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 @item Code Generation Options
433 @xref{Code Gen Options,,Options for Code Generation Conventions}.
435 -fcall-saved-@var{reg} -fcall-used-@var{reg}
436 -fexceptions -funwind-tables -ffixed-@var{reg} -finhibit-size-directive
437 -fcheck-memory-usage -fprefix-function-name
438 -fno-common -fno-ident -fno-gnu-linker
439 -fpcc-struct-return -fpic -fPIC
440 -freg-struct-return -fshared-data -fshort-enums
441 -fshort-double -fvolatile -fvolatile-global -fvolatile-static
442 -fverbose-asm -fpack-struct -fstack-check
443 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym}
444 -fargument-alias -fargument-noalias
445 -fargument-noalias-global
451 * Overall Options:: Controlling the kind of output:
452 an executable, object files, assembler files,
453 or preprocessed source.
454 * C Dialect Options:: Controlling the variant of C language compiled.
455 * C++ Dialect Options:: Variations on C++.
456 * Warning Options:: How picky should the compiler be?
457 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
458 * Optimize Options:: How much optimization?
459 * Preprocessor Options:: Controlling header files and macro definitions.
460 Also, getting dependency information for Make.
461 * Assembler Options:: Passing options to the assembler.
462 * Link Options:: Specifying libraries and so on.
463 * Directory Options:: Where to find header files and libraries.
464 Where to find the compiler executable files.
465 * Spec Files:: How to pass switches to sub-processes.
466 * Target Options:: Running a cross-compiler, or an old version of GCC.
469 @node Overall Options
470 @section Options Controlling the Kind of Output
472 Compilation can involve up to four stages: preprocessing, compilation
473 proper, assembly and linking, always in that order. The first three
474 stages apply to an individual source file, and end by producing an
475 object file; linking combines all the object files (those newly
476 compiled, and those specified as input) into an executable file.
478 @cindex file name suffix
479 For any given input file, the file name suffix determines what kind of
484 C source code which must be preprocessed.
487 C source code which should not be preprocessed.
490 C++ source code which should not be preprocessed.
493 Objective-C source code. Note that you must link with the library
494 @file{libobjc.a} to make an Objective-C program work.
497 C header file (not to be compiled or linked).
500 @itemx @var{file}.cxx
501 @itemx @var{file}.cpp
503 C++ source code which must be preprocessed. Note that in @samp{.cxx},
504 the last two letters must both be literally @samp{x}. Likewise,
505 @samp{.C} refers to a literal capital C.
511 Assembler code which must be preprocessed.
514 An object file to be fed straight into linking.
515 Any file name with no recognized suffix is treated this way.
518 You can specify the input language explicitly with the @samp{-x} option:
521 @item -x @var{language}
522 Specify explicitly the @var{language} for the following input files
523 (rather than letting the compiler choose a default based on the file
524 name suffix). This option applies to all following input files until
525 the next @samp{-x} option. Possible values for @var{language} are:
528 c-header cpp-output c++-cpp-output
529 assembler assembler-with-cpp
533 Turn off any specification of a language, so that subsequent files are
534 handled according to their file name suffixes (as they are if @samp{-x}
535 has not been used at all).
537 @item -pass-exit-codes
538 Normally the @code{gcc} program will exit with the code of 1 if any
539 phase of the compiler returns a non-success return code. If you specify
540 @samp{-pass-exit-codes}, the @code{gcc} program will instead return with
541 numerically highest error produced by any phase that returned an error
545 If you only want some of the stages of compilation, you can use
546 @samp{-x} (or filename suffixes) to tell @code{gcc} where to start, and
547 one of the options @samp{-c}, @samp{-S}, or @samp{-E} to say where
548 @code{gcc} is to stop. Note that some combinations (for example,
549 @samp{-x cpp-output -E} instruct @code{gcc} to do nothing at all.
553 Compile or assemble the source files, but do not link. The linking
554 stage simply is not done. The ultimate output is in the form of an
555 object file for each source file.
557 By default, the object file name for a source file is made by replacing
558 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
560 Unrecognized input files, not requiring compilation or assembly, are
564 Stop after the stage of compilation proper; do not assemble. The output
565 is in the form of an assembler code file for each non-assembler input
568 By default, the assembler file name for a source file is made by
569 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
571 Input files that don't require compilation are ignored.
574 Stop after the preprocessing stage; do not run the compiler proper. The
575 output is in the form of preprocessed source code, which is sent to the
578 Input files which don't require preprocessing are ignored.
580 @cindex output file option
582 Place output in file @var{file}. This applies regardless to whatever
583 sort of output is being produced, whether it be an executable file,
584 an object file, an assembler file or preprocessed C code.
586 Since only one output file can be specified, it does not make sense to
587 use @samp{-o} when compiling more than one input file, unless you are
588 producing an executable file as output.
590 If @samp{-o} is not specified, the default is to put an executable file
591 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
592 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
593 all preprocessed C source on standard output.@refill
596 Print (on standard error output) the commands executed to run the stages
597 of compilation. Also print the version number of the compiler driver
598 program and of the preprocessor and the compiler proper.
601 Use pipes rather than temporary files for communication between the
602 various stages of compilation. This fails to work on some systems where
603 the assembler is unable to read from a pipe; but the GNU assembler has
607 Print (on the standard output) a description of the command line options
608 understood by @code{gcc}. If the @code{-v} option is also specified
609 then @code{--help} will also be passed on to the various processes
610 invoked by @code{gcc}, so that they can display the command line options
611 they accept. If the @code{-W} option is also specified then command
612 line options which have no documentation associated with them will also
617 @section Compiling C++ Programs
619 @cindex suffixes for C++ source
620 @cindex C++ source file suffixes
621 C++ source files conventionally use one of the suffixes @samp{.C},
622 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
623 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
624 files with these names and compiles them as C++ programs even if you
625 call the compiler the same way as for compiling C programs (usually with
626 the name @code{gcc}).
630 However, C++ programs often require class libraries as well as a
631 compiler that understands the C++ language---and under some
632 circumstances, you might want to compile programs from standard input,
633 or otherwise without a suffix that flags them as C++ programs.
634 @code{g++} is a program that calls GCC with the default language
635 set to C++, and automatically specifies linking against the C++
636 library. On many systems, the script @code{g++} is also
637 installed with the name @code{c++}.
639 @cindex invoking @code{g++}
640 When you compile C++ programs, you may specify many of the same
641 command-line options that you use for compiling programs in any
642 language; or command-line options meaningful for C and related
643 languages; or options that are meaningful only for C++ programs.
644 @xref{C Dialect Options,,Options Controlling C Dialect}, for
645 explanations of options for languages related to C.
646 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
647 explanations of options that are meaningful only for C++ programs.
649 @node C Dialect Options
650 @section Options Controlling C Dialect
651 @cindex dialect options
652 @cindex language dialect options
653 @cindex options, dialect
655 The following options control the dialect of C (or languages derived
656 from C, such as C++ and Objective C) that the compiler accepts:
661 In C mode, support all ANSI standard C programs. In C++ mode,
662 remove GNU extensions that conflict with ANSI C++.
663 @c shouldn't we be saying "ISO"?
665 This turns off certain features of GCC that are incompatible with ANSI
666 C (when compiling C code), or of ANSI standard C++ (when compiling C++ code),
667 such as the @code{asm} and @code{typeof} keywords, and
668 predefined macros such as @code{unix} and @code{vax} that identify the
669 type of system you are using. It also enables the undesirable and
670 rarely used ANSI trigraph feature. For the C compiler,
671 it disables recognition of C++ style @samp{//} comments as well as
672 the @code{inline} keyword. For the C++ compiler,
673 @samp{-foperator-names} is enabled as well.
676 The alternate keywords @code{__asm__}, @code{__extension__},
677 @code{__inline__} and @code{__typeof__} continue to work despite
678 @samp{-ansi}. You would not want to use them in an ANSI C program, of
679 course, but it is useful to put them in header files that might be included
680 in compilations done with @samp{-ansi}. Alternate predefined macros
681 such as @code{__unix__} and @code{__vax__} are also available, with or
682 without @samp{-ansi}.
684 The @samp{-ansi} option does not cause non-ANSI programs to be
685 rejected gratuitously. For that, @samp{-pedantic} is required in
686 addition to @samp{-ansi}. @xref{Warning Options}.
688 The macro @code{__STRICT_ANSI__} is predefined when the @samp{-ansi}
689 option is used. Some header files may notice this macro and refrain
690 from declaring certain functions or defining certain macros that the
691 ANSI standard doesn't call for; this is to avoid interfering with any
692 programs that might use these names for other things.
694 The functions @code{alloca}, @code{abort}, @code{exit}, and
695 @code{_exit} are not builtin functions when @samp{-ansi} is used.
698 Determine the language standard. A value for this option must be provided;
706 ISO C as modified in amend. 1
712 same as -std=iso9899:1990
715 same as -std=iso9899:199x
718 default, iso9899:1990 + gnu extensions
721 iso9899:199x + gnu extensions
724 Even when this option is not specified, you can still use some of the
725 features of newer standards in so far as they do not conflict with
726 previous C standards. For example, you may use @code{__restrict__} even
727 when -fstd=c9x is not specified.
730 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
731 keyword, so that code can use these words as identifiers. You can use
732 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
733 instead. @samp{-ansi} implies @samp{-fno-asm}.
735 In C++, this switch only affects the @code{typeof} keyword, since
736 @code{asm} and @code{inline} are standard keywords. You may want to
737 use the @samp{-fno-gnu-keywords} flag instead, as it also disables the
738 other, C++-specific, extension keywords such as @code{headof}.
741 @cindex builtin functions
767 Don't recognize builtin functions that do not begin with @samp{__builtin_}
768 as prefix. Currently, the functions affected include @code{abort},
769 @code{abs}, @code{alloca}, @code{cos}, @code{cosf}, @code{cosl},
770 @code{exit}, @code{_exit}, @code{fabs}, @code{fabsf}, @code{fabsl},
771 @code{ffs}, @code{labs}, @code{memcmp}, @code{memcpy}, @code{memset},
772 @code{sin}, @code{sinf}, @code{sinl}, @code{sqrt}, @code{sqrtf},
773 @code{sqrtl}, @code{strcmp}, @code{strcpy}, and @code{strlen}.
775 GCC normally generates special code to handle certain builtin functions
776 more efficiently; for instance, calls to @code{alloca} may become single
777 instructions that adjust the stack directly, and calls to @code{memcpy}
778 may become inline copy loops. The resulting code is often both smaller
779 and faster, but since the function calls no longer appear as such, you
780 cannot set a breakpoint on those calls, nor can you change the behavior
781 of the functions by linking with a different library.
783 The @samp{-ansi} option prevents @code{alloca}, @code{ffs} and @code{_exit}
784 from being builtin functions, since these functions do not have an ANSI
788 @cindex hosted environment
790 Assert that compilation takes place in a hosted environment. This implies
791 @samp{-fbuiltin}. A hosted environment is one in which the
792 entire standard library is available, and in which @code{main} has a return
793 type of @code{int}. Examples are nearly everything except a kernel.
794 This is equivalent to @samp{-fno-freestanding}.
797 @cindex hosted environment
799 Assert that compilation takes place in a freestanding environment. This
800 implies @samp{-fno-builtin}. A freestanding environment
801 is one in which the standard library may not exist, and program startup may
802 not necessarily be at @code{main}. The most obvious example is an OS kernel.
803 This is equivalent to @samp{-fno-hosted}.
806 Support ANSI C trigraphs. You don't want to know about this
807 brain-damage. The @samp{-ansi} option implies @samp{-trigraphs}.
809 @cindex traditional C language
810 @cindex C language, traditional
812 Attempt to support some aspects of traditional C compilers.
817 All @code{extern} declarations take effect globally even if they
818 are written inside of a function definition. This includes implicit
819 declarations of functions.
822 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
823 and @code{volatile} are not recognized. (You can still use the
824 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
828 Comparisons between pointers and integers are always allowed.
831 Integer types @code{unsigned short} and @code{unsigned char} promote
832 to @code{unsigned int}.
835 Out-of-range floating point literals are not an error.
838 Certain constructs which ANSI regards as a single invalid preprocessing
839 number, such as @samp{0xe-0xd}, are treated as expressions instead.
842 String ``constants'' are not necessarily constant; they are stored in
843 writable space, and identical looking constants are allocated
844 separately. (This is the same as the effect of
845 @samp{-fwritable-strings}.)
847 @cindex @code{longjmp} and automatic variables
849 All automatic variables not declared @code{register} are preserved by
850 @code{longjmp}. Ordinarily, GNU C follows ANSI C: automatic variables
851 not declared @code{volatile} may be clobbered.
856 @cindex escape sequences, traditional
857 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
858 literal characters @samp{x} and @samp{a} respectively. Without
859 @w{@samp{-traditional}}, @samp{\x} is a prefix for the hexadecimal
860 representation of a character, and @samp{\a} produces a bell.
863 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
864 if your program uses names that are normally GNU C builtin functions for
865 other purposes of its own.
867 You cannot use @samp{-traditional} if you include any header files that
868 rely on ANSI C features. Some vendors are starting to ship systems with
869 ANSI C header files and you cannot use @samp{-traditional} on such
870 systems to compile files that include any system headers.
872 The @samp{-traditional} option also enables @samp{-traditional-cpp},
873 which is described next.
875 @item -traditional-cpp
876 Attempt to support some aspects of traditional C preprocessors.
881 Comments convert to nothing at all, rather than to a space. This allows
882 traditional token concatenation.
885 In a preprocessing directive, the @samp{#} symbol must appear as the first
889 Macro arguments are recognized within string constants in a macro
890 definition (and their values are stringified, though without additional
891 quote marks, when they appear in such a context). The preprocessor
892 always considers a string constant to end at a newline.
895 @cindex detecting @w{@samp{-traditional}}
896 The predefined macro @code{__STDC__} is not defined when you use
897 @samp{-traditional}, but @code{__GNUC__} is (since the GNU extensions
898 which @code{__GNUC__} indicates are not affected by
899 @samp{-traditional}). If you need to write header files that work
900 differently depending on whether @samp{-traditional} is in use, by
901 testing both of these predefined macros you can distinguish four
902 situations: GNU C, traditional GNU C, other ANSI C compilers, and other
903 old C compilers. The predefined macro @code{__STDC_VERSION__} is also
904 not defined when you use @samp{-traditional}. @xref{Standard
905 Predefined,,Standard Predefined Macros,cpp.info,The C Preprocessor},
906 for more discussion of these and other predefined macros.
909 @cindex string constants vs newline
910 @cindex newline vs string constants
911 The preprocessor considers a string constant to end at a newline (unless
912 the newline is escaped with @samp{\}). (Without @w{@samp{-traditional}},
913 string constants can contain the newline character as typed.)
916 @item -fcond-mismatch
917 Allow conditional expressions with mismatched types in the second and
918 third arguments. The value of such an expression is void.
920 @item -funsigned-char
921 Let the type @code{char} be unsigned, like @code{unsigned char}.
923 Each kind of machine has a default for what @code{char} should
924 be. It is either like @code{unsigned char} by default or like
925 @code{signed char} by default.
927 Ideally, a portable program should always use @code{signed char} or
928 @code{unsigned char} when it depends on the signedness of an object.
929 But many programs have been written to use plain @code{char} and
930 expect it to be signed, or expect it to be unsigned, depending on the
931 machines they were written for. This option, and its inverse, let you
932 make such a program work with the opposite default.
934 The type @code{char} is always a distinct type from each of
935 @code{signed char} or @code{unsigned char}, even though its behavior
936 is always just like one of those two.
939 Let the type @code{char} be signed, like @code{signed char}.
941 Note that this is equivalent to @samp{-fno-unsigned-char}, which is
942 the negative form of @samp{-funsigned-char}. Likewise, the option
943 @samp{-fno-signed-char} is equivalent to @samp{-funsigned-char}.
945 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
946 if your program uses names that are normally GNU C builtin functions for
947 other purposes of its own.
949 You cannot use @samp{-traditional} if you include any header files that
950 rely on ANSI C features. Some vendors are starting to ship systems with
951 ANSI C header files and you cannot use @samp{-traditional} on such
952 systems to compile files that include any system headers.
954 @item -fsigned-bitfields
955 @itemx -funsigned-bitfields
956 @itemx -fno-signed-bitfields
957 @itemx -fno-unsigned-bitfields
958 These options control whether a bitfield is signed or unsigned, when the
959 declaration does not use either @code{signed} or @code{unsigned}. By
960 default, such a bitfield is signed, because this is consistent: the
961 basic integer types such as @code{int} are signed types.
963 However, when @samp{-traditional} is used, bitfields are all unsigned
966 @item -fwritable-strings
967 Store string constants in the writable data segment and don't uniquize
968 them. This is for compatibility with old programs which assume they can
969 write into string constants. The option @samp{-traditional} also has
972 Writing into string constants is a very bad idea; ``constants'' should
975 @item -fallow-single-precision
976 Do not promote single precision math operations to double precision,
977 even when compiling with @samp{-traditional}.
979 Traditional K&R C promotes all floating point operations to double
980 precision, regardless of the sizes of the operands. On the
981 architecture for which you are compiling, single precision may be faster
982 than double precision. If you must use @samp{-traditional}, but want
983 to use single precision operations when the operands are single
984 precision, use this option. This option has no effect when compiling
985 with ANSI or GNU C conventions (the default).
988 Override the underlying type for @samp{wchar_t} to be @samp{short
989 unsigned int} instead of the default for the target. This option is
990 useful for building programs to run under WINE.
993 @node C++ Dialect Options
994 @section Options Controlling C++ Dialect
996 @cindex compiler options, C++
997 @cindex C++ options, command line
999 This section describes the command-line options that are only meaningful
1000 for C++ programs; but you can also use most of the GNU compiler options
1001 regardless of what language your program is in. For example, you
1002 might compile a file @code{firstClass.C} like this:
1005 g++ -g -frepo -O -c firstClass.C
1009 In this example, only @samp{-frepo} is an option meant
1010 only for C++ programs; you can use the other options with any
1011 language supported by GCC.
1013 Here is a list of options that are @emph{only} for compiling C++ programs:
1016 @item -fno-access-control
1017 Turn off all access checking. This switch is mainly useful for working
1018 around bugs in the access control code.
1021 Check that the pointer returned by @code{operator new} is non-null
1022 before attempting to modify the storage allocated. The current Working
1023 Paper requires that @code{operator new} never return a null pointer, so
1024 this check is normally unnecessary.
1026 An alternative to using this option is to specify that your
1027 @code{operator new} does not throw any exceptions; if you declare it
1028 @samp{throw()}, g++ will check the return value. See also @samp{new
1031 @item -fconserve-space
1032 Put uninitialized or runtime-initialized global variables into the
1033 common segment, as C does. This saves space in the executable at the
1034 cost of not diagnosing duplicate definitions. If you compile with this
1035 flag and your program mysteriously crashes after @code{main()} has
1036 completed, you may have an object that is being destroyed twice because
1037 two definitions were merged.
1039 This option is no longer useful on most targets, now that support has
1040 been added for putting variables into BSS without making them common.
1042 @item -fdollars-in-identifiers
1043 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1044 @samp{$} with the option @samp{-fno-dollars-in-identifiers}. (GNU C allows
1045 @samp{$} by default on most target systems, but there are a few exceptions.)
1046 Traditional C allowed the character @samp{$} to form part of
1047 identifiers. However, ANSI C and C++ forbid @samp{$} in identifiers.
1049 @item -fno-elide-constructors
1050 The C++ standard allows an implementation to omit creating a temporary
1051 which is only used to initialize another object of the same type.
1052 Specifying this option disables that optimization, and forces g++ to
1053 call the copy constructor in all cases.
1055 @item -fexternal-templates
1056 Cause template instantiations to obey @samp{#pragma interface} and
1057 @samp{implementation}; template instances are emitted or not according
1058 to the location of the template definition. @xref{Template
1059 Instantiation}, for more information.
1061 This option is deprecated.
1063 @item -falt-external-templates
1064 Similar to -fexternal-templates, but template instances are emitted or
1065 not according to the place where they are first instantiated.
1066 @xref{Template Instantiation}, for more information.
1068 This option is deprecated.
1071 @itemx -fno-for-scope
1072 If -ffor-scope is specified, the scope of variables declared in
1073 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1074 as specified by the draft C++ standard.
1075 If -fno-for-scope is specified, the scope of variables declared in
1076 a @i{for-init-statement} extends to the end of the enclosing scope,
1077 as was the case in old versions of gcc, and other (traditional)
1078 implementations of C++.
1080 The default if neither flag is given to follow the standard,
1081 but to allow and give a warning for old-style code that would
1082 otherwise be invalid, or have different behavior.
1084 @item -fno-gnu-keywords
1085 Do not recognize @code{classof}, @code{headof}, or @code{typeof} as a
1086 keyword, so that code can use these words as identifiers. You can use
1087 the keywords @code{__classof__}, @code{__headof__}, and
1088 @code{__typeof__} instead. @samp{-ansi} implies
1089 @samp{-fno-gnu-keywords}.
1091 @item -fguiding-decls
1092 Treat a function declaration with the same type as a potential function
1093 template instantiation as though it declares that instantiation, not a
1094 normal function. If a definition is given for the function later in the
1095 translation unit (or another translation unit if the target supports
1096 weak symbols), that definition will be used; otherwise the template will
1097 be instantiated. This behavior reflects the C++ language prior to
1098 September 1996, when guiding declarations were removed.
1100 This option implies @samp{-fname-mangling-version-0}, and will not work
1101 with other name mangling versions. Like all options that change the
1102 ABI, all C++ code, @emph{including libgcc.a} must be built with the same
1103 setting of this option.
1106 Treat the @code{namespace std} as a namespace, instead of ignoring
1107 it. For compatibility with earlier versions of g++, the compiler will,
1108 by default, ignore @code{namespace-declarations},
1109 @code{using-declarations}, @code{using-directives}, and
1110 @code{namespace-names}, if they involve @code{std}.
1112 @item -fhuge-objects
1113 Support virtual function calls for objects that exceed the size
1114 representable by a @samp{short int}. Users should not use this flag by
1115 default; if you need to use it, the compiler will tell you so.
1117 This flag is not useful when compiling with -fvtable-thunks.
1119 Like all options that change the ABI, all C++ code, @emph{including
1120 libgcc} must be built with the same setting of this option.
1122 @item -fno-implicit-templates
1123 Never emit code for non-inline templates which are instantiated
1124 implicitly (i.e. by use); only emit code for explicit instantiations.
1125 @xref{Template Instantiation}, for more information.
1127 @item -fno-implicit-inline-templates
1128 Don't emit code for implicit instantiations of inline templates, either.
1129 The default is to handle inlines differently so that compiles with and
1130 without optimization will need the same set of explicit instantiations.
1132 @item -finit-priority
1133 Support @samp{__attribute__ ((init_priority (n)))} for controlling the
1134 order of initialization of file-scope objects. On ELF targets, this
1135 requires GNU ld 2.10 or later.
1137 @item -fno-implement-inlines
1138 To save space, do not emit out-of-line copies of inline functions
1139 controlled by @samp{#pragma implementation}. This will cause linker
1140 errors if these functions are not inlined everywhere they are called.
1142 @item -fms-extensions
1143 Disable pedwarns about constructs used in MFC, such as implicit int and
1144 getting a pointer to member function via non-standard syntax.
1146 @item -fname-mangling-version-@var{n}
1147 Control the way in which names are mangled. Version 0 is compatible
1148 with versions of g++ before 2.8. Version 1 is the default. Version 1
1149 will allow correct mangling of function templates. For example,
1150 version 0 mangling does not mangle foo<int, double> and foo<int, char>
1151 given this declaration:
1154 template <class T, class U> void foo(T t);
1157 Like all options that change the ABI, all C++ code, @emph{including
1158 libgcc} must be built with the same setting of this option.
1160 @item -foperator-names
1161 Recognize the operator name keywords @code{and}, @code{bitand},
1162 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1163 synonyms for the symbols they refer to. @samp{-ansi} implies
1164 @samp{-foperator-names}.
1166 @item -fno-optional-diags
1167 Disable diagnostics that the standard says a compiler does not need to
1168 issue. Currently, the only such diagnostic issued by g++ is the one for
1169 a name having multiple meanings within a class.
1172 Downgrade messages about nonconformant code from errors to warnings. By
1173 default, g++ effectively sets @samp{-pedantic-errors} without
1174 @samp{-pedantic}; this option reverses that. This behavior and this
1175 option are superseded by @samp{-pedantic}, which works as it does for GNU C.
1178 Enable automatic template instantiation. This option also implies
1179 @samp{-fno-implicit-templates}. @xref{Template Instantiation}, for more
1183 Disable generation of information about every class with virtual
1184 functions for use by the C++ runtime type identification features
1185 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1186 of the language, you can save some space by using this flag. Note that
1187 exception handling uses the same information, but it will generate it as
1190 @item -fstrict-prototype
1191 Within an @samp{extern "C"} linkage specification, treat a function
1192 declaration with no arguments, such as @samp{int foo ();}, as declaring
1193 the function to take no arguments. Normally, such a declaration means
1194 that the function @code{foo} can take any combination of arguments, as
1195 in C. @samp{-pedantic} implies @samp{-fstrict-prototype} unless
1196 overridden with @samp{-fno-strict-prototype}.
1198 Specifying this option will also suppress implicit declarations of
1201 This flag no longer affects declarations with C++ linkage.
1204 @itemx -fno-squangle
1205 @samp{-fsquangle} will enable a compressed form of name mangling for
1206 identifiers. In particular, it helps to shorten very long names by recognizing
1207 types and class names which occur more than once, replacing them with special
1208 short ID codes. This option also requires any C++ libraries being used to
1209 be compiled with this option as well. The compiler has this disabled (the
1210 equivalent of @samp{-fno-squangle}) by default.
1212 Like all options that change the ABI, all C++ code, @emph{including
1213 libgcc.a} must be built with the same setting of this option.
1215 @item -ftemplate-depth-@var{n}
1216 Set the maximum instantiation depth for template classes to @var{n}.
1217 A limit on the template instantiation depth is needed to detect
1218 endless recursions during template class instantiation. ANSI/ISO C++
1219 conforming programs must not rely on a maximum depth greater than 17.
1221 @item -fuse-cxa-atexit
1222 Register destructors for objects with static storage duration with the
1223 @code{__cxa_atexit} function rather than the @code{atexit} function.
1224 This option is required for fully standards-compliant handling of static
1225 destructors, but will only work if your C library supports
1226 @code{__cxa_atexit}.
1228 @item -fvtable-thunks
1229 Use @samp{thunks} to implement the virtual function dispatch table
1230 (@samp{vtable}). The traditional (cfront-style) approach to
1231 implementing vtables was to store a pointer to the function and two
1232 offsets for adjusting the @samp{this} pointer at the call site. Newer
1233 implementations store a single pointer to a @samp{thunk} function which
1234 does any necessary adjustment and then calls the target function.
1236 This option also enables a heuristic for controlling emission of
1237 vtables; if a class has any non-inline virtual functions, the vtable
1238 will be emitted in the translation unit containing the first one of
1241 Like all options that change the ABI, all C++ code, @emph{including
1242 libgcc.a} must be built with the same setting of this option.
1245 Do not search for header files in the standard directories specific to
1246 C++, but do still search the other standard directories. (This option
1247 is used when building the C++ library.)
1250 In addition, these optimization, warning, and code generation options
1251 have meanings only for C++ programs:
1254 @item -fno-default-inline
1255 Do not assume @samp{inline} for functions defined inside a class scope.
1256 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1257 functions will have linkage like inline functions; they just won't be
1260 @item -Wctor-dtor-privacy (C++ only)
1261 Warn when a class seems unusable, because all the constructors or
1262 destructors in a class are private and the class has no friends or
1263 public static member functions.
1265 @item -Wnon-virtual-dtor (C++ only)
1266 Warn when a class declares a non-virtual destructor that should probably
1267 be virtual, because it looks like the class will be used polymorphically.
1269 @item -Wreorder (C++ only)
1270 @cindex reordering, warning
1271 @cindex warning for reordering of member initializers
1272 Warn when the order of member initializers given in the code does not
1273 match the order in which they must be executed. For instance:
1279 A(): j (0), i (1) @{ @}
1283 Here the compiler will warn that the member initializers for @samp{i}
1284 and @samp{j} will be rearranged to match the declaration order of the
1288 The following @samp{-W@dots{}} options are not affected by @samp{-Wall}.
1291 @item -Weffc++ (C++ only)
1292 Warn about violations of various style guidelines from Scott Meyers'
1293 @cite{Effective C++} books. If you use this option, you should be aware
1294 that the standard library headers do not obey all of these guidelines;
1295 you can use @samp{grep -v} to filter out those warnings.
1297 @item -Wno-deprecated (C++ only)
1298 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1300 @item -Wno-non-template-friend (C++ only)
1301 Disable warnings when non-templatized friend functions are declared
1302 within a template. With the advent of explicit template specification
1303 support in g++, if the name of the friend is an unqualified-id (ie,
1304 @samp{friend foo(int)}), the C++ language specification demands that the
1305 friend declare or define an ordinary, nontemplate function. (Section
1306 14.5.3). Before g++ implemented explicit specification, unqualified-ids
1307 could be interpreted as a particular specialization of a templatized
1308 function. Because this non-conforming behavior is no longer the default
1309 behavior for g++, @samp{-Wnon-template-friend} allows the compiler to
1310 check existing code for potential trouble spots, and is on by default.
1311 This new compiler behavior can also be turned off with the flag
1312 @samp{-fguiding-decls}, which activates the older, non-specification
1313 compiler code, or with @samp{-Wno-non-template-friend} which keeps the
1314 conformant compiler code but disables the helpful warning.
1316 @item -Wold-style-cast (C++ only)
1317 Warn if an old-style (C-style) cast is used within a C++ program. The
1318 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1319 @samp{const_cast}) are less vulnerable to unintended effects.
1321 @item -Woverloaded-virtual (C++ only)
1322 @cindex overloaded virtual fn, warning
1323 @cindex warning for overloaded virtual fn
1324 Warn when a derived class function declaration may be an error in
1325 defining a virtual function. In a derived class, the
1326 definitions of virtual functions must match the type signature of a
1327 virtual function declared in the base class. With this option, the
1328 compiler warns when you define a function with the same name as a
1329 virtual function, but with a type signature that does not match any
1330 declarations from the base class.
