1 @c Copyright (C) 1988, 89, 92-98, 1999 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 -flang-isoc9x -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} -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 -mno-new-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}
365 -march=@var{architecture type}
366 -mbig-switch -mdisable-fpregs -mdisable-indexing
367 -mfast-indirect-calls -mgas -mjump-in-delay
368 -mlong-load-store -mno-big-switch -mno-disable-fpregs
369 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas
370 -mno-jump-in-delay -mno-long-load-store
371 -mno-portable-runtime -mno-soft-float
372 -mno-space-regs -msoft-float -mpa-risc-1-0
373 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime
374 -mschedule=@var{cpu type} -mspace-regs
376 @emph{Intel 960 Options}
377 -m@var{cpu type} -masm-compat -mclean-linkage
378 -mcode-align -mcomplex-addr -mleaf-procedures
379 -mic-compat -mic2.0-compat -mic3.0-compat
380 -mintel-asm -mno-clean-linkage -mno-code-align
381 -mno-complex-addr -mno-leaf-procedures
382 -mno-old-align -mno-strict-align -mno-tail-call
383 -mnumerics -mold-align -msoft-float -mstrict-align
386 @emph{DEC Alpha Options}
387 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float
389 -mieee -mieee-with-inexact -mieee-conformant
390 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode}
391 -mtrap-precision=@var{mode} -mbuild-constants
393 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max
394 -mmemory-latency=@var{time}
396 @emph{Clipper Options}
399 @emph{H8/300 Options}
400 -mrelax -mh -ms -mint32 -malign-300
403 -m1 -m2 -m3 -m3e -mb -ml -mdalign -mrelax
405 @emph{System V Options}
406 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}
410 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text section}
411 -mdata=@var{data section} -mrodata=@var{readonly data section}
413 @emph{TMS320C3x/C4x Options}
414 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm
415 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload
416 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned
417 -mparallel-insns -mparallel-mpy -mpreserve-float
420 -mlong-calls -mno-long-calls -mep -mno-ep
421 -mprolog-function -mno-prolog-function -mspace
422 -mtda=@var{n} -msda=@var{n} -mzda=@var{n}
426 -m32032 -m32332 -m32532 -m32081 -m32381 -mmult-add -mnomult-add
427 -msoft-float -mrtd -mnortd -mregparam -mnoregparam -msb -mnosb
428 -mbitfield -mnobitfield -mhimem -mnohimem
431 @item Code Generation Options
432 @xref{Code Gen Options,,Options for Code Generation Conventions}.
434 -fcall-saved-@var{reg} -fcall-used-@var{reg}
435 -fexceptions -funwind-tables -ffixed-@var{reg} -finhibit-size-directive
436 -fcheck-memory-usage -fprefix-function-name
437 -fno-common -fno-ident -fno-gnu-linker
438 -fpcc-struct-return -fpic -fPIC
439 -freg-struct-return -fshared-data -fshort-enums
440 -fshort-double -fvolatile -fvolatile-global -fvolatile-static
441 -fverbose-asm -fpack-struct -fstack-check
442 -fargument-alias -fargument-noalias
443 -fargument-noalias-global
449 * Overall Options:: Controlling the kind of output:
450 an executable, object files, assembler files,
451 or preprocessed source.
452 * C Dialect Options:: Controlling the variant of C language compiled.
453 * C++ Dialect Options:: Variations on C++.
454 * Warning Options:: How picky should the compiler be?
455 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
456 * Optimize Options:: How much optimization?
457 * Preprocessor Options:: Controlling header files and macro definitions.
458 Also, getting dependency information for Make.
459 * Assembler Options:: Passing options to the assembler.
460 * Link Options:: Specifying libraries and so on.
461 * Directory Options:: Where to find header files and libraries.
462 Where to find the compiler executable files.
463 * Spec Files:: How to pass switches to sub-processes.
464 * Target Options:: Running a cross-compiler, or an old version of GCC.
467 @node Overall Options
468 @section Options Controlling the Kind of Output
470 Compilation can involve up to four stages: preprocessing, compilation
471 proper, assembly and linking, always in that order. The first three
472 stages apply to an individual source file, and end by producing an
473 object file; linking combines all the object files (those newly
474 compiled, and those specified as input) into an executable file.
476 @cindex file name suffix
477 For any given input file, the file name suffix determines what kind of
482 C source code which must be preprocessed.
485 C source code which should not be preprocessed.
488 C++ source code which should not be preprocessed.
491 Objective-C source code. Note that you must link with the library
492 @file{libobjc.a} to make an Objective-C program work.
495 C header file (not to be compiled or linked).
498 @itemx @var{file}.cxx
499 @itemx @var{file}.cpp
501 C++ source code which must be preprocessed. Note that in @samp{.cxx},
502 the last two letters must both be literally @samp{x}. Likewise,
503 @samp{.C} refers to a literal capital C.
509 Assembler code which must be preprocessed.
512 An object file to be fed straight into linking.
513 Any file name with no recognized suffix is treated this way.
516 You can specify the input language explicitly with the @samp{-x} option:
519 @item -x @var{language}
520 Specify explicitly the @var{language} for the following input files
521 (rather than letting the compiler choose a default based on the file
522 name suffix). This option applies to all following input files until
523 the next @samp{-x} option. Possible values for @var{language} are:
526 c-header cpp-output c++-cpp-output
527 assembler assembler-with-cpp
531 Turn off any specification of a language, so that subsequent files are
532 handled according to their file name suffixes (as they are if @samp{-x}
533 has not been used at all).
535 @item -pass-exit-codes
536 Normally the @code{gcc} program will exit with the code of 1 if any
537 phase of the compiler returns a non-success return code. If you specify
538 @samp{-pass-exit-codes}, the @code{gcc} program will instead return with
539 numerically highest error produced by any phase that returned an error
543 If you only want some of the stages of compilation, you can use
544 @samp{-x} (or filename suffixes) to tell @code{gcc} where to start, and
545 one of the options @samp{-c}, @samp{-S}, or @samp{-E} to say where
546 @code{gcc} is to stop. Note that some combinations (for example,
547 @samp{-x cpp-output -E} instruct @code{gcc} to do nothing at all.
551 Compile or assemble the source files, but do not link. The linking
552 stage simply is not done. The ultimate output is in the form of an
553 object file for each source file.
555 By default, the object file name for a source file is made by replacing
556 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
558 Unrecognized input files, not requiring compilation or assembly, are
562 Stop after the stage of compilation proper; do not assemble. The output
563 is in the form of an assembler code file for each non-assembler input
566 By default, the assembler file name for a source file is made by
567 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
569 Input files that don't require compilation are ignored.
572 Stop after the preprocessing stage; do not run the compiler proper. The
573 output is in the form of preprocessed source code, which is sent to the
576 Input files which don't require preprocessing are ignored.
578 @cindex output file option
580 Place output in file @var{file}. This applies regardless to whatever
581 sort of output is being produced, whether it be an executable file,
582 an object file, an assembler file or preprocessed C code.
584 Since only one output file can be specified, it does not make sense to
585 use @samp{-o} when compiling more than one input file, unless you are
586 producing an executable file as output.
588 If @samp{-o} is not specified, the default is to put an executable file
589 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
590 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
591 all preprocessed C source on standard output.@refill
594 Print (on standard error output) the commands executed to run the stages
595 of compilation. Also print the version number of the compiler driver
596 program and of the preprocessor and the compiler proper.
599 Use pipes rather than temporary files for communication between the
600 various stages of compilation. This fails to work on some systems where
601 the assembler is unable to read from a pipe; but the GNU assembler has
605 Print (on the standard output) a description of the command line options
606 understood by @code{gcc}. If the @code{-v} option is also specified
607 then @code{--help} will also be passed on to the various processes
608 invoked by @code{gcc}, so that they can display the command line options
609 they accept. If the @code{-W} option is also specified then command
610 line options which have no documentation associated with them will also
615 @section Compiling C++ Programs
617 @cindex suffixes for C++ source
618 @cindex C++ source file suffixes
619 C++ source files conventionally use one of the suffixes @samp{.C},
620 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
621 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
622 files with these names and compiles them as C++ programs even if you
623 call the compiler the same way as for compiling C programs (usually with
624 the name @code{gcc}).
628 However, C++ programs often require class libraries as well as a
629 compiler that understands the C++ language---and under some
630 circumstances, you might want to compile programs from standard input,
631 or otherwise without a suffix that flags them as C++ programs.
632 @code{g++} is a program that calls GCC with the default language
633 set to C++, and automatically specifies linking against the C++
634 library. On many systems, the script @code{g++} is also
635 installed with the name @code{c++}.
637 @cindex invoking @code{g++}
638 When you compile C++ programs, you may specify many of the same
639 command-line options that you use for compiling programs in any
640 language; or command-line options meaningful for C and related
641 languages; or options that are meaningful only for C++ programs.
642 @xref{C Dialect Options,,Options Controlling C Dialect}, for
643 explanations of options for languages related to C.
644 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
645 explanations of options that are meaningful only for C++ programs.
647 @node C Dialect Options
648 @section Options Controlling C Dialect
649 @cindex dialect options
650 @cindex language dialect options
651 @cindex options, dialect
653 The following options control the dialect of C (or languages derived
654 from C, such as C++ and Objective C) that the compiler accepts:
659 In C mode, support all ANSI standard C programs. In C++ mode,
660 remove GNU extensions that conflict with ANSI C++.
661 @c shouldn't we be saying "ISO"?
663 This turns off certain features of GCC that are incompatible with ANSI
664 C (when compiling C code), or of ANSI standard C++ (when compiling C++ code),
665 such as the @code{asm} and @code{typeof} keywords, and
666 predefined macros such as @code{unix} and @code{vax} that identify the
667 type of system you are using. It also enables the undesirable and
668 rarely used ANSI trigraph feature. For the C compiler,
669 it disables recognition of C++ style @samp{//} comments as well as
670 the @code{inline} keyword. For the C++ compiler,
671 @samp{-foperator-names} is enabled as well.
674 The alternate keywords @code{__asm__}, @code{__extension__},
675 @code{__inline__} and @code{__typeof__} continue to work despite
676 @samp{-ansi}. You would not want to use them in an ANSI C program, of
677 course, but it is useful to put them in header files that might be included
678 in compilations done with @samp{-ansi}. Alternate predefined macros
679 such as @code{__unix__} and @code{__vax__} are also available, with or
680 without @samp{-ansi}.
682 The @samp{-ansi} option does not cause non-ANSI programs to be
683 rejected gratuitously. For that, @samp{-pedantic} is required in
684 addition to @samp{-ansi}. @xref{Warning Options}.
686 The macro @code{__STRICT_ANSI__} is predefined when the @samp{-ansi}
687 option is used. Some header files may notice this macro and refrain
688 from declaring certain functions or defining certain macros that the
689 ANSI standard doesn't call for; this is to avoid interfering with any
690 programs that might use these names for other things.
692 The functions @code{alloca}, @code{abort}, @code{exit}, and
693 @code{_exit} are not builtin functions when @samp{-ansi} is used.
696 Enable support for features found in the C9X standard. In particular,
697 enable support for the C9X @code{restrict} keyword.
699 Even when this option is not specified, you can still use some C9X
700 features in so far as they do not conflict with previous C standards.
701 For example, you may use @code{__restrict__} even when -flang-isoc9x
705 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
706 keyword, so that code can use these words as identifiers. You can use
707 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
708 instead. @samp{-ansi} implies @samp{-fno-asm}.
710 In C++, this switch only affects the @code{typeof} keyword, since
711 @code{asm} and @code{inline} are standard keywords. You may want to
712 use the @samp{-fno-gnu-keywords} flag instead, as it also disables the
713 other, C++-specific, extension keywords such as @code{headof}.
716 @cindex builtin functions
742 Don't recognize builtin functions that do not begin with @samp{__builtin_}
743 as prefix. Currently, the functions affected include @code{abort},
744 @code{abs}, @code{alloca}, @code{cos}, @code{cosf}, @code{cosl},
745 @code{exit}, @code{_exit}, @code{fabs}, @code{fabsf}, @code{fabsl},
746 @code{ffs}, @code{labs}, @code{memcmp}, @code{memcpy}, @code{memset},
747 @code{sin}, @code{sinf}, @code{sinl}, @code{sqrt}, @code{sqrtf},
748 @code{sqrtl}, @code{strcmp}, @code{strcpy}, and @code{strlen}.
750 GCC normally generates special code to handle certain builtin functions
751 more efficiently; for instance, calls to @code{alloca} may become single
752 instructions that adjust the stack directly, and calls to @code{memcpy}
753 may become inline copy loops. The resulting code is often both smaller
754 and faster, but since the function calls no longer appear as such, you
755 cannot set a breakpoint on those calls, nor can you change the behavior
756 of the functions by linking with a different library.
758 The @samp{-ansi} option prevents @code{alloca}, @code{ffs} and @code{_exit}
759 from being builtin functions, since these functions do not have an ANSI
763 @cindex hosted environment
765 Assert that compilation takes place in a hosted environment. This implies
766 @samp{-fbuiltin}. A hosted environment is one in which the
767 entire standard library is available, and in which @code{main} has a return
768 type of @code{int}. Examples are nearly everything except a kernel.
769 This is equivalent to @samp{-fno-freestanding}.
772 @cindex hosted environment
774 Assert that compilation takes place in a freestanding environment. This
775 implies @samp{-fno-builtin}. A freestanding environment
776 is one in which the standard library may not exist, and program startup may
777 not necessarily be at @code{main}. The most obvious example is an OS kernel.
778 This is equivalent to @samp{-fno-hosted}.
781 Support ANSI C trigraphs. You don't want to know about this
782 brain-damage. The @samp{-ansi} option implies @samp{-trigraphs}.
784 @cindex traditional C language
785 @cindex C language, traditional
787 Attempt to support some aspects of traditional C compilers.
792 All @code{extern} declarations take effect globally even if they
793 are written inside of a function definition. This includes implicit
794 declarations of functions.
797 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
798 and @code{volatile} are not recognized. (You can still use the
799 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
803 Comparisons between pointers and integers are always allowed.
806 Integer types @code{unsigned short} and @code{unsigned char} promote
807 to @code{unsigned int}.
810 Out-of-range floating point literals are not an error.
813 Certain constructs which ANSI regards as a single invalid preprocessing
814 number, such as @samp{0xe-0xd}, are treated as expressions instead.
817 String ``constants'' are not necessarily constant; they are stored in
818 writable space, and identical looking constants are allocated
819 separately. (This is the same as the effect of
820 @samp{-fwritable-strings}.)
822 @cindex @code{longjmp} and automatic variables
824 All automatic variables not declared @code{register} are preserved by
825 @code{longjmp}. Ordinarily, GNU C follows ANSI C: automatic variables
826 not declared @code{volatile} may be clobbered.
831 @cindex escape sequences, traditional
832 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
833 literal characters @samp{x} and @samp{a} respectively. Without
834 @w{@samp{-traditional}}, @samp{\x} is a prefix for the hexadecimal
835 representation of a character, and @samp{\a} produces a bell.
838 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
839 if your program uses names that are normally GNU C builtin functions for
840 other purposes of its own.
842 You cannot use @samp{-traditional} if you include any header files that
843 rely on ANSI C features. Some vendors are starting to ship systems with
844 ANSI C header files and you cannot use @samp{-traditional} on such
845 systems to compile files that include any system headers.
847 The @samp{-traditional} option also enables @samp{-traditional-cpp},
848 which is described next.
850 @item -traditional-cpp
851 Attempt to support some aspects of traditional C preprocessors.
856 Comments convert to nothing at all, rather than to a space. This allows
857 traditional token concatenation.
860 In a preprocessing directive, the @samp{#} symbol must appear as the first
864 Macro arguments are recognized within string constants in a macro
865 definition (and their values are stringified, though without additional
866 quote marks, when they appear in such a context). The preprocessor
867 always considers a string constant to end at a newline.
870 @cindex detecting @w{@samp{-traditional}}
871 The predefined macro @code{__STDC__} is not defined when you use
872 @samp{-traditional}, but @code{__GNUC__} is (since the GNU extensions
873 which @code{__GNUC__} indicates are not affected by
874 @samp{-traditional}). If you need to write header files that work
875 differently depending on whether @samp{-traditional} is in use, by
876 testing both of these predefined macros you can distinguish four
877 situations: GNU C, traditional GNU C, other ANSI C compilers, and other
878 old C compilers. The predefined macro @code{__STDC_VERSION__} is also
879 not defined when you use @samp{-traditional}. @xref{Standard
880 Predefined,,Standard Predefined Macros,cpp.info,The C Preprocessor},
881 for more discussion of these and other predefined macros.
884 @cindex string constants vs newline
885 @cindex newline vs string constants
886 The preprocessor considers a string constant to end at a newline (unless
887 the newline is escaped with @samp{\}). (Without @w{@samp{-traditional}},
888 string constants can contain the newline character as typed.)
891 @item -fcond-mismatch
892 Allow conditional expressions with mismatched types in the second and
893 third arguments. The value of such an expression is void.
895 @item -funsigned-char
896 Let the type @code{char} be unsigned, like @code{unsigned char}.
898 Each kind of machine has a default for what @code{char} should
899 be. It is either like @code{unsigned char} by default or like
900 @code{signed char} by default.
902 Ideally, a portable program should always use @code{signed char} or
903 @code{unsigned char} when it depends on the signedness of an object.
904 But many programs have been written to use plain @code{char} and
905 expect it to be signed, or expect it to be unsigned, depending on the
906 machines they were written for. This option, and its inverse, let you
907 make such a program work with the opposite default.
909 The type @code{char} is always a distinct type from each of
910 @code{signed char} or @code{unsigned char}, even though its behavior
911 is always just like one of those two.
914 Let the type @code{char} be signed, like @code{signed char}.
916 Note that this is equivalent to @samp{-fno-unsigned-char}, which is
917 the negative form of @samp{-funsigned-char}. Likewise, the option
918 @samp{-fno-signed-char} is equivalent to @samp{-funsigned-char}.
920 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
921 if your program uses names that are normally GNU C builtin functions for
922 other purposes of its own.
924 You cannot use @samp{-traditional} if you include any header files that
925 rely on ANSI C features. Some vendors are starting to ship systems with
926 ANSI C header files and you cannot use @samp{-traditional} on such
927 systems to compile files that include any system headers.
929 @item -fsigned-bitfields
930 @itemx -funsigned-bitfields
931 @itemx -fno-signed-bitfields
932 @itemx -fno-unsigned-bitfields
933 These options control whether a bitfield is signed or unsigned, when the
934 declaration does not use either @code{signed} or @code{unsigned}. By
935 default, such a bitfield is signed, because this is consistent: the
936 basic integer types such as @code{int} are signed types.
938 However, when @samp{-traditional} is used, bitfields are all unsigned
941 @item -fwritable-strings
942 Store string constants in the writable data segment and don't uniquize
943 them. This is for compatibility with old programs which assume they can
944 write into string constants. The option @samp{-traditional} also has
947 Writing into string constants is a very bad idea; ``constants'' should
950 @item -fallow-single-precision
951 Do not promote single precision math operations to double precision,
952 even when compiling with @samp{-traditional}.
954 Traditional K&R C promotes all floating point operations to double
955 precision, regardless of the sizes of the operands. On the
956 architecture for which you are compiling, single precision may be faster
957 than double precision. If you must use @samp{-traditional}, but want
958 to use single precision operations when the operands are single
959 precision, use this option. This option has no effect when compiling
960 with ANSI or GNU C conventions (the default).
963 Override the underlying type for @samp{wchar_t} to be @samp{short
964 unsigned int} instead of the default for the target. This option is
965 useful for building programs to run under WINE.
968 @node C++ Dialect Options
969 @section Options Controlling C++ Dialect
971 @cindex compiler options, C++
972 @cindex C++ options, command line
974 This section describes the command-line options that are only meaningful
975 for C++ programs; but you can also use most of the GNU compiler options
976 regardless of what language your program is in. For example, you
977 might compile a file @code{firstClass.C} like this:
980 g++ -g -frepo -O -c firstClass.C
984 In this example, only @samp{-frepo} is an option meant
985 only for C++ programs; you can use the other options with any
986 language supported by GCC.
988 Here is a list of options that are @emph{only} for compiling C++ programs:
991 @item -fno-access-control
992 Turn off all access checking. This switch is mainly useful for working
993 around bugs in the access control code.
996 Check that the pointer returned by @code{operator new} is non-null
997 before attempting to modify the storage allocated. The current Working
998 Paper requires that @code{operator new} never return a null pointer, so
999 this check is normally unnecessary.
1001 An alternative to using this option is to specify that your
1002 @code{operator new} does not throw any exceptions; if you declare it
1003 @samp{throw()}, g++ will check the return value. See also @samp{new
1006 @item -fconserve-space
1007 Put uninitialized or runtime-initialized global variables into the
1008 common segment, as C does. This saves space in the executable at the
1009 cost of not diagnosing duplicate definitions. If you compile with this
1010 flag and your program mysteriously crashes after @code{main()} has
1011 completed, you may have an object that is being destroyed twice because
1012 two definitions were merged.
1014 This option is no longer useful on most targets, now that support has
1015 been added for putting variables into BSS without making them common.
1017 @item -fdollars-in-identifiers
1018 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1019 @samp{$} with the option @samp{-fno-dollars-in-identifiers}. (GNU C allows
1020 @samp{$} by default on most target systems, but there are a few exceptions.)
1021 Traditional C allowed the character @samp{$} to form part of
1022 identifiers. However, ANSI C and C++ forbid @samp{$} in identifiers.
1024 @item -fno-elide-constructors
1025 The C++ standard allows an implementation to omit creating a temporary
1026 which is only used to initialize another object of the same type.
1027 Specifying this option disables that optimization, and forces g++ to
1028 call the copy constructor in all cases.
1030 @item -fexternal-templates
1031 Cause template instantiations to obey @samp{#pragma interface} and
1032 @samp{implementation}; template instances are emitted or not according
1033 to the location of the template definition. @xref{Template
1034 Instantiation}, for more information.
1036 This option is deprecated.
1038 @item -falt-external-templates
1039 Similar to -fexternal-templates, but template instances are emitted or
1040 not according to the place where they are first instantiated.
1041 @xref{Template Instantiation}, for more information.
1043 This option is deprecated.
1046 @itemx -fno-for-scope
1047 If -ffor-scope is specified, the scope of variables declared in
1048 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1049 as specified by the draft C++ standard.
1050 If -fno-for-scope is specified, the scope of variables declared in
1051 a @i{for-init-statement} extends to the end of the enclosing scope,
1052 as was the case in old versions of gcc, and other (traditional)
1053 implementations of C++.
1055 The default if neither flag is given to follow the standard,
1056 but to allow and give a warning for old-style code that would
1057 otherwise be invalid, or have different behavior.
1059 @item -fno-gnu-keywords
1060 Do not recognize @code{classof}, @code{headof}, or @code{typeof} as a
1061 keyword, so that code can use these words as identifiers. You can use
1062 the keywords @code{__classof__}, @code{__headof__}, and
1063 @code{__typeof__} instead. @samp{-ansi} implies
1064 @samp{-fno-gnu-keywords}.
1066 @item -fguiding-decls
1067 Treat a function declaration with the same type as a potential function
1068 template instantiation as though it declares that instantiation, not a
1069 normal function. If a definition is given for the function later in the
1070 translation unit (or another translation unit if the target supports
1071 weak symbols), that definition will be used; otherwise the template will
1072 be instantiated. This behavior reflects the C++ language prior to
1073 September 1996, when guiding declarations were removed.
1075 This option implies @samp{-fname-mangling-version-0}, and will not work
1076 with other name mangling versions. Like all options that change the
1077 ABI, all C++ code, @emph{including libgcc.a} must be built with the same
1078 setting of this option.
1081 Treat the @code{namespace std} as a namespace, instead of ignoring
1082 it. For compatibility with earlier versions of g++, the compiler will,
1083 by default, ignore @code{namespace-declarations},
1084 @code{using-declarations}, @code{using-directives}, and
1085 @code{namespace-names}, if they involve @code{std}.
1087 @item -fhuge-objects
1088 Support virtual function calls for objects that exceed the size
1089 representable by a @samp{short int}. Users should not use this flag by
1090 default; if you need to use it, the compiler will tell you so.
1092 This flag is not useful when compiling with -fvtable-thunks.
1094 Like all options that change the ABI, all C++ code, @emph{including
1095 libgcc} must be built with the same setting of this option.
1097 @item -fno-implicit-templates
1098 Never emit code for non-inline templates which are instantiated
1099 implicitly (i.e. by use); only emit code for explicit instantiations.
1100 @xref{Template Instantiation}, for more information.
1102 @item -fno-implicit-inline-templates
1103 Don't emit code for implicit instantiations of inline templates, either.
1104 The default is to handle inlines differently so that compiles with and
1105 without optimization will need the same set of explicit instantiations.
1107 @item -finit-priority
1108 Support @samp{__attribute__ ((init_priority (n)))} for controlling the
1109 order of initialization of file-scope objects. On ELF targets, this
1110 requires GNU ld 2.10 or later.
1112 @item -fno-implement-inlines
1113 To save space, do not emit out-of-line copies of inline functions
1114 controlled by @samp{#pragma implementation}. This will cause linker
1115 errors if these functions are not inlined everywhere they are called.
1117 @item -fms-extensions
1118 Disable pedwarns about constructs used in MFC, such as implicit int and
1119 getting a pointer to member function via non-standard syntax.
1121 @item -fname-mangling-version-@var{n}
1122 Control the way in which names are mangled. Version 0 is compatible
1123 with versions of g++ before 2.8. Version 1 is the default. Version 1
1124 will allow correct mangling of function templates. For example,
1125 version 0 mangling does not mangle foo<int, double> and foo<int, char>
1126 given this declaration:
1129 template <class T, class U> void foo(T t);
1132 Like all options that change the ABI, all C++ code, @emph{including
1133 libgcc} must be built with the same setting of this option.
1135 @item -foperator-names
1136 Recognize the operator name keywords @code{and}, @code{bitand},
1137 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1138 synonyms for the symbols they refer to. @samp{-ansi} implies
1139 @samp{-foperator-names}.
1141 @item -fno-optional-diags
1142 Disable diagnostics that the standard says a compiler does not need to
1143 issue. Currently, the only such diagnostic issued by g++ is the one for
1144 a name having multiple meanings within a class.
1147 Downgrade messages about nonconformant code from errors to warnings. By
1148 default, g++ effectively sets @samp{-pedantic-errors} without
1149 @samp{-pedantic}; this option reverses that. This behavior and this
1150 option are superseded by @samp{-pedantic}, which works as it does for GNU C.
1153 Enable automatic template instantiation. This option also implies
1154 @samp{-fno-implicit-templates}. @xref{Template Instantiation}, for more
1158 Disable generation of information about every class with virtual
1159 functions for use by the C++ runtime type identification features
1160 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1161 of the language, you can save some space by using this flag. Note that
1162 exception handling uses the same information, but it will generate it as
1165 @item -fstrict-prototype
1166 Within an @samp{extern "C"} linkage specification, treat a function
1167 declaration with no arguments, such as @samp{int foo ();}, as declaring
1168 the function to take no arguments. Normally, such a declaration means
1169 that the function @code{foo} can take any combination of arguments, as
1170 in C. @samp{-pedantic} implies @samp{-fstrict-prototype} unless
1171 overridden with @samp{-fno-strict-prototype}.
1173 Specifying this option will also suppress implicit declarations of
1176 This flag no longer affects declarations with C++ linkage.
1179 @itemx -fno-squangle
1180 @samp{-fsquangle} will enable a compressed form of name mangling for
1181 identifiers. In particular, it helps to shorten very long names by recognizing
1182 types and class names which occur more than once, replacing them with special
1183 short ID codes. This option also requires any C++ libraries being used to
1184 be compiled with this option as well. The compiler has this disabled (the
1185 equivalent of @samp{-fno-squangle}) by default.
1187 Like all options that change the ABI, all C++ code, @emph{including
1188 libgcc.a} must be built with the same setting of this option.
1190 @item -ftemplate-depth-@var{n}
1191 Set the maximum instantiation depth for template classes to @var{n}.
1192 A limit on the template instantiation depth is needed to detect
1193 endless recursions during template class instantiation. ANSI/ISO C++
1194 conforming programs must not rely on a maximum depth greater than 17.
1196 @item -fvtable-thunks
1197 Use @samp{thunks} to implement the virtual function dispatch table
1198 (@samp{vtable}). The traditional (cfront-style) approach to
1199 implementing vtables was to store a pointer to the function and two
1200 offsets for adjusting the @samp{this} pointer at the call site. Newer
1201 implementations store a single pointer to a @samp{thunk} function which
1202 does any necessary adjustment and then calls the target function.
1204 This option also enables a heuristic for controlling emission of
1205 vtables; if a class has any non-inline virtual functions, the vtable
1206 will be emitted in the translation unit containing the first one of
1209 Like all options that change the ABI, all C++ code, @emph{including
1210 libgcc.a} must be built with the same setting of this option.
1213 Do not search for header files in the standard directories specific to
1214 C++, but do still search the other standard directories. (This option
1215 is used when building the C++ library.)
1218 In addition, these optimization, warning, and code generation options
1219 have meanings only for C++ programs:
1222 @item -fno-default-inline
1223 Do not assume @samp{inline} for functions defined inside a class scope.
1224 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1225 functions will have linkage like inline functions; they just won't be
1228 @item -Wctor-dtor-privacy (C++ only)
1229 Warn when a class seems unusable, because all the constructors or
1230 destructors in a class are private and the class has no friends or
1231 public static member functions.
1233 @item -Wnon-virtual-dtor (C++ only)
1234 Warn when a class declares a non-virtual destructor that should probably
1235 be virtual, because it looks like the class will be used polymorphically.
1237 @item -Wreorder (C++ only)
1238 @cindex reordering, warning
1239 @cindex warning for reordering of member initializers
1240 Warn when the order of member initializers given in the code does not
1241 match the order in which they must be executed. For instance:
1247 A(): j (0), i (1) @{ @}
1251 Here the compiler will warn that the member initializers for @samp{i}
1252 and @samp{j} will be rearranged to match the declaration order of the
1256 The following @samp{-W@dots{}} options are not affected by @samp{-Wall}.
1259 @item -Weffc++ (C++ only)
1260 Warn about violations of various style guidelines from Scott Meyers'
1261 @cite{Effective C++} books. If you use this option, you should be aware
1262 that the standard library headers do not obey all of these guidelines;
1263 you can use @samp{grep -v} to filter out those warnings.
