 Warn when a #pragma directive is encountered which is not understood by
1625 GCC. If this command line option is used, warnings will even be issued
1626 for unknown pragmas in system header files. This is not the case if
1627 the warnings were only enabled by the @samp{-Wall} command line option.
1630 All of the above @samp{-W} options combined. This enables all the
1631 warnings about constructions that some users consider questionable, and
1632 that are easy to avoid (or modify to prevent the warning), even in
1633 conjunction with macros.
1636 The following @samp{-W@dots{}} options are not implied by @samp{-Wall}.
1637 Some of them warn about constructions that users generally do not
1638 consider questionable, but which occasionally you might wish to check
1639 for; others warn about constructions that are necessary or hard to avoid
1640 in some cases, and there is no simple way to modify the code to suppress
1645 Print extra warning messages for these events:
1649 A function can return either with or without a value. (Falling
1650 off the end of the function body is considered returning without
1651 a value.) For example, this function would evoke such a
1665 An expression-statement or the left-hand side of a comma expression
1666 contains no side effects.
1667 To suppress the warning, cast the unused expression to void.
1668 For example, an expression such as @samp{x[i,j]} will cause a warning,
1669 but @samp{x[(void)i,j]} will not.
1672 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
1675 A comparison like @samp{x<=y<=z} appears; this is equivalent to
1676 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
1677 that of ordinary mathematical notation.
1680 Storage-class specifiers like @code{static} are not the first things in
1681 a declaration. According to the C Standard, this usage is obsolescent.
1684 If @samp{-Wall} or @samp{-Wunused} is also specified, warn about unused
1688 A comparison between signed and unsigned values could produce an
1689 incorrect result when the signed value is converted to unsigned.
1690 (But don't warn if @samp{-Wno-sign-compare} is also specified.)
1693 An aggregate has a partly bracketed initializer.
1694 For example, the following code would evoke such a warning,
1695 because braces are missing around the initializer for @code{x.h}:
1698 struct s @{ int f, g; @};
1699 struct t @{ struct s h; int i; @};
1700 struct t x = @{ 1, 2, 3 @};
1704 An aggregate has an initializer which does not initialize all members.
1705 For example, the following code would cause such a warning, because
1706 @code{x.h} would be implicitly initialized to zero:
1709 struct s @{ int f, g, h; @};
1710 struct s x = @{ 3, 4 @};
1715 Warn if floating point values are used in equality comparisons.
1717 @item -Wtraditional (C only)
1718 Warn about certain constructs that behave differently in traditional and
1723 Macro arguments occurring within string constants in the macro body.
1724 These would substitute the argument in traditional C, but are part of
1725 the constant in ANSI C.
1728 A function declared external in one block and then used after the end of
1732 A @code{switch} statement has an operand of type @code{long}.
1735 A non-@code{static} function declaration follows a @code{static} one.
1736 This construct is not accepted by some traditional C compilers.
1739 The ANSI type of an integer constant has a different width or
1740 signedness from its traditional type. This warning is only issued if
1741 the base of the constant is ten. I.e. hexadecimal or octal values, which
1742 typically represent bit patterns, are not warned about.
1745 Usage of ANSI string concatenation is detected.
1749 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
1752 Warn whenever a local variable shadows another local variable.
1754 @item -Wid-clash-@var{len}
1755 Warn whenever two distinct identifiers match in the first @var{len}
1756 characters. This may help you prepare a program that will compile
1757 with certain obsolete, brain-damaged compilers.
1759 @item -Wlarger-than-@var{len}
1760 Warn whenever an object of larger than @var{len} bytes is defined.
1762 @item -Wpointer-arith
1763 Warn about anything that depends on the ``size of'' a function type or
1764 of @code{void}. GNU C assigns these types a size of 1, for
1765 convenience in calculations with @code{void *} pointers and pointers
1768 @item -Wbad-function-cast (C only)
1769 Warn whenever a function call is cast to a non-matching type.
1770 For example, warn if @code{int malloc()} is cast to @code{anything *}.
1773 Warn whenever a pointer is cast so as to remove a type qualifier from
1774 the target type. For example, warn if a @code{const char *} is cast
1775 to an ordinary @code{char *}.
1778 Warn whenever a pointer is cast such that the required alignment of the
1779 target is increased. For example, warn if a @code{char *} is cast to
1780 an @code{int *} on machines where integers can only be accessed at
1781 two- or four-byte boundaries.
1783 @item -Wwrite-strings
1784 Give string constants the type @code{const char[@var{length}]} so that
1785 copying the address of one into a non-@code{const} @code{char *}
1786 pointer will get a warning. These warnings will help you find at
1787 compile time code that can try to write into a string constant, but
1788 only if you have been very careful about using @code{const} in
1789 declarations and prototypes. Otherwise, it will just be a nuisance;
1790 this is why we did not make @samp{-Wall} request these warnings.
1793 Warn if a prototype causes a type conversion that is different from what
1794 would happen to the same argument in the absence of a prototype. This
1795 includes conversions of fixed point to floating and vice versa, and
1796 conversions changing the width or signedness of a fixed point argument
1797 except when the same as the default promotion.
1799 Also, warn if a negative integer constant expression is implicitly
1800 converted to an unsigned type. For example, warn about the assignment
1801 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
1802 casts like @code{(unsigned) -1}.
1804 @item -Wsign-compare
1805 @cindex warning for comparison of signed and unsigned values
1806 @cindex comparison of signed and unsigned values, warning
1807 @cindex signed and unsigned values, comparison warning
1808 Warn when a comparison between signed and unsigned values could produce
1809 an incorrect result when the signed value is converted to unsigned.
1810 This warning is also enabled by @samp{-W}; to get the other warnings
1811 of @samp{-W} without this warning, use @samp{-W -Wno-sign-compare}.
1813 @item -Waggregate-return
1814 Warn if any functions that return structures or unions are defined or
1815 called. (In languages where you can return an array, this also elicits
1818 @item -Wstrict-prototypes (C only)
1819 Warn if a function is declared or defined without specifying the
1820 argument types. (An old-style function definition is permitted without
1821 a warning if preceded by a declaration which specifies the argument
1824 @item -Wmissing-prototypes (C only)
1825 Warn if a global function is defined without a previous prototype
1826 declaration. This warning is issued even if the definition itself
1827 provides a prototype. The aim is to detect global functions that fail
1828 to be declared in header files.
1830 @item -Wmissing-declarations
1831 Warn if a global function is defined without a previous declaration.
1832 Do so even if the definition itself provides a prototype.
1833 Use this option to detect global functions that are not declared in
1836 @item -Wmissing-noreturn
1837 Warn about functions which might be candidates for attribute @code{noreturn}.
1838 Note these are only possible candidates, not absolute ones. Care should
1839 be taken to manually verify functions actually do not ever return before
1840 adding the @code{noreturn} attribute, otherwise subtle code generation
1841 bugs could be introduced.
1844 Warn if a structure is given the packed attribute, but the packed
1845 attribute has no effect on the layout or size of the structure.
1846 Such structures may be mis-aligned for little benefit. For
1847 instance, in this code, the variable @code{f.x} in @code{struct bar}
1848 will be misaligned even though @code{struct bar} does not itself
1849 have the packed attribute:
1856 @} __attribute__((packed));
1865 Warn if padding is included in a structure, either to align an element
1866 of the structure or to align the whole structure. Sometimes when this
1867 happens it is possible to rearrange the fields of the structure to
1868 reduce the padding and so make the structure smaller.
1870 @item -Wredundant-decls
1871 Warn if anything is declared more than once in the same scope, even in
1872 cases where multiple declaration is valid and changes nothing.
1874 @item -Wnested-externs (C only)
1875 Warn if an @code{extern} declaration is encountered within a function.
1877 @item -Wunreachable-code
1878 Warn if the compiler detects that code will never be executed.
1880 This option is intended to warn when the compiler detects that at
1881 least a whole line of source code will never be executed, because
1882 some condition is never satisfied or because it is after a
1883 procedure that never returns.
1885 It is possible for this option to produce a warning even though there
1886 are circumstances under which part of the affected line can be executed,
1887 so care should be taken when removing apparently-unreachable code.
1889 For instance, when a function is inlined, a warning may mean that the
1890 line is unreachable in only one inlined copy of the function.
1892 This option is not made part of @samp{-Wall} because in a debugging
1893 version of a program there is often substantial code which checks
1894 correct functioning of the program and is, hopefully, unreachable
1895 because the program does work. Another common use of unreachable
1896 code is to provide behaviour which is selectable at compile-time.
1899 Warn if a function can not be inlined and it was declared as inline.
1902 Warn if @samp{long long} type is used. This is default. To inhibit
1903 the warning messages, use @samp{-Wno-long-long}. Flags
1904 @samp{-Wlong-long} and @samp{-Wno-long-long} are taken into account
1905 only when @samp{-pedantic} flag is used.
1908 Make all warnings into errors.
1911 @node Debugging Options
1912 @section Options for Debugging Your Program or GCC
1913 @cindex options, debugging
1914 @cindex debugging information options
1916 GCC has various special options that are used for debugging
1917 either your program or GCC:
1921 Produce debugging information in the operating system's native format
1922 (stabs, COFF, XCOFF, or DWARF). GDB can work with this debugging
1925 On most systems that use stabs format, @samp{-g} enables use of extra
1926 debugging information that only GDB can use; this extra information
1927 makes debugging work better in GDB but will probably make other debuggers
1929 refuse to read the program. If you want to control for certain whether
1930 to generate the extra information, use @samp{-gstabs+}, @samp{-gstabs},
1931 @samp{-gxcoff+}, @samp{-gxcoff}, @samp{-gdwarf-1+}, or @samp{-gdwarf-1}
1934 Unlike most other C compilers, GCC allows you to use @samp{-g} with
1935 @samp{-O}. The shortcuts taken by optimized code may occasionally
1936 produce surprising results: some variables you declared may not exist
1937 at all; flow of control may briefly move where you did not expect it;
1938 some statements may not be executed because they compute constant
1939 results or their values were already at hand; some statements may
1940 execute in different places because they were moved out of loops.
1942 Nevertheless it proves possible to debug optimized output. This makes
1943 it reasonable to use the optimizer for programs that might have bugs.
1945 The following options are useful when GCC is generated with the
1946 capability for more than one debugging format.
1949 Produce debugging information for use by GDB. This means to use the
1950 most expressive format available (DWARF 2, stabs, or the native format
1951 if neither of those are supported), including GDB extensions if at all
1955 Produce debugging information in stabs format (if that is supported),
1956 without GDB extensions. This is the format used by DBX on most BSD
1957 systems. On MIPS, Alpha and System V Release 4 systems this option
1958 produces stabs debugging output which is not understood by DBX or SDB.
1959 On System V Release 4 systems this option requires the GNU assembler.
1962 Produce debugging information in stabs format (if that is supported),
1963 using GNU extensions understood only by the GNU debugger (GDB). The
1964 use of these extensions is likely to make other debuggers crash or
1965 refuse to read the program.
1968 Produce debugging information in COFF format (if that is supported).
1969 This is the format used by SDB on most System V systems prior to
1973 Produce debugging information in XCOFF format (if that is supported).
1974 This is the format used by the DBX debugger on IBM RS/6000 systems.
1977 Produce debugging information in XCOFF format (if that is supported),
1978 using GNU extensions understood only by the GNU debugger (GDB). The
1979 use of these extensions is likely to make other debuggers crash or
1980 refuse to read the program, and may cause assemblers other than the GNU
1981 assembler (GAS) to fail with an error.
1984 Produce debugging information in DWARF version 1 format (if that is
1985 supported). This is the format used by SDB on most System V Release 4
1989 Produce debugging information in DWARF version 1 format (if that is
1990 supported), using GNU extensions understood only by the GNU debugger
1991 (GDB). The use of these extensions is likely to make other debuggers
1992 crash or refuse to read the program.
1995 Produce debugging information in DWARF version 2 format (if that is
1996 supported). This is the format used by DBX on IRIX 6.
1999 @itemx -ggdb@var{level}
2000 @itemx -gstabs@var{level}
2001 @itemx -gcoff@var{level}
2002 @itemx -gxcoff@var{level}
2003 @itemx -gdwarf@var{level}
2004 @itemx -gdwarf-2@var{level}
2005 Request debugging information and also use @var{level} to specify how
2006 much information. The default level is 2.
2008 Level 1 produces minimal information, enough for making backtraces in
2009 parts of the program that you don't plan to debug. This includes
2010 descriptions of functions and external variables, but no information
2011 about local variables and no line numbers.
2013 Level 3 includes extra information, such as all the macro definitions
2014 present in the program. Some debuggers support macro expansion when
2019 Generate extra code to write profile information suitable for the
2020 analysis program @code{prof}. You must use this option when compiling
2021 the source files you want data about, and you must also use it when
2024 @cindex @code{gprof}
2026 Generate extra code to write profile information suitable for the
2027 analysis program @code{gprof}. You must use this option when compiling
2028 the source files you want data about, and you must also use it when
2033 Generate extra code to write profile information for basic blocks, which will
2034 record the number of times each basic block is executed, the basic block start
2035 address, and the function name containing the basic block. If @samp{-g} is
2036 used, the line number and filename of the start of the basic block will also be
2037 recorded. If not overridden by the machine description, the default action is
2038 to append to the text file @file{bb.out}.
2040 This data could be analyzed by a program like @code{tcov}. Note,
2041 however, that the format of the data is not what @code{tcov} expects.
2042 Eventually GNU @code{gprof} should be extended to process this data.
2045 Makes the compiler print out each function name as it is compiled, and
2046 print some statistics about each pass when it finishes.
2049 Generate extra code to profile basic blocks. Your executable will
2050 produce output that is a superset of that produced when @samp{-a} is
2051 used. Additional output is the source and target address of the basic
2052 blocks where a jump takes place, the number of times a jump is executed,
2053 and (optionally) the complete sequence of basic blocks being executed.
2054 The output is appended to file @file{bb.out}.
2056 You can examine different profiling aspects without recompilation. Your
2057 executable will read a list of function names from file @file{bb.in}.
2058 Profiling starts when a function on the list is entered and stops when
2059 that invocation is exited. To exclude a function from profiling, prefix
2060 its name with `-'. If a function name is not unique, you can
2061 disambiguate it by writing it in the form
2062 @samp{/path/filename.d:functionname}. Your executable will write the
2063 available paths and filenames in file @file{bb.out}.
2065 Several function names have a special meaning:
2068 Write source, target and frequency of jumps to file @file{bb.out}.
2069 @item __bb_hidecall__
2070 Exclude function calls from frequency count.
2071 @item __bb_showret__
2072 Include function returns in frequency count.
2074 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
2075 The file will be compressed using the program @samp{gzip}, which must
2076 exist in your @code{PATH}. On systems without the @samp{popen}
2077 function, the file will be named @file{bbtrace} and will not be
2078 compressed. @strong{Profiling for even a few seconds on these systems
2079 will produce a very large file.} Note: @code{__bb_hidecall__} and
2080 @code{__bb_showret__} will not affect the sequence written to
2084 Here's a short example using different profiling parameters
2085 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
2086 1 and 2 and is called twice from block 3 of function @code{main}. After
2087 the calls, block 3 transfers control to block 4 of @code{main}.
2089 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
2090 the following sequence of blocks is written to file @file{bbtrace.gz}:
2091 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
2092 the return is to a point inside the block and not to the top. The
2093 block address 0 always indicates, that control is transferred
2094 to the trace from somewhere outside the observed functions. With
2095 @samp{-foo} added to @file{bb.in}, the blocks of function
2096 @code{foo} are removed from the trace, so only 0 3 4 remains.
2098 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
2099 jump frequencies will be written to file @file{bb.out}. The
2100 frequencies are obtained by constructing a trace of blocks
2101 and incrementing a counter for every neighbouring pair of blocks
2102 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2106 Jump from block 0x0 to block 0x3 executed 1 time(s)
2107 Jump from block 0x3 to block 0x1 executed 1 time(s)
2108 Jump from block 0x1 to block 0x2 executed 2 time(s)
2109 Jump from block 0x2 to block 0x1 executed 1 time(s)
2110 Jump from block 0x2 to block 0x4 executed 1 time(s)
2113 With @code{__bb_hidecall__}, control transfer due to call instructions
2114 is removed from the trace, that is the trace is cut into three parts: 0
2115 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2116 to return instructions is added to the trace. The trace becomes: 0 3 1
2117 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2118 written to @file{bbtrace.gz}. It is solely used for counting jump
2121 @item -fprofile-arcs
2122 Instrument @dfn{arcs} during compilation. For each function of your
2123 program, GCC creates a program flow graph, then finds a spanning tree
2124 for the graph. Only arcs that are not on the spanning tree have to be
2125 instrumented: the compiler adds code to count the number of times that these
2126 arcs are executed. When an arc is the only exit or only entrance to a
2127 block, the instrumentation code can be added to the block; otherwise, a
2128 new basic block must be created to hold the instrumentation code.
2130 Since not every arc in the program must be instrumented, programs
2131 compiled with this option run faster than programs compiled with
2132 @samp{-a}, which adds instrumentation code to every basic block in the
2133 program. The tradeoff: since @code{gcov} does not have
2134 execution counts for all branches, it must start with the execution
2135 counts for the instrumented branches, and then iterate over the program
2136 flow graph until the entire graph has been solved. Hence, @code{gcov}
2137 runs a little more slowly than a program which uses information from
2140 @samp{-fprofile-arcs} also makes it possible to estimate branch
2141 probabilities, and to calculate basic block execution counts. In
2142 general, basic block execution counts do not give enough information to
2143 estimate all branch probabilities. When the compiled program exits, it
2144 saves the arc execution counts to a file called
2145 @file{@var{sourcename}.da}. Use the compiler option
2146 @samp{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2147 Control Optimization}) when recompiling, to optimize using estimated
2148 branch probabilities.
2151 @item -ftest-coverage
2152 Create data files for the @code{gcov} code-coverage utility
2153 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2154 The data file names begin with the name of your source file:
2157 @item @var{sourcename}.bb
2158 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2159 associate basic block execution counts with line numbers.
2161 @item @var{sourcename}.bbg
2162 A list of all arcs in the program flow graph. This allows @code{gcov}
2163 to reconstruct the program flow graph, so that it can compute all basic
2164 block and arc execution counts from the information in the
2165 @code{@var{sourcename}.da} file (this last file is the output from
2166 @samp{-fprofile-arcs}).
2169 @item -d@var{letters}
2170 Says to make debugging dumps during compilation at times specified by
2171 @var{letters}. This is used for debugging the compiler. The file names
2172 for most of the dumps are made by appending a pass number and a word to
2173 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.jump}).
2174 Here are the possible letters for use in @var{letters}, and their meanings:
2178 Annotate the assembler output with miscellaneous debugging information.
2180 Dump after computing branch probabilities, to @file{@var{file}.07.bp}.
2182 Dump after instruction combination, to the file @file{@var{file}.09.combine}.
2184 Dump after delayed branch scheduling, to @file{@var{file}.19.dbr}.
2186 Dump all macro definitions, at the end of preprocessing, in addition to
2189 Dump after purging ADDRESSOF, to @file{@var{file}.03.addressof}.
2191 Dump after flow analysis, to @file{@var{file}.08.flow}.
2193 Dump after global register allocation, to @file{@var{file}.13.greg}.
2195 Dump after GCSE, to @file{@var{file}.04.gcse}.
2197 Dump after first jump optimization, to @file{@var{file}.01.jump}.
2199 Dump after last jump optimization, to @file{@var{file}.17.jump2}.
2201 Dump after conversion from registers to stack, to @file{@var{file}.20.stack}.
2203 Dump after local register allocation, to @file{@var{file}.12.lreg}.
2205 Dump after loop optimization, to @file{@var{file}.05.loop}.
2207 Dump after performing the machine dependent reorganisation pass, to
2208 @file{@var{file}.18.mach}.
2210 Dump after the register move pass, to @file{@var{file}.10.regmove}.
2212 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2214 Dump after the second instruction scheduling pass, to
2215 @file{@var{file}.16.sched2}.
2217 Dump after CSE (including the jump optimization that sometimes follows
2218 CSE), to @file{@var{file}.02.cse}.
2220 Dump after the first instruction scheduling pass, to
2221 @file{@var{file}.11.sched}.
2223 Dump after the second CSE pass (including the jump optimization that
2224 sometimes follows CSE), to @file{@var{file}.06.cse2}.
2226 Produce all the dumps listed above.
2228 Print statistics on memory usage, at the end of the run, to
2231 Annotate the assembler output with a comment indicating which
2232 pattern and alternative was used. The length of each instruction is
2235 For each of the other indicated dump files (except for
2236 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
2237 suitible for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
2239 Dump after the second flow pass to @file{@var{file}.14.flow2}.
2241 Just generate RTL for a function instead of compiling it. Usually used
2244 Dump debugging information during parsing, to standard error.
2246 Dump after the peephole2 pass to @file{@var{file}.15.peephole2}.
2249 @item -fdump-unnumbered
2250 When doing debugging dumps (see -d option above), suppress instruction
2251 numbers and line number note output. This makes it more feasible to
2252 use diff on debugging dumps for compiler invokations with different
2253 options, in particular with and without -g.
2255 @item -fdump-translation-unit-@var{file} (C++ only)
2256 Dump a representation of the tree structure for the entire translation
2259 @item -fpretend-float
2260 When running a cross-compiler, pretend that the target machine uses the
2261 same floating point format as the host machine. This causes incorrect
2262 output of the actual floating constants, but the actual instruction
2263 sequence will probably be the same as GCC would make when running on
2267 Store the usual ``temporary'' intermediate files permanently; place them
2268 in the current directory and name them based on the source file. Thus,
2269 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
2270 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}.
2273 Report the CPU time taken by each subprocess in the compilation
2274 sequence. For C source files, this is the preprocessor, compiler
2275 proper, and assembler. The output looks like this:
2283 The first number on each line is the ``user time,'' that is time spent
2284 executing the program itself. The second number is ``system time,''
2285 time spent executing operating system routines on behalf of the program.
2286 Both numbers are in seconds.
2288 @item -print-file-name=@var{library}
2289 Print the full absolute name of the library file @var{library} that
2290 would be used when linking---and don't do anything else. With this
2291 option, GCC does not compile or link anything; it just prints the
2294 @item -print-prog-name=@var{program}
2295 Like @samp{-print-file-name}, but searches for a program such as @samp{cpp}.
2297 @item -print-libgcc-file-name
2298 Same as @samp{-print-file-name=libgcc.a}.
2300 This is useful when you use @samp{-nostdlib} or @samp{-nodefaultlibs}
2301 but you do want to link with @file{libgcc.a}. You can do
2304 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
2307 @item -print-search-dirs
2308 Print the name of the configured installation directory and a list of
2309 program and library directories gcc will search---and don't do anything else.
2311 This is useful when gcc prints the error message
2312 @samp{installation problem, cannot exec cpp: No such file or directory}.
2313 To resolve this you either need to put @file{cpp} and the other compiler
2314 components where gcc expects to find them, or you can set the environment
2315 variable @code{GCC_EXEC_PREFIX} to the directory where you installed them.
2316 Don't forget the trailing '/'.
2317 @xref{Environment Variables}.
2320 @node Optimize Options
2321 @section Options That Control Optimization
2322 @cindex optimize options
2323 @cindex options, optimization
2325 These options control various sorts of optimizations:
2330 Optimize. Optimizing compilation takes somewhat more time, and a lot
2331 more memory for a large function.
2333 Without @samp{-O}, the compiler's goal is to reduce the cost of
2334 compilation and to make debugging produce the expected results.
2335 Statements are independent: if you stop the program with a breakpoint
2336 between statements, you can then assign a new value to any variable or
2337 change the program counter to any other statement in the function and
2338 get exactly the results you would expect from the source code.
2340 Without @samp{-O}, the compiler only allocates variables declared
2341 @code{register} in registers. The resulting compiled code is a little
2342 worse than produced by PCC without @samp{-O}.
2344 With @samp{-O}, the compiler tries to reduce code size and execution
2347 When you specify @samp{-O}, the compiler turns on @samp{-fthread-jumps}
2348 and @samp{-fdefer-pop} on all machines. The compiler turns on
2349 @samp{-fdelayed-branch} on machines that have delay slots, and
2350 @samp{-fomit-frame-pointer} on machines that can support debugging even
2351 without a frame pointer. On some machines the compiler also turns
2352 on other flags.@refill
2355 Optimize even more. GCC performs nearly all supported optimizations
2356 that do not involve a space-speed tradeoff. The compiler does not
2357 perform loop unrolling or function inlining when you specify @samp{-O2}.
2358 As compared to @samp{-O}, this option increases both compilation time
2359 and the performance of the generated code.
2361 @samp{-O2} turns on all optional optimizations except for loop unrolling
2362 and function inlining. It also turns on the @samp{-fforce-mem} option
2363 on all machines and frame pointer elimination on machines where doing so
2364 does not interfere with debugging.
2367 Optimize yet more. @samp{-O3} turns on all optimizations specified by
2368 @samp{-O2} and also turns on the @samp{inline-functions} option.
2374 Optimize for size. @samp{-Os} enables all @samp{-O2} optimizations that
2375 do not typically increase code size. It also performs further
2376 optimizations designed to reduce code size.
2378 If you use multiple @samp{-O} options, with or without level numbers,
2379 the last such option is the one that is effective.
2382 Options of the form @samp{-f@var{flag}} specify machine-independent
2383 flags. Most flags have both positive and negative forms; the negative
2384 form of @samp{-ffoo} would be @samp{-fno-foo}. In the table below,
2385 only one of the forms is listed---the one which is not the default.