1332 @item -Wno-pmf-conversions (C++ only)
1333 Disable the diagnostic for converting a bound pointer to member function
1336 @item -Wsign-promo (C++ only)
1337 Warn when overload resolution chooses a promotion from unsigned or
1338 enumeral type to a signed type over a conversion to an unsigned type of
1339 the same size. Previous versions of g++ would try to preserve
1340 unsignedness, but the standard mandates the current behavior.
1342 @item -Wsynth (C++ only)
1343 @cindex warning for synthesized methods
1344 @cindex synthesized methods, warning
1345 Warn when g++'s synthesis behavior does not match that of cfront. For
1351 A& operator = (int);
1361 In this example, g++ will synthesize a default @samp{A& operator =
1362 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1365 @node Warning Options
1366 @section Options to Request or Suppress Warnings
1367 @cindex options to control warnings
1368 @cindex warning messages
1369 @cindex messages, warning
1370 @cindex suppressing warnings
1372 Warnings are diagnostic messages that report constructions which
1373 are not inherently erroneous but which are risky or suggest there
1374 may have been an error.
1376 You can request many specific warnings with options beginning @samp{-W},
1377 for example @samp{-Wimplicit} to request warnings on implicit
1378 declarations. Each of these specific warning options also has a
1379 negative form beginning @samp{-Wno-} to turn off warnings;
1380 for example, @samp{-Wno-implicit}. This manual lists only one of the
1381 two forms, whichever is not the default.
1383 These options control the amount and kinds of warnings produced by GCC:
1386 @cindex syntax checking
1388 Check the code for syntax errors, but don't do anything beyond that.
1391 Issue all the warnings demanded by strict ANSI C and ISO C++;
1392 reject all programs that use forbidden extensions.
1394 Valid ANSI C and ISO C++ programs should compile properly with or without
1395 this option (though a rare few will require @samp{-ansi}). However,
1396 without this option, certain GNU extensions and traditional C and C++
1397 features are supported as well. With this option, they are rejected.
1399 @samp{-pedantic} does not cause warning messages for use of the
1400 alternate keywords whose names begin and end with @samp{__}. Pedantic
1401 warnings are also disabled in the expression that follows
1402 @code{__extension__}. However, only system header files should use
1403 these escape routes; application programs should avoid them.
1404 @xref{Alternate Keywords}.
1406 This option is not intended to be @i{useful}; it exists only to satisfy
1407 pedants who would otherwise claim that GCC fails to support the ANSI
1410 Some users try to use @samp{-pedantic} to check programs for strict ANSI
1411 C conformance. They soon find that it does not do quite what they want:
1412 it finds some non-ANSI practices, but not all---only those for which
1413 ANSI C @emph{requires} a diagnostic.
1415 A feature to report any failure to conform to ANSI C might be useful in
1416 some instances, but would require considerable additional work and would
1417 be quite different from @samp{-pedantic}. We don't have plans to
1418 support such a feature in the near future.
1420 @item -pedantic-errors
1421 Like @samp{-pedantic}, except that errors are produced rather than
1425 Inhibit all warning messages.
1428 Inhibit warning messages about the use of @samp{#import}.
1430 @item -Wchar-subscripts
1431 Warn if an array subscript has type @code{char}. This is a common cause
1432 of error, as programmers often forget that this type is signed on some
1436 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1437 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1440 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1441 the arguments supplied have types appropriate to the format string
1444 @item -Wimplicit-int
1445 Warn when a declaration does not specify a type.
1447 @item -Wimplicit-function-declaration
1448 @itemx -Werror-implicit-function-declaration
1449 Give a warning (or error) whenever a function is used before being
1453 Same as @samp{-Wimplicit-int} and @samp{-Wimplicit-function-}@*
1457 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1458 function with external linkage, returning int, taking either zero
1459 arguments, two, or three arguments of appropriate types.
1462 Warn if a multicharacter constant (@samp{'FOOF'}) is used. Usually they
1463 indicate a typo in the user's code, as they have implementation-defined
1464 values, and should not be used in portable code.
1467 Warn if parentheses are omitted in certain contexts, such
1468 as when there is an assignment in a context where a truth value
1469 is expected, or when operators are nested whose precedence people
1470 often get confused about.
1472 Also warn about constructions where there may be confusion to which
1473 @code{if} statement an @code{else} branch belongs. Here is an example of
1486 In C, every @code{else} branch belongs to the innermost possible @code{if}
1487 statement, which in this example is @code{if (b)}. This is often not
1488 what the programmer expected, as illustrated in the above example by
1489 indentation the programmer chose. When there is the potential for this
1490 confusion, GNU C will issue a warning when this flag is specified.
1491 To eliminate the warning, add explicit braces around the innermost
1492 @code{if} statement so there is no way the @code{else} could belong to
1493 the enclosing @code{if}. The resulting code would look like this:
1508 Warn whenever a function is defined with a return-type that defaults
1509 to @code{int}. Also warn about any @code{return} statement with no
1510 return-value in a function whose return-type is not @code{void}.
1513 Warn whenever a @code{switch} statement has an index of enumeral type
1514 and lacks a @code{case} for one or more of the named codes of that
1515 enumeration. (The presence of a @code{default} label prevents this
1516 warning.) @code{case} labels outside the enumeration range also
1517 provoke warnings when this option is used.
1520 Warn if any trigraphs are encountered (assuming they are enabled).
1523 Warn whenever a variable is unused aside from its declaration,
1524 whenever a function is declared static but never defined, whenever a
1525 label is declared but not used, and whenever a statement computes a
1526 result that is explicitly not used.
1528 In order to get a warning about an unused function parameter, you must
1529 specify both @samp{-W} and @samp{-Wunused}.
1531 To suppress this warning for an expression, simply cast it to void. For
1532 unused variables, parameters and labels, use the @samp{unused} attribute
1533 (@pxref{Variable Attributes}).
1535 @item -Wuninitialized
1536 Warn if an automatic variable is used without first being initialized or
1537 if a variable may be clobbered by a @code{setjmp} call.
1539 These warnings are possible only in optimizing compilation,
1540 because they require data flow information that is computed only
1541 when optimizing. If you don't specify @samp{-O}, you simply won't
1544 These warnings occur only for variables that are candidates for
1545 register allocation. Therefore, they do not occur for a variable that
1546 is declared @code{volatile}, or whose address is taken, or whose size
1547 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
1548 structures, unions or arrays, even when they are in registers.
1550 Note that there may be no warning about a variable that is used only
1551 to compute a value that itself is never used, because such
1552 computations may be deleted by data flow analysis before the warnings
1555 These warnings are made optional because GCC is not smart
1556 enough to see all the reasons why the code might be correct
1557 despite appearing to have an error. Here is one example of how
1576 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
1577 always initialized, but GCC doesn't know this. Here is
1578 another common case:
1583 if (change_y) save_y = y, y = new_y;
1585 if (change_y) y = save_y;
1590 This has no bug because @code{save_y} is used only if it is set.
1592 @cindex @code{longjmp} warnings
1593 This option also warns when a nonvolatile automatic variable might be
1594 changed by a call to @code{longjmp}. These warnings as well are possible
1595 only in optimizing compilation.
1597 The compiler sees only the calls to @code{setjmp}. It cannot know
1598 where @code{longjmp} will be called; in fact, a signal handler could
1599 call it at any point in the code. As a result, you may get a warning
1600 even when there is in fact no problem because @code{longjmp} cannot
1601 in fact be called at the place which would cause a problem.
1603 Some spurious warnings can be avoided if you declare all the functions
1604 you use that never return as @code{noreturn}. @xref{Function
1607 @item -Wreorder (C++ only)
1608 @cindex reordering, warning
1609 @cindex warning for reordering of member initializers
1610 Warn when the order of member initializers given in the code does not
1611 match the order in which they must be executed. For instance:
1613 @item -Wunknown-pragmas
1614 @cindex warning for unknown pragmas
1615 @cindex unknown pragmas, warning
1616 @cindex pragmas, warning of unknown
1617 Warn when a #pragma directive is encountered which is not understood by
1618 GCC. If this command line option is used, warnings will even be issued
1619 for unknown pragmas in system header files. This is not the case if
1620 the warnings were only enabled by the @samp{-Wall} command line option.
1623 All of the above @samp{-W} options combined. This enables all the
1624 warnings about constructions that some users consider questionable, and
1625 that are easy to avoid (or modify to prevent the warning), even in
1626 conjunction with macros.
1629 The following @samp{-W@dots{}} options are not implied by @samp{-Wall}.
1630 Some of them warn about constructions that users generally do not
1631 consider questionable, but which occasionally you might wish to check
1632 for; others warn about constructions that are necessary or hard to avoid
1633 in some cases, and there is no simple way to modify the code to suppress
1638 Print extra warning messages for these events:
1642 A function can return either with or without a value. (Falling
1643 off the end of the function body is considered returning without
1644 a value.) For example, this function would evoke such a
1658 An expression-statement or the left-hand side of a comma expression
1659 contains no side effects.
1660 To suppress the warning, cast the unused expression to void.
1661 For example, an expression such as @samp{x[i,j]} will cause a warning,
1662 but @samp{x[(void)i,j]} will not.
1665 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
1668 A comparison like @samp{x<=y<=z} appears; this is equivalent to
1669 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
1670 that of ordinary mathematical notation.
1673 Storage-class specifiers like @code{static} are not the first things in
1674 a declaration. According to the C Standard, this usage is obsolescent.
1677 If @samp{-Wall} or @samp{-Wunused} is also specified, warn about unused
1681 A comparison between signed and unsigned values could produce an
1682 incorrect result when the signed value is converted to unsigned.
1683 (But don't warn if @samp{-Wno-sign-compare} is also specified.)
1686 An aggregate has a partly bracketed initializer.
1687 For example, the following code would evoke such a warning,
1688 because braces are missing around the initializer for @code{x.h}:
1691 struct s @{ int f, g; @};
1692 struct t @{ struct s h; int i; @};
1693 struct t x = @{ 1, 2, 3 @};
1697 An aggregate has an initializer which does not initialize all members.
1698 For example, the following code would cause such a warning, because
1699 @code{x.h} would be implicitly initialized to zero:
1702 struct s @{ int f, g, h; @};
1703 struct s x = @{ 3, 4 @};
1708 Warn if floating point values are used in equality comparisons.
1710 @item -Wtraditional (C only)
1711 Warn about certain constructs that behave differently in traditional and
1716 Macro arguments occurring within string constants in the macro body.
1717 These would substitute the argument in traditional C, but are part of
1718 the constant in ANSI C.
1721 A function declared external in one block and then used after the end of
1725 A @code{switch} statement has an operand of type @code{long}.
1728 A non-@code{static} function declaration follows a @code{static} one.
1729 This construct is not accepted by some traditional C compilers.
1732 The ANSI type of an integer constant has a different width or
1733 signedness from its traditional type. This warning is only issued if
1734 the base of the constant is ten. I.e. hexadecimal or octal values, which
1735 typically represent bit patterns, are not warned about.
1738 Usage of ANSI string concatenation is detected.
1742 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
1745 Warn whenever a local variable shadows another local variable.
1747 @item -Wid-clash-@var{len}
1748 Warn whenever two distinct identifiers match in the first @var{len}
1749 characters. This may help you prepare a program that will compile
1750 with certain obsolete, brain-damaged compilers.
1752 @item -Wlarger-than-@var{len}
1753 Warn whenever an object of larger than @var{len} bytes is defined.
1755 @item -Wpointer-arith
1756 Warn about anything that depends on the ``size of'' a function type or
1757 of @code{void}. GNU C assigns these types a size of 1, for
1758 convenience in calculations with @code{void *} pointers and pointers
1761 @item -Wbad-function-cast (C only)
1762 Warn whenever a function call is cast to a non-matching type.
1763 For example, warn if @code{int malloc()} is cast to @code{anything *}.
1766 Warn whenever a pointer is cast so as to remove a type qualifier from
1767 the target type. For example, warn if a @code{const char *} is cast
1768 to an ordinary @code{char *}.
1771 Warn whenever a pointer is cast such that the required alignment of the
1772 target is increased. For example, warn if a @code{char *} is cast to
1773 an @code{int *} on machines where integers can only be accessed at
1774 two- or four-byte boundaries.
1776 @item -Wwrite-strings
1777 Give string constants the type @code{const char[@var{length}]} so that
1778 copying the address of one into a non-@code{const} @code{char *}
1779 pointer will get a warning. These warnings will help you find at
1780 compile time code that can try to write into a string constant, but
1781 only if you have been very careful about using @code{const} in
1782 declarations and prototypes. Otherwise, it will just be a nuisance;
1783 this is why we did not make @samp{-Wall} request these warnings.
1786 Warn if a prototype causes a type conversion that is different from what
1787 would happen to the same argument in the absence of a prototype. This
1788 includes conversions of fixed point to floating and vice versa, and
1789 conversions changing the width or signedness of a fixed point argument
1790 except when the same as the default promotion.
1792 Also, warn if a negative integer constant expression is implicitly
1793 converted to an unsigned type. For example, warn about the assignment
1794 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
1795 casts like @code{(unsigned) -1}.
1797 @item -Wsign-compare
1798 @cindex warning for comparison of signed and unsigned values
1799 @cindex comparison of signed and unsigned values, warning
1800 @cindex signed and unsigned values, comparison warning
1801 Warn when a comparison between signed and unsigned values could produce
1802 an incorrect result when the signed value is converted to unsigned.
1803 This warning is also enabled by @samp{-W}; to get the other warnings
1804 of @samp{-W} without this warning, use @samp{-W -Wno-sign-compare}.
1806 @item -Waggregate-return
1807 Warn if any functions that return structures or unions are defined or
1808 called. (In languages where you can return an array, this also elicits
1811 @item -Wstrict-prototypes (C only)
1812 Warn if a function is declared or defined without specifying the
1813 argument types. (An old-style function definition is permitted without
1814 a warning if preceded by a declaration which specifies the argument
1817 @item -Wmissing-prototypes (C only)
1818 Warn if a global function is defined without a previous prototype
1819 declaration. This warning is issued even if the definition itself
1820 provides a prototype. The aim is to detect global functions that fail
1821 to be declared in header files.
1823 @item -Wmissing-declarations
1824 Warn if a global function is defined without a previous declaration.
1825 Do so even if the definition itself provides a prototype.
1826 Use this option to detect global functions that are not declared in
1829 @item -Wmissing-noreturn
1830 Warn about functions which might be candidates for attribute @code{noreturn}.
1831 Note these are only possible candidates, not absolute ones. Care should
1832 be taken to manually verify functions actually do not ever return before
1833 adding the @code{noreturn} attribute, otherwise subtle code generation
1834 bugs could be introduced.
1837 Warn if a structure is given the packed attribute, but the packed
1838 attribute has no effect on the layout or size of the structure.
1839 Such structures may be mis-aligned for little benefit. For
1840 instance, in this code, the variable @code{f.x} in @code{struct bar}
1841 will be misaligned even though @code{struct bar} does not itself
1842 have the packed attribute:
1849 @} __attribute__((packed));
1858 Warn if padding is included in a structure, either to align an element
1859 of the structure or to align the whole structure. Sometimes when this
1860 happens it is possible to rearrange the fields of the structure to
1861 reduce the padding and so make the structure smaller.
1863 @item -Wredundant-decls
1864 Warn if anything is declared more than once in the same scope, even in
1865 cases where multiple declaration is valid and changes nothing.
1867 @item -Wnested-externs (C only)
1868 Warn if an @code{extern} declaration is encountered within a function.
1870 @item -Wunreachable-code
1871 Warn if the compiler detects that code will never be executed.
1873 This option is intended to warn when the compiler detects that at
1874 least a whole line of source code will never be executed, because
1875 some condition is never satisfied or because it is after a
1876 procedure that never returns.
1878 It is possible for this option to produce a warning even though there
1879 are circumstances under which part of the affected line can be executed,
1880 so care should be taken when removing apparently-unreachable code.
1882 For instance, when a function is inlined, a warning may mean that the
1883 line is unreachable in only one inlined copy of the function.
1885 This option is not made part of @samp{-Wall} because in a debugging
1886 version of a program there is often substantial code which checks
1887 correct functioning of the program and is, hopefully, unreachable
1888 because the program does work. Another common use of unreachable
1889 code is to provide behaviour which is selectable at compile-time.
1892 Warn if a function can not be inlined and it was declared as inline.
1895 Warn if @samp{long long} type is used. This is default. To inhibit
1896 the warning messages, use @samp{-Wno-long-long}. Flags
1897 @samp{-Wlong-long} and @samp{-Wno-long-long} are taken into account
1898 only when @samp{-pedantic} flag is used.
1901 Make all warnings into errors.
1904 @node Debugging Options
1905 @section Options for Debugging Your Program or GCC
1906 @cindex options, debugging
1907 @cindex debugging information options
1909 GCC has various special options that are used for debugging
1910 either your program or GCC:
1914 Produce debugging information in the operating system's native format
1915 (stabs, COFF, XCOFF, or DWARF). GDB can work with this debugging
1918 On most systems that use stabs format, @samp{-g} enables use of extra
1919 debugging information that only GDB can use; this extra information
1920 makes debugging work better in GDB but will probably make other debuggers
1922 refuse to read the program. If you want to control for certain whether
1923 to generate the extra information, use @samp{-gstabs+}, @samp{-gstabs},
1924 @samp{-gxcoff+}, @samp{-gxcoff}, @samp{-gdwarf-1+}, or @samp{-gdwarf-1}
1927 Unlike most other C compilers, GCC allows you to use @samp{-g} with
1928 @samp{-O}. The shortcuts taken by optimized code may occasionally
1929 produce surprising results: some variables you declared may not exist
1930 at all; flow of control may briefly move where you did not expect it;
1931 some statements may not be executed because they compute constant
1932 results or their values were already at hand; some statements may
1933 execute in different places because they were moved out of loops.
1935 Nevertheless it proves possible to debug optimized output. This makes
1936 it reasonable to use the optimizer for programs that might have bugs.
1938 The following options are useful when GCC is generated with the
1939 capability for more than one debugging format.
1942 Produce debugging information for use by GDB. This means to use the
1943 most expressive format available (DWARF 2, stabs, or the native format
1944 if neither of those are supported), including GDB extensions if at all
1948 Produce debugging information in stabs format (if that is supported),
1949 without GDB extensions. This is the format used by DBX on most BSD
1950 systems. On MIPS, Alpha and System V Release 4 systems this option
1951 produces stabs debugging output which is not understood by DBX or SDB.
1952 On System V Release 4 systems this option requires the GNU assembler.
1955 Produce debugging information in stabs format (if that is supported),
1956 using GNU extensions understood only by the GNU debugger (GDB). The
1957 use of these extensions is likely to make other debuggers crash or
1958 refuse to read the program.
1961 Produce debugging information in COFF format (if that is supported).
1962 This is the format used by SDB on most System V systems prior to
1966 Produce debugging information in XCOFF format (if that is supported).
1967 This is the format used by the DBX debugger on IBM RS/6000 systems.
1970 Produce debugging information in XCOFF format (if that is supported),
1971 using GNU extensions understood only by the GNU debugger (GDB). The
1972 use of these extensions is likely to make other debuggers crash or
1973 refuse to read the program, and may cause assemblers other than the GNU
1974 assembler (GAS) to fail with an error.
1977 Produce debugging information in DWARF version 1 format (if that is
1978 supported). This is the format used by SDB on most System V Release 4
1982 Produce debugging information in DWARF version 1 format (if that is
1983 supported), using GNU extensions understood only by the GNU debugger
1984 (GDB). The use of these extensions is likely to make other debuggers
1985 crash or refuse to read the program.
1988 Produce debugging information in DWARF version 2 format (if that is
1989 supported). This is the format used by DBX on IRIX 6.
1992 @itemx -ggdb@var{level}
1993 @itemx -gstabs@var{level}
1994 @itemx -gcoff@var{level}
1995 @itemx -gxcoff@var{level}
1996 @itemx -gdwarf@var{level}
1997 @itemx -gdwarf-2@var{level}
1998 Request debugging information and also use @var{level} to specify how
1999 much information. The default level is 2.
2001 Level 1 produces minimal information, enough for making backtraces in
2002 parts of the program that you don't plan to debug. This includes
2003 descriptions of functions and external variables, but no information
2004 about local variables and no line numbers.
2006 Level 3 includes extra information, such as all the macro definitions
2007 present in the program. Some debuggers support macro expansion when
2012 Generate extra code to write profile information suitable for the
2013 analysis program @code{prof}. You must use this option when compiling
2014 the source files you want data about, and you must also use it when
2017 @cindex @code{gprof}
2019 Generate extra code to write profile information suitable for the
2020 analysis program @code{gprof}. You must use this option when compiling
2021 the source files you want data about, and you must also use it when
2026 Generate extra code to write profile information for basic blocks, which will
2027 record the number of times each basic block is executed, the basic block start
2028 address, and the function name containing the basic block. If @samp{-g} is
2029 used, the line number and filename of the start of the basic block will also be
2030 recorded. If not overridden by the machine description, the default action is
2031 to append to the text file @file{bb.out}.
2033 This data could be analyzed by a program like @code{tcov}. Note,
2034 however, that the format of the data is not what @code{tcov} expects.
2035 Eventually GNU @code{gprof} should be extended to process this data.
2038 Makes the compiler print out each function name as it is compiled, and
2039 print some statistics about each pass when it finishes.
2042 Generate extra code to profile basic blocks. Your executable will
2043 produce output that is a superset of that produced when @samp{-a} is
2044 used. Additional output is the source and target address of the basic
2045 blocks where a jump takes place, the number of times a jump is executed,
2046 and (optionally) the complete sequence of basic blocks being executed.
2047 The output is appended to file @file{bb.out}.
2049 You can examine different profiling aspects without recompilation. Your
2050 executable will read a list of function names from file @file{bb.in}.
2051 Profiling starts when a function on the list is entered and stops when
2052 that invocation is exited. To exclude a function from profiling, prefix
2053 its name with `-'. If a function name is not unique, you can
2054 disambiguate it by writing it in the form
2055 @samp{/path/filename.d:functionname}. Your executable will write the
2056 available paths and filenames in file @file{bb.out}.
2058 Several function names have a special meaning:
2061 Write source, target and frequency of jumps to file @file{bb.out}.
2062 @item __bb_hidecall__
2063 Exclude function calls from frequency count.
2064 @item __bb_showret__
2065 Include function returns in frequency count.
2067 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
2068 The file will be compressed using the program @samp{gzip}, which must
2069 exist in your @code{PATH}. On systems without the @samp{popen}
2070 function, the file will be named @file{bbtrace} and will not be
2071 compressed. @strong{Profiling for even a few seconds on these systems
2072 will produce a very large file.} Note: @code{__bb_hidecall__} and
2073 @code{__bb_showret__} will not affect the sequence written to
2077 Here's a short example using different profiling parameters
2078 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
2079 1 and 2 and is called twice from block 3 of function @code{main}. After
2080 the calls, block 3 transfers control to block 4 of @code{main}.
2082 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
2083 the following sequence of blocks is written to file @file{bbtrace.gz}:
2084 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
2085 the return is to a point inside the block and not to the top. The
2086 block address 0 always indicates, that control is transferred
2087 to the trace from somewhere outside the observed functions. With
2088 @samp{-foo} added to @file{bb.in}, the blocks of function
2089 @code{foo} are removed from the trace, so only 0 3 4 remains.
2091 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
2092 jump frequencies will be written to file @file{bb.out}. The
2093 frequencies are obtained by constructing a trace of blocks
2094 and incrementing a counter for every neighbouring pair of blocks
2095 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2099 Jump from block 0x0 to block 0x3 executed 1 time(s)
2100 Jump from block 0x3 to block 0x1 executed 1 time(s)
2101 Jump from block 0x1 to block 0x2 executed 2 time(s)
2102 Jump from block 0x2 to block 0x1 executed 1 time(s)
2103 Jump from block 0x2 to block 0x4 executed 1 time(s)
2106 With @code{__bb_hidecall__}, control transfer due to call instructions
2107 is removed from the trace, that is the trace is cut into three parts: 0
2108 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2109 to return instructions is added to the trace. The trace becomes: 0 3 1
2110 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2111 written to @file{bbtrace.gz}. It is solely used for counting jump
2114 @item -fprofile-arcs
2115 Instrument @dfn{arcs} during compilation. For each function of your
2116 program, GCC creates a program flow graph, then finds a spanning tree
2117 for the graph. Only arcs that are not on the spanning tree have to be
2118 instrumented: the compiler adds code to count the number of times that these
2119 arcs are executed. When an arc is the only exit or only entrance to a
2120 block, the instrumentation code can be added to the block; otherwise, a
2121 new basic block must be created to hold the instrumentation code.
2123 Since not every arc in the program must be instrumented, programs
2124 compiled with this option run faster than programs compiled with
2125 @samp{-a}, which adds instrumentation code to every basic block in the
2126 program. The tradeoff: since @code{gcov} does not have
2127 execution counts for all branches, it must start with the execution
2128 counts for the instrumented branches, and then iterate over the program
2129 flow graph until the entire graph has been solved. Hence, @code{gcov}
2130 runs a little more slowly than a program which uses information from
2133 @samp{-fprofile-arcs} also makes it possible to estimate branch
2134 probabilities, and to calculate basic block execution counts. In
2135 general, basic block execution counts do not give enough information to
2136 estimate all branch probabilities. When the compiled program exits, it
2137 saves the arc execution counts to a file called
2138 @file{@var{sourcename}.da}. Use the compiler option
2139 @samp{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2140 Control Optimization}) when recompiling, to optimize using estimated
2141 branch probabilities.
2144 @item -ftest-coverage
2145 Create data files for the @code{gcov} code-coverage utility
2146 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2147 The data file names begin with the name of your source file:
2150 @item @var{sourcename}.bb
2151 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2152 associate basic block execution counts with line numbers.
2154 @item @var{sourcename}.bbg
2155 A list of all arcs in the program flow graph. This allows @code{gcov}
2156 to reconstruct the program flow graph, so that it can compute all basic
2157 block and arc execution counts from the information in the
2158 @code{@var{sourcename}.da} file (this last file is the output from
2159 @samp{-fprofile-arcs}).
2162 @item -d@var{letters}
2163 Says to make debugging dumps during compilation at times specified by
2164 @var{letters}. This is used for debugging the compiler. The file names
2165 for most of the dumps are made by appending a pass number and a word to
2166 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.jump}).
2167 Here are the possible letters for use in @var{letters}, and their meanings:
2171 Annotate the assembler output with miscellaneous debugging information.
2173 Dump after computing branch probabilities, to @file{@var{file}.07.bp}.
2175 Dump after instruction combination, to the file @file{@var{file}.09.combine}.
2177 Dump after delayed branch scheduling, to @file{@var{file}.19.dbr}.
2179 Dump all macro definitions, at the end of preprocessing, in addition to
2182 Dump after purging ADDRESSOF, to @file{@var{file}.03.addressof}.
2184 Dump after flow analysis, to @file{@var{file}.08.flow}.
2186 Dump after global register allocation, to @file{@var{file}.13.greg}.
2188 Dump after GCSE, to @file{@var{file}.04.gcse}.
2190 Dump after first jump optimization, to @file{@var{file}.01.jump}.
2192 Dump after last jump optimization, to @file{@var{file}.17.jump2}.
2194 Dump after conversion from registers to stack, to @file{@var{file}.20.stack}.
2196 Dump after local register allocation, to @file{@var{file}.12.lreg}.
2198 Dump after loop optimization, to @file{@var{file}.05.loop}.
2200 Dump after performing the machine dependent reorganisation pass, to
2201 @file{@var{file}.18.mach}.
2203 Dump after the register move pass, to @file{@var{file}.10.regmove}.
2205 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2207 Dump after the second instruction scheduling pass, to
2208 @file{@var{file}.16.sched2}.
2210 Dump after CSE (including the jump optimization that sometimes follows
2211 CSE), to @file{@var{file}.02.cse}.
2213 Dump after the first instruction scheduling pass, to
2214 @file{@var{file}.11.sched}.
2216 Dump after the second CSE pass (including the jump optimization that
2217 sometimes follows CSE), to @file{@var{file}.06.cse2}.
2219 Produce all the dumps listed above.
2221 Print statistics on memory usage, at the end of the run, to
2224 Annotate the assembler output with a comment indicating which
2225 pattern and alternative was used. The length of each instruction is
2228 For each of the other indicated dump files (except for
2229 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
2230 suitible for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
2232 Dump after the second flow pass to @file{@var{file}.14.flow2}.
2234 Just generate RTL for a function instead of compiling it. Usually used
2237 Dump debugging information during parsing, to standard error.
2239 Dump after the peephole2 pass to @file{@var{file}.15.peephole2}.
2242 @item -fdump-unnumbered
2243 When doing debugging dumps (see -d option above), suppress instruction
2244 numbers and line number note output. This makes it more feasible to
2245 use diff on debugging dumps for compiler invokations with different
2246 options, in particular with and without -g.
2248 @item -fdump-translation-unit-@var{file} (C++ only)
2249 Dump a representation of the tree structure for the entire translation
2252 @item -fpretend-float
2253 When running a cross-compiler, pretend that the target machine uses the
2254 same floating point format as the host machine. This causes incorrect
2255 output of the actual floating constants, but the actual instruction
2256 sequence will probably be the same as GCC would make when running on
2260 Store the usual ``temporary'' intermediate files permanently; place them
2261 in the current directory and name them based on the source file. Thus,
2262 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
2263 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}.
2266 Report the CPU time taken by each subprocess in the compilation
2267 sequence. For C source files, this is the preprocessor, compiler
2268 proper, and assembler. The output looks like this:
2276 The first number on each line is the ``user time,'' that is time spent
2277 executing the program itself. The second number is ``system time,''
2278 time spent executing operating system routines on behalf of the program.