1265 @item -Wno-deprecated (C++ only)
1266 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1268 @item -Wno-non-template-friend (C++ only)
1269 Disable warnings when non-templatized friend functions are declared
1270 within a template. With the advent of explicit template specification
1271 support in g++, if the name of the friend is an unqualified-id (ie,
1272 @samp{friend foo(int)}), the C++ language specification demands that the
1273 friend declare or define an ordinary, nontemplate function. (Section
1274 14.5.3). Before g++ implemented explicit specification, unqualified-ids
1275 could be interpreted as a particular specialization of a templatized
1276 function. Because this non-conforming behavior is no longer the default
1277 behavior for g++, @samp{-Wnon-template-friend} allows the compiler to
1278 check existing code for potential trouble spots, and is on by default.
1279 This new compiler behavior can also be turned off with the flag
1280 @samp{-fguiding-decls}, which activates the older, non-specification
1281 compiler code, or with @samp{-Wno-non-template-friend} which keeps the
1282 conformant compiler code but disables the helpful warning.
1284 @item -Wold-style-cast (C++ only)
1285 Warn if an old-style (C-style) cast is used within a C++ program. The
1286 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1287 @samp{const_cast}) are less vulnerable to unintended effects.
1289 @item -Woverloaded-virtual (C++ only)
1290 @cindex overloaded virtual fn, warning
1291 @cindex warning for overloaded virtual fn
1292 Warn when a derived class function declaration may be an error in
1293 defining a virtual function. In a derived class, the
1294 definitions of virtual functions must match the type signature of a
1295 virtual function declared in the base class. With this option, the
1296 compiler warns when you define a function with the same name as a
1297 virtual function, but with a type signature that does not match any
1298 declarations from the base class.
1300 @item -Wno-pmf-conversions (C++ only)
1301 Disable the diagnostic for converting a bound pointer to member function
1304 @item -Wsign-promo (C++ only)
1305 Warn when overload resolution chooses a promotion from unsigned or
1306 enumeral type to a signed type over a conversion to an unsigned type of
1307 the same size. Previous versions of g++ would try to preserve
1308 unsignedness, but the standard mandates the current behavior.
1310 @item -Wsynth (C++ only)
1311 @cindex warning for synthesized methods
1312 @cindex synthesized methods, warning
1313 Warn when g++'s synthesis behavior does not match that of cfront. For
1319 A& operator = (int);
1329 In this example, g++ will synthesize a default @samp{A& operator =
1330 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1333 @node Warning Options
1334 @section Options to Request or Suppress Warnings
1335 @cindex options to control warnings
1336 @cindex warning messages
1337 @cindex messages, warning
1338 @cindex suppressing warnings
1340 Warnings are diagnostic messages that report constructions which
1341 are not inherently erroneous but which are risky or suggest there
1342 may have been an error.
1344 You can request many specific warnings with options beginning @samp{-W},
1345 for example @samp{-Wimplicit} to request warnings on implicit
1346 declarations. Each of these specific warning options also has a
1347 negative form beginning @samp{-Wno-} to turn off warnings;
1348 for example, @samp{-Wno-implicit}. This manual lists only one of the
1349 two forms, whichever is not the default.
1351 These options control the amount and kinds of warnings produced by GCC:
1354 @cindex syntax checking
1356 Check the code for syntax errors, but don't do anything beyond that.
1359 Issue all the warnings demanded by strict ANSI C and ISO C++;
1360 reject all programs that use forbidden extensions.
1362 Valid ANSI C and ISO C++ programs should compile properly with or without
1363 this option (though a rare few will require @samp{-ansi}). However,
1364 without this option, certain GNU extensions and traditional C and C++
1365 features are supported as well. With this option, they are rejected.
1367 @samp{-pedantic} does not cause warning messages for use of the
1368 alternate keywords whose names begin and end with @samp{__}. Pedantic
1369 warnings are also disabled in the expression that follows
1370 @code{__extension__}. However, only system header files should use
1371 these escape routes; application programs should avoid them.
1372 @xref{Alternate Keywords}.
1374 This option is not intended to be @i{useful}; it exists only to satisfy
1375 pedants who would otherwise claim that GCC fails to support the ANSI
1378 Some users try to use @samp{-pedantic} to check programs for strict ANSI
1379 C conformance. They soon find that it does not do quite what they want:
1380 it finds some non-ANSI practices, but not all---only those for which
1381 ANSI C @emph{requires} a diagnostic.
1383 A feature to report any failure to conform to ANSI C might be useful in
1384 some instances, but would require considerable additional work and would
1385 be quite different from @samp{-pedantic}. We don't have plans to
1386 support such a feature in the near future.
1388 @item -pedantic-errors
1389 Like @samp{-pedantic}, except that errors are produced rather than
1393 Inhibit all warning messages.
1396 Inhibit warning messages about the use of @samp{#import}.
1398 @item -Wchar-subscripts
1399 Warn if an array subscript has type @code{char}. This is a common cause
1400 of error, as programmers often forget that this type is signed on some
1404 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1405 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1408 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1409 the arguments supplied have types appropriate to the format string
1412 @item -Wimplicit-int
1413 Warn when a declaration does not specify a type.
1415 @item -Wimplicit-function-declaration
1416 @itemx -Werror-implicit-function-declaration
1417 Give a warning (or error) whenever a function is used before being
1421 Same as @samp{-Wimplicit-int} and @samp{-Wimplicit-function-}@*
1425 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1426 function with external linkage, returning int, taking either zero
1427 arguments, two, or three arguments of appropriate types.
1430 Warn if a multicharacter constant (@samp{'FOOF'}) is used. Usually they
1431 indicate a typo in the user's code, as they have implementation-defined
1432 values, and should not be used in portable code.
1435 Warn if parentheses are omitted in certain contexts, such
1436 as when there is an assignment in a context where a truth value
1437 is expected, or when operators are nested whose precedence people
1438 often get confused about.
1440 Also warn about constructions where there may be confusion to which
1441 @code{if} statement an @code{else} branch belongs. Here is an example of
1454 In C, every @code{else} branch belongs to the innermost possible @code{if}
1455 statement, which in this example is @code{if (b)}. This is often not
1456 what the programmer expected, as illustrated in the above example by
1457 indentation the programmer chose. When there is the potential for this
1458 confusion, GNU C will issue a warning when this flag is specified.
1459 To eliminate the warning, add explicit braces around the innermost
1460 @code{if} statement so there is no way the @code{else} could belong to
1461 the enclosing @code{if}. The resulting code would look like this:
1476 Warn whenever a function is defined with a return-type that defaults
1477 to @code{int}. Also warn about any @code{return} statement with no
1478 return-value in a function whose return-type is not @code{void}.
1481 Warn whenever a @code{switch} statement has an index of enumeral type
1482 and lacks a @code{case} for one or more of the named codes of that
1483 enumeration. (The presence of a @code{default} label prevents this
1484 warning.) @code{case} labels outside the enumeration range also
1485 provoke warnings when this option is used.
1488 Warn if any trigraphs are encountered (assuming they are enabled).
1491 Warn whenever a variable is unused aside from its declaration,
1492 whenever a function is declared static but never defined, whenever a
1493 label is declared but not used, and whenever a statement computes a
1494 result that is explicitly not used.
1496 In order to get a warning about an unused function parameter, you must
1497 specify both @samp{-W} and @samp{-Wunused}.
1499 To suppress this warning for an expression, simply cast it to void. For
1500 unused variables, parameters and labels, use the @samp{unused} attribute
1501 (@pxref{Variable Attributes}).
1503 @item -Wuninitialized
1504 Warn if an automatic variable is used without first being initialized or
1505 if a variable may be clobbered by a @code{setjmp} call.
1507 These warnings are possible only in optimizing compilation,
1508 because they require data flow information that is computed only
1509 when optimizing. If you don't specify @samp{-O}, you simply won't
1512 These warnings occur only for variables that are candidates for
1513 register allocation. Therefore, they do not occur for a variable that
1514 is declared @code{volatile}, or whose address is taken, or whose size
1515 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
1516 structures, unions or arrays, even when they are in registers.
1518 Note that there may be no warning about a variable that is used only
1519 to compute a value that itself is never used, because such
1520 computations may be deleted by data flow analysis before the warnings
1523 These warnings are made optional because GCC is not smart
1524 enough to see all the reasons why the code might be correct
1525 despite appearing to have an error. Here is one example of how
1544 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
1545 always initialized, but GCC doesn't know this. Here is
1546 another common case:
1551 if (change_y) save_y = y, y = new_y;
1553 if (change_y) y = save_y;
1558 This has no bug because @code{save_y} is used only if it is set.
1560 @cindex @code{longjmp} warnings
1561 This option also warns when a nonvolatile automatic variable might be
1562 changed by a call to @code{longjmp}. These warnings as well are possible
1563 only in optimizing compilation.
1565 The compiler sees only the calls to @code{setjmp}. It cannot know
1566 where @code{longjmp} will be called; in fact, a signal handler could
1567 call it at any point in the code. As a result, you may get a warning
1568 even when there is in fact no problem because @code{longjmp} cannot
1569 in fact be called at the place which would cause a problem.
1571 Some spurious warnings can be avoided if you declare all the functions
1572 you use that never return as @code{noreturn}. @xref{Function
1575 @item -Wreorder (C++ only)
1576 @cindex reordering, warning
1577 @cindex warning for reordering of member initializers
1578 Warn when the order of member initializers given in the code does not
1579 match the order in which they must be executed. For instance:
1581 @item -Wunknown-pragmas
1582 @cindex warning for unknown pragmas
1583 @cindex unknown pragmas, warning
1584 @cindex pragmas, warning of unknown
1585 Warn when a #pragma directive is encountered which is not understood by
1586 GCC. If this command line option is used, warnings will even be issued
1587 for unknown pragmas in system header files. This is not the case if
1588 the warnings were only enabled by the @samp{-Wall} command line option.
1591 All of the above @samp{-W} options combined. This enables all the
1592 warnings about constructions that some users consider questionable, and
1593 that are easy to avoid (or modify to prevent the warning), even in
1594 conjunction with macros.
1597 The following @samp{-W@dots{}} options are not implied by @samp{-Wall}.
1598 Some of them warn about constructions that users generally do not
1599 consider questionable, but which occasionally you might wish to check
1600 for; others warn about constructions that are necessary or hard to avoid
1601 in some cases, and there is no simple way to modify the code to suppress
1606 Print extra warning messages for these events:
1610 A function can return either with or without a value. (Falling
1611 off the end of the function body is considered returning without
1612 a value.) For example, this function would evoke such a
1626 An expression-statement or the left-hand side of a comma expression
1627 contains no side effects.
1628 To suppress the warning, cast the unused expression to void.
1629 For example, an expression such as @samp{x[i,j]} will cause a warning,
1630 but @samp{x[(void)i,j]} will not.
1633 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
1636 A comparison like @samp{x<=y<=z} appears; this is equivalent to
1637 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
1638 that of ordinary mathematical notation.
1641 Storage-class specifiers like @code{static} are not the first things in
1642 a declaration. According to the C Standard, this usage is obsolescent.
1645 If @samp{-Wall} or @samp{-Wunused} is also specified, warn about unused
1649 A comparison between signed and unsigned values could produce an
1650 incorrect result when the signed value is converted to unsigned.
1651 (But don't warn if @samp{-Wno-sign-compare} is also specified.)
1654 An aggregate has a partly bracketed initializer.
1655 For example, the following code would evoke such a warning,
1656 because braces are missing around the initializer for @code{x.h}:
1659 struct s @{ int f, g; @};
1660 struct t @{ struct s h; int i; @};
1661 struct t x = @{ 1, 2, 3 @};
1665 An aggregate has an initializer which does not initialize all members.
1666 For example, the following code would cause such a warning, because
1667 @code{x.h} would be implicitly initialized to zero:
1670 struct s @{ int f, g, h; @};
1671 struct s x = @{ 3, 4 @};
1676 Warn if floating point values are used in equality comparisons.
1678 @item -Wtraditional (C only)
1679 Warn about certain constructs that behave differently in traditional and
1684 Macro arguments occurring within string constants in the macro body.
1685 These would substitute the argument in traditional C, but are part of
1686 the constant in ANSI C.
1689 A function declared external in one block and then used after the end of
1693 A @code{switch} statement has an operand of type @code{long}.
1696 A non-@code{static} function declaration follows a @code{static} one.
1697 This construct is not accepted by some traditional C compilers.
1700 The ANSI type of an integer constant has a different width or
1701 signedness from its traditional type. This warning is only issued if
1702 the base of the constant is ten. I.e. hexadecimal or octal values, which
1703 typically represent bit patterns, are not warned about.
1707 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
1710 Warn whenever a local variable shadows another local variable.
1712 @item -Wid-clash-@var{len}
1713 Warn whenever two distinct identifiers match in the first @var{len}
1714 characters. This may help you prepare a program that will compile
1715 with certain obsolete, brain-damaged compilers.
1717 @item -Wlarger-than-@var{len}
1718 Warn whenever an object of larger than @var{len} bytes is defined.
1720 @item -Wpointer-arith
1721 Warn about anything that depends on the ``size of'' a function type or
1722 of @code{void}. GNU C assigns these types a size of 1, for
1723 convenience in calculations with @code{void *} pointers and pointers
1726 @item -Wbad-function-cast (C only)
1727 Warn whenever a function call is cast to a non-matching type.
1728 For example, warn if @code{int malloc()} is cast to @code{anything *}.
1731 Warn whenever a pointer is cast so as to remove a type qualifier from
1732 the target type. For example, warn if a @code{const char *} is cast
1733 to an ordinary @code{char *}.
1736 Warn whenever a pointer is cast such that the required alignment of the
1737 target is increased. For example, warn if a @code{char *} is cast to
1738 an @code{int *} on machines where integers can only be accessed at
1739 two- or four-byte boundaries.
1741 @item -Wwrite-strings
1742 Give string constants the type @code{const char[@var{length}]} so that
1743 copying the address of one into a non-@code{const} @code{char *}
1744 pointer will get a warning. These warnings will help you find at
1745 compile time code that can try to write into a string constant, but
1746 only if you have been very careful about using @code{const} in
1747 declarations and prototypes. Otherwise, it will just be a nuisance;
1748 this is why we did not make @samp{-Wall} request these warnings.
1751 Warn if a prototype causes a type conversion that is different from what
1752 would happen to the same argument in the absence of a prototype. This
1753 includes conversions of fixed point to floating and vice versa, and
1754 conversions changing the width or signedness of a fixed point argument
1755 except when the same as the default promotion.
1757 Also, warn if a negative integer constant expression is implicitly
1758 converted to an unsigned type. For example, warn about the assignment
1759 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
1760 casts like @code{(unsigned) -1}.
1762 @item -Wsign-compare
1763 @cindex warning for comparison of signed and unsigned values
1764 @cindex comparison of signed and unsigned values, warning
1765 @cindex signed and unsigned values, comparison warning
1766 Warn when a comparison between signed and unsigned values could produce
1767 an incorrect result when the signed value is converted to unsigned.
1768 This warning is also enabled by @samp{-W}; to get the other warnings
1769 of @samp{-W} without this warning, use @samp{-W -Wno-sign-compare}.
1771 @item -Waggregate-return
1772 Warn if any functions that return structures or unions are defined or
1773 called. (In languages where you can return an array, this also elicits
1776 @item -Wstrict-prototypes (C only)
1777 Warn if a function is declared or defined without specifying the
1778 argument types. (An old-style function definition is permitted without
1779 a warning if preceded by a declaration which specifies the argument
1782 @item -Wmissing-prototypes (C only)
1783 Warn if a global function is defined without a previous prototype
1784 declaration. This warning is issued even if the definition itself
1785 provides a prototype. The aim is to detect global functions that fail
1786 to be declared in header files.
1788 @item -Wmissing-declarations
1789 Warn if a global function is defined without a previous declaration.
1790 Do so even if the definition itself provides a prototype.
1791 Use this option to detect global functions that are not declared in
1794 @item -Wmissing-noreturn
1795 Warn about functions which might be candidates for attribute @code{noreturn}.
1796 Note these are only possible candidates, not absolute ones. Care should
1797 be taken to manually verify functions actually do not ever return before
1798 adding the @code{noreturn} attribute, otherwise subtle code generation
1799 bugs could be introduced.
1802 Warn if a structure is given the packed attribute, but the packed
1803 attribute has no effect on the layout or size of the structure.
1804 Such structures may be mis-aligned for little benefit. For
1805 instance, in this code, the variable @code{f.x} in @code{struct bar}
1806 will be misaligned even though @code{struct bar} does not itself
1807 have the packed attribute:
1814 @} __attribute__((packed));
1823 Warn if padding is included in a structure, either to align an element
1824 of the structure or to align the whole structure. Sometimes when this
1825 happens it is possible to rearrange the fields of the structure to
1826 reduce the padding and so make the structure smaller.
1828 @item -Wredundant-decls
1829 Warn if anything is declared more than once in the same scope, even in
1830 cases where multiple declaration is valid and changes nothing.
1832 @item -Wnested-externs (C only)
1833 Warn if an @code{extern} declaration is encountered within a function.
1835 @item -Wunreachable-code
1836 Warn if the compiler detects that code will never be executed.
1838 This option is intended to warn when the compiler detects that at
1839 least a whole line of source code will never be executed, because
1840 some condition is never satisfied or because it is after a
1841 procedure that never returns.
1843 It is possible for this option to produce a warning even though there
1844 are circumstances under which part of the affected line can be executed,
1845 so care should be taken when removing apparently-unreachable code.
1847 For instance, when a function is inlined, a warning may mean that the
1848 line is unreachable in only one inlined copy of the function.
1850 This option is not made part of @samp{-Wall} because in a debugging
1851 version of a program there is often substantial code which checks
1852 correct functioning of the program and is, hopefully, unreachable
1853 because the program does work. Another common use of unreachable
1854 code is to provide behaviour which is selectable at compile-time.
1857 Warn if a function can not be inlined and it was declared as inline.
1860 Warn if @samp{long long} type is used. This is default. To inhibit
1861 the warning messages, use @samp{-Wno-long-long}. Flags
1862 @samp{-Wlong-long} and @samp{-Wno-long-long} are taken into account
1863 only when @samp{-pedantic} flag is used.
1866 Make all warnings into errors.
1869 @node Debugging Options
1870 @section Options for Debugging Your Program or GCC
1871 @cindex options, debugging
1872 @cindex debugging information options
1874 GCC has various special options that are used for debugging
1875 either your program or GCC:
1879 Produce debugging information in the operating system's native format
1880 (stabs, COFF, XCOFF, or DWARF). GDB can work with this debugging
1883 On most systems that use stabs format, @samp{-g} enables use of extra
1884 debugging information that only GDB can use; this extra information
1885 makes debugging work better in GDB but will probably make other debuggers
1887 refuse to read the program. If you want to control for certain whether
1888 to generate the extra information, use @samp{-gstabs+}, @samp{-gstabs},
1889 @samp{-gxcoff+}, @samp{-gxcoff}, @samp{-gdwarf-1+}, or @samp{-gdwarf-1}
1892 Unlike most other C compilers, GCC allows you to use @samp{-g} with
1893 @samp{-O}. The shortcuts taken by optimized code may occasionally
1894 produce surprising results: some variables you declared may not exist
1895 at all; flow of control may briefly move where you did not expect it;
1896 some statements may not be executed because they compute constant
1897 results or their values were already at hand; some statements may
1898 execute in different places because they were moved out of loops.
1900 Nevertheless it proves possible to debug optimized output. This makes
1901 it reasonable to use the optimizer for programs that might have bugs.
1903 The following options are useful when GCC is generated with the
1904 capability for more than one debugging format.
1907 Produce debugging information for use by GDB. This means to use the
1908 most expressive format available (DWARF 2, stabs, or the native format
1909 if neither of those are supported), including GDB extensions if at all
1913 Produce debugging information in stabs format (if that is supported),
1914 without GDB extensions. This is the format used by DBX on most BSD
1915 systems. On MIPS, Alpha and System V Release 4 systems this option
1916 produces stabs debugging output which is not understood by DBX or SDB.
1917 On System V Release 4 systems this option requires the GNU assembler.
1920 Produce debugging information in stabs format (if that is supported),
1921 using GNU extensions understood only by the GNU debugger (GDB). The
1922 use of these extensions is likely to make other debuggers crash or
1923 refuse to read the program.
1926 Produce debugging information in COFF format (if that is supported).
1927 This is the format used by SDB on most System V systems prior to
1931 Produce debugging information in XCOFF format (if that is supported).
1932 This is the format used by the DBX debugger on IBM RS/6000 systems.
1935 Produce debugging information in XCOFF format (if that is supported),
1936 using GNU extensions understood only by the GNU debugger (GDB). The
1937 use of these extensions is likely to make other debuggers crash or
1938 refuse to read the program, and may cause assemblers other than the GNU
1939 assembler (GAS) to fail with an error.
1942 Produce debugging information in DWARF version 1 format (if that is
1943 supported). This is the format used by SDB on most System V Release 4
1947 Produce debugging information in DWARF version 1 format (if that is
1948 supported), using GNU extensions understood only by the GNU debugger
1949 (GDB). The use of these extensions is likely to make other debuggers
1950 crash or refuse to read the program.
1953 Produce debugging information in DWARF version 2 format (if that is
1954 supported). This is the format used by DBX on IRIX 6.
1957 @itemx -ggdb@var{level}
1958 @itemx -gstabs@var{level}
1959 @itemx -gcoff@var{level}
1960 @itemx -gxcoff@var{level}
1961 @itemx -gdwarf@var{level}
1962 @itemx -gdwarf-2@var{level}
1963 Request debugging information and also use @var{level} to specify how
1964 much information. The default level is 2.
1966 Level 1 produces minimal information, enough for making backtraces in
1967 parts of the program that you don't plan to debug. This includes
1968 descriptions of functions and external variables, but no information
1969 about local variables and no line numbers.
1971 Level 3 includes extra information, such as all the macro definitions
1972 present in the program. Some debuggers support macro expansion when
1977 Generate extra code to write profile information suitable for the
1978 analysis program @code{prof}. You must use this option when compiling
1979 the source files you want data about, and you must also use it when
1982 @cindex @code{gprof}
1984 Generate extra code to write profile information suitable for the
1985 analysis program @code{gprof}. You must use this option when compiling
1986 the source files you want data about, and you must also use it when
1991 Generate extra code to write profile information for basic blocks, which will
1992 record the number of times each basic block is executed, the basic block start
1993 address, and the function name containing the basic block. If @samp{-g} is
1994 used, the line number and filename of the start of the basic block will also be
1995 recorded. If not overridden by the machine description, the default action is
1996 to append to the text file @file{bb.out}.
1998 This data could be analyzed by a program like @code{tcov}. Note,
1999 however, that the format of the data is not what @code{tcov} expects.
2000 Eventually GNU @code{gprof} should be extended to process this data.
2003 Makes the compiler print out each function name as it is compiled, and
2004 print some statistics about each pass when it finishes.
2007 Generate extra code to profile basic blocks. Your executable will
2008 produce output that is a superset of that produced when @samp{-a} is
2009 used. Additional output is the source and target address of the basic
2010 blocks where a jump takes place, the number of times a jump is executed,
2011 and (optionally) the complete sequence of basic blocks being executed.
2012 The output is appended to file @file{bb.out}.
2014 You can examine different profiling aspects without recompilation. Your
2015 executable will read a list of function names from file @file{bb.in}.
2016 Profiling starts when a function on the list is entered and stops when
2017 that invocation is exited. To exclude a function from profiling, prefix
2018 its name with `-'. If a function name is not unique, you can
2019 disambiguate it by writing it in the form
2020 @samp{/path/filename.d:functionname}. Your executable will write the
2021 available paths and filenames in file @file{bb.out}.
2023 Several function names have a special meaning:
2026 Write source, target and frequency of jumps to file @file{bb.out}.
2027 @item __bb_hidecall__
2028 Exclude function calls from frequency count.
2029 @item __bb_showret__
2030 Include function returns in frequency count.
2032 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
2033 The file will be compressed using the program @samp{gzip}, which must
2034 exist in your @code{PATH}. On systems without the @samp{popen}
2035 function, the file will be named @file{bbtrace} and will not be
2036 compressed. @strong{Profiling for even a few seconds on these systems
2037 will produce a very large file.} Note: @code{__bb_hidecall__} and
2038 @code{__bb_showret__} will not affect the sequence written to
2042 Here's a short example using different profiling parameters
2043 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
2044 1 and 2 and is called twice from block 3 of function @code{main}. After
2045 the calls, block 3 transfers control to block 4 of @code{main}.
2047 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
2048 the following sequence of blocks is written to file @file{bbtrace.gz}:
2049 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
2050 the return is to a point inside the block and not to the top. The
2051 block address 0 always indicates, that control is transferred
2052 to the trace from somewhere outside the observed functions. With
2053 @samp{-foo} added to @file{bb.in}, the blocks of function
2054 @code{foo} are removed from the trace, so only 0 3 4 remains.
2056 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
2057 jump frequencies will be written to file @file{bb.out}. The
2058 frequencies are obtained by constructing a trace of blocks
2059 and incrementing a counter for every neighbouring pair of blocks
2060 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2064 Jump from block 0x0 to block 0x3 executed 1 time(s)
2065 Jump from block 0x3 to block 0x1 executed 1 time(s)
2066 Jump from block 0x1 to block 0x2 executed 2 time(s)
2067 Jump from block 0x2 to block 0x1 executed 1 time(s)
2068 Jump from block 0x2 to block 0x4 executed 1 time(s)
2071 With @code{__bb_hidecall__}, control transfer due to call instructions
2072 is removed from the trace, that is the trace is cut into three parts: 0
2073 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2074 to return instructions is added to the trace. The trace becomes: 0 3 1
2075 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2076 written to @file{bbtrace.gz}. It is solely used for counting jump
2079 @item -fprofile-arcs
2080 Instrument @dfn{arcs} during compilation. For each function of your
2081 program, GCC creates a program flow graph, then finds a spanning tree
2082 for the graph. Only arcs that are not on the spanning tree have to be
2083 instrumented: the compiler adds code to count the number of times that these
2084 arcs are executed. When an arc is the only exit or only entrance to a
2085 block, the instrumentation code can be added to the block; otherwise, a
2086 new basic block must be created to hold the instrumentation code.
2088 Since not every arc in the program must be instrumented, programs
2089 compiled with this option run faster than programs compiled with
2090 @samp{-a}, which adds instrumentation code to every basic block in the
2091 program. The tradeoff: since @code{gcov} does not have
2092 execution counts for all branches, it must start with the execution
2093 counts for the instrumented branches, and then iterate over the program
2094 flow graph until the entire graph has been solved. Hence, @code{gcov}
2095 runs a little more slowly than a program which uses information from
2098 @samp{-fprofile-arcs} also makes it possible to estimate branch
2099 probabilities, and to calculate basic block execution counts. In
2100 general, basic block execution counts do not give enough information to
2101 estimate all branch probabilities. When the compiled program exits, it
2102 saves the arc execution counts to a file called
2103 @file{@var{sourcename}.da}. Use the compiler option
2104 @samp{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2105 Control Optimization}) when recompiling, to optimize using estimated
2106 branch probabilities.
2109 @item -ftest-coverage
2110 Create data files for the @code{gcov} code-coverage utility
2111 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2112 The data file names begin with the name of your source file:
2115 @item @var{sourcename}.bb
2116 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2117 associate basic block execution counts with line numbers.
2119 @item @var{sourcename}.bbg
2120 A list of all arcs in the program flow graph. This allows @code{gcov}
2121 to reconstruct the program flow graph, so that it can compute all basic
2122 block and arc execution counts from the information in the
2123 @code{@var{sourcename}.da} file (this last file is the output from
2124 @samp{-fprofile-arcs}).
2127 @item -d@var{letters}
2128 Says to make debugging dumps during compilation at times specified by
2129 @var{letters}. This is used for debugging the compiler. The file names
2130 for most of the dumps are made by appending a pass number and a word to
2131 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.jump}).
2132 Here are the possible letters for use in @var{letters}, and their meanings:
2136 Annotate the assembler output with miscellaneous debugging information.
2138 Dump after computing branch probabilities, to @file{@var{file}.07.bp}.
2140 Dump after instruction combination, to the file @file{@var{file}.09.combine}.
2142 Dump after delayed branch scheduling, to @file{@var{file}.19.dbr}.
2144 Dump all macro definitions, at the end of preprocessing, in addition to
2147 Dump after purging ADDRESSOF, to @file{@var{file}.03.addressof}.
2149 Dump after flow analysis, to @file{@var{file}.08.flow}.
2151 Dump after global register allocation, to @file{@var{file}.13.greg}.
2153 Dump after GCSE, to @file{@var{file}.04.gcse}.
2155 Dump after first jump optimization, to @file{@var{file}.01.jump}.
2157 Dump after last jump optimization, to @file{@var{file}.17.jump2}.
2159 Dump after conversion from registers to stack, to @file{@var{file}.20.stack}.
2161 Dump after local register allocation, to @file{@var{file}.12.lreg}.
2163 Dump after loop optimization, to @file{@var{file}.05.loop}.
2165 Dump after performing the machine dependent reorganisation pass, to
2166 @file{@var{file}.18.mach}.
2168 Dump after the register move pass, to @file{@var{file}.10.regmove}.
2170 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2172 Dump after the second instruction scheduling pass, to
2173 @file{@var{file}.16.sched2}.
2175 Dump after CSE (including the jump optimization that sometimes follows
2176 CSE), to @file{@var{file}.02.cse}.
2178 Dump after the first instruction scheduling pass, to
2179 @file{@var{file}.11.sched}.
2181 Dump after the second CSE pass (including the jump optimization that
2182 sometimes follows CSE), to @file{@var{file}.06.cse2}.