2386 You can figure out the other form by either removing @samp{no-} or
2391 Do not store floating point variables in registers, and inhibit other
2392 options that might change whether a floating point value is taken from a
2395 @cindex floating point precision
2396 This option prevents undesirable excess precision on machines such as
2397 the 68000 where the floating registers (of the 68881) keep more
2398 precision than a @code{double} is supposed to have. Similarly for the
2399 x86 architecture. For most programs, the excess precision does only
2400 good, but a few programs rely on the precise definition of IEEE floating
2401 point. Use @samp{-ffloat-store} for such programs, after modifying
2402 them to store all pertinent intermediate computations into variables.
2404 @item -fno-default-inline
2405 Do not make member functions inline by default merely because they are
2406 defined inside the class scope (C++ only). Otherwise, when you specify
2407 @w{@samp{-O}}, member functions defined inside class scope are compiled
2408 inline by default; i.e., you don't need to add @samp{inline} in front of
2409 the member function name.
2411 @item -fno-defer-pop
2412 Always pop the arguments to each function call as soon as that function
2413 returns. For machines which must pop arguments after a function call,
2414 the compiler normally lets arguments accumulate on the stack for several
2415 function calls and pops them all at once.
2418 Force memory operands to be copied into registers before doing
2419 arithmetic on them. This produces better code by making all memory
2420 references potential common subexpressions. When they are not common
2421 subexpressions, instruction combination should eliminate the separate
2422 register-load. The @samp{-O2} option turns on this option.
2425 Force memory address constants to be copied into registers before
2426 doing arithmetic on them. This may produce better code just as
2427 @samp{-fforce-mem} may.
2429 @item -fomit-frame-pointer
2430 Don't keep the frame pointer in a register for functions that
2431 don't need one. This avoids the instructions to save, set up and
2432 restore frame pointers; it also makes an extra register available
2433 in many functions. @strong{It also makes debugging impossible on
2437 On some machines, such as the Vax, this flag has no effect, because
2438 the standard calling sequence automatically handles the frame pointer
2439 and nothing is saved by pretending it doesn't exist. The
2440 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2441 whether a target machine supports this flag. @xref{Registers}.@refill
2444 On some machines, such as the Vax, this flag has no effect, because
2445 the standard calling sequence automatically handles the frame pointer
2446 and nothing is saved by pretending it doesn't exist. The
2447 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2448 whether a target machine supports this flag. @xref{Registers,,Register
2449 Usage, gcc.info, Using and Porting GCC}.@refill
2453 Don't pay attention to the @code{inline} keyword. Normally this option
2454 is used to keep the compiler from expanding any functions inline.
2455 Note that if you are not optimizing, no functions can be expanded inline.
2457 @item -finline-functions
2458 Integrate all simple functions into their callers. The compiler
2459 heuristically decides which functions are simple enough to be worth
2460 integrating in this way.
2462 If all calls to a given function are integrated, and the function is
2463 declared @code{static}, then the function is normally not output as
2464 assembler code in its own right.
2466 @item -finline-limit=@var{n}
2467 By default, gcc limits the size of functions that can be inlined. This flag
2468 allows the control of this limit for functions that are explicitly marked as
2469 inline (ie marked with the inline keyword or defined within the class
2470 definition in c++). @var{n} is the size of functions that can be inlined in
2471 number of pseudo instructions (not counting parameter handling). The default
2472 value of n is 10000. Increasing this value can result in more inlined code at
2473 the cost of compilation time and memory consumption. Decreasing usually makes
2474 the compilation faster and less code will be inlined (which presumably
2475 means slower programs). This option is particularly useful for programs that
2476 use inlining heavily such as those based on recursive templates with c++.
2478 @emph{Note:} pseudo instruction represents, in this particular context, an
2479 abstract measurement of function's size. In no way, it represents a count
2480 of assembly instructions and as such its exact meaning might change from one
2481 release to an another.
2483 @item -fkeep-inline-functions
2484 Even if all calls to a given function are integrated, and the function
2485 is declared @code{static}, nevertheless output a separate run-time
2486 callable version of the function. This switch does not affect
2487 @code{extern inline} functions.
2489 @item -fkeep-static-consts
2490 Emit variables declared @code{static const} when optimization isn't turned
2491 on, even if the variables aren't referenced.
2493 GCC enables this option by default. If you want to force the compiler to
2494 check if the variable was referenced, regardless of whether or not
2495 optimization is turned on, use the @samp{-fno-keep-static-consts} option.
2497 @item -fno-function-cse
2498 Do not put function addresses in registers; make each instruction that
2499 calls a constant function contain the function's address explicitly.
2501 This option results in less efficient code, but some strange hacks
2502 that alter the assembler output may be confused by the optimizations
2503 performed when this option is not used.
2506 This option allows GCC to violate some ANSI or IEEE rules and/or
2507 specifications in the interest of optimizing code for speed. For
2508 example, it allows the compiler to assume arguments to the @code{sqrt}
2509 function are non-negative numbers and that no floating-point values
2512 This option should never be turned on by any @samp{-O} option since
2513 it can result in incorrect output for programs which depend on
2514 an exact implementation of IEEE or ANSI rules/specifications for
2517 @item -fno-math-errno
2518 Do not set ERRNO after calling math functions that are executed
2519 with a single instruction, e.g., sqrt. A program that relies on
2520 IEEE exceptions for math error handling may want to use this flag
2521 for speed while maintaining IEEE arithmetic compatibility.
2523 The default is @samp{-fmath-errno}. The @samp{-ffast-math} option
2524 sets @samp{-fno-math-errno}.
2527 @c following causes underfulls.. they don't look great, but we deal.
2529 The following options control specific optimizations. The @samp{-O2}
2530 option turns on all of these optimizations except @samp{-funroll-loops}
2531 and @samp{-funroll-all-loops}. On most machines, the @samp{-O} option
2532 turns on the @samp{-fthread-jumps} and @samp{-fdelayed-branch} options,
2533 but specific machines may handle it differently.
2535 You can use the following flags in the rare cases when ``fine-tuning''
2536 of optimizations to be performed is desired.
2539 @item -fstrength-reduce
2540 Perform the optimizations of loop strength reduction and
2541 elimination of iteration variables.
2543 @item -fthread-jumps
2544 Perform optimizations where we check to see if a jump branches to a
2545 location where another comparison subsumed by the first is found. If
2546 so, the first branch is redirected to either the destination of the
2547 second branch or a point immediately following it, depending on whether
2548 the condition is known to be true or false.
2550 @item -fcse-follow-jumps
2551 In common subexpression elimination, scan through jump instructions
2552 when the target of the jump is not reached by any other path. For
2553 example, when CSE encounters an @code{if} statement with an
2554 @code{else} clause, CSE will follow the jump when the condition
2557 @item -fcse-skip-blocks
2558 This is similar to @samp{-fcse-follow-jumps}, but causes CSE to
2559 follow jumps which conditionally skip over blocks. When CSE
2560 encounters a simple @code{if} statement with no else clause,
2561 @samp{-fcse-skip-blocks} causes CSE to follow the jump around the
2562 body of the @code{if}.
2564 @item -frerun-cse-after-loop
2565 Re-run common subexpression elimination after loop optimizations has been
2568 @item -frerun-loop-opt
2569 Run the loop optimizer twice.
2572 Perform a global common subexpression elimination pass.
2573 This pass also performs global constant and copy propagation.
2575 @item -fdelete-null-pointer-checks
2576 Use global dataflow analysis to identify and eliminate useless null
2577 pointer checks. Programs which rely on NULL pointer dereferences @emph{not}
2578 halting the program may not work properly with this option. Use
2579 -fno-delete-null-pointer-checks to disable this optimizing for programs
2580 which depend on that behavior.
2583 @item -fexpensive-optimizations
2584 Perform a number of minor optimizations that are relatively expensive.
2586 @item -foptimize-register-moves
2588 Attempt to reassign register numbers in move instructions and as
2589 operands of other simple instructions in order to maximize the amount of
2590 register tying. This is especially helpful on machines with two-operand
2591 instructions. GCC enables this optimization by default with @samp{-O2}
2594 Note @code{-fregmove} and @code{-foptimize-register-moves} are the same
2597 @item -fdelayed-branch
2598 If supported for the target machine, attempt to reorder instructions
2599 to exploit instruction slots available after delayed branch
2602 @item -fschedule-insns
2603 If supported for the target machine, attempt to reorder instructions to
2604 eliminate execution stalls due to required data being unavailable. This
2605 helps machines that have slow floating point or memory load instructions
2606 by allowing other instructions to be issued until the result of the load
2607 or floating point instruction is required.
2609 @item -fschedule-insns2
2610 Similar to @samp{-fschedule-insns}, but requests an additional pass of
2611 instruction scheduling after register allocation has been done. This is
2612 especially useful on machines with a relatively small number of
2613 registers and where memory load instructions take more than one cycle.
2615 @item -ffunction-sections
2616 @itemx -fdata-sections
2617 Place each function or data item into its own section in the output
2618 file if the target supports arbitrary sections. The name of the
2619 function or the name of the data item determines the section's name
2622 Use these options on systems where the linker can perform optimizations
2623 to improve locality of reference in the instruction space. HPPA
2624 processors running HP-UX and Sparc processors running Solaris 2 have
2625 linkers with such optimizations. Other systems using the ELF object format
2626 as well as AIX may have these optimizations in the future.
2628 Only use these options when there are significant benefits from doing
2629 so. When you specify these options, the assembler and linker will
2630 create larger object and executable files and will also be slower.
2631 You will not be able to use @code{gprof} on all systems if you
2632 specify this option and you may have problems with debugging if
2633 you specify both this option and @samp{-g}.
2635 @item -fcaller-saves
2636 Enable values to be allocated in registers that will be clobbered by
2637 function calls, by emitting extra instructions to save and restore the
2638 registers around such calls. Such allocation is done only when it
2639 seems to result in better code than would otherwise be produced.
2641 This option is always enabled by default on certain machines, usually
2642 those which have no call-preserved registers to use instead.
2644 For all machines, optimization level 2 and higher enables this flag by
2647 @item -funroll-loops
2648 Perform the optimization of loop unrolling. This is only done for loops
2649 whose number of iterations can be determined at compile time or run time.
2650 @samp{-funroll-loops} implies both @samp{-fstrength-reduce} and
2651 @samp{-frerun-cse-after-loop}.
2653 @item -funroll-all-loops
2654 Perform the optimization of loop unrolling. This is done for all loops
2655 and usually makes programs run more slowly. @samp{-funroll-all-loops}
2656 implies @samp{-fstrength-reduce} as well as @samp{-frerun-cse-after-loop}.
2658 @item -fmove-all-movables
2659 Forces all invariant computations in loops to be moved
2662 @item -freduce-all-givs
2663 Forces all general-induction variables in loops to be
2666 @emph{Note:} When compiling programs written in Fortran,
2667 @samp{-fmove-all-movables} and @samp{-freduce-all-givs} are enabled
2668 by default when you use the optimizer.
2670 These options may generate better or worse code; results are highly
2671 dependent on the structure of loops within the source code.
2673 These two options are intended to be removed someday, once
2674 they have helped determine the efficacy of various
2675 approaches to improving loop optimizations.
2677 Please let us (@code{gcc@@gcc.gnu.org} and @code{fortran@@gnu.org})
2678 know how use of these options affects
2679 the performance of your production code.
2680 We're very interested in code that runs @emph{slower}
2681 when these options are @emph{enabled}.
2684 Disable any machine-specific peephole optimizations.
2686 @item -fbranch-probabilities
2687 After running a program compiled with @samp{-fprofile-arcs}
2688 (@pxref{Debugging Options,, Options for Debugging Your Program or
2689 @code{gcc}}), you can compile it a second time using
2690 @samp{-fbranch-probabilities}, to improve optimizations based on
2691 guessing the path a branch might take.
2694 With @samp{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
2695 note on the first instruction of each basic block, and a
2696 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
2697 These can be used to improve optimization. Currently, they are only
2698 used in one place: in @file{reorg.c}, instead of guessing which path a
2699 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
2700 exactly determine which path is taken more often.
2703 @item -fstrict-aliasing
2704 Allows the compiler to assume the strictest aliasing rules applicable to
2705 the language being compiled. For C (and C++), this activates
2706 optimizations based on the type of expressions. In particular, an
2707 object of one type is assumed never to reside at the same address as an
2708 object of a different type, unless the types are almost the same. For
2709 example, an @code{unsigned int} can alias an @code{int}, but not a
2710 @code{void*} or a @code{double}. A character type may alias any other
2713 Pay special attention to code like this:
2726 The practice of reading from a different union member than the one most
2727 recently written to (called ``type-punning'') is common. Even with
2728 @samp{-fstrict-aliasing}, type-punning is allowed, provided the memory
2729 is accessed through the union type. So, the code above will work as
2730 expected. However, this code might not:
2742 Every language that wishes to perform language-specific alias analysis
2743 should define a function that computes, given an @code{tree}
2744 node, an alias set for the node. Nodes in different alias sets are not
2745 allowed to alias. For an example, see the C front-end function
2746 @code{c_get_alias_set}.
2749 @item -falign-functions
2750 @itemx -falign-functions=@var{n}
2751 Align the start of functions to the next power-of-two greater than
2752 @var{n}, skipping up to @var{n} bytes. For instance,
2753 @samp{-falign-functions=32} aligns functions to the next 32-byte
2754 boundary, but @samp{-falign-functions=24} would align to the next
2755 32-byte boundary only if this can be done by skipping 23 bytes or less.
2757 @samp{-fno-align-functions} and @samp{-falign-functions=1} are
2758 equivalent and mean that functions will not be aligned.
2760 Some assemblers only support this flag when @var{n} is a power of two;
2761 in that case, it is rounded up.
2763 If @var{n} is not specified, use a machine-dependent default.
2765 @item -falign-labels
2766 @itemx -falign-labels=@var{n}
2767 Align all branch targets to a power-of-two boundary, skipping up to
2768 @var{n} bytes like @samp{-falign-functions}. This option can easily
2769 make code slower, because it must insert dummy operations for when the
2770 branch target is reached in the usual flow of the code.
2772 If @samp{-falign-loops} or @samp{-falign-jumps} are applicable and
2773 are greater than this value, then their values are used instead.
2775 If @var{n} is not specified, use a machine-dependent default which is
2776 very likely to be @samp{1}, meaning no alignment.
2779 @itemx -falign-loops=@var{n}
2780 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
2781 like @samp{-falign-functions}. The hope is that the loop will be
2782 executed many times, which will make up for any execution of the dummy
2785 If @var{n} is not specified, use a machine-dependent default.
2788 @itemx -falign-jumps=@var{n}
2789 Align branch targets to a power-of-two boundary, for branch targets
2790 where the targets can only be reached by jumping, skipping up to @var{n}
2791 bytes like @samp{-falign-functions}. In this case, no dummy operations
2794 If @var{n} is not specified, use a machine-dependent default.
2798 @node Preprocessor Options
2799 @section Options Controlling the Preprocessor
2800 @cindex preprocessor options
2801 @cindex options, preprocessor
2803 These options control the C preprocessor, which is run on each C source
2804 file before actual compilation.
2806 If you use the @samp{-E} option, nothing is done except preprocessing.
2807 Some of these options make sense only together with @samp{-E} because
2808 they cause the preprocessor output to be unsuitable for actual
2812 @item -include @var{file}
2813 Process @var{file} as input before processing the regular input file.
2814 In effect, the contents of @var{file} are compiled first. Any @samp{-D}
2815 and @samp{-U} options on the command line are always processed before
2816 @samp{-include @var{file}}, regardless of the order in which they are
2817 written. All the @samp{-include} and @samp{-imacros} options are
2818 processed in the order in which they are written.
2820 @item -imacros @var{file}
2821 Process @var{file} as input, discarding the resulting output, before
2822 processing the regular input file. Because the output generated from
2823 @var{file} is discarded, the only effect of @samp{-imacros @var{file}}
2824 is to make the macros defined in @var{file} available for use in the
2827 Any @samp{-D} and @samp{-U} options on the command line are always
2828 processed before @samp{-imacros @var{file}}, regardless of the order in
2829 which they are written. All the @samp{-include} and @samp{-imacros}
2830 options are processed in the order in which they are written.
2832 @item -idirafter @var{dir}
2833 @cindex second include path
2834 Add the directory @var{dir} to the second include path. The directories
2835 on the second include path are searched when a header file is not found
2836 in any of the directories in the main include path (the one that
2839 @item -iprefix @var{prefix}
2840 Specify @var{prefix} as the prefix for subsequent @samp{-iwithprefix}
2843 @item -iwithprefix @var{dir}
2844 Add a directory to the second include path. The directory's name is
2845 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
2846 specified previously with @samp{-iprefix}. If you have not specified a
2847 prefix yet, the directory containing the installed passes of the
2848 compiler is used as the default.
2850 @item -iwithprefixbefore @var{dir}
2851 Add a directory to the main include path. The directory's name is made
2852 by concatenating @var{prefix} and @var{dir}, as in the case of
2853 @samp{-iwithprefix}.
2855 @item -isystem @var{dir}
2856 Add a directory to the beginning of the second include path, marking it
2857 as a system directory, so that it gets the same special treatment as
2858 is applied to the standard system directories.
2861 Do not search the standard system directories for header files. Only
2862 the directories you have specified with @samp{-I} options (and the
2863 current directory, if appropriate) are searched. @xref{Directory
2864 Options}, for information on @samp{-I}.
2866 By using both @samp{-nostdinc} and @samp{-I-}, you can limit the include-file
2867 search path to only those directories you specify explicitly.
2870 Do not predefine any nonstandard macros. (Including architecture flags).
2873 Run only the C preprocessor. Preprocess all the C source files
2874 specified and output the results to standard output or to the
2875 specified output file.
2878 Tell the preprocessor not to discard comments. Used with the
2882 Tell the preprocessor not to generate @samp{#line} directives.
2883 Used with the @samp{-E} option.
2886 @cindex dependencies, make
2888 Tell the preprocessor to output a rule suitable for @code{make}
2889 describing the dependencies of each object file. For each source file,
2890 the preprocessor outputs one @code{make}-rule whose target is the object
2891 file name for that source file and whose dependencies are all the
2892 @code{#include} header files it uses. This rule may be a single line or
2893 may be continued with @samp{\}-newline if it is long. The list of rules
2894 is printed on standard output instead of the preprocessed C program.
2896 @samp{-M} implies @samp{-E}.
2898 Another way to specify output of a @code{make} rule is by setting
2899 the environment variable @code{DEPENDENCIES_OUTPUT} (@pxref{Environment
2903 Like @samp{-M} but the output mentions only the user header files
2904 included with @samp{#include "@var{file}"}. System header files
2905 included with @samp{#include <@var{file}>} are omitted.
2908 Like @samp{-M} but the dependency information is written to a file made by
2909 replacing ".c" with ".d" at the end of the input file names.
2910 This is in addition to compiling the file as specified---@samp{-MD} does
2911 not inhibit ordinary compilation the way @samp{-M} does.
2913 In Mach, you can use the utility @code{md} to merge multiple dependency
2914 files into a single dependency file suitable for using with the @samp{make}
2918 Like @samp{-MD} except mention only user header files, not system
2922 Treat missing header files as generated files and assume they live in the
2923 same directory as the source file. If you specify @samp{-MG}, you
2924 must also specify either @samp{-M} or @samp{-MM}. @samp{-MG} is not
2925 supported with @samp{-MD} or @samp{-MMD}.
2928 Print the name of each header file used, in addition to other normal
2931 @item -A@var{question}(@var{answer})
2932 Assert the answer @var{answer} for @var{question}, in case it is tested
2933 with a preprocessing conditional such as @samp{#if
2934 #@var{question}(@var{answer})}. @samp{-A-} disables the standard
2935 assertions that normally describe the target machine.
2938 Define macro @var{macro} with the string @samp{1} as its definition.
2940 @item -D@var{macro}=@var{defn}
2941 Define macro @var{macro} as @var{defn}. All instances of @samp{-D} on
2942 the command line are processed before any @samp{-U} options.
2945 Undefine macro @var{macro}. @samp{-U} options are evaluated after all
2946 @samp{-D} options, but before any @samp{-include} and @samp{-imacros}
2950 Tell the preprocessor to output only a list of the macro definitions
2951 that are in effect at the end of preprocessing. Used with the @samp{-E}
2955 Tell the preprocessing to pass all macro definitions into the output, in
2956 their proper sequence in the rest of the output.
2959 Like @samp{-dD} except that the macro arguments and contents are omitted.
2960 Only @samp{#define @var{name}} is included in the output.
2963 Support ANSI C trigraphs. The @samp{-ansi} option also has this effect.
2965 @item -Wp,@var{option}
2966 Pass @var{option} as an option to the preprocessor. If @var{option}
2967 contains commas, it is split into multiple options at the commas.
2970 @node Assembler Options
2971 @section Passing Options to the Assembler
2973 @c prevent bad page break with this line
2974 You can pass options to the assembler.
2977 @item -Wa,@var{option}
2978 Pass @var{option} as an option to the assembler. If @var{option}
2979 contains commas, it is split into multiple options at the commas.
2983 @section Options for Linking
2984 @cindex link options
2985 @cindex options, linking
2987 These options come into play when the compiler links object files into
2988 an executable output file. They are meaningless if the compiler is
2989 not doing a link step.
2993 @item @var{object-file-name}
2994 A file name that does not end in a special recognized suffix is
2995 considered to name an object file or library. (Object files are
2996 distinguished from libraries by the linker according to the file
2997 contents.) If linking is done, these object files are used as input
3003 If any of these options is used, then the linker is not run, and
3004 object file names should not be used as arguments. @xref{Overall
3008 @item -l@var{library}
3009 Search the library named @var{library} when linking.
3011 It makes a difference where in the command you write this option; the
3012 linker searches processes libraries and object files in the order they
3013 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
3014 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
3015 to functions in @samp{z}, those functions may not be loaded.
3017 The linker searches a standard list of directories for the library,
3018 which is actually a file named @file{lib@var{library}.a}. The linker
3019 then uses this file as if it had been specified precisely by name.
3021 The directories searched include several standard system directories
3022 plus any that you specify with @samp{-L}.
3024 Normally the files found this way are library files---archive files
3025 whose members are object files. The linker handles an archive file by
3026 scanning through it for members which define symbols that have so far
3027 been referenced but not defined. But if the file that is found is an
3028 ordinary object file, it is linked in the usual fashion. The only
3029 difference between using an @samp{-l} option and specifying a file name
3030 is that @samp{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
3031 and searches several directories.
3034 You need this special case of the @samp{-l} option in order to
3035 link an Objective C program.
3038 Do not use the standard system startup files when linking.
3039 The standard system libraries are used normally, unless @code{-nostdlib}
3040 or @code{-nodefaultlibs} is used.
3042 @item -nodefaultlibs
3043 Do not use the standard system libraries when linking.
3044 Only the libraries you specify will be passed to the linker.
3045 The standard startup files are used normally, unless @code{-nostartfiles}
3046 is used. The compiler may generate calls to memcmp, memset, and memcpy
3047 for System V (and ANSI C) environments or to bcopy and bzero for
3048 BSD environments. These entries are usually resolved by entries in
3049 libc. These entry points should be supplied through some other
3050 mechanism when this option is specified.
3053 Do not use the standard system startup files or libraries when linking.
3054 No startup files and only the libraries you specify will be passed to
3055 the linker. The compiler may generate calls to memcmp, memset, and memcpy
3056 for System V (and ANSI C) environments or to bcopy and bzero for
3057 BSD environments. These entries are usually resolved by entries in
3058 libc. These entry points should be supplied through some other
3059 mechanism when this option is specified.
3061 @cindex @code{-lgcc}, use with @code{-nostdlib}
3062 @cindex @code{-nostdlib} and unresolved references
3063 @cindex unresolved references and @code{-nostdlib}
3064 @cindex @code{-lgcc}, use with @code{-nodefaultlibs}
3065 @cindex @code{-nodefaultlibs} and unresolved references
3066 @cindex unresolved references and @code{-nodefaultlibs}
3067 One of the standard libraries bypassed by @samp{-nostdlib} and
3068 @samp{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
3069 that GCC uses to overcome shortcomings of particular machines, or special
3070 needs for some languages.
3072 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
3076 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
3077 for more discussion of @file{libgcc.a}.)
3079 In most cases, you need @file{libgcc.a} even when you want to avoid
3080 other standard libraries. In other words, when you specify @samp{-nostdlib}
3081 or @samp{-nodefaultlibs} you should usually specify @samp{-lgcc} as well.
3082 This ensures that you have no unresolved references to internal GCC
3083 library subroutines. (For example, @samp{__main}, used to ensure C++
3084 constructors will be called; @pxref{Collect2,,@code{collect2}}.)
3087 Remove all symbol table and relocation information from the executable.
3090 On systems that support dynamic linking, this prevents linking with the shared
3091 libraries. On other systems, this option has no effect.
3094 Produce a shared object which can then be linked with other objects to
3095 form an executable. Not all systems support this option. You must
3096 also specify @samp{-fpic} or @samp{-fPIC} on some systems when
3097 you specify this option.
3100 Bind references to global symbols when building a shared object. Warn
3101 about any unresolved references (unless overridden by the link editor
3102 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
3105 @item -Xlinker @var{option}
3106 Pass @var{option} as an option to the linker. You can use this to
3107 supply system-specific linker options which GCC does not know how to
3110 If you want to pass an option that takes an argument, you must use
3111 @samp{-Xlinker} twice, once for the option and once for the argument.
3112 For example, to pass @samp{-assert definitions}, you must write
3113 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
3114 @samp{-Xlinker "-assert definitions"}, because this passes the entire
3115 string as a single argument, which is not what the linker expects.