2279 Both numbers are in seconds.
2281 @item -print-file-name=@var{library}
2282 Print the full absolute name of the library file @var{library} that
2283 would be used when linking---and don't do anything else. With this
2284 option, GCC does not compile or link anything; it just prints the
2287 @item -print-prog-name=@var{program}
2288 Like @samp{-print-file-name}, but searches for a program such as @samp{cpp}.
2290 @item -print-libgcc-file-name
2291 Same as @samp{-print-file-name=libgcc.a}.
2293 This is useful when you use @samp{-nostdlib} or @samp{-nodefaultlibs}
2294 but you do want to link with @file{libgcc.a}. You can do
2297 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
2300 @item -print-search-dirs
2301 Print the name of the configured installation directory and a list of
2302 program and library directories gcc will search---and don't do anything else.
2304 This is useful when gcc prints the error message
2305 @samp{installation problem, cannot exec cpp: No such file or directory}.
2306 To resolve this you either need to put @file{cpp} and the other compiler
2307 components where gcc expects to find them, or you can set the environment
2308 variable @code{GCC_EXEC_PREFIX} to the directory where you installed them.
2309 Don't forget the trailing '/'.
2310 @xref{Environment Variables}.
2313 @node Optimize Options
2314 @section Options That Control Optimization
2315 @cindex optimize options
2316 @cindex options, optimization
2318 These options control various sorts of optimizations:
2323 Optimize. Optimizing compilation takes somewhat more time, and a lot
2324 more memory for a large function.
2326 Without @samp{-O}, the compiler's goal is to reduce the cost of
2327 compilation and to make debugging produce the expected results.
2328 Statements are independent: if you stop the program with a breakpoint
2329 between statements, you can then assign a new value to any variable or
2330 change the program counter to any other statement in the function and
2331 get exactly the results you would expect from the source code.
2333 Without @samp{-O}, the compiler only allocates variables declared
2334 @code{register} in registers. The resulting compiled code is a little
2335 worse than produced by PCC without @samp{-O}.
2337 With @samp{-O}, the compiler tries to reduce code size and execution
2340 When you specify @samp{-O}, the compiler turns on @samp{-fthread-jumps}
2341 and @samp{-fdefer-pop} on all machines. The compiler turns on
2342 @samp{-fdelayed-branch} on machines that have delay slots, and
2343 @samp{-fomit-frame-pointer} on machines that can support debugging even
2344 without a frame pointer. On some machines the compiler also turns
2345 on other flags.@refill
2348 Optimize even more. GCC performs nearly all supported optimizations
2349 that do not involve a space-speed tradeoff. The compiler does not
2350 perform loop unrolling or function inlining when you specify @samp{-O2}.
2351 As compared to @samp{-O}, this option increases both compilation time
2352 and the performance of the generated code.
2354 @samp{-O2} turns on all optional optimizations except for loop unrolling
2355 and function inlining. It also turns on the @samp{-fforce-mem} option
2356 on all machines and frame pointer elimination on machines where doing so
2357 does not interfere with debugging.
2360 Optimize yet more. @samp{-O3} turns on all optimizations specified by
2361 @samp{-O2} and also turns on the @samp{inline-functions} option.
2367 Optimize for size. @samp{-Os} enables all @samp{-O2} optimizations that
2368 do not typically increase code size. It also performs further
2369 optimizations designed to reduce code size.
2371 If you use multiple @samp{-O} options, with or without level numbers,
2372 the last such option is the one that is effective.
2375 Options of the form @samp{-f@var{flag}} specify machine-independent
2376 flags. Most flags have both positive and negative forms; the negative
2377 form of @samp{-ffoo} would be @samp{-fno-foo}. In the table below,
2378 only one of the forms is listed---the one which is not the default.
2379 You can figure out the other form by either removing @samp{no-} or
2384 Do not store floating point variables in registers, and inhibit other
2385 options that might change whether a floating point value is taken from a
2388 @cindex floating point precision
2389 This option prevents undesirable excess precision on machines such as
2390 the 68000 where the floating registers (of the 68881) keep more
2391 precision than a @code{double} is supposed to have. Similarly for the
2392 x86 architecture. For most programs, the excess precision does only
2393 good, but a few programs rely on the precise definition of IEEE floating
2394 point. Use @samp{-ffloat-store} for such programs, after modifying
2395 them to store all pertinent intermediate computations into variables.
2397 @item -fno-default-inline
2398 Do not make member functions inline by default merely because they are
2399 defined inside the class scope (C++ only). Otherwise, when you specify
2400 @w{@samp{-O}}, member functions defined inside class scope are compiled
2401 inline by default; i.e., you don't need to add @samp{inline} in front of
2402 the member function name.
2404 @item -fno-defer-pop
2405 Always pop the arguments to each function call as soon as that function
2406 returns. For machines which must pop arguments after a function call,
2407 the compiler normally lets arguments accumulate on the stack for several
2408 function calls and pops them all at once.
2411 Force memory operands to be copied into registers before doing
2412 arithmetic on them. This produces better code by making all memory
2413 references potential common subexpressions. When they are not common
2414 subexpressions, instruction combination should eliminate the separate
2415 register-load. The @samp{-O2} option turns on this option.
2418 Force memory address constants to be copied into registers before
2419 doing arithmetic on them. This may produce better code just as
2420 @samp{-fforce-mem} may.
2422 @item -fomit-frame-pointer
2423 Don't keep the frame pointer in a register for functions that
2424 don't need one. This avoids the instructions to save, set up and
2425 restore frame pointers; it also makes an extra register available
2426 in many functions. @strong{It also makes debugging impossible on
2430 On some machines, such as the Vax, this flag has no effect, because
2431 the standard calling sequence automatically handles the frame pointer
2432 and nothing is saved by pretending it doesn't exist. The
2433 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2434 whether a target machine supports this flag. @xref{Registers}.@refill
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,,Register
2442 Usage, gcc.info, Using and Porting GCC}.@refill
2446 Don't pay attention to the @code{inline} keyword. Normally this option
2447 is used to keep the compiler from expanding any functions inline.
2448 Note that if you are not optimizing, no functions can be expanded inline.
2450 @item -finline-functions
2451 Integrate all simple functions into their callers. The compiler
2452 heuristically decides which functions are simple enough to be worth
2453 integrating in this way.
2455 If all calls to a given function are integrated, and the function is
2456 declared @code{static}, then the function is normally not output as
2457 assembler code in its own right.
2459 @item -finline-limit=@var{n}
2460 By default, gcc limits the size of functions that can be inlined. This flag
2461 allows the control of this limit for functions that are explicitly marked as
2462 inline (ie marked with the inline keyword or defined within the class
2463 definition in c++). @var{n} is the size of functions that can be inlined in
2464 number of pseudo instructions (not counting parameter handling). The default
2465 value of n is 10000. Increasing this value can result in more inlined code at
2466 the cost of compilation time and memory consumption. Decreasing usually makes
2467 the compilation faster and less code will be inlined (which presumably
2468 means slower programs). This option is particularly useful for programs that
2469 use inlining heavily such as those based on recursive templates with c++.
2471 @emph{Note:} pseudo instruction represents, in this particular context, an
2472 abstract measurement of function's size. In no way, it represents a count
2473 of assembly instructions and as such its exact meaning might change from one
2474 release to an another.
2476 @item -fkeep-inline-functions
2477 Even if all calls to a given function are integrated, and the function
2478 is declared @code{static}, nevertheless output a separate run-time
2479 callable version of the function. This switch does not affect
2480 @code{extern inline} functions.
2482 @item -fkeep-static-consts
2483 Emit variables declared @code{static const} when optimization isn't turned
2484 on, even if the variables aren't referenced.
2486 GCC enables this option by default. If you want to force the compiler to
2487 check if the variable was referenced, regardless of whether or not
2488 optimization is turned on, use the @samp{-fno-keep-static-consts} option.
2490 @item -fno-function-cse
2491 Do not put function addresses in registers; make each instruction that
2492 calls a constant function contain the function's address explicitly.
2494 This option results in less efficient code, but some strange hacks
2495 that alter the assembler output may be confused by the optimizations
2496 performed when this option is not used.
2499 This option allows GCC to violate some ANSI or IEEE rules and/or
2500 specifications in the interest of optimizing code for speed. For
2501 example, it allows the compiler to assume arguments to the @code{sqrt}
2502 function are non-negative numbers and that no floating-point values
2505 This option should never be turned on by any @samp{-O} option since
2506 it can result in incorrect output for programs which depend on
2507 an exact implementation of IEEE or ANSI rules/specifications for
2510 @item -fno-math-errno
2511 Do not set ERRNO after calling math functions that are executed
2512 with a single instruction, e.g., sqrt. A program that relies on
2513 IEEE exceptions for math error handling may want to use this flag
2514 for speed while maintaining IEEE arithmetic compatibility.
2516 The default is @samp{-fmath-errno}. The @samp{-ffast-math} option
2517 sets @samp{-fno-math-errno}.
2520 @c following causes underfulls.. they don't look great, but we deal.
2522 The following options control specific optimizations. The @samp{-O2}
2523 option turns on all of these optimizations except @samp{-funroll-loops}
2524 and @samp{-funroll-all-loops}. On most machines, the @samp{-O} option
2525 turns on the @samp{-fthread-jumps} and @samp{-fdelayed-branch} options,
2526 but specific machines may handle it differently.
2528 You can use the following flags in the rare cases when ``fine-tuning''
2529 of optimizations to be performed is desired.
2532 @item -fstrength-reduce
2533 Perform the optimizations of loop strength reduction and
2534 elimination of iteration variables.
2536 @item -fthread-jumps
2537 Perform optimizations where we check to see if a jump branches to a
2538 location where another comparison subsumed by the first is found. If
2539 so, the first branch is redirected to either the destination of the
2540 second branch or a point immediately following it, depending on whether
2541 the condition is known to be true or false.
2543 @item -fcse-follow-jumps
2544 In common subexpression elimination, scan through jump instructions
2545 when the target of the jump is not reached by any other path. For
2546 example, when CSE encounters an @code{if} statement with an
2547 @code{else} clause, CSE will follow the jump when the condition
2550 @item -fcse-skip-blocks
2551 This is similar to @samp{-fcse-follow-jumps}, but causes CSE to
2552 follow jumps which conditionally skip over blocks. When CSE
2553 encounters a simple @code{if} statement with no else clause,
2554 @samp{-fcse-skip-blocks} causes CSE to follow the jump around the
2555 body of the @code{if}.
2557 @item -frerun-cse-after-loop
2558 Re-run common subexpression elimination after loop optimizations has been
2561 @item -frerun-loop-opt
2562 Run the loop optimizer twice.
2565 Perform a global common subexpression elimination pass.
2566 This pass also performs global constant and copy propagation.
2568 @item -fdelete-null-pointer-checks
2569 Use global dataflow analysis to identify and eliminate useless null
2570 pointer checks. Programs which rely on NULL pointer dereferences @emph{not}
2571 halting the program may not work properly with this option. Use
2572 -fno-delete-null-pointer-checks to disable this optimizing for programs
2573 which depend on that behavior.
2576 @item -fexpensive-optimizations
2577 Perform a number of minor optimizations that are relatively expensive.
2579 @item -foptimize-register-moves
2581 Attempt to reassign register numbers in move instructions and as
2582 operands of other simple instructions in order to maximize the amount of
2583 register tying. This is especially helpful on machines with two-operand
2584 instructions. GCC enables this optimization by default with @samp{-O2}
2587 Note @code{-fregmove} and @code{-foptimize-register-moves} are the same
2590 @item -fdelayed-branch
2591 If supported for the target machine, attempt to reorder instructions
2592 to exploit instruction slots available after delayed branch
2595 @item -fschedule-insns
2596 If supported for the target machine, attempt to reorder instructions to
2597 eliminate execution stalls due to required data being unavailable. This
2598 helps machines that have slow floating point or memory load instructions
2599 by allowing other instructions to be issued until the result of the load
2600 or floating point instruction is required.
2602 @item -fschedule-insns2
2603 Similar to @samp{-fschedule-insns}, but requests an additional pass of
2604 instruction scheduling after register allocation has been done. This is
2605 especially useful on machines with a relatively small number of
2606 registers and where memory load instructions take more than one cycle.
2608 @item -ffunction-sections
2609 @itemx -fdata-sections
2610 Place each function or data item into its own section in the output
2611 file if the target supports arbitrary sections. The name of the
2612 function or the name of the data item determines the section's name
2615 Use these options on systems where the linker can perform optimizations
2616 to improve locality of reference in the instruction space. HPPA
2617 processors running HP-UX and Sparc processors running Solaris 2 have
2618 linkers with such optimizations. Other systems using the ELF object format
2619 as well as AIX may have these optimizations in the future.
2621 Only use these options when there are significant benefits from doing
2622 so. When you specify these options, the assembler and linker will
2623 create larger object and executable files and will also be slower.
2624 You will not be able to use @code{gprof} on all systems if you
2625 specify this option and you may have problems with debugging if
2626 you specify both this option and @samp{-g}.
2628 @item -fcaller-saves
2629 Enable values to be allocated in registers that will be clobbered by
2630 function calls, by emitting extra instructions to save and restore the
2631 registers around such calls. Such allocation is done only when it
2632 seems to result in better code than would otherwise be produced.
2634 This option is always enabled by default on certain machines, usually
2635 those which have no call-preserved registers to use instead.
2637 For all machines, optimization level 2 and higher enables this flag by
2640 @item -funroll-loops
2641 Perform the optimization of loop unrolling. This is only done for loops
2642 whose number of iterations can be determined at compile time or run time.
2643 @samp{-funroll-loops} implies both @samp{-fstrength-reduce} and
2644 @samp{-frerun-cse-after-loop}.
2646 @item -funroll-all-loops
2647 Perform the optimization of loop unrolling. This is done for all loops
2648 and usually makes programs run more slowly. @samp{-funroll-all-loops}
2649 implies @samp{-fstrength-reduce} as well as @samp{-frerun-cse-after-loop}.
2651 @item -fmove-all-movables
2652 Forces all invariant computations in loops to be moved
2655 @item -freduce-all-givs
2656 Forces all general-induction variables in loops to be
2659 @emph{Note:} When compiling programs written in Fortran,
2660 @samp{-fmove-all-movables} and @samp{-freduce-all-givs} are enabled
2661 by default when you use the optimizer.
2663 These options may generate better or worse code; results are highly
2664 dependent on the structure of loops within the source code.
2666 These two options are intended to be removed someday, once
2667 they have helped determine the efficacy of various
2668 approaches to improving loop optimizations.
2670 Please let us (@code{gcc@@gcc.gnu.org} and @code{fortran@@gnu.org})
2671 know how use of these options affects
2672 the performance of your production code.
2673 We're very interested in code that runs @emph{slower}
2674 when these options are @emph{enabled}.
2677 Disable any machine-specific peephole optimizations.
2679 @item -fbranch-probabilities
2680 After running a program compiled with @samp{-fprofile-arcs}
2681 (@pxref{Debugging Options,, Options for Debugging Your Program or
2682 @code{gcc}}), you can compile it a second time using
2683 @samp{-fbranch-probabilities}, to improve optimizations based on
2684 guessing the path a branch might take.
2687 With @samp{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
2688 note on the first instruction of each basic block, and a
2689 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
2690 These can be used to improve optimization. Currently, they are only
2691 used in one place: in @file{reorg.c}, instead of guessing which path a
2692 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
2693 exactly determine which path is taken more often.
2696 @item -fstrict-aliasing
2697 Allows the compiler to assume the strictest aliasing rules applicable to
2698 the language being compiled. For C (and C++), this activates
2699 optimizations based on the type of expressions. In particular, an
2700 object of one type is assumed never to reside at the same address as an
2701 object of a different type, unless the types are almost the same. For
2702 example, an @code{unsigned int} can alias an @code{int}, but not a
2703 @code{void*} or a @code{double}. A character type may alias any other
2706 Pay special attention to code like this:
2719 The practice of reading from a different union member than the one most
2720 recently written to (called ``type-punning'') is common. Even with
2721 @samp{-fstrict-aliasing}, type-punning is allowed, provided the memory
2722 is accessed through the union type. So, the code above will work as
2723 expected. However, this code might not:
2735 Every language that wishes to perform language-specific alias analysis
2736 should define a function that computes, given an @code{tree}
2737 node, an alias set for the node. Nodes in different alias sets are not
2738 allowed to alias. For an example, see the C front-end function
2739 @code{c_get_alias_set}.
2742 @item -falign-functions
2743 @itemx -falign-functions=@var{n}
2744 Align the start of functions to the next power-of-two greater than
2745 @var{n}, skipping up to @var{n} bytes. For instance,
2746 @samp{-falign-functions=32} aligns functions to the next 32-byte
2747 boundary, but @samp{-falign-functions=24} would align to the next
2748 32-byte boundary only if this can be done by skipping 23 bytes or less.
2750 @samp{-fno-align-functions} and @samp{-falign-functions=1} are
2751 equivalent and mean that functions will not be aligned.
2753 Some assemblers only support this flag when @var{n} is a power of two;
2754 in that case, it is rounded up.
2756 If @var{n} is not specified, use a machine-dependent default.
2758 @item -falign-labels
2759 @itemx -falign-labels=@var{n}
2760 Align all branch targets to a power-of-two boundary, skipping up to
2761 @var{n} bytes like @samp{-falign-functions}. This option can easily
2762 make code slower, because it must insert dummy operations for when the
2763 branch target is reached in the usual flow of the code.
2765 If @samp{-falign-loops} or @samp{-falign-jumps} are applicable and
2766 are greater than this value, then their values are used instead.
2768 If @var{n} is not specified, use a machine-dependent default which is
2769 very likely to be @samp{1}, meaning no alignment.
2772 @itemx -falign-loops=@var{n}
2773 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
2774 like @samp{-falign-functions}. The hope is that the loop will be
2775 executed many times, which will make up for any execution of the dummy
2778 If @var{n} is not specified, use a machine-dependent default.
2781 @itemx -falign-jumps=@var{n}
2782 Align branch targets to a power-of-two boundary, for branch targets
2783 where the targets can only be reached by jumping, skipping up to @var{n}
2784 bytes like @samp{-falign-functions}. In this case, no dummy operations
2787 If @var{n} is not specified, use a machine-dependent default.
2791 @node Preprocessor Options
2792 @section Options Controlling the Preprocessor
2793 @cindex preprocessor options
2794 @cindex options, preprocessor
2796 These options control the C preprocessor, which is run on each C source
2797 file before actual compilation.
2799 If you use the @samp{-E} option, nothing is done except preprocessing.
2800 Some of these options make sense only together with @samp{-E} because
2801 they cause the preprocessor output to be unsuitable for actual
2805 @item -include @var{file}
2806 Process @var{file} as input before processing the regular input file.
2807 In effect, the contents of @var{file} are compiled first. Any @samp{-D}
2808 and @samp{-U} options on the command line are always processed before
2809 @samp{-include @var{file}}, regardless of the order in which they are
2810 written. All the @samp{-include} and @samp{-imacros} options are
2811 processed in the order in which they are written.
2813 @item -imacros @var{file}
2814 Process @var{file} as input, discarding the resulting output, before
2815 processing the regular input file. Because the output generated from
2816 @var{file} is discarded, the only effect of @samp{-imacros @var{file}}
2817 is to make the macros defined in @var{file} available for use in the
2820 Any @samp{-D} and @samp{-U} options on the command line are always
2821 processed before @samp{-imacros @var{file}}, regardless of the order in
2822 which they are written. All the @samp{-include} and @samp{-imacros}
2823 options are processed in the order in which they are written.
2825 @item -idirafter @var{dir}
2826 @cindex second include path
2827 Add the directory @var{dir} to the second include path. The directories
2828 on the second include path are searched when a header file is not found
2829 in any of the directories in the main include path (the one that
2832 @item -iprefix @var{prefix}
2833 Specify @var{prefix} as the prefix for subsequent @samp{-iwithprefix}
2836 @item -iwithprefix @var{dir}
2837 Add a directory to the second include path. The directory's name is
2838 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
2839 specified previously with @samp{-iprefix}. If you have not specified a
2840 prefix yet, the directory containing the installed passes of the
2841 compiler is used as the default.
2843 @item -iwithprefixbefore @var{dir}
2844 Add a directory to the main include path. The directory's name is made
2845 by concatenating @var{prefix} and @var{dir}, as in the case of
2846 @samp{-iwithprefix}.
2848 @item -isystem @var{dir}
2849 Add a directory to the beginning of the second include path, marking it
2850 as a system directory, so that it gets the same special treatment as
2851 is applied to the standard system directories.
2854 Do not search the standard system directories for header files. Only
2855 the directories you have specified with @samp{-I} options (and the
2856 current directory, if appropriate) are searched. @xref{Directory
2857 Options}, for information on @samp{-I}.
2859 By using both @samp{-nostdinc} and @samp{-I-}, you can limit the include-file
2860 search path to only those directories you specify explicitly.
2863 Do not predefine any nonstandard macros. (Including architecture flags).
2866 Run only the C preprocessor. Preprocess all the C source files
2867 specified and output the results to standard output or to the
2868 specified output file.
2871 Tell the preprocessor not to discard comments. Used with the
2875 Tell the preprocessor not to generate @samp{#line} directives.
2876 Used with the @samp{-E} option.
2879 @cindex dependencies, make
2881 Tell the preprocessor to output a rule suitable for @code{make}
2882 describing the dependencies of each object file. For each source file,
2883 the preprocessor outputs one @code{make}-rule whose target is the object
2884 file name for that source file and whose dependencies are all the
2885 @code{#include} header files it uses. This rule may be a single line or
2886 may be continued with @samp{\}-newline if it is long. The list of rules
2887 is printed on standard output instead of the preprocessed C program.
2889 @samp{-M} implies @samp{-E}.
2891 Another way to specify output of a @code{make} rule is by setting
2892 the environment variable @code{DEPENDENCIES_OUTPUT} (@pxref{Environment
2896 Like @samp{-M} but the output mentions only the user header files
2897 included with @samp{#include "@var{file}"}. System header files
2898 included with @samp{#include <@var{file}>} are omitted.
2901 Like @samp{-M} but the dependency information is written to a file made by
2902 replacing ".c" with ".d" at the end of the input file names.
2903 This is in addition to compiling the file as specified---@samp{-MD} does
2904 not inhibit ordinary compilation the way @samp{-M} does.
2906 In Mach, you can use the utility @code{md} to merge multiple dependency
2907 files into a single dependency file suitable for using with the @samp{make}
2911 Like @samp{-MD} except mention only user header files, not system
2915 Treat missing header files as generated files and assume they live in the
2916 same directory as the source file. If you specify @samp{-MG}, you
2917 must also specify either @samp{-M} or @samp{-MM}. @samp{-MG} is not
2918 supported with @samp{-MD} or @samp{-MMD}.
2921 Print the name of each header file used, in addition to other normal
2924 @item -A@var{question}(@var{answer})
2925 Assert the answer @var{answer} for @var{question}, in case it is tested
2926 with a preprocessing conditional such as @samp{#if
2927 #@var{question}(@var{answer})}. @samp{-A-} disables the standard
2928 assertions that normally describe the target machine.
2931 Define macro @var{macro} with the string @samp{1} as its definition.
2933 @item -D@var{macro}=@var{defn}
2934 Define macro @var{macro} as @var{defn}. All instances of @samp{-D} on
2935 the command line are processed before any @samp{-U} options.
2938 Undefine macro @var{macro}. @samp{-U} options are evaluated after all
2939 @samp{-D} options, but before any @samp{-include} and @samp{-imacros}
2943 Tell the preprocessor to output only a list of the macro definitions
2944 that are in effect at the end of preprocessing. Used with the @samp{-E}
2948 Tell the preprocessing to pass all macro definitions into the output, in
2949 their proper sequence in the rest of the output.
2952 Like @samp{-dD} except that the macro arguments and contents are omitted.
2953 Only @samp{#define @var{name}} is included in the output.
2956 Support ANSI C trigraphs. The @samp{-ansi} option also has this effect.
2958 @item -Wp,@var{option}
2959 Pass @var{option} as an option to the preprocessor. If @var{option}
2960 contains commas, it is split into multiple options at the commas.
2963 @node Assembler Options
2964 @section Passing Options to the Assembler
2966 @c prevent bad page break with this line
2967 You can pass options to the assembler.
2970 @item -Wa,@var{option}
2971 Pass @var{option} as an option to the assembler. If @var{option}
2972 contains commas, it is split into multiple options at the commas.
2976 @section Options for Linking
2977 @cindex link options
2978 @cindex options, linking
2980 These options come into play when the compiler links object files into
2981 an executable output file. They are meaningless if the compiler is
2982 not doing a link step.
2986 @item @var{object-file-name}
2987 A file name that does not end in a special recognized suffix is
2988 considered to name an object file or library. (Object files are
2989 distinguished from libraries by the linker according to the file
2990 contents.) If linking is done, these object files are used as input
2996 If any of these options is used, then the linker is not run, and
2997 object file names should not be used as arguments. @xref{Overall
3001 @item -l@var{library}
3002 Search the library named @var{library} when linking.
3004 It makes a difference where in the command you write this option; the
3005 linker searches processes libraries and object files in the order they
3006 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
3007 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
3008 to functions in @samp{z}, those functions may not be loaded.
3010 The linker searches a standard list of directories for the library,
3011 which is actually a file named @file{lib@var{library}.a}. The linker
3012 then uses this file as if it had been specified precisely by name.
3014 The directories searched include several standard system directories
3015 plus any that you specify with @samp{-L}.
3017 Normally the files found this way are library files---archive files
3018 whose members are object files. The linker handles an archive file by
3019 scanning through it for members which define symbols that have so far
3020 been referenced but not defined. But if the file that is found is an
3021 ordinary object file, it is linked in the usual fashion. The only
3022 difference between using an @samp{-l} option and specifying a file name
3023 is that @samp{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
3024 and searches several directories.
3027 You need this special case of the @samp{-l} option in order to
3028 link an Objective C program.
3031 Do not use the standard system startup files when linking.
3032 The standard system libraries are used normally, unless @code{-nostdlib}
3033 or @code{-nodefaultlibs} is used.
3035 @item -nodefaultlibs
3036 Do not use the standard system libraries when linking.
3037 Only the libraries you specify will be passed to the linker.
3038 The standard startup files are used normally, unless @code{-nostartfiles}
3039 is used. The compiler may generate calls to memcmp, memset, and memcpy
3040 for System V (and ANSI C) environments or to bcopy and bzero for
3041 BSD environments. These entries are usually resolved by entries in
3042 libc. These entry points should be supplied through some other
3043 mechanism when this option is specified.
3046 Do not use the standard system startup files or libraries when linking.
3047 No startup files and only the libraries you specify will be passed to
3048 the linker. The compiler may generate calls to memcmp, memset, and memcpy
3049 for System V (and ANSI C) environments or to bcopy and bzero for
3050 BSD environments. These entries are usually resolved by entries in
3051 libc. These entry points should be supplied through some other
3052 mechanism when this option is specified.
3054 @cindex @code{-lgcc}, use with @code{-nostdlib}
3055 @cindex @code{-nostdlib} and unresolved references
3056 @cindex unresolved references and @code{-nostdlib}
3057 @cindex @code{-lgcc}, use with @code{-nodefaultlibs}
3058 @cindex @code{-nodefaultlibs} and unresolved references
3059 @cindex unresolved references and @code{-nodefaultlibs}
3060 One of the standard libraries bypassed by @samp{-nostdlib} and
3061 @samp{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
3062 that GCC uses to overcome shortcomings of particular machines, or special
3063 needs for some languages.
3065 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
3069 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
3070 for more discussion of @file{libgcc.a}.)
3072 In most cases, you need @file{libgcc.a} even when you want to avoid
3073 other standard libraries. In other words, when you specify @samp{-nostdlib}
3074 or @samp{-nodefaultlibs} you should usually specify @samp{-lgcc} as well.
3075 This ensures that you have no unresolved references to internal GCC
3076 library subroutines. (For example, @samp{__main}, used to ensure C++
3077 constructors will be called; @pxref{Collect2,,@code{collect2}}.)
3080 Remove all symbol table and relocation information from the executable.
3083 On systems that support dynamic linking, this prevents linking with the shared
3084 libraries. On other systems, this option has no effect.
3087 Produce a shared object which can then be linked with other objects to
3088 form an executable. Not all systems support this option. You must
3089 also specify @samp{-fpic} or @samp{-fPIC} on some systems when
3090 you specify this option.
3093 Bind references to global symbols when building a shared object. Warn
3094 about any unresolved references (unless overridden by the link editor
3095 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
3098 @item -Xlinker @var{option}
3099 Pass @var{option} as an option to the linker. You can use this to
3100 supply system-specific linker options which GCC does not know how to
3103 If you want to pass an option that takes an argument, you must use
3104 @samp{-Xlinker} twice, once for the option and once for the argument.
3105 For example, to pass @samp{-assert definitions}, you must write
3106 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
3107 @samp{-Xlinker "-assert definitions"}, because this passes the entire
3108 string as a single argument, which is not what the linker expects.
3110 @item -Wl,@var{option}
3111 Pass @var{option} as an option to the linker. If @var{option} contains
3112 commas, it is split into multiple options at the commas.
3114 @item -u @var{symbol}
3115 Pretend the symbol @var{symbol} is undefined, to force linking of
3116 library modules to define it. You can use @samp{-u} multiple times with
3117 different symbols to force loading of additional library modules.
3120 @node Directory Options
3121 @section Options for Directory Search
3122 @cindex directory options
3123 @cindex options, directory search
3126 These options specify directories to search for header files, for
3127 libraries and for parts of the compiler:
3131 Add the directory @var{dir} to the head of the list of directories to be
3132 searched for header files. This can be used to override a system header
3133 file, substituting your own version, since these directories are
3134 searched before the system header file directories. If you use more
3135 than one @samp{-I} option, the directories are scanned in left-to-right
3136 order; the standard system directories come after.
3139 Any directories you specify with @samp{-I} options before the @samp{-I-}
3140 option are searched only for the case of @samp{#include "@var{file}"};
3141 they are not searched for @samp{#include <@var{file}>}.
3143 If additional directories are specified with @samp{-I} options after
3144 the @samp{-I-}, these directories are searched for all @samp{#include}
3145 directives. (Ordinarily @emph{all} @samp{-I} directories are used
3148 In addition, the @samp{-I-} option inhibits the use of the current
3149 directory (where the current input file came from) as the first search
3150 directory for @samp{#include "@var{file}"}. There is no way to
3151 override this effect of @samp{-I-}. With @samp{-I.} you can specify
3152 searching the directory which was current when the compiler was
3153 invoked. That is not exactly the same as what the preprocessor does
3154 by default, but it is often satisfactory.