2184 Produce all the dumps listed above.
2186 Print statistics on memory usage, at the end of the run, to
2189 Annotate the assembler output with a comment indicating which
2190 pattern and alternative was used. The length of each instruction is
2193 For each of the other indicated dump files (except for
2194 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
2195 suitible for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
2197 Dump after the second flow pass to @file{@var{file}.14.flow2}.
2199 Just generate RTL for a function instead of compiling it. Usually used
2202 Dump debugging information during parsing, to standard error.
2204 Dump after the peephole2 pass to @file{@var{file}.15.peephole2}.
2207 @item -fdump-unnumbered
2208 When doing debugging dumps (see -d option above), suppress instruction
2209 numbers and line number note output. This makes it more feasible to
2210 use diff on debugging dumps for compiler invokations with different
2211 options, in particular with and without -g.
2213 @item -fdump-translation-unit-@var{file} (C++ only)
2214 Dump a representation of the tree structure for the entire translation
2217 @item -fpretend-float
2218 When running a cross-compiler, pretend that the target machine uses the
2219 same floating point format as the host machine. This causes incorrect
2220 output of the actual floating constants, but the actual instruction
2221 sequence will probably be the same as GCC would make when running on
2225 Store the usual ``temporary'' intermediate files permanently; place them
2226 in the current directory and name them based on the source file. Thus,
2227 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
2228 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}.
2231 Report the CPU time taken by each subprocess in the compilation
2232 sequence. For C source files, this is the preprocessor, compiler
2233 proper, and assembler. The output looks like this:
2241 The first number on each line is the ``user time,'' that is time spent
2242 executing the program itself. The second number is ``system time,''
2243 time spent executing operating system routines on behalf of the program.
2244 Both numbers are in seconds.
2246 @item -print-file-name=@var{library}
2247 Print the full absolute name of the library file @var{library} that
2248 would be used when linking---and don't do anything else. With this
2249 option, GCC does not compile or link anything; it just prints the
2252 @item -print-prog-name=@var{program}
2253 Like @samp{-print-file-name}, but searches for a program such as @samp{cpp}.
2255 @item -print-libgcc-file-name
2256 Same as @samp{-print-file-name=libgcc.a}.
2258 This is useful when you use @samp{-nostdlib} or @samp{-nodefaultlibs}
2259 but you do want to link with @file{libgcc.a}. You can do
2262 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
2265 @item -print-search-dirs
2266 Print the name of the configured installation directory and a list of
2267 program and library directories gcc will search---and don't do anything else.
2269 This is useful when gcc prints the error message
2270 @samp{installation problem, cannot exec cpp: No such file or directory}.
2271 To resolve this you either need to put @file{cpp} and the other compiler
2272 components where gcc expects to find them, or you can set the environment
2273 variable @code{GCC_EXEC_PREFIX} to the directory where you installed them.
2274 Don't forget the trailing '/'.
2275 @xref{Environment Variables}.
2278 @node Optimize Options
2279 @section Options That Control Optimization
2280 @cindex optimize options
2281 @cindex options, optimization
2283 These options control various sorts of optimizations:
2288 Optimize. Optimizing compilation takes somewhat more time, and a lot
2289 more memory for a large function.
2291 Without @samp{-O}, the compiler's goal is to reduce the cost of
2292 compilation and to make debugging produce the expected results.
2293 Statements are independent: if you stop the program with a breakpoint
2294 between statements, you can then assign a new value to any variable or
2295 change the program counter to any other statement in the function and
2296 get exactly the results you would expect from the source code.
2298 Without @samp{-O}, the compiler only allocates variables declared
2299 @code{register} in registers. The resulting compiled code is a little
2300 worse than produced by PCC without @samp{-O}.
2302 With @samp{-O}, the compiler tries to reduce code size and execution
2305 When you specify @samp{-O}, the compiler turns on @samp{-fthread-jumps}
2306 and @samp{-fdefer-pop} on all machines. The compiler turns on
2307 @samp{-fdelayed-branch} on machines that have delay slots, and
2308 @samp{-fomit-frame-pointer} on machines that can support debugging even
2309 without a frame pointer. On some machines the compiler also turns
2310 on other flags.@refill
2313 Optimize even more. GCC performs nearly all supported optimizations
2314 that do not involve a space-speed tradeoff. The compiler does not
2315 perform loop unrolling or function inlining when you specify @samp{-O2}.
2316 As compared to @samp{-O}, this option increases both compilation time
2317 and the performance of the generated code.
2319 @samp{-O2} turns on all optional optimizations except for loop unrolling
2320 and function inlining. It also turns on the @samp{-fforce-mem} option
2321 on all machines and frame pointer elimination on machines where doing so
2322 does not interfere with debugging.
2325 Optimize yet more. @samp{-O3} turns on all optimizations specified by
2326 @samp{-O2} and also turns on the @samp{inline-functions} option.
2332 Optimize for size. @samp{-Os} enables all @samp{-O2} optimizations that
2333 do not typically increase code size. It also performs further
2334 optimizations designed to reduce code size.
2336 If you use multiple @samp{-O} options, with or without level numbers,
2337 the last such option is the one that is effective.
2340 Options of the form @samp{-f@var{flag}} specify machine-independent
2341 flags. Most flags have both positive and negative forms; the negative
2342 form of @samp{-ffoo} would be @samp{-fno-foo}. In the table below,
2343 only one of the forms is listed---the one which is not the default.
2344 You can figure out the other form by either removing @samp{no-} or
2349 Do not store floating point variables in registers, and inhibit other
2350 options that might change whether a floating point value is taken from a
2353 @cindex floating point precision
2354 This option prevents undesirable excess precision on machines such as
2355 the 68000 where the floating registers (of the 68881) keep more
2356 precision than a @code{double} is supposed to have. Similarly for the
2357 x86 architecture. For most programs, the excess precision does only
2358 good, but a few programs rely on the precise definition of IEEE floating
2359 point. Use @samp{-ffloat-store} for such programs, after modifying
2360 them to store all pertinent intermediate computations into variables.
2362 @item -fno-default-inline
2363 Do not make member functions inline by default merely because they are
2364 defined inside the class scope (C++ only). Otherwise, when you specify
2365 @w{@samp{-O}}, member functions defined inside class scope are compiled
2366 inline by default; i.e., you don't need to add @samp{inline} in front of
2367 the member function name.
2369 @item -fno-defer-pop
2370 Always pop the arguments to each function call as soon as that function
2371 returns. For machines which must pop arguments after a function call,
2372 the compiler normally lets arguments accumulate on the stack for several
2373 function calls and pops them all at once.
2376 Force memory operands to be copied into registers before doing
2377 arithmetic on them. This produces better code by making all memory
2378 references potential common subexpressions. When they are not common
2379 subexpressions, instruction combination should eliminate the separate
2380 register-load. The @samp{-O2} option turns on this option.
2383 Force memory address constants to be copied into registers before
2384 doing arithmetic on them. This may produce better code just as
2385 @samp{-fforce-mem} may.
2387 @item -fomit-frame-pointer
2388 Don't keep the frame pointer in a register for functions that
2389 don't need one. This avoids the instructions to save, set up and
2390 restore frame pointers; it also makes an extra register available
2391 in many functions. @strong{It also makes debugging impossible on
2395 On some machines, such as the Vax, this flag has no effect, because
2396 the standard calling sequence automatically handles the frame pointer
2397 and nothing is saved by pretending it doesn't exist. The
2398 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2399 whether a target machine supports this flag. @xref{Registers}.@refill
2402 On some machines, such as the Vax, this flag has no effect, because
2403 the standard calling sequence automatically handles the frame pointer
2404 and nothing is saved by pretending it doesn't exist. The
2405 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2406 whether a target machine supports this flag. @xref{Registers,,Register
2407 Usage, gcc.info, Using and Porting GCC}.@refill
2411 Don't pay attention to the @code{inline} keyword. Normally this option
2412 is used to keep the compiler from expanding any functions inline.
2413 Note that if you are not optimizing, no functions can be expanded inline.
2415 @item -finline-functions
2416 Integrate all simple functions into their callers. The compiler
2417 heuristically decides which functions are simple enough to be worth
2418 integrating in this way.
2420 If all calls to a given function are integrated, and the function is
2421 declared @code{static}, then the function is normally not output as
2422 assembler code in its own right.
2424 @item -finline-limit=@var{n}
2425 By default, gcc limits the size of functions that can be inlined. This flag
2426 allows the control of this limit for functions that are explicitly marked as
2427 inline (ie marked with the inline keyword or defined within the class
2428 definition in c++). @var{n} is the size of functions that can be inlined in
2429 number of pseudo instructions (not counting parameter handling). The default
2430 value of n is 10000. Increasing this value can result in more inlined code at
2431 the cost of compilation time and memory consumption. Decreasing usually makes
2432 the compilation faster and less code will be inlined (which presumably
2433 means slower programs). This option is particularly useful for programs that
2434 use inlining heavily such as those based on recursive templates with c++.
2436 @emph{Note:} pseudo instruction represents, in this particular context, an
2437 abstract measurement of function's size. In no way, it represents a count
2438 of assembly instructions and as such its exact meaning might change from one
2439 release to an another.
2441 @item -fkeep-inline-functions
2442 Even if all calls to a given function are integrated, and the function
2443 is declared @code{static}, nevertheless output a separate run-time
2444 callable version of the function. This switch does not affect
2445 @code{extern inline} functions.
2447 @item -fkeep-static-consts
2448 Emit variables declared @code{static const} when optimization isn't turned
2449 on, even if the variables aren't referenced.
2451 GCC enables this option by default. If you want to force the compiler to
2452 check if the variable was referenced, regardless of whether or not
2453 optimization is turned on, use the @samp{-fno-keep-static-consts} option.
2455 @item -fno-function-cse
2456 Do not put function addresses in registers; make each instruction that
2457 calls a constant function contain the function's address explicitly.
2459 This option results in less efficient code, but some strange hacks
2460 that alter the assembler output may be confused by the optimizations
2461 performed when this option is not used.
2464 This option allows GCC to violate some ANSI or IEEE rules and/or
2465 specifications in the interest of optimizing code for speed. For
2466 example, it allows the compiler to assume arguments to the @code{sqrt}
2467 function are non-negative numbers and that no floating-point values
2470 This option should never be turned on by any @samp{-O} option since
2471 it can result in incorrect output for programs which depend on
2472 an exact implementation of IEEE or ANSI rules/specifications for
2475 @item -fno-math-errno
2476 Do not set ERRNO after calling math functions that are executed
2477 with a single instruction, e.g., sqrt. A program that relies on
2478 IEEE exceptions for math error handling may want to use this flag
2479 for speed while maintaining IEEE arithmetic compatibility.
2481 The default is @samp{-fmath-errno}. The @samp{-ffast-math} option
2482 sets @samp{-fno-math-errno}.
2485 @c following causes underfulls.. they don't look great, but we deal.
2487 The following options control specific optimizations. The @samp{-O2}
2488 option turns on all of these optimizations except @samp{-funroll-loops}
2489 and @samp{-funroll-all-loops}. On most machines, the @samp{-O} option
2490 turns on the @samp{-fthread-jumps} and @samp{-fdelayed-branch} options,
2491 but specific machines may handle it differently.
2493 You can use the following flags in the rare cases when ``fine-tuning''
2494 of optimizations to be performed is desired.
2497 @item -fstrength-reduce
2498 Perform the optimizations of loop strength reduction and
2499 elimination of iteration variables.
2501 @item -fthread-jumps
2502 Perform optimizations where we check to see if a jump branches to a
2503 location where another comparison subsumed by the first is found. If
2504 so, the first branch is redirected to either the destination of the
2505 second branch or a point immediately following it, depending on whether
2506 the condition is known to be true or false.
2508 @item -fcse-follow-jumps
2509 In common subexpression elimination, scan through jump instructions
2510 when the target of the jump is not reached by any other path. For
2511 example, when CSE encounters an @code{if} statement with an
2512 @code{else} clause, CSE will follow the jump when the condition
2515 @item -fcse-skip-blocks
2516 This is similar to @samp{-fcse-follow-jumps}, but causes CSE to
2517 follow jumps which conditionally skip over blocks. When CSE
2518 encounters a simple @code{if} statement with no else clause,
2519 @samp{-fcse-skip-blocks} causes CSE to follow the jump around the
2520 body of the @code{if}.
2522 @item -frerun-cse-after-loop
2523 Re-run common subexpression elimination after loop optimizations has been
2526 @item -frerun-loop-opt
2527 Run the loop optimizer twice.
2530 Perform a global common subexpression elimination pass.
2531 This pass also performs global constant and copy propagation.
2533 @item -fdelete-null-pointer-checks
2534 Use global dataflow analysis to identify and eliminate useless null
2535 pointer checks. Programs which rely on NULL pointer dereferences @emph{not}
2536 halting the program may not work properly with this option. Use
2537 -fno-delete-null-pointer-checks to disable this optimizing for programs
2538 which depend on that behavior.
2541 @item -fexpensive-optimizations
2542 Perform a number of minor optimizations that are relatively expensive.
2544 @item -foptimize-register-moves
2546 Attempt to reassign register numbers in move instructions and as
2547 operands of other simple instructions in order to maximize the amount of
2548 register tying. This is especially helpful on machines with two-operand
2549 instructions. GCC enables this optimization by default with @samp{-O2}
2552 Note @code{-fregmove} and @code{-foptimize-register-moves} are the same
2555 @item -fdelayed-branch
2556 If supported for the target machine, attempt to reorder instructions
2557 to exploit instruction slots available after delayed branch
2560 @item -fschedule-insns
2561 If supported for the target machine, attempt to reorder instructions to
2562 eliminate execution stalls due to required data being unavailable. This
2563 helps machines that have slow floating point or memory load instructions
2564 by allowing other instructions to be issued until the result of the load
2565 or floating point instruction is required.
2567 @item -fschedule-insns2
2568 Similar to @samp{-fschedule-insns}, but requests an additional pass of
2569 instruction scheduling after register allocation has been done. This is
2570 especially useful on machines with a relatively small number of
2571 registers and where memory load instructions take more than one cycle.
2573 @item -ffunction-sections
2574 @itemx -fdata-sections
2575 Place each function or data item into its own section in the output
2576 file if the target supports arbitrary sections. The name of the
2577 function or the name of the data item determines the section's name
2580 Use these options on systems where the linker can perform optimizations
2581 to improve locality of reference in the instruction space. HPPA
2582 processors running HP-UX and Sparc processors running Solaris 2 have
2583 linkers with such optimizations. Other systems using the ELF object format
2584 as well as AIX may have these optimizations in the future.
2586 Only use these options when there are significant benefits from doing
2587 so. When you specify these options, the assembler and linker will
2588 create larger object and executable files and will also be slower.
2589 You will not be able to use @code{gprof} on all systems if you
2590 specify this option and you may have problems with debugging if
2591 you specify both this option and @samp{-g}.
2593 @item -fcaller-saves
2594 Enable values to be allocated in registers that will be clobbered by
2595 function calls, by emitting extra instructions to save and restore the
2596 registers around such calls. Such allocation is done only when it
2597 seems to result in better code than would otherwise be produced.
2599 This option is always enabled by default on certain machines, usually
2600 those which have no call-preserved registers to use instead.
2602 For all machines, optimization level 2 and higher enables this flag by
2605 @item -funroll-loops
2606 Perform the optimization of loop unrolling. This is only done for loops
2607 whose number of iterations can be determined at compile time or run time.
2608 @samp{-funroll-loops} implies both @samp{-fstrength-reduce} and
2609 @samp{-frerun-cse-after-loop}.
2611 @item -funroll-all-loops
2612 Perform the optimization of loop unrolling. This is done for all loops
2613 and usually makes programs run more slowly. @samp{-funroll-all-loops}
2614 implies @samp{-fstrength-reduce} as well as @samp{-frerun-cse-after-loop}.
2616 @item -fmove-all-movables
2617 Forces all invariant computations in loops to be moved
2620 @item -freduce-all-givs
2621 Forces all general-induction variables in loops to be
2624 @emph{Note:} When compiling programs written in Fortran,
2625 @samp{-fmove-all-movables} and @samp{-freduce-all-givs} are enabled
2626 by default when you use the optimizer.
2628 These options may generate better or worse code; results are highly
2629 dependent on the structure of loops within the source code.
2631 These two options are intended to be removed someday, once
2632 they have helped determine the efficacy of various
2633 approaches to improving loop optimizations.
2635 Please let us (@code{gcc@@gcc.gnu.org} and @code{fortran@@gnu.org})
2636 know how use of these options affects
2637 the performance of your production code.
2638 We're very interested in code that runs @emph{slower}
2639 when these options are @emph{enabled}.
2642 Disable any machine-specific peephole optimizations.
2644 @item -fbranch-probabilities
2645 After running a program compiled with @samp{-fprofile-arcs}
2646 (@pxref{Debugging Options,, Options for Debugging Your Program or
2647 @code{gcc}}), you can compile it a second time using
2648 @samp{-fbranch-probabilities}, to improve optimizations based on
2649 guessing the path a branch might take.
2652 With @samp{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
2653 note on the first instruction of each basic block, and a
2654 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
2655 These can be used to improve optimization. Currently, they are only
2656 used in one place: in @file{reorg.c}, instead of guessing which path a
2657 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
2658 exactly determine which path is taken more often.
2661 @item -fstrict-aliasing
2662 Allows the compiler to assume the strictest aliasing rules applicable to
2663 the language being compiled. For C (and C++), this activates
2664 optimizations based on the type of expressions. In particular, an
2665 object of one type is assumed never to reside at the same address as an
2666 object of a different type, unless the types are almost the same. For
2667 example, an @code{unsigned int} can alias an @code{int}, but not a
2668 @code{void*} or a @code{double}. A character type may alias any other
2671 Pay special attention to code like this:
2684 The practice of reading from a different union member than the one most
2685 recently written to (called ``type-punning'') is common. Even with
2686 @samp{-fstrict-aliasing}, type-punning is allowed, provided the memory
2687 is accessed through the union type. So, the code above will work as
2688 expected. However, this code might not:
2700 Every language that wishes to perform language-specific alias analysis
2701 should define a function that computes, given an @code{tree}
2702 node, an alias set for the node. Nodes in different alias sets are not
2703 allowed to alias. For an example, see the C front-end function
2704 @code{c_get_alias_set}.
2707 @item -falign-functions
2708 @itemx -falign-functions=@var{n}
2709 Align the start of functions to the next power-of-two greater than
2710 @var{n}, skipping up to @var{n} bytes. For instance,
2711 @samp{-falign-functions=32} aligns functions to the next 32-byte
2712 boundary, but @samp{-falign-functions=24} would align to the next
2713 32-byte boundary only if this can be done by skipping 23 bytes or less.
2715 @samp{-fno-align-functions} and @samp{-falign-functions=1} are
2716 equivalent and mean that functions will not be aligned.
2718 Some assemblers only support this flag when @var{n} is a power of two;
2719 in that case, it is rounded up.
2721 If @var{n} is not specified, use a machine-dependent default.
2723 @item -falign-labels
2724 @itemx -falign-labels=@var{n}
2725 Align all branch targets to a power-of-two boundary, skipping up to
2726 @var{n} bytes like @samp{-falign-functions}. This option can easily
2727 make code slower, because it must insert dummy operations for when the
2728 branch target is reached in the usual flow of the code.
2730 If @samp{-falign-loops} or @samp{-falign-jumps} are applicable and
2731 are greater than this value, then their values are used instead.
2733 If @var{n} is not specified, use a machine-dependent default which is
2734 very likely to be @samp{1}, meaning no alignment.
2737 @itemx -falign-loops=@var{n}
2738 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
2739 like @samp{-falign-functions}. The hope is that the loop will be
2740 executed many times, which will make up for any execution of the dummy
2743 If @var{n} is not specified, use a machine-dependent default.
2746 @itemx -falign-jumps=@var{n}
2747 Align branch targets to a power-of-two boundary, for branch targets
2748 where the targets can only be reached by jumping, skipping up to @var{n}
2749 bytes like @samp{-falign-functions}. In this case, no dummy operations
2752 If @var{n} is not specified, use a machine-dependent default.
2756 @node Preprocessor Options
2757 @section Options Controlling the Preprocessor
2758 @cindex preprocessor options
2759 @cindex options, preprocessor
2761 These options control the C preprocessor, which is run on each C source
2762 file before actual compilation.
2764 If you use the @samp{-E} option, nothing is done except preprocessing.
2765 Some of these options make sense only together with @samp{-E} because
2766 they cause the preprocessor output to be unsuitable for actual
2770 @item -include @var{file}
2771 Process @var{file} as input before processing the regular input file.
2772 In effect, the contents of @var{file} are compiled first. Any @samp{-D}
2773 and @samp{-U} options on the command line are always processed before
2774 @samp{-include @var{file}}, regardless of the order in which they are
2775 written. All the @samp{-include} and @samp{-imacros} options are
2776 processed in the order in which they are written.
2778 @item -imacros @var{file}
2779 Process @var{file} as input, discarding the resulting output, before
2780 processing the regular input file. Because the output generated from
2781 @var{file} is discarded, the only effect of @samp{-imacros @var{file}}
2782 is to make the macros defined in @var{file} available for use in the
2785 Any @samp{-D} and @samp{-U} options on the command line are always
2786 processed before @samp{-imacros @var{file}}, regardless of the order in
2787 which they are written. All the @samp{-include} and @samp{-imacros}
2788 options are processed in the order in which they are written.
2790 @item -idirafter @var{dir}
2791 @cindex second include path
2792 Add the directory @var{dir} to the second include path. The directories
2793 on the second include path are searched when a header file is not found
2794 in any of the directories in the main include path (the one that
2797 @item -iprefix @var{prefix}
2798 Specify @var{prefix} as the prefix for subsequent @samp{-iwithprefix}
2801 @item -iwithprefix @var{dir}
2802 Add a directory to the second include path. The directory's name is
2803 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
2804 specified previously with @samp{-iprefix}. If you have not specified a
2805 prefix yet, the directory containing the installed passes of the
2806 compiler is used as the default.
2808 @item -iwithprefixbefore @var{dir}
2809 Add a directory to the main include path. The directory's name is made
2810 by concatenating @var{prefix} and @var{dir}, as in the case of
2811 @samp{-iwithprefix}.
2813 @item -isystem @var{dir}
2814 Add a directory to the beginning of the second include path, marking it
2815 as a system directory, so that it gets the same special treatment as
2816 is applied to the standard system directories.
2819 Do not search the standard system directories for header files. Only
2820 the directories you have specified with @samp{-I} options (and the
2821 current directory, if appropriate) are searched. @xref{Directory
2822 Options}, for information on @samp{-I}.
2824 By using both @samp{-nostdinc} and @samp{-I-}, you can limit the include-file
2825 search path to only those directories you specify explicitly.
2828 Do not predefine any nonstandard macros. (Including architecture flags).
2831 Run only the C preprocessor. Preprocess all the C source files
2832 specified and output the results to standard output or to the
2833 specified output file.
2836 Tell the preprocessor not to discard comments. Used with the
2840 Tell the preprocessor not to generate @samp{#line} directives.
2841 Used with the @samp{-E} option.
2844 @cindex dependencies, make
2846 Tell the preprocessor to output a rule suitable for @code{make}
2847 describing the dependencies of each object file. For each source file,
2848 the preprocessor outputs one @code{make}-rule whose target is the object
2849 file name for that source file and whose dependencies are all the
2850 @code{#include} header files it uses. This rule may be a single line or
2851 may be continued with @samp{\}-newline if it is long. The list of rules
2852 is printed on standard output instead of the preprocessed C program.
2854 @samp{-M} implies @samp{-E}.
2856 Another way to specify output of a @code{make} rule is by setting
2857 the environment variable @code{DEPENDENCIES_OUTPUT} (@pxref{Environment
2861 Like @samp{-M} but the output mentions only the user header files
2862 included with @samp{#include "@var{file}"}. System header files
2863 included with @samp{#include <@var{file}>} are omitted.
2866 Like @samp{-M} but the dependency information is written to a file made by
2867 replacing ".c" with ".d" at the end of the input file names.
2868 This is in addition to compiling the file as specified---@samp{-MD} does
2869 not inhibit ordinary compilation the way @samp{-M} does.
2871 In Mach, you can use the utility @code{md} to merge multiple dependency
2872 files into a single dependency file suitable for using with the @samp{make}
2876 Like @samp{-MD} except mention only user header files, not system
2880 Treat missing header files as generated files and assume they live in the
2881 same directory as the source file. If you specify @samp{-MG}, you
2882 must also specify either @samp{-M} or @samp{-MM}. @samp{-MG} is not
2883 supported with @samp{-MD} or @samp{-MMD}.
2886 Print the name of each header file used, in addition to other normal
2889 @item -A@var{question}(@var{answer})
2890 Assert the answer @var{answer} for @var{question}, in case it is tested
2891 with a preprocessing conditional such as @samp{#if
2892 #@var{question}(@var{answer})}. @samp{-A-} disables the standard
2893 assertions that normally describe the target machine.
2896 Define macro @var{macro} with the string @samp{1} as its definition.
2898 @item -D@var{macro}=@var{defn}
2899 Define macro @var{macro} as @var{defn}. All instances of @samp{-D} on
2900 the command line are processed before any @samp{-U} options.
2903 Undefine macro @var{macro}. @samp{-U} options are evaluated after all
2904 @samp{-D} options, but before any @samp{-include} and @samp{-imacros}
2908 Tell the preprocessor to output only a list of the macro definitions
2909 that are in effect at the end of preprocessing. Used with the @samp{-E}
2913 Tell the preprocessing to pass all macro definitions into the output, in
2914 their proper sequence in the rest of the output.
2917 Like @samp{-dD} except that the macro arguments and contents are omitted.
2918 Only @samp{#define @var{name}} is included in the output.
2921 Support ANSI C trigraphs. The @samp{-ansi} option also has this effect.
2923 @item -Wp,@var{option}
2924 Pass @var{option} as an option to the preprocessor. If @var{option}
2925 contains commas, it is split into multiple options at the commas.
2928 @node Assembler Options
2929 @section Passing Options to the Assembler
2931 @c prevent bad page break with this line
2932 You can pass options to the assembler.
2935 @item -Wa,@var{option}
2936 Pass @var{option} as an option to the assembler. If @var{option}
2937 contains commas, it is split into multiple options at the commas.
2941 @section Options for Linking
2942 @cindex link options
2943 @cindex options, linking
2945 These options come into play when the compiler links object files into
2946 an executable output file. They are meaningless if the compiler is
2947 not doing a link step.
2951 @item @var{object-file-name}
2952 A file name that does not end in a special recognized suffix is
2953 considered to name an object file or library. (Object files are
2954 distinguished from libraries by the linker according to the file
2955 contents.) If linking is done, these object files are used as input
2961 If any of these options is used, then the linker is not run, and
2962 object file names should not be used as arguments. @xref{Overall
2966 @item -l@var{library}
2967 Search the library named @var{library} when linking.
2969 It makes a difference where in the command you write this option; the
2970 linker searches processes libraries and object files in the order they
2971 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
2972 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
2973 to functions in @samp{z}, those functions may not be loaded.
2975 The linker searches a standard list of directories for the library,
2976 which is actually a file named @file{lib@var{library}.a}. The linker
2977 then uses this file as if it had been specified precisely by name.
2979 The directories searched include several standard system directories
2980 plus any that you specify with @samp{-L}.
2982 Normally the files found this way are library files---archive files
2983 whose members are object files. The linker handles an archive file by
2984 scanning through it for members which define symbols that have so far
2985 been referenced but not defined. But if the file that is found is an
2986 ordinary object file, it is linked in the usual fashion. The only
2987 difference between using an @samp{-l} option and specifying a file name
2988 is that @samp{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
2989 and searches several directories.
2992 You need this special case of the @samp{-l} option in order to
2993 link an Objective C program.
2996 Do not use the standard system startup files when linking.
2997 The standard system libraries are used normally, unless @code{-nostdlib}
2998 or @code{-nodefaultlibs} is used.
3000 @item -nodefaultlibs
3001 Do not use the standard system libraries when linking.
3002 Only the libraries you specify will be passed to the linker.
3003 The standard startup files are used normally, unless @code{-nostartfiles}
3004 is used. The compiler may generate calls to memcmp, memset, and memcpy
3005 for System V (and ANSI C) environments or to bcopy and bzero for
3006 BSD environments. These entries are usually resolved by entries in
3007 libc. These entry points should be supplied through some other
3008 mechanism when this option is specified.
3011 Do not use the standard system startup files or libraries when linking.
3012 No startup files and only the libraries you specify will be passed to
3013 the linker. The compiler may generate calls to memcmp, memset, and memcpy
3014 for System V (and ANSI C) environments or to bcopy and bzero for
3015 BSD environments. These entries are usually resolved by entries in
3016 libc. These entry points should be supplied through some other
3017 mechanism when this option is specified.
3019 @cindex @code{-lgcc}, use with @code{-nostdlib}
3020 @cindex @code{-nostdlib} and unresolved references
3021 @cindex unresolved references and @code{-nostdlib}
3022 @cindex @code{-lgcc}, use with @code{-nodefaultlibs}
3023 @cindex @code{-nodefaultlibs} and unresolved references
3024 @cindex unresolved references and @code{-nodefaultlibs}
3025 One of the standard libraries bypassed by @samp{-nostdlib} and
3026 @samp{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
3027 that GCC uses to overcome shortcomings of particular machines, or special
3028 needs for some languages.
3030 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
3034 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
3035 for more discussion of @file{libgcc.a}.)
3037 In most cases, you need @file{libgcc.a} even when you want to avoid
3038 other standard libraries. In other words, when you specify @samp{-nostdlib}
3039 or @samp{-nodefaultlibs} you should usually specify @samp{-lgcc} as well.
3040 This ensures that you have no unresolved references to internal GCC
3041 library subroutines. (For example, @samp{__main}, used to ensure C++
3042 constructors will be called; @pxref{Collect2,,@code{collect2}}.)
3045 Remove all symbol table and relocation information from the executable.
3048 On systems that support dynamic linking, this prevents linking with the shared
3049 libraries. On other systems, this option has no effect.
3052 Produce a shared object which can then be linked with other objects to
3053 form an executable. Not all systems support this option. You must
3054 also specify @samp{-fpic} or @samp{-fPIC} on some systems when
3055 you specify this option.
3058 Bind references to global symbols when building a shared object. Warn
3059 about any unresolved references (unless overridden by the link editor
3060 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
3063 @item -Xlinker @var{option}
3064 Pass @var{option} as an option to the linker. You can use this to
3065 supply system-specific linker options which GCC does not know how to
3068 If you want to pass an option that takes an argument, you must use
3069 @samp{-Xlinker} twice, once for the option and once for the argument.
