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 GNU CC Command Options
7 @cindex GNU CC command options
8 @cindex command options
9 @cindex options, GNU CC command
11 When you invoke GNU CC, 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 GNU CC 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 GNU CC.
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 GNU CC.
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 -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 -fhandle-signatures
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} -fthis-is-variable -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 -Wbad-function-cast
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 -Winline
129 -Wlarger-than-@var{len} -Wlong-long
130 -Wmain -Wmissing-declarations -Wmissing-noreturn
131 -Wmissing-prototypes -Wmultichar -Wnested-externs -Wno-import
132 -Wparentheses -Wpointer-arith -Wredundant-decls
133 -Wreturn-type -Wshadow -Wsign-compare -Wstrict-prototypes
134 -Wswitch -Wtraditional
135 -Wtrigraphs -Wundef -Wuninitialized -Wunused -Wwrite-strings
139 @item Debugging Options
140 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
142 -a -ax -d@var{letters} -fdump-unnumbered -fpretend-float
143 -fprofile-arcs -ftest-coverage
144 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2
145 -ggdb -gstabs -gstabs+ -gxcoff -gxcoff+
146 -p -pg -print-file-name=@var{library} -print-libgcc-file-name
147 -print-prog-name=@var{program} -print-search-dirs -save-temps
150 @item Optimization Options
151 @xref{Optimize Options,,Options that Control Optimization}.
153 -fbranch-probabilities -foptimize-register-moves
154 -fcaller-saves -fcse-follow-jumps -fcse-skip-blocks
155 -fdelayed-branch -fexpensive-optimizations
156 -ffast-math -ffloat-store -fforce-addr -fforce-mem
157 -fdata-sections -ffunction-sections -fgcse
158 -finline-functions -finline-limit-@var{n} -fkeep-inline-functions
159 -fno-default-inline -fno-defer-pop -fno-function-cse
160 -fno-inline -fno-peephole -fomit-frame-pointer -fregmove
161 -frerun-cse-after-loop -frerun-loop-opt -fschedule-insns
162 -fschedule-insns2 -fstrength-reduce -fthread-jumps
163 -funroll-all-loops -funroll-loops
164 -fmove-all-movables -freduce-all-givs -fstrict-aliasing
165 -O -O0 -O1 -O2 -O3 -Os
168 @item Preprocessor Options
169 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
171 -A@var{question}(@var{answer}) -C -dD -dM -dN
172 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H
174 -include @var{file} -imacros @var{file}
175 -iprefix @var{file} -iwithprefix @var{dir}
176 -iwithprefixbefore @var{dir} -isystem @var{dir} -isystem-c++ @var{dir}
177 -M -MD -MM -MMD -MG -nostdinc -P -trigraphs
178 -undef -U@var{macro} -Wp,@var{option}
181 @item Assembler Option
182 @xref{Assembler Options,,Passing Options to the Assembler}.
188 @xref{Link Options,,Options for Linking}.
190 @var{object-file-name} -l@var{library}
191 -nostartfiles -nodefaultlibs -nostdlib
192 -s -static -shared -symbolic
193 -Wl,@var{option} -Xlinker @var{option}
197 @item Directory Options
198 @xref{Directory Options,,Options for Directory Search}.
200 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}
204 @c I wrote this xref this way to avoid overfull hbox. -- rms
205 @xref{Target Options}.
207 -b @var{machine} -V @var{version}
210 @item Machine Dependent Options
211 @xref{Submodel Options,,Hardware Models and Configurations}.
213 @emph{M680x0 Options}
214 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040
215 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020
216 -mfpa -mnobitfield -mrtd -mshort -msoft-float -mpcrel
224 -mtune=@var{cpu type}
225 -mcmodel=@var{code model}
226 -malign-jumps=@var{num} -malign-loops=@var{num}
227 -malign-functions=@var{num}
229 -mapp-regs -mbroken-saverestore -mcypress -mepilogue
230 -mflat -mfpu -mhard-float -mhard-quad-float
231 -mimpure-text -mlive-g0 -mno-app-regs -mno-epilogue
232 -mno-flat -mno-fpu -mno-impure-text
233 -mno-stack-bias -mno-unaligned-doubles
234 -msoft-float -msoft-quad-float -msparclite -mstack-bias
235 -msupersparc -munaligned-doubles -mv8
237 @emph{Convex Options}
238 -mc1 -mc2 -mc32 -mc34 -mc38
239 -margcount -mnoargcount
241 -mvolatile-cache -mvolatile-nocache
243 @emph{AMD29K Options}
244 -m29000 -m29050 -mbw -mnbw -mdw -mndw
245 -mlarge -mnormal -msmall
246 -mkernel-registers -mno-reuse-arg-regs
247 -mno-stack-check -mno-storem-bug
248 -mreuse-arg-regs -msoft-float -mstack-check
249 -mstorem-bug -muser-registers
252 -mapcs-frame -mno-apcs-frame
254 -mapcs-stack-check -mno-apcs-stack-check
255 -mapcs-float -mno-apcs-float
256 -mapcs-reentrant -mno-apcs-reentrant
257 -msched-prolog -mno-sched-prolog
258 -mlittle-endian -mbig-endian -mwords-little-endian
259 -mshort-load-bytes -mno-short-load-bytes -mshort-load-words -mno-short-load-words
260 -msoft-float -mhard-float -mfpe
261 -mthumb-interwork -mno-thumb-interwork
262 -mcpu= -march= -mfpe=
263 -mstructure-size-boundary=
264 -mbsd -mxopen -mno-symrename
267 -mnop-fun-dllimport -mno-nop-fun-dllimport
270 -mtpcs-frame -mno-tpcs-frame
271 -mtpcs-leaf-frame -mno-tpcs-leaf-frame
272 -mlittle-endian -mbig-endian
273 -mthumb-interwork -mno-thumb-interwork
274 -mstructure-size-boundary=
275 -mnop-fun-dllimport -mno-nop-fun-dllimport
276 -mcallee-super-interworking -mno-callee-super-interworking
277 -mcaller-super-interworking -mno-caller-super-interworking
279 @emph{MN10200 Options}
282 @emph{MN10300 Options}
287 @emph{M32R/D Options}
288 -mcode-model=@var{model type} -msdata=@var{sdata type}
292 -m88000 -m88100 -m88110 -mbig-pic
293 -mcheck-zero-division -mhandle-large-shift
294 -midentify-revision -mno-check-zero-division
295 -mno-ocs-debug-info -mno-ocs-frame-position
296 -mno-optimize-arg-area -mno-serialize-volatile
297 -mno-underscores -mocs-debug-info
298 -mocs-frame-position -moptimize-arg-area
299 -mserialize-volatile -mshort-data-@var{num} -msvr3
300 -msvr4 -mtrap-large-shift -muse-div-instruction
301 -mversion-03.00 -mwarn-passed-structs
303 @emph{RS/6000 and PowerPC Options}
305 -mtune=@var{cpu type}
306 -mpower -mno-power -mpower2 -mno-power2
307 -mpowerpc -mno-powerpc
308 -mpowerpc-gpopt -mno-powerpc-gpopt
309 -mpowerpc-gfxopt -mno-powerpc-gfxopt
310 -mnew-mnemonics -mno-new-mnemonics
311 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc
312 -maix64 -maix32 -mxl-call -mno-xl-call -mthreads -mpe
313 -msoft-float -mhard-float -mmultiple -mno-multiple
314 -mstring -mno-string -mupdate -mno-update
315 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align
316 -mstrict-align -mno-strict-align -mrelocatable
317 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib
318 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian
319 -mcall-aix -mcall-sysv -mprototype -mno-prototype
320 -msim -mmvme -mads -myellowknife -memb -msdata
321 -msdata=@var{opt} -G @var{num}
324 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs
325 -mfull-fp-blocks -mhc-struct-return -min-line-mul
326 -mminimum-fp-blocks -mnohc-struct-return
329 -mabicalls -mcpu=@var{cpu type} -membedded-data
330 -membedded-pic -mfp32 -mfp64 -mgas -mgp32 -mgp64
331 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1
332 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy
333 -mmips-as -mmips-tfile -mno-abicalls
334 -mno-embedded-data -mno-embedded-pic
335 -mno-gpopt -mno-long-calls
336 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats
337 -mrnames -msoft-float
338 -m4650 -msingle-float -mmad
339 -mstats -EL -EB -G @var{num} -nocpp
340 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi
344 -march=@var{cpu type}
345 -mieee-fp -mno-fancy-math-387
346 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib
347 -mno-wide-multiply -mrtd -malign-double
348 -mreg-alloc=@var{list} -mregparm=@var{num}
349 -malign-jumps=@var{num} -malign-loops=@var{num}
350 -malign-functions=@var{num} -mpreferred-stack-boundary=@var{num}
353 -march=@var{architecture type}
354 -mbig-switch -mdisable-fpregs -mdisable-indexing
355 -mfast-indirect-calls -mgas -mjump-in-delay
356 -mlong-load-store -mno-big-switch -mno-disable-fpregs
357 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas
358 -mno-jump-in-delay -mno-long-load-store
359 -mno-portable-runtime -mno-soft-float
360 -mno-space-regs -msoft-float -mpa-risc-1-0
361 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime
362 -mschedule=@var{cpu type} -mspace-regs
364 @emph{Intel 960 Options}
365 -m@var{cpu type} -masm-compat -mclean-linkage
366 -mcode-align -mcomplex-addr -mleaf-procedures
367 -mic-compat -mic2.0-compat -mic3.0-compat
368 -mintel-asm -mno-clean-linkage -mno-code-align
369 -mno-complex-addr -mno-leaf-procedures
370 -mno-old-align -mno-strict-align -mno-tail-call
371 -mnumerics -mold-align -msoft-float -mstrict-align
374 @emph{DEC Alpha Options}
375 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float
377 -mieee -mieee-with-inexact -mieee-conformant
378 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode}
379 -mtrap-precision=@var{mode} -mbuild-constants
381 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max
382 -mmemory-latency=@var{time}
384 @emph{Clipper Options}
387 @emph{H8/300 Options}
388 -mrelax -mh -ms -mint32 -malign-300
391 -m1 -m2 -m3 -m3e -mb -ml -mdalign -mrelax
393 @emph{System V Options}
394 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}
398 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text section}
399 -mdata=@var{data section} -mrodata=@var{readonly data section}
401 @emph{TMS320C3x/C4x Options}
402 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm
403 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload
404 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned
405 -mparallel-insns -mparallel-mpy -mpreserve-float
408 -mlong-calls -mno-long-calls -mep -mno-ep
409 -mprolog-function -mno-prolog-function -mspace
410 -mtda=@var{n} -msda=@var{n} -mzda=@var{n}
414 -m32032 -m32332 -m32532 -m32081 -m32381 -mmult-add -mnomult-add
415 -msoft-float -mrtd -mnortd -mregparam -mnoregparam -msb -mnosb
416 -mbitfield -mnobitfield -mhimem -mnohimem
419 @item Code Generation Options
420 @xref{Code Gen Options,,Options for Code Generation Conventions}.
422 -fcall-saved-@var{reg} -fcall-used-@var{reg}
423 -fexceptions -ffixed-@var{reg} -finhibit-size-directive
424 -fcheck-memory-usage -fprefix-function-name
425 -fno-common -fno-ident -fno-gnu-linker
426 -fpcc-struct-return -fpic -fPIC
427 -freg-struct-return -fshared-data -fshort-enums
428 -fshort-double -fvolatile -fvolatile-global -fvolatile-static
429 -fverbose-asm -fpack-struct -fstack-check
430 -fargument-alias -fargument-noalias
431 -fargument-noalias-global
437 * Overall Options:: Controlling the kind of output:
438 an executable, object files, assembler files,
439 or preprocessed source.
440 * C Dialect Options:: Controlling the variant of C language compiled.
441 * C++ Dialect Options:: Variations on C++.
442 * Warning Options:: How picky should the compiler be?
443 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
444 * Optimize Options:: How much optimization?
445 * Preprocessor Options:: Controlling header files and macro definitions.
446 Also, getting dependency information for Make.
447 * Assembler Options:: Passing options to the assembler.
448 * Link Options:: Specifying libraries and so on.
449 * Directory Options:: Where to find header files and libraries.
450 Where to find the compiler executable files.
451 * Spec Files:: How to pass switches to sub-processes.
452 * Target Options:: Running a cross-compiler, or an old version of GNU CC.
455 @node Overall Options
456 @section Options Controlling the Kind of Output
458 Compilation can involve up to four stages: preprocessing, compilation
459 proper, assembly and linking, always in that order. The first three
460 stages apply to an individual source file, and end by producing an
461 object file; linking combines all the object files (those newly
462 compiled, and those specified as input) into an executable file.
464 @cindex file name suffix
465 For any given input file, the file name suffix determines what kind of
470 C source code which must be preprocessed.
473 C source code which should not be preprocessed.
476 C++ source code which should not be preprocessed.
479 Objective-C source code. Note that you must link with the library
480 @file{libobjc.a} to make an Objective-C program work.
483 C header file (not to be compiled or linked).
486 @itemx @var{file}.cxx
487 @itemx @var{file}.cpp
489 C++ source code which must be preprocessed. Note that in @samp{.cxx},
490 the last two letters must both be literally @samp{x}. Likewise,
491 @samp{.C} refers to a literal capital C.