3156 @samp{-I-} does not inhibit the use of the standard system directories
3157 for header files. Thus, @samp{-I-} and @samp{-nostdinc} are
3161 Add directory @var{dir} to the list of directories to be searched
3164 @item -B@var{prefix}
3165 This option specifies where to find the executables, libraries,
3166 include files, and data files of the compiler itself.
3168 The compiler driver program runs one or more of the subprograms
3169 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
3170 @var{prefix} as a prefix for each program it tries to run, both with and
3171 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
3173 For each subprogram to be run, the compiler driver first tries the
3174 @samp{-B} prefix, if any. If that name is not found, or if @samp{-B}
3175 was not specified, the driver tries two standard prefixes, which are
3176 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
3177 those results in a file name that is found, the unmodified program
3178 name is searched for using the directories specified in your
3179 @samp{PATH} environment variable.
3181 @samp{-B} prefixes that effectively specify directory names also apply
3182 to libraries in the linker, because the compiler translates these
3183 options into @samp{-L} options for the linker. They also apply to
3184 includes files in the preprocessor, because the compiler translates these
3185 options into @samp{-isystem} options for the preprocessor. In this case,
3186 the compiler appends @samp{include} to the prefix.
3188 The run-time support file @file{libgcc.a} can also be searched for using
3189 the @samp{-B} prefix, if needed. If it is not found there, the two
3190 standard prefixes above are tried, and that is all. The file is left
3191 out of the link if it is not found by those means.
3193 Another way to specify a prefix much like the @samp{-B} prefix is to use
3194 the environment variable @code{GCC_EXEC_PREFIX}. @xref{Environment
3197 @item -specs=@var{file}
3198 Process @var{file} after the compiler reads in the standard @file{specs}
3199 file, in order to override the defaults that the @file{gcc} driver
3200 program uses when determining what switches to pass to @file{cc1},
3201 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
3202 @samp{-specs=}@var{file} can be specified on the command line, and they
3203 are processed in order, from left to right.
3207 @section Specifying subprocesses and the switches to pass to them
3209 @code{GCC} is a driver program. It performs its job by invoking a
3210 sequence of other programs to do the work of compiling, assembling and
3211 linking. GCC interprets its command-line parameters and uses these to
3212 deduce which programs it should invoke, and which command-line options
3213 it ought to place on their command lines. This behaviour is controlled
3214 by @dfn{spec strings}. In most cases there is one spec string for each
3215 program that GCC can invoke, but a few programs have multiple spec
3216 strings to control their behaviour. The spec strings built into GCC can
3217 be overridden by using the @samp{-specs=} command-line switch to specify
3220 @dfn{Spec files} are plaintext files that are used to construct spec
3221 strings. They consist of a sequence of directives separated by blank
3222 lines. The type of directive is determined by the first non-whitespace
3223 character on the line and it can be one of the following:
3226 @item %@var{command}
3227 Issues a @var{command} to the spec file processor. The commands that can
3231 @item %include <@var{file}>
3233 Search for @var{file} and insert its text at the current point in the
3236 @item %include_noerr <@var{file}>
3237 @cindex %include_noerr
3238 Just like @samp{%include}, but do not generate an error message if the include
3239 file cannot be found.
3241 @item %rename @var{old_name} @var{new_name}
3243 Rename the spec string @var{old_name} to @var{new_name}.
3247 @item *[@var{spec_name}]:
3248 This tells the compiler to create, override or delete the named spec
3249 string. All lines after this directive up to the next directive or
3250 blank line are considered to be the text for the spec string. If this
3251 results in an empty string then the spec will be deleted. (Or, if the
3252 spec did not exist, then nothing will happened.) Otherwise, if the spec
3253 does not currently exist a new spec will be created. If the spec does
3254 exist then its contents will be overridden by the text of this
3255 directive, unless the first character of that text is the @samp{+}
3256 character, in which case the text will be appended to the spec.
3258 @item [@var{suffix}]:
3259 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
3260 and up to the next directive or blank line are considered to make up the
3261 spec string for the indicated suffix. When the compiler encounters an
3262 input file with the named suffix, it will processes the spec string in
3263 order to work out how to compile that file. For example:
3270 This says that any input file whose name ends in @samp{.ZZ} should be
3271 passed to the program @samp{z-compile}, which should be invoked with the
3272 command-line switch @samp{-input} and with the result of performing the
3273 @samp{%i} substitution. (See below.)
3275 As an alternative to providing a spec string, the text that follows a
3276 suffix directive can be one of the following:
3279 @item @@@var{language}
3280 This says that the suffix is an alias for a known @var{language}. This is
3281 similar to using the @code{-x} command-line switch to GCC to specify a
3282 language explicitly. For example:
3289 Says that .ZZ files are, in fact, C++ source files.
3292 This causes an error messages saying:
3295 @var{name} compiler not installed on this system.
3299 GCC already has an extensive list of suffixes built into it.
3300 This directive will add an entry to the end of the list of suffixes, but
3301 since the list is searched from the end backwards, it is effectively
3302 possible to override earlier entries using this technique.
3306 GCC has the following spec strings built into it. Spec files can
3307 override these strings or create their own. Note that individual
3308 targets can also add their own spec strings to this list.
3311 asm Options to pass to the assembler
3312 asm_final Options to pass to the assembler post-processor
3313 cpp Options to pass to the C preprocessor
3314 cc1 Options to pass to the C compiler
3315 cc1plus Options to pass to the C++ compiler
3316 endfile Object files to include at the end of the link
3317 link Options to pass to the linker
3318 lib Libraries to include on the command line to the linker
3319 libgcc Decides which GCC support library to pass to the linker
3320 linker Sets the name of the linker
3321 predefines Defines to be passed to the C preprocessor
3322 signed_char Defines to pass to CPP to say whether @code{char} is signed by default
3323 startfile Object files to include at the start of the link
3326 Here is a small example of a spec file:
3332 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
3335 This example renames the spec called @samp{lib} to @samp{old_lib} and
3336 then overrides the previous definition of @samp{lib} with a new one.
3337 The new definition adds in some extra command-line options before
3338 including the text of the old definition.
3340 @dfn{Spec strings} are a list of command-line options to be passed to their
3341 corresponding program. In addition, the spec strings can contain
3342 @samp{%}-prefixed sequences to substitute variable text or to
3343 conditionally insert text into the command line. Using these constructs
3344 it is possible to generate quite complex command lines.
3346 Here is a table of all defined @samp{%}-sequences for spec
3347 strings. Note that spaces are not generated automatically around the
3348 results of expanding these sequences. Therefore you can concatenate them
3349 together or combine them with constant text in a single argument.
3353 Substitute one @samp{%} into the program name or argument.
3356 Substitute the name of the input file being processed.
3359 Substitute the basename of the input file being processed.
3360 This is the substring up to (and not including) the last period
3361 and not including the directory.
3364 Marks the argument containing or following the @samp{%d} as a
3365 temporary file name, so that that file will be deleted if GCC exits
3366 successfully. Unlike @samp{%g}, this contributes no text to the
3369 @item %g@var{suffix}
3370 Substitute a file name that has suffix @var{suffix} and is chosen
3371 once per compilation, and mark the argument in the same way as
3372 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
3373 name is now chosen in a way that is hard to predict even when previously
3374 chosen file names are known. For example, @samp{%g.s ... %g.o ... %g.s}
3375 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
3376 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
3377 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
3378 was simply substituted with a file name chosen once per compilation,
3379 without regard to any appended suffix (which was therefore treated
3380 just like ordinary text), making such attacks more likely to succeed.
3382 @item %u@var{suffix}
3383 Like @samp{%g}, but generates a new temporary file name even if
3384 @samp{%u@var{suffix}} was already seen.
3386 @item %U@var{suffix}
3387 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
3388 new one if there is no such last file name. In the absence of any
3389 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
3390 the same suffix @emph{space}, so @samp{%g.s ... %U.s ... %g.s ... %U.s}
3391 would involve the generation of two distinct file names, one
3392 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
3393 simply substituted with a file name chosen for the previous @samp{%u},
3394 without regard to any appended suffix.
3397 Marks the argument containing or following the @samp{%w} as the
3398 designated output file of this compilation. This puts the argument
3399 into the sequence of arguments that @samp{%o} will substitute later.
3402 Substitutes the names of all the output files, with spaces
3403 automatically placed around them. You should write spaces
3404 around the @samp{%o} as well or the results are undefined.
3405 @samp{%o} is for use in the specs for running the linker.
3406 Input files whose names have no recognized suffix are not compiled
3407 at all, but they are included among the output files, so they will
3411 Substitutes the suffix for object files. Note that this is
3412 handled specially when it immediately follows @samp{%g, %u, or %U},
3413 because of the need for those to form complete file names. The
3414 handling is such that @samp{%O} is treated exactly as if it had already
3415 been substituted, except that @samp{%g, %u, and %U} do not currently
3416 support additional @var{suffix} characters following @samp{%O} as they would
3417 following, for example, @samp{.o}.
3420 Substitutes the standard macro predefinitions for the
3421 current target machine. Use this when running @code{cpp}.
3424 Like @samp{%p}, but puts @samp{__} before and after the name of each
3425 predefined macro, except for macros that start with @samp{__} or with
3426 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ANSI
3430 Substitute a @samp{-iprefix} option made from GCC_EXEC_PREFIX.
3433 Current argument is the name of a library or startup file of some sort.
3434 Search for that file in a standard list of directories and substitute
3435 the full name found.
3438 Print @var{str} as an error message. @var{str} is terminated by a newline.
3439 Use this when inconsistent options are detected.
3442 Output @samp{-} if the input for the current command is coming from a pipe.
3445 Substitute the contents of spec string @var{name} at this point.
3448 Like @samp{%(...)} but put @samp{__} around @samp{-D} arguments.
3450 @item %x@{@var{option}@}
3451 Accumulate an option for @samp{%X}.
3454 Output the accumulated linker options specified by @samp{-Wl} or a @samp{%x}
3458 Output the accumulated assembler options specified by @samp{-Wa}.
3461 Output the accumulated preprocessor options specified by @samp{-Wp}.
3464 Substitute the major version number of GCC.
3465 (For version 2.9.5, this is 2.)
3468 Substitute the minor version number of GCC.
3469 (For version 2.9.5, this is 9.)
3472 Process the @code{asm} spec. This is used to compute the
3473 switches to be passed to the assembler.
3476 Process the @code{asm_final} spec. This is a spec string for
3477 passing switches to an assembler post-processor, if such a program is
3481 Process the @code{link} spec. This is the spec for computing the
3482 command line passed to the linker. Typically it will make use of the
3483 @samp{%L %G %S %D and %E} sequences.
3486 Dump out a @samp{-L} option for each directory that GCC believes might
3487 contain startup files. If the target supports multilibs then the
3488 current multilib directory will be prepended to each of these paths.
3491 Process the @code{lib} spec. This is a spec string for deciding which
3492 libraries should be included on the command line to the linker.
3495 Process the @code{libgcc} spec. This is a spec string for deciding
3496 which GCC support library should be included on the command line to the linker.
3499 Process the @code{startfile} spec. This is a spec for deciding which
3500 object files should be the first ones passed to the linker. Typically
3501 this might be a file named @file{crt0.o}.
3504 Process the @code{endfile} spec. This is a spec string that specifies
3505 the last object files that will be passed to the linker.
3508 Process the @code{cpp} spec. This is used to construct the arguments
3509 to be passed to the C preprocessor.
3512 Process the @code{signed_char} spec. This is intended to be used
3513 to tell cpp whether a char is signed. It typically has the definition:
3515 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
3519 Process the @code{cc1} spec. This is used to construct the options to be
3520 passed to the actual C compiler (@samp{cc1}).
3523 Process the @code{cc1plus} spec. This is used to construct the options to be
3524 passed to the actual C++ compiler (@samp{cc1plus}).
3527 Substitute the variable part of a matched option. See below.
3528 Note that each comma in the substituted string is replaced by
3532 Substitutes the @code{-S} switch, if that switch was given to GCC.
3533 If that switch was not specified, this substitutes nothing. Note that
3534 the leading dash is omitted when specifying this option, and it is
3535 automatically inserted if the substitution is performed. Thus the spec
3536 string @samp{%@{foo@}} would match the command-line option @samp{-foo}
3537 and would output the command line option @samp{-foo}.
3539 @item %W@{@code{S}@}
3540 Like %@{@code{S}@} but mark last argument supplied within as a file to be
3543 @item %@{@code{S}*@}
3544 Substitutes all the switches specified to GCC whose names start
3545 with @code{-S}, but which also take an argument. This is used for
3546 switches like @samp{-o, -D, -I}, etc. GCC considers @samp{-o foo} as being
3547 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
3548 text, including the space. Thus two arguments would be generated.
3550 @item %@{^@code{S}*@}
3551 Like %@{@code{S}*@}, but don't put a blank between a switch and its
3552 argument. Thus %@{^o*@} would only generate one argument, not two.
3554 @item %@{@code{S}*:@code{X}@}
3555 Substitutes @code{X} if one or more switches whose names start with
3556 @code{-S} are specified to GCC. Note that the tail part of the
3557 @code{-S} option (i.e. the part matched by the @samp{*}) will be substituted
3558 for each occurrence of @samp{%*} within @code{X}.
3560 @item %@{@code{S}:@code{X}@}
3561 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC.
3563 @item %@{!@code{S}:@code{X}@}
3564 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC.
3566 @item %@{|@code{S}:@code{X}@}
3567 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
3569 @item %@{|!@code{S}:@code{X}@}
3570 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
3572 @item %@{.@code{S}:@code{X}@}
3573 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
3575 @item %@{!.@code{S}:@code{X}@}
3576 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
3578 @item %@{@code{S}|@code{P}:@code{X}@}
3579 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC. This may be
3580 combined with @samp{!} and @samp{.} sequences as well, although they
3581 have a stronger binding than the @samp{|}. For example a spec string
3585 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
3588 will output the following command-line options from the following input
3589 command-line options:
3594 -d fred.c -foo -baz -boggle
3595 -d jim.d -bar -baz -boggle
3600 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
3601 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
3602 or spaces, or even newlines. They are processed as usual, as described
3605 The @samp{-O, -f, -m, and -W} switches are handled specifically in these
3606 constructs. If another value of @samp{-O} or the negated form of a @samp{-f, -m, or
3607 -W} switch is found later in the command line, the earlier switch
3608 value is ignored, except with @{@code{S}*@} where @code{S} is just one
3609 letter, which passes all matching options.
3611 The character @samp{|} at the beginning of the predicate text is used to indicate
3612 that a command should be piped to the following command, but only if @samp{-pipe}
3615 It is built into GCC which switches take arguments and which do not.
3616 (You might think it would be useful to generalize this to allow each
3617 compiler's spec to say which switches take arguments. But this cannot
3618 be done in a consistent fashion. GCC cannot even decide which input
3619 files have been specified without knowing which switches take arguments,
3620 and it must know which input files to compile in order to tell which
3623 GCC also knows implicitly that arguments starting in @samp{-l} are to be
3624 treated as compiler output files, and passed to the linker in their
3625 proper position among the other output files.
3627 @node Target Options
3628 @section Specifying Target Machine and Compiler Version
3629 @cindex target options
3630 @cindex cross compiling
3631 @cindex specifying machine version
3632 @cindex specifying compiler version and target machine
3633 @cindex compiler version, specifying
3634 @cindex target machine, specifying
3636 By default, GCC compiles code for the same type of machine that you
3637 are using. However, it can also be installed as a cross-compiler, to
3638 compile for some other type of machine. In fact, several different
3639 configurations of GCC, for different target machines, can be
3640 installed side by side. Then you specify which one to use with the
3643 In addition, older and newer versions of GCC can be installed side
3644 by side. One of them (probably the newest) will be the default, but
3645 you may sometimes wish to use another.
3648 @item -b @var{machine}
3649 The argument @var{machine} specifies the target machine for compilation.
3650 This is useful when you have installed GCC as a cross-compiler.
3652 The value to use for @var{machine} is the same as was specified as the
3653 machine type when configuring GCC as a cross-compiler. For
3654 example, if a cross-compiler was configured with @samp{configure
3655 i386v}, meaning to compile for an 80386 running System V, then you
3656 would specify @samp{-b i386v} to run that cross compiler.
3658 When you do not specify @samp{-b}, it normally means to compile for
3659 the same type of machine that you are using.
3661 @item -V @var{version}
3662 The argument @var{version} specifies which version of GCC to run.
3663 This is useful when multiple versions are installed. For example,
3664 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
3666 The default version, when you do not specify @samp{-V}, is the last
3667 version of GCC that you installed.
3670 The @samp{-b} and @samp{-V} options actually work by controlling part of
3671 the file name used for the executable files and libraries used for
3672 compilation. A given version of GCC, for a given target machine, is
3673 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.@refill
3675 Thus, sites can customize the effect of @samp{-b} or @samp{-V} either by
3676 changing the names of these directories or adding alternate names (or
3677 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
3678 file @file{80386} is a link to the file @file{i386v}, then @samp{-b
3679 80386} becomes an alias for @samp{-b i386v}.
3681 In one respect, the @samp{-b} or @samp{-V} do not completely change
3682 to a different compiler: the top-level driver program @code{gcc}
3683 that you originally invoked continues to run and invoke the other
3684 executables (preprocessor, compiler per se, assembler and linker)
3685 that do the real work. However, since no real work is done in the
3686 driver program, it usually does not matter that the driver program
3687 in use is not the one for the specified target and version.
3689 The only way that the driver program depends on the target machine is
3690 in the parsing and handling of special machine-specific options.
3691 However, this is controlled by a file which is found, along with the
3692 other executables, in the directory for the specified version and
3693 target machine. As a result, a single installed driver program adapts
3694 to any specified target machine and compiler version.
3696 The driver program executable does control one significant thing,
3697 however: the default version and target machine. Therefore, you can
3698 install different instances of the driver program, compiled for
3699 different targets or versions, under different names.
3701 For example, if the driver for version 2.0 is installed as @code{ogcc}
3702 and that for version 2.1 is installed as @code{gcc}, then the command
3703 @code{gcc} will use version 2.1 by default, while @code{ogcc} will use
3704 2.0 by default. However, you can choose either version with either
3705 command with the @samp{-V} option.
3707 @node Submodel Options
3708 @section Hardware Models and Configurations
3709 @cindex submodel options
3710 @cindex specifying hardware config
3711 @cindex hardware models and configurations, specifying
3712 @cindex machine dependent options
3714 Earlier we discussed the standard option @samp{-b} which chooses among
3715 different installed compilers for completely different target
3716 machines, such as Vax vs. 68000 vs. 80386.
3718 In addition, each of these target machine types can have its own
3719 special options, starting with @samp{-m}, to choose among various
3720 hardware models or configurations---for example, 68010 vs 68020,
3721 floating coprocessor or none. A single installed version of the
3722 compiler can compile for any model or configuration, according to the
3725 Some configurations of the compiler also support additional special
3726 options, usually for compatibility with other compilers on the same
3730 These options are defined by the macro @code{TARGET_SWITCHES} in the
3731 machine description. The default for the options is also defined by
3732 that macro, which enables you to change the defaults.
3747 * RS/6000 and PowerPC Options::
3752 * Intel 960 Options::
3753 * DEC Alpha Options::
3757 * System V Options::
3758 * TMS320C3x/C4x Options::
3764 @node M680x0 Options
3765 @subsection M680x0 Options
3766 @cindex M680x0 options
3768 These are the @samp{-m} options defined for the 68000 series. The default
3769 values for these options depends on which style of 68000 was selected when
3770 the compiler was configured; the defaults for the most common choices are
3776 Generate output for a 68000. This is the default
3777 when the compiler is configured for 68000-based systems.
3779 Use this option for microcontrollers with a 68000 or EC000 core,
3780 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
3784 Generate output for a 68020. This is the default
3785 when the compiler is configured for 68020-based systems.
3788 Generate output containing 68881 instructions for floating point.
3789 This is the default for most 68020 systems unless @samp{-nfp} was
3790 specified when the compiler was configured.
3793 Generate output for a 68030. This is the default when the compiler is
3794 configured for 68030-based systems.
3797 Generate output for a 68040. This is the default when the compiler is
3798 configured for 68040-based systems.
3800 This option inhibits the use of 68881/68882 instructions that have to be
3801 emulated by software on the 68040. Use this option if your 68040 does not
3802 have code to emulate those instructions.
3805 Generate output for a 68060. This is the default when the compiler is
3806 configured for 68060-based systems.
3808 This option inhibits the use of 68020 and 68881/68882 instructions that
3809 have to be emulated by software on the 68060. Use this option if your 68060
3810 does not have code to emulate those instructions.
3813 Generate output for a CPU32. This is the default
3814 when the compiler is configured for CPU32-based systems.
3816 Use this option for microcontrollers with a
3817 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
3818 68336, 68340, 68341, 68349 and 68360.
3821 Generate output for a 520X "coldfire" family cpu. This is the default
3822 when the compiler is configured for 520X-based systems.
3824 Use this option for microcontroller with a 5200 core, including
3825 the MCF5202, MCF5203, MCF5204 and MCF5202.
3829 Generate output for a 68040, without using any of the new instructions.
3830 This results in code which can run relatively efficiently on either a
3831 68020/68881 or a 68030 or a 68040. The generated code does use the
3832 68881 instructions that are emulated on the 68040.
3835 Generate output for a 68060, without using any of the new instructions.
3836 This results in code which can run relatively efficiently on either a
3837 68020/68881 or a 68030 or a 68040. The generated code does use the
3838 68881 instructions that are emulated on the 68060.
3841 Generate output containing Sun FPA instructions for floating point.
3844 Generate output containing library calls for floating point.
3845 @strong{Warning:} the requisite libraries are not available for all m68k
3846 targets. Normally the facilities of the machine's usual C compiler are
3847 used, but this can't be done directly in cross-compilation. You must
3848 make your own arrangements to provide suitable library functions for
3849 cross-compilation. The embedded targets @samp{m68k-*-aout} and
3850 @samp{m68k-*-coff} do provide software floating point support.
3853 Consider type @code{int} to be 16 bits wide, like @code{short int}.
3856 Do not use the bit-field instructions. The @samp{-m68000}, @samp{-mcpu32}
3857 and @samp{-m5200} options imply @w{@samp{-mnobitfield}}.
3860 Do use the bit-field instructions. The @samp{-m68020} option implies
3861 @samp{-mbitfield}. This is the default if you use a configuration
3862 designed for a 68020.
3865 Use a different function-calling convention, in which functions
3866 that take a fixed number of arguments return with the @code{rtd}
3867 instruction, which pops their arguments while returning. This
3868 saves one instruction in the caller since there is no need to pop
3869 the arguments there.
3871 This calling convention is incompatible with the one normally
3872 used on Unix, so you cannot use it if you need to call libraries
3873 compiled with the Unix compiler.
3875 Also, you must provide function prototypes for all functions that
3876 take variable numbers of arguments (including @code{printf});
3877 otherwise incorrect code will be generated for calls to those
3880 In addition, seriously incorrect code will result if you call a
3881 function with too many arguments. (Normally, extra arguments are
3882 harmlessly ignored.)
3884 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
3885 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
3888 @itemx -mno-align-int
3889 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
3890 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
3891 boundary (@samp{-malign-int}) or a 16-bit boundary (@samp{-mno-align-int}).
3892 Aligning variables on 32-bit boundaries produces code that runs somewhat
3893 faster on processors with 32-bit busses at the expense of more memory.
3895 @strong{Warning:} if you use the @samp{-malign-int} switch, GCC will
3896 align structures containing the above types differently than
3897 most published application binary interface specifications for the m68k.
3900 Use the pc-relative addressing mode of the 68000 directly, instead of
3901 using a global offset table. At present, this option implies -fpic,
3902 allowing at most a 16-bit offset for pc-relative addressing. -fPIC is
3903 not presently supported with -mpcrel, though this could be supported for
3904 68020 and higher processors.
3906 @item -mno-strict-align
3907 @itemx -mstrict-align
3908 @kindex -mstrict-align
3909 Do not (do) assume that unaligned memory references will be handled by
3915 @subsection VAX Options
3918 These @samp{-m} options are defined for the Vax:
3922 Do not output certain jump instructions (@code{aobleq} and so on)
3923 that the Unix assembler for the Vax cannot handle across long
3927 Do output those jump instructions, on the assumption that you
3928 will assemble with the GNU assembler.
3931 Output code for g-format floating point numbers instead of d-format.
3935 @subsection SPARC Options
3936 @cindex SPARC options
3938 These @samp{-m} switches are supported on the SPARC:
3943 Specify @samp{-mapp-regs} to generate output using the global registers
3944 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
3947 To be fully SVR4 ABI compliant at the cost of some performance loss,
3948 specify @samp{-mno-app-regs}. You should compile libraries and system
3949 software with this option.
3953 Generate output containing floating point instructions. This is the
3958 Generate output containing library calls for floating point.
3959 @strong{Warning:} the requisite libraries are not available for all SPARC
3960 targets. Normally the facilities of the machine's usual C compiler are
3961 used, but this cannot be done directly in cross-compilation. You must make
3962 your own arrangements to provide suitable library functions for
3963 cross-compilation. The embedded targets @samp{sparc-*-aout} and
3964 @samp{sparclite-*-*} do provide software floating point support.
3966 @samp{-msoft-float} changes the calling convention in the output file;
3967 therefore, it is only useful if you compile @emph{all} of a program with
3968 this option. In particular, you need to compile @file{libgcc.a}, the
3969 library that comes with GCC, with @samp{-msoft-float} in order for
3972 @item -mhard-quad-float
3973 Generate output containing quad-word (long double) floating point
3976 @item -msoft-quad-float
3977 Generate output containing library calls for quad-word (long double)
3978 floating point instructions. The functions called are those specified
3979 in the SPARC ABI. This is the default.
3981 As of this writing, there are no sparc implementations that have hardware
3982 support for the quad-word floating point instructions. They all invoke
3983 a trap handler for one of these instructions, and then the trap handler
3984 emulates the effect of the instruction. Because of the trap handler overhead,
3985 this is much slower than calling the ABI library routines. Thus the
3986 @samp{-msoft-quad-float} option is the default.
3990 With @samp{-mepilogue} (the default), the compiler always emits code for
3991 function exit at the end of each function. Any function exit in
3992 the middle of the function (such as a return statement in C) will
3993 generate a jump to the exit code at the end of the function.
3995 With @samp{-mno-epilogue}, the compiler tries to emit exit code inline
3996 at every function exit.
4000 With @samp{-mflat}, the compiler does not generate save/restore instructions
4001 and will use a "flat" or single register window calling convention.
4002 This model uses %i7 as the frame pointer and is compatible with the normal
4003 register window model. Code from either may be intermixed.
4004 The local registers and the input registers (0-5) are still treated as
4005 "call saved" registers and will be saved on the stack as necessary.
4007 With @samp{-mno-flat} (the default), the compiler emits save/restore
4008 instructions (except for leaf functions) and is the normal mode of operation.
4010 @item -mno-unaligned-doubles
4011 @itemx -munaligned-doubles
4012 Assume that doubles have 8 byte alignment. This is the default.
4014 With @samp{-munaligned-doubles}, GCC assumes that doubles have 8 byte
4015 alignment only if they are contained in another type, or if they have an
4016 absolute address. Otherwise, it assumes they have 4 byte alignment.
4017 Specifying this option avoids some rare compatibility problems with code
4018 generated by other compilers. It is not the default because it results
4019 in a performance loss, especially for floating point code.
4021 @item -mno-faster-structs
4022 @itemx -mfaster-structs
4023 With @samp{-mfaster-structs}, the compiler assumes that structures
4024 should have 8 byte alignment. This enables the use of pairs of
4025 @code{ldd} and @code{std} instructions for copies in structure
4026 assignment, in place of twice as many @code{ld} and @code{st} pairs.
4027 However, the use of this changed alignment directly violates the Sparc
4028 ABI. Thus, it's intended only for use on targets where the developer
4029 acknowledges that their resulting code will not be directly in line with
4030 the rules of the ABI.
4034 These two options select variations on the SPARC architecture.
4036 By default (unless specifically configured for the Fujitsu SPARClite),
4037 GCC generates code for the v7 variant of the SPARC architecture.
4039 @samp{-mv8} will give you SPARC v8 code. The only difference from v7
4040 code is that the compiler emits the integer multiply and integer
4041 divide instructions which exist in SPARC v8 but not in SPARC v7.
4043 @samp{-msparclite} will give you SPARClite code. This adds the integer
4044 multiply, integer divide step and scan (@code{ffs}) instructions which
4045 exist in SPARClite but not in SPARC v7.
4047 These options are deprecated and will be deleted in a future GCC release.
4048 They have been replaced with @samp{-mcpu=xxx}.
4052 These two options select the processor for which the code is optimised.
4054 With @samp{-mcypress} (the default), the compiler optimizes code for the
4055 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
4056 This is also appropriate for the older SparcStation 1, 2, IPX etc.
4058 With @samp{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
4059 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
4060 of the full SPARC v8 instruction set.
4062 These options are deprecated and will be deleted in a future GCC release.
4063 They have been replaced with @samp{-mcpu=xxx}.
4065 @item -mcpu=@var{cpu_type}
4066 Set the instruction set, register set, and instruction scheduling parameters
4067 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
4068 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
4069 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
4070 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
4072 Default instruction scheduling parameters are used for values that select
4073 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
4074 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
4076 Here is a list of each supported architecture and their supported
4081 v8: supersparc, hypersparc
4082 sparclite: f930, f934, sparclite86x
4087 @item -mtune=@var{cpu_type}
4088 Set the instruction scheduling parameters for machine type
4089 @var{cpu_type}, but do not set the instruction set or register set that the
4090 option @samp{-mcpu=}@var{cpu_type} would.
4092 The same values for @samp{-mcpu=}@var{cpu_type} are used for
4093 @samp{-mtune=}@*@var{cpu_type}, though the only useful values are those that
4094 select a particular cpu implementation: @samp{cypress}, @samp{supersparc},
4095 @samp{hypersparc}, @samp{f930}, @samp{f934}, @samp{sparclite86x},
4096 @samp{tsc701}, @samp{ultrasparc}.