3070 For example, to pass @samp{-assert definitions}, you must write
3071 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
3072 @samp{-Xlinker "-assert definitions"}, because this passes the entire
3073 string as a single argument, which is not what the linker expects.
3075 @item -Wl,@var{option}
3076 Pass @var{option} as an option to the linker. If @var{option} contains
3077 commas, it is split into multiple options at the commas.
3079 @item -u @var{symbol}
3080 Pretend the symbol @var{symbol} is undefined, to force linking of
3081 library modules to define it. You can use @samp{-u} multiple times with
3082 different symbols to force loading of additional library modules.
3085 @node Directory Options
3086 @section Options for Directory Search
3087 @cindex directory options
3088 @cindex options, directory search
3091 These options specify directories to search for header files, for
3092 libraries and for parts of the compiler:
3096 Add the directory @var{dir} to the head of the list of directories to be
3097 searched for header files. This can be used to override a system header
3098 file, substituting your own version, since these directories are
3099 searched before the system header file directories. If you use more
3100 than one @samp{-I} option, the directories are scanned in left-to-right
3101 order; the standard system directories come after.
3104 Any directories you specify with @samp{-I} options before the @samp{-I-}
3105 option are searched only for the case of @samp{#include "@var{file}"};
3106 they are not searched for @samp{#include <@var{file}>}.
3108 If additional directories are specified with @samp{-I} options after
3109 the @samp{-I-}, these directories are searched for all @samp{#include}
3110 directives. (Ordinarily @emph{all} @samp{-I} directories are used
3113 In addition, the @samp{-I-} option inhibits the use of the current
3114 directory (where the current input file came from) as the first search
3115 directory for @samp{#include "@var{file}"}. There is no way to
3116 override this effect of @samp{-I-}. With @samp{-I.} you can specify
3117 searching the directory which was current when the compiler was
3118 invoked. That is not exactly the same as what the preprocessor does
3119 by default, but it is often satisfactory.
3121 @samp{-I-} does not inhibit the use of the standard system directories
3122 for header files. Thus, @samp{-I-} and @samp{-nostdinc} are
3126 Add directory @var{dir} to the list of directories to be searched
3129 @item -B@var{prefix}
3130 This option specifies where to find the executables, libraries,
3131 include files, and data files of the compiler itself.
3133 The compiler driver program runs one or more of the subprograms
3134 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
3135 @var{prefix} as a prefix for each program it tries to run, both with and
3136 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
3138 For each subprogram to be run, the compiler driver first tries the
3139 @samp{-B} prefix, if any. If that name is not found, or if @samp{-B}
3140 was not specified, the driver tries two standard prefixes, which are
3141 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
3142 those results in a file name that is found, the unmodified program
3143 name is searched for using the directories specified in your
3144 @samp{PATH} environment variable.
3146 @samp{-B} prefixes that effectively specify directory names also apply
3147 to libraries in the linker, because the compiler translates these
3148 options into @samp{-L} options for the linker. They also apply to
3149 includes files in the preprocessor, because the compiler translates these
3150 options into @samp{-isystem} options for the preprocessor. In this case,
3151 the compiler appends @samp{include} to the prefix.
3153 The run-time support file @file{libgcc.a} can also be searched for using
3154 the @samp{-B} prefix, if needed. If it is not found there, the two
3155 standard prefixes above are tried, and that is all. The file is left
3156 out of the link if it is not found by those means.
3158 Another way to specify a prefix much like the @samp{-B} prefix is to use
3159 the environment variable @code{GCC_EXEC_PREFIX}. @xref{Environment
3162 @item -specs=@var{file}
3163 Process @var{file} after the compiler reads in the standard @file{specs}
3164 file, in order to override the defaults that the @file{gcc} driver
3165 program uses when determining what switches to pass to @file{cc1},
3166 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
3167 @samp{-specs=}@var{file} can be specified on the command line, and they
3168 are processed in order, from left to right.
3172 @section Specifying subprocesses and the switches to pass to them
3174 @code{GCC} is a driver program. It performs its job by invoking a
3175 sequence of other programs to do the work of compiling, assembling and
3176 linking. GCC interprets its command-line parameters and uses these to
3177 deduce which programs it should invoke, and which command-line options
3178 it ought to place on their command lines. This behaviour is controlled
3179 by @dfn{spec strings}. In most cases there is one spec string for each
3180 program that GCC can invoke, but a few programs have multiple spec
3181 strings to control their behaviour. The spec strings built into GCC can
3182 be overridden by using the @samp{-specs=} command-line switch to specify
3185 @dfn{Spec files} are plaintext files that are used to construct spec
3186 strings. They consist of a sequence of directives separated by blank
3187 lines. The type of directive is determined by the first non-whitespace
3188 character on the line and it can be one of the following:
3191 @item %@var{command}
3192 Issues a @var{command} to the spec file processor. The commands that can
3196 @item %include <@var{file}>
3198 Search for @var{file} and insert its text at the current point in the
3201 @item %include_noerr <@var{file}>
3202 @cindex %include_noerr
3203 Just like @samp{%include}, but do not generate an error message if the include
3204 file cannot be found.
3206 @item %rename @var{old_name} @var{new_name}
3208 Rename the spec string @var{old_name} to @var{new_name}.
3212 @item *[@var{spec_name}]:
3213 This tells the compiler to create, override or delete the named spec
3214 string. All lines after this directive up to the next directive or
3215 blank line are considered to be the text for the spec string. If this
3216 results in an empty string then the spec will be deleted. (Or, if the
3217 spec did not exist, then nothing will happened.) Otherwise, if the spec
3218 does not currently exist a new spec will be created. If the spec does
3219 exist then its contents will be overridden by the text of this
3220 directive, unless the first character of that text is the @samp{+}
3221 character, in which case the text will be appended to the spec.
3223 @item [@var{suffix}]:
3224 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
3225 and up to the next directive or blank line are considered to make up the
3226 spec string for the indicated suffix. When the compiler encounters an
3227 input file with the named suffix, it will processes the spec string in
3228 order to work out how to compile that file. For example:
3235 This says that any input file whose name ends in @samp{.ZZ} should be
3236 passed to the program @samp{z-compile}, which should be invoked with the
3237 command-line switch @samp{-input} and with the result of performing the
3238 @samp{%i} substitution. (See below.)
3240 As an alternative to providing a spec string, the text that follows a
3241 suffix directive can be one of the following:
3244 @item @@@var{language}
3245 This says that the suffix is an alias for a known @var{language}. This is
3246 similar to using the @code{-x} command-line switch to GCC to specify a
3247 language explicitly. For example:
3254 Says that .ZZ files are, in fact, C++ source files.
3257 This causes an error messages saying:
3260 @var{name} compiler not installed on this system.
3264 GCC already has an extensive list of suffixes built into it.
3265 This directive will add an entry to the end of the list of suffixes, but
3266 since the list is searched from the end backwards, it is effectively
3267 possible to override earlier entries using this technique.
3271 GCC has the following spec strings built into it. Spec files can
3272 override these strings or create their own. Note that individual
3273 targets can also add their own spec strings to this list.
3276 asm Options to pass to the assembler
3277 asm_final Options to pass to the assembler post-processor
3278 cpp Options to pass to the C preprocessor
3279 cc1 Options to pass to the C compiler
3280 cc1plus Options to pass to the C++ compiler
3281 endfile Object files to include at the end of the link
3282 link Options to pass to the linker
3283 lib Libraries to include on the command line to the linker
3284 libgcc Decides which GCC support library to pass to the linker
3285 linker Sets the name of the linker
3286 predefines Defines to be passed to the C preprocessor
3287 signed_char Defines to pass to CPP to say whether @code{char} is signed by default
3288 startfile Object files to include at the start of the link
3291 Here is a small example of a spec file:
3297 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
3300 This example renames the spec called @samp{lib} to @samp{old_lib} and
3301 then overrides the previous definition of @samp{lib} with a new one.
3302 The new definition adds in some extra command-line options before
3303 including the text of the old definition.
3305 @dfn{Spec strings} are a list of command-line options to be passed to their
3306 corresponding program. In addition, the spec strings can contain
3307 @samp{%}-prefixed sequences to substitute variable text or to
3308 conditionally insert text into the command line. Using these constructs
3309 it is possible to generate quite complex command lines.
3311 Here is a table of all defined @samp{%}-sequences for spec
3312 strings. Note that spaces are not generated automatically around the
3313 results of expanding these sequences. Therefore you can concatenate them
3314 together or combine them with constant text in a single argument.
3318 Substitute one @samp{%} into the program name or argument.
3321 Substitute the name of the input file being processed.
3324 Substitute the basename of the input file being processed.
3325 This is the substring up to (and not including) the last period
3326 and not including the directory.
3329 Marks the argument containing or following the @samp{%d} as a
3330 temporary file name, so that that file will be deleted if GCC exits
3331 successfully. Unlike @samp{%g}, this contributes no text to the
3334 @item %g@var{suffix}
3335 Substitute a file name that has suffix @var{suffix} and is chosen
3336 once per compilation, and mark the argument in the same way as
3337 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
3338 name is now chosen in a way that is hard to predict even when previously
3339 chosen file names are known. For example, @samp{%g.s ... %g.o ... %g.s}
3340 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
3341 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
3342 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
3343 was simply substituted with a file name chosen once per compilation,
3344 without regard to any appended suffix (which was therefore treated
3345 just like ordinary text), making such attacks more likely to succeed.
3347 @item %u@var{suffix}
3348 Like @samp{%g}, but generates a new temporary file name even if
3349 @samp{%u@var{suffix}} was already seen.
3351 @item %U@var{suffix}
3352 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
3353 new one if there is no such last file name. In the absence of any
3354 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
3355 the same suffix @emph{space}, so @samp{%g.s ... %U.s ... %g.s ... %U.s}
3356 would involve the generation of two distinct file names, one
3357 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
3358 simply substituted with a file name chosen for the previous @samp{%u},
3359 without regard to any appended suffix.
3362 Marks the argument containing or following the @samp{%w} as the
3363 designated output file of this compilation. This puts the argument
3364 into the sequence of arguments that @samp{%o} will substitute later.
3367 Substitutes the names of all the output files, with spaces
3368 automatically placed around them. You should write spaces
3369 around the @samp{%o} as well or the results are undefined.
3370 @samp{%o} is for use in the specs for running the linker.
3371 Input files whose names have no recognized suffix are not compiled
3372 at all, but they are included among the output files, so they will
3376 Substitutes the suffix for object files. Note that this is
3377 handled specially when it immediately follows @samp{%g, %u, or %U},
3378 because of the need for those to form complete file names. The
3379 handling is such that @samp{%O} is treated exactly as if it had already
3380 been substituted, except that @samp{%g, %u, and %U} do not currently
3381 support additional @var{suffix} characters following @samp{%O} as they would
3382 following, for example, @samp{.o}.
3385 Substitutes the standard macro predefinitions for the
3386 current target machine. Use this when running @code{cpp}.
3389 Like @samp{%p}, but puts @samp{__} before and after the name of each
3390 predefined macro, except for macros that start with @samp{__} or with
3391 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ANSI
3395 Substitute a @samp{-iprefix} option made from GCC_EXEC_PREFIX.
3398 Current argument is the name of a library or startup file of some sort.
3399 Search for that file in a standard list of directories and substitute
3400 the full name found.
3403 Print @var{str} as an error message. @var{str} is terminated by a newline.
3404 Use this when inconsistent options are detected.
3407 Output @samp{-} if the input for the current command is coming from a pipe.
3410 Substitute the contents of spec string @var{name} at this point.
3413 Like @samp{%(...)} but put @samp{__} around @samp{-D} arguments.
3415 @item %x@{@var{option}@}
3416 Accumulate an option for @samp{%X}.
3419 Output the accumulated linker options specified by @samp{-Wl} or a @samp{%x}
3423 Output the accumulated assembler options specified by @samp{-Wa}.
3426 Output the accumulated preprocessor options specified by @samp{-Wp}.
3429 Substitute the major version number of GCC.
3430 (For version 2.9.5, this is 2.)
3433 Substitute the minor version number of GCC.
3434 (For version 2.9.5, this is 9.)
3437 Process the @code{asm} spec. This is used to compute the
3438 switches to be passed to the assembler.
3441 Process the @code{asm_final} spec. This is a spec string for
3442 passing switches to an assembler post-processor, if such a program is
3446 Process the @code{link} spec. This is the spec for computing the
3447 command line passed to the linker. Typically it will make use of the
3448 @samp{%L %G %S %D and %E} sequences.
3451 Dump out a @samp{-L} option for each directory that GCC believes might
3452 contain startup files. If the target supports multilibs then the
3453 current multilib directory will be prepended to each of these paths.
3456 Process the @code{lib} spec. This is a spec string for deciding which
3457 libraries should be included on the command line to the linker.
3460 Process the @code{libgcc} spec. This is a spec string for deciding
3461 which GCC support library should be included on the command line to the linker.
3464 Process the @code{startfile} spec. This is a spec for deciding which
3465 object files should be the first ones passed to the linker. Typically
3466 this might be a file named @file{crt0.o}.
3469 Process the @code{endfile} spec. This is a spec string that specifies
3470 the last object files that will be passed to the linker.
3473 Process the @code{cpp} spec. This is used to construct the arguments
3474 to be passed to the C preprocessor.
3477 Process the @code{signed_char} spec. This is intended to be used
3478 to tell cpp whether a char is signed. It typically has the definition:
3480 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
3484 Process the @code{cc1} spec. This is used to construct the options to be
3485 passed to the actual C compiler (@samp{cc1}).
3488 Process the @code{cc1plus} spec. This is used to construct the options to be
3489 passed to the actual C++ compiler (@samp{cc1plus}).
3492 Substitute the variable part of a matched option. See below.
3493 Note that each comma in the substituted string is replaced by
3497 Substitutes the @code{-S} switch, if that switch was given to GCC.
3498 If that switch was not specified, this substitutes nothing. Note that
3499 the leading dash is omitted when specifying this option, and it is
3500 automatically inserted if the substitution is performed. Thus the spec
3501 string @samp{%@{foo@}} would match the command-line option @samp{-foo}
3502 and would output the command line option @samp{-foo}.
3504 @item %W@{@code{S}@}
3505 Like %@{@code{S}@} but mark last argument supplied within as a file to be
3508 @item %@{@code{S}*@}
3509 Substitutes all the switches specified to GCC whose names start
3510 with @code{-S}, but which also take an argument. This is used for
3511 switches like @samp{-o, -D, -I}, etc. GCC considers @samp{-o foo} as being
3512 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
3513 text, including the space. Thus two arguments would be generated.
3515 @item %@{^@code{S}*@}
3516 Like %@{@code{S}*@}, but don't put a blank between a switch and its
3517 argument. Thus %@{^o*@} would only generate one argument, not two.
3519 @item %@{@code{S}*:@code{X}@}
3520 Substitutes @code{X} if one or more switches whose names start with
3521 @code{-S} are specified to GCC. Note that the tail part of the
3522 @code{-S} option (i.e. the part matched by the @samp{*}) will be substituted
3523 for each occurrence of @samp{%*} within @code{X}.
3525 @item %@{@code{S}:@code{X}@}
3526 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC.
3528 @item %@{!@code{S}:@code{X}@}
3529 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC.
3531 @item %@{|@code{S}:@code{X}@}
3532 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
3534 @item %@{|!@code{S}:@code{X}@}
3535 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
3537 @item %@{.@code{S}:@code{X}@}
3538 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
3540 @item %@{!.@code{S}:@code{X}@}
3541 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
3543 @item %@{@code{S}|@code{P}:@code{X}@}
3544 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC. This may be
3545 combined with @samp{!} and @samp{.} sequences as well, although they
3546 have a stronger binding than the @samp{|}. For example a spec string
3550 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
3553 will output the following command-line options from the following input
3554 command-line options:
3559 -d fred.c -foo -baz -boggle
3560 -d jim.d -bar -baz -boggle
3565 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
3566 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
3567 or spaces, or even newlines. They are processed as usual, as described
3570 The @samp{-O, -f, -m, and -W} switches are handled specifically in these
3571 constructs. If another value of @samp{-O} or the negated form of a @samp{-f, -m, or
3572 -W} switch is found later in the command line, the earlier switch
3573 value is ignored, except with @{@code{S}*@} where @code{S} is just one
3574 letter, which passes all matching options.
3576 The character @samp{|} at the beginning of the predicate text is used to indicate
3577 that a command should be piped to the following command, but only if @samp{-pipe}
3580 It is built into GCC which switches take arguments and which do not.
3581 (You might think it would be useful to generalize this to allow each
3582 compiler's spec to say which switches take arguments. But this cannot
3583 be done in a consistent fashion. GCC cannot even decide which input
3584 files have been specified without knowing which switches take arguments,
3585 and it must know which input files to compile in order to tell which
3588 GCC also knows implicitly that arguments starting in @samp{-l} are to be
3589 treated as compiler output files, and passed to the linker in their
3590 proper position among the other output files.
3592 @node Target Options
3593 @section Specifying Target Machine and Compiler Version
3594 @cindex target options
3595 @cindex cross compiling
3596 @cindex specifying machine version
3597 @cindex specifying compiler version and target machine
3598 @cindex compiler version, specifying
3599 @cindex target machine, specifying
3601 By default, GCC compiles code for the same type of machine that you
3602 are using. However, it can also be installed as a cross-compiler, to
3603 compile for some other type of machine. In fact, several different
3604 configurations of GCC, for different target machines, can be
3605 installed side by side. Then you specify which one to use with the
3608 In addition, older and newer versions of GCC can be installed side
3609 by side. One of them (probably the newest) will be the default, but
3610 you may sometimes wish to use another.
3613 @item -b @var{machine}
3614 The argument @var{machine} specifies the target machine for compilation.
3615 This is useful when you have installed GCC as a cross-compiler.
3617 The value to use for @var{machine} is the same as was specified as the
3618 machine type when configuring GCC as a cross-compiler. For
3619 example, if a cross-compiler was configured with @samp{configure
3620 i386v}, meaning to compile for an 80386 running System V, then you
3621 would specify @samp{-b i386v} to run that cross compiler.
3623 When you do not specify @samp{-b}, it normally means to compile for
3624 the same type of machine that you are using.
3626 @item -V @var{version}
3627 The argument @var{version} specifies which version of GCC to run.
3628 This is useful when multiple versions are installed. For example,
3629 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
3631 The default version, when you do not specify @samp{-V}, is the last
3632 version of GCC that you installed.
3635 The @samp{-b} and @samp{-V} options actually work by controlling part of
3636 the file name used for the executable files and libraries used for
3637 compilation. A given version of GCC, for a given target machine, is
3638 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.@refill
3640 Thus, sites can customize the effect of @samp{-b} or @samp{-V} either by
3641 changing the names of these directories or adding alternate names (or
3642 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
3643 file @file{80386} is a link to the file @file{i386v}, then @samp{-b
3644 80386} becomes an alias for @samp{-b i386v}.
3646 In one respect, the @samp{-b} or @samp{-V} do not completely change
3647 to a different compiler: the top-level driver program @code{gcc}
3648 that you originally invoked continues to run and invoke the other
3649 executables (preprocessor, compiler per se, assembler and linker)
3650 that do the real work. However, since no real work is done in the
3651 driver program, it usually does not matter that the driver program
3652 in use is not the one for the specified target and version.
3654 The only way that the driver program depends on the target machine is
3655 in the parsing and handling of special machine-specific options.
3656 However, this is controlled by a file which is found, along with the
3657 other executables, in the directory for the specified version and
3658 target machine. As a result, a single installed driver program adapts
3659 to any specified target machine and compiler version.
3661 The driver program executable does control one significant thing,
3662 however: the default version and target machine. Therefore, you can
3663 install different instances of the driver program, compiled for
3664 different targets or versions, under different names.
3666 For example, if the driver for version 2.0 is installed as @code{ogcc}
3667 and that for version 2.1 is installed as @code{gcc}, then the command
3668 @code{gcc} will use version 2.1 by default, while @code{ogcc} will use
3669 2.0 by default. However, you can choose either version with either
3670 command with the @samp{-V} option.
3672 @node Submodel Options
3673 @section Hardware Models and Configurations
3674 @cindex submodel options
3675 @cindex specifying hardware config
3676 @cindex hardware models and configurations, specifying
3677 @cindex machine dependent options
3679 Earlier we discussed the standard option @samp{-b} which chooses among
3680 different installed compilers for completely different target
3681 machines, such as Vax vs. 68000 vs. 80386.
3683 In addition, each of these target machine types can have its own
3684 special options, starting with @samp{-m}, to choose among various
3685 hardware models or configurations---for example, 68010 vs 68020,
3686 floating coprocessor or none. A single installed version of the
3687 compiler can compile for any model or configuration, according to the
3690 Some configurations of the compiler also support additional special
3691 options, usually for compatibility with other compilers on the same
3695 These options are defined by the macro @code{TARGET_SWITCHES} in the
3696 machine description. The default for the options is also defined by
3697 that macro, which enables you to change the defaults.
3712 * RS/6000 and PowerPC Options::
3717 * Intel 960 Options::
3718 * DEC Alpha Options::
3722 * System V Options::
3723 * TMS320C3x/C4x Options::
3729 @node M680x0 Options
3730 @subsection M680x0 Options
3731 @cindex M680x0 options
3733 These are the @samp{-m} options defined for the 68000 series. The default
3734 values for these options depends on which style of 68000 was selected when
3735 the compiler was configured; the defaults for the most common choices are
3741 Generate output for a 68000. This is the default
3742 when the compiler is configured for 68000-based systems.
3744 Use this option for microcontrollers with a 68000 or EC000 core,
3745 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
3749 Generate output for a 68020. This is the default
3750 when the compiler is configured for 68020-based systems.
3753 Generate output containing 68881 instructions for floating point.
3754 This is the default for most 68020 systems unless @samp{-nfp} was
3755 specified when the compiler was configured.
3758 Generate output for a 68030. This is the default when the compiler is
3759 configured for 68030-based systems.
3762 Generate output for a 68040. This is the default when the compiler is
3763 configured for 68040-based systems.
3765 This option inhibits the use of 68881/68882 instructions that have to be
3766 emulated by software on the 68040. Use this option if your 68040 does not
3767 have code to emulate those instructions.
3770 Generate output for a 68060. This is the default when the compiler is
3771 configured for 68060-based systems.
3773 This option inhibits the use of 68020 and 68881/68882 instructions that
3774 have to be emulated by software on the 68060. Use this option if your 68060
3775 does not have code to emulate those instructions.
3778 Generate output for a CPU32. This is the default
3779 when the compiler is configured for CPU32-based systems.
3781 Use this option for microcontrollers with a
3782 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
3783 68336, 68340, 68341, 68349 and 68360.
3786 Generate output for a 520X "coldfire" family cpu. This is the default
3787 when the compiler is configured for 520X-based systems.
3789 Use this option for microcontroller with a 5200 core, including
3790 the MCF5202, MCF5203, MCF5204 and MCF5202.
3794 Generate output for a 68040, without using any of the new instructions.
3795 This results in code which can run relatively efficiently on either a
3796 68020/68881 or a 68030 or a 68040. The generated code does use the
3797 68881 instructions that are emulated on the 68040.
3800 Generate output for a 68060, without using any of the new instructions.
3801 This results in code which can run relatively efficiently on either a
3802 68020/68881 or a 68030 or a 68040. The generated code does use the
3803 68881 instructions that are emulated on the 68060.
3806 Generate output containing Sun FPA instructions for floating point.
3809 Generate output containing library calls for floating point.
3810 @strong{Warning:} the requisite libraries are not available for all m68k
3811 targets. Normally the facilities of the machine's usual C compiler are
3812 used, but this can't be done directly in cross-compilation. You must
3813 make your own arrangements to provide suitable library functions for
3814 cross-compilation. The embedded targets @samp{m68k-*-aout} and
3815 @samp{m68k-*-coff} do provide software floating point support.
3818 Consider type @code{int} to be 16 bits wide, like @code{short int}.
3821 Do not use the bit-field instructions. The @samp{-m68000}, @samp{-mcpu32}
3822 and @samp{-m5200} options imply @w{@samp{-mnobitfield}}.
3825 Do use the bit-field instructions. The @samp{-m68020} option implies
3826 @samp{-mbitfield}. This is the default if you use a configuration
3827 designed for a 68020.
3830 Use a different function-calling convention, in which functions
3831 that take a fixed number of arguments return with the @code{rtd}
3832 instruction, which pops their arguments while returning. This
3833 saves one instruction in the caller since there is no need to pop
3834 the arguments there.
3836 This calling convention is incompatible with the one normally
3837 used on Unix, so you cannot use it if you need to call libraries
3838 compiled with the Unix compiler.
3840 Also, you must provide function prototypes for all functions that
3841 take variable numbers of arguments (including @code{printf});
3842 otherwise incorrect code will be generated for calls to those
3845 In addition, seriously incorrect code will result if you call a
3846 function with too many arguments. (Normally, extra arguments are
3847 harmlessly ignored.)
3849 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
3850 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
3853 @itemx -mno-align-int
3854 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
3855 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
3856 boundary (@samp{-malign-int}) or a 16-bit boundary (@samp{-mno-align-int}).
3857 Aligning variables on 32-bit boundaries produces code that runs somewhat
3858 faster on processors with 32-bit busses at the expense of more memory.
3860 @strong{Warning:} if you use the @samp{-malign-int} switch, GCC will
3861 align structures containing the above types differently than
3862 most published application binary interface specifications for the m68k.
3865 Use the pc-relative addressing mode of the 68000 directly, instead of
3866 using a global offset table. At present, this option implies -fpic,
3867 allowing at most a 16-bit offset for pc-relative addressing. -fPIC is
3868 not presently supported with -mpcrel, though this could be supported for
3869 68020 and higher processors.
3871 @item -mno-strict-align
3872 @itemx -mstrict-align
3873 @kindex -mstrict-align
3874 Do not (do) assume that unaligned memory references will be handled by
3880 @subsection VAX Options
3883 These @samp{-m} options are defined for the Vax:
3887 Do not output certain jump instructions (@code{aobleq} and so on)
3888 that the Unix assembler for the Vax cannot handle across long
3892 Do output those jump instructions, on the assumption that you
3893 will assemble with the GNU assembler.
3896 Output code for g-format floating point numbers instead of d-format.
3900 @subsection SPARC Options
3901 @cindex SPARC options
3903 These @samp{-m} switches are supported on the SPARC:
3908 Specify @samp{-mapp-regs} to generate output using the global registers
3909 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
3912 To be fully SVR4 ABI compliant at the cost of some performance loss,
3913 specify @samp{-mno-app-regs}. You should compile libraries and system
3914 software with this option.
3918 Generate output containing floating point instructions. This is the
3923 Generate output containing library calls for floating point.
3924 @strong{Warning:} the requisite libraries are not available for all SPARC
3925 targets. Normally the facilities of the machine's usual C compiler are
3926 used, but this cannot be done directly in cross-compilation. You must make
3927 your own arrangements to provide suitable library functions for
3928 cross-compilation. The embedded targets @samp{sparc-*-aout} and
3929 @samp{sparclite-*-*} do provide software floating point support.
3931 @samp{-msoft-float} changes the calling convention in the output file;
3932 therefore, it is only useful if you compile @emph{all} of a program with
3933 this option. In particular, you need to compile @file{libgcc.a}, the
3934 library that comes with GCC, with @samp{-msoft-float} in order for
3937 @item -mhard-quad-float
3938 Generate output containing quad-word (long double) floating point
3941 @item -msoft-quad-float
3942 Generate output containing library calls for quad-word (long double)
3943 floating point instructions. The functions called are those specified
3944 in the SPARC ABI. This is the default.
3946 As of this writing, there are no sparc implementations that have hardware
3947 support for the quad-word floating point instructions. They all invoke
3948 a trap handler for one of these instructions, and then the trap handler
3949 emulates the effect of the instruction. Because of the trap handler overhead,
3950 this is much slower than calling the ABI library routines. Thus the
3951 @samp{-msoft-quad-float} option is the default.
3955 With @samp{-mepilogue} (the default), the compiler always emits code for
3956 function exit at the end of each function. Any function exit in
3957 the middle of the function (such as a return statement in C) will
3958 generate a jump to the exit code at the end of the function.
3960 With @samp{-mno-epilogue}, the compiler tries to emit exit code inline
3961 at every function exit.
3965 With @samp{-mflat}, the compiler does not generate save/restore instructions
3966 and will use a "flat" or single register window calling convention.
3967 This model uses %i7 as the frame pointer and is compatible with the normal
3968 register window model. Code from either may be intermixed.
3969 The local registers and the input registers (0-5) are still treated as
3970 "call saved" registers and will be saved on the stack as necessary.
3972 With @samp{-mno-flat} (the default), the compiler emits save/restore
3973 instructions (except for leaf functions) and is the normal mode of operation.
3975 @item -mno-unaligned-doubles
3976 @itemx -munaligned-doubles
3977 Assume that doubles have 8 byte alignment. This is the default.
3979 With @samp{-munaligned-doubles}, GCC assumes that doubles have 8 byte
3980 alignment only if they are contained in another type, or if they have an
3981 absolute address. Otherwise, it assumes they have 4 byte alignment.
3982 Specifying this option avoids some rare compatibility problems with code
3983 generated by other compilers. It is not the default because it results
3984 in a performance loss, especially for floating point code.
3986 @item -mno-faster-structs
3987 @itemx -mfaster-structs
3988 With @samp{-mfaster-structs}, the compiler assumes that structures
3989 should have 8 byte alignment. This enables the use of pairs of
3990 @code{ldd} and @code{std} instructions for copies in structure
3991 assignment, in place of twice as many @code{ld} and @code{st} pairs.
3992 However, the use of this changed alignment directly violates the Sparc
3993 ABI. Thus, it's intended only for use on targets where the developer
3994 acknowledges that their resulting code will not be directly in line with
3995 the rules of the ABI.
3999 These two options select variations on the SPARC architecture.
4001 By default (unless specifically configured for the Fujitsu SPARClite),
4002 GCC generates code for the v7 variant of the SPARC architecture.
4004 @samp{-mv8} will give you SPARC v8 code. The only difference from v7
4005 code is that the compiler emits the integer multiply and integer
4006 divide instructions which exist in SPARC v8 but not in SPARC v7.