497 Assembler code which must be preprocessed.
500 An object file to be fed straight into linking.
501 Any file name with no recognized suffix is treated this way.
504 You can specify the input language explicitly with the @samp{-x} option:
507 @item -x @var{language}
508 Specify explicitly the @var{language} for the following input files
509 (rather than letting the compiler choose a default based on the file
510 name suffix). This option applies to all following input files until
511 the next @samp{-x} option. Possible values for @var{language} are:
514 c-header cpp-output c++-cpp-output
515 assembler assembler-with-cpp
519 Turn off any specification of a language, so that subsequent files are
520 handled according to their file name suffixes (as they are if @samp{-x}
521 has not been used at all).
524 If you only want some of the stages of compilation, you can use
525 @samp{-x} (or filename suffixes) to tell @code{gcc} where to start, and
526 one of the options @samp{-c}, @samp{-S}, or @samp{-E} to say where
527 @code{gcc} is to stop. Note that some combinations (for example,
528 @samp{-x cpp-output -E} instruct @code{gcc} to do nothing at all.
532 Compile or assemble the source files, but do not link. The linking
533 stage simply is not done. The ultimate output is in the form of an
534 object file for each source file.
536 By default, the object file name for a source file is made by replacing
537 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
539 Unrecognized input files, not requiring compilation or assembly, are
543 Stop after the stage of compilation proper; do not assemble. The output
544 is in the form of an assembler code file for each non-assembler input
547 By default, the assembler file name for a source file is made by
548 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
550 Input files that don't require compilation are ignored.
553 Stop after the preprocessing stage; do not run the compiler proper. The
554 output is in the form of preprocessed source code, which is sent to the
557 Input files which don't require preprocessing are ignored.
559 @cindex output file option
561 Place output in file @var{file}. This applies regardless to whatever
562 sort of output is being produced, whether it be an executable file,
563 an object file, an assembler file or preprocessed C code.
565 Since only one output file can be specified, it does not make sense to
566 use @samp{-o} when compiling more than one input file, unless you are
567 producing an executable file as output.
569 If @samp{-o} is not specified, the default is to put an executable file
570 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
571 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
572 all preprocessed C source on standard output.@refill
575 Print (on standard error output) the commands executed to run the stages
576 of compilation. Also print the version number of the compiler driver
577 program and of the preprocessor and the compiler proper.
580 Use pipes rather than temporary files for communication between the
581 various stages of compilation. This fails to work on some systems where
582 the assembler is unable to read from a pipe; but the GNU assembler has
586 Print (on the standard output) a description of the command line options
587 understood by @code{gcc}. If the @code{-v} option is also specified
588 then @code{--help} will also be passed on to the various processes
589 invoked by @code{gcc}, so that they can display the command line options
590 they accept. If the @code{-W} option is also specified then command
591 line options which have no documentation associated with them will also
596 @section Compiling C++ Programs
598 @cindex suffixes for C++ source
599 @cindex C++ source file suffixes
600 C++ source files conventionally use one of the suffixes @samp{.C},
601 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
602 preprocessed C++ files use the suffix @samp{.ii}. GNU CC recognizes
603 files with these names and compiles them as C++ programs even if you
604 call the compiler the same way as for compiling C programs (usually with
605 the name @code{gcc}).
609 However, C++ programs often require class libraries as well as a
610 compiler that understands the C++ language---and under some
611 circumstances, you might want to compile programs from standard input,
612 or otherwise without a suffix that flags them as C++ programs.
613 @code{g++} is a program that calls GNU CC with the default language
614 set to C++, and automatically specifies linking against the C++
615 library. On many systems, the script @code{g++} is also
616 installed with the name @code{c++}.
618 @cindex invoking @code{g++}
619 When you compile C++ programs, you may specify many of the same
620 command-line options that you use for compiling programs in any
621 language; or command-line options meaningful for C and related
622 languages; or options that are meaningful only for C++ programs.
623 @xref{C Dialect Options,,Options Controlling C Dialect}, for
624 explanations of options for languages related to C.
625 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
626 explanations of options that are meaningful only for C++ programs.
628 @node C Dialect Options
629 @section Options Controlling C Dialect
630 @cindex dialect options
631 @cindex language dialect options
632 @cindex options, dialect
634 The following options control the dialect of C (or languages derived
635 from C, such as C++ and Objective C) that the compiler accepts:
640 Support all ANSI standard C programs.
642 This turns off certain features of GNU C that are incompatible with ANSI
643 C, such as the @code{asm}, @code{inline} and @code{typeof} keywords, and
644 predefined macros such as @code{unix} and @code{vax} that identify the
645 type of system you are using. It also enables the undesirable and
646 rarely used ANSI trigraph feature, and it disables recognition of C++
647 style @samp{//} comments.
649 The alternate keywords @code{__asm__}, @code{__extension__},
650 @code{__inline__} and @code{__typeof__} continue to work despite
651 @samp{-ansi}. You would not want to use them in an ANSI C program, of
652 course, but it is useful to put them in header files that might be included
653 in compilations done with @samp{-ansi}. Alternate predefined macros
654 such as @code{__unix__} and @code{__vax__} are also available, with or
655 without @samp{-ansi}.
657 The @samp{-ansi} option does not cause non-ANSI programs to be
658 rejected gratuitously. For that, @samp{-pedantic} is required in
659 addition to @samp{-ansi}. @xref{Warning Options}.
661 The macro @code{__STRICT_ANSI__} is predefined when the @samp{-ansi}
662 option is used. Some header files may notice this macro and refrain
663 from declaring certain functions or defining certain macros that the
664 ANSI standard doesn't call for; this is to avoid interfering with any
665 programs that might use these names for other things.
667 The functions @code{alloca}, @code{abort}, @code{exit}, and
668 @code{_exit} are not builtin functions when @samp{-ansi} is used.
671 Enable support for features found in the C9X standard. In particular,
672 enable support for the C9X @code{restrict} keyword.
674 Even when this option is not specified, you can still use some C9X
675 features in so far as they do not conflict with previous C standards.
676 For example, you may use @code{__restrict__} even when -flang-isoc9x
680 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
681 keyword, so that code can use these words as identifiers. You can use
682 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
683 instead. @samp{-ansi} implies @samp{-fno-asm}.
685 In C++, this switch only affects the @code{typeof} keyword, since
686 @code{asm} and @code{inline} are standard keywords. You may want to
687 use the @samp{-fno-gnu-keywords} flag instead, as it also disables the
688 other, C++-specific, extension keywords such as @code{headof}.
691 @cindex builtin functions
707 Don't recognize builtin functions that do not begin with @samp{__builtin_}
708 as prefix. Currently, the functions affected include @code{abort},
709 @code{abs}, @code{alloca}, @code{cos}, @code{exit}, @code{fabs},
710 @code{ffs}, @code{labs}, @code{memcmp}, @code{memcpy}, @code{sin},
711 @code{sqrt}, @code{strcmp}, @code{strcpy}, and @code{strlen}.