3117 @item -Wl,@var{option}
3118 Pass @var{option} as an option to the linker. If @var{option} contains
3119 commas, it is split into multiple options at the commas.
3121 @item -u @var{symbol}
3122 Pretend the symbol @var{symbol} is undefined, to force linking of
3123 library modules to define it. You can use @samp{-u} multiple times with
3124 different symbols to force loading of additional library modules.
3127 @node Directory Options
3128 @section Options for Directory Search
3129 @cindex directory options
3130 @cindex options, directory search
3133 These options specify directories to search for header files, for
3134 libraries and for parts of the compiler:
3138 Add the directory @var{dir} to the head of the list of directories to be
3139 searched for header files. This can be used to override a system header
3140 file, substituting your own version, since these directories are
3141 searched before the system header file directories. If you use more
3142 than one @samp{-I} option, the directories are scanned in left-to-right
3143 order; the standard system directories come after.
3146 Any directories you specify with @samp{-I} options before the @samp{-I-}
3147 option are searched only for the case of @samp{#include "@var{file}"};
3148 they are not searched for @samp{#include <@var{file}>}.
3150 If additional directories are specified with @samp{-I} options after
3151 the @samp{-I-}, these directories are searched for all @samp{#include}
3152 directives. (Ordinarily @emph{all} @samp{-I} directories are used
3155 In addition, the @samp{-I-} option inhibits the use of the current
3156 directory (where the current input file came from) as the first search
3157 directory for @samp{#include "@var{file}"}. There is no way to
3158 override this effect of @samp{-I-}. With @samp{-I.} you can specify
3159 searching the directory which was current when the compiler was
3160 invoked. That is not exactly the same as what the preprocessor does
3161 by default, but it is often satisfactory.
3163 @samp{-I-} does not inhibit the use of the standard system directories
3164 for header files. Thus, @samp{-I-} and @samp{-nostdinc} are
3168 Add directory @var{dir} to the list of directories to be searched
3171 @item -B@var{prefix}
3172 This option specifies where to find the executables, libraries,
3173 include files, and data files of the compiler itself.
3175 The compiler driver program runs one or more of the subprograms
3176 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
3177 @var{prefix} as a prefix for each program it tries to run, both with and
3178 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
3180 For each subprogram to be run, the compiler driver first tries the
3181 @samp{-B} prefix, if any. If that name is not found, or if @samp{-B}
3182 was not specified, the driver tries two standard prefixes, which are
3183 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
3184 those results in a file name that is found, the unmodified program
3185 name is searched for using the directories specified in your
3186 @samp{PATH} environment variable.
3188 @samp{-B} prefixes that effectively specify directory names also apply
3189 to libraries in the linker, because the compiler translates these
3190 options into @samp{-L} options for the linker. They also apply to
3191 includes files in the preprocessor, because the compiler translates these
3192 options into @samp{-isystem} options for the preprocessor. In this case,
3193 the compiler appends @samp{include} to the prefix.
3195 The run-time support file @file{libgcc.a} can also be searched for using
3196 the @samp{-B} prefix, if needed. If it is not found there, the two
3197 standard prefixes above are tried, and that is all. The file is left
3198 out of the link if it is not found by those means.
3200 Another way to specify a prefix much like the @samp{-B} prefix is to use
3201 the environment variable @code{GCC_EXEC_PREFIX}. @xref{Environment
3204 @item -specs=@var{file}
3205 Process @var{file} after the compiler reads in the standard @file{specs}
3206 file, in order to override the defaults that the @file{gcc} driver
3207 program uses when determining what switches to pass to @file{cc1},
3208 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
3209 @samp{-specs=}@var{file} can be specified on the command line, and they
3210 are processed in order, from left to right.
3214 @section Specifying subprocesses and the switches to pass to them
3216 @code{GCC} is a driver program. It performs its job by invoking a
3217 sequence of other programs to do the work of compiling, assembling and
3218 linking. GCC interprets its command-line parameters and uses these to
3219 deduce which programs it should invoke, and which command-line options
3220 it ought to place on their command lines. This behaviour is controlled
3221 by @dfn{spec strings}. In most cases there is one spec string for each
3222 program that GCC can invoke, but a few programs have multiple spec
3223 strings to control their behaviour. The spec strings built into GCC can
3224 be overridden by using the @samp{-specs=} command-line switch to specify
3227 @dfn{Spec files} are plaintext files that are used to construct spec
3228 strings. They consist of a sequence of directives separated by blank
3229 lines. The type of directive is determined by the first non-whitespace
3230 character on the line and it can be one of the following:
3233 @item %@var{command}
3234 Issues a @var{command} to the spec file processor. The commands that can
3238 @item %include <@var{file}>
3240 Search for @var{file} and insert its text at the current point in the
3243 @item %include_noerr <@var{file}>
3244 @cindex %include_noerr
3245 Just like @samp{%include}, but do not generate an error message if the include
3246 file cannot be found.
3248 @item %rename @var{old_name} @var{new_name}
3250 Rename the spec string @var{old_name} to @var{new_name}.
3254 @item *[@var{spec_name}]:
3255 This tells the compiler to create, override or delete the named spec
3256 string. All lines after this directive up to the next directive or
3257 blank line are considered to be the text for the spec string. If this
3258 results in an empty string then the spec will be deleted. (Or, if the
3259 spec did not exist, then nothing will happened.) Otherwise, if the spec
3260 does not currently exist a new spec will be created. If the spec does
3261 exist then its contents will be overridden by the text of this
3262 directive, unless the first character of that text is the @samp{+}
3263 character, in which case the text will be appended to the spec.
3265 @item [@var{suffix}]:
3266 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
3267 and up to the next directive or blank line are considered to make up the
3268 spec string for the indicated suffix. When the compiler encounters an
3269 input file with the named suffix, it will processes the spec string in
3270 order to work out how to compile that file. For example:
3277 This says that any input file whose name ends in @samp{.ZZ} should be
3278 passed to the program @samp{z-compile}, which should be invoked with the
3279 command-line switch @samp{-input} and with the result of performing the
3280 @samp{%i} substitution. (See below.)
3282 As an alternative to providing a spec string, the text that follows a
3283 suffix directive can be one of the following:
3286 @item @@@var{language}
3287 This says that the suffix is an alias for a known @var{language}. This is
3288 similar to using the @code{-x} command-line switch to GCC to specify a
3289 language explicitly. For example:
3296 Says that .ZZ files are, in fact, C++ source files.
3299 This causes an error messages saying:
3302 @var{name} compiler not installed on this system.
3306 GCC already has an extensive list of suffixes built into it.
3307 This directive will add an entry to the end of the list of suffixes, but
3308 since the list is searched from the end backwards, it is effectively
3309 possible to override earlier entries using this technique.
3313 GCC has the following spec strings built into it. Spec files can
3314 override these strings or create their own. Note that individual
3315 targets can also add their own spec strings to this list.
3318 asm Options to pass to the assembler
3319 asm_final Options to pass to the assembler post-processor
3320 cpp Options to pass to the C preprocessor
3321 cc1 Options to pass to the C compiler
3322 cc1plus Options to pass to the C++ compiler
3323 endfile Object files to include at the end of the link
3324 link Options to pass to the linker
3325 lib Libraries to include on the command line to the linker
3326 libgcc Decides which GCC support library to pass to the linker
3327 linker Sets the name of the linker
3328 predefines Defines to be passed to the C preprocessor
3329 signed_char Defines to pass to CPP to say whether @code{char} is signed by default
3330 startfile Object files to include at the start of the link
3333 Here is a small example of a spec file:
3339 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
3342 This example renames the spec called @samp{lib} to @samp{old_lib} and
3343 then overrides the previous definition of @samp{lib} with a new one.
3344 The new definition adds in some extra command-line options before
3345 including the text of the old definition.
3347 @dfn{Spec strings} are a list of command-line options to be passed to their
3348 corresponding program. In addition, the spec strings can contain
3349 @samp{%}-prefixed sequences to substitute variable text or to
3350 conditionally insert text into the command line. Using these constructs
3351 it is possible to generate quite complex command lines.
3353 Here is a table of all defined @samp{%}-sequences for spec
3354 strings. Note that spaces are not generated automatically around the
3355 results of expanding these sequences. Therefore you can concatenate them
3356 together or combine them with constant text in a single argument.
3360 Substitute one @samp{%} into the program name or argument.
3363 Substitute the name of the input file being processed.
3366 Substitute the basename of the input file being processed.
3367 This is the substring up to (and not including) the last period
3368 and not including the directory.
3371 Marks the argument containing or following the @samp{%d} as a
3372 temporary file name, so that that file will be deleted if GCC exits
3373 successfully. Unlike @samp{%g}, this contributes no text to the
3376 @item %g@var{suffix}
3377 Substitute a file name that has suffix @var{suffix} and is chosen
3378 once per compilation, and mark the argument in the same way as
3379 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
3380 name is now chosen in a way that is hard to predict even when previously
3381 chosen file names are known. For example, @samp{%g.s ... %g.o ... %g.s}
3382 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
3383 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
3384 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
3385 was simply substituted with a file name chosen once per compilation,
3386 without regard to any appended suffix (which was therefore treated
3387 just like ordinary text), making such attacks more likely to succeed.
3389 @item %u@var{suffix}
3390 Like @samp{%g}, but generates a new temporary file name even if
3391 @samp{%u@var{suffix}} was already seen.
3393 @item %U@var{suffix}
3394 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
3395 new one if there is no such last file name. In the absence of any
3396 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
3397 the same suffix @emph{space}, so @samp{%g.s ... %U.s ... %g.s ... %U.s}
3398 would involve the generation of two distinct file names, one
3399 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
3400 simply substituted with a file name chosen for the previous @samp{%u},
3401 without regard to any appended suffix.
3404 Marks the argument containing or following the @samp{%w} as the
3405 designated output file of this compilation. This puts the argument
3406 into the sequence of arguments that @samp{%o} will substitute later.
3409 Substitutes the names of all the output files, with spaces
3410 automatically placed around them. You should write spaces
3411 around the @samp{%o} as well or the results are undefined.
3412 @samp{%o} is for use in the specs for running the linker.
3413 Input files whose names have no recognized suffix are not compiled
3414 at all, but they are included among the output files, so they will
3418 Substitutes the suffix for object files. Note that this is
3419 handled specially when it immediately follows @samp{%g, %u, or %U},
3420 because of the need for those to form complete file names. The
3421 handling is such that @samp{%O} is treated exactly as if it had already
3422 been substituted, except that @samp{%g, %u, and %U} do not currently
3423 support additional @var{suffix} characters following @samp{%O} as they would
3424 following, for example, @samp{.o}.
3427 Substitutes the standard macro predefinitions for the
3428 current target machine. Use this when running @code{cpp}.
3431 Like @samp{%p}, but puts @samp{__} before and after the name of each
3432 predefined macro, except for macros that start with @samp{__} or with
3433 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ANSI
3437 Substitute a @samp{-iprefix} option made from GCC_EXEC_PREFIX.
3440 Current argument is the name of a library or startup file of some sort.
3441 Search for that file in a standard list of directories and substitute
3442 the full name found.
3445 Print @var{str} as an error message. @var{str} is terminated by a newline.
3446 Use this when inconsistent options are detected.
3449 Output @samp{-} if the input for the current command is coming from a pipe.
3452 Substitute the contents of spec string @var{name} at this point.
3455 Like @samp{%(...)} but put @samp{__} around @samp{-D} arguments.
3457 @item %x@{@var{option}@}
3458 Accumulate an option for @samp{%X}.
3461 Output the accumulated linker options specified by @samp{-Wl} or a @samp{%x}
3465 Output the accumulated assembler options specified by @samp{-Wa}.
3468 Output the accumulated preprocessor options specified by @samp{-Wp}.
3471 Substitute the major version number of GCC.
3472 (For version 2.9.5, this is 2.)
3475 Substitute the minor version number of GCC.
3476 (For version 2.9.5, this is 9.)
3479 Process the @code{asm} spec. This is used to compute the
3480 switches to be passed to the assembler.
3483 Process the @code{asm_final} spec. This is a spec string for
3484 passing switches to an assembler post-processor, if such a program is
3488 Process the @code{link} spec. This is the spec for computing the
3489 command line passed to the linker. Typically it will make use of the
3490 @samp{%L %G %S %D and %E} sequences.
3493 Dump out a @samp{-L} option for each directory that GCC believes might
3494 contain startup files. If the target supports multilibs then the
3495 current multilib directory will be prepended to each of these paths.
3498 Process the @code{lib} spec. This is a spec string for deciding which
3499 libraries should be included on the command line to the linker.
3502 Process the @code{libgcc} spec. This is a spec string for deciding
3503 which GCC support library should be included on the command line to the linker.
3506 Process the @code{startfile} spec. This is a spec for deciding which
3507 object files should be the first ones passed to the linker. Typically
3508 this might be a file named @file{crt0.o}.
3511 Process the @code{endfile} spec. This is a spec string that specifies
3512 the last object files that will be passed to the linker.
3515 Process the @code{cpp} spec. This is used to construct the arguments
3516 to be passed to the C preprocessor.
3519 Process the @code{signed_char} spec. This is intended to be used
3520 to tell cpp whether a char is signed. It typically has the definition:
3522 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
3526 Process the @code{cc1} spec. This is used to construct the options to be
3527 passed to the actual C compiler (@samp{cc1}).
3530 Process the @code{cc1plus} spec. This is used to construct the options to be
3531 passed to the actual C++ compiler (@samp{cc1plus}).
3534 Substitute the variable part of a matched option. See below.
3535 Note that each comma in the substituted string is replaced by
3539 Substitutes the @code{-S} switch, if that switch was given to GCC.
3540 If that switch was not specified, this substitutes nothing. Note that
3541 the leading dash is omitted when specifying this option, and it is
3542 automatically inserted if the substitution is performed. Thus the spec
3543 string @samp{%@{foo@}} would match the command-line option @samp{-foo}
3544 and would output the command line option @samp{-foo}.
3546 @item %W@{@code{S}@}
3547 Like %@{@code{S}@} but mark last argument supplied within as a file to be
3550 @item %@{@code{S}*@}
3551 Substitutes all the switches specified to GCC whose names start
3552 with @code{-S}, but which also take an argument. This is used for
3553 switches like @samp{-o, -D, -I}, etc. GCC considers @samp{-o foo} as being
3554 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
3555 text, including the space. Thus two arguments would be generated.
3557 @item %@{^@code{S}*@}
3558 Like %@{@code{S}*@}, but don't put a blank between a switch and its
3559 argument. Thus %@{^o*@} would only generate one argument, not two.
3561 @item %@{@code{S}*:@code{X}@}
3562 Substitutes @code{X} if one or more switches whose names start with
3563 @code{-S} are specified to GCC. Note that the tail part of the
3564 @code{-S} option (i.e. the part matched by the @samp{*}) will be substituted
3565 for each occurrence of @samp{%*} within @code{X}.
3567 @item %@{@code{S}:@code{X}@}
3568 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC.
3570 @item %@{!@code{S}:@code{X}@}
3571 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC.
3573 @item %@{|@code{S}:@code{X}@}
3574 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
3576 @item %@{|!@code{S}:@code{X}@}
3577 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
3579 @item %@{.@code{S}:@code{X}@}
3580 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
3582 @item %@{!.@code{S}:@code{X}@}
3583 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
3585 @item %@{@code{S}|@code{P}:@code{X}@}
3586 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC. This may be
3587 combined with @samp{!} and @samp{.} sequences as well, although they
3588 have a stronger binding than the @samp{|}. For example a spec string
3592 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
3595 will output the following command-line options from the following input
3596 command-line options:
3601 -d fred.c -foo -baz -boggle
3602 -d jim.d -bar -baz -boggle
3607 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
3608 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
3609 or spaces, or even newlines. They are processed as usual, as described
3612 The @samp{-O, -f, -m, and -W} switches are handled specifically in these
3613 constructs. If another value of @samp{-O} or the negated form of a @samp{-f, -m, or
3614 -W} switch is found later in the command line, the earlier switch
3615 value is ignored, except with @{@code{S}*@} where @code{S} is just one
3616 letter, which passes all matching options.
3618 The character @samp{|} at the beginning of the predicate text is used to indicate
3619 that a command should be piped to the following command, but only if @samp{-pipe}
3622 It is built into GCC which switches take arguments and which do not.
3623 (You might think it would be useful to generalize this to allow each
3624 compiler's spec to say which switches take arguments. But this cannot
3625 be done in a consistent fashion. GCC cannot even decide which input
3626 files have been specified without knowing which switches take arguments,
3627 and it must know which input files to compile in order to tell which
3630 GCC also knows implicitly that arguments starting in @samp{-l} are to be
3631 treated as compiler output files, and passed to the linker in their
3632 proper position among the other output files.
3634 @node Target Options
3635 @section Specifying Target Machine and Compiler Version
3636 @cindex target options
3637 @cindex cross compiling
3638 @cindex specifying machine version
3639 @cindex specifying compiler version and target machine
3640 @cindex compiler version, specifying
3641 @cindex target machine, specifying
3643 By default, GCC compiles code for the same type of machine that you
3644 are using. However, it can also be installed as a cross-compiler, to
3645 compile for some other type of machine. In fact, several different
3646 configurations of GCC, for different target machines, can be
3647 installed side by side. Then you specify which one to use with the
3650 In addition, older and newer versions of GCC can be installed side
3651 by side. One of them (probably the newest) will be the default, but
3652 you may sometimes wish to use another.
3655 @item -b @var{machine}
3656 The argument @var{machine} specifies the target machine for compilation.
3657 This is useful when you have installed GCC as a cross-compiler.
3659 The value to use for @var{machine} is the same as was specified as the
3660 machine type when configuring GCC as a cross-compiler. For
3661 example, if a cross-compiler was configured with @samp{configure
3662 i386v}, meaning to compile for an 80386 running System V, then you
3663 would specify @samp{-b i386v} to run that cross compiler.
3665 When you do not specify @samp{-b}, it normally means to compile for
3666 the same type of machine that you are using.
3668 @item -V @var{version}
3669 The argument @var{version} specifies which version of GCC to run.
3670 This is useful when multiple versions are installed. For example,
3671 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
3673 The default version, when you do not specify @samp{-V}, is the last
3674 version of GCC that you installed.
3677 The @samp{-b} and @samp{-V} options actually work by controlling part of
3678 the file name used for the executable files and libraries used for
3679 compilation. A given version of GCC, for a given target machine, is
3680 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.@refill
3682 Thus, sites can customize the effect of @samp{-b} or @samp{-V} either by
3683 changing the names of these directories or adding alternate names (or
3684 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
3685 file @file{80386} is a link to the file @file{i386v}, then @samp{-b
3686 80386} becomes an alias for @samp{-b i386v}.
3688 In one respect, the @samp{-b} or @samp{-V} do not completely change
3689 to a different compiler: the top-level driver program @code{gcc}
3690 that you originally invoked continues to run and invoke the other
3691 executables (preprocessor, compiler per se, assembler and linker)
3692 that do the real work. However, since no real work is done in the
3693 driver program, it usually does not matter that the driver program
3694 in use is not the one for the specified target and version.
3696 The only way that the driver program depends on the target machine is
3697 in the parsing and handling of special machine-specific options.
3698 However, this is controlled by a file which is found, along with the
3699 other executables, in the directory for the specified version and
3700 target machine. As a result, a single installed driver program adapts
3701 to any specified target machine and compiler version.
3703 The driver program executable does control one significant thing,
3704 however: the default version and target machine. Therefore, you can
3705 install different instances of the driver program, compiled for
3706 different targets or versions, under different names.
3708 For example, if the driver for version 2.0 is installed as @code{ogcc}
3709 and that for version 2.1 is installed as @code{gcc}, then the command
3710 @code{gcc} will use version 2.1 by default, while @code{ogcc} will use
3711 2.0 by default. However, you can choose either version with either
3712 command with the @samp{-V} option.
3714 @node Submodel Options
3715 @section Hardware Models and Configurations
3716 @cindex submodel options
3717 @cindex specifying hardware config
3718 @cindex hardware models and configurations, specifying
3719 @cindex machine dependent options
3721 Earlier we discussed the standard option @samp{-b} which chooses among
3722 different installed compilers for completely different target
3723 machines, such as Vax vs. 68000 vs. 80386.
3725 In addition, each of these target machine types can have its own
3726 special options, starting with @samp{-m}, to choose among various
3727 hardware models or configurations---for example, 68010 vs 68020,
3728 floating coprocessor or none. A single installed version of the
3729 compiler can compile for any model or configuration, according to the
3732 Some configurations of the compiler also support additional special
3733 options, usually for compatibility with other compilers on the same
3737 These options are defined by the macro @code{TARGET_SWITCHES} in the
3738 machine description. The default for the options is also defined by
3739 that macro, which enables you to change the defaults.
3754 * RS/6000 and PowerPC Options::
3759 * Intel 960 Options::
3760 * DEC Alpha Options::
3764 * System V Options::
3765 * TMS320C3x/C4x Options::
3772 @node M680x0 Options
3773 @subsection M680x0 Options
3774 @cindex M680x0 options
3776 These are the @samp{-m} options defined for the 68000 series. The default
3777 values for these options depends on which style of 68000 was selected when
3778 the compiler was configured; the defaults for the most common choices are
3784 Generate output for a 68000. This is the default
3785 when the compiler is configured for 68000-based systems.
3787 Use this option for microcontrollers with a 68000 or EC000 core,
3788 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
3792 Generate output for a 68020. This is the default
3793 when the compiler is configured for 68020-based systems.
3796 Generate output containing 68881 instructions for floating point.
3797 This is the default for most 68020 systems unless @samp{-nfp} was
3798 specified when the compiler was configured.
3801 Generate output for a 68030. This is the default when the compiler is
3802 configured for 68030-based systems.
3805 Generate output for a 68040. This is the default when the compiler is
3806 configured for 68040-based systems.
3808 This option inhibits the use of 68881/68882 instructions that have to be
3809 emulated by software on the 68040. Use this option if your 68040 does not
3810 have code to emulate those instructions.
3813 Generate output for a 68060. This is the default when the compiler is
3814 configured for 68060-based systems.
3816 This option inhibits the use of 68020 and 68881/68882 instructions that
3817 have to be emulated by software on the 68060. Use this option if your 68060
3818 does not have code to emulate those instructions.
3821 Generate output for a CPU32. This is the default
3822 when the compiler is configured for CPU32-based systems.
3824 Use this option for microcontrollers with a
3825 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
3826 68336, 68340, 68341, 68349 and 68360.
3829 Generate output for a 520X "coldfire" family cpu. This is the default
3830 when the compiler is configured for 520X-based systems.
3832 Use this option for microcontroller with a 5200 core, including
3833 the MCF5202, MCF5203, MCF5204 and MCF5202.
3837 Generate output for a 68040, without using any of the new instructions.
3838 This results in code which can run relatively efficiently on either a
3839 68020/68881 or a 68030 or a 68040. The generated code does use the
3840 68881 instructions that are emulated on the 68040.
3843 Generate output for a 68060, without using any of the new instructions.
3844 This results in code which can run relatively efficiently on either a
3845 68020/68881 or a 68030 or a 68040. The generated code does use the
3846 68881 instructions that are emulated on the 68060.
3849 Generate output containing Sun FPA instructions for floating point.
3852 Generate output containing library calls for floating point.
3853 @strong{Warning:} the requisite libraries are not available for all m68k
3854 targets. Normally the facilities of the machine's usual C compiler are
3855 used, but this can't be done directly in cross-compilation. You must
3856 make your own arrangements to provide suitable library functions for
3857 cross-compilation. The embedded targets @samp{m68k-*-aout} and
3858 @samp{m68k-*-coff} do provide software floating point support.
3861 Consider type @code{int} to be 16 bits wide, like @code{short int}.
3864 Do not use the bit-field instructions. The @samp{-m68000}, @samp{-mcpu32}
3865 and @samp{-m5200} options imply @w{@samp{-mnobitfield}}.
3868 Do use the bit-field instructions. The @samp{-m68020} option implies
3869 @samp{-mbitfield}. This is the default if you use a configuration
3870 designed for a 68020.
3873 Use a different function-calling convention, in which functions
3874 that take a fixed number of arguments return with the @code{rtd}
3875 instruction, which pops their arguments while returning. This
3876 saves one instruction in the caller since there is no need to pop
3877 the arguments there.
3879 This calling convention is incompatible with the one normally
3880 used on Unix, so you cannot use it if you need to call libraries
3881 compiled with the Unix compiler.
3883 Also, you must provide function prototypes for all functions that
3884 take variable numbers of arguments (including @code{printf});
3885 otherwise incorrect code will be generated for calls to those
3888 In addition, seriously incorrect code will result if you call a
3889 function with too many arguments. (Normally, extra arguments are
3890 harmlessly ignored.)
3892 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
3893 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
3896 @itemx -mno-align-int
3897 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
3898 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
3899 boundary (@samp{-malign-int}) or a 16-bit boundary (@samp{-mno-align-int}).
3900 Aligning variables on 32-bit boundaries produces code that runs somewhat
3901 faster on processors with 32-bit busses at the expense of more memory.
3903 @strong{Warning:} if you use the @samp{-malign-int} switch, GCC will
3904 align structures containing the above types differently than
3905 most published application binary interface specifications for the m68k.
3908 Use the pc-relative addressing mode of the 68000 directly, instead of
3909 using a global offset table. At present, this option implies -fpic,
3910 allowing at most a 16-bit offset for pc-relative addressing. -fPIC is
3911 not presently supported with -mpcrel, though this could be supported for
3912 68020 and higher processors.