4008 @samp{-msparclite} will give you SPARClite code. This adds the integer
4009 multiply, integer divide step and scan (@code{ffs}) instructions which
4010 exist in SPARClite but not in SPARC v7.
4012 These options are deprecated and will be deleted in a future GCC release.
4013 They have been replaced with @samp{-mcpu=xxx}.
4017 These two options select the processor for which the code is optimised.
4019 With @samp{-mcypress} (the default), the compiler optimizes code for the
4020 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
4021 This is also appropriate for the older SparcStation 1, 2, IPX etc.
4023 With @samp{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
4024 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
4025 of the full SPARC v8 instruction set.
4027 These options are deprecated and will be deleted in a future GCC release.
4028 They have been replaced with @samp{-mcpu=xxx}.
4030 @item -mcpu=@var{cpu_type}
4031 Set the instruction set, register set, and instruction scheduling parameters
4032 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
4033 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
4034 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
4035 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
4037 Default instruction scheduling parameters are used for values that select
4038 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
4039 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
4041 Here is a list of each supported architecture and their supported
4046 v8: supersparc, hypersparc
4047 sparclite: f930, f934, sparclite86x
4052 @item -mtune=@var{cpu_type}
4053 Set the instruction scheduling parameters for machine type
4054 @var{cpu_type}, but do not set the instruction set or register set that the
4055 option @samp{-mcpu=}@var{cpu_type} would.
4057 The same values for @samp{-mcpu=}@var{cpu_type} are used for
4058 @samp{-mtune=}@*@var{cpu_type}, though the only useful values are those that
4059 select a particular cpu implementation: @samp{cypress}, @samp{supersparc},
4060 @samp{hypersparc}, @samp{f930}, @samp{f934}, @samp{sparclite86x},
4061 @samp{tsc701}, @samp{ultrasparc}.
4065 These @samp{-m} switches are supported in addition to the above
4066 on the SPARCLET processor.
4069 @item -mlittle-endian
4070 Generate code for a processor running in little-endian mode.
4073 Treat register @code{%g0} as a normal register.
4074 GCC will continue to clobber it as necessary but will not assume
4075 it always reads as 0.
4077 @item -mbroken-saverestore
4078 Generate code that does not use non-trivial forms of the @code{save} and
4079 @code{restore} instructions. Early versions of the SPARCLET processor do
4080 not correctly handle @code{save} and @code{restore} instructions used with
4081 arguments. They correctly handle them used without arguments. A @code{save}
4082 instruction used without arguments increments the current window pointer
4083 but does not allocate a new stack frame. It is assumed that the window
4084 overflow trap handler will properly handle this case as will interrupt
4088 These @samp{-m} switches are supported in addition to the above
4089 on SPARC V9 processors in 64 bit environments.
4092 @item -mlittle-endian
4093 Generate code for a processor running in little-endian mode.
4097 Generate code for a 32 bit or 64 bit environment.
4098 The 32 bit environment sets int, long and pointer to 32 bits.
4099 The 64 bit environment sets int to 32 bits and long and pointer
4102 @item -mcmodel=medlow
4103 Generate code for the Medium/Low code model: the program must be linked
4104 in the low 32 bits of the address space. Pointers are 64 bits.
4105 Programs can be statically or dynamically linked.
4107 @item -mcmodel=medmid
4108 Generate code for the Medium/Middle code model: the program must be linked
4109 in the low 44 bits of the address space, the text segment must be less than
4110 2G bytes, and data segment must be within 2G of the text segment.
4111 Pointers are 64 bits.
4113 @item -mcmodel=medany
4114 Generate code for the Medium/Anywhere code model: the program may be linked
4115 anywhere in the address space, the text segment must be less than
4116 2G bytes, and data segment must be within 2G of the text segment.
4117 Pointers are 64 bits.
4119 @item -mcmodel=embmedany
4120 Generate code for the Medium/Anywhere code model for embedded systems:
4121 assume a 32 bit text and a 32 bit data segment, both starting anywhere
4122 (determined at link time). Register %g4 points to the base of the
4123 data segment. Pointers still 64 bits.
4124 Programs are statically linked, PIC is not supported.
4127 @itemx -mno-stack-bias
4128 With @samp{-mstack-bias}, GCC assumes that the stack pointer, and
4129 frame pointer if present, are offset by -2047 which must be added back
4130 when making stack frame references.
4131 Otherwise, assume no such offset is present.
4134 @node Convex Options
4135 @subsection Convex Options
4136 @cindex Convex options
4138 These @samp{-m} options are defined for Convex:
4142 Generate output for C1. The code will run on any Convex machine.
4143 The preprocessor symbol @code{__convex__c1__} is defined.
4146 Generate output for C2. Uses instructions not available on C1.
4147 Scheduling and other optimizations are chosen for max performance on C2.
4148 The preprocessor symbol @code{__convex_c2__} is defined.
4151 Generate output for C32xx. Uses instructions not available on C1.
4152 Scheduling and other optimizations are chosen for max performance on C32.
4153 The preprocessor symbol @code{__convex_c32__} is defined.
4156 Generate output for C34xx. Uses instructions not available on C1.
4157 Scheduling and other optimizations are chosen for max performance on C34.
4158 The preprocessor symbol @code{__convex_c34__} is defined.
4161 Generate output for C38xx. Uses instructions not available on C1.
4162 Scheduling and other optimizations are chosen for max performance on C38.
4163 The preprocessor symbol @code{__convex_c38__} is defined.
4166 Generate code which puts an argument count in the word preceding each
4167 argument list. This is compatible with regular CC, and a few programs
4168 may need the argument count word. GDB and other source-level debuggers
4169 do not need it; this info is in the symbol table.
4172 Omit the argument count word. This is the default.
4174 @item -mvolatile-cache
4175 Allow volatile references to be cached. This is the default.
4177 @item -mvolatile-nocache
4178 Volatile references bypass the data cache, going all the way to memory.
4179 This is only needed for multi-processor code that does not use standard
4180 synchronization instructions. Making non-volatile references to volatile
4181 locations will not necessarily work.
4184 Type long is 32 bits, the same as type int. This is the default.
4187 Type long is 64 bits, the same as type long long. This option is useless,
4188 because no library support exists for it.
4191 @node AMD29K Options
4192 @subsection AMD29K Options
4193 @cindex AMD29K options
4195 These @samp{-m} options are defined for the AMD Am29000:
4200 @cindex DW bit (29k)
4201 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
4202 halfword operations are directly supported by the hardware. This is the
4207 Generate code that assumes the @code{DW} bit is not set.
4211 @cindex byte writes (29k)
4212 Generate code that assumes the system supports byte and halfword write
4213 operations. This is the default.
4217 Generate code that assumes the systems does not support byte and
4218 halfword write operations. @samp{-mnbw} implies @samp{-mndw}.
4222 @cindex memory model (29k)
4223 Use a small memory model that assumes that all function addresses are
4224 either within a single 256 KB segment or at an absolute address of less
4225 than 256k. This allows the @code{call} instruction to be used instead
4226 of a @code{const}, @code{consth}, @code{calli} sequence.
4230 Use the normal memory model: Generate @code{call} instructions only when
4231 calling functions in the same file and @code{calli} instructions
4232 otherwise. This works if each file occupies less than 256 KB but allows
4233 the entire executable to be larger than 256 KB. This is the default.
4236 Always use @code{calli} instructions. Specify this option if you expect
4237 a single file to compile into more than 256 KB of code.
4241 @cindex processor selection (29k)
4242 Generate code for the Am29050.
4246 Generate code for the Am29000. This is the default.
4248 @item -mkernel-registers
4249 @kindex -mkernel-registers
4250 @cindex kernel and user registers (29k)
4251 Generate references to registers @code{gr64-gr95} instead of to
4252 registers @code{gr96-gr127}. This option can be used when compiling
4253 kernel code that wants a set of global registers disjoint from that used
4256 Note that when this option is used, register names in @samp{-f} flags
4257 must use the normal, user-mode, names.
4259 @item -muser-registers
4260 @kindex -muser-registers
4261 Use the normal set of global registers, @code{gr96-gr127}. This is the
4265 @itemx -mno-stack-check
4266 @kindex -mstack-check
4267 @cindex stack checks (29k)
4268 Insert (or do not insert) a call to @code{__msp_check} after each stack
4269 adjustment. This is often used for kernel code.
4272 @itemx -mno-storem-bug
4273 @kindex -mstorem-bug
4274 @cindex storem bug (29k)
4275 @samp{-mstorem-bug} handles 29k processors which cannot handle the
4276 separation of a mtsrim insn and a storem instruction (most 29000 chips
4277 to date, but not the 29050).
4279 @item -mno-reuse-arg-regs
4280 @itemx -mreuse-arg-regs
4281 @kindex -mreuse-arg-regs
4282 @samp{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
4283 registers for copying out arguments. This helps detect calling a function
4284 with fewer arguments than it was declared with.
4286 @item -mno-impure-text
4287 @itemx -mimpure-text
4288 @kindex -mimpure-text
4289 @samp{-mimpure-text}, used in addition to @samp{-shared}, tells the compiler to
4290 not pass @samp{-assert pure-text} to the linker when linking a shared object.
4293 @kindex -msoft-float
4294 Generate output containing library calls for floating point.
4295 @strong{Warning:} the requisite libraries are not part of GCC.
4296 Normally the facilities of the machine's usual C compiler are used, but
4297 this can't be done directly in cross-compilation. You must make your
4298 own arrangements to provide suitable library functions for
4303 Do not generate multm or multmu instructions. This is useful for some embedded
4304 systems which do not have trap handlers for these instructions.
4308 @subsection ARM Options
4311 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
4316 @kindex -mapcs-frame
4317 Generate a stack frame that is compliant with the ARM Procedure Call
4318 Standard for all functions, even if this is not strictly necessary for
4319 correct execution of the code. Specifying @samp{-fomit-frame-pointer}
4320 with this option will cause the stack frames not to be generated for
4321 leaf functions. The default is @samp{-mno-apcs-frame}.
4325 This is a synonym for @samp{-mapcs-frame}.
4329 Generate code for a processor running with a 26-bit program counter,
4330 and conforming to the function calling standards for the APCS 26-bit
4331 option. This option replaces the @samp{-m2} and @samp{-m3} options
4332 of previous releases of the compiler.
4336 Generate code for a processor running with a 32-bit program counter,
4337 and conforming to the function calling standards for the APCS 32-bit
4338 option. This option replaces the @samp{-m6} option of previous releases
4341 @item -mapcs-stack-check
4342 @kindex -mapcs-stack-check
4343 @kindex -mno-apcs-stack-check
4344 Generate code to check the amount of stack space available upon entry to
4345 every function (that actually uses some stack space). If there is
4346 insufficient space available then either the function
4347 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
4348 called, depending upon the amount of stack space required. The run time
4349 system is required to provide these functions. The default is
4350 @samp{-mno-apcs-stack-check}, since this produces smaller code.
4353 @kindex -mapcs-float
4354 @kindex -mno-apcs-float
4355 Pass floating point arguments using the float point registers. This is
4356 one of the variants of the APCS. This option is recommended if the
4357 target hardware has a floating point unit or if a lot of floating point
4358 arithmetic is going to be performed by the code. The default is
4359 @samp{-mno-apcs-float}, since integer only code is slightly increased in
4360 size if @samp{-mapcs-float} is used.
4362 @item -mapcs-reentrant
4363 @kindex -mapcs-reentrant
4364 @kindex -mno-apcs-reentrant
4365 Generate reentrant, position independent code. This is the equivalent
4366 to specifying the @samp{-fpic} option. The default is
4367 @samp{-mno-apcs-reentrant}.
4369 @item -mthumb-interwork
4370 @kindex -mthumb-interwork
4371 @kindex -mno-thumb-interwork
4372 Generate code which supports calling between the ARM and THUMB
4373 instruction sets. Without this option the two instruction sets cannot
4374 be reliably used inside one program. The default is
4375 @samp{-mno-thumb-interwork}, since slightly larger code is generated
4376 when @samp{-mthumb-interwork} is specified.
4378 @item -mno-sched-prolog
4379 @kindex -mno-sched-prolog
4380 @kindex -msched-prolog
4381 Prevent the reordering of instructions in the function prolog, or the
4382 merging of those instruction with the instructions in the function's
4383 body. This means that all functions will start with a recognizable set
4384 of instructions (or in fact one of a choice from a small set of
4385 different function prologues), and this information can be used to
4386 locate the start if functions inside an executable piece of code. The
4387 default is @samp{-msched-prolog}.
4390 Generate output containing floating point instructions. This is the
4394 Generate output containing library calls for floating point.
4395 @strong{Warning:} the requisite libraries are not available for all ARM
4396 targets. Normally the facilities of the machine's usual C compiler are
4397 used, but this cannot be done directly in cross-compilation. You must make
4398 your own arrangements to provide suitable library functions for
4401 @samp{-msoft-float} changes the calling convention in the output file;
4402 therefore, it is only useful if you compile @emph{all} of a program with
4403 this option. In particular, you need to compile @file{libgcc.a}, the
4404 library that comes with GCC, with @samp{-msoft-float} in order for
4407 @item -mlittle-endian
4408 Generate code for a processor running in little-endian mode. This is
4409 the default for all standard configurations.
4412 Generate code for a processor running in big-endian mode; the default is
4413 to compile code for a little-endian processor.
4415 @item -mwords-little-endian
4416 This option only applies when generating code for big-endian processors.
4417 Generate code for a little-endian word order but a big-endian byte
4418 order. That is, a byte order of the form @samp{32107654}. Note: this
4419 option should only be used if you require compatibility with code for
4420 big-endian ARM processors generated by versions of the compiler prior to
4423 @item -malignment-traps
4424 @kindex -malignment-traps
4425 Generate code that will not trap if the MMU has alignment traps enabled.
4426 On ARM architectures prior to ARMv4, there were no instructions to
4427 access half-word objects stored in memory. However, when reading from
4428 memory a feature of the ARM architecture allows a word load to be used,
4429 even if the address is unaligned, and the processor core will rotate the
4430 data as it is being loaded. This option tells the compiler that such
4431 misaligned accesses will cause a MMU trap and that it should instead
4432 synthesise the access as a series of byte accesses. The compiler can
4433 still use word accesses to load half-word data if it knows that the
4434 address is aligned to a word boundary.
4436 This option is ignored when compiling for ARM architecture 4 or later,
4437 since these processors have instructions to directly access half-word
4440 @item -mno-alignment-traps
4441 @kindex -mno-alignment-traps
4442 Generate code that assumes that the MMU will not trap unaligned
4443 accesses. This produces better code when the target instruction set
4444 does not have half-word memory operations (implementations prior to
4447 Note that you cannot use this option to access unaligned word objects,
4448 since the processor will only fetch one 32-bit aligned object from
4451 The default setting for most targets is -mno-alignment-traps, since
4452 this produces better code when there are no half-word memory
4453 instructions available.
4455 @item -mshort-load-bytes
4456 @kindex -mshort-load-bytes
4457 This is a depreciated alias for @samp{-malignment-traps}.
4459 @item -mno-short-load-bytes
4460 @kindex -mno-short-load-bytes
4461 This is a depreciated alias for @samp{-mno-alignment-traps}.
4463 @item -mshort-load-words
4464 @kindex -mshort-load-words
4465 This is a depreciated alias for @samp{-mno-alignment-traps}.
4467 @item -mno-short-load-words
4468 @kindex -mno-short-load-words
4469 This is a depreciated alias for @samp{-malignment-traps}.
4473 This option only applies to RISC iX. Emulate the native BSD-mode
4474 compiler. This is the default if @samp{-ansi} is not specified.
4478 This option only applies to RISC iX. Emulate the native X/Open-mode
4481 @item -mno-symrename
4482 @kindex -mno-symrename
4483 This option only applies to RISC iX. Do not run the assembler
4484 post-processor, @samp{symrename}, after code has been assembled.
4485 Normally it is necessary to modify some of the standard symbols in
4486 preparation for linking with the RISC iX C library; this option
4487 suppresses this pass. The post-processor is never run when the
4488 compiler is built for cross-compilation.
4492 This specifies the name of the target ARM processor. GCC uses this name
4493 to determine what kind of instructions it can use when generating
4494 assembly code. Permissible names are: arm2, arm250, arm3, arm6, arm60,
4495 arm600, arm610, arm620, arm7, arm7m, arm7d, arm7dm, arm7di, arm7dmi,
4496 arm70, arm700, arm700i, arm710, arm710c, arm7100, arm7500, arm7500fe,
4497 arm7tdmi, arm8, strongarm, strongarm110, strongarm1100, arm8, arm810,
4498 arm9, arm920, arm920t, arm9tdmi.
4500 @itemx -mtune=<name>
4502 This option is very similar to the @samp{-mcpu=} option, except that
4503 instead of specifying the actual target processor type, and hence
4504 restricting which instructions can be used, it specifies that GCC should
4505 tune the performance of the code as if the target were of the type
4506 specified in this option, but still choosing the instructions that it
4507 will generate based on the cpu specified by a @samp{-mcpu=} option.
4508 For some arm implementations better performance can be obtained by using
4513 This specifies the name of the target ARM architecture. GCC uses this
4514 name to determine what kind of instructions it can use when generating
4515 assembly code. This option can be used in conjunction with or instead
4516 of the @samp{-mcpu=} option. Permissible names are: armv2, armv2a,
4517 armv3, armv3m, armv4, armv4t, armv5.
4519 @item -mfpe=<number>
4520 @itemx -mfp=<number>
4523 This specifes the version of the floating point emulation available on
4524 the target. Permissible values are 2 and 3. @samp{-mfp=} is a synonym
4525 for @samp{-mfpe=} to support older versions of GCC.
4527 @item -mstructure-size-boundary=<n>
4528 @kindex -mstructure-size-boundary
4529 The size of all structures and unions will be rounded up to a multiple
4530 of the number of bits set by this option. Permissible values are 8 and
4531 32. The default value varies for different toolchains. For the COFF
4532 targeted toolchain the default value is 8. Specifying the larger number
4533 can produce faster, more efficient code, but can also increase the size
4534 of the program. The two values are potentially incompatible. Code
4535 compiled with one value cannot necessarily expect to work with code or
4536 libraries compiled with the other value, if they exchange information
4537 using structures or unions. Programmers are encouraged to use the 32
4538 value as future versions of the toolchain may default to this value.
4540 @item -mabort-on-noreturn
4541 @kindex -mabort-on-noreturn
4542 @kindex -mnoabort-on-noreturn
4543 Generate a call to the function abort at the end of a noreturn function.
4544 It will be executed if the function tries to return.
4546 @item -mnop-fun-dllimport
4547 @kindex -mnop-fun-dllimport
4548 Disable the support for the @emph{dllimport} attribute.
4550 @item -msingle-pic-base
4551 @kindex -msingle-pic-base
4552 Treat the register used for PIC addressing as read-only, rather than
4553 loading it in the prologue for each function. The run-time system is
4554 responsible for initialising this register with an appropriate value
4555 before execution begins.
4557 @item -mpic-register=<reg>
4558 @kindex -mpic-register=
4559 Specify the register to be used for PIC addressing. The default is R10
4560 unless stack-checking is enabled, when R9 is used.
4565 @subsection Thumb Options
4566 @cindex Thumb Options
4570 @item -mthumb-interwork
4571 @kindex -mthumb-interwork
4572 @kindex -mno-thumb-interwork
4573 Generate code which supports calling between the THUMB and ARM
4574 instruction sets. Without this option the two instruction sets cannot
4575 be reliably used inside one program. The default is
4576 @samp{-mno-thumb-interwork}, since slightly smaller code is generated
4580 @kindex -mtpcs-frame
4581 @kindex -mno-tpcs-frame
4582 Generate a stack frame that is compliant with the Thumb Procedure Call
4583 Standard for all non-leaf functions. (A leaf function is one that does
4584 not call any other functions). The default is @samp{-mno-apcs-frame}.
4586 @item -mtpcs-leaf-frame
4587 @kindex -mtpcs-leaf-frame
4588 @kindex -mno-tpcs-leaf-frame
4589 Generate a stack frame that is compliant with the Thumb Procedure Call
4590 Standard for all leaf functions. (A leaf function is one that does
4591 not call any other functions). The default is @samp{-mno-apcs-leaf-frame}.
4593 @item -mlittle-endian
4594 @kindex -mlittle-endian
4595 Generate code for a processor running in little-endian mode. This is
4596 the default for all standard configurations.
4599 @kindex -mbig-endian
4600 Generate code for a processor running in big-endian mode.
4602 @item -mstructure-size-boundary=<n>
4603 @kindex -mstructure-size-boundary
4604 The size of all structures and unions will be rounded up to a multiple
4605 of the number of bits set by this option. Permissible values are 8 and
4606 32. The default value varies for different toolchains. For the COFF
4607 targeted toolchain the default value is 8. Specifying the larger number
4608 can produced faster, more efficient code, but can also increase the size
4609 of the program. The two values are potentially incompatible. Code
4610 compiled with one value cannot necessarily expect to work with code or
4611 libraries compiled with the other value, if they exchange information
4612 using structures or unions. Programmers are encouraged to use the 32
4613 value as future versions of the toolchain may default to this value.
4615 @item -mnop-fun-dllimport
4616 @kindex -mnop-fun-dllimport
4617 Disable the support for the @emph{dllimport} attribute.
4619 @item -mcallee-super-interworking
4620 @kindex -mcallee-super-interworking
4621 Gives all externally visible functions in the file being compiled an ARM
4622 instruction set header which switches to Thumb mode before executing the
4623 rest of the function. This allows these functions to be called from
4624 non-interworking code.
4626 @item -mcaller-super-interworking
4627 @kindex -mcaller-super-interworking
4628 Allows calls via function pointers (including virtual functions) to
4629 execute correctly regardless of whether the target code has been
4630 compiled for interworking or not. There is a small overhead in the cost
4631 of executing a function pointer if this option is enabled.
4633 @item -msingle-pic-base
4634 @kindex -msingle-pic-base
4635 Treat the register used for PIC addressing as read-only, rather than
4636 loading it in the prologue for each function. The run-time system is
4637 responsible for initialising this register with an appropriate value
4638 before execution begins.
4640 @item -mpic-register=<reg>
4641 @kindex -mpic-register=
4642 Specify the register to be used for PIC addressing. The default is R10.
4646 @node MN10200 Options
4647 @subsection MN10200 Options
4648 @cindex MN10200 options
4649 These @samp{-m} options are defined for Matsushita MN10200 architectures:
4653 Indicate to the linker that it should perform a relaxation optimization pass
4654 to shorten branches, calls and absolute memory addresses. This option only
4655 has an effect when used on the command line for the final link step.
4657 This option makes symbolic debugging impossible.
4660 @node MN10300 Options
4661 @subsection MN10300 Options
4662 @cindex MN10300 options
4663 These @samp{-m} options are defined for Matsushita MN10300 architectures:
4667 Generate code to avoid bugs in the multiply instructions for the MN10300
4668 processors. This is the default.
4671 Do not generate code to avoid bugs in the multiply instructions for the
4675 Generate code which uses features specific to the AM33 processor.
4678 Do not generate code which uses features specific to the AM33 processor. This
4682 Indicate to the linker that it should perform a relaxation optimization pass
4683 to shorten branches, calls and absolute memory addresses. This option only
4684 has an effect when used on the command line for the final link step.
4686 This option makes symbolic debugging impossible.
4690 @node M32R/D Options
4691 @subsection M32R/D Options
4692 @cindex M32R/D options
4694 These @samp{-m} options are defined for Mitsubishi M32R/D architectures:
4697 @item -mcode-model=small
4698 Assume all objects live in the lower 16MB of memory (so that their addresses
4699 can be loaded with the @code{ld24} instruction), and assume all subroutines
4700 are reachable with the @code{bl} instruction.
4701 This is the default.
4703 The addressability of a particular object can be set with the
4704 @code{model} attribute.
4706 @item -mcode-model=medium
4707 Assume objects may be anywhere in the 32 bit address space (the compiler
4708 will generate @code{seth/add3} instructions to load their addresses), and
4709 assume all subroutines are reachable with the @code{bl} instruction.
4711 @item -mcode-model=large
4712 Assume objects may be anywhere in the 32 bit address space (the compiler
4713 will generate @code{seth/add3} instructions to load their addresses), and
4714 assume subroutines may not be reachable with the @code{bl} instruction
4715 (the compiler will generate the much slower @code{seth/add3/jl}
4716 instruction sequence).
4719 Disable use of the small data area. Variables will be put into
4720 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
4721 @code{section} attribute has been specified).
4722 This is the default.
4724 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
4725 Objects may be explicitly put in the small data area with the
4726 @code{section} attribute using one of these sections.
4729 Put small global and static data in the small data area, but do not
4730 generate special code to reference them.
4733 Put small global and static data in the small data area, and generate
4734 special instructions to reference them.
4737 @cindex smaller data references
4738 Put global and static objects less than or equal to @var{num} bytes
4739 into the small data or bss sections instead of the normal data or bss
4740 sections. The default value of @var{num} is 8.
4741 The @samp{-msdata} option must be set to one of @samp{sdata} or @samp{use}
4742 for this option to have any effect.
4744 All modules should be compiled with the same @samp{-G @var{num}} value.
4745 Compiling with different values of @var{num} may or may not work; if it
4746 doesn't the linker will give an error message - incorrect code will not be
4752 @subsection M88K Options
4753 @cindex M88k options
4755 These @samp{-m} options are defined for Motorola 88k architectures:
4760 Generate code that works well on both the m88100 and the
4765 Generate code that works best for the m88100, but that also
4770 Generate code that works best for the m88110, and may not run
4775 Obsolete option to be removed from the next revision.
4778 @item -midentify-revision
4779 @kindex -midentify-revision
4781 @cindex identifying source, compiler (88k)
4782 Include an @code{ident} directive in the assembler output recording the
4783 source file name, compiler name and version, timestamp, and compilation
4786 @item -mno-underscores
4787 @kindex -mno-underscores
4788 @cindex underscores, avoiding (88k)
4789 In assembler output, emit symbol names without adding an underscore
4790 character at the beginning of each name. The default is to use an
4791 underscore as prefix on each name.
4793 @item -mocs-debug-info
4794 @itemx -mno-ocs-debug-info
4795 @kindex -mocs-debug-info
4796 @kindex -mno-ocs-debug-info
4798 @cindex debugging, 88k OCS
4799 Include (or omit) additional debugging information (about registers used
4800 in each stack frame) as specified in the 88open Object Compatibility
4801 Standard, ``OCS''. This extra information allows debugging of code that
4802 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
4803 Delta 88 SVr3.2 is to include this information; other 88k configurations
4804 omit this information by default.
4806 @item -mocs-frame-position
4807 @kindex -mocs-frame-position
4808 @cindex register positions in frame (88k)
4809 When emitting COFF debugging information for automatic variables and
4810 parameters stored on the stack, use the offset from the canonical frame
4811 address, which is the stack pointer (register 31) on entry to the
4812 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
4813 @samp{-mocs-frame-position}; other 88k configurations have the default
4814 @samp{-mno-ocs-frame-position}.
4816 @item -mno-ocs-frame-position
4817 @kindex -mno-ocs-frame-position
4818 @cindex register positions in frame (88k)
4819 When emitting COFF debugging information for automatic variables and
4820 parameters stored on the stack, use the offset from the frame pointer
4821 register (register 30). When this option is in effect, the frame
4822 pointer is not eliminated when debugging information is selected by the
4825 @item -moptimize-arg-area
4826 @itemx -mno-optimize-arg-area
4827 @kindex -moptimize-arg-area
4828 @kindex -mno-optimize-arg-area
4829 @cindex arguments in frame (88k)
4830 Control how function arguments are stored in stack frames.
4831 @samp{-moptimize-arg-area} saves space by optimizing them, but this
4832 conflicts with the 88open specifications. The opposite alternative,
4833 @samp{-mno-optimize-arg-area}, agrees with 88open standards. By default
4834 GCC does not optimize the argument area.
4836 @item -mshort-data-@var{num}
4837 @kindex -mshort-data-@var{num}
4838 @cindex smaller data references (88k)
4839 @cindex r0-relative references (88k)
4840 Generate smaller data references by making them relative to @code{r0},
4841 which allows loading a value using a single instruction (rather than the
4842 usual two). You control which data references are affected by
4843 specifying @var{num} with this option. For example, if you specify
4844 @samp{-mshort-data-512}, then the data references affected are those
4845 involving displacements of less than 512 bytes.
4846 @samp{-mshort-data-@var{num}} is not effective for @var{num} greater
4849 @item -mserialize-volatile
4850 @kindex -mserialize-volatile
4851 @itemx -mno-serialize-volatile
4852 @kindex -mno-serialize-volatile
4853 @cindex sequential consistency on 88k
4854 Do, or don't, generate code to guarantee sequential consistency
4855 of volatile memory references. By default, consistency is
4858 The order of memory references made by the MC88110 processor does
4859 not always match the order of the instructions requesting those
4860 references. In particular, a load instruction may execute before
4861 a preceding store instruction. Such reordering violates
4862 sequential consistency of volatile memory references, when there
4863 are multiple processors. When consistency must be guaranteed,
4864 GNU C generates special instructions, as needed, to force
4865 execution in the proper order.
4867 The MC88100 processor does not reorder memory references and so
4868 always provides sequential consistency. However, by default, GNU
4869 C generates the special instructions to guarantee consistency
4870 even when you use @samp{-m88100}, so that the code may be run on an
4871 MC88110 processor. If you intend to run your code only on the
4872 MC88100 processor, you may use @samp{-mno-serialize-volatile}.
4874 The extra code generated to guarantee consistency may affect the
4875 performance of your application. If you know that you can safely
4876 forgo this guarantee, you may use @samp{-mno-serialize-volatile}.
4882 @cindex assembler syntax, 88k
4884 Turn on (@samp{-msvr4}) or off (@samp{-msvr3}) compiler extensions
4885 related to System V release 4 (SVr4). This controls the following:
4889 Which variant of the assembler syntax to emit.
4891 @samp{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
4892 that is used on System V release 4.