713 GCC normally generates special code to handle certain builtin functions
714 more efficiently; for instance, calls to @code{alloca} may become single
715 instructions that adjust the stack directly, and calls to @code{memcpy}
716 may become inline copy loops. The resulting code is often both smaller
717 and faster, but since the function calls no longer appear as such, you
718 cannot set a breakpoint on those calls, nor can you change the behavior
719 of the functions by linking with a different library.
721 The @samp{-ansi} option prevents @code{alloca} and @code{ffs} from being
722 builtin functions, since these functions do not have an ANSI standard
726 @cindex hosted environment
728 Assert that compilation takes place in a hosted environment. This implies
729 @samp{-fbuiltin}. A hosted environment is one in which the
730 entire standard library is available, and in which @code{main} has a return
731 type of @code{int}. Examples are nearly everything except a kernel.
732 This is equivalent to @samp{-fno-freestanding}.
735 @cindex hosted environment
737 Assert that compilation takes place in a freestanding environment. This
738 implies @samp{-fno-builtin}. A freestanding environment
739 is one in which the standard library may not exist, and program startup may
740 not necessarily be at @code{main}. The most obvious example is an OS kernel.
741 This is equivalent to @samp{-fno-hosted}.
744 Support ANSI C trigraphs. You don't want to know about this
745 brain-damage. The @samp{-ansi} option implies @samp{-trigraphs}.
747 @cindex traditional C language
748 @cindex C language, traditional
750 Attempt to support some aspects of traditional C compilers.
755 All @code{extern} declarations take effect globally even if they
756 are written inside of a function definition. This includes implicit
757 declarations of functions.
760 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
761 and @code{volatile} are not recognized. (You can still use the
762 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
766 Comparisons between pointers and integers are always allowed.
769 Integer types @code{unsigned short} and @code{unsigned char} promote
770 to @code{unsigned int}.
773 Out-of-range floating point literals are not an error.
776 Certain constructs which ANSI regards as a single invalid preprocessing
777 number, such as @samp{0xe-0xd}, are treated as expressions instead.
780 String ``constants'' are not necessarily constant; they are stored in
781 writable space, and identical looking constants are allocated
782 separately. (This is the same as the effect of
783 @samp{-fwritable-strings}.)
785 @cindex @code{longjmp} and automatic variables
787 All automatic variables not declared @code{register} are preserved by
788 @code{longjmp}. Ordinarily, GNU C follows ANSI C: automatic variables
789 not declared @code{volatile} may be clobbered.
794 @cindex escape sequences, traditional
795 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
796 literal characters @samp{x} and @samp{a} respectively. Without
797 @w{@samp{-traditional}}, @samp{\x} is a prefix for the hexadecimal
798 representation of a character, and @samp{\a} produces a bell.
801 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
802 if your program uses names that are normally GNU C builtin functions for
803 other purposes of its own.
805 You cannot use @samp{-traditional} if you include any header files that
806 rely on ANSI C features. Some vendors are starting to ship systems with
807 ANSI C header files and you cannot use @samp{-traditional} on such
808 systems to compile files that include any system headers.
810 The @samp{-traditional} option also enables @samp{-traditional-cpp},
811 which is described next.
813 @item -traditional-cpp
814 Attempt to support some aspects of traditional C preprocessors.
819 Comments convert to nothing at all, rather than to a space. This allows
820 traditional token concatenation.
823 In a preprocessing directive, the @samp{#} symbol must appear as the first
827 Macro arguments are recognized within string constants in a macro
828 definition (and their values are stringified, though without additional
829 quote marks, when they appear in such a context). The preprocessor
830 always considers a string constant to end at a newline.
833 @cindex detecting @w{@samp{-traditional}}
834 The predefined macro @code{__STDC__} is not defined when you use
835 @samp{-traditional}, but @code{__GNUC__} is (since the GNU extensions
836 which @code{__GNUC__} indicates are not affected by
837 @samp{-traditional}). If you need to write header files that work
838 differently depending on whether @samp{-traditional} is in use, by
839 testing both of these predefined macros you can distinguish four
840 situations: GNU C, traditional GNU C, other ANSI C compilers, and other
841 old C compilers. The predefined macro @code{__STDC_VERSION__} is also
842 not defined when you use @samp{-traditional}. @xref{Standard
843 Predefined,,Standard Predefined Macros,cpp.info,The C Preprocessor},
844 for more discussion of these and other predefined macros.
847 @cindex string constants vs newline
848 @cindex newline vs string constants
849 The preprocessor considers a string constant to end at a newline (unless
850 the newline is escaped with @samp{\}). (Without @w{@samp{-traditional}},
851 string constants can contain the newline character as typed.)
854 @item -fcond-mismatch
855 Allow conditional expressions with mismatched types in the second and
856 third arguments. The value of such an expression is void.
858 @item -funsigned-char
859 Let the type @code{char} be unsigned, like @code{unsigned char}.
861 Each kind of machine has a default for what @code{char} should
862 be. It is either like @code{unsigned char} by default or like
863 @code{signed char} by default.
865 Ideally, a portable program should always use @code{signed char} or
866 @code{unsigned char} when it depends on the signedness of an object.
867 But many programs have been written to use plain @code{char} and
868 expect it to be signed, or expect it to be unsigned, depending on the
869 machines they were written for. This option, and its inverse, let you
870 make such a program work with the opposite default.
872 The type @code{char} is always a distinct type from each of
873 @code{signed char} or @code{unsigned char}, even though its behavior
874 is always just like one of those two.
877 Let the type @code{char} be signed, like @code{signed char}.
879 Note that this is equivalent to @samp{-fno-unsigned-char}, which is
880 the negative form of @samp{-funsigned-char}. Likewise, the option
881 @samp{-fno-signed-char} is equivalent to @samp{-funsigned-char}.
883 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
884 if your program uses names that are normally GNU C builtin functions for
885 other purposes of its own.
887 You cannot use @samp{-traditional} if you include any header files that
888 rely on ANSI C features. Some vendors are starting to ship systems with
889 ANSI C header files and you cannot use @samp{-traditional} on such
890 systems to compile files that include any system headers.
892 @item -fsigned-bitfields
893 @itemx -funsigned-bitfields
894 @itemx -fno-signed-bitfields
895 @itemx -fno-unsigned-bitfields
896 These options control whether a bitfield is signed or unsigned, when the
897 declaration does not use either @code{signed} or @code{unsigned}. By
898 default, such a bitfield is signed, because this is consistent: the
899 basic integer types such as @code{int} are signed types.
901 However, when @samp{-traditional} is used, bitfields are all unsigned
904 @item -fwritable-strings
905 Store string constants in the writable data segment and don't uniquize
906 them. This is for compatibility with old programs which assume they can
907 write into string constants. The option @samp{-traditional} also has
910 Writing into string constants is a very bad idea; ``constants'' should
913 @item -fallow-single-precision
914 Do not promote single precision math operations to double precision,
915 even when compiling with @samp{-traditional}.