3914 @item -mno-strict-align
3915 @itemx -mstrict-align
3916 @kindex -mstrict-align
3917 Do not (do) assume that unaligned memory references will be handled by
3923 @subsection VAX Options
3926 These @samp{-m} options are defined for the Vax:
3930 Do not output certain jump instructions (@code{aobleq} and so on)
3931 that the Unix assembler for the Vax cannot handle across long
3935 Do output those jump instructions, on the assumption that you
3936 will assemble with the GNU assembler.
3939 Output code for g-format floating point numbers instead of d-format.
3943 @subsection SPARC Options
3944 @cindex SPARC options
3946 These @samp{-m} switches are supported on the SPARC:
3951 Specify @samp{-mapp-regs} to generate output using the global registers
3952 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
3955 To be fully SVR4 ABI compliant at the cost of some performance loss,
3956 specify @samp{-mno-app-regs}. You should compile libraries and system
3957 software with this option.
3961 Generate output containing floating point instructions. This is the
3966 Generate output containing library calls for floating point.
3967 @strong{Warning:} the requisite libraries are not available for all SPARC
3968 targets. Normally the facilities of the machine's usual C compiler are
3969 used, but this cannot be done directly in cross-compilation. You must make
3970 your own arrangements to provide suitable library functions for
3971 cross-compilation. The embedded targets @samp{sparc-*-aout} and
3972 @samp{sparclite-*-*} do provide software floating point support.
3974 @samp{-msoft-float} changes the calling convention in the output file;
3975 therefore, it is only useful if you compile @emph{all} of a program with
3976 this option. In particular, you need to compile @file{libgcc.a}, the
3977 library that comes with GCC, with @samp{-msoft-float} in order for
3980 @item -mhard-quad-float
3981 Generate output containing quad-word (long double) floating point
3984 @item -msoft-quad-float
3985 Generate output containing library calls for quad-word (long double)
3986 floating point instructions. The functions called are those specified
3987 in the SPARC ABI. This is the default.
3989 As of this writing, there are no sparc implementations that have hardware
3990 support for the quad-word floating point instructions. They all invoke
3991 a trap handler for one of these instructions, and then the trap handler
3992 emulates the effect of the instruction. Because of the trap handler overhead,
3993 this is much slower than calling the ABI library routines. Thus the
3994 @samp{-msoft-quad-float} option is the default.
3998 With @samp{-mepilogue} (the default), the compiler always emits code for
3999 function exit at the end of each function. Any function exit in
4000 the middle of the function (such as a return statement in C) will
4001 generate a jump to the exit code at the end of the function.
4003 With @samp{-mno-epilogue}, the compiler tries to emit exit code inline
4004 at every function exit.
4008 With @samp{-mflat}, the compiler does not generate save/restore instructions
4009 and will use a "flat" or single register window calling convention.
4010 This model uses %i7 as the frame pointer and is compatible with the normal
4011 register window model. Code from either may be intermixed.
4012 The local registers and the input registers (0-5) are still treated as
4013 "call saved" registers and will be saved on the stack as necessary.
4015 With @samp{-mno-flat} (the default), the compiler emits save/restore
4016 instructions (except for leaf functions) and is the normal mode of operation.
4018 @item -mno-unaligned-doubles
4019 @itemx -munaligned-doubles
4020 Assume that doubles have 8 byte alignment. This is the default.
4022 With @samp{-munaligned-doubles}, GCC assumes that doubles have 8 byte
4023 alignment only if they are contained in another type, or if they have an
4024 absolute address. Otherwise, it assumes they have 4 byte alignment.
4025 Specifying this option avoids some rare compatibility problems with code
4026 generated by other compilers. It is not the default because it results
4027 in a performance loss, especially for floating point code.
4029 @item -mno-faster-structs
4030 @itemx -mfaster-structs
4031 With @samp{-mfaster-structs}, the compiler assumes that structures
4032 should have 8 byte alignment. This enables the use of pairs of
4033 @code{ldd} and @code{std} instructions for copies in structure
4034 assignment, in place of twice as many @code{ld} and @code{st} pairs.
4035 However, the use of this changed alignment directly violates the Sparc
4036 ABI. Thus, it's intended only for use on targets where the developer
4037 acknowledges that their resulting code will not be directly in line with
4038 the rules of the ABI.
4042 These two options select variations on the SPARC architecture.
4044 By default (unless specifically configured for the Fujitsu SPARClite),
4045 GCC generates code for the v7 variant of the SPARC architecture.
4047 @samp{-mv8} will give you SPARC v8 code. The only difference from v7
4048 code is that the compiler emits the integer multiply and integer
4049 divide instructions which exist in SPARC v8 but not in SPARC v7.
4051 @samp{-msparclite} will give you SPARClite code. This adds the integer
4052 multiply, integer divide step and scan (@code{ffs}) instructions which
4053 exist in SPARClite but not in SPARC v7.
4055 These options are deprecated and will be deleted in a future GCC release.
4056 They have been replaced with @samp{-mcpu=xxx}.
4060 These two options select the processor for which the code is optimised.
4062 With @samp{-mcypress} (the default), the compiler optimizes code for the
4063 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
4064 This is also appropriate for the older SparcStation 1, 2, IPX etc.
4066 With @samp{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
4067 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
4068 of the full SPARC v8 instruction set.
4070 These options are deprecated and will be deleted in a future GCC release.
4071 They have been replaced with @samp{-mcpu=xxx}.
4073 @item -mcpu=@var{cpu_type}
4074 Set the instruction set, register set, and instruction scheduling parameters
4075 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
4076 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
4077 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
4078 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
4080 Default instruction scheduling parameters are used for values that select
4081 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
4082 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
4084 Here is a list of each supported architecture and their supported
4089 v8: supersparc, hypersparc
4090 sparclite: f930, f934, sparclite86x
4095 @item -mtune=@var{cpu_type}
4096 Set the instruction scheduling parameters for machine type
4097 @var{cpu_type}, but do not set the instruction set or register set that the
4098 option @samp{-mcpu=}@var{cpu_type} would.
4100 The same values for @samp{-mcpu=}@var{cpu_type} are used for
4101 @samp{-mtune=}@*@var{cpu_type}, though the only useful values are those that
4102 select a particular cpu implementation: @samp{cypress}, @samp{supersparc},
4103 @samp{hypersparc}, @samp{f930}, @samp{f934}, @samp{sparclite86x},
4104 @samp{tsc701}, @samp{ultrasparc}.
4108 These @samp{-m} switches are supported in addition to the above
4109 on the SPARCLET processor.
4112 @item -mlittle-endian
4113 Generate code for a processor running in little-endian mode.
4116 Treat register @code{%g0} as a normal register.
4117 GCC will continue to clobber it as necessary but will not assume
4118 it always reads as 0.
4120 @item -mbroken-saverestore
4121 Generate code that does not use non-trivial forms of the @code{save} and
4122 @code{restore} instructions. Early versions of the SPARCLET processor do
4123 not correctly handle @code{save} and @code{restore} instructions used with
4124 arguments. They correctly handle them used without arguments. A @code{save}
4125 instruction used without arguments increments the current window pointer
4126 but does not allocate a new stack frame. It is assumed that the window
4127 overflow trap handler will properly handle this case as will interrupt
4131 These @samp{-m} switches are supported in addition to the above
4132 on SPARC V9 processors in 64 bit environments.
4135 @item -mlittle-endian
4136 Generate code for a processor running in little-endian mode.
4140 Generate code for a 32 bit or 64 bit environment.
4141 The 32 bit environment sets int, long and pointer to 32 bits.
4142 The 64 bit environment sets int to 32 bits and long and pointer
4145 @item -mcmodel=medlow
4146 Generate code for the Medium/Low code model: the program must be linked
4147 in the low 32 bits of the address space. Pointers are 64 bits.
4148 Programs can be statically or dynamically linked.
4150 @item -mcmodel=medmid
4151 Generate code for the Medium/Middle code model: the program must be linked
4152 in the low 44 bits of the address space, the text segment must be less than
4153 2G bytes, and data segment must be within 2G of the text segment.
4154 Pointers are 64 bits.
4156 @item -mcmodel=medany
4157 Generate code for the Medium/Anywhere code model: the program may be linked
4158 anywhere in the address space, the text segment must be less than
4159 2G bytes, and data segment must be within 2G of the text segment.
4160 Pointers are 64 bits.
4162 @item -mcmodel=embmedany
4163 Generate code for the Medium/Anywhere code model for embedded systems:
4164 assume a 32 bit text and a 32 bit data segment, both starting anywhere
4165 (determined at link time). Register %g4 points to the base of the
4166 data segment. Pointers still 64 bits.
4167 Programs are statically linked, PIC is not supported.
4170 @itemx -mno-stack-bias
4171 With @samp{-mstack-bias}, GCC assumes that the stack pointer, and
4172 frame pointer if present, are offset by -2047 which must be added back
4173 when making stack frame references.
4174 Otherwise, assume no such offset is present.
4177 @node Convex Options
4178 @subsection Convex Options
4179 @cindex Convex options
4181 These @samp{-m} options are defined for Convex:
4185 Generate output for C1. The code will run on any Convex machine.
4186 The preprocessor symbol @code{__convex__c1__} is defined.
4189 Generate output for C2. Uses instructions not available on C1.
4190 Scheduling and other optimizations are chosen for max performance on C2.
4191 The preprocessor symbol @code{__convex_c2__} is defined.
4194 Generate output for C32xx. Uses instructions not available on C1.
4195 Scheduling and other optimizations are chosen for max performance on C32.
4196 The preprocessor symbol @code{__convex_c32__} is defined.
4199 Generate output for C34xx. Uses instructions not available on C1.
4200 Scheduling and other optimizations are chosen for max performance on C34.
4201 The preprocessor symbol @code{__convex_c34__} is defined.
4204 Generate output for C38xx. Uses instructions not available on C1.
4205 Scheduling and other optimizations are chosen for max performance on C38.
4206 The preprocessor symbol @code{__convex_c38__} is defined.
4209 Generate code which puts an argument count in the word preceding each
4210 argument list. This is compatible with regular CC, and a few programs
4211 may need the argument count word. GDB and other source-level debuggers
4212 do not need it; this info is in the symbol table.
4215 Omit the argument count word. This is the default.
4217 @item -mvolatile-cache
4218 Allow volatile references to be cached. This is the default.
4220 @item -mvolatile-nocache
4221 Volatile references bypass the data cache, going all the way to memory.
4222 This is only needed for multi-processor code that does not use standard
4223 synchronization instructions. Making non-volatile references to volatile
4224 locations will not necessarily work.
4227 Type long is 32 bits, the same as type int. This is the default.
4230 Type long is 64 bits, the same as type long long. This option is useless,
4231 because no library support exists for it.
4234 @node AMD29K Options
4235 @subsection AMD29K Options
4236 @cindex AMD29K options
4238 These @samp{-m} options are defined for the AMD Am29000:
4243 @cindex DW bit (29k)
4244 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
4245 halfword operations are directly supported by the hardware. This is the
4250 Generate code that assumes the @code{DW} bit is not set.
4254 @cindex byte writes (29k)
4255 Generate code that assumes the system supports byte and halfword write
4256 operations. This is the default.
4260 Generate code that assumes the systems does not support byte and
4261 halfword write operations. @samp{-mnbw} implies @samp{-mndw}.
4265 @cindex memory model (29k)
4266 Use a small memory model that assumes that all function addresses are
4267 either within a single 256 KB segment or at an absolute address of less
4268 than 256k. This allows the @code{call} instruction to be used instead
4269 of a @code{const}, @code{consth}, @code{calli} sequence.
4273 Use the normal memory model: Generate @code{call} instructions only when
4274 calling functions in the same file and @code{calli} instructions
4275 otherwise. This works if each file occupies less than 256 KB but allows
4276 the entire executable to be larger than 256 KB. This is the default.
4279 Always use @code{calli} instructions. Specify this option if you expect
4280 a single file to compile into more than 256 KB of code.
4284 @cindex processor selection (29k)
4285 Generate code for the Am29050.
4289 Generate code for the Am29000. This is the default.
4291 @item -mkernel-registers
4292 @kindex -mkernel-registers
4293 @cindex kernel and user registers (29k)
4294 Generate references to registers @code{gr64-gr95} instead of to
4295 registers @code{gr96-gr127}. This option can be used when compiling
4296 kernel code that wants a set of global registers disjoint from that used
4299 Note that when this option is used, register names in @samp{-f} flags
4300 must use the normal, user-mode, names.
4302 @item -muser-registers
4303 @kindex -muser-registers
4304 Use the normal set of global registers, @code{gr96-gr127}. This is the
4308 @itemx -mno-stack-check
4309 @kindex -mstack-check
4310 @cindex stack checks (29k)
4311 Insert (or do not insert) a call to @code{__msp_check} after each stack
4312 adjustment. This is often used for kernel code.
4315 @itemx -mno-storem-bug
4316 @kindex -mstorem-bug
4317 @cindex storem bug (29k)
4318 @samp{-mstorem-bug} handles 29k processors which cannot handle the
4319 separation of a mtsrim insn and a storem instruction (most 29000 chips
4320 to date, but not the 29050).
4322 @item -mno-reuse-arg-regs
4323 @itemx -mreuse-arg-regs
4324 @kindex -mreuse-arg-regs
4325 @samp{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
4326 registers for copying out arguments. This helps detect calling a function
4327 with fewer arguments than it was declared with.
4329 @item -mno-impure-text
4330 @itemx -mimpure-text
4331 @kindex -mimpure-text
4332 @samp{-mimpure-text}, used in addition to @samp{-shared}, tells the compiler to
4333 not pass @samp{-assert pure-text} to the linker when linking a shared object.
4336 @kindex -msoft-float
4337 Generate output containing library calls for floating point.
4338 @strong{Warning:} the requisite libraries are not part of GCC.
4339 Normally the facilities of the machine's usual C compiler are used, but
4340 this can't be done directly in cross-compilation. You must make your
4341 own arrangements to provide suitable library functions for
4346 Do not generate multm or multmu instructions. This is useful for some embedded
4347 systems which do not have trap handlers for these instructions.
4351 @subsection ARM Options
4354 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
4359 @kindex -mapcs-frame
4360 Generate a stack frame that is compliant with the ARM Procedure Call
4361 Standard for all functions, even if this is not strictly necessary for
4362 correct execution of the code. Specifying @samp{-fomit-frame-pointer}
4363 with this option will cause the stack frames not to be generated for
4364 leaf functions. The default is @samp{-mno-apcs-frame}.
4368 This is a synonym for @samp{-mapcs-frame}.
4372 Generate code for a processor running with a 26-bit program counter,
4373 and conforming to the function calling standards for the APCS 26-bit
4374 option. This option replaces the @samp{-m2} and @samp{-m3} options
4375 of previous releases of the compiler.
4379 Generate code for a processor running with a 32-bit program counter,
4380 and conforming to the function calling standards for the APCS 32-bit
4381 option. This option replaces the @samp{-m6} option of previous releases
4384 @item -mapcs-stack-check
4385 @kindex -mapcs-stack-check
4386 @kindex -mno-apcs-stack-check
4387 Generate code to check the amount of stack space available upon entry to
4388 every function (that actually uses some stack space). If there is
4389 insufficient space available then either the function
4390 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
4391 called, depending upon the amount of stack space required. The run time
4392 system is required to provide these functions. The default is
4393 @samp{-mno-apcs-stack-check}, since this produces smaller code.
4396 @kindex -mapcs-float
4397 @kindex -mno-apcs-float
4398 Pass floating point arguments using the float point registers. This is
4399 one of the variants of the APCS. This option is recommended if the
4400 target hardware has a floating point unit or if a lot of floating point
4401 arithmetic is going to be performed by the code. The default is
4402 @samp{-mno-apcs-float}, since integer only code is slightly increased in
4403 size if @samp{-mapcs-float} is used.
4405 @item -mapcs-reentrant
4406 @kindex -mapcs-reentrant
4407 @kindex -mno-apcs-reentrant
4408 Generate reentrant, position independent code. This is the equivalent
4409 to specifying the @samp{-fpic} option. The default is
4410 @samp{-mno-apcs-reentrant}.
4412 @item -mthumb-interwork
4413 @kindex -mthumb-interwork
4414 @kindex -mno-thumb-interwork
4415 Generate code which supports calling between the ARM and THUMB
4416 instruction sets. Without this option the two instruction sets cannot
4417 be reliably used inside one program. The default is
4418 @samp{-mno-thumb-interwork}, since slightly larger code is generated
4419 when @samp{-mthumb-interwork} is specified.
4421 @item -mno-sched-prolog
4422 @kindex -mno-sched-prolog
4423 @kindex -msched-prolog
4424 Prevent the reordering of instructions in the function prolog, or the
4425 merging of those instruction with the instructions in the function's
4426 body. This means that all functions will start with a recognizable set
4427 of instructions (or in fact one of a choice from a small set of
4428 different function prologues), and this information can be used to
4429 locate the start if functions inside an executable piece of code. The
4430 default is @samp{-msched-prolog}.
4433 Generate output containing floating point instructions. This is the
4437 Generate output containing library calls for floating point.
4438 @strong{Warning:} the requisite libraries are not available for all ARM
4439 targets. Normally the facilities of the machine's usual C compiler are
4440 used, but this cannot be done directly in cross-compilation. You must make
4441 your own arrangements to provide suitable library functions for
4444 @samp{-msoft-float} changes the calling convention in the output file;
4445 therefore, it is only useful if you compile @emph{all} of a program with
4446 this option. In particular, you need to compile @file{libgcc.a}, the
4447 library that comes with GCC, with @samp{-msoft-float} in order for
4450 @item -mlittle-endian
4451 Generate code for a processor running in little-endian mode. This is
4452 the default for all standard configurations.
4455 Generate code for a processor running in big-endian mode; the default is
4456 to compile code for a little-endian processor.
4458 @item -mwords-little-endian
4459 This option only applies when generating code for big-endian processors.
4460 Generate code for a little-endian word order but a big-endian byte
4461 order. That is, a byte order of the form @samp{32107654}. Note: this
4462 option should only be used if you require compatibility with code for
4463 big-endian ARM processors generated by versions of the compiler prior to
4466 @item -malignment-traps
4467 @kindex -malignment-traps
4468 Generate code that will not trap if the MMU has alignment traps enabled.
4469 On ARM architectures prior to ARMv4, there were no instructions to
4470 access half-word objects stored in memory. However, when reading from
4471 memory a feature of the ARM architecture allows a word load to be used,
4472 even if the address is unaligned, and the processor core will rotate the
4473 data as it is being loaded. This option tells the compiler that such
4474 misaligned accesses will cause a MMU trap and that it should instead
4475 synthesise the access as a series of byte accesses. The compiler can
4476 still use word accesses to load half-word data if it knows that the
4477 address is aligned to a word boundary.
4479 This option is ignored when compiling for ARM architecture 4 or later,
4480 since these processors have instructions to directly access half-word
4483 @item -mno-alignment-traps
4484 @kindex -mno-alignment-traps
4485 Generate code that assumes that the MMU will not trap unaligned
4486 accesses. This produces better code when the target instruction set
4487 does not have half-word memory operations (implementations prior to
4490 Note that you cannot use this option to access unaligned word objects,
4491 since the processor will only fetch one 32-bit aligned object from
4494 The default setting for most targets is -mno-alignment-traps, since
4495 this produces better code when there are no half-word memory
4496 instructions available.
4498 @item -mshort-load-bytes
4499 @kindex -mshort-load-bytes
4500 This is a depreciated alias for @samp{-malignment-traps}.
4502 @item -mno-short-load-bytes
4503 @kindex -mno-short-load-bytes
4504 This is a depreciated alias for @samp{-mno-alignment-traps}.
4506 @item -mshort-load-words
4507 @kindex -mshort-load-words
4508 This is a depreciated alias for @samp{-mno-alignment-traps}.
4510 @item -mno-short-load-words
4511 @kindex -mno-short-load-words
4512 This is a depreciated alias for @samp{-malignment-traps}.
4516 This option only applies to RISC iX. Emulate the native BSD-mode
4517 compiler. This is the default if @samp{-ansi} is not specified.
4521 This option only applies to RISC iX. Emulate the native X/Open-mode
4524 @item -mno-symrename
4525 @kindex -mno-symrename
4526 This option only applies to RISC iX. Do not run the assembler
4527 post-processor, @samp{symrename}, after code has been assembled.
4528 Normally it is necessary to modify some of the standard symbols in
4529 preparation for linking with the RISC iX C library; this option
4530 suppresses this pass. The post-processor is never run when the
4531 compiler is built for cross-compilation.
4535 This specifies the name of the target ARM processor. GCC uses this name
4536 to determine what kind of instructions it can use when generating
4537 assembly code. Permissible names are: arm2, arm250, arm3, arm6, arm60,
4538 arm600, arm610, arm620, arm7, arm7m, arm7d, arm7dm, arm7di, arm7dmi,
4539 arm70, arm700, arm700i, arm710, arm710c, arm7100, arm7500, arm7500fe,
4540 arm7tdmi, arm8, strongarm, strongarm110, strongarm1100, arm8, arm810,
4541 arm9, arm920, arm920t, arm9tdmi.
4543 @itemx -mtune=<name>
4545 This option is very similar to the @samp{-mcpu=} option, except that
4546 instead of specifying the actual target processor type, and hence
4547 restricting which instructions can be used, it specifies that GCC should
4548 tune the performance of the code as if the target were of the type
4549 specified in this option, but still choosing the instructions that it
4550 will generate based on the cpu specified by a @samp{-mcpu=} option.
4551 For some arm implementations better performance can be obtained by using
4556 This specifies the name of the target ARM architecture. GCC uses this
4557 name to determine what kind of instructions it can use when generating
4558 assembly code. This option can be used in conjunction with or instead
4559 of the @samp{-mcpu=} option. Permissible names are: armv2, armv2a,
4560 armv3, armv3m, armv4, armv4t, armv5.
4562 @item -mfpe=<number>
4563 @itemx -mfp=<number>
4566 This specifes the version of the floating point emulation available on
4567 the target. Permissible values are 2 and 3. @samp{-mfp=} is a synonym
4568 for @samp{-mfpe=} to support older versions of GCC.
4570 @item -mstructure-size-boundary=<n>
4571 @kindex -mstructure-size-boundary
4572 The size of all structures and unions will be rounded up to a multiple
4573 of the number of bits set by this option. Permissible values are 8 and
4574 32. The default value varies for different toolchains. For the COFF
4575 targeted toolchain the default value is 8. Specifying the larger number
4576 can produce faster, more efficient code, but can also increase the size
4577 of the program. The two values are potentially incompatible. Code
4578 compiled with one value cannot necessarily expect to work with code or
4579 libraries compiled with the other value, if they exchange information
4580 using structures or unions. Programmers are encouraged to use the 32
4581 value as future versions of the toolchain may default to this value.
4583 @item -mabort-on-noreturn
4584 @kindex -mabort-on-noreturn
4585 @kindex -mnoabort-on-noreturn
4586 Generate a call to the function abort at the end of a noreturn function.
4587 It will be executed if the function tries to return.
4589 @item -mnop-fun-dllimport
4590 @kindex -mnop-fun-dllimport
4591 Disable the support for the @emph{dllimport} attribute.
4593 @item -msingle-pic-base
4594 @kindex -msingle-pic-base
4595 Treat the register used for PIC addressing as read-only, rather than
4596 loading it in the prologue for each function. The run-time system is
4597 responsible for initialising this register with an appropriate value
4598 before execution begins.
4600 @item -mpic-register=<reg>
4601 @kindex -mpic-register=
4602 Specify the register to be used for PIC addressing. The default is R10
4603 unless stack-checking is enabled, when R9 is used.
4608 @subsection Thumb Options
4609 @cindex Thumb Options
4613 @item -mthumb-interwork
4614 @kindex -mthumb-interwork
4615 @kindex -mno-thumb-interwork
4616 Generate code which supports calling between the THUMB and ARM
4617 instruction sets. Without this option the two instruction sets cannot
4618 be reliably used inside one program. The default is
4619 @samp{-mno-thumb-interwork}, since slightly smaller code is generated
4623 @kindex -mtpcs-frame
4624 @kindex -mno-tpcs-frame
4625 Generate a stack frame that is compliant with the Thumb Procedure Call
4626 Standard for all non-leaf functions. (A leaf function is one that does
4627 not call any other functions). The default is @samp{-mno-apcs-frame}.
4629 @item -mtpcs-leaf-frame
4630 @kindex -mtpcs-leaf-frame
4631 @kindex -mno-tpcs-leaf-frame
4632 Generate a stack frame that is compliant with the Thumb Procedure Call
4633 Standard for all leaf functions. (A leaf function is one that does
4634 not call any other functions). The default is @samp{-mno-apcs-leaf-frame}.
4636 @item -mlittle-endian
4637 @kindex -mlittle-endian
4638 Generate code for a processor running in little-endian mode. This is
4639 the default for all standard configurations.
4642 @kindex -mbig-endian
4643 Generate code for a processor running in big-endian mode.
4645 @item -mstructure-size-boundary=<n>
4646 @kindex -mstructure-size-boundary
4647 The size of all structures and unions will be rounded up to a multiple
4648 of the number of bits set by this option. Permissible values are 8 and
4649 32. The default value varies for different toolchains. For the COFF
4650 targeted toolchain the default value is 8. Specifying the larger number
4651 can produced faster, more efficient code, but can also increase the size
4652 of the program. The two values are potentially incompatible. Code
4653 compiled with one value cannot necessarily expect to work with code or
4654 libraries compiled with the other value, if they exchange information
4655 using structures or unions. Programmers are encouraged to use the 32
4656 value as future versions of the toolchain may default to this value.