4894 @samp{-msvr4} makes GCC issue additional declaration directives used in
4898 @samp{-msvr4} is the default for the m88k-motorola-sysv4 and
4899 m88k-dg-dgux m88k configurations. @samp{-msvr3} is the default for all
4900 other m88k configurations.
4902 @item -mversion-03.00
4903 @kindex -mversion-03.00
4904 This option is obsolete, and is ignored.
4905 @c ??? which asm syntax better for GAS? option there too?
4907 @item -mno-check-zero-division
4908 @itemx -mcheck-zero-division
4909 @kindex -mno-check-zero-division
4910 @kindex -mcheck-zero-division
4911 @cindex zero division on 88k
4912 Do, or don't, generate code to guarantee that integer division by
4913 zero will be detected. By default, detection is guaranteed.
4915 Some models of the MC88100 processor fail to trap upon integer
4916 division by zero under certain conditions. By default, when
4917 compiling code that might be run on such a processor, GNU C
4918 generates code that explicitly checks for zero-valued divisors
4919 and traps with exception number 503 when one is detected. Use of
4920 mno-check-zero-division suppresses such checking for code
4921 generated to run on an MC88100 processor.
4923 GNU C assumes that the MC88110 processor correctly detects all
4924 instances of integer division by zero. When @samp{-m88110} is
4925 specified, both @samp{-mcheck-zero-division} and
4926 @samp{-mno-check-zero-division} are ignored, and no explicit checks for
4927 zero-valued divisors are generated.
4929 @item -muse-div-instruction
4930 @kindex -muse-div-instruction
4931 @cindex divide instruction, 88k
4932 Use the div instruction for signed integer division on the
4933 MC88100 processor. By default, the div instruction is not used.
4935 On the MC88100 processor the signed integer division instruction
4936 div) traps to the operating system on a negative operand. The
4937 operating system transparently completes the operation, but at a
4938 large cost in execution time. By default, when compiling code
4939 that might be run on an MC88100 processor, GNU C emulates signed
4940 integer division using the unsigned integer division instruction
4941 divu), thereby avoiding the large penalty of a trap to the
4942 operating system. Such emulation has its own, smaller, execution
4943 cost in both time and space. To the extent that your code's
4944 important signed integer division operations are performed on two
4945 nonnegative operands, it may be desirable to use the div
4946 instruction directly.
4948 On the MC88110 processor the div instruction (also known as the
4949 divs instruction) processes negative operands without trapping to
4950 the operating system. When @samp{-m88110} is specified,
4951 @samp{-muse-div-instruction} is ignored, and the div instruction is used
4952 for signed integer division.
4954 Note that the result of dividing INT_MIN by -1 is undefined. In
4955 particular, the behavior of such a division with and without
4956 @samp{-muse-div-instruction} may differ.
4958 @item -mtrap-large-shift
4959 @itemx -mhandle-large-shift
4960 @kindex -mtrap-large-shift
4961 @kindex -mhandle-large-shift
4962 @cindex bit shift overflow (88k)
4963 @cindex large bit shifts (88k)
4964 Include code to detect bit-shifts of more than 31 bits; respectively,
4965 trap such shifts or emit code to handle them properly. By default GCC
4966 makes no special provision for large bit shifts.
4968 @item -mwarn-passed-structs
4969 @kindex -mwarn-passed-structs
4970 @cindex structure passing (88k)
4971 Warn when a function passes a struct as an argument or result.
4972 Structure-passing conventions have changed during the evolution of the C
4973 language, and are often the source of portability problems. By default,
4974 GCC issues no such warning.
4977 @node RS/6000 and PowerPC Options
4978 @subsection IBM RS/6000 and PowerPC Options
4979 @cindex RS/6000 and PowerPC Options
4980 @cindex IBM RS/6000 and PowerPC Options
4982 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
4990 @itemx -mpowerpc-gpopt
4991 @itemx -mno-powerpc-gpopt
4992 @itemx -mpowerpc-gfxopt
4993 @itemx -mno-powerpc-gfxopt
4995 @itemx -mno-powerpc64
4999 @kindex -mpowerpc-gpopt
5000 @kindex -mpowerpc-gfxopt
5002 GCC supports two related instruction set architectures for the
5003 RS/6000 and PowerPC. The @dfn{POWER} instruction set are those
5004 instructions supported by the @samp{rios} chip set used in the original
5005 RS/6000 systems and the @dfn{PowerPC} instruction set is the
5006 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
5007 the IBM 4xx microprocessors.
5009 Neither architecture is a subset of the other. However there is a
5010 large common subset of instructions supported by both. An MQ
5011 register is included in processors supporting the POWER architecture.
5013 You use these options to specify which instructions are available on the
5014 processor you are using. The default value of these options is
5015 determined when configuring GCC. Specifying the
5016 @samp{-mcpu=@var{cpu_type}} overrides the specification of these
5017 options. We recommend you use the @samp{-mcpu=@var{cpu_type}} option
5018 rather than the options listed above.
5020 The @samp{-mpower} option allows GCC to generate instructions that
5021 are found only in the POWER architecture and to use the MQ register.
5022 Specifying @samp{-mpower2} implies @samp{-power} and also allows GCC
5023 to generate instructions that are present in the POWER2 architecture but
5024 not the original POWER architecture.
5026 The @samp{-mpowerpc} option allows GCC to generate instructions that
5027 are found only in the 32-bit subset of the PowerPC architecture.
5028 Specifying @samp{-mpowerpc-gpopt} implies @samp{-mpowerpc} and also allows
5029 GCC to use the optional PowerPC architecture instructions in the
5030 General Purpose group, including floating-point square root. Specifying
5031 @samp{-mpowerpc-gfxopt} implies @samp{-mpowerpc} and also allows GCC to
5032 use the optional PowerPC architecture instructions in the Graphics
5033 group, including floating-point select.
5035 The @samp{-mpowerpc64} option allows GCC to generate the additional
5036 64-bit instructions that are found in the full PowerPC64 architecture
5037 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
5038 @samp{-mno-powerpc64}.
5040 If you specify both @samp{-mno-power} and @samp{-mno-powerpc}, GCC
5041 will use only the instructions in the common subset of both
5042 architectures plus some special AIX common-mode calls, and will not use
5043 the MQ register. Specifying both @samp{-mpower} and @samp{-mpowerpc}
5044 permits GCC to use any instruction from either architecture and to
5045 allow use of the MQ register; specify this for the Motorola MPC601.
5047 @item -mnew-mnemonics
5048 @itemx -mold-mnemonics
5049 @kindex -mnew-mnemonics
5050 @kindex -mold-mnemonics
5051 Select which mnemonics to use in the generated assembler code.
5052 @samp{-mnew-mnemonics} requests output that uses the assembler mnemonics
5053 defined for the PowerPC architecture, while @samp{-mold-mnemonics}
5054 requests the assembler mnemonics defined for the POWER architecture.
5055 Instructions defined in only one architecture have only one mnemonic;
5056 GCC uses that mnemonic irrespective of which of these options is
5059 GCC defaults to the mnemonics appropriate for the architecture in
5060 use. Specifying @samp{-mcpu=@var{cpu_type}} sometimes overrides the
5061 value of these option. Unless you are building a cross-compiler, you
5062 should normally not specify either @samp{-mnew-mnemonics} or
5063 @samp{-mold-mnemonics}, but should instead accept the default.
5065 @item -mcpu=@var{cpu_type}
5067 Set architecture type, register usage, choice of mnemonics, and
5068 instruction scheduling parameters for machine type @var{cpu_type}.
5069 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
5070 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
5071 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
5072 @samp{630}, @samp{740}, @samp{750}, @samp{power}, @samp{power2},
5073 @samp{powerpc}, @samp{403}, @samp{505}, @samp{801}, @samp{821},
5074 @samp{823}, and @samp{860} and @samp{common}. @samp{-mcpu=power},
5075 @samp{-mcpu=power2}, @samp{-mcpu=powerpc}, and @samp{-mcpu=powerpc64}
5076 specify generic POWER, POWER2, pure 32-bit PowerPC (i.e., not MPC601),
5077 and 64-bit PowerPC architecture machine types, with an appropriate,
5078 generic processor model assumed for scheduling purposes.@refill
5080 Specifying any of the following options:
5081 @samp{-mcpu=rios1}, @samp{-mcpu=rios2}, @samp{-mcpu=rsc},
5082 @samp{-mcpu=power}, or @samp{-mcpu=power2}
5083 enables the @samp{-mpower} option and disables the @samp{-mpowerpc} option;
5084 @samp{-mcpu=601} enables both the @samp{-mpower} and @samp{-mpowerpc} options.
5085 All of @samp{-mcpu=rs64a}, @samp{-mcpu=602}, @samp{-mcpu=603},
5086 @samp{-mcpu=603e}, @samp{-mcpu=604}, @samp{-mcpu=620}, @samp{-mcpu=630},
5087 @samp{-mcpu=740}, and @samp{-mcpu=750}
5088 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5089 Exactly similarly, all of @samp{-mcpu=403},
5090 @samp{-mcpu=505}, @samp{-mcpu=821}, @samp{-mcpu=860} and @samp{-mcpu=powerpc}
5091 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5092 @samp{-mcpu=common} disables both the
5093 @samp{-mpower} and @samp{-mpowerpc} options.@refill
5095 AIX versions 4 or greater selects @samp{-mcpu=common} by default, so
5096 that code will operate on all members of the RS/6000 POWER and PowerPC
5097 families. In that case, GCC will use only the instructions in the
5098 common subset of both architectures plus some special AIX common-mode
5099 calls, and will not use the MQ register. GCC assumes a generic
5100 processor model for scheduling purposes.
5102 Specifying any of the options @samp{-mcpu=rios1}, @samp{-mcpu=rios2},
5103 @samp{-mcpu=rsc}, @samp{-mcpu=power}, or @samp{-mcpu=power2} also
5104 disables the @samp{new-mnemonics} option. Specifying @samp{-mcpu=601},
5105 @samp{-mcpu=602}, @samp{-mcpu=603}, @samp{-mcpu=603e}, @samp{-mcpu=604},
5106 @samp{-mcpu=620}, @samp{-mcpu=630}, @samp{-mcpu=403}, @samp{-mcpu=505},
5107 @samp{-mcpu=821}, @samp{-mcpu=860} or @samp{-mcpu=powerpc} also enables
5108 the @samp{new-mnemonics} option.@refill
5110 Specifying @samp{-mcpu=403}, @samp{-mcpu=821}, or @samp{-mcpu=860} also
5111 enables the @samp{-msoft-float} option.
5113 @item -mtune=@var{cpu_type}
5114 Set the instruction scheduling parameters for machine type
5115 @var{cpu_type}, but do not set the architecture type, register usage,
5116 choice of mnemonics like @samp{-mcpu=}@var{cpu_type} would. The same
5117 values for @var{cpu_type} are used for @samp{-mtune=}@var{cpu_type} as
5118 for @samp{-mcpu=}@var{cpu_type}. The @samp{-mtune=}@var{cpu_type}
5119 option overrides the @samp{-mcpu=}@var{cpu_type} option in terms of
5120 instruction scheduling parameters.
5123 @itemx -mno-fp-in-toc
5124 @itemx -mno-sum-in-toc
5125 @itemx -mminimal-toc
5126 @kindex -mminimal-toc
5127 Modify generation of the TOC (Table Of Contents), which is created for
5128 every executable file. The @samp{-mfull-toc} option is selected by
5129 default. In that case, GCC will allocate at least one TOC entry for
5130 each unique non-automatic variable reference in your program. GCC
5131 will also place floating-point constants in the TOC. However, only
5132 16,384 entries are available in the TOC.
5134 If you receive a linker error message that saying you have overflowed
5135 the available TOC space, you can reduce the amount of TOC space used
5136 with the @samp{-mno-fp-in-toc} and @samp{-mno-sum-in-toc} options.
5137 @samp{-mno-fp-in-toc} prevents GCC from putting floating-point
5138 constants in the TOC and @samp{-mno-sum-in-toc} forces GCC to
5139 generate code to calculate the sum of an address and a constant at
5140 run-time instead of putting that sum into the TOC. You may specify one
5141 or both of these options. Each causes GCC to produce very slightly
5142 slower and larger code at the expense of conserving TOC space.
5144 If you still run out of space in the TOC even when you specify both of
5145 these options, specify @samp{-mminimal-toc} instead. This option causes
5146 GCC to make only one TOC entry for every file. When you specify this
5147 option, GCC will produce code that is slower and larger but which
5148 uses extremely little TOC space. You may wish to use this option
5149 only on files that contain less frequently executed code. @refill
5155 Enable 64-bit PowerPC ABI and calling convention: 64-bit pointers, 64-bit
5156 @code{long} type, and the infrastructure needed to support them.
5157 Specifying @samp{-m64} implies @samp{-mpowerpc64} and
5158 @samp{-mpowerpc}, while @samp{-m32} disables the 64-bit ABI and
5159 implies @samp{-mno-powerpc64}. GCC defaults to @samp{-m32}.
5164 On AIX, pass floating-point arguments to prototyped functions beyond the
5165 register save area (RSA) on the stack in addition to argument FPRs. The
5166 AIX calling convention was extended but not initially documented to
5167 handle an obscure K&R C case of calling a function that takes the
5168 address of its arguments with fewer arguments than declared. AIX XL
5169 compilers access floating point arguments which do not fit in the
5170 RSA from the stack when a subroutine is compiled without
5171 optimization. Because always storing floating-point arguments on the
5172 stack is inefficient and rarely needed, this option is not enabled by
5173 default and only is necessary when calling subroutines compiled by AIX
5174 XL compilers without optimization.
5178 Support @dfn{AIX Threads}. Link an application written to use
5179 @dfn{pthreads} with special libraries and startup code to enable the
5184 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE). Link an
5185 application written to use message passing with special startup code to
5186 enable the application to run. The system must have PE installed in the
5187 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
5188 must be overridden with the @samp{-specs=} option to specify the
5189 appropriate directory location. The Parallel Environment does not
5190 support threads, so the @samp{-mpe} option and the @samp{-mthreads}
5191 option are incompatible.
5195 @kindex -msoft-float
5196 Generate code that does not use (uses) the floating-point register set.
5197 Software floating point emulation is provided if you use the
5198 @samp{-msoft-float} option, and pass the option to GCC when linking.
5201 @itemx -mno-multiple
5202 Generate code that uses (does not use) the load multiple word
5203 instructions and the store multiple word instructions. These
5204 instructions are generated by default on POWER systems, and not
5205 generated on PowerPC systems. Do not use @samp{-mmultiple} on little
5206 endian PowerPC systems, since those instructions do not work when the
5207 processor is in little endian mode. The exceptions are PPC740 and
5208 PPC750 which permit the instructions usage in little endian mode.
5213 Generate code that uses (does not use) the load string instructions
5214 and the store string word instructions to save multiple registers and
5215 do small block moves. These instructions are generated by default on
5216 POWER systems, and not generated on PowerPC systems. Do not use
5217 @samp{-mstring} on little endian PowerPC systems, since those
5218 instructions do not work when the processor is in little endian mode.
5219 The exceptions are PPC740 and PPC750 which permit the instructions
5220 usage in little endian mode.
5225 Generate code that uses (does not use) the load or store instructions
5226 that update the base register to the address of the calculated memory
5227 location. These instructions are generated by default. If you use
5228 @samp{-mno-update}, there is a small window between the time that the
5229 stack pointer is updated and the address of the previous frame is
5230 stored, which means code that walks the stack frame across interrupts or
5231 signals may get corrupted data.
5234 @itemx -mno-fused-madd
5235 @kindex -mfused-madd
5236 Generate code that uses (does not use) the floating point multiply and
5237 accumulate instructions. These instructions are generated by default if
5238 hardware floating is used.
5240 @item -mno-bit-align
5243 On System V.4 and embedded PowerPC systems do not (do) force structures
5244 and unions that contain bit fields to be aligned to the base type of the
5247 For example, by default a structure containing nothing but 8
5248 @code{unsigned} bitfields of length 1 would be aligned to a 4 byte
5249 boundary and have a size of 4 bytes. By using @samp{-mno-bit-align},
5250 the structure would be aligned to a 1 byte boundary and be one byte in
5253 @item -mno-strict-align
5254 @itemx -mstrict-align
5255 @kindex -mstrict-align
5256 On System V.4 and embedded PowerPC systems do not (do) assume that
5257 unaligned memory references will be handled by the system.
5260 @itemx -mno-relocatable
5261 @kindex -mrelocatable
5262 On embedded PowerPC systems generate code that allows (does not allow)
5263 the program to be relocated to a different address at runtime. If you
5264 use @samp{-mrelocatable} on any module, all objects linked together must
5265 be compiled with @samp{-mrelocatable} or @samp{-mrelocatable-lib}.
5267 @item -mrelocatable-lib
5268 @itemx -mno-relocatable-lib
5269 On embedded PowerPC systems generate code that allows (does not allow)
5270 the program to be relocated to a different address at runtime. Modules
5271 compiled with @samp{-mrelocatable-lib} can be linked with either modules
5272 compiled without @samp{-mrelocatable} and @samp{-mrelocatable-lib} or
5273 with modules compiled with the @samp{-mrelocatable} options.
5277 On System V.4 and embedded PowerPC systems do not (do) assume that
5278 register 2 contains a pointer to a global area pointing to the addresses
5279 used in the program.
5282 @itemx -mlittle-endian
5283 On System V.4 and embedded PowerPC systems compile code for the
5284 processor in little endian mode. The @samp{-mlittle-endian} option is
5285 the same as @samp{-mlittle}.
5289 On System V.4 and embedded PowerPC systems compile code for the
5290 processor in big endian mode. The @samp{-mbig-endian} option is
5291 the same as @samp{-mbig}.
5294 On System V.4 and embedded PowerPC systems compile code using calling
5295 conventions that adheres to the March 1995 draft of the System V
5296 Application Binary Interface, PowerPC processor supplement. This is the
5297 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
5299 @item -mcall-sysv-eabi
5300 Specify both @samp{-mcall-sysv} and @samp{-meabi} options.
5302 @item -mcall-sysv-noeabi
5303 Specify both @samp{-mcall-sysv} and @samp{-mno-eabi} options.
5306 On System V.4 and embedded PowerPC systems compile code using calling
5307 conventions that are similar to those used on AIX. This is the
5308 default if you configured GCC using @samp{powerpc-*-eabiaix}.
5310 @item -mcall-solaris
5311 On System V.4 and embedded PowerPC systems compile code for the Solaris
5315 On System V.4 and embedded PowerPC systems compile code for the
5316 Linux-based GNU system.
5319 @itemx -mno-prototype
5320 On System V.4 and embedded PowerPC systems assume that all calls to
5321 variable argument functions are properly prototyped. Otherwise, the
5322 compiler must insert an instruction before every non prototyped call to
5323 set or clear bit 6 of the condition code register (@var{CR}) to
5324 indicate whether floating point values were passed in the floating point
5325 registers in case the function takes a variable arguments. With
5326 @samp{-mprototype}, only calls to prototyped variable argument functions
5327 will set or clear the bit.
5330 On embedded PowerPC systems, assume that the startup module is called
5331 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
5332 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
5336 On embedded PowerPC systems, assume that the startup module is called
5337 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
5341 On embedded PowerPC systems, assume that the startup module is called
5342 @file{crt0.o} and the standard C libraries are @file{libads.a} and
5346 On embedded PowerPC systems, assume that the startup module is called
5347 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
5351 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
5352 header to indicate that @samp{eabi} extended relocations are used.
5356 On System V.4 and embedded PowerPC systems do (do not) adhere to the
5357 Embedded Applications Binary Interface (eabi) which is a set of
5358 modifications to the System V.4 specifications. Selecting @code{-meabi}
5359 means that the stack is aligned to an 8 byte boundary, a function
5360 @code{__eabi} is called to from @code{main} to set up the eabi
5361 environment, and the @samp{-msdata} option can use both @code{r2} and
5362 @code{r13} to point to two separate small data areas. Selecting
5363 @code{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
5364 do not call an initialization function from @code{main}, and the
5365 @samp{-msdata} option will only use @code{r13} to point to a single
5366 small data area. The @samp{-meabi} option is on by default if you
5367 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
5370 On System V.4 and embedded PowerPC systems, put small initialized
5371 @code{const} global and static data in the @samp{.sdata2} section, which
5372 is pointed to by register @code{r2}. Put small initialized
5373 non-@code{const} global and static data in the @samp{.sdata} section,
5374 which is pointed to by register @code{r13}. Put small uninitialized
5375 global and static data in the @samp{.sbss} section, which is adjacent to
5376 the @samp{.sdata} section. The @samp{-msdata=eabi} option is
5377 incompatible with the @samp{-mrelocatable} option. The
5378 @samp{-msdata=eabi} option also sets the @samp{-memb} option.
5381 On System V.4 and embedded PowerPC systems, put small global and static
5382 data in the @samp{.sdata} section, which is pointed to by register
5383 @code{r13}. Put small uninitialized global and static data in the
5384 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
5385 The @samp{-msdata=sysv} option is incompatible with the
5386 @samp{-mrelocatable} option.
5388 @item -msdata=default
5390 On System V.4 and embedded PowerPC systems, if @samp{-meabi} is used,
5391 compile code the same as @samp{-msdata=eabi}, otherwise compile code the
5392 same as @samp{-msdata=sysv}.
5395 On System V.4 and embedded PowerPC systems, put small global and static
5396 data in the @samp{.sdata} section. Put small uninitialized global and
5397 static data in the @samp{.sbss} section. Do not use register @code{r13}
5398 to address small data however. This is the default behavior unless
5399 other @samp{-msdata} options are used.
5403 On embedded PowerPC systems, put all initialized global and static data
5404 in the @samp{.data} section, and all uninitialized data in the
5405 @samp{.bss} section.
5408 @cindex smaller data references (PowerPC)
5409 @cindex .sdata/.sdata2 references (PowerPC)
5410 On embedded PowerPC systems, put global and static items less than or
5411 equal to @var{num} bytes into the small data or bss sections instead of
5412 the normal data or bss section. By default, @var{num} is 8. The
5413 @samp{-G @var{num}} switch is also passed to the linker.
5414 All modules should be compiled with the same @samp{-G @var{num}} value.
5417 @itemx -mno-regnames
5418 On System V.4 and embedded PowerPC systems do (do not) emit register
5419 names in the assembly language output using symbolic forms.
5424 @subsection IBM RT Options
5426 @cindex IBM RT options
5428 These @samp{-m} options are defined for the IBM RT PC:
5432 Use an in-line code sequence for integer multiplies. This is the
5435 @item -mcall-lib-mul
5436 Call @code{lmul$$} for integer multiples.
5438 @item -mfull-fp-blocks
5439 Generate full-size floating point data blocks, including the minimum
5440 amount of scratch space recommended by IBM. This is the default.
5442 @item -mminimum-fp-blocks
5443 Do not include extra scratch space in floating point data blocks. This
5444 results in smaller code, but slower execution, since scratch space must
5445 be allocated dynamically.
5447 @cindex @file{varargs.h} and RT PC
5448 @cindex @file{stdarg.h} and RT PC
5449 @item -mfp-arg-in-fpregs
5450 Use a calling sequence incompatible with the IBM calling convention in
5451 which floating point arguments are passed in floating point registers.
5452 Note that @code{varargs.h} and @code{stdargs.h} will not work with
5453 floating point operands if this option is specified.
5455 @item -mfp-arg-in-gregs
5456 Use the normal calling convention for floating point arguments. This is
5459 @item -mhc-struct-return
5460 Return structures of more than one word in memory, rather than in a
5461 register. This provides compatibility with the MetaWare HighC (hc)
5462 compiler. Use the option @samp{-fpcc-struct-return} for compatibility
5463 with the Portable C Compiler (pcc).
5465 @item -mnohc-struct-return
5466 Return some structures of more than one word in registers, when
5467 convenient. This is the default. For compatibility with the
5468 IBM-supplied compilers, use the option @samp{-fpcc-struct-return} or the
5469 option @samp{-mhc-struct-return}.
5473 @subsection MIPS Options
5474 @cindex MIPS options
5476 These @samp{-m} options are defined for the MIPS family of computers:
5479 @item -mcpu=@var{cpu type}
5480 Assume the defaults for the machine type @var{cpu type} when scheduling
5481 instructions. The choices for @var{cpu type} are @samp{r2000}, @samp{r3000},
5482 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
5483 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
5484 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
5485 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
5486 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
5487 @var{cpu type} will schedule things appropriately for that particular
5488 chip, the compiler will not generate any code that does not meet level 1
5489 of the MIPS ISA (instruction set architecture) without a @samp{-mipsX}
5490 or @samp{-mabi} switch being used.
5493 Issue instructions from level 1 of the MIPS ISA. This is the default.
5494 @samp{r3000} is the default @var{cpu type} at this ISA level.
5497 Issue instructions from level 2 of the MIPS ISA (branch likely, square
5498 root instructions). @samp{r6000} is the default @var{cpu type} at this
5502 Issue instructions from level 3 of the MIPS ISA (64 bit instructions).
5503 @samp{r4000} is the default @var{cpu type} at this ISA level.
5506 Issue instructions from level 4 of the MIPS ISA (conditional move,
5507 prefetch, enhanced FPU instructions). @samp{r8000} is the default
5508 @var{cpu type} at this ISA level.
5511 Assume that 32 32-bit floating point registers are available. This is
5515 Assume that 32 64-bit floating point registers are available. This is
5516 the default when the @samp{-mips3} option is used.
5519 Assume that 32 32-bit general purpose registers are available. This is
5523 Assume that 32 64-bit general purpose registers are available. This is
5524 the default when the @samp{-mips3} option is used.
5527 Force int and long types to be 64 bits wide. See @samp{-mlong32} for an
5528 explanation of the default, and the width of pointers.
5531 Force long types to be 64 bits wide. See @samp{-mlong32} for an
5532 explanation of the default, and the width of pointers.
5535 Force long, int, and pointer types to be 32 bits wide.
5537 If none of @samp{-mlong32}, @samp{-mlong64}, or @samp{-mint64} are set,
5538 the size of ints, longs, and pointers depends on the ABI and ISA choosen.
5539 For @samp{-mabi=32}, and @samp{-mabi=n32}, ints and longs are 32 bits
5540 wide. For @samp{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
5541 For @samp{-mabi=eabi} and either @samp{-mips1} or @samp{-mips2}, ints
5542 and longs are 32 bits wide. For @samp{-mabi=eabi} and higher ISAs, ints
5543 are 32 bits, and longs are 64 bits wide. The width of pointer types is
5544 the smaller of the width of longs or the width of general purpose
5545 registers (which in turn depends on the ISA).
5552 Generate code for the indicated ABI. The default instruction level is
5553 @samp{-mips1} for @samp{32}, @samp{-mips3} for @samp{n32}, and
5554 @samp{-mips4} otherwise. Conversely, with @samp{-mips1} or
5555 @samp{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
5559 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
5560 add normal debug information. This is the default for all
5561 platforms except for the OSF/1 reference platform, using the OSF/rose
5562 object format. If the either of the @samp{-gstabs} or @samp{-gstabs+}
5563 switches are used, the @file{mips-tfile} program will encapsulate the
5564 stabs within MIPS ECOFF.
5567 Generate code for the GNU assembler. This is the default on the OSF/1
5568 reference platform, using the OSF/rose object format. Also, this is
5569 the default if the configure option @samp{--with-gnu-as} is used.
5571 @item -msplit-addresses
5572 @itemx -mno-split-addresses
5573 Generate code to load the high and low parts of address constants separately.
5574 This allows @code{gcc} to optimize away redundant loads of the high order
5575 bits of addresses. This optimization requires GNU as and GNU ld.
5576 This optimization is enabled by default for some embedded targets where
5577 GNU as and GNU ld are standard.
5581 The @samp{-mrnames} switch says to output code using the MIPS software
5582 names for the registers, instead of the hardware names (ie, @var{a0}
5583 instead of @var{$4}). The only known assembler that supports this option
5584 is the Algorithmics assembler.
5588 The @samp{-mgpopt} switch says to write all of the data declarations
5589 before the instructions in the text section, this allows the MIPS
5590 assembler to generate one word memory references instead of using two
5591 words for short global or static data items. This is on by default if
5592 optimization is selected.
5596 For each non-inline function processed, the @samp{-mstats} switch
5597 causes the compiler to emit one line to the standard error file to
5598 print statistics about the program (number of registers saved, stack
5603 The @samp{-mmemcpy} switch makes all block moves call the appropriate
5604 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
5605 generating inline code.
5608 @itemx -mno-mips-tfile
5609 The @samp{-mno-mips-tfile} switch causes the compiler not
5610 postprocess the object file with the @file{mips-tfile} program,
5611 after the MIPS assembler has generated it to add debug support. If
5612 @file{mips-tfile} is not run, then no local variables will be
5613 available to the debugger. In addition, @file{stage2} and
5614 @file{stage3} objects will have the temporary file names passed to the
5615 assembler embedded in the object file, which means the objects will
5616 not compare the same. The @samp{-mno-mips-tfile} switch should only
5617 be used when there are bugs in the @file{mips-tfile} program that
5618 prevents compilation.
5621 Generate output containing library calls for floating point.
5622 @strong{Warning:} the requisite libraries are not part of GCC.
5623 Normally the facilities of the machine's usual C compiler are used, but
5624 this can't be done directly in cross-compilation. You must make your
5625 own arrangements to provide suitable library functions for
5629 Generate output containing floating point instructions. This is the
5630 default if you use the unmodified sources.
5633 @itemx -mno-abicalls
5634 Emit (or do not emit) the pseudo operations @samp{.abicalls},
5635 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
5636 position independent code.
5639 @itemx -mno-long-calls
5640 Do all calls with the @samp{JALR} instruction, which requires
5641 loading up a function's address into a register before the call.
5642 You need to use this switch, if you call outside of the current
5643 512 megabyte segment to functions that are not through pointers.
5646 @itemx -mno-half-pic
5647 Put pointers to extern references into the data section and load them
5648 up, rather than put the references in the text section.
5650 @item -membedded-pic
5651 @itemx -mno-embedded-pic
5652 Generate PIC code suitable for some embedded systems. All calls are
5653 made using PC relative address, and all data is addressed using the $gp
5654 register. No more than 65536 bytes of global data may be used. This
5655 requires GNU as and GNU ld which do most of the work. This currently
5656 only works on targets which use ECOFF; it does not work with ELF.