917 Traditional K&R C promotes all floating point operations to double
918 precision, regardless of the sizes of the operands. On the
919 architecture for which you are compiling, single precision may be faster
920 than double precision. If you must use @samp{-traditional}, but want
921 to use single precision operations when the operands are single
922 precision, use this option. This option has no effect when compiling
923 with ANSI or GNU C conventions (the default).
927 @node C++ Dialect Options
928 @section Options Controlling C++ Dialect
930 @cindex compiler options, C++
931 @cindex C++ options, command line
933 This section describes the command-line options that are only meaningful
934 for C++ programs; but you can also use most of the GNU compiler options
935 regardless of what language your program is in. For example, you
936 might compile a file @code{firstClass.C} like this:
939 g++ -g -frepo -O -c firstClass.C
943 In this example, only @samp{-frepo} is an option meant
944 only for C++ programs; you can use the other options with any
945 language supported by GNU CC.
947 Here is a list of options that are @emph{only} for compiling C++ programs:
950 @item -fno-access-control
951 Turn off all access checking. This switch is mainly useful for working
952 around bugs in the access control code.
955 Check that the pointer returned by @code{operator new} is non-null
956 before attempting to modify the storage allocated. The current Working
957 Paper requires that @code{operator new} never return a null pointer, so
958 this check is normally unnecessary.
960 An alternative to using this option is to specify that your
961 @code{operator new} does not throw any exceptions; if you declare it
962 @samp{throw()}, g++ will check the return value. See also @samp{new
965 @item -fconserve-space
966 Put uninitialized or runtime-initialized global variables into the
967 common segment, as C does. This saves space in the executable at the
968 cost of not diagnosing duplicate definitions. If you compile with this
969 flag and your program mysteriously crashes after @code{main()} has
970 completed, you may have an object that is being destroyed twice because
971 two definitions were merged.
973 This option is no longer useful on most targets, now that support has
974 been added for putting variables into BSS without making them common.
976 @item -fdollars-in-identifiers
977 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
978 @samp{$} with the option @samp{-fno-dollars-in-identifiers}. (GNU C allows
979 @samp{$} by default on most target systems, but there are a few exceptions.)
980 Traditional C allowed the character @samp{$} to form part of
981 identifiers. However, ANSI C and C++ forbid @samp{$} in identifiers.
983 @item -fno-elide-constructors
984 The C++ standard allows an implementation to omit creating a temporary
985 which is only used to initialize another object of the same type.
986 Specifying this option disables that optimization, and forces g++ to
987 call the copy constructor in all cases.
989 @item -fexternal-templates
990 Cause template instantiations to obey @samp{#pragma interface} and
991 @samp{implementation}; template instances are emitted or not according
992 to the location of the template definition. @xref{Template
993 Instantiation}, for more information.
995 This option is deprecated.
997 @item -falt-external-templates
998 Similar to -fexternal-templates, but template instances are emitted or
999 not according to the place where they are first instantiated.
1000 @xref{Template Instantiation}, for more information.
1002 This option is deprecated.
1005 @itemx -fno-for-scope
1006 If -ffor-scope is specified, the scope of variables declared in
1007 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1008 as specified by the draft C++ standard.
1009 If -fno-for-scope is specified, the scope of variables declared in
1010 a @i{for-init-statement} extends to the end of the enclosing scope,
1011 as was the case in old versions of gcc, and other (traditional)
1012 implementations of C++.
1014 The default if neither flag is given to follow the standard,
1015 but to allow and give a warning for old-style code that would
1016 otherwise be invalid, or have different behavior.
1018 @item -fno-gnu-keywords
1019 Do not recognize @code{classof}, @code{headof}, @code{signature},
1020 @code{sigof} or @code{typeof} as a keyword, so that code can use these
1021 words as identifiers. You can use the keywords @code{__classof__},
1022 @code{__headof__}, @code{__signature__}, @code{__sigof__}, and
1023 @code{__typeof__} instead. @samp{-ansi} implies
1024 @samp{-fno-gnu-keywords}.
1026 @item -fguiding-decls
1027 Treat a function declaration with the same type as a potential function
1028 template instantiation as though it declares that instantiation, not a
1029 normal function. If a definition is given for the function later in the
1030 translation unit (or another translation unit if the target supports
1031 weak symbols), that definition will be used; otherwise the template will
1032 be instantiated. This behavior reflects the C++ language prior to
1033 September 1996, when guiding declarations were removed.
1035 This option implies @samp{-fname-mangling-version-0}, and will not work
1036 with other name mangling versions. Like all options that change the
1037 ABI, all C++ code, @emph{including libgcc.a} must be built with the same
1038 setting of this option.
1040 @item -fhandle-signatures
1041 Recognize the @code{signature} and @code{sigof} keywords for specifying
1042 abstract types. The default (@samp{-fno-handle-signatures}) is not to
1043 recognize them. @xref{C++ Signatures, Type Abstraction using
1047 Treat the @code{namespace std} as a namespace, instead of ignoring
1048 it. For compatibility with earlier versions of g++, the compiler will,
1049 by default, ignore @code{namespace-declarations},
1050 @code{using-declarations}, @code{using-directives}, and
1051 @code{namespace-names}, if they involve @code{std}.
1053 @item -fhuge-objects
1054 Support virtual function calls for objects that exceed the size
1055 representable by a @samp{short int}. Users should not use this flag by
1056 default; if you need to use it, the compiler will tell you so.
1058 This flag is not useful when compiling with -fvtable-thunks.
1060 Like all options that change the ABI, all C++ code, @emph{including
1061 libgcc} must be built with the same setting of this option.
1063 @item -fno-implicit-templates
1064 Never emit code for non-inline templates which are instantiated
1065 implicitly (i.e. by use); only emit code for explicit instantiations.
1066 @xref{Template Instantiation}, for more information.
1068 @item -fno-implicit-inline-templates
1069 Don't emit code for implicit instantiations of inline templates, either.
1070 The default is to handle inlines differently so that compiles with and
1071 without optimization will need the same set of explicit instantiations.
1073 @item -finit-priority
1074 Support @samp{__attribute__ ((init_priority (n)))} for controlling the
1075 order of initialization of file-scope objects. On ELF targets, this
1076 requires GNU ld 2.10 or later.
1078 @item -fno-implement-inlines
1079 To save space, do not emit out-of-line copies of inline functions
1080 controlled by @samp{#pragma implementation}. This will cause linker
1081 errors if these functions are not inlined everywhere they are called.
1083 @item -fname-mangling-version-@var{n}
1084 Control the way in which names are mangled. Version 0 is compatible
1085 with versions of g++ before 2.8. Version 1 is the default. Version 1
1086 will allow correct mangling of function templates. For example,
1087 version 0 mangling does not mangle foo<int, double> and foo<int, char>
1088 given this declaration:
1091 template <class T, class U> void foo(T t);
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 -foperator-names
1098 Recognize the operator name keywords @code{and}, @code{bitand},
1099 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1100 synonyms for the symbols they refer to. @samp{-ansi} implies
1101 @samp{-foperator-names}.