4658 @item -mnop-fun-dllimport
4659 @kindex -mnop-fun-dllimport
4660 Disable the support for the @emph{dllimport} attribute.
4662 @item -mcallee-super-interworking
4663 @kindex -mcallee-super-interworking
4664 Gives all externally visible functions in the file being compiled an ARM
4665 instruction set header which switches to Thumb mode before executing the
4666 rest of the function. This allows these functions to be called from
4667 non-interworking code.
4669 @item -mcaller-super-interworking
4670 @kindex -mcaller-super-interworking
4671 Allows calls via function pointers (including virtual functions) to
4672 execute correctly regardless of whether the target code has been
4673 compiled for interworking or not. There is a small overhead in the cost
4674 of executing a function pointer if this option is enabled.
4676 @item -msingle-pic-base
4677 @kindex -msingle-pic-base
4678 Treat the register used for PIC addressing as read-only, rather than
4679 loading it in the prologue for each function. The run-time system is
4680 responsible for initialising this register with an appropriate value
4681 before execution begins.
4683 @item -mpic-register=<reg>
4684 @kindex -mpic-register=
4685 Specify the register to be used for PIC addressing. The default is R10.
4689 @node MN10200 Options
4690 @subsection MN10200 Options
4691 @cindex MN10200 options
4692 These @samp{-m} options are defined for Matsushita MN10200 architectures:
4696 Indicate to the linker that it should perform a relaxation optimization pass
4697 to shorten branches, calls and absolute memory addresses. This option only
4698 has an effect when used on the command line for the final link step.
4700 This option makes symbolic debugging impossible.
4703 @node MN10300 Options
4704 @subsection MN10300 Options
4705 @cindex MN10300 options
4706 These @samp{-m} options are defined for Matsushita MN10300 architectures:
4710 Generate code to avoid bugs in the multiply instructions for the MN10300
4711 processors. This is the default.
4714 Do not generate code to avoid bugs in the multiply instructions for the
4718 Generate code which uses features specific to the AM33 processor.
4721 Do not generate code which uses features specific to the AM33 processor. This
4725 Indicate to the linker that it should perform a relaxation optimization pass
4726 to shorten branches, calls and absolute memory addresses. This option only
4727 has an effect when used on the command line for the final link step.
4729 This option makes symbolic debugging impossible.
4733 @node M32R/D Options
4734 @subsection M32R/D Options
4735 @cindex M32R/D options
4737 These @samp{-m} options are defined for Mitsubishi M32R/D architectures:
4740 @item -mcode-model=small
4741 Assume all objects live in the lower 16MB of memory (so that their addresses
4742 can be loaded with the @code{ld24} instruction), and assume all subroutines
4743 are reachable with the @code{bl} instruction.
4744 This is the default.
4746 The addressability of a particular object can be set with the
4747 @code{model} attribute.
4749 @item -mcode-model=medium
4750 Assume objects may be anywhere in the 32 bit address space (the compiler
4751 will generate @code{seth/add3} instructions to load their addresses), and
4752 assume all subroutines are reachable with the @code{bl} instruction.
4754 @item -mcode-model=large
4755 Assume objects may be anywhere in the 32 bit address space (the compiler
4756 will generate @code{seth/add3} instructions to load their addresses), and
4757 assume subroutines may not be reachable with the @code{bl} instruction
4758 (the compiler will generate the much slower @code{seth/add3/jl}
4759 instruction sequence).
4762 Disable use of the small data area. Variables will be put into
4763 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
4764 @code{section} attribute has been specified).
4765 This is the default.
4767 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
4768 Objects may be explicitly put in the small data area with the
4769 @code{section} attribute using one of these sections.
4772 Put small global and static data in the small data area, but do not
4773 generate special code to reference them.
4776 Put small global and static data in the small data area, and generate
4777 special instructions to reference them.
4780 @cindex smaller data references
4781 Put global and static objects less than or equal to @var{num} bytes
4782 into the small data or bss sections instead of the normal data or bss
4783 sections. The default value of @var{num} is 8.
4784 The @samp{-msdata} option must be set to one of @samp{sdata} or @samp{use}
4785 for this option to have any effect.
4787 All modules should be compiled with the same @samp{-G @var{num}} value.
4788 Compiling with different values of @var{num} may or may not work; if it
4789 doesn't the linker will give an error message - incorrect code will not be
4795 @subsection M88K Options
4796 @cindex M88k options
4798 These @samp{-m} options are defined for Motorola 88k architectures:
4803 Generate code that works well on both the m88100 and the
4808 Generate code that works best for the m88100, but that also
4813 Generate code that works best for the m88110, and may not run
4818 Obsolete option to be removed from the next revision.
4821 @item -midentify-revision
4822 @kindex -midentify-revision
4824 @cindex identifying source, compiler (88k)
4825 Include an @code{ident} directive in the assembler output recording the
4826 source file name, compiler name and version, timestamp, and compilation
4829 @item -mno-underscores
4830 @kindex -mno-underscores
4831 @cindex underscores, avoiding (88k)
4832 In assembler output, emit symbol names without adding an underscore
4833 character at the beginning of each name. The default is to use an
4834 underscore as prefix on each name.
4836 @item -mocs-debug-info
4837 @itemx -mno-ocs-debug-info
4838 @kindex -mocs-debug-info
4839 @kindex -mno-ocs-debug-info
4841 @cindex debugging, 88k OCS
4842 Include (or omit) additional debugging information (about registers used
4843 in each stack frame) as specified in the 88open Object Compatibility
4844 Standard, ``OCS''. This extra information allows debugging of code that
4845 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
4846 Delta 88 SVr3.2 is to include this information; other 88k configurations
4847 omit this information by default.
4849 @item -mocs-frame-position
4850 @kindex -mocs-frame-position
4851 @cindex register positions in frame (88k)
4852 When emitting COFF debugging information for automatic variables and
4853 parameters stored on the stack, use the offset from the canonical frame
4854 address, which is the stack pointer (register 31) on entry to the
4855 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
4856 @samp{-mocs-frame-position}; other 88k configurations have the default
4857 @samp{-mno-ocs-frame-position}.
4859 @item -mno-ocs-frame-position
4860 @kindex -mno-ocs-frame-position
4861 @cindex register positions in frame (88k)
4862 When emitting COFF debugging information for automatic variables and
4863 parameters stored on the stack, use the offset from the frame pointer
4864 register (register 30). When this option is in effect, the frame
4865 pointer is not eliminated when debugging information is selected by the
4868 @item -moptimize-arg-area
4869 @itemx -mno-optimize-arg-area
4870 @kindex -moptimize-arg-area
4871 @kindex -mno-optimize-arg-area
4872 @cindex arguments in frame (88k)
4873 Control how function arguments are stored in stack frames.
4874 @samp{-moptimize-arg-area} saves space by optimizing them, but this
4875 conflicts with the 88open specifications. The opposite alternative,
4876 @samp{-mno-optimize-arg-area}, agrees with 88open standards. By default
4877 GCC does not optimize the argument area.
4879 @item -mshort-data-@var{num}
4880 @kindex -mshort-data-@var{num}
4881 @cindex smaller data references (88k)
4882 @cindex r0-relative references (88k)
4883 Generate smaller data references by making them relative to @code{r0},
4884 which allows loading a value using a single instruction (rather than the
4885 usual two). You control which data references are affected by
4886 specifying @var{num} with this option. For example, if you specify
4887 @samp{-mshort-data-512}, then the data references affected are those
4888 involving displacements of less than 512 bytes.
4889 @samp{-mshort-data-@var{num}} is not effective for @var{num} greater
4892 @item -mserialize-volatile
4893 @kindex -mserialize-volatile
4894 @itemx -mno-serialize-volatile
4895 @kindex -mno-serialize-volatile
4896 @cindex sequential consistency on 88k
4897 Do, or don't, generate code to guarantee sequential consistency
4898 of volatile memory references. By default, consistency is
4901 The order of memory references made by the MC88110 processor does
4902 not always match the order of the instructions requesting those
4903 references. In particular, a load instruction may execute before
4904 a preceding store instruction. Such reordering violates
4905 sequential consistency of volatile memory references, when there
4906 are multiple processors. When consistency must be guaranteed,
4907 GNU C generates special instructions, as needed, to force
4908 execution in the proper order.
4910 The MC88100 processor does not reorder memory references and so
4911 always provides sequential consistency. However, by default, GNU
4912 C generates the special instructions to guarantee consistency
4913 even when you use @samp{-m88100}, so that the code may be run on an
4914 MC88110 processor. If you intend to run your code only on the
4915 MC88100 processor, you may use @samp{-mno-serialize-volatile}.
4917 The extra code generated to guarantee consistency may affect the
4918 performance of your application. If you know that you can safely
4919 forgo this guarantee, you may use @samp{-mno-serialize-volatile}.
4925 @cindex assembler syntax, 88k
4927 Turn on (@samp{-msvr4}) or off (@samp{-msvr3}) compiler extensions
4928 related to System V release 4 (SVr4). This controls the following:
4932 Which variant of the assembler syntax to emit.
4934 @samp{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
4935 that is used on System V release 4.
4937 @samp{-msvr4} makes GCC issue additional declaration directives used in
4941 @samp{-msvr4} is the default for the m88k-motorola-sysv4 and
4942 m88k-dg-dgux m88k configurations. @samp{-msvr3} is the default for all
4943 other m88k configurations.
4945 @item -mversion-03.00
4946 @kindex -mversion-03.00
4947 This option is obsolete, and is ignored.
4948 @c ??? which asm syntax better for GAS? option there too?
4950 @item -mno-check-zero-division
4951 @itemx -mcheck-zero-division
4952 @kindex -mno-check-zero-division
4953 @kindex -mcheck-zero-division
4954 @cindex zero division on 88k
4955 Do, or don't, generate code to guarantee that integer division by
4956 zero will be detected. By default, detection is guaranteed.
4958 Some models of the MC88100 processor fail to trap upon integer
4959 division by zero under certain conditions. By default, when
4960 compiling code that might be run on such a processor, GNU C
4961 generates code that explicitly checks for zero-valued divisors
4962 and traps with exception number 503 when one is detected. Use of
4963 mno-check-zero-division suppresses such checking for code
4964 generated to run on an MC88100 processor.
4966 GNU C assumes that the MC88110 processor correctly detects all
4967 instances of integer division by zero. When @samp{-m88110} is
4968 specified, both @samp{-mcheck-zero-division} and
4969 @samp{-mno-check-zero-division} are ignored, and no explicit checks for
4970 zero-valued divisors are generated.
4972 @item -muse-div-instruction
4973 @kindex -muse-div-instruction
4974 @cindex divide instruction, 88k
4975 Use the div instruction for signed integer division on the
4976 MC88100 processor. By default, the div instruction is not used.
4978 On the MC88100 processor the signed integer division instruction
4979 div) traps to the operating system on a negative operand. The
4980 operating system transparently completes the operation, but at a
4981 large cost in execution time. By default, when compiling code
4982 that might be run on an MC88100 processor, GNU C emulates signed
4983 integer division using the unsigned integer division instruction
4984 divu), thereby avoiding the large penalty of a trap to the
4985 operating system. Such emulation has its own, smaller, execution
4986 cost in both time and space. To the extent that your code's
4987 important signed integer division operations are performed on two
4988 nonnegative operands, it may be desirable to use the div
4989 instruction directly.
4991 On the MC88110 processor the div instruction (also known as the
4992 divs instruction) processes negative operands without trapping to
4993 the operating system. When @samp{-m88110} is specified,
4994 @samp{-muse-div-instruction} is ignored, and the div instruction is used
4995 for signed integer division.
4997 Note that the result of dividing INT_MIN by -1 is undefined. In
4998 particular, the behavior of such a division with and without
4999 @samp{-muse-div-instruction} may differ.
5001 @item -mtrap-large-shift
5002 @itemx -mhandle-large-shift
5003 @kindex -mtrap-large-shift
5004 @kindex -mhandle-large-shift
5005 @cindex bit shift overflow (88k)
5006 @cindex large bit shifts (88k)
5007 Include code to detect bit-shifts of more than 31 bits; respectively,
5008 trap such shifts or emit code to handle them properly. By default GCC
5009 makes no special provision for large bit shifts.
5011 @item -mwarn-passed-structs
5012 @kindex -mwarn-passed-structs
5013 @cindex structure passing (88k)
5014 Warn when a function passes a struct as an argument or result.
5015 Structure-passing conventions have changed during the evolution of the C
5016 language, and are often the source of portability problems. By default,
5017 GCC issues no such warning.
5020 @node RS/6000 and PowerPC Options
5021 @subsection IBM RS/6000 and PowerPC Options
5022 @cindex RS/6000 and PowerPC Options
5023 @cindex IBM RS/6000 and PowerPC Options
5025 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
5033 @itemx -mpowerpc-gpopt
5034 @itemx -mno-powerpc-gpopt
5035 @itemx -mpowerpc-gfxopt
5036 @itemx -mno-powerpc-gfxopt
5038 @itemx -mno-powerpc64
5042 @kindex -mpowerpc-gpopt
5043 @kindex -mpowerpc-gfxopt
5045 GCC supports two related instruction set architectures for the
5046 RS/6000 and PowerPC. The @dfn{POWER} instruction set are those
5047 instructions supported by the @samp{rios} chip set used in the original
5048 RS/6000 systems and the @dfn{PowerPC} instruction set is the
5049 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
5050 the IBM 4xx microprocessors.
5052 Neither architecture is a subset of the other. However there is a
5053 large common subset of instructions supported by both. An MQ
5054 register is included in processors supporting the POWER architecture.
5056 You use these options to specify which instructions are available on the
5057 processor you are using. The default value of these options is
5058 determined when configuring GCC. Specifying the
5059 @samp{-mcpu=@var{cpu_type}} overrides the specification of these
5060 options. We recommend you use the @samp{-mcpu=@var{cpu_type}} option
5061 rather than the options listed above.
5063 The @samp{-mpower} option allows GCC to generate instructions that
5064 are found only in the POWER architecture and to use the MQ register.
5065 Specifying @samp{-mpower2} implies @samp{-power} and also allows GCC
5066 to generate instructions that are present in the POWER2 architecture but
5067 not the original POWER architecture.
5069 The @samp{-mpowerpc} option allows GCC to generate instructions that
5070 are found only in the 32-bit subset of the PowerPC architecture.
5071 Specifying @samp{-mpowerpc-gpopt} implies @samp{-mpowerpc} and also allows
5072 GCC to use the optional PowerPC architecture instructions in the
5073 General Purpose group, including floating-point square root. Specifying
5074 @samp{-mpowerpc-gfxopt} implies @samp{-mpowerpc} and also allows GCC to
5075 use the optional PowerPC architecture instructions in the Graphics
5076 group, including floating-point select.
5078 The @samp{-mpowerpc64} option allows GCC to generate the additional
5079 64-bit instructions that are found in the full PowerPC64 architecture
5080 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
5081 @samp{-mno-powerpc64}.
5083 If you specify both @samp{-mno-power} and @samp{-mno-powerpc}, GCC
5084 will use only the instructions in the common subset of both
5085 architectures plus some special AIX common-mode calls, and will not use
5086 the MQ register. Specifying both @samp{-mpower} and @samp{-mpowerpc}
5087 permits GCC to use any instruction from either architecture and to
5088 allow use of the MQ register; specify this for the Motorola MPC601.
5090 @item -mnew-mnemonics
5091 @itemx -mold-mnemonics
5092 @kindex -mnew-mnemonics
5093 @kindex -mold-mnemonics
5094 Select which mnemonics to use in the generated assembler code.
5095 @samp{-mnew-mnemonics} requests output that uses the assembler mnemonics
5096 defined for the PowerPC architecture, while @samp{-mold-mnemonics}
5097 requests the assembler mnemonics defined for the POWER architecture.
5098 Instructions defined in only one architecture have only one mnemonic;
5099 GCC uses that mnemonic irrespective of which of these options is
5102 GCC defaults to the mnemonics appropriate for the architecture in
5103 use. Specifying @samp{-mcpu=@var{cpu_type}} sometimes overrides the
5104 value of these option. Unless you are building a cross-compiler, you
5105 should normally not specify either @samp{-mnew-mnemonics} or
5106 @samp{-mold-mnemonics}, but should instead accept the default.
5108 @item -mcpu=@var{cpu_type}
5110 Set architecture type, register usage, choice of mnemonics, and
5111 instruction scheduling parameters for machine type @var{cpu_type}.
5112 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
5113 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
5114 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
5115 @samp{630}, @samp{740}, @samp{750}, @samp{power}, @samp{power2},
5116 @samp{powerpc}, @samp{403}, @samp{505}, @samp{801}, @samp{821},
5117 @samp{823}, and @samp{860} and @samp{common}. @samp{-mcpu=power},
5118 @samp{-mcpu=power2}, @samp{-mcpu=powerpc}, and @samp{-mcpu=powerpc64}
5119 specify generic POWER, POWER2, pure 32-bit PowerPC (i.e., not MPC601),
5120 and 64-bit PowerPC architecture machine types, with an appropriate,
5121 generic processor model assumed for scheduling purposes.@refill
5123 Specifying any of the following options:
5124 @samp{-mcpu=rios1}, @samp{-mcpu=rios2}, @samp{-mcpu=rsc},
5125 @samp{-mcpu=power}, or @samp{-mcpu=power2}
5126 enables the @samp{-mpower} option and disables the @samp{-mpowerpc} option;
5127 @samp{-mcpu=601} enables both the @samp{-mpower} and @samp{-mpowerpc} options.
5128 All of @samp{-mcpu=rs64a}, @samp{-mcpu=602}, @samp{-mcpu=603},
5129 @samp{-mcpu=603e}, @samp{-mcpu=604}, @samp{-mcpu=620}, @samp{-mcpu=630},
5130 @samp{-mcpu=740}, and @samp{-mcpu=750}
5131 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5132 Exactly similarly, all of @samp{-mcpu=403},
5133 @samp{-mcpu=505}, @samp{-mcpu=821}, @samp{-mcpu=860} and @samp{-mcpu=powerpc}
5134 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5135 @samp{-mcpu=common} disables both the
5136 @samp{-mpower} and @samp{-mpowerpc} options.@refill
5138 AIX versions 4 or greater selects @samp{-mcpu=common} by default, so
5139 that code will operate on all members of the RS/6000 POWER and PowerPC
5140 families. In that case, GCC will use only the instructions in the
5141 common subset of both architectures plus some special AIX common-mode
5142 calls, and will not use the MQ register. GCC assumes a generic
5143 processor model for scheduling purposes.
5145 Specifying any of the options @samp{-mcpu=rios1}, @samp{-mcpu=rios2},
5146 @samp{-mcpu=rsc}, @samp{-mcpu=power}, or @samp{-mcpu=power2} also
5147 disables the @samp{new-mnemonics} option. Specifying @samp{-mcpu=601},
5148 @samp{-mcpu=602}, @samp{-mcpu=603}, @samp{-mcpu=603e}, @samp{-mcpu=604},
5149 @samp{-mcpu=620}, @samp{-mcpu=630}, @samp{-mcpu=403}, @samp{-mcpu=505},
5150 @samp{-mcpu=821}, @samp{-mcpu=860} or @samp{-mcpu=powerpc} also enables
5151 the @samp{new-mnemonics} option.@refill
5153 Specifying @samp{-mcpu=403}, @samp{-mcpu=821}, or @samp{-mcpu=860} also
5154 enables the @samp{-msoft-float} option.
5156 @item -mtune=@var{cpu_type}
5157 Set the instruction scheduling parameters for machine type
5158 @var{cpu_type}, but do not set the architecture type, register usage,
5159 choice of mnemonics like @samp{-mcpu=}@var{cpu_type} would. The same
5160 values for @var{cpu_type} are used for @samp{-mtune=}@var{cpu_type} as
5161 for @samp{-mcpu=}@var{cpu_type}. The @samp{-mtune=}@var{cpu_type}
5162 option overrides the @samp{-mcpu=}@var{cpu_type} option in terms of
5163 instruction scheduling parameters.
5166 @itemx -mno-fp-in-toc
5167 @itemx -mno-sum-in-toc
5168 @itemx -mminimal-toc
5169 @kindex -mminimal-toc
5170 Modify generation of the TOC (Table Of Contents), which is created for
5171 every executable file. The @samp{-mfull-toc} option is selected by
5172 default. In that case, GCC will allocate at least one TOC entry for
5173 each unique non-automatic variable reference in your program. GCC
5174 will also place floating-point constants in the TOC. However, only
5175 16,384 entries are available in the TOC.
5177 If you receive a linker error message that saying you have overflowed
5178 the available TOC space, you can reduce the amount of TOC space used
5179 with the @samp{-mno-fp-in-toc} and @samp{-mno-sum-in-toc} options.
5180 @samp{-mno-fp-in-toc} prevents GCC from putting floating-point
5181 constants in the TOC and @samp{-mno-sum-in-toc} forces GCC to
5182 generate code to calculate the sum of an address and a constant at
5183 run-time instead of putting that sum into the TOC. You may specify one
5184 or both of these options. Each causes GCC to produce very slightly
5185 slower and larger code at the expense of conserving TOC space.
5187 If you still run out of space in the TOC even when you specify both of
5188 these options, specify @samp{-mminimal-toc} instead. This option causes
5189 GCC to make only one TOC entry for every file. When you specify this
5190 option, GCC will produce code that is slower and larger but which
5191 uses extremely little TOC space. You may wish to use this option
5192 only on files that contain less frequently executed code. @refill
5198 Enable 64-bit PowerPC ABI and calling convention: 64-bit pointers, 64-bit
5199 @code{long} type, and the infrastructure needed to support them.
5200 Specifying @samp{-m64} implies @samp{-mpowerpc64} and
5201 @samp{-mpowerpc}, while @samp{-m32} disables the 64-bit ABI and
5202 implies @samp{-mno-powerpc64}. GCC defaults to @samp{-m32}.
5207 On AIX, pass floating-point arguments to prototyped functions beyond the
5208 register save area (RSA) on the stack in addition to argument FPRs. The
5209 AIX calling convention was extended but not initially documented to
5210 handle an obscure K&R C case of calling a function that takes the
5211 address of its arguments with fewer arguments than declared. AIX XL
5212 compilers access floating point arguments which do not fit in the
5213 RSA from the stack when a subroutine is compiled without
5214 optimization. Because always storing floating-point arguments on the
5215 stack is inefficient and rarely needed, this option is not enabled by
5216 default and only is necessary when calling subroutines compiled by AIX
5217 XL compilers without optimization.
5221 Support @dfn{AIX Threads}. Link an application written to use
5222 @dfn{pthreads} with special libraries and startup code to enable the
5227 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE). Link an
5228 application written to use message passing with special startup code to
5229 enable the application to run. The system must have PE installed in the
5230 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
5231 must be overridden with the @samp{-specs=} option to specify the
5232 appropriate directory location. The Parallel Environment does not
5233 support threads, so the @samp{-mpe} option and the @samp{-mthreads}
5234 option are incompatible.
5238 @kindex -msoft-float
5239 Generate code that does not use (uses) the floating-point register set.
5240 Software floating point emulation is provided if you use the
5241 @samp{-msoft-float} option, and pass the option to GCC when linking.
5244 @itemx -mno-multiple
5245 Generate code that uses (does not use) the load multiple word
5246 instructions and the store multiple word instructions. These
5247 instructions are generated by default on POWER systems, and not
5248 generated on PowerPC systems. Do not use @samp{-mmultiple} on little
5249 endian PowerPC systems, since those instructions do not work when the
5250 processor is in little endian mode. The exceptions are PPC740 and
5251 PPC750 which permit the instructions usage in little endian mode.
5256 Generate code that uses (does not use) the load string instructions
5257 and the store string word instructions to save multiple registers and
5258 do small block moves. These instructions are generated by default on
5259 POWER systems, and not generated on PowerPC systems. Do not use
5260 @samp{-mstring} on little endian PowerPC systems, since those
5261 instructions do not work when the processor is in little endian mode.
5262 The exceptions are PPC740 and PPC750 which permit the instructions
5263 usage in little endian mode.
5268 Generate code that uses (does not use) the load or store instructions
5269 that update the base register to the address of the calculated memory
5270 location. These instructions are generated by default. If you use
5271 @samp{-mno-update}, there is a small window between the time that the
5272 stack pointer is updated and the address of the previous frame is
5273 stored, which means code that walks the stack frame across interrupts or
5274 signals may get corrupted data.
5277 @itemx -mno-fused-madd
5278 @kindex -mfused-madd
5279 Generate code that uses (does not use) the floating point multiply and
5280 accumulate instructions. These instructions are generated by default if
5281 hardware floating is used.
5283 @item -mno-bit-align
5286 On System V.4 and embedded PowerPC systems do not (do) force structures
5287 and unions that contain bit fields to be aligned to the base type of the
5290 For example, by default a structure containing nothing but 8
5291 @code{unsigned} bitfields of length 1 would be aligned to a 4 byte
5292 boundary and have a size of 4 bytes. By using @samp{-mno-bit-align},
5293 the structure would be aligned to a 1 byte boundary and be one byte in
5296 @item -mno-strict-align
5297 @itemx -mstrict-align
5298 @kindex -mstrict-align
5299 On System V.4 and embedded PowerPC systems do not (do) assume that
5300 unaligned memory references will be handled by the system.
5303 @itemx -mno-relocatable
5304 @kindex -mrelocatable
5305 On embedded PowerPC systems generate code that allows (does not allow)
5306 the program to be relocated to a different address at runtime. If you
5307 use @samp{-mrelocatable} on any module, all objects linked together must
5308 be compiled with @samp{-mrelocatable} or @samp{-mrelocatable-lib}.