5658 @item -membedded-data
5659 @itemx -mno-embedded-data
5660 Allocate variables to the read-only data section first if possible, then
5661 next in the small data section if possible, otherwise in data. This gives
5662 slightly slower code than the default, but reduces the amount of RAM required
5663 when executing, and thus may be preferred for some embedded systems.
5665 @item -muninit-const-in-rodata
5666 @itemx -mno-uninit-const-in-rodata
5667 When used together with -membedded-data, it will always store uninitialized
5668 const variables in the read-only data section.
5670 @item -msingle-float
5671 @itemx -mdouble-float
5672 The @samp{-msingle-float} switch tells gcc to assume that the floating
5673 point coprocessor only supports single precision operations, as on the
5674 @samp{r4650} chip. The @samp{-mdouble-float} switch permits gcc to use
5675 double precision operations. This is the default.
5679 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
5680 as on the @samp{r4650} chip.
5683 Turns on @samp{-msingle-float}, @samp{-mmad}, and, at least for now,
5688 Enable 16-bit instructions.
5691 Use the entry and exit pseudo ops. This option can only be used with
5695 Compile code for the processor in little endian mode.
5696 The requisite libraries are assumed to exist.
5699 Compile code for the processor in big endian mode.
5700 The requisite libraries are assumed to exist.
5703 @cindex smaller data references (MIPS)
5704 @cindex gp-relative references (MIPS)
5705 Put global and static items less than or equal to @var{num} bytes into
5706 the small data or bss sections instead of the normal data or bss
5707 section. This allows the assembler to emit one word memory reference
5708 instructions based on the global pointer (@var{gp} or @var{$28}),
5709 instead of the normal two words used. By default, @var{num} is 8 when
5710 the MIPS assembler is used, and 0 when the GNU assembler is used. The
5711 @samp{-G @var{num}} switch is also passed to the assembler and linker.
5712 All modules should be compiled with the same @samp{-G @var{num}}
5716 Tell the MIPS assembler to not run its preprocessor over user
5717 assembler files (with a @samp{.s} suffix) when assembling them.
5720 Do not include the default crt0.
5724 These options are defined by the macro
5725 @code{TARGET_SWITCHES} in the machine description. The default for the
5726 options is also defined by that macro, which enables you to change the
5731 @subsection Intel 386 Options
5732 @cindex i386 Options
5733 @cindex Intel 386 Options
5735 These @samp{-m} options are defined for the i386 family of computers:
5738 @item -mcpu=@var{cpu type}
5739 Assume the defaults for the machine type @var{cpu type} when scheduling
5740 instructions. The choices for @var{cpu type} are:
5742 @multitable @columnfractions .20 .20 .20 .20
5743 @item @samp{i386} @tab @samp{i486} @tab @samp{i586} @tab @samp{i686}
5744 @item @samp{pentium} @tab @samp{pentiumpro} @tab @samp{k6}
5747 While picking a specific @var{cpu type} will schedule things appropriately
5748 for that particular chip, the compiler will not generate any code that
5749 does not run on the i386 without the @samp{-march=@var{cpu type}} option
5750 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
5751 is equivalent to @samp{pentiumpro}. @samp{k6} is the AMD chip as
5752 opposed to the Intel ones.
5754 @item -march=@var{cpu type}
5755 Generate instructions for the machine type @var{cpu type}. The choices
5756 for @var{cpu type} are the same as for @samp{-mcpu}. Moreover,
5757 specifying @samp{-march=@var{cpu type}} implies @samp{-mcpu=@var{cpu type}}.
5763 Synonyms for -mcpu=i386, -mcpu=i486, -mcpu=pentium, and -mcpu=pentiumpro
5764 respectively. These synonyms are deprecated.
5768 Control whether or not the compiler uses IEEE floating point
5769 comparisons. These handle correctly the case where the result of a
5770 comparison is unordered.
5773 Generate output containing library calls for floating point.
5774 @strong{Warning:} the requisite libraries are not part of GCC.
5775 Normally the facilities of the machine's usual C compiler are used, but
5776 this can't be done directly in cross-compilation. You must make your
5777 own arrangements to provide suitable library functions for
5780 On machines where a function returns floating point results in the 80387
5781 register stack, some floating point opcodes may be emitted even if
5782 @samp{-msoft-float} is used.
5784 @item -mno-fp-ret-in-387
5785 Do not use the FPU registers for return values of functions.
5787 The usual calling convention has functions return values of types
5788 @code{float} and @code{double} in an FPU register, even if there
5789 is no FPU. The idea is that the operating system should emulate
5792 The option @samp{-mno-fp-ret-in-387} causes such values to be returned
5793 in ordinary CPU registers instead.
5795 @item -mno-fancy-math-387
5796 Some 387 emulators do not support the @code{sin}, @code{cos} and
5797 @code{sqrt} instructions for the 387. Specify this option to avoid
5798 generating those instructions. This option is the default on FreeBSD.
5799 As of revision 2.6.1, these instructions are not generated unless you
5800 also use the @samp{-ffast-math} switch.
5802 @item -malign-double
5803 @itemx -mno-align-double
5804 Control whether GCC aligns @code{double}, @code{long double}, and
5805 @code{long long} variables on a two word boundary or a one word
5806 boundary. Aligning @code{double} variables on a two word boundary will
5807 produce code that runs somewhat faster on a @samp{Pentium} at the
5808 expense of more memory.
5810 @strong{Warning:} if you use the @samp{-malign-double} switch,
5811 structures containing the above types will be aligned differently than
5812 the published application binary interface specifications for the 386.
5815 @itemx -mno-svr3-shlib
5816 Control whether GCC places uninitialized locals into @code{bss} or
5817 @code{data}. @samp{-msvr3-shlib} places these locals into @code{bss}.
5818 These options are meaningful only on System V Release 3.
5820 @item -mno-wide-multiply
5821 @itemx -mwide-multiply
5822 Control whether GCC uses the @code{mul} and @code{imul} that produce
5823 64 bit results in @code{eax:edx} from 32 bit operands to do @code{long
5824 long} multiplies and 32-bit division by constants.
5827 Use a different function-calling convention, in which functions that
5828 take a fixed number of arguments return with the @code{ret} @var{num}
5829 instruction, which pops their arguments while returning. This saves one
5830 instruction in the caller since there is no need to pop the arguments
5833 You can specify that an individual function is called with this calling
5834 sequence with the function attribute @samp{stdcall}. You can also
5835 override the @samp{-mrtd} option by using the function attribute
5836 @samp{cdecl}. @xref{Function Attributes}.
5838 @strong{Warning:} this calling convention is incompatible with the one
5839 normally used on Unix, so you cannot use it if you need to call
5840 libraries compiled with the Unix compiler.
5842 Also, you must provide function prototypes for all functions that
5843 take variable numbers of arguments (including @code{printf});
5844 otherwise incorrect code will be generated for calls to those
5847 In addition, seriously incorrect code will result if you call a
5848 function with too many arguments. (Normally, extra arguments are
5849 harmlessly ignored.)
5851 @item -mreg-alloc=@var{regs}
5852 Control the default allocation order of integer registers. The
5853 string @var{regs} is a series of letters specifying a register. The
5854 supported letters are: @code{a} allocate EAX; @code{b} allocate EBX;
5855 @code{c} allocate ECX; @code{d} allocate EDX; @code{S} allocate ESI;
5856 @code{D} allocate EDI; @code{B} allocate EBP.
5858 @item -mregparm=@var{num}
5859 Control how many registers are used to pass integer arguments. By
5860 default, no registers are used to pass arguments, and at most 3
5861 registers can be used. You can control this behavior for a specific
5862 function by using the function attribute @samp{regparm}.
5863 @xref{Function Attributes}.
5865 @strong{Warning:} if you use this switch, and
5866 @var{num} is nonzero, then you must build all modules with the same
5867 value, including any libraries. This includes the system libraries and
5870 @item -malign-loops=@var{num}
5871 Align loops to a 2 raised to a @var{num} byte boundary. If
5872 @samp{-malign-loops} is not specified, the default is 2 unless
5873 gas 2.8 (or later) is being used in which case the default is
5874 to align the loop on a 16 byte boundary if it is less than 8
5877 @item -malign-jumps=@var{num}
5878 Align instructions that are only jumped to to a 2 raised to a @var{num}
5879 byte boundary. If @samp{-malign-jumps} is not specified, the default is
5880 2 if optimizing for a 386, and 4 if optimizing for a 486 unless
5881 gas 2.8 (or later) is being used in which case the default is
5882 to align the instruction on a 16 byte boundary if it is less
5885 @item -malign-functions=@var{num}
5886 Align the start of functions to a 2 raised to @var{num} byte boundary.
5887 If @samp{-malign-functions} is not specified, the default is 2 if optimizing
5888 for a 386, and 4 if optimizing for a 486.
5890 @item -mpreferred-stack-boundary=@var{num}
5891 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
5892 byte boundary. If @samp{-mpreferred-stack-boundary} is not specified,
5893 the default is 4 (16 bytes or 128 bits).
5895 The stack is required to be aligned on a 4 byte boundary. On Pentium
5896 and PentiumPro, @code{double} and @code{long double} values should be
5897 aligned to an 8 byte boundary (see @samp{-malign-double}) or suffer
5898 significant run time performance penalties. On Pentium III, the
5899 Streaming SIMD Extention (SSE) data type @code{__m128} suffers similar
5900 penalties if it is not 16 byte aligned.
5902 To ensure proper alignment of this values on the stack, the stack boundary
5903 must be as aligned as that required by any value stored on the stack.
5904 Further, every function must be generated such that it keeps the stack
5905 aligned. Thus calling a function compiled with a higher preferred
5906 stack boundary from a function compiled with a lower preferred stack
5907 boundary will most likely misalign the stack. It is recommended that
5908 libraries that use callbacks always use the default setting.
5910 This extra alignment does consume extra stack space. Code that is sensitive
5911 to stack space usage, such as embedded systems and operating system kernels,
5912 may want to reduce the preferred alignment to
5913 @samp{-mpreferred-stack-boundary=2}.
5917 @subsection HPPA Options
5918 @cindex HPPA Options
5920 These @samp{-m} options are defined for the HPPA family of computers:
5923 @item -march=@var{architecture type}
5924 Generate code for the specified architecture. The choices for
5925 @var{architecture type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
5926 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
5927 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
5928 architecture option for your machine. Code compiled for lower numbered
5929 architectures will run on higher numbered architectures, but not the
5932 PA 2.0 support currently requires gas snapshot 19990413 or later. The
5933 next release of binutils (current is 2.9.1) will probably contain PA 2.0
5937 @itemx -mpa-risc-1-1
5938 @itemx -mpa-risc-2-0
5939 Synonyms for -march=1.0, -march=1.1, and -march=2.0 respectively.
5942 Generate code suitable for big switch tables. Use this option only if
5943 the assembler/linker complain about out of range branches within a switch
5946 @item -mjump-in-delay
5947 Fill delay slots of function calls with unconditional jump instructions
5948 by modifying the return pointer for the function call to be the target
5949 of the conditional jump.
5951 @item -mdisable-fpregs
5952 Prevent floating point registers from being used in any manner. This is
5953 necessary for compiling kernels which perform lazy context switching of
5954 floating point registers. If you use this option and attempt to perform
5955 floating point operations, the compiler will abort.
5957 @item -mdisable-indexing
5958 Prevent the compiler from using indexing address modes. This avoids some
5959 rather obscure problems when compiling MIG generated code under MACH.
5961 @item -mno-space-regs
5962 Generate code that assumes the target has no space registers. This allows
5963 GCC to generate faster indirect calls and use unscaled index address modes.
5965 Such code is suitable for level 0 PA systems and kernels.
5967 @item -mfast-indirect-calls
5968 Generate code that assumes calls never cross space boundaries. This
5969 allows GCC to emit code which performs faster indirect calls.
5971 This option will not work in the presense of shared libraries or nested
5974 @item -mlong-load-store
5975 Generate 3-instruction load and store sequences as sometimes required by
5976 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
5979 @item -mportable-runtime
5980 Use the portable calling conventions proposed by HP for ELF systems.
5983 Enable the use of assembler directives only GAS understands.
5985 @item -mschedule=@var{cpu type}
5986 Schedule code according to the constraints for the machine type
5987 @var{cpu type}. The choices for @var{cpu type} are @samp{700}
5988 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
5989 @file{/usr/lib/sched.models} on an HP-UX system to determine the
5990 proper scheduling option for your machine.
5993 Enable the optimization pass in the HPUX linker. Note this makes symbolic
5994 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
5995 in which they give bogus error messages when linking some programs.
5998 Generate output containing library calls for floating point.
5999 @strong{Warning:} the requisite libraries are not available for all HPPA
6000 targets. Normally the facilities of the machine's usual C compiler are
6001 used, but this cannot be done directly in cross-compilation. You must make
6002 your own arrangements to provide suitable library functions for
6003 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
6004 does provide software floating point support.
6006 @samp{-msoft-float} changes the calling convention in the output file;
6007 therefore, it is only useful if you compile @emph{all} of a program with
6008 this option. In particular, you need to compile @file{libgcc.a}, the
6009 library that comes with GCC, with @samp{-msoft-float} in order for
6013 @node Intel 960 Options
6014 @subsection Intel 960 Options
6016 These @samp{-m} options are defined for the Intel 960 implementations:
6019 @item -m@var{cpu type}
6020 Assume the defaults for the machine type @var{cpu type} for some of
6021 the other options, including instruction scheduling, floating point
6022 support, and addressing modes. The choices for @var{cpu type} are
6023 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
6024 @samp{sa}, and @samp{sb}.
6030 The @samp{-mnumerics} option indicates that the processor does support
6031 floating-point instructions. The @samp{-msoft-float} option indicates
6032 that floating-point support should not be assumed.
6034 @item -mleaf-procedures
6035 @itemx -mno-leaf-procedures
6036 Do (or do not) attempt to alter leaf procedures to be callable with the
6037 @code{bal} instruction as well as @code{call}. This will result in more
6038 efficient code for explicit calls when the @code{bal} instruction can be
6039 substituted by the assembler or linker, but less efficient code in other
6040 cases, such as calls via function pointers, or using a linker that doesn't
6041 support this optimization.
6044 @itemx -mno-tail-call
6045 Do (or do not) make additional attempts (beyond those of the
6046 machine-independent portions of the compiler) to optimize tail-recursive
6047 calls into branches. You may not want to do this because the detection of
6048 cases where this is not valid is not totally complete. The default is
6049 @samp{-mno-tail-call}.
6051 @item -mcomplex-addr
6052 @itemx -mno-complex-addr
6053 Assume (or do not assume) that the use of a complex addressing mode is a
6054 win on this implementation of the i960. Complex addressing modes may not
6055 be worthwhile on the K-series, but they definitely are on the C-series.
6056 The default is currently @samp{-mcomplex-addr} for all processors except
6060 @itemx -mno-code-align
6061 Align code to 8-byte boundaries for faster fetching (or don't bother).
6062 Currently turned on by default for C-series implementations only.
6065 @item -mclean-linkage
6066 @itemx -mno-clean-linkage
6067 These options are not fully implemented.
6071 @itemx -mic2.0-compat
6072 @itemx -mic3.0-compat
6073 Enable compatibility with iC960 v2.0 or v3.0.
6077 Enable compatibility with the iC960 assembler.
6079 @item -mstrict-align
6080 @itemx -mno-strict-align
6081 Do not permit (do permit) unaligned accesses.
6084 Enable structure-alignment compatibility with Intel's gcc release version
6085 1.3 (based on gcc 1.37). This option implies @samp{-mstrict-align}.
6087 @item -mlong-double-64
6088 Implement type @samp{long double} as 64-bit floating point numbers.
6089 Without the option @samp{long double} is implemented by 80-bit
6090 floating point numbers. The only reason we have it because there is
6091 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
6092 is only useful for people using soft-float targets. Otherwise, we
6093 should recommend against use of it.
6097 @node DEC Alpha Options
6098 @subsection DEC Alpha Options
6100 These @samp{-m} options are defined for the DEC Alpha implementations:
6103 @item -mno-soft-float
6105 Use (do not use) the hardware floating-point instructions for
6106 floating-point operations. When @code{-msoft-float} is specified,
6107 functions in @file{libgcc1.c} will be used to perform floating-point
6108 operations. Unless they are replaced by routines that emulate the
6109 floating-point operations, or compiled in such a way as to call such
6110 emulations routines, these routines will issue floating-point
6111 operations. If you are compiling for an Alpha without floating-point
6112 operations, you must ensure that the library is built so as not to call
6115 Note that Alpha implementations without floating-point operations are
6116 required to have floating-point registers.
6120 Generate code that uses (does not use) the floating-point register set.
6121 @code{-mno-fp-regs} implies @code{-msoft-float}. If the floating-point
6122 register set is not used, floating point operands are passed in integer
6123 registers as if they were integers and floating-point results are passed
6124 in $0 instead of $f0. This is a non-standard calling sequence, so any
6125 function with a floating-point argument or return value called by code
6126 compiled with @code{-mno-fp-regs} must also be compiled with that
6129 A typical use of this option is building a kernel that does not use,
6130 and hence need not save and restore, any floating-point registers.
6133 The Alpha architecture implements floating-point hardware optimized for
6134 maximum performance. It is mostly compliant with the IEEE floating
6135 point standard. However, for full compliance, software assistance is
6136 required. This option generates code fully IEEE compliant code
6137 @emph{except} that the @var{inexact flag} is not maintained (see below).
6138 If this option is turned on, the CPP macro @code{_IEEE_FP} is defined
6139 during compilation. The option is a shorthand for: @samp{-D_IEEE_FP
6140 -mfp-trap-mode=su -mtrap-precision=i -mieee-conformant}. The resulting
6141 code is less efficient but is able to correctly support denormalized
6142 numbers and exceptional IEEE values such as not-a-number and plus/minus
6143 infinity. Other Alpha compilers call this option
6144 @code{-ieee_with_no_inexact}.
6146 @item -mieee-with-inexact
6147 @c overfull hbox here --bob 22 jul96
6148 @c original text between ignore ... end ignore
6150 This is like @samp{-mieee} except the generated code also maintains the
6151 IEEE @var{inexact flag}. Turning on this option causes the generated
6152 code to implement fully-compliant IEEE math. The option is a shorthand
6153 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus @samp{-mieee-conformant},
6154 @samp{-mfp-trap-mode=sui}, and @samp{-mtrap-precision=i}. On some Alpha
6155 implementations the resulting code may execute significantly slower than
6156 the code generated by default. Since there is very little code that
6157 depends on the @var{inexact flag}, you should normally not specify this
6158 option. Other Alpha compilers call this option
6159 @samp{-ieee_with_inexact}.
6161 @c changed paragraph
6162 This is like @samp{-mieee} except the generated code also maintains the
6163 IEEE @var{inexact flag}. Turning on this option causes the generated
6164 code to implement fully-compliant IEEE math. The option is a shorthand
6165 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus the three following:
6166 @samp{-mieee-conformant},
6167 @samp{-mfp-trap-mode=sui},
6168 and @samp{-mtrap-precision=i}.
6169 On some Alpha implementations the resulting code may execute
6170 significantly slower than the code generated by default. Since there
6171 is very little code that depends on the @var{inexact flag}, you should
6172 normally not specify this option. Other Alpha compilers call this
6173 option @samp{-ieee_with_inexact}.
6174 @c end changes to prevent overfull hboxes
6176 @item -mfp-trap-mode=@var{trap mode}
6177 This option controls what floating-point related traps are enabled.
6178 Other Alpha compilers call this option @samp{-fptm }@var{trap mode}.
6179 The trap mode can be set to one of four values:
6183 This is the default (normal) setting. The only traps that are enabled
6184 are the ones that cannot be disabled in software (e.g., division by zero
6188 In addition to the traps enabled by @samp{n}, underflow traps are enabled
6192 Like @samp{su}, but the instructions are marked to be safe for software
6193 completion (see Alpha architecture manual for details).
6196 Like @samp{su}, but inexact traps are enabled as well.
6199 @item -mfp-rounding-mode=@var{rounding mode}
6200 Selects the IEEE rounding mode. Other Alpha compilers call this option
6201 @samp{-fprm }@var{rounding mode}. The @var{rounding mode} can be one
6206 Normal IEEE rounding mode. Floating point numbers are rounded towards
6207 the nearest machine number or towards the even machine number in case
6211 Round towards minus infinity.
6214 Chopped rounding mode. Floating point numbers are rounded towards zero.
6217 Dynamic rounding mode. A field in the floating point control register
6218 (@var{fpcr}, see Alpha architecture reference manual) controls the
6219 rounding mode in effect. The C library initializes this register for
6220 rounding towards plus infinity. Thus, unless your program modifies the
6221 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
6224 @item -mtrap-precision=@var{trap precision}
6225 In the Alpha architecture, floating point traps are imprecise. This
6226 means without software assistance it is impossible to recover from a
6227 floating trap and program execution normally needs to be terminated.
6228 GCC can generate code that can assist operating system trap handlers
6229 in determining the exact location that caused a floating point trap.
6230 Depending on the requirements of an application, different levels of
6231 precisions can be selected:
6235 Program precision. This option is the default and means a trap handler
6236 can only identify which program caused a floating point exception.
6239 Function precision. The trap handler can determine the function that
6240 caused a floating point exception.
6243 Instruction precision. The trap handler can determine the exact
6244 instruction that caused a floating point exception.
6247 Other Alpha compilers provide the equivalent options called
6248 @samp{-scope_safe} and @samp{-resumption_safe}.
6250 @item -mieee-conformant
6251 This option marks the generated code as IEEE conformant. You must not
6252 use this option unless you also specify @samp{-mtrap-precision=i} and either
6253 @samp{-mfp-trap-mode=su} or @samp{-mfp-trap-mode=sui}. Its only effect
6254 is to emit the line @samp{.eflag 48} in the function prologue of the
6255 generated assembly file. Under DEC Unix, this has the effect that
6256 IEEE-conformant math library routines will be linked in.
6258 @item -mbuild-constants
6259 Normally GCC examines a 32- or 64-bit integer constant to
6260 see if it can construct it from smaller constants in two or three
6261 instructions. If it cannot, it will output the constant as a literal and
6262 generate code to load it from the data segment at runtime.
6264 Use this option to require GCC to construct @emph{all} integer constants
6265 using code, even if it takes more instructions (the maximum is six).
6267 You would typically use this option to build a shared library dynamic
6268 loader. Itself a shared library, it must relocate itself in memory
6269 before it can find the variables and constants in its own data segment.
6273 Select whether to generate code to be assembled by the vendor-supplied
6274 assembler (@samp{-malpha-as}) or by the GNU assembler @samp{-mgas}.
6282 Indicate whether GCC should generate code to use the optional BWX,
6283 CIX, and MAX instruction sets. The default is to use the instruction sets
6284 supported by the CPU type specified via @samp{-mcpu=} option or that
6285 of the CPU on which GCC was built if none was specified.
6287 @item -mcpu=@var{cpu_type}
6288 Set the instruction set, register set, and instruction scheduling
6289 parameters for machine type @var{cpu_type}. You can specify either the
6290 @samp{EV} style name or the corresponding chip number. GCC
6291 supports scheduling parameters for the EV4 and EV5 family of processors
6292 and will choose the default values for the instruction set from
6293 the processor you specify. If you do not specify a processor type,
6294 GCC will default to the processor on which the compiler was built.
6296 Supported values for @var{cpu_type} are
6301 Schedules as an EV4 and has no instruction set extensions.
6305 Schedules as an EV5 and has no instruction set extensions.
6309 Schedules as an EV5 and supports the BWX extension.
6314 Schedules as an EV5 and supports the BWX and MAX extensions.
6318 Schedules as an EV5 (until Digital releases the scheduling parameters
6319 for the EV6) and supports the BWX, CIX, and MAX extensions.
6322 @item -mmemory-latency=@var{time}
6323 Sets the latency the scheduler should assume for typical memory
6324 references as seen by the application. This number is highly
6325 dependant on the memory access patterns used by the application
6326 and the size of the external cache on the machine.
6328 Valid options for @var{time} are
6332 A decimal number representing clock cycles.
6338 The compiler contains estimates of the number of clock cycles for
6339 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
6340 (also called Dcache, Scache, and Bcache), as well as to main memory.
6341 Note that L3 is only valid for EV5.
6346 @node Clipper Options
6347 @subsection Clipper Options
6349 These @samp{-m} options are defined for the Clipper implementations:
6353 Produce code for a C300 Clipper processor. This is the default.
6356 Produce code for a C400 Clipper processor i.e. use floating point
6360 @node H8/300 Options
6361 @subsection H8/300 Options
6363 These @samp{-m} options are defined for the H8/300 implementations:
6367 Shorten some address references at link time, when possible; uses the
6368 linker option @samp{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
6369 ld.info, Using ld}, for a fuller description.
6372 Generate code for the H8/300H.
6375 Generate code for the H8/S.
6378 Make @code{int} data 32 bits by default.
6381 On the h8/300h, use the same alignment rules as for the h8/300.
6382 The default for the h8/300h is to align longs and floats on 4 byte boundaries.
6383 @samp{-malign-300} causes them to be aligned on 2 byte boundaries.
6384 This option has no effect on the h8/300.
6388 @subsection SH Options
6390 These @samp{-m} options are defined for the SH implementations:
6394 Generate code for the SH1.
6397 Generate code for the SH2.
6400 Generate code for the SH3.
6403 Generate code for the SH3e.
6406 Compile code for the processor in big endian mode.
6409 Compile code for the processor in little endian mode.
6412 Align doubles at 64 bit boundaries. Note that this changes the calling
6413 conventions, and thus some functions from the standard C library will
6414 not work unless you recompile it first with -mdalign.
6417 Shorten some address references at link time, when possible; uses the
6418 linker option @samp{-relax}.
6421 @node System V Options
6422 @subsection Options for System V
6424 These additional options are available on System V Release 4 for
6425 compatibility with other compilers on those systems:
6429 Create a shared object.
6430 It is recommended that @samp{-symbolic} or @samp{-shared} be used instead.
6433 Identify the versions of each tool used by the compiler, in a
6434 @code{.ident} assembler directive in the output.
6437 Refrain from adding @code{.ident} directives to the output file (this is
6440 @item -YP,@var{dirs}
6441 Search the directories @var{dirs}, and no others, for libraries
6442 specified with @samp{-l}.
6445 Look in the directory @var{dir} to find the M4 preprocessor.
6446 The assembler uses this option.
6447 @c This is supposed to go with a -Yd for predefined M4 macro files, but
6448 @c the generic assembler that comes with Solaris takes just -Ym.
6451 @node TMS320C3x/C4x Options
6452 @subsection TMS320C3x/C4x Options
6453 @cindex TMS320C3x/C4x Options
6455 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
6459 @item -mcpu=@var{cpu_type}
6460 Set the instruction set, register set, and instruction scheduling
6461 parameters for machine type @var{cpu_type}. Supported values for
6462 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
6463 @samp{c44}. The default is @samp{c40} to generate code for the
6468 @itemx -msmall-memory
6470 Generates code for the big or small memory model. The small memory
6471 model assumed that all data fits into one 64K word page. At run-time
6472 the data page (DP) register must be set to point to the 64K page
6473 containing the .bss and .data program sections. The big memory model is
6474 the default and requires reloading of the DP register for every direct
6479 Allow (disallow) allocation of general integer operands into the block
6484 Enable (disable) generation of code using decrement and branch,
6485 DBcond(D), instructions. This is enabled by default for the C4x. To be
6486 on the safe side, this is disabled for the C3x, since the maximum
6487 iteration count on the C3x is 2^23 + 1 (but who iterates loops more than
6488 2^23 times on the C3x?). Note that GCC will try to reverse a loop so
6489 that it can utilise the decrement and branch instruction, but will give
6490 up if there is more than one memory reference in the loop. Thus a loop
6491 where the loop counter is decremented can generate slightly more
6492 efficient code, in cases where the RPTB instruction cannot be utilised.
6494 @item -mdp-isr-reload
6496 Force the DP register to be saved on entry to an interrupt service
6497 routine (ISR), reloaded to point to the data section, and restored on
6498 exit from the ISR. This should not be required unless someone has
6499 violated the small memory model by modifying the DP register, say within
6504 For the C3x use the 24-bit MPYI instruction for integer multiplies
6505 instead of a library call to guarantee 32-bit results. Note that if one
6506 of the operands is a constant, then the multiplication will be performed
6507 using shifts and adds. If the -mmpyi option is not specified for the C3x,
6508 then squaring operations are performed inline instead of a library call.
6511 @itemx -mno-fast-fix
6512 The C3x/C4x FIX instruction to convert a floating point value to an
6513 integer value chooses the nearest integer less than or equal to the
6514 floating point value rather than to the nearest integer. Thus if the
6515 floating point number is negative, the result will be incorrectly
6516 truncated an additional code is necessary to detect and correct this
6517 case. This option can be used to disable generation of the additional
6518 code required to correct the result.
6522 Enable (disable) generation of repeat block sequences using the RPTB
6523 instruction for zero overhead looping. The RPTB construct is only used
6524 for innermost loops that do not call functions or jump across the loop
6525 boundaries. There is no advantage having nested RPTB loops due to the
6526 overhead required to save and restore the RC, RS, and RE registers.
6527 This is enabled by default with -O2.
6529 @item -mrpts=@var{count}
6531 Enable (disable) the use of the single instruction repeat instruction
6532 RPTS. If a repeat block contains a single instruction, and the loop
6533 count can be guaranteed to be less than the value @var{count}, GCC will
6534 emit a RPTS instruction instead of a RPTB. If no value is specified,
6535 then a RPTS will be emitted even if the loop count cannot be determined
6536 at compile time. Note that the repeated instruction following RPTS does
6537 not have to be reloaded from memory each iteration, thus freeing up the
6538 CPU buses for oeprands. However, since interrupts are blocked by this
6539 instruction, it is disabled by default.
6541 @item -mloop-unsigned
6542 @itemx -mno-loop-unsigned
6543 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
6544 is 2^31 + 1 since these instructions test if the iteration count is
6545 negative to terminate the loop. If the iteration count is unsigned
6546 there is a possibility than the 2^31 + 1 maximum iteration count may be
6547 exceeded. This switch allows an unsigned iteration count.