1103 @item -fno-optional-diags
1104 Disable diagnostics that the standard says a compiler does not need to
1105 issue. Currently, the only such diagnostic issued by g++ is the one for
1106 a name having multiple meanings within a class.
1109 Downgrade messages about nonconformant code from errors to warnings. By
1110 default, g++ effectively sets @samp{-pedantic-errors} without
1111 @samp{-pedantic}; this option reverses that. This behavior and this
1112 option are superceded by @samp{-pedantic}, which works as it does for GNU C.
1115 Enable automatic template instantiation. This option also implies
1116 @samp{-fno-implicit-templates}. @xref{Template Instantiation}, for more
1120 Disable generation of the information used by C++ runtime type
1121 identification features (@samp{dynamic_cast} and @samp{typeid}). If you
1122 don't use those parts of the language (or exception handling, which uses
1123 @samp{dynamic_cast} internally), you can save some space by using this
1126 @item -fstrict-prototype
1127 Within an @samp{extern "C"} linkage specification, treat a function
1128 declaration with no arguments, such as @samp{int foo ();}, as declaring
1129 the function to take no arguments. Normally, such a declaration means
1130 that the function @code{foo} can take any combination of arguments, as
1131 in C. @samp{-pedantic} implies @samp{-fstrict-prototype} unless
1132 overridden with @samp{-fno-strict-prototype}.
1134 Specifying this option will also suppress implicit declarations of
1137 This flag no longer affects declarations with C++ linkage.
1140 @itemx -fno-squangle
1141 @samp{-fsquangle} will enable a compressed form of name mangling for
1142 identifiers. In particular, it helps to shorten very long names by recognizing
1143 types and class names which occur more than once, replacing them with special
1144 short ID codes. This option also requires any C++ libraries being used to
1145 be compiled with this option as well. The compiler has this disabled (the
1146 equivalent of @samp{-fno-squangle}) by default.
1148 Like all options that change the ABI, all C++ code, @emph{including
1149 libgcc.a} must be built with the same setting of this option.
1151 @item -ftemplate-depth-@var{n}
1152 Set the maximum instantiation depth for template classes to @var{n}.
1153 A limit on the template instantiation depth is needed to detect
1154 endless recursions during template class instantiation. ANSI/ISO C++
1155 conforming programs must not rely on a maximum depth greater than 17.
1157 @item -fthis-is-variable
1158 Permit assignment to @code{this}. The incorporation of user-defined
1159 free store management into C++ has made assignment to @samp{this} an
1160 anachronism. Therefore, by default it is invalid to assign to
1161 @code{this} within a class member function; that is, GNU C++ treats
1162 @samp{this} in a member function of class @code{X} as a non-lvalue of
1163 type @samp{X *}. However, for backwards compatibility, you can make it
1164 valid with @samp{-fthis-is-variable}.
1166 @item -fvtable-thunks
1167 Use @samp{thunks} to implement the virtual function dispatch table
1168 (@samp{vtable}). The traditional (cfront-style) approach to
1169 implementing vtables was to store a pointer to the function and two
1170 offsets for adjusting the @samp{this} pointer at the call site. Newer
1171 implementations store a single pointer to a @samp{thunk} function which
1172 does any necessary adjustment and then calls the target function.
1174 This option also enables a heuristic for controlling emission of
1175 vtables; if a class has any non-inline virtual functions, the vtable
1176 will be emitted in the translation unit containing the first one of
1179 Like all options that change the ABI, all C++ code, @emph{including
1180 libgcc.a} must be built with the same setting of this option.
1183 Do not search for header files in the standard directories specific to
1184 C++, but do still search the other standard directories. (This option
1185 is used when building the C++ library.)
1188 In addition, these optimization, warning, and code generation options
1189 have meanings only for C++ programs:
1192 @item -fno-default-inline
1193 Do not assume @samp{inline} for functions defined inside a class scope.
1194 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1195 functions will have linkage like inline functions; they just won't be
1198 @item -Wctor-dtor-privacy (C++ only)
1199 Warn when a class seems unusable, because all the constructors or
1200 destructors in a class are private and the class has no friends or
1201 public static member functions.
1203 @item -Wnon-virtual-dtor (C++ only)
1204 Warn when a class declares a non-virtual destructor that should probably
1205 be virtual, because it looks like the class will be used polymorphically.
1207 @item -Wreorder (C++ only)
1208 @cindex reordering, warning
1209 @cindex warning for reordering of member initializers
1210 Warn when the order of member initializers given in the code does not
1211 match the order in which they must be executed. For instance:
1217 A(): j (0), i (1) @{ @}
1221 Here the compiler will warn that the member initializers for @samp{i}
1222 and @samp{j} will be rearranged to match the declaration order of the
1226 The following @samp{-W@dots{}} options are not affected by @samp{-Wall}.
1229 @item -Weffc++ (C++ only)
1230 Warn about violations of various style guidelines from Scott Meyers'
1231 @cite{Effective C++} books. If you use this option, you should be aware
1232 that the standard library headers do not obey all of these guidelines;
1233 you can use @samp{grep -v} to filter out those warnings.
1235 @item -Wno-deprecated (C++ only)
1236 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1238 @item -Wno-non-template-friend (C++ only)
1239 Disable warnings when non-templatized friend functions are declared
1240 within a template. With the advent of explicit template specification
1241 support in g++, if the name of the friend is an unqualified-id (ie,
1242 @samp{friend foo(int)}), the C++ language specification demands that the
1243 friend declare or define an ordinary, nontemplate function. (Section
1244 14.5.3). Before g++ implemented explicit specification, unqualified-ids
1245 could be interpreted as a particular specialization of a templatized
1246 function. Because this non-conforming behavior is no longer the default
1247 behavior for g++, @samp{-Wnon-template-friend} allows the compiler to
1248 check existing code for potential trouble spots, and is on by default.
1249 This new compiler behavior can also be turned off with the flag
1250 @samp{-fguiding-decls}, which activates the older, non-specification
1251 compiler code, or with @samp{-Wno-non-template-friend} which keeps the
1252 conformant compiler code but disables the helpful warning.
1254 @item -Wold-style-cast (C++ only)
1255 Warn if an old-style (C-style) cast is used within a C++ program. The
1256 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1257 @samp{const_cast}) are less vulnerable to unintended effects.
1259 @item -Woverloaded-virtual (C++ only)
1260 @cindex overloaded virtual fn, warning
1261 @cindex warning for overloaded virtual fn
1262 Warn when a derived class function declaration may be an error in
1263 defining a virtual function. In a derived class, the
1264 definitions of virtual functions must match the type signature of a
1265 virtual function declared in the base class. With this option, the
1266 compiler warns when you define a function with the same name as a
1267 virtual function, but with a type signature that does not match any
1268 declarations from the base class.