4100 These @samp{-m} switches are supported in addition to the above
4101 on the SPARCLET processor.
4104 @item -mlittle-endian
4105 Generate code for a processor running in little-endian mode.
4108 Treat register @code{%g0} as a normal register.
4109 GCC will continue to clobber it as necessary but will not assume
4110 it always reads as 0.
4112 @item -mbroken-saverestore
4113 Generate code that does not use non-trivial forms of the @code{save} and
4114 @code{restore} instructions. Early versions of the SPARCLET processor do
4115 not correctly handle @code{save} and @code{restore} instructions used with
4116 arguments. They correctly handle them used without arguments. A @code{save}
4117 instruction used without arguments increments the current window pointer
4118 but does not allocate a new stack frame. It is assumed that the window
4119 overflow trap handler will properly handle this case as will interrupt
4123 These @samp{-m} switches are supported in addition to the above
4124 on SPARC V9 processors in 64 bit environments.
4127 @item -mlittle-endian
4128 Generate code for a processor running in little-endian mode.
4132 Generate code for a 32 bit or 64 bit environment.
4133 The 32 bit environment sets int, long and pointer to 32 bits.
4134 The 64 bit environment sets int to 32 bits and long and pointer
4137 @item -mcmodel=medlow
4138 Generate code for the Medium/Low code model: the program must be linked
4139 in the low 32 bits of the address space. Pointers are 64 bits.
4140 Programs can be statically or dynamically linked.
4142 @item -mcmodel=medmid
4143 Generate code for the Medium/Middle code model: the program must be linked
4144 in the low 44 bits of the address space, the text segment must be less than
4145 2G bytes, and data segment must be within 2G of the text segment.
4146 Pointers are 64 bits.
4148 @item -mcmodel=medany
4149 Generate code for the Medium/Anywhere code model: the program may be linked
4150 anywhere in the address space, the text segment must be less than
4151 2G bytes, and data segment must be within 2G of the text segment.
4152 Pointers are 64 bits.
4154 @item -mcmodel=embmedany
4155 Generate code for the Medium/Anywhere code model for embedded systems:
4156 assume a 32 bit text and a 32 bit data segment, both starting anywhere
4157 (determined at link time). Register %g4 points to the base of the
4158 data segment. Pointers still 64 bits.
4159 Programs are statically linked, PIC is not supported.
4162 @itemx -mno-stack-bias
4163 With @samp{-mstack-bias}, GCC assumes that the stack pointer, and
4164 frame pointer if present, are offset by -2047 which must be added back
4165 when making stack frame references.
4166 Otherwise, assume no such offset is present.
4169 @node Convex Options
4170 @subsection Convex Options
4171 @cindex Convex options
4173 These @samp{-m} options are defined for Convex:
4177 Generate output for C1. The code will run on any Convex machine.
4178 The preprocessor symbol @code{__convex__c1__} is defined.
4181 Generate output for C2. Uses instructions not available on C1.
4182 Scheduling and other optimizations are chosen for max performance on C2.
4183 The preprocessor symbol @code{__convex_c2__} is defined.
4186 Generate output for C32xx. Uses instructions not available on C1.
4187 Scheduling and other optimizations are chosen for max performance on C32.
4188 The preprocessor symbol @code{__convex_c32__} is defined.
4191 Generate output for C34xx. Uses instructions not available on C1.
4192 Scheduling and other optimizations are chosen for max performance on C34.
4193 The preprocessor symbol @code{__convex_c34__} is defined.
4196 Generate output for C38xx. Uses instructions not available on C1.
4197 Scheduling and other optimizations are chosen for max performance on C38.
4198 The preprocessor symbol @code{__convex_c38__} is defined.
4201 Generate code which puts an argument count in the word preceding each
4202 argument list. This is compatible with regular CC, and a few programs
4203 may need the argument count word. GDB and other source-level debuggers
4204 do not need it; this info is in the symbol table.
4207 Omit the argument count word. This is the default.
4209 @item -mvolatile-cache
4210 Allow volatile references to be cached. This is the default.
4212 @item -mvolatile-nocache
4213 Volatile references bypass the data cache, going all the way to memory.
4214 This is only needed for multi-processor code that does not use standard
4215 synchronization instructions. Making non-volatile references to volatile
4216 locations will not necessarily work.
4219 Type long is 32 bits, the same as type int. This is the default.
4222 Type long is 64 bits, the same as type long long. This option is useless,
4223 because no library support exists for it.
4226 @node AMD29K Options
4227 @subsection AMD29K Options
4228 @cindex AMD29K options
4230 These @samp{-m} options are defined for the AMD Am29000:
4235 @cindex DW bit (29k)
4236 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
4237 halfword operations are directly supported by the hardware. This is the
4242 Generate code that assumes the @code{DW} bit is not set.
4246 @cindex byte writes (29k)
4247 Generate code that assumes the system supports byte and halfword write
4248 operations. This is the default.
4252 Generate code that assumes the systems does not support byte and
4253 halfword write operations. @samp{-mnbw} implies @samp{-mndw}.
4257 @cindex memory model (29k)
4258 Use a small memory model that assumes that all function addresses are
4259 either within a single 256 KB segment or at an absolute address of less
4260 than 256k. This allows the @code{call} instruction to be used instead
4261 of a @code{const}, @code{consth}, @code{calli} sequence.
4265 Use the normal memory model: Generate @code{call} instructions only when
4266 calling functions in the same file and @code{calli} instructions
4267 otherwise. This works if each file occupies less than 256 KB but allows
4268 the entire executable to be larger than 256 KB. This is the default.
4271 Always use @code{calli} instructions. Specify this option if you expect
4272 a single file to compile into more than 256 KB of code.
4276 @cindex processor selection (29k)
4277 Generate code for the Am29050.
4281 Generate code for the Am29000. This is the default.
4283 @item -mkernel-registers
4284 @kindex -mkernel-registers
4285 @cindex kernel and user registers (29k)
4286 Generate references to registers @code{gr64-gr95} instead of to
4287 registers @code{gr96-gr127}. This option can be used when compiling
4288 kernel code that wants a set of global registers disjoint from that used
4291 Note that when this option is used, register names in @samp{-f} flags
4292 must use the normal, user-mode, names.
4294 @item -muser-registers
4295 @kindex -muser-registers
4296 Use the normal set of global registers, @code{gr96-gr127}. This is the
4300 @itemx -mno-stack-check
4301 @kindex -mstack-check
4302 @cindex stack checks (29k)
4303 Insert (or do not insert) a call to @code{__msp_check} after each stack
4304 adjustment. This is often used for kernel code.
4307 @itemx -mno-storem-bug
4308 @kindex -mstorem-bug
4309 @cindex storem bug (29k)
4310 @samp{-mstorem-bug} handles 29k processors which cannot handle the
4311 separation of a mtsrim insn and a storem instruction (most 29000 chips
4312 to date, but not the 29050).
4314 @item -mno-reuse-arg-regs
4315 @itemx -mreuse-arg-regs
4316 @kindex -mreuse-arg-regs
4317 @samp{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
4318 registers for copying out arguments. This helps detect calling a function
4319 with fewer arguments than it was declared with.
4321 @item -mno-impure-text
4322 @itemx -mimpure-text
4323 @kindex -mimpure-text
4324 @samp{-mimpure-text}, used in addition to @samp{-shared}, tells the compiler to
4325 not pass @samp{-assert pure-text} to the linker when linking a shared object.
4328 @kindex -msoft-float
4329 Generate output containing library calls for floating point.
4330 @strong{Warning:} the requisite libraries are not part of GCC.
4331 Normally the facilities of the machine's usual C compiler are used, but
4332 this can't be done directly in cross-compilation. You must make your
4333 own arrangements to provide suitable library functions for
4338 Do not generate multm or multmu instructions. This is useful for some embedded
4339 systems which do not have trap handlers for these instructions.
4343 @subsection ARM Options
4346 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
4351 @kindex -mapcs-frame
4352 Generate a stack frame that is compliant with the ARM Procedure Call
4353 Standard for all functions, even if this is not strictly necessary for
4354 correct execution of the code. Specifying @samp{-fomit-frame-pointer}
4355 with this option will cause the stack frames not to be generated for
4356 leaf functions. The default is @samp{-mno-apcs-frame}.
4360 This is a synonym for @samp{-mapcs-frame}.
4364 Generate code for a processor running with a 26-bit program counter,
4365 and conforming to the function calling standards for the APCS 26-bit
4366 option. This option replaces the @samp{-m2} and @samp{-m3} options
4367 of previous releases of the compiler.
4371 Generate code for a processor running with a 32-bit program counter,
4372 and conforming to the function calling standards for the APCS 32-bit
4373 option. This option replaces the @samp{-m6} option of previous releases
4376 @item -mapcs-stack-check
4377 @kindex -mapcs-stack-check
4378 @kindex -mno-apcs-stack-check
4379 Generate code to check the amount of stack space available upon entry to
4380 every function (that actually uses some stack space). If there is
4381 insufficient space available then either the function
4382 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
4383 called, depending upon the amount of stack space required. The run time
4384 system is required to provide these functions. The default is
4385 @samp{-mno-apcs-stack-check}, since this produces smaller code.
4388 @kindex -mapcs-float
4389 @kindex -mno-apcs-float
4390 Pass floating point arguments using the float point registers. This is
4391 one of the variants of the APCS. This option is recommended if the
4392 target hardware has a floating point unit or if a lot of floating point
4393 arithmetic is going to be performed by the code. The default is
4394 @samp{-mno-apcs-float}, since integer only code is slightly increased in
4395 size if @samp{-mapcs-float} is used.
4397 @item -mapcs-reentrant
4398 @kindex -mapcs-reentrant
4399 @kindex -mno-apcs-reentrant
4400 Generate reentrant, position independent code. This is the equivalent
4401 to specifying the @samp{-fpic} option. The default is
4402 @samp{-mno-apcs-reentrant}.
4404 @item -mthumb-interwork
4405 @kindex -mthumb-interwork
4406 @kindex -mno-thumb-interwork
4407 Generate code which supports calling between the ARM and THUMB
4408 instruction sets. Without this option the two instruction sets cannot
4409 be reliably used inside one program. The default is
4410 @samp{-mno-thumb-interwork}, since slightly larger code is generated
4411 when @samp{-mthumb-interwork} is specified.
4413 @item -mno-sched-prolog
4414 @kindex -mno-sched-prolog
4415 @kindex -msched-prolog
4416 Prevent the reordering of instructions in the function prolog, or the
4417 merging of those instruction with the instructions in the function's
4418 body. This means that all functions will start with a recognizable set
4419 of instructions (or in fact one of a choice from a small set of
4420 different function prologues), and this information can be used to
4421 locate the start if functions inside an executable piece of code. The
4422 default is @samp{-msched-prolog}.
4425 Generate output containing floating point instructions. This is the
4429 Generate output containing library calls for floating point.
4430 @strong{Warning:} the requisite libraries are not available for all ARM
4431 targets. Normally the facilities of the machine's usual C compiler are
4432 used, but this cannot be done directly in cross-compilation. You must make
4433 your own arrangements to provide suitable library functions for
4436 @samp{-msoft-float} changes the calling convention in the output file;
4437 therefore, it is only useful if you compile @emph{all} of a program with
4438 this option. In particular, you need to compile @file{libgcc.a}, the
4439 library that comes with GCC, with @samp{-msoft-float} in order for
4442 @item -mlittle-endian
4443 Generate code for a processor running in little-endian mode. This is
4444 the default for all standard configurations.
4447 Generate code for a processor running in big-endian mode; the default is
4448 to compile code for a little-endian processor.
4450 @item -mwords-little-endian
4451 This option only applies when generating code for big-endian processors.
4452 Generate code for a little-endian word order but a big-endian byte
4453 order. That is, a byte order of the form @samp{32107654}. Note: this
4454 option should only be used if you require compatibility with code for
4455 big-endian ARM processors generated by versions of the compiler prior to
4458 @item -malignment-traps
4459 @kindex -malignment-traps
4460 Generate code that will not trap if the MMU has alignment traps enabled.
4461 On ARM architectures prior to ARMv4, there were no instructions to
4462 access half-word objects stored in memory. However, when reading from
4463 memory a feature of the ARM architecture allows a word load to be used,
4464 even if the address is unaligned, and the processor core will rotate the
4465 data as it is being loaded. This option tells the compiler that such
4466 misaligned accesses will cause a MMU trap and that it should instead
4467 synthesise the access as a series of byte accesses. The compiler can
4468 still use word accesses to load half-word data if it knows that the
4469 address is aligned to a word boundary.
4471 This option is ignored when compiling for ARM architecture 4 or later,
4472 since these processors have instructions to directly access half-word
4475 @item -mno-alignment-traps
4476 @kindex -mno-alignment-traps
4477 Generate code that assumes that the MMU will not trap unaligned
4478 accesses. This produces better code when the target instruction set
4479 does not have half-word memory operations (implementations prior to
4482 Note that you cannot use this option to access unaligned word objects,
4483 since the processor will only fetch one 32-bit aligned object from
4486 The default setting for most targets is -mno-alignment-traps, since
4487 this produces better code when there are no half-word memory
4488 instructions available.
4490 @item -mshort-load-bytes
4491 @kindex -mshort-load-bytes
4492 This is a depreciated alias for @samp{-malignment-traps}.
4494 @item -mno-short-load-bytes
4495 @kindex -mno-short-load-bytes
4496 This is a depreciated alias for @samp{-mno-alignment-traps}.
4498 @item -mshort-load-words
4499 @kindex -mshort-load-words
4500 This is a depreciated alias for @samp{-mno-alignment-traps}.
4502 @item -mno-short-load-words
4503 @kindex -mno-short-load-words
4504 This is a depreciated alias for @samp{-malignment-traps}.
4508 This option only applies to RISC iX. Emulate the native BSD-mode
4509 compiler. This is the default if @samp{-ansi} is not specified.
4513 This option only applies to RISC iX. Emulate the native X/Open-mode
4516 @item -mno-symrename
4517 @kindex -mno-symrename
4518 This option only applies to RISC iX. Do not run the assembler
4519 post-processor, @samp{symrename}, after code has been assembled.
4520 Normally it is necessary to modify some of the standard symbols in
4521 preparation for linking with the RISC iX C library; this option
4522 suppresses this pass. The post-processor is never run when the
4523 compiler is built for cross-compilation.
4527 This specifies the name of the target ARM processor. GCC uses this name
4528 to determine what kind of instructions it can use when generating
4529 assembly code. Permissible names are: arm2, arm250, arm3, arm6, arm60,
4530 arm600, arm610, arm620, arm7, arm7m, arm7d, arm7dm, arm7di, arm7dmi,
4531 arm70, arm700, arm700i, arm710, arm710c, arm7100, arm7500, arm7500fe,
4532 arm7tdmi, arm8, strongarm, strongarm110, strongarm1100, arm8, arm810,
4533 arm9, arm920, arm920t, arm9tdmi.
4535 @itemx -mtune=<name>
4537 This option is very similar to the @samp{-mcpu=} option, except that
4538 instead of specifying the actual target processor type, and hence
4539 restricting which instructions can be used, it specifies that GCC should
4540 tune the performance of the code as if the target were of the type
4541 specified in this option, but still choosing the instructions that it
4542 will generate based on the cpu specified by a @samp{-mcpu=} option.
4543 For some arm implementations better performance can be obtained by using
4548 This specifies the name of the target ARM architecture. GCC uses this
4549 name to determine what kind of instructions it can use when generating
4550 assembly code. This option can be used in conjunction with or instead
4551 of the @samp{-mcpu=} option. Permissible names are: armv2, armv2a,
4552 armv3, armv3m, armv4, armv4t, armv5.
4554 @item -mfpe=<number>
4555 @itemx -mfp=<number>
4558 This specifes the version of the floating point emulation available on
4559 the target. Permissible values are 2 and 3. @samp{-mfp=} is a synonym
4560 for @samp{-mfpe=} to support older versions of GCC.
4562 @item -mstructure-size-boundary=<n>
4563 @kindex -mstructure-size-boundary
4564 The size of all structures and unions will be rounded up to a multiple
4565 of the number of bits set by this option. Permissible values are 8 and
4566 32. The default value varies for different toolchains. For the COFF
4567 targeted toolchain the default value is 8. Specifying the larger number
4568 can produce faster, more efficient code, but can also increase the size
4569 of the program. The two values are potentially incompatible. Code
4570 compiled with one value cannot necessarily expect to work with code or
4571 libraries compiled with the other value, if they exchange information
4572 using structures or unions. Programmers are encouraged to use the 32
4573 value as future versions of the toolchain may default to this value.
4575 @item -mabort-on-noreturn
4576 @kindex -mabort-on-noreturn
4577 @kindex -mnoabort-on-noreturn
4578 Generate a call to the function abort at the end of a noreturn function.
4579 It will be executed if the function tries to return.
4581 @item -mnop-fun-dllimport
4582 @kindex -mnop-fun-dllimport
4583 Disable the support for the @emph{dllimport} attribute.
4585 @item -msingle-pic-base
4586 @kindex -msingle-pic-base
4587 Treat the register used for PIC addressing as read-only, rather than
4588 loading it in the prologue for each function. The run-time system is
4589 responsible for initialising this register with an appropriate value
4590 before execution begins.
4592 @item -mpic-register=<reg>
4593 @kindex -mpic-register=
4594 Specify the register to be used for PIC addressing. The default is R10
4595 unless stack-checking is enabled, when R9 is used.
4600 @subsection Thumb Options
4601 @cindex Thumb Options
4605 @item -mthumb-interwork
4606 @kindex -mthumb-interwork
4607 @kindex -mno-thumb-interwork
4608 Generate code which supports calling between the THUMB and ARM
4609 instruction sets. Without this option the two instruction sets cannot
4610 be reliably used inside one program. The default is
4611 @samp{-mno-thumb-interwork}, since slightly smaller code is generated
4615 @kindex -mtpcs-frame
4616 @kindex -mno-tpcs-frame
4617 Generate a stack frame that is compliant with the Thumb Procedure Call
4618 Standard for all non-leaf functions. (A leaf function is one that does
4619 not call any other functions). The default is @samp{-mno-apcs-frame}.
4621 @item -mtpcs-leaf-frame
4622 @kindex -mtpcs-leaf-frame
4623 @kindex -mno-tpcs-leaf-frame
4624 Generate a stack frame that is compliant with the Thumb Procedure Call
4625 Standard for all leaf functions. (A leaf function is one that does
4626 not call any other functions). The default is @samp{-mno-apcs-leaf-frame}.
4628 @item -mlittle-endian
4629 @kindex -mlittle-endian
4630 Generate code for a processor running in little-endian mode. This is
4631 the default for all standard configurations.
4634 @kindex -mbig-endian
4635 Generate code for a processor running in big-endian mode.
4637 @item -mstructure-size-boundary=<n>
4638 @kindex -mstructure-size-boundary
4639 The size of all structures and unions will be rounded up to a multiple
4640 of the number of bits set by this option. Permissible values are 8 and
4641 32. The default value varies for different toolchains. For the COFF
4642 targeted toolchain the default value is 8. Specifying the larger number
4643 can produced faster, more efficient code, but can also increase the size
4644 of the program. The two values are potentially incompatible. Code
4645 compiled with one value cannot necessarily expect to work with code or
4646 libraries compiled with the other value, if they exchange information
4647 using structures or unions. Programmers are encouraged to use the 32
4648 value as future versions of the toolchain may default to this value.
4650 @item -mnop-fun-dllimport
4651 @kindex -mnop-fun-dllimport
4652 Disable the support for the @emph{dllimport} attribute.
4654 @item -mcallee-super-interworking
4655 @kindex -mcallee-super-interworking
4656 Gives all externally visible functions in the file being compiled an ARM
4657 instruction set header which switches to Thumb mode before executing the
4658 rest of the function. This allows these functions to be called from
4659 non-interworking code.
4661 @item -mcaller-super-interworking
4662 @kindex -mcaller-super-interworking
4663 Allows calls via function pointers (including virtual functions) to
4664 execute correctly regardless of whether the target code has been
4665 compiled for interworking or not. There is a small overhead in the cost
4666 of executing a function pointer if this option is enabled.
4668 @item -msingle-pic-base
4669 @kindex -msingle-pic-base
4670 Treat the register used for PIC addressing as read-only, rather than
4671 loading it in the prologue for each function. The run-time system is
4672 responsible for initialising this register with an appropriate value
4673 before execution begins.
4675 @item -mpic-register=<reg>
4676 @kindex -mpic-register=
4677 Specify the register to be used for PIC addressing. The default is R10.
4681 @node MN10200 Options
4682 @subsection MN10200 Options
4683 @cindex MN10200 options
4684 These @samp{-m} options are defined for Matsushita MN10200 architectures:
4688 Indicate to the linker that it should perform a relaxation optimization pass
4689 to shorten branches, calls and absolute memory addresses. This option only
4690 has an effect when used on the command line for the final link step.
4692 This option makes symbolic debugging impossible.
4695 @node MN10300 Options
4696 @subsection MN10300 Options
4697 @cindex MN10300 options
4698 These @samp{-m} options are defined for Matsushita MN10300 architectures:
4702 Generate code to avoid bugs in the multiply instructions for the MN10300
4703 processors. This is the default.
4706 Do not generate code to avoid bugs in the multiply instructions for the
4710 Generate code which uses features specific to the AM33 processor.
4713 Do not generate code which uses features specific to the AM33 processor. This
4717 Indicate to the linker that it should perform a relaxation optimization pass
4718 to shorten branches, calls and absolute memory addresses. This option only
4719 has an effect when used on the command line for the final link step.
4721 This option makes symbolic debugging impossible.
4725 @node M32R/D Options
4726 @subsection M32R/D Options
4727 @cindex M32R/D options
4729 These @samp{-m} options are defined for Mitsubishi M32R/D architectures:
4732 @item -mcode-model=small
4733 Assume all objects live in the lower 16MB of memory (so that their addresses
4734 can be loaded with the @code{ld24} instruction), and assume all subroutines
4735 are reachable with the @code{bl} instruction.
4736 This is the default.
4738 The addressability of a particular object can be set with the
4739 @code{model} attribute.
4741 @item -mcode-model=medium
4742 Assume objects may be anywhere in the 32 bit address space (the compiler
4743 will generate @code{seth/add3} instructions to load their addresses), and
4744 assume all subroutines are reachable with the @code{bl} instruction.
4746 @item -mcode-model=large
4747 Assume objects may be anywhere in the 32 bit address space (the compiler
4748 will generate @code{seth/add3} instructions to load their addresses), and
4749 assume subroutines may not be reachable with the @code{bl} instruction
4750 (the compiler will generate the much slower @code{seth/add3/jl}
4751 instruction sequence).
4754 Disable use of the small data area. Variables will be put into
4755 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
4756 @code{section} attribute has been specified).
4757 This is the default.
4759 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
4760 Objects may be explicitly put in the small data area with the
4761 @code{section} attribute using one of these sections.
4764 Put small global and static data in the small data area, but do not
4765 generate special code to reference them.
4768 Put small global and static data in the small data area, and generate
4769 special instructions to reference them.
4772 @cindex smaller data references
4773 Put global and static objects less than or equal to @var{num} bytes
4774 into the small data or bss sections instead of the normal data or bss
4775 sections. The default value of @var{num} is 8.
4776 The @samp{-msdata} option must be set to one of @samp{sdata} or @samp{use}
4777 for this option to have any effect.
4779 All modules should be compiled with the same @samp{-G @var{num}} value.
4780 Compiling with different values of @var{num} may or may not work; if it
4781 doesn't the linker will give an error message - incorrect code will not be
4787 @subsection M88K Options
4788 @cindex M88k options
4790 These @samp{-m} options are defined for Motorola 88k architectures:
4795 Generate code that works well on both the m88100 and the
4800 Generate code that works best for the m88100, but that also
4805 Generate code that works best for the m88110, and may not run
4810 Obsolete option to be removed from the next revision.
4813 @item -midentify-revision
4814 @kindex -midentify-revision
4816 @cindex identifying source, compiler (88k)
4817 Include an @code{ident} directive in the assembler output recording the
4818 source file name, compiler name and version, timestamp, and compilation
4821 @item -mno-underscores
4822 @kindex -mno-underscores
4823 @cindex underscores, avoiding (88k)
4824 In assembler output, emit symbol names without adding an underscore
4825 character at the beginning of each name. The default is to use an
4826 underscore as prefix on each name.
4828 @item -mocs-debug-info
4829 @itemx -mno-ocs-debug-info
4830 @kindex -mocs-debug-info
4831 @kindex -mno-ocs-debug-info
4833 @cindex debugging, 88k OCS
4834 Include (or omit) additional debugging information (about registers used
4835 in each stack frame) as specified in the 88open Object Compatibility
4836 Standard, ``OCS''. This extra information allows debugging of code that
4837 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
4838 Delta 88 SVr3.2 is to include this information; other 88k configurations
4839 omit this information by default.
4841 @item -mocs-frame-position
4842 @kindex -mocs-frame-position
4843 @cindex register positions in frame (88k)
4844 When emitting COFF debugging information for automatic variables and
4845 parameters stored on the stack, use the offset from the canonical frame
4846 address, which is the stack pointer (register 31) on entry to the
4847 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
4848 @samp{-mocs-frame-position}; other 88k configurations have the default
4849 @samp{-mno-ocs-frame-position}.
4851 @item -mno-ocs-frame-position
4852 @kindex -mno-ocs-frame-position
4853 @cindex register positions in frame (88k)
4854 When emitting COFF debugging information for automatic variables and
4855 parameters stored on the stack, use the offset from the frame pointer
4856 register (register 30). When this option is in effect, the frame
4857 pointer is not eliminated when debugging information is selected by the
4860 @item -moptimize-arg-area
4861 @itemx -mno-optimize-arg-area
4862 @kindex -moptimize-arg-area
4863 @kindex -mno-optimize-arg-area
4864 @cindex arguments in frame (88k)
4865 Control how function arguments are stored in stack frames.
4866 @samp{-moptimize-arg-area} saves space by optimizing them, but this
4867 conflicts with the 88open specifications. The opposite alternative,
4868 @samp{-mno-optimize-arg-area}, agrees with 88open standards. By default
4869 GCC does not optimize the argument area.
4871 @item -mshort-data-@var{num}
4872 @kindex -mshort-data-@var{num}
4873 @cindex smaller data references (88k)
4874 @cindex r0-relative references (88k)
4875 Generate smaller data references by making them relative to @code{r0},
4876 which allows loading a value using a single instruction (rather than the
4877 usual two). You control which data references are affected by
4878 specifying @var{num} with this option. For example, if you specify
4879 @samp{-mshort-data-512}, then the data references affected are those
4880 involving displacements of less than 512 bytes.
4881 @samp{-mshort-data-@var{num}} is not effective for @var{num} greater
4884 @item -mserialize-volatile
4885 @kindex -mserialize-volatile
4886 @itemx -mno-serialize-volatile
4887 @kindex -mno-serialize-volatile
4888 @cindex sequential consistency on 88k
4889 Do, or don't, generate code to guarantee sequential consistency
4890 of volatile memory references. By default, consistency is
4893 The order of memory references made by the MC88110 processor does
4894 not always match the order of the instructions requesting those
4895 references. In particular, a load instruction may execute before
4896 a preceding store instruction. Such reordering violates
4897 sequential consistency of volatile memory references, when there
4898 are multiple processors. When consistency must be guaranteed,
4899 GNU C generates special instructions, as needed, to force
4900 execution in the proper order.
4902 The MC88100 processor does not reorder memory references and so
4903 always provides sequential consistency. However, by default, GNU
4904 C generates the special instructions to guarantee consistency
4905 even when you use @samp{-m88100}, so that the code may be run on an
4906 MC88110 processor. If you intend to run your code only on the
4907 MC88100 processor, you may use @samp{-mno-serialize-volatile}.
4909 The extra code generated to guarantee consistency may affect the
4910 performance of your application. If you know that you can safely
4911 forgo this guarantee, you may use @samp{-mno-serialize-volatile}.
4917 @cindex assembler syntax, 88k
4919 Turn on (@samp{-msvr4}) or off (@samp{-msvr3}) compiler extensions
4920 related to System V release 4 (SVr4). This controls the following:
4924 Which variant of the assembler syntax to emit.
4926 @samp{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
4927 that is used on System V release 4.
4929 @samp{-msvr4} makes GCC issue additional declaration directives used in
4933 @samp{-msvr4} is the default for the m88k-motorola-sysv4 and
4934 m88k-dg-dgux m88k configurations. @samp{-msvr3} is the default for all
4935 other m88k configurations.
4937 @item -mversion-03.00
4938 @kindex -mversion-03.00
4939 This option is obsolete, and is ignored.
4940 @c ??? which asm syntax better for GAS? option there too?
4942 @item -mno-check-zero-division
4943 @itemx -mcheck-zero-division
4944 @kindex -mno-check-zero-division
4945 @kindex -mcheck-zero-division
4946 @cindex zero division on 88k
4947 Do, or don't, generate code to guarantee that integer division by
4948 zero will be detected. By default, detection is guaranteed.
4950 Some models of the MC88100 processor fail to trap upon integer
4951 division by zero under certain conditions. By default, when
4952 compiling code that might be run on such a processor, GNU C
4953 generates code that explicitly checks for zero-valued divisors
4954 and traps with exception number 503 when one is detected. Use of
4955 mno-check-zero-division suppresses such checking for code
4956 generated to run on an MC88100 processor.
4958 GNU C assumes that the MC88110 processor correctly detects all
4959 instances of integer division by zero. When @samp{-m88110} is
4960 specified, both @samp{-mcheck-zero-division} and
4961 @samp{-mno-check-zero-division} are ignored, and no explicit checks for
4962 zero-valued divisors are generated.
4964 @item -muse-div-instruction
4965 @kindex -muse-div-instruction
4966 @cindex divide instruction, 88k
4967 Use the div instruction for signed integer division on the
4968 MC88100 processor. By default, the div instruction is not used.
4970 On the MC88100 processor the signed integer division instruction
4971 div) traps to the operating system on a negative operand. The
4972 operating system transparently completes the operation, but at a
4973 large cost in execution time. By default, when compiling code
4974 that might be run on an MC88100 processor, GNU C emulates signed
4975 integer division using the unsigned integer division instruction
4976 divu), thereby avoiding the large penalty of a trap to the
4977 operating system. Such emulation has its own, smaller, execution
4978 cost in both time and space. To the extent that your code's
4979 important signed integer division operations are performed on two
4980 nonnegative operands, it may be desirable to use the div
4981 instruction directly.