5310 @item -mrelocatable-lib
5311 @itemx -mno-relocatable-lib
5312 On embedded PowerPC systems generate code that allows (does not allow)
5313 the program to be relocated to a different address at runtime. Modules
5314 compiled with @samp{-mrelocatable-lib} can be linked with either modules
5315 compiled without @samp{-mrelocatable} and @samp{-mrelocatable-lib} or
5316 with modules compiled with the @samp{-mrelocatable} options.
5320 On System V.4 and embedded PowerPC systems do not (do) assume that
5321 register 2 contains a pointer to a global area pointing to the addresses
5322 used in the program.
5325 @itemx -mlittle-endian
5326 On System V.4 and embedded PowerPC systems compile code for the
5327 processor in little endian mode. The @samp{-mlittle-endian} option is
5328 the same as @samp{-mlittle}.
5332 On System V.4 and embedded PowerPC systems compile code for the
5333 processor in big endian mode. The @samp{-mbig-endian} option is
5334 the same as @samp{-mbig}.
5337 On System V.4 and embedded PowerPC systems compile code using calling
5338 conventions that adheres to the March 1995 draft of the System V
5339 Application Binary Interface, PowerPC processor supplement. This is the
5340 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
5342 @item -mcall-sysv-eabi
5343 Specify both @samp{-mcall-sysv} and @samp{-meabi} options.
5345 @item -mcall-sysv-noeabi
5346 Specify both @samp{-mcall-sysv} and @samp{-mno-eabi} options.
5349 On System V.4 and embedded PowerPC systems compile code using calling
5350 conventions that are similar to those used on AIX. This is the
5351 default if you configured GCC using @samp{powerpc-*-eabiaix}.
5353 @item -mcall-solaris
5354 On System V.4 and embedded PowerPC systems compile code for the Solaris
5358 On System V.4 and embedded PowerPC systems compile code for the
5359 Linux-based GNU system.
5362 @itemx -mno-prototype
5363 On System V.4 and embedded PowerPC systems assume that all calls to
5364 variable argument functions are properly prototyped. Otherwise, the
5365 compiler must insert an instruction before every non prototyped call to
5366 set or clear bit 6 of the condition code register (@var{CR}) to
5367 indicate whether floating point values were passed in the floating point
5368 registers in case the function takes a variable arguments. With
5369 @samp{-mprototype}, only calls to prototyped variable argument functions
5370 will set or clear the bit.
5373 On embedded PowerPC systems, assume that the startup module is called
5374 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
5375 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
5379 On embedded PowerPC systems, assume that the startup module is called
5380 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
5384 On embedded PowerPC systems, assume that the startup module is called
5385 @file{crt0.o} and the standard C libraries are @file{libads.a} and
5389 On embedded PowerPC systems, assume that the startup module is called
5390 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
5394 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
5395 header to indicate that @samp{eabi} extended relocations are used.
5399 On System V.4 and embedded PowerPC systems do (do not) adhere to the
5400 Embedded Applications Binary Interface (eabi) which is a set of
5401 modifications to the System V.4 specifications. Selecting @code{-meabi}
5402 means that the stack is aligned to an 8 byte boundary, a function
5403 @code{__eabi} is called to from @code{main} to set up the eabi
5404 environment, and the @samp{-msdata} option can use both @code{r2} and
5405 @code{r13} to point to two separate small data areas. Selecting
5406 @code{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
5407 do not call an initialization function from @code{main}, and the
5408 @samp{-msdata} option will only use @code{r13} to point to a single
5409 small data area. The @samp{-meabi} option is on by default if you
5410 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
5413 On System V.4 and embedded PowerPC systems, put small initialized
5414 @code{const} global and static data in the @samp{.sdata2} section, which
5415 is pointed to by register @code{r2}. Put small initialized
5416 non-@code{const} global and static data in the @samp{.sdata} section,
5417 which is pointed to by register @code{r13}. Put small uninitialized
5418 global and static data in the @samp{.sbss} section, which is adjacent to
5419 the @samp{.sdata} section. The @samp{-msdata=eabi} option is
5420 incompatible with the @samp{-mrelocatable} option. The
5421 @samp{-msdata=eabi} option also sets the @samp{-memb} option.
5424 On System V.4 and embedded PowerPC systems, put small global and static
5425 data in the @samp{.sdata} section, which is pointed to by register
5426 @code{r13}. Put small uninitialized global and static data in the
5427 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
5428 The @samp{-msdata=sysv} option is incompatible with the
5429 @samp{-mrelocatable} option.
5431 @item -msdata=default
5433 On System V.4 and embedded PowerPC systems, if @samp{-meabi} is used,
5434 compile code the same as @samp{-msdata=eabi}, otherwise compile code the
5435 same as @samp{-msdata=sysv}.
5438 On System V.4 and embedded PowerPC systems, put small global and static
5439 data in the @samp{.sdata} section. Put small uninitialized global and
5440 static data in the @samp{.sbss} section. Do not use register @code{r13}
5441 to address small data however. This is the default behavior unless
5442 other @samp{-msdata} options are used.
5446 On embedded PowerPC systems, put all initialized global and static data
5447 in the @samp{.data} section, and all uninitialized data in the
5448 @samp{.bss} section.
5451 @cindex smaller data references (PowerPC)
5452 @cindex .sdata/.sdata2 references (PowerPC)
5453 On embedded PowerPC systems, put global and static items less than or
5454 equal to @var{num} bytes into the small data or bss sections instead of
5455 the normal data or bss section. By default, @var{num} is 8. The
5456 @samp{-G @var{num}} switch is also passed to the linker.
5457 All modules should be compiled with the same @samp{-G @var{num}} value.
5460 @itemx -mno-regnames
5461 On System V.4 and embedded PowerPC systems do (do not) emit register
5462 names in the assembly language output using symbolic forms.
5467 @subsection IBM RT Options
5469 @cindex IBM RT options
5471 These @samp{-m} options are defined for the IBM RT PC:
5475 Use an in-line code sequence for integer multiplies. This is the
5478 @item -mcall-lib-mul
5479 Call @code{lmul$$} for integer multiples.
5481 @item -mfull-fp-blocks
5482 Generate full-size floating point data blocks, including the minimum
5483 amount of scratch space recommended by IBM. This is the default.
5485 @item -mminimum-fp-blocks
5486 Do not include extra scratch space in floating point data blocks. This
5487 results in smaller code, but slower execution, since scratch space must
5488 be allocated dynamically.
5490 @cindex @file{varargs.h} and RT PC
5491 @cindex @file{stdarg.h} and RT PC
5492 @item -mfp-arg-in-fpregs
5493 Use a calling sequence incompatible with the IBM calling convention in
5494 which floating point arguments are passed in floating point registers.
5495 Note that @code{varargs.h} and @code{stdargs.h} will not work with
5496 floating point operands if this option is specified.
5498 @item -mfp-arg-in-gregs
5499 Use the normal calling convention for floating point arguments. This is
5502 @item -mhc-struct-return
5503 Return structures of more than one word in memory, rather than in a
5504 register. This provides compatibility with the MetaWare HighC (hc)
5505 compiler. Use the option @samp{-fpcc-struct-return} for compatibility
5506 with the Portable C Compiler (pcc).
5508 @item -mnohc-struct-return
5509 Return some structures of more than one word in registers, when
5510 convenient. This is the default. For compatibility with the
5511 IBM-supplied compilers, use the option @samp{-fpcc-struct-return} or the
5512 option @samp{-mhc-struct-return}.
5516 @subsection MIPS Options
5517 @cindex MIPS options
5519 These @samp{-m} options are defined for the MIPS family of computers:
5522 @item -mcpu=@var{cpu type}
5523 Assume the defaults for the machine type @var{cpu type} when scheduling
5524 instructions. The choices for @var{cpu type} are @samp{r2000}, @samp{r3000},
5525 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
5526 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
5527 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
5528 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
5529 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
5530 @var{cpu type} will schedule things appropriately for that particular
5531 chip, the compiler will not generate any code that does not meet level 1
5532 of the MIPS ISA (instruction set architecture) without a @samp{-mipsX}
5533 or @samp{-mabi} switch being used.
5536 Issue instructions from level 1 of the MIPS ISA. This is the default.
5537 @samp{r3000} is the default @var{cpu type} at this ISA level.
5540 Issue instructions from level 2 of the MIPS ISA (branch likely, square
5541 root instructions). @samp{r6000} is the default @var{cpu type} at this
5545 Issue instructions from level 3 of the MIPS ISA (64 bit instructions).
5546 @samp{r4000} is the default @var{cpu type} at this ISA level.
5549 Issue instructions from level 4 of the MIPS ISA (conditional move,
5550 prefetch, enhanced FPU instructions). @samp{r8000} is the default
5551 @var{cpu type} at this ISA level.
5554 Assume that 32 32-bit floating point registers are available. This is
5558 Assume that 32 64-bit floating point registers are available. This is
5559 the default when the @samp{-mips3} option is used.
5562 Assume that 32 32-bit general purpose registers are available. This is
5566 Assume that 32 64-bit general purpose registers are available. This is
5567 the default when the @samp{-mips3} option is used.
5570 Force int and long types to be 64 bits wide. See @samp{-mlong32} for an
5571 explanation of the default, and the width of pointers.
5574 Force long types to be 64 bits wide. See @samp{-mlong32} for an
5575 explanation of the default, and the width of pointers.
5578 Force long, int, and pointer types to be 32 bits wide.
5580 If none of @samp{-mlong32}, @samp{-mlong64}, or @samp{-mint64} are set,
5581 the size of ints, longs, and pointers depends on the ABI and ISA choosen.
5582 For @samp{-mabi=32}, and @samp{-mabi=n32}, ints and longs are 32 bits
5583 wide. For @samp{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
5584 For @samp{-mabi=eabi} and either @samp{-mips1} or @samp{-mips2}, ints
5585 and longs are 32 bits wide. For @samp{-mabi=eabi} and higher ISAs, ints
5586 are 32 bits, and longs are 64 bits wide. The width of pointer types is
5587 the smaller of the width of longs or the width of general purpose
5588 registers (which in turn depends on the ISA).
5595 Generate code for the indicated ABI. The default instruction level is
5596 @samp{-mips1} for @samp{32}, @samp{-mips3} for @samp{n32}, and
5597 @samp{-mips4} otherwise. Conversely, with @samp{-mips1} or
5598 @samp{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
5602 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
5603 add normal debug information. This is the default for all
5604 platforms except for the OSF/1 reference platform, using the OSF/rose
5605 object format. If the either of the @samp{-gstabs} or @samp{-gstabs+}
5606 switches are used, the @file{mips-tfile} program will encapsulate the
5607 stabs within MIPS ECOFF.
5610 Generate code for the GNU assembler. This is the default on the OSF/1
5611 reference platform, using the OSF/rose object format. Also, this is
5612 the default if the configure option @samp{--with-gnu-as} is used.
5614 @item -msplit-addresses
5615 @itemx -mno-split-addresses
5616 Generate code to load the high and low parts of address constants separately.
5617 This allows @code{gcc} to optimize away redundant loads of the high order
5618 bits of addresses. This optimization requires GNU as and GNU ld.
5619 This optimization is enabled by default for some embedded targets where
5620 GNU as and GNU ld are standard.
5624 The @samp{-mrnames} switch says to output code using the MIPS software
5625 names for the registers, instead of the hardware names (ie, @var{a0}
5626 instead of @var{$4}). The only known assembler that supports this option
5627 is the Algorithmics assembler.
5631 The @samp{-mgpopt} switch says to write all of the data declarations
5632 before the instructions in the text section, this allows the MIPS
5633 assembler to generate one word memory references instead of using two
5634 words for short global or static data items. This is on by default if
5635 optimization is selected.
5639 For each non-inline function processed, the @samp{-mstats} switch
5640 causes the compiler to emit one line to the standard error file to
5641 print statistics about the program (number of registers saved, stack
5646 The @samp{-mmemcpy} switch makes all block moves call the appropriate
5647 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
5648 generating inline code.
5651 @itemx -mno-mips-tfile
5652 The @samp{-mno-mips-tfile} switch causes the compiler not
5653 postprocess the object file with the @file{mips-tfile} program,
5654 after the MIPS assembler has generated it to add debug support. If
5655 @file{mips-tfile} is not run, then no local variables will be
5656 available to the debugger. In addition, @file{stage2} and
5657 @file{stage3} objects will have the temporary file names passed to the
5658 assembler embedded in the object file, which means the objects will
5659 not compare the same. The @samp{-mno-mips-tfile} switch should only
5660 be used when there are bugs in the @file{mips-tfile} program that
5661 prevents compilation.
5664 Generate output containing library calls for floating point.
5665 @strong{Warning:} the requisite libraries are not part of GCC.
5666 Normally the facilities of the machine's usual C compiler are used, but
5667 this can't be done directly in cross-compilation. You must make your
5668 own arrangements to provide suitable library functions for
5672 Generate output containing floating point instructions. This is the
5673 default if you use the unmodified sources.
5676 @itemx -mno-abicalls
5677 Emit (or do not emit) the pseudo operations @samp{.abicalls},
5678 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
5679 position independent code.
5682 @itemx -mno-long-calls
5683 Do all calls with the @samp{JALR} instruction, which requires
5684 loading up a function's address into a register before the call.
5685 You need to use this switch, if you call outside of the current
5686 512 megabyte segment to functions that are not through pointers.
5689 @itemx -mno-half-pic
5690 Put pointers to extern references into the data section and load them
5691 up, rather than put the references in the text section.
5693 @item -membedded-pic
5694 @itemx -mno-embedded-pic
5695 Generate PIC code suitable for some embedded systems. All calls are
5696 made using PC relative address, and all data is addressed using the $gp
5697 register. No more than 65536 bytes of global data may be used. This
5698 requires GNU as and GNU ld which do most of the work. This currently
5699 only works on targets which use ECOFF; it does not work with ELF.
5701 @item -membedded-data
5702 @itemx -mno-embedded-data
5703 Allocate variables to the read-only data section first if possible, then
5704 next in the small data section if possible, otherwise in data. This gives
5705 slightly slower code than the default, but reduces the amount of RAM required
5706 when executing, and thus may be preferred for some embedded systems.
5708 @item -muninit-const-in-rodata
5709 @itemx -mno-uninit-const-in-rodata
5710 When used together with -membedded-data, it will always store uninitialized
5711 const variables in the read-only data section.
5713 @item -msingle-float
5714 @itemx -mdouble-float
5715 The @samp{-msingle-float} switch tells gcc to assume that the floating
5716 point coprocessor only supports single precision operations, as on the
5717 @samp{r4650} chip. The @samp{-mdouble-float} switch permits gcc to use
5718 double precision operations. This is the default.
5722 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
5723 as on the @samp{r4650} chip.
5726 Turns on @samp{-msingle-float}, @samp{-mmad}, and, at least for now,
5731 Enable 16-bit instructions.
5734 Use the entry and exit pseudo ops. This option can only be used with
5738 Compile code for the processor in little endian mode.
5739 The requisite libraries are assumed to exist.
5742 Compile code for the processor in big endian mode.
5743 The requisite libraries are assumed to exist.
5746 @cindex smaller data references (MIPS)
5747 @cindex gp-relative references (MIPS)
5748 Put global and static items less than or equal to @var{num} bytes into
5749 the small data or bss sections instead of the normal data or bss
5750 section. This allows the assembler to emit one word memory reference
5751 instructions based on the global pointer (@var{gp} or @var{$28}),
5752 instead of the normal two words used. By default, @var{num} is 8 when
5753 the MIPS assembler is used, and 0 when the GNU assembler is used. The
5754 @samp{-G @var{num}} switch is also passed to the assembler and linker.
5755 All modules should be compiled with the same @samp{-G @var{num}}
5759 Tell the MIPS assembler to not run its preprocessor over user
5760 assembler files (with a @samp{.s} suffix) when assembling them.
5763 Do not include the default crt0.
5767 These options are defined by the macro
5768 @code{TARGET_SWITCHES} in the machine description. The default for the
5769 options is also defined by that macro, which enables you to change the
5774 @subsection Intel 386 Options
5775 @cindex i386 Options
5776 @cindex Intel 386 Options
5778 These @samp{-m} options are defined for the i386 family of computers:
5781 @item -mcpu=@var{cpu type}
5782 Assume the defaults for the machine type @var{cpu type} when scheduling
5783 instructions. The choices for @var{cpu type} are:
5785 @multitable @columnfractions .20 .20 .20 .20
5786 @item @samp{i386} @tab @samp{i486} @tab @samp{i586} @tab @samp{i686}
5787 @item @samp{pentium} @tab @samp{pentiumpro} @tab @samp{k6}
5790 While picking a specific @var{cpu type} will schedule things appropriately
5791 for that particular chip, the compiler will not generate any code that
5792 does not run on the i386 without the @samp{-march=@var{cpu type}} option
5793 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
5794 is equivalent to @samp{pentiumpro}. @samp{k6} is the AMD chip as
5795 opposed to the Intel ones.
5797 @item -march=@var{cpu type}
5798 Generate instructions for the machine type @var{cpu type}. The choices
5799 for @var{cpu type} are the same as for @samp{-mcpu}. Moreover,
5800 specifying @samp{-march=@var{cpu type}} implies @samp{-mcpu=@var{cpu type}}.
5806 Synonyms for -mcpu=i386, -mcpu=i486, -mcpu=pentium, and -mcpu=pentiumpro
5807 respectively. These synonyms are deprecated.
5811 Control whether or not the compiler uses IEEE floating point
5812 comparisons. These handle correctly the case where the result of a
5813 comparison is unordered.
5816 Generate output containing library calls for floating point.
5817 @strong{Warning:} the requisite libraries are not part of GCC.
5818 Normally the facilities of the machine's usual C compiler are used, but
5819 this can't be done directly in cross-compilation. You must make your
5820 own arrangements to provide suitable library functions for
5823 On machines where a function returns floating point results in the 80387
5824 register stack, some floating point opcodes may be emitted even if
5825 @samp{-msoft-float} is used.
5827 @item -mno-fp-ret-in-387
5828 Do not use the FPU registers for return values of functions.
5830 The usual calling convention has functions return values of types
5831 @code{float} and @code{double} in an FPU register, even if there
5832 is no FPU. The idea is that the operating system should emulate
5835 The option @samp{-mno-fp-ret-in-387} causes such values to be returned
5836 in ordinary CPU registers instead.
5838 @item -mno-fancy-math-387
5839 Some 387 emulators do not support the @code{sin}, @code{cos} and
5840 @code{sqrt} instructions for the 387. Specify this option to avoid
5841 generating those instructions. This option is the default on FreeBSD.
5842 As of revision 2.6.1, these instructions are not generated unless you
5843 also use the @samp{-ffast-math} switch.
5845 @item -malign-double
5846 @itemx -mno-align-double
5847 Control whether GCC aligns @code{double}, @code{long double}, and
5848 @code{long long} variables on a two word boundary or a one word
5849 boundary. Aligning @code{double} variables on a two word boundary will
5850 produce code that runs somewhat faster on a @samp{Pentium} at the
5851 expense of more memory.
5853 @strong{Warning:} if you use the @samp{-malign-double} switch,
5854 structures containing the above types will be aligned differently than
5855 the published application binary interface specifications for the 386.
5858 @itemx -mno-svr3-shlib
5859 Control whether GCC places uninitialized locals into @code{bss} or
5860 @code{data}. @samp{-msvr3-shlib} places these locals into @code{bss}.
5861 These options are meaningful only on System V Release 3.
5863 @item -mno-wide-multiply
5864 @itemx -mwide-multiply
5865 Control whether GCC uses the @code{mul} and @code{imul} that produce
5866 64 bit results in @code{eax:edx} from 32 bit operands to do @code{long
5867 long} multiplies and 32-bit division by constants.
5870 Use a different function-calling convention, in which functions that
5871 take a fixed number of arguments return with the @code{ret} @var{num}
5872 instruction, which pops their arguments while returning. This saves one
5873 instruction in the caller since there is no need to pop the arguments
5876 You can specify that an individual function is called with this calling
5877 sequence with the function attribute @samp{stdcall}. You can also
5878 override the @samp{-mrtd} option by using the function attribute
5879 @samp{cdecl}. @xref{Function Attributes}.
5881 @strong{Warning:} this calling convention is incompatible with the one
5882 normally used on Unix, so you cannot use it if you need to call
5883 libraries compiled with the Unix compiler.
5885 Also, you must provide function prototypes for all functions that
5886 take variable numbers of arguments (including @code{printf});
5887 otherwise incorrect code will be generated for calls to those
5890 In addition, seriously incorrect code will result if you call a
5891 function with too many arguments. (Normally, extra arguments are
5892 harmlessly ignored.)
5894 @item -mreg-alloc=@var{regs}
5895 Control the default allocation order of integer registers. The
5896 string @var{regs} is a series of letters specifying a register. The
5897 supported letters are: @code{a} allocate EAX; @code{b} allocate EBX;
5898 @code{c} allocate ECX; @code{d} allocate EDX; @code{S} allocate ESI;
5899 @code{D} allocate EDI; @code{B} allocate EBP.
5901 @item -mregparm=@var{num}
5902 Control how many registers are used to pass integer arguments. By
5903 default, no registers are used to pass arguments, and at most 3
5904 registers can be used. You can control this behavior for a specific
5905 function by using the function attribute @samp{regparm}.
5906 @xref{Function Attributes}.
5908 @strong{Warning:} if you use this switch, and
5909 @var{num} is nonzero, then you must build all modules with the same
5910 value, including any libraries. This includes the system libraries and
5913 @item -malign-loops=@var{num}
5914 Align loops to a 2 raised to a @var{num} byte boundary. If
5915 @samp{-malign-loops} is not specified, the default is 2 unless
5916 gas 2.8 (or later) is being used in which case the default is
5917 to align the loop on a 16 byte boundary if it is less than 8
5920 @item -malign-jumps=@var{num}
5921 Align instructions that are only jumped to to a 2 raised to a @var{num}
5922 byte boundary. If @samp{-malign-jumps} is not specified, the default is
5923 2 if optimizing for a 386, and 4 if optimizing for a 486 unless
5924 gas 2.8 (or later) is being used in which case the default is
5925 to align the instruction on a 16 byte boundary if it is less
5928 @item -malign-functions=@var{num}
5929 Align the start of functions to a 2 raised to @var{num} byte boundary.
5930 If @samp{-malign-functions} is not specified, the default is 2 if optimizing
5931 for a 386, and 4 if optimizing for a 486.
5933 @item -mpreferred-stack-boundary=@var{num}
5934 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
5935 byte boundary. If @samp{-mpreferred-stack-boundary} is not specified,
5936 the default is 4 (16 bytes or 128 bits).
5938 The stack is required to be aligned on a 4 byte boundary. On Pentium
5939 and PentiumPro, @code{double} and @code{long double} values should be
5940 aligned to an 8 byte boundary (see @samp{-malign-double}) or suffer
5941 significant run time performance penalties. On Pentium III, the
5942 Streaming SIMD Extention (SSE) data type @code{__m128} suffers similar
5943 penalties if it is not 16 byte aligned.
5945 To ensure proper alignment of this values on the stack, the stack boundary
5946 must be as aligned as that required by any value stored on the stack.
5947 Further, every function must be generated such that it keeps the stack
5948 aligned. Thus calling a function compiled with a higher preferred
5949 stack boundary from a function compiled with a lower preferred stack
5950 boundary will most likely misalign the stack. It is recommended that
5951 libraries that use callbacks always use the default setting.
5953 This extra alignment does consume extra stack space. Code that is sensitive
5954 to stack space usage, such as embedded systems and operating system kernels,
5955 may want to reduce the preferred alignment to
5956 @samp{-mpreferred-stack-boundary=2}.
5960 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
5961 on thread-safe exception handling must compile and link all code with the
5962 @samp{-mthreads} option. When compiling, @samp{-mthreads} defines
5963 @samp{-D_MT}; when linking, it links in a special thread helper library
5964 @samp{-lmingwthrd} which cleans up per thread exception handling data.
5966 @item -mno-align-stringops
5967 @kindex -mno-align-stringops
5968 Do not align destination of inlined string operations. This switch reduces
5969 code size and improves performance in case the destination is already aligned,
5970 but gcc don't know about it.
5972 @item -minline-all-stringops
5973 @kindex -minline-all-stringops
5974 By default GCC inlines string operations only when destination is known to be
5975 aligned at least to 4 byte boundary. This enables more inlining, increase code
5976 size, but may improve performance of code that depends on fast memcpy, strlen
5977 and memset for short lengths.
5981 @subsection HPPA Options
5982 @cindex HPPA Options
5984 These @samp{-m} options are defined for the HPPA family of computers:
5987 @item -march=@var{architecture type}
5988 Generate code for the specified architecture. The choices for
5989 @var{architecture type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
5990 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
5991 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
5992 architecture option for your machine. Code compiled for lower numbered
5993 architectures will run on higher numbered architectures, but not the
5996 PA 2.0 support currently requires gas snapshot 19990413 or later. The
5997 next release of binutils (current is 2.9.1) will probably contain PA 2.0
6001 @itemx -mpa-risc-1-1
6002 @itemx -mpa-risc-2-0
6003 Synonyms for -march=1.0, -march=1.1, and -march=2.0 respectively.
6006 Generate code suitable for big switch tables. Use this option only if
6007 the assembler/linker complain about out of range branches within a switch
6010 @item -mjump-in-delay
6011 Fill delay slots of function calls with unconditional jump instructions
6012 by modifying the return pointer for the function call to be the target
6013 of the conditional jump.
6015 @item -mdisable-fpregs
6016 Prevent floating point registers from being used in any manner. This is
6017 necessary for compiling kernels which perform lazy context switching of
6018 floating point registers. If you use this option and attempt to perform
6019 floating point operations, the compiler will abort.
6021 @item -mdisable-indexing
6022 Prevent the compiler from using indexing address modes. This avoids some
6023 rather obscure problems when compiling MIG generated code under MACH.