6550 Try to emit an assembler syntax that the TI assembler (asm30) is happy
6551 with. This also enforces compatibility with the API employed by the TI
6552 C3x C compiler. For example, long doubles are passed as structures
6553 rather than in floating point registers.
6557 Generate code that uses registers (stack) for passing arguments to functions.
6558 By default, arguments are passed in registers where possible rather
6559 than by pushing arguments on to the stack.
6561 @item -mparallel-insns
6562 @itemx -mno-parallel-insns
6563 Allow the generation of parallel instructions. This is enabled by
6566 @item -mparallel-mpy
6567 @itemx -mno-parallel-mpy
6568 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
6569 provided -mparallel-insns is also specified. These instructions have
6570 tight register constraints which can pessimize the code generation
6576 @subsection V850 Options
6577 @cindex V850 Options
6579 These @samp{-m} options are defined for V850 implementations:
6583 @itemx -mno-long-calls
6584 Treat all calls as being far away (near). If calls are assumed to be
6585 far away, the compiler will always load the functions address up into a
6586 register, and call indirect through the pointer.
6590 Do not optimize (do optimize) basic blocks that use the same index
6591 pointer 4 or more times to copy pointer into the @code{ep} register, and
6592 use the shorter @code{sld} and @code{sst} instructions. The @samp{-mep}
6593 option is on by default if you optimize.
6595 @item -mno-prolog-function
6596 @itemx -mprolog-function
6597 Do not use (do use) external functions to save and restore registers at
6598 the prolog and epilog of a function. The external functions are slower,
6599 but use less code space if more than one function saves the same number
6600 of registers. The @samp{-mprolog-function} option is on by default if
6604 Try to make the code as small as possible. At present, this just turns
6605 on the @samp{-mep} and @samp{-mprolog-function} options.
6608 Put static or global variables whose size is @var{n} bytes or less into
6609 the tiny data area that register @code{ep} points to. The tiny data
6610 area can hold up to 256 bytes in total (128 bytes for byte references).
6613 Put static or global variables whose size is @var{n} bytes or less into
6614 the small data area that register @code{gp} points to. The small data
6615 area can hold up to 64 kilobytes.
6618 Put static or global variables whose size is @var{n} bytes or less into
6619 the first 32 kilobytes of memory.
6622 Specify that the target processor is the V850.
6625 Generate code suitable for big switch tables. Use this option only if
6626 the assembler/linker complain about out of range branches within a switch
6631 @subsection ARC Options
6634 These options are defined for ARC implementations:
6638 Compile code for little endian mode. This is the default.
6641 Compile code for big endian mode.
6644 Prepend the name of the cpu to all public symbol names.
6645 In multiple-processor systems, there are many ARC variants with different
6646 instruction and register set characteristics. This flag prevents code
6647 compiled for one cpu to be linked with code compiled for another.
6648 No facility exists for handling variants that are "almost identical".
6649 This is an all or nothing option.
6651 @item -mcpu=@var{cpu}
6652 Compile code for ARC variant @var{cpu}.
6653 Which variants are supported depend on the configuration.
6654 All variants support @samp{-mcpu=base}, this is the default.
6656 @item -mtext=@var{text section}
6657 @itemx -mdata=@var{data section}
6658 @itemx -mrodata=@var{readonly data section}
6659 Put functions, data, and readonly data in @var{text section},
6660 @var{data section}, and @var{readonly data section} respectively
6661 by default. This can be overridden with the @code{section} attribute.
6662 @xref{Variable Attributes}.
6667 @subsection NS32K Options
6668 @cindex NS32K options
6670 These are the @samp{-m} options defined for the 32000 series. The default
6671 values for these options depends on which style of 32000 was selected when
6672 the compiler was configured; the defaults for the most common choices are
6678 Generate output for a 32032. This is the default
6679 when the compiler is configured for 32032 and 32016 based systems.
6683 Generate output for a 32332. This is the default
6684 when the compiler is configured for 32332-based systems.
6688 Generate output for a 32532. This is the default
6689 when the compiler is configured for 32532-based systems.
6692 Generate output containing 32081 instructions for floating point.
6693 This is the default for all systems.
6696 Generate output containing 32381 instructions for floating point. This
6697 also implies @samp{-m32081}. The 32381 is only compatible with the 32332
6698 and 32532 cpus. This is the default for the pc532-netbsd configuration.
6701 Try and generate multiply-add floating point instructions @code{polyF}
6702 and @code{dotF}. This option is only available if the @samp{-m32381}
6703 option is in effect. Using these instructions requires changes to to
6704 register allocation which generally has a negative impact on
6705 performance. This option should only be enabled when compiling code
6706 particularly likely to make heavy use of multiply-add instructions.
6709 Do not try and generate multiply-add floating point instructions
6710 @code{polyF} and @code{dotF}. This is the default on all platforms.
6713 Generate output containing library calls for floating point.
6714 @strong{Warning:} the requisite libraries may not be available.
6717 Do not use the bit-field instructions. On some machines it is faster to
6718 use shifting and masking operations. This is the default for the pc532.
6721 Do use the bit-field instructions. This is the default for all platforms
6725 Use a different function-calling convention, in which functions
6726 that take a fixed number of arguments return pop their
6727 arguments on return with the @code{ret} instruction.
6729 This calling convention is incompatible with the one normally
6730 used on Unix, so you cannot use it if you need to call libraries
6731 compiled with the Unix compiler.
6733 Also, you must provide function prototypes for all functions that
6734 take variable numbers of arguments (including @code{printf});
6735 otherwise incorrect code will be generated for calls to those
6738 In addition, seriously incorrect code will result if you call a
6739 function with too many arguments. (Normally, extra arguments are
6740 harmlessly ignored.)
6742 This option takes its name from the 680x0 @code{rtd} instruction.
6746 Use a different function-calling convention where the first two arguments
6747 are passed in registers.
6749 This calling convention is incompatible with the one normally
6750 used on Unix, so you cannot use it if you need to call libraries
6751 compiled with the Unix compiler.
6754 Do not pass any arguments in registers. This is the default for all
6758 It is OK to use the sb as an index register which is always loaded with
6759 zero. This is the default for the pc532-netbsd target.
6762 The sb register is not available for use or has not been initialized to
6763 zero by the run time system. This is the default for all targets except
6764 the pc532-netbsd. It is also implied whenever @samp{-mhimem} or
6765 @samp{-fpic} is set.
6768 Many ns32000 series addressing modes use displacements of up to 512MB.
6769 If an address is above 512MB then displacements from zero can not be used.
6770 This option causes code to be generated which can be loaded above 512MB.
6771 This may be useful for operating systems or ROM code.
6774 Assume code will be loaded in the first 512MB of virtual address space.
6775 This is the default for all platforms.
6782 @node Code Gen Options
6783 @section Options for Code Generation Conventions
6784 @cindex code generation conventions
6785 @cindex options, code generation
6786 @cindex run-time options
6788 These machine-independent options control the interface conventions
6789 used in code generation.
6791 Most of them have both positive and negative forms; the negative form
6792 of @samp{-ffoo} would be @samp{-fno-foo}. In the table below, only
6793 one of the forms is listed---the one which is not the default. You
6794 can figure out the other form by either removing @samp{no-} or adding
6799 Enable exception handling. Generates extra code needed to propagate
6800 exceptions. For some targets, this implies GNU CC will generate frame
6801 unwind information for all functions, which can produce significant data
6802 size overhead, although it does not affect execution. If you do not
6803 specify this option, GNU CC will enable it by default for languages like
6804 C++ which normally require exception handling, and disable itfor
6805 languages like C that do not normally require it. However, you may need
6806 to enable this option when compiling C code that needs to interoperate
6807 properly with exception handlers written in C++. You may also wish to
6808 disable this option if you are compiling older C++ programs that don't
6809 use exception handling.
6811 @item -funwind-tables
6812 Similar to @code{-fexceptions}, except that it will just generate any needed
6813 static data, but will not affect the generated code in any other way.
6814 You will normally not enable this option; instead, a language processor
6815 that needs this handling would enable it on your behalf.
6817 @item -fpcc-struct-return
6818 Return ``short'' @code{struct} and @code{union} values in memory like
6819 longer ones, rather than in registers. This convention is less
6820 efficient, but it has the advantage of allowing intercallability between
6821 GCC-compiled files and files compiled with other compilers.
6823 The precise convention for returning structures in memory depends
6824 on the target configuration macros.
6826 Short structures and unions are those whose size and alignment match
6827 that of some integer type.
6829 @item -freg-struct-return
6830 Use the convention that @code{struct} and @code{union} values are
6831 returned in registers when possible. This is more efficient for small
6832 structures than @samp{-fpcc-struct-return}.
6834 If you specify neither @samp{-fpcc-struct-return} nor its contrary
6835 @samp{-freg-struct-return}, GCC defaults to whichever convention is
6836 standard for the target. If there is no standard convention, GCC
6837 defaults to @samp{-fpcc-struct-return}, except on targets where GCC
6838 is the principal compiler. In those cases, we can choose the standard,
6839 and we chose the more efficient register return alternative.
6842 Allocate to an @code{enum} type only as many bytes as it needs for the
6843 declared range of possible values. Specifically, the @code{enum} type
6844 will be equivalent to the smallest integer type which has enough room.
6846 @item -fshort-double
6847 Use the same size for @code{double} as for @code{float}.
6850 Requests that the data and non-@code{const} variables of this
6851 compilation be shared data rather than private data. The distinction
6852 makes sense only on certain operating systems, where shared data is
6853 shared between processes running the same program, while private data
6854 exists in one copy per process.
6857 Allocate even uninitialized global variables in the data section of the
6858 object file, rather than generating them as common blocks. This has the
6859 effect that if the same variable is declared (without @code{extern}) in
6860 two different compilations, you will get an error when you link them.
6861 The only reason this might be useful is if you wish to verify that the
6862 program will work on other systems which always work this way.
6865 Ignore the @samp{#ident} directive.
6867 @item -fno-gnu-linker
6868 Do not output global initializations (such as C++ constructors and
6869 destructors) in the form used by the GNU linker (on systems where the GNU
6870 linker is the standard method of handling them). Use this option when
6871 you want to use a non-GNU linker, which also requires using the
6872 @code{collect2} program to make sure the system linker includes
6873 constructors and destructors. (@code{collect2} is included in the GCC
6874 distribution.) For systems which @emph{must} use @code{collect2}, the
6875 compiler driver @code{gcc} is configured to do this automatically.
6877 @item -finhibit-size-directive
6878 Don't output a @code{.size} assembler directive, or anything else that
6879 would cause trouble if the function is split in the middle, and the
6880 two halves are placed at locations far apart in memory. This option is
6881 used when compiling @file{crtstuff.c}; you should not need to use it
6885 Put extra commentary information in the generated assembly code to
6886 make it more readable. This option is generally only of use to those
6887 who actually need to read the generated assembly code (perhaps while
6888 debugging the compiler itself).
6890 @samp{-fno-verbose-asm}, the default, causes the
6891 extra information to be omitted and is useful when comparing two assembler
6895 Consider all memory references through pointers to be volatile.
6897 @item -fvolatile-global
6898 Consider all memory references to extern and global data items to
6899 be volatile. GCC does not consider static data items to be volatile
6900 because of this switch.
6902 @item -fvolatile-static
6903 Consider all memory references to static data to be volatile.
6906 @cindex global offset table
6908 Generate position-independent code (PIC) suitable for use in a shared
6909 library, if supported for the target machine. Such code accesses all
6910 constant addresses through a global offset table (GOT). The dynamic
6911 loader resolves the GOT entries when the program starts (the dynamic
6912 loader is not part of GCC; it is part of the operating system). If
6913 the GOT size for the linked executable exceeds a machine-specific
6914 maximum size, you get an error message from the linker indicating that
6915 @samp{-fpic} does not work; in that case, recompile with @samp{-fPIC}
6916 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
6917 on the m68k and RS/6000. The 386 has no such limit.)
6919 Position-independent code requires special support, and therefore works
6920 only on certain machines. For the 386, GCC supports PIC for System V
6921 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
6922 position-independent.
6925 If supported for the target machine, emit position-independent code,
6926 suitable for dynamic linking and avoiding any limit on the size of the
6927 global offset table. This option makes a difference on the m68k, m88k,
6930 Position-independent code requires special support, and therefore works
6931 only on certain machines.
6933 @item -ffixed-@var{reg}
6934 Treat the register named @var{reg} as a fixed register; generated code
6935 should never refer to it (except perhaps as a stack pointer, frame
6936 pointer or in some other fixed role).
6938 @var{reg} must be the name of a register. The register names accepted
6939 are machine-specific and are defined in the @code{REGISTER_NAMES}
6940 macro in the machine description macro file.
6942 This flag does not have a negative form, because it specifies a
6945 @item -fcall-used-@var{reg}
6946 Treat the register named @var{reg} as an allocable register that is
6947 clobbered by function calls. It may be allocated for temporaries or
6948 variables that do not live across a call. Functions compiled this way
6949 will not save and restore the register @var{reg}.
6951 It is an error to used this flag with the frame pointer or stack pointer.
6952 Use of this flag for other registers that have fixed pervasive roles in
6953 the machine's execution model will produce disastrous results.
6955 This flag does not have a negative form, because it specifies a
6958 @item -fcall-saved-@var{reg}
6959 Treat the register named @var{reg} as an allocable register saved by
6960 functions. It may be allocated even for temporaries or variables that
6961 live across a call. Functions compiled this way will save and restore
6962 the register @var{reg} if they use it.
6964 It is an error to used this flag with the frame pointer or stack pointer.
6965 Use of this flag for other registers that have fixed pervasive roles in
6966 the machine's execution model will produce disastrous results.
6968 A different sort of disaster will result from the use of this flag for
6969 a register in which function values may be returned.
6971 This flag does not have a negative form, because it specifies a
6975 Pack all structure members together without holes. Usually you would
6976 not want to use this option, since it makes the code suboptimal, and
6977 the offsets of structure members won't agree with system libraries.
6979 @item -fcheck-memory-usage
6980 Generate extra code to check each memory access. GCC will generate
6981 code that is suitable for a detector of bad memory accesses such as
6984 Normally, you should compile all, or none, of your code with this option.
6986 If you do mix code compiled with and without this option,
6987 you must ensure that all code that has side effects
6988 and that is called by code compiled with this option
6989 is, itself, compiled with this option.
6990 If you do not, you might get erroneous messages from the detector.
6992 If you use functions from a library that have side-effects (such as
6993 @code{read}), you might not be able to recompile the library and
6994 specify this option. In that case, you can enable the
6995 @samp{-fprefix-function-name} option, which requests GCC to encapsulate
6996 your code and make other functions look as if they were compiled with
6997 @samp{-fcheck-memory-usage}. This is done by calling ``stubs'',
6998 which are provided by the detector. If you cannot find or build
6999 stubs for every function you call, you might have to specify
7000 @samp{-fcheck-memory-usage} without @samp{-fprefix-function-name}.
7002 If you specify this option, you can not use the @code{asm} or
7003 @code{__asm__} keywords in functions with memory checking enabled. GNU
7004 CC cannot understand what the @code{asm} statement may do, and therefore
7005 cannot generate the appropriate code, so it will reject it. However, if
7006 you specify the function attribute @code{no_check_memory_usage} (see
7007 @pxref{Function Attributes}, GNU CC will disable memory checking within a
7008 function; you may use @code{asm} statements inside such functions. You
7009 may have an inline expansion of a non-checked function within a checked
7010 function; in that case GNU CC will not generate checks for the inlined
7011 function's memory accesses.
7013 If you move your @code{asm} statements to non-checked inline functions
7014 and they do access memory, you can add calls to the support code in your
7015 inline function, to indicate any reads, writes, or copies being done.
7016 These calls would be similar to those done in the stubs described above.
7018 @item -fprefix-function-name
7019 Request GCC to add a prefix to the symbols generated for function names.
7020 GCC adds a prefix to the names of functions defined as well as
7021 functions called. Code compiled with this option and code compiled
7022 without the option can't be linked together, unless stubs are used.
7024 If you compile the following code with @samp{-fprefix-function-name}
7026 extern void bar (int);
7035 GCC will compile the code as if it was written:
7037 extern void prefix_bar (int);
7041 return prefix_bar (a + 5);
7044 This option is designed to be used with @samp{-fcheck-memory-usage}.
7046 @item -finstrument-functions
7047 Generate instrumentation calls for entry and exit to functions. Just
7048 after function entry and just before function exit, the following
7049 profiling functions will be called with the address of the current
7050 function and its call site. (On some platforms,
7051 @code{__builtin_return_address} does not work beyond the current
7052 function, so the call site information may not be available to the
7053 profiling functions otherwise.)
7056 void __cyg_profile_func_enter (void *this_fn, void *call_site);
7057 void __cyg_profile_func_exit (void *this_fn, void *call_site);
7060 The first argument is the address of the start of the current function,
7061 which may be looked up exactly in the symbol table.
7063 This instrumentation is also done for functions expanded inline in other
7064 functions. The profiling calls will indicate where, conceptually, the
7065 inline function is entered and exited. This means that addressable
7066 versions of such functions must be available. If all your uses of a
7067 function are expanded inline, this may mean an additional expansion of
7068 code size. If you use @samp{extern inline} in your C code, an
7069 addressable version of such functions must be provided. (This is
7070 normally the case anyways, but if you get lucky and the optimizer always
7071 expands the functions inline, you might have gotten away without
7072 providing static copies.)
7074 A function may be given the attribute @code{no_instrument_function}, in
7075 which case this instrumentation will not be done. This can be used, for
7076 example, for the profiling functions listed above, high-priority
7077 interrupt routines, and any functions from which the profiling functions
7078 cannot safely be called (perhaps signal handlers, if the profiling
7079 routines generate output or allocate memory).
7082 Generate code to verify that you do not go beyond the boundary of the
7083 stack. You should specify this flag if you are running in an
7084 environment with multiple threads, but only rarely need to specify it in
7085 a single-threaded environment since stack overflow is automatically
7086 detected on nearly all systems if there is only one stack.
7088 @cindex aliasing of parameters
7089 @cindex parameters, aliased
7090 @item -fargument-alias
7091 @itemx -fargument-noalias
7092 @itemx -fargument-noalias-global
7093 Specify the possible relationships among parameters and between
7094 parameters and global data.
7096 @samp{-fargument-alias} specifies that arguments (parameters) may
7097 alias each other and may alias global storage.
7098 @samp{-fargument-noalias} specifies that arguments do not alias
7099 each other, but may alias global storage.
7100 @samp{-fargument-noalias-global} specifies that arguments do not
7101 alias each other and do not alias global storage.
7103 Each language will automatically use whatever option is required by
7104 the language standard. You should not need to use these options yourself.
7106 @item -fleading-underscore
7107 This option and its counterpart, -fno-leading-underscore, forcibly
7108 change the way C symbols are represented in the object file. One use
7109 is to help link with legacy assembly code.
7111 Be warned that you should know what you are doing when invoking this
7112 option, and that not all targets provide complete support for it.
7115 @node Environment Variables
7116 @section Environment Variables Affecting GCC
7117 @cindex environment variables
7119 This section describes several environment variables that affect how GCC
7120 operates. Some of them work by specifying directories or prefixes to use
7121 when searching for various kinds of files. Some are used to specify other
7122 aspects of the compilation environment.
7125 Note that you can also specify places to search using options such as
7126 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
7127 take precedence over places specified using environment variables, which
7128 in turn take precedence over those specified by the configuration of GCC.
7132 Note that you can also specify places to search using options such as
7133 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
7134 take precedence over places specified using environment variables, which
7135 in turn take precedence over those specified by the configuration of GCC.
7142 @c @itemx LC_COLLATE
7144 @c @itemx LC_MONETARY
7145 @c @itemx LC_NUMERIC
7150 @c @findex LC_COLLATE
7152 @c @findex LC_MONETARY
7153 @c @findex LC_NUMERIC
7157 These environment variables control the way that GCC uses
7158 localization information that allow GCC to work with different
7159 national conventions. GCC inspects the locale categories
7160 @code{LC_CTYPE} and @code{LC_MESSAGES} if it has been configured to do
7161 so. These locale categories can be set to any value supported by your
7162 installation. A typical value is @samp{en_UK} for English in the United
7165 The @code{LC_CTYPE} environment variable specifies character
7166 classification. GCC uses it to determine the character boundaries in
7167 a string; this is needed for some multibyte encodings that contain quote
7168 and escape characters that would otherwise be interpreted as a string
7171 The @code{LC_MESSAGES} environment variable specifies the language to
7172 use in diagnostic messages.
7174 If the @code{LC_ALL} environment variable is set, it overrides the value
7175 of @code{LC_CTYPE} and @code{LC_MESSAGES}; otherwise, @code{LC_CTYPE}
7176 and @code{LC_MESSAGES} default to the value of the @code{LANG}
7177 environment variable. If none of these variables are set, GCC
7178 defaults to traditional C English behavior.
7182 If @code{TMPDIR} is set, it specifies the directory to use for temporary
7183 files. GCC uses temporary files to hold the output of one stage of
7184 compilation which is to be used as input to the next stage: for example,
7185 the output of the preprocessor, which is the input to the compiler
7188 @item GCC_EXEC_PREFIX
7189 @findex GCC_EXEC_PREFIX
7190 If @code{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
7191 names of the subprograms executed by the compiler. No slash is added
7192 when this prefix is combined with the name of a subprogram, but you can
7193 specify a prefix that ends with a slash if you wish.
7195 If @code{GCC_EXEC_PREFIX} is not set, GNU CC will attempt to figure out
7196 an appropriate prefix to use based on the pathname it was invoked with.
7198 If GCC cannot find the subprogram using the specified prefix, it
7199 tries looking in the usual places for the subprogram.
7201 The default value of @code{GCC_EXEC_PREFIX} is
7202 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
7203 of @code{prefix} when you ran the @file{configure} script.
7205 Other prefixes specified with @samp{-B} take precedence over this prefix.
7207 This prefix is also used for finding files such as @file{crt0.o} that are
7210 In addition, the prefix is used in an unusual way in finding the
7211 directories to search for header files. For each of the standard
7212 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
7213 (more precisely, with the value of @code{GCC_INCLUDE_DIR}), GCC tries
7214 replacing that beginning with the specified prefix to produce an
7215 alternate directory name. Thus, with @samp{-Bfoo/}, GCC will search
7216 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
7217 These alternate directories are searched first; the standard directories
7221 @findex COMPILER_PATH
7222 The value of @code{COMPILER_PATH} is a colon-separated list of
7223 directories, much like @code{PATH}. GCC tries the directories thus
7224 specified when searching for subprograms, if it can't find the
7225 subprograms using @code{GCC_EXEC_PREFIX}.
7228 @findex LIBRARY_PATH
7229 The value of @code{LIBRARY_PATH} is a colon-separated list of
7230 directories, much like @code{PATH}. When configured as a native compiler,
7231 GCC tries the directories thus specified when searching for special
7232 linker files, if it can't find them using @code{GCC_EXEC_PREFIX}. Linking
7233 using GCC also uses these directories when searching for ordinary
7234 libraries for the @samp{-l} option (but directories specified with
7235 @samp{-L} come first).
7237 @item C_INCLUDE_PATH
7238 @itemx CPLUS_INCLUDE_PATH
7239 @itemx OBJC_INCLUDE_PATH
7240 @findex C_INCLUDE_PATH
7241 @findex CPLUS_INCLUDE_PATH
7242 @findex OBJC_INCLUDE_PATH
7243 @c @itemx OBJCPLUS_INCLUDE_PATH
7244 These environment variables pertain to particular languages. Each
7245 variable's value is a colon-separated list of directories, much like
7246 @code{PATH}. When GCC searches for header files, it tries the
7247 directories listed in the variable for the language you are using, after
7248 the directories specified with @samp{-I} but before the standard header
7251 @item DEPENDENCIES_OUTPUT
7252 @findex DEPENDENCIES_OUTPUT
7253 @cindex dependencies for make as output
7254 If this variable is set, its value specifies how to output dependencies
7255 for Make based on the header files processed by the compiler. This
7256 output looks much like the output from the @samp{-M} option
7257 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
7258 in addition to the usual results of compilation.
7260 The value of @code{DEPENDENCIES_OUTPUT} can be just a file name, in
7261 which case the Make rules are written to that file, guessing the target
7262 name from the source file name. Or the value can have the form
7263 @samp{@var{file} @var{target}}, in which case the rules are written to
7264 file @var{file} using @var{target} as the target name.
7268 @cindex locale definition
7269 This variable is used to pass locale information to the compiler. One way in
7270 which this information is used is to determine the character set to be used
7271 when character literals, string literals and comments are parsed in C and C++.
7272 When the compiler is configured to allow multibyte characters,
7273 the following values for @code{LANG} are recognized:
7277 Recognize JIS characters.
7279 Recognize SJIS characters.
7281 Recognize EUCJP characters.
7284 If @code{LANG} is not defined, or if it has some other value, then the
7285 compiler will use mblen and mbtowc as defined by the default locale to
7286 recognize and translate multibyte characters.
7289 @node Running Protoize
7290 @section Running Protoize
7292 The program @code{protoize} is an optional part of GNU C. You can use
7293 it to add prototypes to a program, thus converting the program to ANSI
7294 C in one respect. The companion program @code{unprotoize} does the
7295 reverse: it removes argument types from any prototypes that are found.
7297 When you run these programs, you must specify a set of source files as
7298 command line arguments. The conversion programs start out by compiling
7299 these files to see what functions they define. The information gathered
7300 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
7302 After scanning comes actual conversion. The specified files are all
7303 eligible to be converted; any files they include (whether sources or
7304 just headers) are eligible as well.
7306 But not all the eligible files are converted. By default,
7307 @code{protoize} and @code{unprotoize} convert only source and header
7308 files in the current directory. You can specify additional directories
7309 whose files should be converted with the @samp{-d @var{directory}}
7310 option. You can also specify particular files to exclude with the
7311 @samp{-x @var{file}} option. A file is converted if it is eligible, its
7312 directory name matches one of the specified directory names, and its
7313 name within the directory has not been excluded.
7315 Basic conversion with @code{protoize} consists of rewriting most
7316 function definitions and function declarations to specify the types of
7317 the arguments. The only ones not rewritten are those for varargs
7320 @code{protoize} optionally inserts prototype declarations at the
7321 beginning of the source file, to make them available for any calls that
7322 precede the function's definition. Or it can insert prototype
7323 declarations with block scope in the blocks where undeclared functions
7326 Basic conversion with @code{unprotoize} consists of rewriting most
7327 function declarations to remove any argument types, and rewriting
7328 function definitions to the old-style pre-ANSI form.
7330 Both conversion programs print a warning for any function declaration or
7331 definition that they can't convert. You can suppress these warnings
7334 The output from @code{protoize} or @code{unprotoize} replaces the
7335 original source file. The original file is renamed to a name ending
7336 with @samp{.save}. If the @samp{.save} file already exists, then
7337 the source file is simply discarded.
7339 @code{protoize} and @code{unprotoize} both depend on GCC itself to
7340 scan the program and collect information about the functions it uses.
7341 So neither of these programs will work until GCC is installed.
7343 Here is a table of the options you can use with @code{protoize} and
7344 @code{unprotoize}. Each option works with both programs unless
7348 @item -B @var{directory}
7349 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
7350 usual directory (normally @file{/usr/local/lib}). This file contains
7351 prototype information about standard system functions. This option
7352 applies only to @code{protoize}.
7354 @item -c @var{compilation-options}
7355 Use @var{compilation-options} as the options when running @code{gcc} to
7356 produce the @samp{.X} files. The special option @samp{-aux-info} is
7357 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
7359 Note that the compilation options must be given as a single argument to
7360 @code{protoize} or @code{unprotoize}. If you want to specify several
7361 @code{gcc} options, you must quote the entire set of compilation options
7362 to make them a single word in the shell.
7364 There are certain @code{gcc} arguments that you cannot use, because they
7365 would produce the wrong kind of output. These include @samp{-g},
7366 @samp{-O}, @samp{-c}, @samp{-S}, and @samp{-o} If you include these in
7367 the @var{compilation-options}, they are ignored.
7370 Rename files to end in @samp{.C} instead of @samp{.c}.
7371 This is convenient if you are converting a C program to C++.
7372 This option applies only to @code{protoize}.
7375 Add explicit global declarations. This means inserting explicit
7376 declarations at the beginning of each source file for each function
7377 that is called in the file and was not declared. These declarations
7378 precede the first function definition that contains a call to an
7379 undeclared function. This option applies only to @code{protoize}.
7381 @item -i @var{string}
7382 Indent old-style parameter declarations with the string @var{string}.
7383 This option applies only to @code{protoize}.
7385 @code{unprotoize} converts prototyped function definitions to old-style
7386 function definitions, where the arguments are declared between the
7387 argument list and the initial @samp{@{}. By default, @code{unprotoize}
7388 uses five spaces as the indentation. If you want to indent with just
7389 one space instead, use @samp{-i " "}.
7392 Keep the @samp{.X} files. Normally, they are deleted after conversion
7396 Add explicit local declarations. @code{protoize} with @samp{-l} inserts
7397 a prototype declaration for each function in each block which calls the
7398 function without any declaration. This option applies only to
7402 Make no real changes. This mode just prints information about the conversions
7403 that would have been done without @samp{-n}.
7406 Make no @samp{.save} files. The original files are simply deleted.
7407 Use this option with caution.
7409 @item -p @var{program}
7410 Use the program @var{program} as the compiler. Normally, the name
7414 Work quietly. Most warnings are suppressed.
7417 Print the version number, just like @samp{-v} for @code{gcc}.
7420 If you need special compiler options to compile one of your program's
7421 source files, then you should generate that file's @samp{.X} file
7422 specially, by running @code{gcc} on that source file with the
7423 appropriate options and the option @samp{-aux-info}. Then run
7424 @code{protoize} on the entire set of files. @code{protoize} will use
7425 the existing @samp{.X} file because it is newer than the source file.
7429 gcc -Dfoo=bar file1.c -aux-info
7434 You need to include the special files along with the rest in the
7435 @code{protoize} command, even though their @samp{.X} files already
7436 exist, because otherwise they won't get converted.
7438 @xref{Protoize Caveats}, for more information on how to use
7439 @code{protoize} successfully.