1270 @item -Wno-pmf-conversions (C++ only)
1271 Disable the diagnostic for converting a bound pointer to member function
1274 @item -Wsign-promo (C++ only)
1275 Warn when overload resolution chooses a promotion from unsigned or
1276 enumeral type to a signed type over a conversion to an unsigned type of
1277 the same size. Previous versions of g++ would try to preserve
1278 unsignedness, but the standard mandates the current behavior.
1280 @item -Wsynth (C++ only)
1281 @cindex warning for synthesized methods
1282 @cindex synthesized methods, warning
1283 Warn when g++'s synthesis behavior does not match that of cfront. For
1289 A& operator = (int);
1299 In this example, g++ will synthesize a default @samp{A& operator =
1300 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1303 @node Warning Options
1304 @section Options to Request or Suppress Warnings
1305 @cindex options to control warnings
1306 @cindex warning messages
1307 @cindex messages, warning
1308 @cindex suppressing warnings
1310 Warnings are diagnostic messages that report constructions which
1311 are not inherently erroneous but which are risky or suggest there
1312 may have been an error.
1314 You can request many specific warnings with options beginning @samp{-W},
1315 for example @samp{-Wimplicit} to request warnings on implicit
1316 declarations. Each of these specific warning options also has a
1317 negative form beginning @samp{-Wno-} to turn off warnings;
1318 for example, @samp{-Wno-implicit}. This manual lists only one of the
1319 two forms, whichever is not the default.
1321 These options control the amount and kinds of warnings produced by GNU
1325 @cindex syntax checking
1327 Check the code for syntax errors, but don't do anything beyond that.
1330 Issue all the warnings demanded by strict ANSI C and ISO C++;
1331 reject all programs that use forbidden extensions.
1333 Valid ANSI C and ISO C++ programs should compile properly with or without
1334 this option (though a rare few will require @samp{-ansi}). However,
1335 without this option, certain GNU extensions and traditional C and C++
1336 features are supported as well. With this option, they are rejected.
1338 @samp{-pedantic} does not cause warning messages for use of the
1339 alternate keywords whose names begin and end with @samp{__}. Pedantic
1340 warnings are also disabled in the expression that follows
1341 @code{__extension__}. However, only system header files should use
1342 these escape routes; application programs should avoid them.
1343 @xref{Alternate Keywords}.
1345 This option is not intended to be @i{useful}; it exists only to satisfy
1346 pedants who would otherwise claim that GNU CC fails to support the ANSI
1349 Some users try to use @samp{-pedantic} to check programs for strict ANSI
1350 C conformance. They soon find that it does not do quite what they want:
1351 it finds some non-ANSI practices, but not all---only those for which
1352 ANSI C @emph{requires} a diagnostic.
1354 A feature to report any failure to conform to ANSI C might be useful in
1355 some instances, but would require considerable additional work and would
1356 be quite different from @samp{-pedantic}. We don't have plans to
1357 support such a feature in the near future.
1359 @item -pedantic-errors
1360 Like @samp{-pedantic}, except that errors are produced rather than
1364 Inhibit all warning messages.
1367 Inhibit warning messages about the use of @samp{#import}.
1369 @item -Wchar-subscripts
1370 Warn if an array subscript has type @code{char}. This is a common cause
1371 of error, as programmers often forget that this type is signed on some
1375 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1376 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1379 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1380 the arguments supplied have types appropriate to the format string
1383 @item -Wimplicit-int
1384 Warn when a declaration does not specify a type.
1386 @item -Wimplicit-function-declaration
1387 @itemx -Werror-implicit-function-declaration
1388 Give a warning (or error) whenever a function is used before being
1392 Same as @samp{-Wimplicit-int} and @samp{-Wimplicit-function-}@*
1396 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1397 function with external linkage, returning int, taking either zero
1398 arguments, two, or three arguments of appropriate types.
1401 Warn if a multicharacter constant (@samp{'FOOF'}) is used. Usually they
1402 indicate a typo in the user's code, as they have implementation-defined
1403 values, and should not be used in portable code.
1406 Warn if parentheses are omitted in certain contexts, such
1407 as when there is an assignment in a context where a truth value
1408 is expected, or when operators are nested whose precedence people
1409 often get confused about.
1411 Also warn about constructions where there may be confusion to which
1412 @code{if} statement an @code{else} branch belongs. Here is an example of
1425 In C, every @code{else} branch belongs to the innermost possible @code{if}
1426 statement, which in this example is @code{if (b)}. This is often not
1427 what the programmer expected, as illustrated in the above example by
1428 indentation the programmer chose. When there is the potential for this
1429 confusion, GNU C will issue a warning when this flag is specified.
1430 To eliminate the warning, add explicit braces around the innermost
1431 @code{if} statement so there is no way the @code{else} could belong to
1432 the enclosing @code{if}. The resulting code would look like this:
1447 Warn whenever a function is defined with a return-type that defaults
1448 to @code{int}. Also warn about any @code{return} statement with no
1449 return-value in a function whose return-type is not @code{void}.
1452 Warn whenever a @code{switch} statement has an index of enumeral type
1453 and lacks a @code{case} for one or more of the named codes of that
1454 enumeration. (The presence of a @code{default} label prevents this
1455 warning.) @code{case} labels outside the enumeration range also
1456 provoke warnings when this option is used.
1459 Warn if any trigraphs are encountered (assuming they are enabled).
1462 Warn whenever a variable is unused aside from its declaration,
1463 whenever a function is declared static but never defined, whenever a
1464 label is declared but not used, and whenever a statement computes a
1465 result that is explicitly not used.
1467 In order to get a warning about an unused function parameter, you must
1468 specify both @samp{-W} and @samp{-Wunused}.
1470 To suppress this warning for an expression, simply cast it to void. For
1471 unused variables, parameters and labels, use the @samp{unused} attribute
1472 (@pxref{Variable Attributes}).
1474 @item -Wuninitialized
1475 An automatic variable is used without first being initialized.
1477 These warnings are possible only in optimizing compilation,
1478 because they require data flow information that is computed only
1479 when optimizing. If you don't specify @samp{-O}, you simply won't
1482 These warnings occur only for variables that are candidates for
1483 register allocation. Therefore, they do not occur for a variable that
1484 is declared @code{volatile}, or whose address is taken, or whose size
1485 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
1486 structures, unions or arrays, even when they are in registers.