4983 On the MC88110 processor the div instruction (also known as the
4984 divs instruction) processes negative operands without trapping to
4985 the operating system. When @samp{-m88110} is specified,
4986 @samp{-muse-div-instruction} is ignored, and the div instruction is used
4987 for signed integer division.
4989 Note that the result of dividing INT_MIN by -1 is undefined. In
4990 particular, the behavior of such a division with and without
4991 @samp{-muse-div-instruction} may differ.
4993 @item -mtrap-large-shift
4994 @itemx -mhandle-large-shift
4995 @kindex -mtrap-large-shift
4996 @kindex -mhandle-large-shift
4997 @cindex bit shift overflow (88k)
4998 @cindex large bit shifts (88k)
4999 Include code to detect bit-shifts of more than 31 bits; respectively,
5000 trap such shifts or emit code to handle them properly. By default GCC
5001 makes no special provision for large bit shifts.
5003 @item -mwarn-passed-structs
5004 @kindex -mwarn-passed-structs
5005 @cindex structure passing (88k)
5006 Warn when a function passes a struct as an argument or result.
5007 Structure-passing conventions have changed during the evolution of the C
5008 language, and are often the source of portability problems. By default,
5009 GCC issues no such warning.
5012 @node RS/6000 and PowerPC Options
5013 @subsection IBM RS/6000 and PowerPC Options
5014 @cindex RS/6000 and PowerPC Options
5015 @cindex IBM RS/6000 and PowerPC Options
5017 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
5025 @itemx -mpowerpc-gpopt
5026 @itemx -mno-powerpc-gpopt
5027 @itemx -mpowerpc-gfxopt
5028 @itemx -mno-powerpc-gfxopt
5030 @itemx -mno-powerpc64
5034 @kindex -mpowerpc-gpopt
5035 @kindex -mpowerpc-gfxopt
5037 GCC supports two related instruction set architectures for the
5038 RS/6000 and PowerPC. The @dfn{POWER} instruction set are those
5039 instructions supported by the @samp{rios} chip set used in the original
5040 RS/6000 systems and the @dfn{PowerPC} instruction set is the
5041 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
5042 the IBM 4xx microprocessors.
5044 Neither architecture is a subset of the other. However there is a
5045 large common subset of instructions supported by both. An MQ
5046 register is included in processors supporting the POWER architecture.
5048 You use these options to specify which instructions are available on the
5049 processor you are using. The default value of these options is
5050 determined when configuring GCC. Specifying the
5051 @samp{-mcpu=@var{cpu_type}} overrides the specification of these
5052 options. We recommend you use the @samp{-mcpu=@var{cpu_type}} option
5053 rather than the options listed above.
5055 The @samp{-mpower} option allows GCC to generate instructions that
5056 are found only in the POWER architecture and to use the MQ register.
5057 Specifying @samp{-mpower2} implies @samp{-power} and also allows GCC
5058 to generate instructions that are present in the POWER2 architecture but
5059 not the original POWER architecture.
5061 The @samp{-mpowerpc} option allows GCC to generate instructions that
5062 are found only in the 32-bit subset of the PowerPC architecture.
5063 Specifying @samp{-mpowerpc-gpopt} implies @samp{-mpowerpc} and also allows
5064 GCC to use the optional PowerPC architecture instructions in the
5065 General Purpose group, including floating-point square root. Specifying
5066 @samp{-mpowerpc-gfxopt} implies @samp{-mpowerpc} and also allows GCC to
5067 use the optional PowerPC architecture instructions in the Graphics
5068 group, including floating-point select.
5070 The @samp{-mpowerpc64} option allows GCC to generate the additional
5071 64-bit instructions that are found in the full PowerPC64 architecture
5072 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
5073 @samp{-mno-powerpc64}.
5075 If you specify both @samp{-mno-power} and @samp{-mno-powerpc}, GCC
5076 will use only the instructions in the common subset of both
5077 architectures plus some special AIX common-mode calls, and will not use
5078 the MQ register. Specifying both @samp{-mpower} and @samp{-mpowerpc}
5079 permits GCC to use any instruction from either architecture and to
5080 allow use of the MQ register; specify this for the Motorola MPC601.
5082 @item -mnew-mnemonics
5083 @itemx -mold-mnemonics
5084 @kindex -mnew-mnemonics
5085 @kindex -mold-mnemonics
5086 Select which mnemonics to use in the generated assembler code.
5087 @samp{-mnew-mnemonics} requests output that uses the assembler mnemonics
5088 defined for the PowerPC architecture, while @samp{-mold-mnemonics}
5089 requests the assembler mnemonics defined for the POWER architecture.
5090 Instructions defined in only one architecture have only one mnemonic;
5091 GCC uses that mnemonic irrespective of which of these options is
5094 GCC defaults to the mnemonics appropriate for the architecture in
5095 use. Specifying @samp{-mcpu=@var{cpu_type}} sometimes overrides the
5096 value of these option. Unless you are building a cross-compiler, you
5097 should normally not specify either @samp{-mnew-mnemonics} or
5098 @samp{-mold-mnemonics}, but should instead accept the default.
5100 @item -mcpu=@var{cpu_type}
5102 Set architecture type, register usage, choice of mnemonics, and
5103 instruction scheduling parameters for machine type @var{cpu_type}.
5104 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
5105 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
5106 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
5107 @samp{630}, @samp{740}, @samp{750}, @samp{power}, @samp{power2},
5108 @samp{powerpc}, @samp{403}, @samp{505}, @samp{801}, @samp{821},
5109 @samp{823}, and @samp{860} and @samp{common}. @samp{-mcpu=power},
5110 @samp{-mcpu=power2}, @samp{-mcpu=powerpc}, and @samp{-mcpu=powerpc64}
5111 specify generic POWER, POWER2, pure 32-bit PowerPC (i.e., not MPC601),
5112 and 64-bit PowerPC architecture machine types, with an appropriate,
5113 generic processor model assumed for scheduling purposes.@refill
5115 Specifying any of the following options:
5116 @samp{-mcpu=rios1}, @samp{-mcpu=rios2}, @samp{-mcpu=rsc},
5117 @samp{-mcpu=power}, or @samp{-mcpu=power2}
5118 enables the @samp{-mpower} option and disables the @samp{-mpowerpc} option;
5119 @samp{-mcpu=601} enables both the @samp{-mpower} and @samp{-mpowerpc} options.
5120 All of @samp{-mcpu=rs64a}, @samp{-mcpu=602}, @samp{-mcpu=603},
5121 @samp{-mcpu=603e}, @samp{-mcpu=604}, @samp{-mcpu=620}, @samp{-mcpu=630},
5122 @samp{-mcpu=740}, and @samp{-mcpu=750}
5123 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5124 Exactly similarly, all of @samp{-mcpu=403},
5125 @samp{-mcpu=505}, @samp{-mcpu=821}, @samp{-mcpu=860} and @samp{-mcpu=powerpc}
5126 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5127 @samp{-mcpu=common} disables both the
5128 @samp{-mpower} and @samp{-mpowerpc} options.@refill
5130 AIX versions 4 or greater selects @samp{-mcpu=common} by default, so
5131 that code will operate on all members of the RS/6000 POWER and PowerPC
5132 families. In that case, GCC will use only the instructions in the
5133 common subset of both architectures plus some special AIX common-mode
5134 calls, and will not use the MQ register. GCC assumes a generic
5135 processor model for scheduling purposes.
5137 Specifying any of the options @samp{-mcpu=rios1}, @samp{-mcpu=rios2},
5138 @samp{-mcpu=rsc}, @samp{-mcpu=power}, or @samp{-mcpu=power2} also
5139 disables the @samp{new-mnemonics} option. Specifying @samp{-mcpu=601},
5140 @samp{-mcpu=602}, @samp{-mcpu=603}, @samp{-mcpu=603e}, @samp{-mcpu=604},
5141 @samp{-mcpu=620}, @samp{-mcpu=630}, @samp{-mcpu=403}, @samp{-mcpu=505},
5142 @samp{-mcpu=821}, @samp{-mcpu=860} or @samp{-mcpu=powerpc} also enables
5143 the @samp{new-mnemonics} option.@refill
5145 Specifying @samp{-mcpu=403}, @samp{-mcpu=821}, or @samp{-mcpu=860} also
5146 enables the @samp{-msoft-float} option.
5148 @item -mtune=@var{cpu_type}
5149 Set the instruction scheduling parameters for machine type
5150 @var{cpu_type}, but do not set the architecture type, register usage,
5151 choice of mnemonics like @samp{-mcpu=}@var{cpu_type} would. The same
5152 values for @var{cpu_type} are used for @samp{-mtune=}@var{cpu_type} as
5153 for @samp{-mcpu=}@var{cpu_type}. The @samp{-mtune=}@var{cpu_type}
5154 option overrides the @samp{-mcpu=}@var{cpu_type} option in terms of
5155 instruction scheduling parameters.
5158 @itemx -mno-fp-in-toc
5159 @itemx -mno-sum-in-toc
5160 @itemx -mminimal-toc
5161 @kindex -mminimal-toc
5162 Modify generation of the TOC (Table Of Contents), which is created for
5163 every executable file. The @samp{-mfull-toc} option is selected by
5164 default. In that case, GCC will allocate at least one TOC entry for
5165 each unique non-automatic variable reference in your program. GCC
5166 will also place floating-point constants in the TOC. However, only
5167 16,384 entries are available in the TOC.
5169 If you receive a linker error message that saying you have overflowed
5170 the available TOC space, you can reduce the amount of TOC space used
5171 with the @samp{-mno-fp-in-toc} and @samp{-mno-sum-in-toc} options.
5172 @samp{-mno-fp-in-toc} prevents GCC from putting floating-point
5173 constants in the TOC and @samp{-mno-sum-in-toc} forces GCC to
5174 generate code to calculate the sum of an address and a constant at
5175 run-time instead of putting that sum into the TOC. You may specify one
5176 or both of these options. Each causes GCC to produce very slightly
5177 slower and larger code at the expense of conserving TOC space.
5179 If you still run out of space in the TOC even when you specify both of
5180 these options, specify @samp{-mminimal-toc} instead. This option causes
5181 GCC to make only one TOC entry for every file. When you specify this
5182 option, GCC will produce code that is slower and larger but which
5183 uses extremely little TOC space. You may wish to use this option
5184 only on files that contain less frequently executed code. @refill
5190 Enable 64-bit PowerPC ABI and calling convention: 64-bit pointers, 64-bit
5191 @code{long} type, and the infrastructure needed to support them.
5192 Specifying @samp{-m64} implies @samp{-mpowerpc64} and
5193 @samp{-mpowerpc}, while @samp{-m32} disables the 64-bit ABI and
5194 implies @samp{-mno-powerpc64}. GCC defaults to @samp{-m32}.
5199 On AIX, pass floating-point arguments to prototyped functions beyond the
5200 register save area (RSA) on the stack in addition to argument FPRs. The
5201 AIX calling convention was extended but not initially documented to
5202 handle an obscure K&R C case of calling a function that takes the
5203 address of its arguments with fewer arguments than declared. AIX XL
5204 compilers access floating point arguments which do not fit in the
5205 RSA from the stack when a subroutine is compiled without
5206 optimization. Because always storing floating-point arguments on the
5207 stack is inefficient and rarely needed, this option is not enabled by
5208 default and only is necessary when calling subroutines compiled by AIX
5209 XL compilers without optimization.
5213 Support @dfn{AIX Threads}. Link an application written to use
5214 @dfn{pthreads} with special libraries and startup code to enable the
5219 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE). Link an
5220 application written to use message passing with special startup code to
5221 enable the application to run. The system must have PE installed in the
5222 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
5223 must be overridden with the @samp{-specs=} option to specify the
5224 appropriate directory location. The Parallel Environment does not
5225 support threads, so the @samp{-mpe} option and the @samp{-mthreads}
5226 option are incompatible.
5230 @kindex -msoft-float
5231 Generate code that does not use (uses) the floating-point register set.
5232 Software floating point emulation is provided if you use the
5233 @samp{-msoft-float} option, and pass the option to GCC when linking.
5236 @itemx -mno-multiple
5237 Generate code that uses (does not use) the load multiple word
5238 instructions and the store multiple word instructions. These
5239 instructions are generated by default on POWER systems, and not
5240 generated on PowerPC systems. Do not use @samp{-mmultiple} on little
5241 endian PowerPC systems, since those instructions do not work when the
5242 processor is in little endian mode. The exceptions are PPC740 and
5243 PPC750 which permit the instructions usage in little endian mode.
5248 Generate code that uses (does not use) the load string instructions
5249 and the store string word instructions to save multiple registers and
5250 do small block moves. These instructions are generated by default on
5251 POWER systems, and not generated on PowerPC systems. Do not use
5252 @samp{-mstring} on little endian PowerPC systems, since those
5253 instructions do not work when the processor is in little endian mode.
5254 The exceptions are PPC740 and PPC750 which permit the instructions
5255 usage in little endian mode.
5260 Generate code that uses (does not use) the load or store instructions
5261 that update the base register to the address of the calculated memory
5262 location. These instructions are generated by default. If you use
5263 @samp{-mno-update}, there is a small window between the time that the
5264 stack pointer is updated and the address of the previous frame is
5265 stored, which means code that walks the stack frame across interrupts or
5266 signals may get corrupted data.
5269 @itemx -mno-fused-madd
5270 @kindex -mfused-madd
5271 Generate code that uses (does not use) the floating point multiply and
5272 accumulate instructions. These instructions are generated by default if
5273 hardware floating is used.
5275 @item -mno-bit-align
5278 On System V.4 and embedded PowerPC systems do not (do) force structures
5279 and unions that contain bit fields to be aligned to the base type of the
5282 For example, by default a structure containing nothing but 8
5283 @code{unsigned} bitfields of length 1 would be aligned to a 4 byte
5284 boundary and have a size of 4 bytes. By using @samp{-mno-bit-align},
5285 the structure would be aligned to a 1 byte boundary and be one byte in
5288 @item -mno-strict-align
5289 @itemx -mstrict-align
5290 @kindex -mstrict-align
5291 On System V.4 and embedded PowerPC systems do not (do) assume that
5292 unaligned memory references will be handled by the system.
5295 @itemx -mno-relocatable
5296 @kindex -mrelocatable
5297 On embedded PowerPC systems generate code that allows (does not allow)
5298 the program to be relocated to a different address at runtime. If you
5299 use @samp{-mrelocatable} on any module, all objects linked together must
5300 be compiled with @samp{-mrelocatable} or @samp{-mrelocatable-lib}.
5302 @item -mrelocatable-lib
5303 @itemx -mno-relocatable-lib
5304 On embedded PowerPC systems generate code that allows (does not allow)
5305 the program to be relocated to a different address at runtime. Modules
5306 compiled with @samp{-mrelocatable-lib} can be linked with either modules
5307 compiled without @samp{-mrelocatable} and @samp{-mrelocatable-lib} or
5308 with modules compiled with the @samp{-mrelocatable} options.
5312 On System V.4 and embedded PowerPC systems do not (do) assume that
5313 register 2 contains a pointer to a global area pointing to the addresses
5314 used in the program.
5317 @itemx -mlittle-endian
5318 On System V.4 and embedded PowerPC systems compile code for the
5319 processor in little endian mode. The @samp{-mlittle-endian} option is
5320 the same as @samp{-mlittle}.
5324 On System V.4 and embedded PowerPC systems compile code for the
5325 processor in big endian mode. The @samp{-mbig-endian} option is
5326 the same as @samp{-mbig}.
5329 On System V.4 and embedded PowerPC systems compile code using calling
5330 conventions that adheres to the March 1995 draft of the System V
5331 Application Binary Interface, PowerPC processor supplement. This is the
5332 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
5334 @item -mcall-sysv-eabi
5335 Specify both @samp{-mcall-sysv} and @samp{-meabi} options.
5337 @item -mcall-sysv-noeabi
5338 Specify both @samp{-mcall-sysv} and @samp{-mno-eabi} options.
5341 On System V.4 and embedded PowerPC systems compile code using calling
5342 conventions that are similar to those used on AIX. This is the
5343 default if you configured GCC using @samp{powerpc-*-eabiaix}.
5345 @item -mcall-solaris
5346 On System V.4 and embedded PowerPC systems compile code for the Solaris
5350 On System V.4 and embedded PowerPC systems compile code for the
5351 Linux-based GNU system.
5354 @itemx -mno-prototype
5355 On System V.4 and embedded PowerPC systems assume that all calls to
5356 variable argument functions are properly prototyped. Otherwise, the
5357 compiler must insert an instruction before every non prototyped call to
5358 set or clear bit 6 of the condition code register (@var{CR}) to
5359 indicate whether floating point values were passed in the floating point
5360 registers in case the function takes a variable arguments. With
5361 @samp{-mprototype}, only calls to prototyped variable argument functions
5362 will set or clear the bit.
5365 On embedded PowerPC systems, assume that the startup module is called
5366 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
5367 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
5371 On embedded PowerPC systems, assume that the startup module is called
5372 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
5376 On embedded PowerPC systems, assume that the startup module is called
5377 @file{crt0.o} and the standard C libraries are @file{libads.a} and
5381 On embedded PowerPC systems, assume that the startup module is called
5382 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
5386 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
5387 header to indicate that @samp{eabi} extended relocations are used.
5391 On System V.4 and embedded PowerPC systems do (do not) adhere to the
5392 Embedded Applications Binary Interface (eabi) which is a set of
5393 modifications to the System V.4 specifications. Selecting @code{-meabi}
5394 means that the stack is aligned to an 8 byte boundary, a function
5395 @code{__eabi} is called to from @code{main} to set up the eabi
5396 environment, and the @samp{-msdata} option can use both @code{r2} and
5397 @code{r13} to point to two separate small data areas. Selecting
5398 @code{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
5399 do not call an initialization function from @code{main}, and the
5400 @samp{-msdata} option will only use @code{r13} to point to a single
5401 small data area. The @samp{-meabi} option is on by default if you
5402 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
5405 On System V.4 and embedded PowerPC systems, put small initialized
5406 @code{const} global and static data in the @samp{.sdata2} section, which
5407 is pointed to by register @code{r2}. Put small initialized
5408 non-@code{const} global and static data in the @samp{.sdata} section,
5409 which is pointed to by register @code{r13}. Put small uninitialized
5410 global and static data in the @samp{.sbss} section, which is adjacent to
5411 the @samp{.sdata} section. The @samp{-msdata=eabi} option is
5412 incompatible with the @samp{-mrelocatable} option. The
5413 @samp{-msdata=eabi} option also sets the @samp{-memb} option.
5416 On System V.4 and embedded PowerPC systems, put small global and static
5417 data in the @samp{.sdata} section, which is pointed to by register
5418 @code{r13}. Put small uninitialized global and static data in the
5419 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
5420 The @samp{-msdata=sysv} option is incompatible with the
5421 @samp{-mrelocatable} option.
5423 @item -msdata=default
5425 On System V.4 and embedded PowerPC systems, if @samp{-meabi} is used,
5426 compile code the same as @samp{-msdata=eabi}, otherwise compile code the
5427 same as @samp{-msdata=sysv}.
5430 On System V.4 and embedded PowerPC systems, put small global and static
5431 data in the @samp{.sdata} section. Put small uninitialized global and
5432 static data in the @samp{.sbss} section. Do not use register @code{r13}
5433 to address small data however. This is the default behavior unless
5434 other @samp{-msdata} options are used.
5438 On embedded PowerPC systems, put all initialized global and static data
5439 in the @samp{.data} section, and all uninitialized data in the
5440 @samp{.bss} section.
5443 @cindex smaller data references (PowerPC)
5444 @cindex .sdata/.sdata2 references (PowerPC)
5445 On embedded PowerPC systems, put global and static items less than or
5446 equal to @var{num} bytes into the small data or bss sections instead of
5447 the normal data or bss section. By default, @var{num} is 8. The
5448 @samp{-G @var{num}} switch is also passed to the linker.
5449 All modules should be compiled with the same @samp{-G @var{num}} value.
5452 @itemx -mno-regnames
5453 On System V.4 and embedded PowerPC systems do (do not) emit register
5454 names in the assembly language output using symbolic forms.
5459 @subsection IBM RT Options
5461 @cindex IBM RT options
5463 These @samp{-m} options are defined for the IBM RT PC:
5467 Use an in-line code sequence for integer multiplies. This is the
5470 @item -mcall-lib-mul
5471 Call @code{lmul$$} for integer multiples.
5473 @item -mfull-fp-blocks
5474 Generate full-size floating point data blocks, including the minimum
5475 amount of scratch space recommended by IBM. This is the default.
5477 @item -mminimum-fp-blocks
5478 Do not include extra scratch space in floating point data blocks. This
5479 results in smaller code, but slower execution, since scratch space must
5480 be allocated dynamically.
5482 @cindex @file{varargs.h} and RT PC
5483 @cindex @file{stdarg.h} and RT PC
5484 @item -mfp-arg-in-fpregs
5485 Use a calling sequence incompatible with the IBM calling convention in
5486 which floating point arguments are passed in floating point registers.
5487 Note that @code{varargs.h} and @code{stdargs.h} will not work with
5488 floating point operands if this option is specified.
5490 @item -mfp-arg-in-gregs
5491 Use the normal calling convention for floating point arguments. This is
5494 @item -mhc-struct-return
5495 Return structures of more than one word in memory, rather than in a
5496 register. This provides compatibility with the MetaWare HighC (hc)
5497 compiler. Use the option @samp{-fpcc-struct-return} for compatibility
5498 with the Portable C Compiler (pcc).
5500 @item -mnohc-struct-return
5501 Return some structures of more than one word in registers, when
5502 convenient. This is the default. For compatibility with the
5503 IBM-supplied compilers, use the option @samp{-fpcc-struct-return} or the
5504 option @samp{-mhc-struct-return}.
5508 @subsection MIPS Options
5509 @cindex MIPS options
5511 These @samp{-m} options are defined for the MIPS family of computers:
5514 @item -mcpu=@var{cpu type}
5515 Assume the defaults for the machine type @var{cpu type} when scheduling
5516 instructions. The choices for @var{cpu type} are @samp{r2000}, @samp{r3000},
5517 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
5518 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
5519 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
5520 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
5521 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
5522 @var{cpu type} will schedule things appropriately for that particular
5523 chip, the compiler will not generate any code that does not meet level 1
5524 of the MIPS ISA (instruction set architecture) without a @samp{-mipsX}
5525 or @samp{-mabi} switch being used.
5528 Issue instructions from level 1 of the MIPS ISA. This is the default.
5529 @samp{r3000} is the default @var{cpu type} at this ISA level.
5532 Issue instructions from level 2 of the MIPS ISA (branch likely, square
5533 root instructions). @samp{r6000} is the default @var{cpu type} at this
5537 Issue instructions from level 3 of the MIPS ISA (64 bit instructions).
5538 @samp{r4000} is the default @var{cpu type} at this ISA level.
5541 Issue instructions from level 4 of the MIPS ISA (conditional move,
5542 prefetch, enhanced FPU instructions). @samp{r8000} is the default
5543 @var{cpu type} at this ISA level.
5546 Assume that 32 32-bit floating point registers are available. This is
5550 Assume that 32 64-bit floating point registers are available. This is
5551 the default when the @samp{-mips3} option is used.
5554 Assume that 32 32-bit general purpose registers are available. This is
5558 Assume that 32 64-bit general purpose registers are available. This is
5559 the default when the @samp{-mips3} option is used.
5562 Force int and long types to be 64 bits wide. See @samp{-mlong32} for an
5563 explanation of the default, and the width of pointers.
5566 Force long types to be 64 bits wide. See @samp{-mlong32} for an
5567 explanation of the default, and the width of pointers.
5570 Force long, int, and pointer types to be 32 bits wide.
5572 If none of @samp{-mlong32}, @samp{-mlong64}, or @samp{-mint64} are set,
5573 the size of ints, longs, and pointers depends on the ABI and ISA choosen.
5574 For @samp{-mabi=32}, and @samp{-mabi=n32}, ints and longs are 32 bits
5575 wide. For @samp{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
5576 For @samp{-mabi=eabi} and either @samp{-mips1} or @samp{-mips2}, ints
5577 and longs are 32 bits wide. For @samp{-mabi=eabi} and higher ISAs, ints
5578 are 32 bits, and longs are 64 bits wide. The width of pointer types is
5579 the smaller of the width of longs or the width of general purpose
5580 registers (which in turn depends on the ISA).
5587 Generate code for the indicated ABI. The default instruction level is
5588 @samp{-mips1} for @samp{32}, @samp{-mips3} for @samp{n32}, and
5589 @samp{-mips4} otherwise. Conversely, with @samp{-mips1} or
5590 @samp{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
5594 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
5595 add normal debug information. This is the default for all
5596 platforms except for the OSF/1 reference platform, using the OSF/rose
5597 object format. If the either of the @samp{-gstabs} or @samp{-gstabs+}
5598 switches are used, the @file{mips-tfile} program will encapsulate the
5599 stabs within MIPS ECOFF.
5602 Generate code for the GNU assembler. This is the default on the OSF/1
5603 reference platform, using the OSF/rose object format. Also, this is
5604 the default if the configure option @samp{--with-gnu-as} is used.
5606 @item -msplit-addresses
5607 @itemx -mno-split-addresses
5608 Generate code to load the high and low parts of address constants separately.
5609 This allows @code{gcc} to optimize away redundant loads of the high order
5610 bits of addresses. This optimization requires GNU as and GNU ld.
5611 This optimization is enabled by default for some embedded targets where
5612 GNU as and GNU ld are standard.
5616 The @samp{-mrnames} switch says to output code using the MIPS software
5617 names for the registers, instead of the hardware names (ie, @var{a0}
5618 instead of @var{$4}). The only known assembler that supports this option
5619 is the Algorithmics assembler.
5623 The @samp{-mgpopt} switch says to write all of the data declarations
5624 before the instructions in the text section, this allows the MIPS
5625 assembler to generate one word memory references instead of using two
5626 words for short global or static data items. This is on by default if
5627 optimization is selected.
5631 For each non-inline function processed, the @samp{-mstats} switch
5632 causes the compiler to emit one line to the standard error file to
5633 print statistics about the program (number of registers saved, stack
5638 The @samp{-mmemcpy} switch makes all block moves call the appropriate
5639 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
5640 generating inline code.
5643 @itemx -mno-mips-tfile
5644 The @samp{-mno-mips-tfile} switch causes the compiler not
5645 postprocess the object file with the @file{mips-tfile} program,
5646 after the MIPS assembler has generated it to add debug support. If
5647 @file{mips-tfile} is not run, then no local variables will be
5648 available to the debugger. In addition, @file{stage2} and
5649 @file{stage3} objects will have the temporary file names passed to the
5650 assembler embedded in the object file, which means the objects will
5651 not compare the same. The @samp{-mno-mips-tfile} switch should only
5652 be used when there are bugs in the @file{mips-tfile} program that
5653 prevents compilation.
5656 Generate output containing library calls for floating point.
5657 @strong{Warning:} the requisite libraries are not part of GCC.
5658 Normally the facilities of the machine's usual C compiler are used, but
5659 this can't be done directly in cross-compilation. You must make your
5660 own arrangements to provide suitable library functions for
5664 Generate output containing floating point instructions. This is the
5665 default if you use the unmodified sources.
5668 @itemx -mno-abicalls
5669 Emit (or do not emit) the pseudo operations @samp{.abicalls},
5670 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
5671 position independent code.
5674 @itemx -mno-long-calls
5675 Do all calls with the @samp{JALR} instruction, which requires
5676 loading up a function's address into a register before the call.
5677 You need to use this switch, if you call outside of the current
5678 512 megabyte segment to functions that are not through pointers.
5681 @itemx -mno-half-pic
5682 Put pointers to extern references into the data section and load them
5683 up, rather than put the references in the text section.
5685 @item -membedded-pic
5686 @itemx -mno-embedded-pic
5687 Generate PIC code suitable for some embedded systems. All calls are
5688 made using PC relative address, and all data is addressed using the $gp
5689 register. No more than 65536 bytes of global data may be used. This
5690 requires GNU as and GNU ld which do most of the work. This currently
5691 only works on targets which use ECOFF; it does not work with ELF.
5693 @item -membedded-data
5694 @itemx -mno-embedded-data
5695 Allocate variables to the read-only data section first if possible, then
5696 next in the small data section if possible, otherwise in data. This gives
5697 slightly slower code than the default, but reduces the amount of RAM required
5698 when executing, and thus may be preferred for some embedded systems.
5700 @item -muninit-const-in-rodata
5701 @itemx -mno-uninit-const-in-rodata
5702 When used together with -membedded-data, it will always store uninitialized
5703 const variables in the read-only data section.
5705 @item -msingle-float
5706 @itemx -mdouble-float
5707 The @samp{-msingle-float} switch tells gcc to assume that the floating
5708 point coprocessor only supports single precision operations, as on the
5709 @samp{r4650} chip. The @samp{-mdouble-float} switch permits gcc to use
5710 double precision operations. This is the default.
5714 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
5715 as on the @samp{r4650} chip.
5718 Turns on @samp{-msingle-float}, @samp{-mmad}, and, at least for now,
5723 Enable 16-bit instructions.
5726 Use the entry and exit pseudo ops. This option can only be used with
5730 Compile code for the processor in little endian mode.
5731 The requisite libraries are assumed to exist.
5734 Compile code for the processor in big endian mode.
5735 The requisite libraries are assumed to exist.
5738 @cindex smaller data references (MIPS)
5739 @cindex gp-relative references (MIPS)
5740 Put global and static items less than or equal to @var{num} bytes into
5741 the small data or bss sections instead of the normal data or bss
5742 section. This allows the assembler to emit one word memory reference
5743 instructions based on the global pointer (@var{gp} or @var{$28}),
5744 instead of the normal two words used. By default, @var{num} is 8 when
5745 the MIPS assembler is used, and 0 when the GNU assembler is used. The
5746 @samp{-G @var{num}} switch is also passed to the assembler and linker.