6025 @item -mno-space-regs
6026 Generate code that assumes the target has no space registers. This allows
6027 GCC to generate faster indirect calls and use unscaled index address modes.
6029 Such code is suitable for level 0 PA systems and kernels.
6031 @item -mfast-indirect-calls
6032 Generate code that assumes calls never cross space boundaries. This
6033 allows GCC to emit code which performs faster indirect calls.
6035 This option will not work in the presense of shared libraries or nested
6038 @item -mlong-load-store
6039 Generate 3-instruction load and store sequences as sometimes required by
6040 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
6043 @item -mportable-runtime
6044 Use the portable calling conventions proposed by HP for ELF systems.
6047 Enable the use of assembler directives only GAS understands.
6049 @item -mschedule=@var{cpu type}
6050 Schedule code according to the constraints for the machine type
6051 @var{cpu type}. The choices for @var{cpu type} are @samp{700}
6052 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
6053 @file{/usr/lib/sched.models} on an HP-UX system to determine the
6054 proper scheduling option for your machine.
6057 Enable the optimization pass in the HPUX linker. Note this makes symbolic
6058 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
6059 in which they give bogus error messages when linking some programs.
6062 Generate output containing library calls for floating point.
6063 @strong{Warning:} the requisite libraries are not available for all HPPA
6064 targets. Normally the facilities of the machine's usual C compiler are
6065 used, but this cannot be done directly in cross-compilation. You must make
6066 your own arrangements to provide suitable library functions for
6067 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
6068 does provide software floating point support.
6070 @samp{-msoft-float} changes the calling convention in the output file;
6071 therefore, it is only useful if you compile @emph{all} of a program with
6072 this option. In particular, you need to compile @file{libgcc.a}, the
6073 library that comes with GCC, with @samp{-msoft-float} in order for
6077 @node Intel 960 Options
6078 @subsection Intel 960 Options
6080 These @samp{-m} options are defined for the Intel 960 implementations:
6083 @item -m@var{cpu type}
6084 Assume the defaults for the machine type @var{cpu type} for some of
6085 the other options, including instruction scheduling, floating point
6086 support, and addressing modes. The choices for @var{cpu type} are
6087 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
6088 @samp{sa}, and @samp{sb}.
6094 The @samp{-mnumerics} option indicates that the processor does support
6095 floating-point instructions. The @samp{-msoft-float} option indicates
6096 that floating-point support should not be assumed.
6098 @item -mleaf-procedures
6099 @itemx -mno-leaf-procedures
6100 Do (or do not) attempt to alter leaf procedures to be callable with the
6101 @code{bal} instruction as well as @code{call}. This will result in more
6102 efficient code for explicit calls when the @code{bal} instruction can be
6103 substituted by the assembler or linker, but less efficient code in other
6104 cases, such as calls via function pointers, or using a linker that doesn't
6105 support this optimization.
6108 @itemx -mno-tail-call
6109 Do (or do not) make additional attempts (beyond those of the
6110 machine-independent portions of the compiler) to optimize tail-recursive
6111 calls into branches. You may not want to do this because the detection of
6112 cases where this is not valid is not totally complete. The default is
6113 @samp{-mno-tail-call}.
6115 @item -mcomplex-addr
6116 @itemx -mno-complex-addr
6117 Assume (or do not assume) that the use of a complex addressing mode is a
6118 win on this implementation of the i960. Complex addressing modes may not
6119 be worthwhile on the K-series, but they definitely are on the C-series.
6120 The default is currently @samp{-mcomplex-addr} for all processors except
6124 @itemx -mno-code-align
6125 Align code to 8-byte boundaries for faster fetching (or don't bother).
6126 Currently turned on by default for C-series implementations only.
6129 @item -mclean-linkage
6130 @itemx -mno-clean-linkage
6131 These options are not fully implemented.
6135 @itemx -mic2.0-compat
6136 @itemx -mic3.0-compat
6137 Enable compatibility with iC960 v2.0 or v3.0.
6141 Enable compatibility with the iC960 assembler.
6143 @item -mstrict-align
6144 @itemx -mno-strict-align
6145 Do not permit (do permit) unaligned accesses.
6148 Enable structure-alignment compatibility with Intel's gcc release version
6149 1.3 (based on gcc 1.37). This option implies @samp{-mstrict-align}.
6151 @item -mlong-double-64
6152 Implement type @samp{long double} as 64-bit floating point numbers.
6153 Without the option @samp{long double} is implemented by 80-bit
6154 floating point numbers. The only reason we have it because there is
6155 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
6156 is only useful for people using soft-float targets. Otherwise, we
6157 should recommend against use of it.
6161 @node DEC Alpha Options
6162 @subsection DEC Alpha Options
6164 These @samp{-m} options are defined for the DEC Alpha implementations:
6167 @item -mno-soft-float
6169 Use (do not use) the hardware floating-point instructions for
6170 floating-point operations. When @code{-msoft-float} is specified,
6171 functions in @file{libgcc1.c} will be used to perform floating-point
6172 operations. Unless they are replaced by routines that emulate the
6173 floating-point operations, or compiled in such a way as to call such
6174 emulations routines, these routines will issue floating-point
6175 operations. If you are compiling for an Alpha without floating-point
6176 operations, you must ensure that the library is built so as not to call
6179 Note that Alpha implementations without floating-point operations are
6180 required to have floating-point registers.
6184 Generate code that uses (does not use) the floating-point register set.
6185 @code{-mno-fp-regs} implies @code{-msoft-float}. If the floating-point
6186 register set is not used, floating point operands are passed in integer
6187 registers as if they were integers and floating-point results are passed
6188 in $0 instead of $f0. This is a non-standard calling sequence, so any
6189 function with a floating-point argument or return value called by code
6190 compiled with @code{-mno-fp-regs} must also be compiled with that
6193 A typical use of this option is building a kernel that does not use,
6194 and hence need not save and restore, any floating-point registers.
6197 The Alpha architecture implements floating-point hardware optimized for
6198 maximum performance. It is mostly compliant with the IEEE floating
6199 point standard. However, for full compliance, software assistance is
6200 required. This option generates code fully IEEE compliant code
6201 @emph{except} that the @var{inexact flag} is not maintained (see below).
6202 If this option is turned on, the CPP macro @code{_IEEE_FP} is defined
6203 during compilation. The option is a shorthand for: @samp{-D_IEEE_FP
6204 -mfp-trap-mode=su -mtrap-precision=i -mieee-conformant}. The resulting
6205 code is less efficient but is able to correctly support denormalized
6206 numbers and exceptional IEEE values such as not-a-number and plus/minus
6207 infinity. Other Alpha compilers call this option
6208 @code{-ieee_with_no_inexact}.
6210 @item -mieee-with-inexact
6211 @c overfull hbox here --bob 22 jul96
6212 @c original text between ignore ... end ignore
6214 This is like @samp{-mieee} except the generated code also maintains the
6215 IEEE @var{inexact flag}. Turning on this option causes the generated
6216 code to implement fully-compliant IEEE math. The option is a shorthand
6217 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus @samp{-mieee-conformant},
6218 @samp{-mfp-trap-mode=sui}, and @samp{-mtrap-precision=i}. On some Alpha
6219 implementations the resulting code may execute significantly slower than
6220 the code generated by default. Since there is very little code that
6221 depends on the @var{inexact flag}, you should normally not specify this
6222 option. Other Alpha compilers call this option
6223 @samp{-ieee_with_inexact}.
6225 @c changed paragraph
6226 This is like @samp{-mieee} except the generated code also maintains the
6227 IEEE @var{inexact flag}. Turning on this option causes the generated
6228 code to implement fully-compliant IEEE math. The option is a shorthand
6229 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus the three following:
6230 @samp{-mieee-conformant},
6231 @samp{-mfp-trap-mode=sui},
6232 and @samp{-mtrap-precision=i}.
6233 On some Alpha implementations the resulting code may execute
6234 significantly slower than the code generated by default. Since there
6235 is very little code that depends on the @var{inexact flag}, you should
6236 normally not specify this option. Other Alpha compilers call this
6237 option @samp{-ieee_with_inexact}.
6238 @c end changes to prevent overfull hboxes
6240 @item -mfp-trap-mode=@var{trap mode}
6241 This option controls what floating-point related traps are enabled.
6242 Other Alpha compilers call this option @samp{-fptm }@var{trap mode}.
6243 The trap mode can be set to one of four values:
6247 This is the default (normal) setting. The only traps that are enabled
6248 are the ones that cannot be disabled in software (e.g., division by zero
6252 In addition to the traps enabled by @samp{n}, underflow traps are enabled
6256 Like @samp{su}, but the instructions are marked to be safe for software
6257 completion (see Alpha architecture manual for details).
6260 Like @samp{su}, but inexact traps are enabled as well.
6263 @item -mfp-rounding-mode=@var{rounding mode}
6264 Selects the IEEE rounding mode. Other Alpha compilers call this option
6265 @samp{-fprm }@var{rounding mode}. The @var{rounding mode} can be one
6270 Normal IEEE rounding mode. Floating point numbers are rounded towards
6271 the nearest machine number or towards the even machine number in case
6275 Round towards minus infinity.
6278 Chopped rounding mode. Floating point numbers are rounded towards zero.
6281 Dynamic rounding mode. A field in the floating point control register
6282 (@var{fpcr}, see Alpha architecture reference manual) controls the
6283 rounding mode in effect. The C library initializes this register for
6284 rounding towards plus infinity. Thus, unless your program modifies the
6285 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
6288 @item -mtrap-precision=@var{trap precision}
6289 In the Alpha architecture, floating point traps are imprecise. This
6290 means without software assistance it is impossible to recover from a
6291 floating trap and program execution normally needs to be terminated.
6292 GCC can generate code that can assist operating system trap handlers
6293 in determining the exact location that caused a floating point trap.
6294 Depending on the requirements of an application, different levels of
6295 precisions can be selected:
6299 Program precision. This option is the default and means a trap handler
6300 can only identify which program caused a floating point exception.
6303 Function precision. The trap handler can determine the function that
6304 caused a floating point exception.
6307 Instruction precision. The trap handler can determine the exact
6308 instruction that caused a floating point exception.
6311 Other Alpha compilers provide the equivalent options called
6312 @samp{-scope_safe} and @samp{-resumption_safe}.
6314 @item -mieee-conformant
6315 This option marks the generated code as IEEE conformant. You must not
6316 use this option unless you also specify @samp{-mtrap-precision=i} and either
6317 @samp{-mfp-trap-mode=su} or @samp{-mfp-trap-mode=sui}. Its only effect
6318 is to emit the line @samp{.eflag 48} in the function prologue of the
6319 generated assembly file. Under DEC Unix, this has the effect that
6320 IEEE-conformant math library routines will be linked in.
6322 @item -mbuild-constants
6323 Normally GCC examines a 32- or 64-bit integer constant to
6324 see if it can construct it from smaller constants in two or three
6325 instructions. If it cannot, it will output the constant as a literal and
6326 generate code to load it from the data segment at runtime.
6328 Use this option to require GCC to construct @emph{all} integer constants
6329 using code, even if it takes more instructions (the maximum is six).
6331 You would typically use this option to build a shared library dynamic
6332 loader. Itself a shared library, it must relocate itself in memory
6333 before it can find the variables and constants in its own data segment.
6337 Select whether to generate code to be assembled by the vendor-supplied
6338 assembler (@samp{-malpha-as}) or by the GNU assembler @samp{-mgas}.
6346 Indicate whether GCC should generate code to use the optional BWX,
6347 CIX, and MAX instruction sets. The default is to use the instruction sets
6348 supported by the CPU type specified via @samp{-mcpu=} option or that
6349 of the CPU on which GCC was built if none was specified.
6351 @item -mcpu=@var{cpu_type}
6352 Set the instruction set, register set, and instruction scheduling
6353 parameters for machine type @var{cpu_type}. You can specify either the
6354 @samp{EV} style name or the corresponding chip number. GCC
6355 supports scheduling parameters for the EV4 and EV5 family of processors
6356 and will choose the default values for the instruction set from
6357 the processor you specify. If you do not specify a processor type,
6358 GCC will default to the processor on which the compiler was built.
6360 Supported values for @var{cpu_type} are
6365 Schedules as an EV4 and has no instruction set extensions.
6369 Schedules as an EV5 and has no instruction set extensions.
6373 Schedules as an EV5 and supports the BWX extension.
6378 Schedules as an EV5 and supports the BWX and MAX extensions.
6382 Schedules as an EV5 (until Digital releases the scheduling parameters
6383 for the EV6) and supports the BWX, CIX, and MAX extensions.
6386 @item -mmemory-latency=@var{time}
6387 Sets the latency the scheduler should assume for typical memory
6388 references as seen by the application. This number is highly
6389 dependant on the memory access patterns used by the application
6390 and the size of the external cache on the machine.
6392 Valid options for @var{time} are
6396 A decimal number representing clock cycles.
6402 The compiler contains estimates of the number of clock cycles for
6403 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
6404 (also called Dcache, Scache, and Bcache), as well as to main memory.
6405 Note that L3 is only valid for EV5.
6410 @node Clipper Options
6411 @subsection Clipper Options
6413 These @samp{-m} options are defined for the Clipper implementations:
6417 Produce code for a C300 Clipper processor. This is the default.
6420 Produce code for a C400 Clipper processor i.e. use floating point
6424 @node H8/300 Options
6425 @subsection H8/300 Options
6427 These @samp{-m} options are defined for the H8/300 implementations:
6431 Shorten some address references at link time, when possible; uses the
6432 linker option @samp{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
6433 ld.info, Using ld}, for a fuller description.
6436 Generate code for the H8/300H.
6439 Generate code for the H8/S.
6442 Make @code{int} data 32 bits by default.
6445 On the h8/300h, use the same alignment rules as for the h8/300.
6446 The default for the h8/300h is to align longs and floats on 4 byte boundaries.
6447 @samp{-malign-300} causes them to be aligned on 2 byte boundaries.
6448 This option has no effect on the h8/300.
6452 @subsection SH Options
6454 These @samp{-m} options are defined for the SH implementations:
6458 Generate code for the SH1.
6461 Generate code for the SH2.
6464 Generate code for the SH3.
6467 Generate code for the SH3e.
6470 Compile code for the processor in big endian mode.
6473 Compile code for the processor in little endian mode.
6476 Align doubles at 64 bit boundaries. Note that this changes the calling
6477 conventions, and thus some functions from the standard C library will
6478 not work unless you recompile it first with -mdalign.
6481 Shorten some address references at link time, when possible; uses the
6482 linker option @samp{-relax}.
6485 @node System V Options
6486 @subsection Options for System V
6488 These additional options are available on System V Release 4 for
6489 compatibility with other compilers on those systems:
6493 Create a shared object.
6494 It is recommended that @samp{-symbolic} or @samp{-shared} be used instead.
6497 Identify the versions of each tool used by the compiler, in a
6498 @code{.ident} assembler directive in the output.
6501 Refrain from adding @code{.ident} directives to the output file (this is
6504 @item -YP,@var{dirs}
6505 Search the directories @var{dirs}, and no others, for libraries
6506 specified with @samp{-l}.
6509 Look in the directory @var{dir} to find the M4 preprocessor.
6510 The assembler uses this option.
6511 @c This is supposed to go with a -Yd for predefined M4 macro files, but
6512 @c the generic assembler that comes with Solaris takes just -Ym.
6515 @node TMS320C3x/C4x Options
6516 @subsection TMS320C3x/C4x Options
6517 @cindex TMS320C3x/C4x Options
6519 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
6523 @item -mcpu=@var{cpu_type}
6524 Set the instruction set, register set, and instruction scheduling
6525 parameters for machine type @var{cpu_type}. Supported values for
6526 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
6527 @samp{c44}. The default is @samp{c40} to generate code for the
6532 @itemx -msmall-memory
6534 Generates code for the big or small memory model. The small memory
6535 model assumed that all data fits into one 64K word page. At run-time
6536 the data page (DP) register must be set to point to the 64K page
6537 containing the .bss and .data program sections. The big memory model is
6538 the default and requires reloading of the DP register for every direct
6543 Allow (disallow) allocation of general integer operands into the block
6548 Enable (disable) generation of code using decrement and branch,
6549 DBcond(D), instructions. This is enabled by default for the C4x. To be
6550 on the safe side, this is disabled for the C3x, since the maximum
6551 iteration count on the C3x is 2^23 + 1 (but who iterates loops more than
6552 2^23 times on the C3x?). Note that GCC will try to reverse a loop so
6553 that it can utilise the decrement and branch instruction, but will give
6554 up if there is more than one memory reference in the loop. Thus a loop
6555 where the loop counter is decremented can generate slightly more
6556 efficient code, in cases where the RPTB instruction cannot be utilised.
6558 @item -mdp-isr-reload
6560 Force the DP register to be saved on entry to an interrupt service
6561 routine (ISR), reloaded to point to the data section, and restored on
6562 exit from the ISR. This should not be required unless someone has
6563 violated the small memory model by modifying the DP register, say within
6568 For the C3x use the 24-bit MPYI instruction for integer multiplies
6569 instead of a library call to guarantee 32-bit results. Note that if one
6570 of the operands is a constant, then the multiplication will be performed
6571 using shifts and adds. If the -mmpyi option is not specified for the C3x,
6572 then squaring operations are performed inline instead of a library call.
6575 @itemx -mno-fast-fix
6576 The C3x/C4x FIX instruction to convert a floating point value to an
6577 integer value chooses the nearest integer less than or equal to the
6578 floating point value rather than to the nearest integer. Thus if the
6579 floating point number is negative, the result will be incorrectly
6580 truncated an additional code is necessary to detect and correct this
6581 case. This option can be used to disable generation of the additional
6582 code required to correct the result.
6586 Enable (disable) generation of repeat block sequences using the RPTB
6587 instruction for zero overhead looping. The RPTB construct is only used
6588 for innermost loops that do not call functions or jump across the loop
6589 boundaries. There is no advantage having nested RPTB loops due to the
6590 overhead required to save and restore the RC, RS, and RE registers.
6591 This is enabled by default with -O2.
6593 @item -mrpts=@var{count}
6595 Enable (disable) the use of the single instruction repeat instruction
6596 RPTS. If a repeat block contains a single instruction, and the loop
6597 count can be guaranteed to be less than the value @var{count}, GCC will
6598 emit a RPTS instruction instead of a RPTB. If no value is specified,
6599 then a RPTS will be emitted even if the loop count cannot be determined
6600 at compile time. Note that the repeated instruction following RPTS does
6601 not have to be reloaded from memory each iteration, thus freeing up the
6602 CPU buses for oeprands. However, since interrupts are blocked by this
6603 instruction, it is disabled by default.
6605 @item -mloop-unsigned
6606 @itemx -mno-loop-unsigned
6607 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
6608 is 2^31 + 1 since these instructions test if the iteration count is
6609 negative to terminate the loop. If the iteration count is unsigned
6610 there is a possibility than the 2^31 + 1 maximum iteration count may be
6611 exceeded. This switch allows an unsigned iteration count.
6614 Try to emit an assembler syntax that the TI assembler (asm30) is happy
6615 with. This also enforces compatibility with the API employed by the TI
6616 C3x C compiler. For example, long doubles are passed as structures
6617 rather than in floating point registers.
6621 Generate code that uses registers (stack) for passing arguments to functions.
6622 By default, arguments are passed in registers where possible rather
6623 than by pushing arguments on to the stack.
6625 @item -mparallel-insns
6626 @itemx -mno-parallel-insns
6627 Allow the generation of parallel instructions. This is enabled by
6630 @item -mparallel-mpy
6631 @itemx -mno-parallel-mpy
6632 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
6633 provided -mparallel-insns is also specified. These instructions have
6634 tight register constraints which can pessimize the code generation
6640 @subsection V850 Options
6641 @cindex V850 Options
6643 These @samp{-m} options are defined for V850 implementations:
6647 @itemx -mno-long-calls
6648 Treat all calls as being far away (near). If calls are assumed to be
6649 far away, the compiler will always load the functions address up into a
6650 register, and call indirect through the pointer.
6654 Do not optimize (do optimize) basic blocks that use the same index
6655 pointer 4 or more times to copy pointer into the @code{ep} register, and
6656 use the shorter @code{sld} and @code{sst} instructions. The @samp{-mep}
6657 option is on by default if you optimize.
6659 @item -mno-prolog-function
6660 @itemx -mprolog-function
6661 Do not use (do use) external functions to save and restore registers at
6662 the prolog and epilog of a function. The external functions are slower,
6663 but use less code space if more than one function saves the same number
6664 of registers. The @samp{-mprolog-function} option is on by default if
6668 Try to make the code as small as possible. At present, this just turns
6669 on the @samp{-mep} and @samp{-mprolog-function} options.
6672 Put static or global variables whose size is @var{n} bytes or less into
6673 the tiny data area that register @code{ep} points to. The tiny data
6674 area can hold up to 256 bytes in total (128 bytes for byte references).
6677 Put static or global variables whose size is @var{n} bytes or less into
6678 the small data area that register @code{gp} points to. The small data
6679 area can hold up to 64 kilobytes.
6682 Put static or global variables whose size is @var{n} bytes or less into
6683 the first 32 kilobytes of memory.
6686 Specify that the target processor is the V850.
6689 Generate code suitable for big switch tables. Use this option only if
6690 the assembler/linker complain about out of range branches within a switch
6695 @subsection ARC Options
6698 These options are defined for ARC implementations:
6702 Compile code for little endian mode. This is the default.
6705 Compile code for big endian mode.
6708 Prepend the name of the cpu to all public symbol names.
6709 In multiple-processor systems, there are many ARC variants with different
6710 instruction and register set characteristics. This flag prevents code
6711 compiled for one cpu to be linked with code compiled for another.
6712 No facility exists for handling variants that are "almost identical".
6713 This is an all or nothing option.
6715 @item -mcpu=@var{cpu}
6716 Compile code for ARC variant @var{cpu}.
6717 Which variants are supported depend on the configuration.
6718 All variants support @samp{-mcpu=base}, this is the default.
6720 @item -mtext=@var{text section}
6721 @itemx -mdata=@var{data section}
6722 @itemx -mrodata=@var{readonly data section}
6723 Put functions, data, and readonly data in @var{text section},
6724 @var{data section}, and @var{readonly data section} respectively
6725 by default. This can be overridden with the @code{section} attribute.
6726 @xref{Variable Attributes}.
6731 @subsection NS32K Options
6732 @cindex NS32K options
6734 These are the @samp{-m} options defined for the 32000 series. The default
6735 values for these options depends on which style of 32000 was selected when
6736 the compiler was configured; the defaults for the most common choices are
6742 Generate output for a 32032. This is the default
6743 when the compiler is configured for 32032 and 32016 based systems.
6747 Generate output for a 32332. This is the default
6748 when the compiler is configured for 32332-based systems.
6752 Generate output for a 32532. This is the default
6753 when the compiler is configured for 32532-based systems.
6756 Generate output containing 32081 instructions for floating point.
6757 This is the default for all systems.
6760 Generate output containing 32381 instructions for floating point. This
6761 also implies @samp{-m32081}. The 32381 is only compatible with the 32332
6762 and 32532 cpus. This is the default for the pc532-netbsd configuration.
6765 Try and generate multiply-add floating point instructions @code{polyF}
6766 and @code{dotF}. This option is only available if the @samp{-m32381}
6767 option is in effect. Using these instructions requires changes to to
6768 register allocation which generally has a negative impact on
6769 performance. This option should only be enabled when compiling code
6770 particularly likely to make heavy use of multiply-add instructions.
6773 Do not try and generate multiply-add floating point instructions
6774 @code{polyF} and @code{dotF}. This is the default on all platforms.
6777 Generate output containing library calls for floating point.
6778 @strong{Warning:} the requisite libraries may not be available.
6781 Do not use the bit-field instructions. On some machines it is faster to
6782 use shifting and masking operations. This is the default for the pc532.
6785 Do use the bit-field instructions. This is the default for all platforms
6789 Use a different function-calling convention, in which functions
6790 that take a fixed number of arguments return pop their
6791 arguments on return with the @code{ret} instruction.
6793 This calling convention is incompatible with the one normally
6794 used on Unix, so you cannot use it if you need to call libraries
6795 compiled with the Unix compiler.
6797 Also, you must provide function prototypes for all functions that
6798 take variable numbers of arguments (including @code{printf});
6799 otherwise incorrect code will be generated for calls to those
6802 In addition, seriously incorrect code will result if you call a
6803 function with too many arguments. (Normally, extra arguments are
6804 harmlessly ignored.)
6806 This option takes its name from the 680x0 @code{rtd} instruction.
6810 Use a different function-calling convention where the first two arguments
6811 are passed in registers.
6813 This calling convention is incompatible with the one normally
6814 used on Unix, so you cannot use it if you need to call libraries
6815 compiled with the Unix compiler.
6818 Do not pass any arguments in registers. This is the default for all
6822 It is OK to use the sb as an index register which is always loaded with
6823 zero. This is the default for the pc532-netbsd target.
6826 The sb register is not available for use or has not been initialized to
6827 zero by the run time system. This is the default for all targets except
6828 the pc532-netbsd. It is also implied whenever @samp{-mhimem} or
6829 @samp{-fpic} is set.
6832 Many ns32000 series addressing modes use displacements of up to 512MB.
6833 If an address is above 512MB then displacements from zero can not be used.
6834 This option causes code to be generated which can be loaded above 512MB.
6835 This may be useful for operating systems or ROM code.
6838 Assume code will be loaded in the first 512MB of virtual address space.
6839 This is the default for all platforms.
6845 @subsection MCore Options
6846 @cindex MCore options
6848 These are the @samp{-m} options defined for the Motorola M*Core
6856 Inline constants into the code stream if it can be done in two
6857 instructions or less.