5747 All modules should be compiled with the same @samp{-G @var{num}}
5751 Tell the MIPS assembler to not run its preprocessor over user
5752 assembler files (with a @samp{.s} suffix) when assembling them.
5755 Do not include the default crt0.
5759 These options are defined by the macro
5760 @code{TARGET_SWITCHES} in the machine description. The default for the
5761 options is also defined by that macro, which enables you to change the
5766 @subsection Intel 386 Options
5767 @cindex i386 Options
5768 @cindex Intel 386 Options
5770 These @samp{-m} options are defined for the i386 family of computers:
5773 @item -mcpu=@var{cpu type}
5774 Assume the defaults for the machine type @var{cpu type} when scheduling
5775 instructions. The choices for @var{cpu type} are:
5777 @multitable @columnfractions .20 .20 .20 .20
5778 @item @samp{i386} @tab @samp{i486} @tab @samp{i586} @tab @samp{i686}
5779 @item @samp{pentium} @tab @samp{pentiumpro} @tab @samp{k6}
5782 While picking a specific @var{cpu type} will schedule things appropriately
5783 for that particular chip, the compiler will not generate any code that
5784 does not run on the i386 without the @samp{-march=@var{cpu type}} option
5785 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
5786 is equivalent to @samp{pentiumpro}. @samp{k6} is the AMD chip as
5787 opposed to the Intel ones.
5789 @item -march=@var{cpu type}
5790 Generate instructions for the machine type @var{cpu type}. The choices
5791 for @var{cpu type} are the same as for @samp{-mcpu}. Moreover,
5792 specifying @samp{-march=@var{cpu type}} implies @samp{-mcpu=@var{cpu type}}.
5798 Synonyms for -mcpu=i386, -mcpu=i486, -mcpu=pentium, and -mcpu=pentiumpro
5799 respectively. These synonyms are deprecated.
5803 Control whether or not the compiler uses IEEE floating point
5804 comparisons. These handle correctly the case where the result of a
5805 comparison is unordered.
5808 Generate output containing library calls for floating point.
5809 @strong{Warning:} the requisite libraries are not part of GCC.
5810 Normally the facilities of the machine's usual C compiler are used, but
5811 this can't be done directly in cross-compilation. You must make your
5812 own arrangements to provide suitable library functions for
5815 On machines where a function returns floating point results in the 80387
5816 register stack, some floating point opcodes may be emitted even if
5817 @samp{-msoft-float} is used.
5819 @item -mno-fp-ret-in-387
5820 Do not use the FPU registers for return values of functions.
5822 The usual calling convention has functions return values of types
5823 @code{float} and @code{double} in an FPU register, even if there
5824 is no FPU. The idea is that the operating system should emulate
5827 The option @samp{-mno-fp-ret-in-387} causes such values to be returned
5828 in ordinary CPU registers instead.
5830 @item -mno-fancy-math-387
5831 Some 387 emulators do not support the @code{sin}, @code{cos} and
5832 @code{sqrt} instructions for the 387. Specify this option to avoid
5833 generating those instructions. This option is the default on FreeBSD.
5834 As of revision 2.6.1, these instructions are not generated unless you
5835 also use the @samp{-ffast-math} switch.
5837 @item -malign-double
5838 @itemx -mno-align-double
5839 Control whether GCC aligns @code{double}, @code{long double}, and
5840 @code{long long} variables on a two word boundary or a one word
5841 boundary. Aligning @code{double} variables on a two word boundary will
5842 produce code that runs somewhat faster on a @samp{Pentium} at the
5843 expense of more memory.
5845 @strong{Warning:} if you use the @samp{-malign-double} switch,
5846 structures containing the above types will be aligned differently than
5847 the published application binary interface specifications for the 386.
5850 @itemx -mno-svr3-shlib
5851 Control whether GCC places uninitialized locals into @code{bss} or
5852 @code{data}. @samp{-msvr3-shlib} places these locals into @code{bss}.
5853 These options are meaningful only on System V Release 3.
5855 @item -mno-wide-multiply
5856 @itemx -mwide-multiply
5857 Control whether GCC uses the @code{mul} and @code{imul} that produce
5858 64 bit results in @code{eax:edx} from 32 bit operands to do @code{long
5859 long} multiplies and 32-bit division by constants.
5862 Use a different function-calling convention, in which functions that
5863 take a fixed number of arguments return with the @code{ret} @var{num}
5864 instruction, which pops their arguments while returning. This saves one
5865 instruction in the caller since there is no need to pop the arguments
5868 You can specify that an individual function is called with this calling
5869 sequence with the function attribute @samp{stdcall}. You can also
5870 override the @samp{-mrtd} option by using the function attribute
5871 @samp{cdecl}. @xref{Function Attributes}.
5873 @strong{Warning:} this calling convention is incompatible with the one
5874 normally used on Unix, so you cannot use it if you need to call
5875 libraries compiled with the Unix compiler.
5877 Also, you must provide function prototypes for all functions that
5878 take variable numbers of arguments (including @code{printf});
5879 otherwise incorrect code will be generated for calls to those
5882 In addition, seriously incorrect code will result if you call a
5883 function with too many arguments. (Normally, extra arguments are
5884 harmlessly ignored.)
5886 @item -mreg-alloc=@var{regs}
5887 Control the default allocation order of integer registers. The
5888 string @var{regs} is a series of letters specifying a register. The
5889 supported letters are: @code{a} allocate EAX; @code{b} allocate EBX;
5890 @code{c} allocate ECX; @code{d} allocate EDX; @code{S} allocate ESI;
5891 @code{D} allocate EDI; @code{B} allocate EBP.
5893 @item -mregparm=@var{num}
5894 Control how many registers are used to pass integer arguments. By
5895 default, no registers are used to pass arguments, and at most 3
5896 registers can be used. You can control this behavior for a specific
5897 function by using the function attribute @samp{regparm}.
5898 @xref{Function Attributes}.
5900 @strong{Warning:} if you use this switch, and
5901 @var{num} is nonzero, then you must build all modules with the same
5902 value, including any libraries. This includes the system libraries and
5905 @item -malign-loops=@var{num}
5906 Align loops to a 2 raised to a @var{num} byte boundary. If
5907 @samp{-malign-loops} is not specified, the default is 2 unless
5908 gas 2.8 (or later) is being used in which case the default is
5909 to align the loop on a 16 byte boundary if it is less than 8
5912 @item -malign-jumps=@var{num}
5913 Align instructions that are only jumped to to a 2 raised to a @var{num}
5914 byte boundary. If @samp{-malign-jumps} is not specified, the default is
5915 2 if optimizing for a 386, and 4 if optimizing for a 486 unless
5916 gas 2.8 (or later) is being used in which case the default is
5917 to align the instruction on a 16 byte boundary if it is less
5920 @item -malign-functions=@var{num}
5921 Align the start of functions to a 2 raised to @var{num} byte boundary.
5922 If @samp{-malign-functions} is not specified, the default is 2 if optimizing
5923 for a 386, and 4 if optimizing for a 486.
5925 @item -mpreferred-stack-boundary=@var{num}
5926 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
5927 byte boundary. If @samp{-mpreferred-stack-boundary} is not specified,
5928 the default is 4 (16 bytes or 128 bits).
5930 The stack is required to be aligned on a 4 byte boundary. On Pentium
5931 and PentiumPro, @code{double} and @code{long double} values should be
5932 aligned to an 8 byte boundary (see @samp{-malign-double}) or suffer
5933 significant run time performance penalties. On Pentium III, the
5934 Streaming SIMD Extention (SSE) data type @code{__m128} suffers similar
5935 penalties if it is not 16 byte aligned.
5937 To ensure proper alignment of this values on the stack, the stack boundary
5938 must be as aligned as that required by any value stored on the stack.
5939 Further, every function must be generated such that it keeps the stack
5940 aligned. Thus calling a function compiled with a higher preferred
5941 stack boundary from a function compiled with a lower preferred stack
5942 boundary will most likely misalign the stack. It is recommended that
5943 libraries that use callbacks always use the default setting.
5945 This extra alignment does consume extra stack space. Code that is sensitive
5946 to stack space usage, such as embedded systems and operating system kernels,
5947 may want to reduce the preferred alignment to
5948 @samp{-mpreferred-stack-boundary=2}.
5952 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
5953 on thread-safe exception handling must compile and link all code with the
5954 @samp{-mthreads} option. When compiling, @samp{-mthreads} defines
5955 @samp{-D_MT}; when linking, it links in a special thread helper library
5956 @samp{-lmingwthrd} which cleans up per thread exception handling data.
5958 @item -mno-align-stringops
5959 @kindex -mno-align-stringops
5960 Do not align destination of inlined string operations. This switch reduces
5961 code size and improves performance in case the destination is already aligned,
5962 but gcc don't know about it.
5964 @item -minline-all-stringops
5965 @kindex -minline-all-stringops
5966 By default GCC inlines string operations only when destination is known to be
5967 aligned at least to 4 byte boundary. This enables more inlining, increase code
5968 size, but may improve performance of code that depends on fast memcpy, strlen
5969 and memset for short lengths.
5973 @subsection HPPA Options
5974 @cindex HPPA Options
5976 These @samp{-m} options are defined for the HPPA family of computers:
5979 @item -march=@var{architecture type}
5980 Generate code for the specified architecture. The choices for
5981 @var{architecture type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
5982 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
5983 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
5984 architecture option for your machine. Code compiled for lower numbered
5985 architectures will run on higher numbered architectures, but not the
5988 PA 2.0 support currently requires gas snapshot 19990413 or later. The
5989 next release of binutils (current is 2.9.1) will probably contain PA 2.0
5993 @itemx -mpa-risc-1-1
5994 @itemx -mpa-risc-2-0
5995 Synonyms for -march=1.0, -march=1.1, and -march=2.0 respectively.
5998 Generate code suitable for big switch tables. Use this option only if
5999 the assembler/linker complain about out of range branches within a switch
6002 @item -mjump-in-delay
6003 Fill delay slots of function calls with unconditional jump instructions
6004 by modifying the return pointer for the function call to be the target
6005 of the conditional jump.
6007 @item -mdisable-fpregs
6008 Prevent floating point registers from being used in any manner. This is
6009 necessary for compiling kernels which perform lazy context switching of
6010 floating point registers. If you use this option and attempt to perform
6011 floating point operations, the compiler will abort.
6013 @item -mdisable-indexing
6014 Prevent the compiler from using indexing address modes. This avoids some
6015 rather obscure problems when compiling MIG generated code under MACH.
6017 @item -mno-space-regs
6018 Generate code that assumes the target has no space registers. This allows
6019 GCC to generate faster indirect calls and use unscaled index address modes.
6021 Such code is suitable for level 0 PA systems and kernels.
6023 @item -mfast-indirect-calls
6024 Generate code that assumes calls never cross space boundaries. This
6025 allows GCC to emit code which performs faster indirect calls.
6027 This option will not work in the presense of shared libraries or nested
6030 @item -mlong-load-store
6031 Generate 3-instruction load and store sequences as sometimes required by
6032 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
6035 @item -mportable-runtime
6036 Use the portable calling conventions proposed by HP for ELF systems.
6039 Enable the use of assembler directives only GAS understands.
6041 @item -mschedule=@var{cpu type}
6042 Schedule code according to the constraints for the machine type
6043 @var{cpu type}. The choices for @var{cpu type} are @samp{700}
6044 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
6045 @file{/usr/lib/sched.models} on an HP-UX system to determine the
6046 proper scheduling option for your machine.
6049 Enable the optimization pass in the HPUX linker. Note this makes symbolic
6050 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
6051 in which they give bogus error messages when linking some programs.
6054 Generate output containing library calls for floating point.
6055 @strong{Warning:} the requisite libraries are not available for all HPPA
6056 targets. Normally the facilities of the machine's usual C compiler are
6057 used, but this cannot be done directly in cross-compilation. You must make
6058 your own arrangements to provide suitable library functions for
6059 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
6060 does provide software floating point support.
6062 @samp{-msoft-float} changes the calling convention in the output file;
6063 therefore, it is only useful if you compile @emph{all} of a program with
6064 this option. In particular, you need to compile @file{libgcc.a}, the
6065 library that comes with GCC, with @samp{-msoft-float} in order for
6069 @node Intel 960 Options
6070 @subsection Intel 960 Options
6072 These @samp{-m} options are defined for the Intel 960 implementations:
6075 @item -m@var{cpu type}
6076 Assume the defaults for the machine type @var{cpu type} for some of
6077 the other options, including instruction scheduling, floating point
6078 support, and addressing modes. The choices for @var{cpu type} are
6079 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
6080 @samp{sa}, and @samp{sb}.
6086 The @samp{-mnumerics} option indicates that the processor does support
6087 floating-point instructions. The @samp{-msoft-float} option indicates
6088 that floating-point support should not be assumed.
6090 @item -mleaf-procedures
6091 @itemx -mno-leaf-procedures
6092 Do (or do not) attempt to alter leaf procedures to be callable with the
6093 @code{bal} instruction as well as @code{call}. This will result in more
6094 efficient code for explicit calls when the @code{bal} instruction can be
6095 substituted by the assembler or linker, but less efficient code in other
6096 cases, such as calls via function pointers, or using a linker that doesn't
6097 support this optimization.
6100 @itemx -mno-tail-call
6101 Do (or do not) make additional attempts (beyond those of the
6102 machine-independent portions of the compiler) to optimize tail-recursive
6103 calls into branches. You may not want to do this because the detection of
6104 cases where this is not valid is not totally complete. The default is
6105 @samp{-mno-tail-call}.
6107 @item -mcomplex-addr
6108 @itemx -mno-complex-addr
6109 Assume (or do not assume) that the use of a complex addressing mode is a
6110 win on this implementation of the i960. Complex addressing modes may not
6111 be worthwhile on the K-series, but they definitely are on the C-series.
6112 The default is currently @samp{-mcomplex-addr} for all processors except
6116 @itemx -mno-code-align
6117 Align code to 8-byte boundaries for faster fetching (or don't bother).
6118 Currently turned on by default for C-series implementations only.
6121 @item -mclean-linkage
6122 @itemx -mno-clean-linkage
6123 These options are not fully implemented.
6127 @itemx -mic2.0-compat
6128 @itemx -mic3.0-compat
6129 Enable compatibility with iC960 v2.0 or v3.0.
6133 Enable compatibility with the iC960 assembler.
6135 @item -mstrict-align
6136 @itemx -mno-strict-align
6137 Do not permit (do permit) unaligned accesses.
6140 Enable structure-alignment compatibility with Intel's gcc release version
6141 1.3 (based on gcc 1.37). This option implies @samp{-mstrict-align}.
6143 @item -mlong-double-64
6144 Implement type @samp{long double} as 64-bit floating point numbers.
6145 Without the option @samp{long double} is implemented by 80-bit
6146 floating point numbers. The only reason we have it because there is
6147 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
6148 is only useful for people using soft-float targets. Otherwise, we
6149 should recommend against use of it.
6153 @node DEC Alpha Options
6154 @subsection DEC Alpha Options
6156 These @samp{-m} options are defined for the DEC Alpha implementations:
6159 @item -mno-soft-float
6161 Use (do not use) the hardware floating-point instructions for
6162 floating-point operations. When @code{-msoft-float} is specified,
6163 functions in @file{libgcc1.c} will be used to perform floating-point
6164 operations. Unless they are replaced by routines that emulate the
6165 floating-point operations, or compiled in such a way as to call such
6166 emulations routines, these routines will issue floating-point
6167 operations. If you are compiling for an Alpha without floating-point
6168 operations, you must ensure that the library is built so as not to call
6171 Note that Alpha implementations without floating-point operations are
6172 required to have floating-point registers.
6176 Generate code that uses (does not use) the floating-point register set.
6177 @code{-mno-fp-regs} implies @code{-msoft-float}. If the floating-point
6178 register set is not used, floating point operands are passed in integer
6179 registers as if they were integers and floating-point results are passed
6180 in $0 instead of $f0. This is a non-standard calling sequence, so any
6181 function with a floating-point argument or return value called by code
6182 compiled with @code{-mno-fp-regs} must also be compiled with that
6185 A typical use of this option is building a kernel that does not use,
6186 and hence need not save and restore, any floating-point registers.
6189 The Alpha architecture implements floating-point hardware optimized for
6190 maximum performance. It is mostly compliant with the IEEE floating
6191 point standard. However, for full compliance, software assistance is
6192 required. This option generates code fully IEEE compliant code
6193 @emph{except} that the @var{inexact flag} is not maintained (see below).
6194 If this option is turned on, the CPP macro @code{_IEEE_FP} is defined
6195 during compilation. The option is a shorthand for: @samp{-D_IEEE_FP
6196 -mfp-trap-mode=su -mtrap-precision=i -mieee-conformant}. The resulting
6197 code is less efficient but is able to correctly support denormalized
6198 numbers and exceptional IEEE values such as not-a-number and plus/minus
6199 infinity. Other Alpha compilers call this option
6200 @code{-ieee_with_no_inexact}.
6202 @item -mieee-with-inexact
6203 @c overfull hbox here --bob 22 jul96
6204 @c original text between ignore ... end ignore
6206 This is like @samp{-mieee} except the generated code also maintains the
6207 IEEE @var{inexact flag}. Turning on this option causes the generated
6208 code to implement fully-compliant IEEE math. The option is a shorthand
6209 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus @samp{-mieee-conformant},
6210 @samp{-mfp-trap-mode=sui}, and @samp{-mtrap-precision=i}. On some Alpha
6211 implementations the resulting code may execute significantly slower than
6212 the code generated by default. Since there is very little code that
6213 depends on the @var{inexact flag}, you should normally not specify this
6214 option. Other Alpha compilers call this option
6215 @samp{-ieee_with_inexact}.
6217 @c changed paragraph
6218 This is like @samp{-mieee} except the generated code also maintains the
6219 IEEE @var{inexact flag}. Turning on this option causes the generated
6220 code to implement fully-compliant IEEE math. The option is a shorthand
6221 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus the three following:
6222 @samp{-mieee-conformant},
6223 @samp{-mfp-trap-mode=sui},
6224 and @samp{-mtrap-precision=i}.
6225 On some Alpha implementations the resulting code may execute
6226 significantly slower than the code generated by default. Since there
6227 is very little code that depends on the @var{inexact flag}, you should
6228 normally not specify this option. Other Alpha compilers call this
6229 option @samp{-ieee_with_inexact}.
6230 @c end changes to prevent overfull hboxes
6232 @item -mfp-trap-mode=@var{trap mode}
6233 This option controls what floating-point related traps are enabled.
6234 Other Alpha compilers call this option @samp{-fptm }@var{trap mode}.
6235 The trap mode can be set to one of four values:
6239 This is the default (normal) setting. The only traps that are enabled
6240 are the ones that cannot be disabled in software (e.g., division by zero
6244 In addition to the traps enabled by @samp{n}, underflow traps are enabled
6248 Like @samp{su}, but the instructions are marked to be safe for software
6249 completion (see Alpha architecture manual for details).
6252 Like @samp{su}, but inexact traps are enabled as well.
6255 @item -mfp-rounding-mode=@var{rounding mode}
6256 Selects the IEEE rounding mode. Other Alpha compilers call this option
6257 @samp{-fprm }@var{rounding mode}. The @var{rounding mode} can be one
6262 Normal IEEE rounding mode. Floating point numbers are rounded towards
6263 the nearest machine number or towards the even machine number in case
6267 Round towards minus infinity.
6270 Chopped rounding mode. Floating point numbers are rounded towards zero.
6273 Dynamic rounding mode. A field in the floating point control register
6274 (@var{fpcr}, see Alpha architecture reference manual) controls the
6275 rounding mode in effect. The C library initializes this register for
6276 rounding towards plus infinity. Thus, unless your program modifies the
6277 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
6280 @item -mtrap-precision=@var{trap precision}
6281 In the Alpha architecture, floating point traps are imprecise. This
6282 means without software assistance it is impossible to recover from a
6283 floating trap and program execution normally needs to be terminated.
6284 GCC can generate code that can assist operating system trap handlers
6285 in determining the exact location that caused a floating point trap.
6286 Depending on the requirements of an application, different levels of
6287 precisions can be selected:
6291 Program precision. This option is the default and means a trap handler
6292 can only identify which program caused a floating point exception.
6295 Function precision. The trap handler can determine the function that
6296 caused a floating point exception.
6299 Instruction precision. The trap handler can determine the exact
6300 instruction that caused a floating point exception.
6303 Other Alpha compilers provide the equivalent options called
6304 @samp{-scope_safe} and @samp{-resumption_safe}.
6306 @item -mieee-conformant
6307 This option marks the generated code as IEEE conformant. You must not
6308 use this option unless you also specify @samp{-mtrap-precision=i} and either
6309 @samp{-mfp-trap-mode=su} or @samp{-mfp-trap-mode=sui}. Its only effect
6310 is to emit the line @samp{.eflag 48} in the function prologue of the
6311 generated assembly file. Under DEC Unix, this has the effect that
6312 IEEE-conformant math library routines will be linked in.
6314 @item -mbuild-constants
6315 Normally GCC examines a 32- or 64-bit integer constant to
6316 see if it can construct it from smaller constants in two or three
6317 instructions. If it cannot, it will output the constant as a literal and
6318 generate code to load it from the data segment at runtime.
6320 Use this option to require GCC to construct @emph{all} integer constants
6321 using code, even if it takes more instructions (the maximum is six).
6323 You would typically use this option to build a shared library dynamic
6324 loader. Itself a shared library, it must relocate itself in memory
6325 before it can find the variables and constants in its own data segment.
6329 Select whether to generate code to be assembled by the vendor-supplied
6330 assembler (@samp{-malpha-as}) or by the GNU assembler @samp{-mgas}.
6338 Indicate whether GCC should generate code to use the optional BWX,
6339 CIX, and MAX instruction sets. The default is to use the instruction sets
6340 supported by the CPU type specified via @samp{-mcpu=} option or that
6341 of the CPU on which GCC was built if none was specified.
6343 @item -mcpu=@var{cpu_type}
6344 Set the instruction set, register set, and instruction scheduling
6345 parameters for machine type @var{cpu_type}. You can specify either the
6346 @samp{EV} style name or the corresponding chip number. GCC
6347 supports scheduling parameters for the EV4 and EV5 family of processors
6348 and will choose the default values for the instruction set from
6349 the processor you specify. If you do not specify a processor type,
6350 GCC will default to the processor on which the compiler was built.
6352 Supported values for @var{cpu_type} are
6357 Schedules as an EV4 and has no instruction set extensions.
6361 Schedules as an EV5 and has no instruction set extensions.
6365 Schedules as an EV5 and supports the BWX extension.
6370 Schedules as an EV5 and supports the BWX and MAX extensions.
6374 Schedules as an EV5 (until Digital releases the scheduling parameters
6375 for the EV6) and supports the BWX, CIX, and MAX extensions.
6378 @item -mmemory-latency=@var{time}
6379 Sets the latency the scheduler should assume for typical memory
6380 references as seen by the application. This number is highly
6381 dependant on the memory access patterns used by the application
6382 and the size of the external cache on the machine.
6384 Valid options for @var{time} are
6388 A decimal number representing clock cycles.
6394 The compiler contains estimates of the number of clock cycles for
6395 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
6396 (also called Dcache, Scache, and Bcache), as well as to main memory.
6397 Note that L3 is only valid for EV5.
6402 @node Clipper Options
6403 @subsection Clipper Options
6405 These @samp{-m} options are defined for the Clipper implementations:
6409 Produce code for a C300 Clipper processor. This is the default.
6412 Produce code for a C400 Clipper processor i.e. use floating point
6416 @node H8/300 Options
6417 @subsection H8/300 Options
6419 These @samp{-m} options are defined for the H8/300 implementations:
6423 Shorten some address references at link time, when possible; uses the
6424 linker option @samp{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
6425 ld.info, Using ld}, for a fuller description.
6428 Generate code for the H8/300H.
6431 Generate code for the H8/S.
6434 Make @code{int} data 32 bits by default.
6437 On the h8/300h, use the same alignment rules as for the h8/300.
6438 The default for the h8/300h is to align longs and floats on 4 byte boundaries.
6439 @samp{-malign-300} causes them to be aligned on 2 byte boundaries.
6440 This option has no effect on the h8/300.
6444 @subsection SH Options
6446 These @samp{-m} options are defined for the SH implementations:
6450 Generate code for the SH1.
6453 Generate code for the SH2.
6456 Generate code for the SH3.
6459 Generate code for the SH3e.
6462 Compile code for the processor in big endian mode.
6465 Compile code for the processor in little endian mode.
6468 Align doubles at 64 bit boundaries. Note that this changes the calling
6469 conventions, and thus some functions from the standard C library will
6470 not work unless you recompile it first with -mdalign.
6473 Shorten some address references at link time, when possible; uses the
6474 linker option @samp{-relax}.
6477 @node System V Options
6478 @subsection Options for System V
6480 These additional options are available on System V Release 4 for
6481 compatibility with other compilers on those systems:
6485 Create a shared object.
6486 It is recommended that @samp{-symbolic} or @samp{-shared} be used instead.
6489 Identify the versions of each tool used by the compiler, in a
6490 @code{.ident} assembler directive in the output.
6493 Refrain from adding @code{.ident} directives to the output file (this is
6496 @item -YP,@var{dirs}
6497 Search the directories @var{dirs}, and no others, for libraries
6498 specified with @samp{-l}.
6501 Look in the directory @var{dir} to find the M4 preprocessor.
6502 The assembler uses this option.
6503 @c This is supposed to go with a -Yd for predefined M4 macro files, but
6504 @c the generic assembler that comes with Solaris takes just -Ym.
6507 @node TMS320C3x/C4x Options
6508 @subsection TMS320C3x/C4x Options
6509 @cindex TMS320C3x/C4x Options
6511 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
6515 @item -mcpu=@var{cpu_type}
6516 Set the instruction set, register set, and instruction scheduling
6517 parameters for machine type @var{cpu_type}. Supported values for
6518 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
6519 @samp{c44}. The default is @samp{c40} to generate code for the
6524 @itemx -msmall-memory
6526 Generates code for the big or small memory model. The small memory
6527 model assumed that all data fits into one 64K word page. At run-time
6528 the data page (DP) register must be set to point to the 64K page
6529 containing the .bss and .data program sections. The big memory model is
6530 the default and requires reloading of the DP register for every direct
6535 Allow (disallow) allocation of general integer operands into the block
6540 Enable (disable) generation of code using decrement and branch,
6541 DBcond(D), instructions. This is enabled by default for the C4x. To be
6542 on the safe side, this is disabled for the C3x, since the maximum
6543 iteration count on the C3x is 2^23 + 1 (but who iterates loops more than
6544 2^23 times on the C3x?). Note that GCC will try to reverse a loop so
6545 that it can utilise the decrement and branch instruction, but will give
6546 up if there is more than one memory reference in the loop. Thus a loop
6547 where the loop counter is decremented can generate slightly more
6548 efficient code, in cases where the RPTB instruction cannot be utilised.
6550 @item -mdp-isr-reload
6552 Force the DP register to be saved on entry to an interrupt service
6553 routine (ISR), reloaded to point to the data section, and restored on
6554 exit from the ISR. This should not be required unless someone has
6555 violated the small memory model by modifying the DP register, say within
6560 For the C3x use the 24-bit MPYI instruction for integer multiplies
6561 instead of a library call to guarantee 32-bit results. Note that if one
6562 of the operands is a constant, then the multiplication will be performed
6563 using shifts and adds. If the -mmpyi option is not specified for the C3x,
6564 then squaring operations are performed inline instead of a library call.
6567 @itemx -mno-fast-fix
6568 The C3x/C4x FIX instruction to convert a floating point value to an
6569 integer value chooses the nearest integer less than or equal to the
6570 floating point value rather than to the nearest integer. Thus if the
6571 floating point number is negative, the result will be incorrectly
6572 truncated an additional code is necessary to detect and correct this
6573 case. This option can be used to disable generation of the additional
6574 code required to correct the result.
6578 Enable (disable) generation of repeat block sequences using the RPTB
6579 instruction for zero overhead looping. The RPTB construct is only used
6580 for innermost loops that do not call functions or jump across the loop
6581 boundaries. There is no advantage having nested RPTB loops due to the
6582 overhead required to save and restore the RC, RS, and RE registers.
6583 This is enabled by default with -O2.
6585 @item -mrpts=@var{count}
6587 Enable (disable) the use of the single instruction repeat instruction
6588 RPTS. If a repeat block contains a single instruction, and the loop
6589 count can be guaranteed to be less than the value @var{count}, GCC will
6590 emit a RPTS instruction instead of a RPTB. If no value is specified,
6591 then a RPTS will be emitted even if the loop count cannot be determined
6592 at compile time. Note that the repeated instruction following RPTS does
6593 not have to be reloaded from memory each iteration, thus freeing up the
6594 CPU buses for oeprands. However, since interrupts are blocked by this
6595 instruction, it is disabled by default.
6597 @item -mloop-unsigned
6598 @itemx -mno-loop-unsigned
6599 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
6600 is 2^31 + 1 since these instructions test if the iteration count is
6601 negative to terminate the loop. If the iteration count is unsigned
6602 there is a possibility than the 2^31 + 1 maximum iteration count may be
6603 exceeded. This switch allows an unsigned iteration count.
6606 Try to emit an assembler syntax that the TI assembler (asm30) is happy
6607 with. This also enforces compatibility with the API employed by the TI
6608 C3x C compiler. For example, long doubles are passed as structures
6609 rather than in floating point registers.
6613 Generate code that uses registers (stack) for passing arguments to functions.
6614 By default, arguments are passed in registers where possible rather
6615 than by pushing arguments on to the stack.
6617 @item -mparallel-insns
6618 @itemx -mno-parallel-insns
6619 Allow the generation of parallel instructions. This is enabled by
6622 @item -mparallel-mpy
6623 @itemx -mno-parallel-mpy
6624 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
6625 provided -mparallel-insns is also specified. These instructions have
6626 tight register constraints which can pessimize the code generation
6632 @subsection V850 Options
6633 @cindex V850 Options
6635 These @samp{-m} options are defined for V850 implementations:
6639 @itemx -mno-long-calls
6640 Treat all calls as being far away (near). If calls are assumed to be
6641 far away, the compiler will always load the functions address up into a
6642 register, and call indirect through the pointer.