6862 Use the divide instruction. (Enabled by default).
6864 @item -mrelax-immediate
6865 @itemx -mrelax-immediate
6866 @itemx -mno-relax-immediate
6867 Allow arbitary sized immediated in bit operations.
6869 @item -mwide-bitfields
6870 @itemx -mwide-bitfields
6871 @itemx -mno-wide-bitfields
6872 Always treat bitfields as int-sized.
6874 @item -m4byte-functions
6875 @itemx -m4byte-functions
6876 @itemx -mno-4byte-functions
6877 Force all functions to be aligfned to a four byte boundary.
6879 @item -mcallgraph-data
6880 @itemx -mcallgraph-data
6881 @itemx -mno-callgraph-data
6882 Emit callgraph information.
6886 @itemx -mno-slow-bytes
6887 Prefer word access when reading byte quantities.
6889 @item -mlittle-endian
6890 @itemx -mlittle-endian
6892 Genreate code for a little endian target.
6897 Generate code for the 210 processor.
6902 @node Code Gen Options
6903 @section Options for Code Generation Conventions
6904 @cindex code generation conventions
6905 @cindex options, code generation
6906 @cindex run-time options
6908 These machine-independent options control the interface conventions
6909 used in code generation.
6911 Most of them have both positive and negative forms; the negative form
6912 of @samp{-ffoo} would be @samp{-fno-foo}. In the table below, only
6913 one of the forms is listed---the one which is not the default. You
6914 can figure out the other form by either removing @samp{no-} or adding
6919 Enable exception handling. Generates extra code needed to propagate
6920 exceptions. For some targets, this implies GNU CC will generate frame
6921 unwind information for all functions, which can produce significant data
6922 size overhead, although it does not affect execution. If you do not
6923 specify this option, GNU CC will enable it by default for languages like
6924 C++ which normally require exception handling, and disable itfor
6925 languages like C that do not normally require it. However, you may need
6926 to enable this option when compiling C code that needs to interoperate
6927 properly with exception handlers written in C++. You may also wish to
6928 disable this option if you are compiling older C++ programs that don't
6929 use exception handling.
6931 @item -funwind-tables
6932 Similar to @code{-fexceptions}, except that it will just generate any needed
6933 static data, but will not affect the generated code in any other way.
6934 You will normally not enable this option; instead, a language processor
6935 that needs this handling would enable it on your behalf.
6937 @item -fpcc-struct-return
6938 Return ``short'' @code{struct} and @code{union} values in memory like
6939 longer ones, rather than in registers. This convention is less
6940 efficient, but it has the advantage of allowing intercallability between
6941 GCC-compiled files and files compiled with other compilers.
6943 The precise convention for returning structures in memory depends
6944 on the target configuration macros.
6946 Short structures and unions are those whose size and alignment match
6947 that of some integer type.
6949 @item -freg-struct-return
6950 Use the convention that @code{struct} and @code{union} values are
6951 returned in registers when possible. This is more efficient for small
6952 structures than @samp{-fpcc-struct-return}.
6954 If you specify neither @samp{-fpcc-struct-return} nor its contrary
6955 @samp{-freg-struct-return}, GCC defaults to whichever convention is
6956 standard for the target. If there is no standard convention, GCC
6957 defaults to @samp{-fpcc-struct-return}, except on targets where GCC
6958 is the principal compiler. In those cases, we can choose the standard,
6959 and we chose the more efficient register return alternative.
6962 Allocate to an @code{enum} type only as many bytes as it needs for the
6963 declared range of possible values. Specifically, the @code{enum} type
6964 will be equivalent to the smallest integer type which has enough room.
6966 @item -fshort-double
6967 Use the same size for @code{double} as for @code{float}.
6970 Requests that the data and non-@code{const} variables of this
6971 compilation be shared data rather than private data. The distinction
6972 makes sense only on certain operating systems, where shared data is
6973 shared between processes running the same program, while private data
6974 exists in one copy per process.
6977 Allocate even uninitialized global variables in the data section of the
6978 object file, rather than generating them as common blocks. This has the
6979 effect that if the same variable is declared (without @code{extern}) in
6980 two different compilations, you will get an error when you link them.
6981 The only reason this might be useful is if you wish to verify that the
6982 program will work on other systems which always work this way.
6985 Ignore the @samp{#ident} directive.
6987 @item -fno-gnu-linker
6988 Do not output global initializations (such as C++ constructors and
6989 destructors) in the form used by the GNU linker (on systems where the GNU
6990 linker is the standard method of handling them). Use this option when
6991 you want to use a non-GNU linker, which also requires using the
6992 @code{collect2} program to make sure the system linker includes
6993 constructors and destructors. (@code{collect2} is included in the GCC
6994 distribution.) For systems which @emph{must} use @code{collect2}, the
6995 compiler driver @code{gcc} is configured to do this automatically.
6997 @item -finhibit-size-directive
6998 Don't output a @code{.size} assembler directive, or anything else that
6999 would cause trouble if the function is split in the middle, and the
7000 two halves are placed at locations far apart in memory. This option is
7001 used when compiling @file{crtstuff.c}; you should not need to use it
7005 Put extra commentary information in the generated assembly code to
7006 make it more readable. This option is generally only of use to those
7007 who actually need to read the generated assembly code (perhaps while
7008 debugging the compiler itself).
7010 @samp{-fno-verbose-asm}, the default, causes the
7011 extra information to be omitted and is useful when comparing two assembler
7015 Consider all memory references through pointers to be volatile.
7017 @item -fvolatile-global
7018 Consider all memory references to extern and global data items to
7019 be volatile. GCC does not consider static data items to be volatile
7020 because of this switch.
7022 @item -fvolatile-static
7023 Consider all memory references to static data to be volatile.
7026 @cindex global offset table
7028 Generate position-independent code (PIC) suitable for use in a shared
7029 library, if supported for the target machine. Such code accesses all
7030 constant addresses through a global offset table (GOT). The dynamic
7031 loader resolves the GOT entries when the program starts (the dynamic
7032 loader is not part of GCC; it is part of the operating system). If
7033 the GOT size for the linked executable exceeds a machine-specific
7034 maximum size, you get an error message from the linker indicating that
7035 @samp{-fpic} does not work; in that case, recompile with @samp{-fPIC}
7036 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
7037 on the m68k and RS/6000. The 386 has no such limit.)
7039 Position-independent code requires special support, and therefore works
7040 only on certain machines. For the 386, GCC supports PIC for System V
7041 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
7042 position-independent.
7045 If supported for the target machine, emit position-independent code,
7046 suitable for dynamic linking and avoiding any limit on the size of the
7047 global offset table. This option makes a difference on the m68k, m88k,
7050 Position-independent code requires special support, and therefore works
7051 only on certain machines.
7053 @item -ffixed-@var{reg}
7054 Treat the register named @var{reg} as a fixed register; generated code
7055 should never refer to it (except perhaps as a stack pointer, frame
7056 pointer or in some other fixed role).
7058 @var{reg} must be the name of a register. The register names accepted
7059 are machine-specific and are defined in the @code{REGISTER_NAMES}
7060 macro in the machine description macro file.
7062 This flag does not have a negative form, because it specifies a
7065 @item -fcall-used-@var{reg}
7066 Treat the register named @var{reg} as an allocable register that is
7067 clobbered by function calls. It may be allocated for temporaries or
7068 variables that do not live across a call. Functions compiled this way
7069 will not save and restore the register @var{reg}.
7071 It is an error to used this flag with the frame pointer or stack pointer.
7072 Use of this flag for other registers that have fixed pervasive roles in
7073 the machine's execution model will produce disastrous results.
7075 This flag does not have a negative form, because it specifies a
7078 @item -fcall-saved-@var{reg}
7079 Treat the register named @var{reg} as an allocable register saved by
7080 functions. It may be allocated even for temporaries or variables that
7081 live across a call. Functions compiled this way will save and restore
7082 the register @var{reg} if they use it.
7084 It is an error to used this flag with the frame pointer or stack pointer.
7085 Use of this flag for other registers that have fixed pervasive roles in
7086 the machine's execution model will produce disastrous results.
7088 A different sort of disaster will result from the use of this flag for
7089 a register in which function values may be returned.
7091 This flag does not have a negative form, because it specifies a
7095 Pack all structure members together without holes. Usually you would
7096 not want to use this option, since it makes the code suboptimal, and
7097 the offsets of structure members won't agree with system libraries.
7099 @item -fcheck-memory-usage
7100 Generate extra code to check each memory access. GCC will generate
7101 code that is suitable for a detector of bad memory accesses such as
7104 Normally, you should compile all, or none, of your code with this option.
7106 If you do mix code compiled with and without this option,
7107 you must ensure that all code that has side effects
7108 and that is called by code compiled with this option
7109 is, itself, compiled with this option.
7110 If you do not, you might get erroneous messages from the detector.
7112 If you use functions from a library that have side-effects (such as
7113 @code{read}), you might not be able to recompile the library and
7114 specify this option. In that case, you can enable the
7115 @samp{-fprefix-function-name} option, which requests GCC to encapsulate
7116 your code and make other functions look as if they were compiled with
7117 @samp{-fcheck-memory-usage}. This is done by calling ``stubs'',
7118 which are provided by the detector. If you cannot find or build
7119 stubs for every function you call, you might have to specify
7120 @samp{-fcheck-memory-usage} without @samp{-fprefix-function-name}.
7122 If you specify this option, you can not use the @code{asm} or
7123 @code{__asm__} keywords in functions with memory checking enabled. GNU
7124 CC cannot understand what the @code{asm} statement may do, and therefore
7125 cannot generate the appropriate code, so it will reject it. However, if
7126 you specify the function attribute @code{no_check_memory_usage} (see
7127 @pxref{Function Attributes}, GNU CC will disable memory checking within a
7128 function; you may use @code{asm} statements inside such functions. You
7129 may have an inline expansion of a non-checked function within a checked
7130 function; in that case GNU CC will not generate checks for the inlined
7131 function's memory accesses.
7133 If you move your @code{asm} statements to non-checked inline functions
7134 and they do access memory, you can add calls to the support code in your
7135 inline function, to indicate any reads, writes, or copies being done.
7136 These calls would be similar to those done in the stubs described above.
7138 @item -fprefix-function-name
7139 Request GCC to add a prefix to the symbols generated for function names.
7140 GCC adds a prefix to the names of functions defined as well as
7141 functions called. Code compiled with this option and code compiled
7142 without the option can't be linked together, unless stubs are used.
7144 If you compile the following code with @samp{-fprefix-function-name}
7146 extern void bar (int);
7155 GCC will compile the code as if it was written:
7157 extern void prefix_bar (int);
7161 return prefix_bar (a + 5);
7164 This option is designed to be used with @samp{-fcheck-memory-usage}.
7166 @item -finstrument-functions
7167 Generate instrumentation calls for entry and exit to functions. Just
7168 after function entry and just before function exit, the following
7169 profiling functions will be called with the address of the current
7170 function and its call site. (On some platforms,
7171 @code{__builtin_return_address} does not work beyond the current
7172 function, so the call site information may not be available to the
7173 profiling functions otherwise.)
7176 void __cyg_profile_func_enter (void *this_fn, void *call_site);
7177 void __cyg_profile_func_exit (void *this_fn, void *call_site);
7180 The first argument is the address of the start of the current function,
7181 which may be looked up exactly in the symbol table.
7183 This instrumentation is also done for functions expanded inline in other
7184 functions. The profiling calls will indicate where, conceptually, the
7185 inline function is entered and exited. This means that addressable
7186 versions of such functions must be available. If all your uses of a
7187 function are expanded inline, this may mean an additional expansion of
7188 code size. If you use @samp{extern inline} in your C code, an
7189 addressable version of such functions must be provided. (This is
7190 normally the case anyways, but if you get lucky and the optimizer always
7191 expands the functions inline, you might have gotten away without
7192 providing static copies.)
7194 A function may be given the attribute @code{no_instrument_function}, in
7195 which case this instrumentation will not be done. This can be used, for
7196 example, for the profiling functions listed above, high-priority
7197 interrupt routines, and any functions from which the profiling functions
7198 cannot safely be called (perhaps signal handlers, if the profiling
7199 routines generate output or allocate memory).
7202 Generate code to verify that you do not go beyond the boundary of the
7203 stack. You should specify this flag if you are running in an
7204 environment with multiple threads, but only rarely need to specify it in
7205 a single-threaded environment since stack overflow is automatically
7206 detected on nearly all systems if there is only one stack.
7208 Note that this switch does not actually cause checking to be done; the
7209 operating system must do that. The switch causes generation of code
7210 to ensure that the operating system sees the stack being extended.
7212 @item -fstack-limit-register=@var{reg}
7213 @itemx -fstack-limit-symbol=@var{sym}
7214 @itemx -fno-stack-limit
7215 Generate code to ensure that the stack does not grow beyond a certain value,
7216 either the value of a register or the address of a symbol. If the stack
7217 would grow beyond the value, a signal is raised. For most targets,
7218 the signal is raised before the stack overruns the boundary, so
7219 it is possible to catch the signal without taking special precautions.
7221 For instance, if the stack starts at address @samp{0x80000000} and grows
7222 downwards you can use the flags
7223 @samp{-fstack-limit-symbol=__stack_limit}
7224 @samp{-Wl,--defsym,__stack_limit=0x7ffe0000} which will enforce a stack
7227 @cindex aliasing of parameters
7228 @cindex parameters, aliased
7229 @item -fargument-alias
7230 @itemx -fargument-noalias
7231 @itemx -fargument-noalias-global
7232 Specify the possible relationships among parameters and between
7233 parameters and global data.
7235 @samp{-fargument-alias} specifies that arguments (parameters) may
7236 alias each other and may alias global storage.
7237 @samp{-fargument-noalias} specifies that arguments do not alias
7238 each other, but may alias global storage.
7239 @samp{-fargument-noalias-global} specifies that arguments do not
7240 alias each other and do not alias global storage.
7242 Each language will automatically use whatever option is required by
7243 the language standard. You should not need to use these options yourself.
7245 @item -fleading-underscore
7246 This option and its counterpart, -fno-leading-underscore, forcibly
7247 change the way C symbols are represented in the object file. One use
7248 is to help link with legacy assembly code.
7250 Be warned that you should know what you are doing when invoking this
7251 option, and that not all targets provide complete support for it.
7254 @node Environment Variables
7255 @section Environment Variables Affecting GCC
7256 @cindex environment variables
7258 This section describes several environment variables that affect how GCC
7259 operates. Some of them work by specifying directories or prefixes to use
7260 when searching for various kinds of files. Some are used to specify other
7261 aspects of the compilation environment.
7264 Note that you can also specify places to search using options such as
7265 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
7266 take precedence over places specified using environment variables, which
7267 in turn take precedence over those specified by the configuration of GCC.
7271 Note that you can also specify places to search using options such as
7272 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
7273 take precedence over places specified using environment variables, which
7274 in turn take precedence over those specified by the configuration of GCC.
7281 @c @itemx LC_COLLATE
7283 @c @itemx LC_MONETARY
7284 @c @itemx LC_NUMERIC
7289 @c @findex LC_COLLATE
7291 @c @findex LC_MONETARY
7292 @c @findex LC_NUMERIC
7296 These environment variables control the way that GCC uses
7297 localization information that allow GCC to work with different
7298 national conventions. GCC inspects the locale categories
7299 @code{LC_CTYPE} and @code{LC_MESSAGES} if it has been configured to do
7300 so. These locale categories can be set to any value supported by your
7301 installation. A typical value is @samp{en_UK} for English in the United
7304 The @code{LC_CTYPE} environment variable specifies character
7305 classification. GCC uses it to determine the character boundaries in
7306 a string; this is needed for some multibyte encodings that contain quote
7307 and escape characters that would otherwise be interpreted as a string
7310 The @code{LC_MESSAGES} environment variable specifies the language to
7311 use in diagnostic messages.
7313 If the @code{LC_ALL} environment variable is set, it overrides the value
7314 of @code{LC_CTYPE} and @code{LC_MESSAGES}; otherwise, @code{LC_CTYPE}
7315 and @code{LC_MESSAGES} default to the value of the @code{LANG}
7316 environment variable. If none of these variables are set, GCC
7317 defaults to traditional C English behavior.
7321 If @code{TMPDIR} is set, it specifies the directory to use for temporary
7322 files. GCC uses temporary files to hold the output of one stage of
7323 compilation which is to be used as input to the next stage: for example,
7324 the output of the preprocessor, which is the input to the compiler
7327 @item GCC_EXEC_PREFIX
7328 @findex GCC_EXEC_PREFIX
7329 If @code{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
7330 names of the subprograms executed by the compiler. No slash is added
7331 when this prefix is combined with the name of a subprogram, but you can
7332 specify a prefix that ends with a slash if you wish.
7334 If @code{GCC_EXEC_PREFIX} is not set, GNU CC will attempt to figure out
7335 an appropriate prefix to use based on the pathname it was invoked with.
7337 If GCC cannot find the subprogram using the specified prefix, it
7338 tries looking in the usual places for the subprogram.
7340 The default value of @code{GCC_EXEC_PREFIX} is
7341 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
7342 of @code{prefix} when you ran the @file{configure} script.
7344 Other prefixes specified with @samp{-B} take precedence over this prefix.
7346 This prefix is also used for finding files such as @file{crt0.o} that are
7349 In addition, the prefix is used in an unusual way in finding the
7350 directories to search for header files. For each of the standard
7351 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
7352 (more precisely, with the value of @code{GCC_INCLUDE_DIR}), GCC tries
7353 replacing that beginning with the specified prefix to produce an
7354 alternate directory name. Thus, with @samp{-Bfoo/}, GCC will search
7355 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
7356 These alternate directories are searched first; the standard directories
7360 @findex COMPILER_PATH
7361 The value of @code{COMPILER_PATH} is a colon-separated list of
7362 directories, much like @code{PATH}. GCC tries the directories thus
7363 specified when searching for subprograms, if it can't find the
7364 subprograms using @code{GCC_EXEC_PREFIX}.
7367 @findex LIBRARY_PATH
7368 The value of @code{LIBRARY_PATH} is a colon-separated list of
7369 directories, much like @code{PATH}. When configured as a native compiler,
7370 GCC tries the directories thus specified when searching for special
7371 linker files, if it can't find them using @code{GCC_EXEC_PREFIX}. Linking
7372 using GCC also uses these directories when searching for ordinary
7373 libraries for the @samp{-l} option (but directories specified with
7374 @samp{-L} come first).
7376 @item C_INCLUDE_PATH
7377 @itemx CPLUS_INCLUDE_PATH
7378 @itemx OBJC_INCLUDE_PATH
7379 @findex C_INCLUDE_PATH
7380 @findex CPLUS_INCLUDE_PATH
7381 @findex OBJC_INCLUDE_PATH
7382 @c @itemx OBJCPLUS_INCLUDE_PATH
7383 These environment variables pertain to particular languages. Each
7384 variable's value is a colon-separated list of directories, much like
7385 @code{PATH}. When GCC searches for header files, it tries the
7386 directories listed in the variable for the language you are using, after
7387 the directories specified with @samp{-I} but before the standard header
7390 @item DEPENDENCIES_OUTPUT
7391 @findex DEPENDENCIES_OUTPUT
7392 @cindex dependencies for make as output
7393 If this variable is set, its value specifies how to output dependencies
7394 for Make based on the header files processed by the compiler. This
7395 output looks much like the output from the @samp{-M} option
7396 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
7397 in addition to the usual results of compilation.
7399 The value of @code{DEPENDENCIES_OUTPUT} can be just a file name, in
7400 which case the Make rules are written to that file, guessing the target
7401 name from the source file name. Or the value can have the form
7402 @samp{@var{file} @var{target}}, in which case the rules are written to
7403 file @var{file} using @var{target} as the target name.
7407 @cindex locale definition
7408 This variable is used to pass locale information to the compiler. One way in
7409 which this information is used is to determine the character set to be used
7410 when character literals, string literals and comments are parsed in C and C++.
7411 When the compiler is configured to allow multibyte characters,
7412 the following values for @code{LANG} are recognized:
7416 Recognize JIS characters.
7418 Recognize SJIS characters.
7420 Recognize EUCJP characters.
7423 If @code{LANG} is not defined, or if it has some other value, then the
7424 compiler will use mblen and mbtowc as defined by the default locale to
7425 recognize and translate multibyte characters.
7428 @node Running Protoize
7429 @section Running Protoize
7431 The program @code{protoize} is an optional part of GNU C. You can use
7432 it to add prototypes to a program, thus converting the program to ANSI
7433 C in one respect. The companion program @code{unprotoize} does the
7434 reverse: it removes argument types from any prototypes that are found.
7436 When you run these programs, you must specify a set of source files as
7437 command line arguments. The conversion programs start out by compiling
7438 these files to see what functions they define. The information gathered
7439 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
7441 After scanning comes actual conversion. The specified files are all
7442 eligible to be converted; any files they include (whether sources or
7443 just headers) are eligible as well.
7445 But not all the eligible files are converted. By default,
7446 @code{protoize} and @code{unprotoize} convert only source and header
7447 files in the current directory. You can specify additional directories
7448 whose files should be converted with the @samp{-d @var{directory}}
7449 option. You can also specify particular files to exclude with the
7450 @samp{-x @var{file}} option. A file is converted if it is eligible, its
7451 directory name matches one of the specified directory names, and its
7452 name within the directory has not been excluded.
7454 Basic conversion with @code{protoize} consists of rewriting most
7455 function definitions and function declarations to specify the types of
7456 the arguments. The only ones not rewritten are those for varargs
7459 @code{protoize} optionally inserts prototype declarations at the
7460 beginning of the source file, to make them available for any calls that
7461 precede the function's definition. Or it can insert prototype
7462 declarations with block scope in the blocks where undeclared functions
7465 Basic conversion with @code{unprotoize} consists of rewriting most
7466 function declarations to remove any argument types, and rewriting
7467 function definitions to the old-style pre-ANSI form.
7469 Both conversion programs print a warning for any function declaration or
7470 definition that they can't convert. You can suppress these warnings
7473 The output from @code{protoize} or @code{unprotoize} replaces the
7474 original source file. The original file is renamed to a name ending
7475 with @samp{.save}. If the @samp{.save} file already exists, then
7476 the source file is simply discarded.
7478 @code{protoize} and @code{unprotoize} both depend on GCC itself to
7479 scan the program and collect information about the functions it uses.
7480 So neither of these programs will work until GCC is installed.
7482 Here is a table of the options you can use with @code{protoize} and
7483 @code{unprotoize}. Each option works with both programs unless
7487 @item -B @var{directory}
7488 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
7489 usual directory (normally @file{/usr/local/lib}). This file contains
7490 prototype information about standard system functions. This option
7491 applies only to @code{protoize}.
7493 @item -c @var{compilation-options}
7494 Use @var{compilation-options} as the options when running @code{gcc} to
7495 produce the @samp{.X} files. The special option @samp{-aux-info} is
7496 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
7498 Note that the compilation options must be given as a single argument to
7499 @code{protoize} or @code{unprotoize}. If you want to specify several
7500 @code{gcc} options, you must quote the entire set of compilation options
7501 to make them a single word in the shell.
7503 There are certain @code{gcc} arguments that you cannot use, because they
7504 would produce the wrong kind of output. These include @samp{-g},
7505 @samp{-O}, @samp{-c}, @samp{-S}, and @samp{-o} If you include these in
7506 the @var{compilation-options}, they are ignored.
7509 Rename files to end in @samp{.C}, or @samp{.cc} for DOS-based file
7510 systems, instead of @samp{.c}. This is convenient if you are converting
7511 a C program to C++. This option applies only to @code{protoize}.
7514 Add explicit global declarations. This means inserting explicit
7515 declarations at the beginning of each source file for each function
7516 that is called in the file and was not declared. These declarations
7517 precede the first function definition that contains a call to an
7518 undeclared function. This option applies only to @code{protoize}.
7520 @item -i @var{string}
7521 Indent old-style parameter declarations with the string @var{string}.
7522 This option applies only to @code{protoize}.
7524 @code{unprotoize} converts prototyped function definitions to old-style
7525 function definitions, where the arguments are declared between the
7526 argument list and the initial @samp{@{}. By default, @code{unprotoize}
7527 uses five spaces as the indentation. If you want to indent with just
7528 one space instead, use @samp{-i " "}.
7531 Keep the @samp{.X} files. Normally, they are deleted after conversion
7535 Add explicit local declarations. @code{protoize} with @samp{-l} inserts
7536 a prototype declaration for each function in each block which calls the
7537 function without any declaration. This option applies only to
7541 Make no real changes. This mode just prints information about the conversions
7542 that would have been done without @samp{-n}.
7545 Make no @samp{.save} files. The original files are simply deleted.
7546 Use this option with caution.
7548 @item -p @var{program}
7549 Use the program @var{program} as the compiler. Normally, the name
7553 Work quietly. Most warnings are suppressed.
7556 Print the version number, just like @samp{-v} for @code{gcc}.
7559 If you need special compiler options to compile one of your program's
7560 source files, then you should generate that file's @samp{.X} file
7561 specially, by running @code{gcc} on that source file with the
7562 appropriate options and the option @samp{-aux-info}. Then run
7563 @code{protoize} on the entire set of files. @code{protoize} will use
7564 the existing @samp{.X} file because it is newer than the source file.
7568 gcc -Dfoo=bar file1.c -aux-info
7573 You need to include the special files along with the rest in the
7574 @code{protoize} command, even though their @samp{.X} files already
7575 exist, because otherwise they won't get converted.
7577 @xref{Protoize Caveats}, for more information on how to use
7578 @code{protoize} successfully.