6646 Do not optimize (do optimize) basic blocks that use the same index
6647 pointer 4 or more times to copy pointer into the @code{ep} register, and
6648 use the shorter @code{sld} and @code{sst} instructions. The @samp{-mep}
6649 option is on by default if you optimize.
6651 @item -mno-prolog-function
6652 @itemx -mprolog-function
6653 Do not use (do use) external functions to save and restore registers at
6654 the prolog and epilog of a function. The external functions are slower,
6655 but use less code space if more than one function saves the same number
6656 of registers. The @samp{-mprolog-function} option is on by default if
6660 Try to make the code as small as possible. At present, this just turns
6661 on the @samp{-mep} and @samp{-mprolog-function} options.
6664 Put static or global variables whose size is @var{n} bytes or less into
6665 the tiny data area that register @code{ep} points to. The tiny data
6666 area can hold up to 256 bytes in total (128 bytes for byte references).
6669 Put static or global variables whose size is @var{n} bytes or less into
6670 the small data area that register @code{gp} points to. The small data
6671 area can hold up to 64 kilobytes.
6674 Put static or global variables whose size is @var{n} bytes or less into
6675 the first 32 kilobytes of memory.
6678 Specify that the target processor is the V850.
6681 Generate code suitable for big switch tables. Use this option only if
6682 the assembler/linker complain about out of range branches within a switch
6687 @subsection ARC Options
6690 These options are defined for ARC implementations:
6694 Compile code for little endian mode. This is the default.
6697 Compile code for big endian mode.
6700 Prepend the name of the cpu to all public symbol names.
6701 In multiple-processor systems, there are many ARC variants with different
6702 instruction and register set characteristics. This flag prevents code
6703 compiled for one cpu to be linked with code compiled for another.
6704 No facility exists for handling variants that are "almost identical".
6705 This is an all or nothing option.
6707 @item -mcpu=@var{cpu}
6708 Compile code for ARC variant @var{cpu}.
6709 Which variants are supported depend on the configuration.
6710 All variants support @samp{-mcpu=base}, this is the default.
6712 @item -mtext=@var{text section}
6713 @itemx -mdata=@var{data section}
6714 @itemx -mrodata=@var{readonly data section}
6715 Put functions, data, and readonly data in @var{text section},
6716 @var{data section}, and @var{readonly data section} respectively
6717 by default. This can be overridden with the @code{section} attribute.
6718 @xref{Variable Attributes}.
6723 @subsection NS32K Options
6724 @cindex NS32K options
6726 These are the @samp{-m} options defined for the 32000 series. The default
6727 values for these options depends on which style of 32000 was selected when
6728 the compiler was configured; the defaults for the most common choices are
6734 Generate output for a 32032. This is the default
6735 when the compiler is configured for 32032 and 32016 based systems.
6739 Generate output for a 32332. This is the default
6740 when the compiler is configured for 32332-based systems.
6744 Generate output for a 32532. This is the default
6745 when the compiler is configured for 32532-based systems.
6748 Generate output containing 32081 instructions for floating point.
6749 This is the default for all systems.
6752 Generate output containing 32381 instructions for floating point. This
6753 also implies @samp{-m32081}. The 32381 is only compatible with the 32332
6754 and 32532 cpus. This is the default for the pc532-netbsd configuration.
6757 Try and generate multiply-add floating point instructions @code{polyF}
6758 and @code{dotF}. This option is only available if the @samp{-m32381}
6759 option is in effect. Using these instructions requires changes to to
6760 register allocation which generally has a negative impact on
6761 performance. This option should only be enabled when compiling code
6762 particularly likely to make heavy use of multiply-add instructions.
6765 Do not try and generate multiply-add floating point instructions
6766 @code{polyF} and @code{dotF}. This is the default on all platforms.
6769 Generate output containing library calls for floating point.
6770 @strong{Warning:} the requisite libraries may not be available.
6773 Do not use the bit-field instructions. On some machines it is faster to
6774 use shifting and masking operations. This is the default for the pc532.
6777 Do use the bit-field instructions. This is the default for all platforms
6781 Use a different function-calling convention, in which functions
6782 that take a fixed number of arguments return pop their
6783 arguments on return with the @code{ret} instruction.
6785 This calling convention is incompatible with the one normally
6786 used on Unix, so you cannot use it if you need to call libraries
6787 compiled with the Unix compiler.
6789 Also, you must provide function prototypes for all functions that
6790 take variable numbers of arguments (including @code{printf});
6791 otherwise incorrect code will be generated for calls to those
6794 In addition, seriously incorrect code will result if you call a
6795 function with too many arguments. (Normally, extra arguments are
6796 harmlessly ignored.)
6798 This option takes its name from the 680x0 @code{rtd} instruction.
6802 Use a different function-calling convention where the first two arguments
6803 are passed in registers.
6805 This calling convention is incompatible with the one normally
6806 used on Unix, so you cannot use it if you need to call libraries
6807 compiled with the Unix compiler.
6810 Do not pass any arguments in registers. This is the default for all
6814 It is OK to use the sb as an index register which is always loaded with
6815 zero. This is the default for the pc532-netbsd target.
6818 The sb register is not available for use or has not been initialized to
6819 zero by the run time system. This is the default for all targets except
6820 the pc532-netbsd. It is also implied whenever @samp{-mhimem} or
6821 @samp{-fpic} is set.
6824 Many ns32000 series addressing modes use displacements of up to 512MB.
6825 If an address is above 512MB then displacements from zero can not be used.
6826 This option causes code to be generated which can be loaded above 512MB.
6827 This may be useful for operating systems or ROM code.
6830 Assume code will be loaded in the first 512MB of virtual address space.
6831 This is the default for all platforms.
6838 @node Code Gen Options
6839 @section Options for Code Generation Conventions
6840 @cindex code generation conventions
6841 @cindex options, code generation
6842 @cindex run-time options
6844 These machine-independent options control the interface conventions
6845 used in code generation.
6847 Most of them have both positive and negative forms; the negative form
6848 of @samp{-ffoo} would be @samp{-fno-foo}. In the table below, only
6849 one of the forms is listed---the one which is not the default. You
6850 can figure out the other form by either removing @samp{no-} or adding
6855 Enable exception handling. Generates extra code needed to propagate
6856 exceptions. For some targets, this implies GNU CC will generate frame
6857 unwind information for all functions, which can produce significant data
6858 size overhead, although it does not affect execution. If you do not
6859 specify this option, GNU CC will enable it by default for languages like
6860 C++ which normally require exception handling, and disable itfor
6861 languages like C that do not normally require it. However, you may need
6862 to enable this option when compiling C code that needs to interoperate
6863 properly with exception handlers written in C++. You may also wish to
6864 disable this option if you are compiling older C++ programs that don't
6865 use exception handling.
6867 @item -funwind-tables
6868 Similar to @code{-fexceptions}, except that it will just generate any needed
6869 static data, but will not affect the generated code in any other way.
6870 You will normally not enable this option; instead, a language processor
6871 that needs this handling would enable it on your behalf.
6873 @item -fpcc-struct-return
6874 Return ``short'' @code{struct} and @code{union} values in memory like
6875 longer ones, rather than in registers. This convention is less
6876 efficient, but it has the advantage of allowing intercallability between
6877 GCC-compiled files and files compiled with other compilers.
6879 The precise convention for returning structures in memory depends
6880 on the target configuration macros.
6882 Short structures and unions are those whose size and alignment match
6883 that of some integer type.
6885 @item -freg-struct-return
6886 Use the convention that @code{struct} and @code{union} values are
6887 returned in registers when possible. This is more efficient for small
6888 structures than @samp{-fpcc-struct-return}.
6890 If you specify neither @samp{-fpcc-struct-return} nor its contrary
6891 @samp{-freg-struct-return}, GCC defaults to whichever convention is
6892 standard for the target. If there is no standard convention, GCC
6893 defaults to @samp{-fpcc-struct-return}, except on targets where GCC
6894 is the principal compiler. In those cases, we can choose the standard,
6895 and we chose the more efficient register return alternative.
6898 Allocate to an @code{enum} type only as many bytes as it needs for the
6899 declared range of possible values. Specifically, the @code{enum} type
6900 will be equivalent to the smallest integer type which has enough room.
6902 @item -fshort-double
6903 Use the same size for @code{double} as for @code{float}.
6906 Requests that the data and non-@code{const} variables of this
6907 compilation be shared data rather than private data. The distinction
6908 makes sense only on certain operating systems, where shared data is
6909 shared between processes running the same program, while private data
6910 exists in one copy per process.
6913 Allocate even uninitialized global variables in the data section of the
6914 object file, rather than generating them as common blocks. This has the
6915 effect that if the same variable is declared (without @code{extern}) in
6916 two different compilations, you will get an error when you link them.
6917 The only reason this might be useful is if you wish to verify that the
6918 program will work on other systems which always work this way.
6921 Ignore the @samp{#ident} directive.
6923 @item -fno-gnu-linker
6924 Do not output global initializations (such as C++ constructors and
6925 destructors) in the form used by the GNU linker (on systems where the GNU
6926 linker is the standard method of handling them). Use this option when
6927 you want to use a non-GNU linker, which also requires using the
6928 @code{collect2} program to make sure the system linker includes
6929 constructors and destructors. (@code{collect2} is included in the GCC
6930 distribution.) For systems which @emph{must} use @code{collect2}, the
6931 compiler driver @code{gcc} is configured to do this automatically.
6933 @item -finhibit-size-directive
6934 Don't output a @code{.size} assembler directive, or anything else that
6935 would cause trouble if the function is split in the middle, and the
6936 two halves are placed at locations far apart in memory. This option is
6937 used when compiling @file{crtstuff.c}; you should not need to use it
6941 Put extra commentary information in the generated assembly code to
6942 make it more readable. This option is generally only of use to those
6943 who actually need to read the generated assembly code (perhaps while
6944 debugging the compiler itself).
6946 @samp{-fno-verbose-asm}, the default, causes the
6947 extra information to be omitted and is useful when comparing two assembler
6951 Consider all memory references through pointers to be volatile.
6953 @item -fvolatile-global
6954 Consider all memory references to extern and global data items to
6955 be volatile. GCC does not consider static data items to be volatile
6956 because of this switch.
6958 @item -fvolatile-static
6959 Consider all memory references to static data to be volatile.
6962 @cindex global offset table
6964 Generate position-independent code (PIC) suitable for use in a shared
6965 library, if supported for the target machine. Such code accesses all
6966 constant addresses through a global offset table (GOT). The dynamic
6967 loader resolves the GOT entries when the program starts (the dynamic
6968 loader is not part of GCC; it is part of the operating system). If
6969 the GOT size for the linked executable exceeds a machine-specific
6970 maximum size, you get an error message from the linker indicating that
6971 @samp{-fpic} does not work; in that case, recompile with @samp{-fPIC}
6972 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
6973 on the m68k and RS/6000. The 386 has no such limit.)
6975 Position-independent code requires special support, and therefore works
6976 only on certain machines. For the 386, GCC supports PIC for System V
6977 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
6978 position-independent.
6981 If supported for the target machine, emit position-independent code,
6982 suitable for dynamic linking and avoiding any limit on the size of the
6983 global offset table. This option makes a difference on the m68k, m88k,
6986 Position-independent code requires special support, and therefore works
6987 only on certain machines.
6989 @item -ffixed-@var{reg}
6990 Treat the register named @var{reg} as a fixed register; generated code
6991 should never refer to it (except perhaps as a stack pointer, frame
6992 pointer or in some other fixed role).
6994 @var{reg} must be the name of a register. The register names accepted
6995 are machine-specific and are defined in the @code{REGISTER_NAMES}
6996 macro in the machine description macro file.
6998 This flag does not have a negative form, because it specifies a
7001 @item -fcall-used-@var{reg}
7002 Treat the register named @var{reg} as an allocable register that is
7003 clobbered by function calls. It may be allocated for temporaries or
7004 variables that do not live across a call. Functions compiled this way
7005 will not save and restore the register @var{reg}.
7007 It is an error to used this flag with the frame pointer or stack pointer.
7008 Use of this flag for other registers that have fixed pervasive roles in
7009 the machine's execution model will produce disastrous results.
7011 This flag does not have a negative form, because it specifies a
7014 @item -fcall-saved-@var{reg}
7015 Treat the register named @var{reg} as an allocable register saved by
7016 functions. It may be allocated even for temporaries or variables that
7017 live across a call. Functions compiled this way will save and restore
7018 the register @var{reg} if they use it.
7020 It is an error to used this flag with the frame pointer or stack pointer.
7021 Use of this flag for other registers that have fixed pervasive roles in
7022 the machine's execution model will produce disastrous results.
7024 A different sort of disaster will result from the use of this flag for
7025 a register in which function values may be returned.
7027 This flag does not have a negative form, because it specifies a
7031 Pack all structure members together without holes. Usually you would
7032 not want to use this option, since it makes the code suboptimal, and
7033 the offsets of structure members won't agree with system libraries.
7035 @item -fcheck-memory-usage
7036 Generate extra code to check each memory access. GCC will generate
7037 code that is suitable for a detector of bad memory accesses such as
7040 Normally, you should compile all, or none, of your code with this option.
7042 If you do mix code compiled with and without this option,
7043 you must ensure that all code that has side effects
7044 and that is called by code compiled with this option
7045 is, itself, compiled with this option.
7046 If you do not, you might get erroneous messages from the detector.
7048 If you use functions from a library that have side-effects (such as
7049 @code{read}), you might not be able to recompile the library and
7050 specify this option. In that case, you can enable the
7051 @samp{-fprefix-function-name} option, which requests GCC to encapsulate
7052 your code and make other functions look as if they were compiled with
7053 @samp{-fcheck-memory-usage}. This is done by calling ``stubs'',
7054 which are provided by the detector. If you cannot find or build
7055 stubs for every function you call, you might have to specify
7056 @samp{-fcheck-memory-usage} without @samp{-fprefix-function-name}.
7058 If you specify this option, you can not use the @code{asm} or
7059 @code{__asm__} keywords in functions with memory checking enabled. GNU
7060 CC cannot understand what the @code{asm} statement may do, and therefore
7061 cannot generate the appropriate code, so it will reject it. However, if
7062 you specify the function attribute @code{no_check_memory_usage} (see
7063 @pxref{Function Attributes}, GNU CC will disable memory checking within a
7064 function; you may use @code{asm} statements inside such functions. You
7065 may have an inline expansion of a non-checked function within a checked
7066 function; in that case GNU CC will not generate checks for the inlined
7067 function's memory accesses.
7069 If you move your @code{asm} statements to non-checked inline functions
7070 and they do access memory, you can add calls to the support code in your
7071 inline function, to indicate any reads, writes, or copies being done.
7072 These calls would be similar to those done in the stubs described above.
7074 @item -fprefix-function-name
7075 Request GCC to add a prefix to the symbols generated for function names.
7076 GCC adds a prefix to the names of functions defined as well as
7077 functions called. Code compiled with this option and code compiled
7078 without the option can't be linked together, unless stubs are used.
7080 If you compile the following code with @samp{-fprefix-function-name}
7082 extern void bar (int);
7091 GCC will compile the code as if it was written:
7093 extern void prefix_bar (int);
7097 return prefix_bar (a + 5);
7100 This option is designed to be used with @samp{-fcheck-memory-usage}.
7102 @item -finstrument-functions
7103 Generate instrumentation calls for entry and exit to functions. Just
7104 after function entry and just before function exit, the following
7105 profiling functions will be called with the address of the current
7106 function and its call site. (On some platforms,
7107 @code{__builtin_return_address} does not work beyond the current
7108 function, so the call site information may not be available to the
7109 profiling functions otherwise.)
7112 void __cyg_profile_func_enter (void *this_fn, void *call_site);
7113 void __cyg_profile_func_exit (void *this_fn, void *call_site);
7116 The first argument is the address of the start of the current function,
7117 which may be looked up exactly in the symbol table.
7119 This instrumentation is also done for functions expanded inline in other
7120 functions. The profiling calls will indicate where, conceptually, the
7121 inline function is entered and exited. This means that addressable
7122 versions of such functions must be available. If all your uses of a
7123 function are expanded inline, this may mean an additional expansion of
7124 code size. If you use @samp{extern inline} in your C code, an
7125 addressable version of such functions must be provided. (This is
7126 normally the case anyways, but if you get lucky and the optimizer always
7127 expands the functions inline, you might have gotten away without
7128 providing static copies.)
7130 A function may be given the attribute @code{no_instrument_function}, in
7131 which case this instrumentation will not be done. This can be used, for
7132 example, for the profiling functions listed above, high-priority
7133 interrupt routines, and any functions from which the profiling functions
7134 cannot safely be called (perhaps signal handlers, if the profiling
7135 routines generate output or allocate memory).
7138 Generate code to verify that you do not go beyond the boundary of the
7139 stack. You should specify this flag if you are running in an
7140 environment with multiple threads, but only rarely need to specify it in
7141 a single-threaded environment since stack overflow is automatically
7142 detected on nearly all systems if there is only one stack.
7144 Note that this switch does not actually cause checking to be done; the
7145 operating system must do that. The switch causes generation of code
7146 to ensure that the operating system sees the stack being extended.
7148 @item -fstack-limit-register=@var{reg}
7149 @itemx -fstack-limit-symbol=@var{sym}
7150 @itemx -fno-stack-limit
7151 Generate code to ensure that the stack does not grow beyond a certain value,
7152 either the value of a register or the address of a symbol. If the stack
7153 would grow beyond the value, a signal is raised. For most targets,
7154 the signal is raised before the stack overruns the boundary, so
7155 it is possible to catch the signal without taking special precautions.
7157 For instance, if the stack starts at address @samp{0x80000000} and grows
7158 downwards you can use the flags
7159 @samp{-fstack-limit-symbol=__stack_limit}
7160 @samp{-Wl,--defsym,__stack_limit=0x7ffe0000} which will enforce a stack
7163 @cindex aliasing of parameters
7164 @cindex parameters, aliased
7165 @item -fargument-alias
7166 @itemx -fargument-noalias
7167 @itemx -fargument-noalias-global
7168 Specify the possible relationships among parameters and between
7169 parameters and global data.
7171 @samp{-fargument-alias} specifies that arguments (parameters) may
7172 alias each other and may alias global storage.
7173 @samp{-fargument-noalias} specifies that arguments do not alias
7174 each other, but may alias global storage.
7175 @samp{-fargument-noalias-global} specifies that arguments do not
7176 alias each other and do not alias global storage.
7178 Each language will automatically use whatever option is required by
7179 the language standard. You should not need to use these options yourself.
7181 @item -fleading-underscore
7182 This option and its counterpart, -fno-leading-underscore, forcibly
7183 change the way C symbols are represented in the object file. One use
7184 is to help link with legacy assembly code.
7186 Be warned that you should know what you are doing when invoking this
7187 option, and that not all targets provide complete support for it.
7190 @node Environment Variables
7191 @section Environment Variables Affecting GCC
7192 @cindex environment variables
7194 This section describes several environment variables that affect how GCC
7195 operates. Some of them work by specifying directories or prefixes to use
7196 when searching for various kinds of files. Some are used to specify other
7197 aspects of the compilation environment.
7200 Note that you can also specify places to search using options such as
7201 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
7202 take precedence over places specified using environment variables, which
7203 in turn take precedence over those specified by the configuration of GCC.
7207 Note that you can also specify places to search using options such as
7208 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
7209 take precedence over places specified using environment variables, which
7210 in turn take precedence over those specified by the configuration of GCC.
7217 @c @itemx LC_COLLATE
7219 @c @itemx LC_MONETARY
7220 @c @itemx LC_NUMERIC
7225 @c @findex LC_COLLATE
7227 @c @findex LC_MONETARY
7228 @c @findex LC_NUMERIC
7232 These environment variables control the way that GCC uses
7233 localization information that allow GCC to work with different
7234 national conventions. GCC inspects the locale categories
7235 @code{LC_CTYPE} and @code{LC_MESSAGES} if it has been configured to do
7236 so. These locale categories can be set to any value supported by your
7237 installation. A typical value is @samp{en_UK} for English in the United
7240 The @code{LC_CTYPE} environment variable specifies character
7241 classification. GCC uses it to determine the character boundaries in
7242 a string; this is needed for some multibyte encodings that contain quote
7243 and escape characters that would otherwise be interpreted as a string
7246 The @code{LC_MESSAGES} environment variable specifies the language to
7247 use in diagnostic messages.
7249 If the @code{LC_ALL} environment variable is set, it overrides the value
7250 of @code{LC_CTYPE} and @code{LC_MESSAGES}; otherwise, @code{LC_CTYPE}
7251 and @code{LC_MESSAGES} default to the value of the @code{LANG}
7252 environment variable. If none of these variables are set, GCC
7253 defaults to traditional C English behavior.
7257 If @code{TMPDIR} is set, it specifies the directory to use for temporary
7258 files. GCC uses temporary files to hold the output of one stage of
7259 compilation which is to be used as input to the next stage: for example,
7260 the output of the preprocessor, which is the input to the compiler
7263 @item GCC_EXEC_PREFIX
7264 @findex GCC_EXEC_PREFIX
7265 If @code{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
7266 names of the subprograms executed by the compiler. No slash is added
7267 when this prefix is combined with the name of a subprogram, but you can
7268 specify a prefix that ends with a slash if you wish.
7270 If @code{GCC_EXEC_PREFIX} is not set, GNU CC will attempt to figure out
7271 an appropriate prefix to use based on the pathname it was invoked with.
7273 If GCC cannot find the subprogram using the specified prefix, it
7274 tries looking in the usual places for the subprogram.
7276 The default value of @code{GCC_EXEC_PREFIX} is
7277 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
7278 of @code{prefix} when you ran the @file{configure} script.
7280 Other prefixes specified with @samp{-B} take precedence over this prefix.
7282 This prefix is also used for finding files such as @file{crt0.o} that are
7285 In addition, the prefix is used in an unusual way in finding the
7286 directories to search for header files. For each of the standard
7287 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
7288 (more precisely, with the value of @code{GCC_INCLUDE_DIR}), GCC tries
7289 replacing that beginning with the specified prefix to produce an
7290 alternate directory name. Thus, with @samp{-Bfoo/}, GCC will search
7291 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
7292 These alternate directories are searched first; the standard directories
7296 @findex COMPILER_PATH
7297 The value of @code{COMPILER_PATH} is a colon-separated list of
7298 directories, much like @code{PATH}. GCC tries the directories thus
7299 specified when searching for subprograms, if it can't find the
7300 subprograms using @code{GCC_EXEC_PREFIX}.
7303 @findex LIBRARY_PATH
7304 The value of @code{LIBRARY_PATH} is a colon-separated list of
7305 directories, much like @code{PATH}. When configured as a native compiler,
7306 GCC tries the directories thus specified when searching for special
7307 linker files, if it can't find them using @code{GCC_EXEC_PREFIX}. Linking
7308 using GCC also uses these directories when searching for ordinary
7309 libraries for the @samp{-l} option (but directories specified with
7310 @samp{-L} come first).
7312 @item C_INCLUDE_PATH
7313 @itemx CPLUS_INCLUDE_PATH
7314 @itemx OBJC_INCLUDE_PATH
7315 @findex C_INCLUDE_PATH
7316 @findex CPLUS_INCLUDE_PATH
7317 @findex OBJC_INCLUDE_PATH
7318 @c @itemx OBJCPLUS_INCLUDE_PATH
7319 These environment variables pertain to particular languages. Each
7320 variable's value is a colon-separated list of directories, much like
7321 @code{PATH}. When GCC searches for header files, it tries the
7322 directories listed in the variable for the language you are using, after
7323 the directories specified with @samp{-I} but before the standard header
7326 @item DEPENDENCIES_OUTPUT
7327 @findex DEPENDENCIES_OUTPUT
7328 @cindex dependencies for make as output
7329 If this variable is set, its value specifies how to output dependencies
7330 for Make based on the header files processed by the compiler. This
7331 output looks much like the output from the @samp{-M} option
7332 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
7333 in addition to the usual results of compilation.
7335 The value of @code{DEPENDENCIES_OUTPUT} can be just a file name, in
7336 which case the Make rules are written to that file, guessing the target
7337 name from the source file name. Or the value can have the form
7338 @samp{@var{file} @var{target}}, in which case the rules are written to
7339 file @var{file} using @var{target} as the target name.
7343 @cindex locale definition
7344 This variable is used to pass locale information to the compiler. One way in
7345 which this information is used is to determine the character set to be used
7346 when character literals, string literals and comments are parsed in C and C++.
7347 When the compiler is configured to allow multibyte characters,
7348 the following values for @code{LANG} are recognized:
7352 Recognize JIS characters.
7354 Recognize SJIS characters.
7356 Recognize EUCJP characters.
7359 If @code{LANG} is not defined, or if it has some other value, then the
7360 compiler will use mblen and mbtowc as defined by the default locale to
7361 recognize and translate multibyte characters.
7364 @node Running Protoize
7365 @section Running Protoize
7367 The program @code{protoize} is an optional part of GNU C. You can use
7368 it to add prototypes to a program, thus converting the program to ANSI
7369 C in one respect. The companion program @code{unprotoize} does the
7370 reverse: it removes argument types from any prototypes that are found.
7372 When you run these programs, you must specify a set of source files as
7373 command line arguments. The conversion programs start out by compiling
7374 these files to see what functions they define. The information gathered
7375 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
7377 After scanning comes actual conversion. The specified files are all
7378 eligible to be converted; any files they include (whether sources or
7379 just headers) are eligible as well.
7381 But not all the eligible files are converted. By default,
7382 @code{protoize} and @code{unprotoize} convert only source and header
7383 files in the current directory. You can specify additional directories
7384 whose files should be converted with the @samp{-d @var{directory}}
7385 option. You can also specify particular files to exclude with the
7386 @samp{-x @var{file}} option. A file is converted if it is eligible, its
7387 directory name matches one of the specified directory names, and its
7388 name within the directory has not been excluded.
7390 Basic conversion with @code{protoize} consists of rewriting most
7391 function definitions and function declarations to specify the types of
7392 the arguments. The only ones not rewritten are those for varargs
7395 @code{protoize} optionally inserts prototype declarations at the
7396 beginning of the source file, to make them available for any calls that
7397 precede the function's definition. Or it can insert prototype
7398 declarations with block scope in the blocks where undeclared functions
7401 Basic conversion with @code{unprotoize} consists of rewriting most
7402 function declarations to remove any argument types, and rewriting
7403 function definitions to the old-style pre-ANSI form.
7405 Both conversion programs print a warning for any function declaration or
7406 definition that they can't convert. You can suppress these warnings
7409 The output from @code{protoize} or @code{unprotoize} replaces the
7410 original source file. The original file is renamed to a name ending
7411 with @samp{.save}. If the @samp{.save} file already exists, then
7412 the source file is simply discarded.
7414 @code{protoize} and @code{unprotoize} both depend on GCC itself to
7415 scan the program and collect information about the functions it uses.
7416 So neither of these programs will work until GCC is installed.
7418 Here is a table of the options you can use with @code{protoize} and
7419 @code{unprotoize}. Each option works with both programs unless
7423 @item -B @var{directory}
7424 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
7425 usual directory (normally @file{/usr/local/lib}). This file contains
7426 prototype information about standard system functions. This option
7427 applies only to @code{protoize}.
7429 @item -c @var{compilation-options}
7430 Use @var{compilation-options} as the options when running @code{gcc} to
7431 produce the @samp{.X} files. The special option @samp{-aux-info} is
7432 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
7434 Note that the compilation options must be given as a single argument to
7435 @code{protoize} or @code{unprotoize}. If you want to specify several
7436 @code{gcc} options, you must quote the entire set of compilation options
7437 to make them a single word in the shell.
7439 There are certain @code{gcc} arguments that you cannot use, because they
7440 would produce the wrong kind of output. These include @samp{-g},
7441 @samp{-O}, @samp{-c}, @samp{-S}, and @samp{-o} If you include these in
7442 the @var{compilation-options}, they are ignored.
7445 Rename files to end in @samp{.C}, or @samp{.cc} for DOS-based file
7446 systems, instead of @samp{.c}. This is convenient if you are converting
7447 a C program to C++. This option applies only to @code{protoize}.
7450 Add explicit global declarations. This means inserting explicit
7451 declarations at the beginning of each source file for each function
7452 that is called in the file and was not declared. These declarations
7453 precede the first function definition that contains a call to an
7454 undeclared function. This option applies only to @code{protoize}.
7456 @item -i @var{string}
7457 Indent old-style parameter declarations with the string @var{string}.
7458 This option applies only to @code{protoize}.
7460 @code{unprotoize} converts prototyped function definitions to old-style
7461 function definitions, where the arguments are declared between the
7462 argument list and the initial @samp{@{}. By default, @code{unprotoize}
7463 uses five spaces as the indentation. If you want to indent with just
7464 one space instead, use @samp{-i " "}.
7467 Keep the @samp{.X} files. Normally, they are deleted after conversion
7471 Add explicit local declarations. @code{protoize} with @samp{-l} inserts
7472 a prototype declaration for each function in each block which calls the
7473 function without any declaration. This option applies only to
7477 Make no real changes. This mode just prints information about the conversions
7478 that would have been done without @samp{-n}.
7481 Make no @samp{.save} files. The original files are simply deleted.
7482 Use this option with caution.
7484 @item -p @var{program}
7485 Use the program @var{program} as the compiler. Normally, the name
7489 Work quietly. Most warnings are suppressed.
7492 Print the version number, just like @samp{-v} for @code{gcc}.
7495 If you need special compiler options to compile one of your program's
7496 source files, then you should generate that file's @samp{.X} file
7497 specially, by running @code{gcc} on that source file with the
7498 appropriate options and the option @samp{-aux-info}. Then run
7499 @code{protoize} on the entire set of files. @code{protoize} will use
7500 the existing @samp{.X} file because it is newer than the source file.
7504 gcc -Dfoo=bar file1.c -aux-info
7509 You need to include the special files along with the rest in the
7510 @code{protoize} command, even though their @samp{.X} files already
7511 exist, because otherwise they won't get converted.
7513 @xref{Protoize Caveats}, for more information on how to use
7514 @code{protoize} successfully.