1488 Note that there may be no warning about a variable that is used only
1489 to compute a value that itself is never used, because such
1490 computations may be deleted by data flow analysis before the warnings
1493 These warnings are made optional because GNU CC is not smart
1494 enough to see all the reasons why the code might be correct
1495 despite appearing to have an error. Here is one example of how
1514 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
1515 always initialized, but GNU CC doesn't know this. Here is
1516 another common case:
1521 if (change_y) save_y = y, y = new_y;
1523 if (change_y) y = save_y;
1528 This has no bug because @code{save_y} is used only if it is set.
1530 Some spurious warnings can be avoided if you declare all the functions
1531 you use that never return as @code{noreturn}. @xref{Function
1534 @item -Wunknown-pragmas
1535 @cindex warning for unknown pragmas
1536 @cindex unknown pragmas, warning
1537 @cindex pragmas, warning of unknown
1538 Warn when a #pragma directive is encountered which is not understood by
1539 GCC. If this command line option is used, warnings will even be issued
1540 for unknown pragmas in system header files. This is not the case if
1541 the warnings were only enabled by the @samp{-Wall} command line option.
1544 All of the above @samp{-W} options combined. This enables all the
1545 warnings about constructions that some users consider questionable, and
1546 that are easy to avoid (or modify to prevent the warning), even in
1547 conjunction with macros.
1550 The following @samp{-W@dots{}} options are not implied by @samp{-Wall}.
1551 Some of them warn about constructions that users generally do not
1552 consider questionable, but which occasionally you might wish to check
1553 for; others warn about constructions that are necessary or hard to avoid
1554 in some cases, and there is no simple way to modify the code to suppress
1559 Print extra warning messages for these events:
1562 @cindex @code{longjmp} warnings
1564 A nonvolatile automatic variable might be changed by a call to
1565 @code{longjmp}. These warnings as well are possible only in
1566 optimizing compilation.
1568 The compiler sees only the calls to @code{setjmp}. It cannot know
1569 where @code{longjmp} will be called; in fact, a signal handler could
1570 call it at any point in the code. As a result, you may get a warning
1571 even when there is in fact no problem because @code{longjmp} cannot
1572 in fact be called at the place which would cause a problem.
1575 A function can return either with or without a value. (Falling
1576 off the end of the function body is considered returning without
1577 a value.) For example, this function would evoke such a
1591 An expression-statement or the left-hand side of a comma expression
1592 contains no side effects.
1593 To suppress the warning, cast the unused expression to void.
1594 For example, an expression such as @samp{x[i,j]} will cause a warning,
1595 but @samp{x[(void)i,j]} will not.
1598 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
1601 A comparison like @samp{x<=y<=z} appears; this is equivalent to
1602 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
1603 that of ordinary mathematical notation.
1606 Storage-class specifiers like @code{static} are not the first things in
1607 a declaration. According to the C Standard, this usage is obsolescent.
1610 If @samp{-Wall} or @samp{-Wunused} is also specified, warn about unused
1614 A comparison between signed and unsigned values could produce an
1615 incorrect result when the signed value is converted to unsigned.
1616 (But don't warn if @samp{-Wno-sign-compare} is also specified.)
1619 An aggregate has a partly bracketed initializer.
1620 For example, the following code would evoke such a warning,
1621 because braces are missing around the initializer for @code{x.h}:
1624 struct s @{ int f, g; @};
1625 struct t @{ struct s h; int i; @};
1626 struct t x = @{ 1, 2, 3 @};
1630 An aggregate has an initializer which does not initialize all members.
1631 For example, the following code would cause such a warning, because
1632 @code{x.h} would be implicitly initialized to zero:
1635 struct s @{ int f, g, h; @};
1636 struct s x = @{ 3, 4 @};
1641 Warn about certain constructs that behave differently in traditional and
1646 Macro arguments occurring within string constants in the macro body.
1647 These would substitute the argument in traditional C, but are part of
1648 the constant in ANSI C.
1651 A function declared external in one block and then used after the end of
1655 A @code{switch} statement has an operand of type @code{long}.
1658 A non-@code{static} function declaration follows a @code{static} one.
1659 This construct is not accepted by some traditional C compilers.
1663 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
1666 Warn whenever a local variable shadows another local variable.
1668 @item -Wid-clash-@var{len}
1669 Warn whenever two distinct identifiers match in the first @var{len}
1670 characters. This may help you prepare a program that will compile
1671 with certain obsolete, brain-damaged compilers.
1673 @item -Wlarger-than-@var{len}
1674 Warn whenever an object of larger than @var{len} bytes is defined.
1676 @item -Wpointer-arith
1677 Warn about anything that depends on the ``size of'' a function type or
1678 of @code{void}. GNU C assigns these types a size of 1, for
1679 convenience in calculations with @code{void *} pointers and pointers
1682 @item -Wbad-function-cast
1683 Warn whenever a function call is cast to a non-matching type.
1684 For example, warn if @code{int malloc()} is cast to @code{anything *}.
1687 Warn whenever a pointer is cast so as to remove a type qualifier from
1688 the target type. For example, warn if a @code{const char *} is cast
1689 to an ordinary @code{char *}.
1692 Warn whenever a pointer is cast such that the required alignment of the
1693 target is increased. For example, warn if a @code{char *} is cast to
1694 an @code{int *} on machines where integers can only be accessed at
1695 two- or four-byte boundaries.
1697 @item -Wwrite-strings
1698 Give string constants the type @code{const char[@var{length}]} so that
1699 copying the address of one into a non-@code{const} @code{char *}
1700 pointer will get a warning. These warnings will help you find at
1701 compile time code that can try to write into a string constant, but
1702 only if you have been very careful about using @code{const} in
1703 declarations and prototypes. Otherwise, it will just be a nuisance;
1704 this is why we did not make @samp{-Wall} request these warnings.
1707 Warn if a prototype causes a type conversion that is different from what
1708 would happen to the same argument in the absence of a prototype. This
1709 includes conversions of fixed point to floating and vice versa, and
1710 conversions changing the width or signedness of a fixed point argument
1711 except when the same as the default promotion.
1713 Also, warn if a negative integer constant expression is implicitly
1714 converted to an unsigned type. For example, warn about the assignment
1715 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
1716 casts like @code{(unsigned) -1}.
1718 @item -Wsign-compare
1719 @cindex warning for comparison of signed and unsigned values
1720 @cindex comparison of signed and unsigned values, warning
1721 @cindex signed and unsigned values, comparison warning
1722 Warn when a comparison between signed and unsigned values could produce
1723 an incorrect result when the signed value is converted to unsigned.
1724 This warning is also enabled by @samp{-W}; to get the other warnings
1725 of @samp{-W} without this warning, use @samp{-W -Wno-sign-compare}.
1727 @item -Waggregate-return
1728 Warn if any functions that return structures or unions are defined or
1729 called. (In languages where you can return an array, this also elicits
1732 @item -Wstrict-prototypes
1733 Warn if a function is declared or defined without specifying the
1734 argument types. (An old-style function definition is permitted without
1735 a warning if preceded by a declaration which specifies the argument
1738 @item -Wmissing-prototypes
1739 Warn if a global function is defined without a previous prototype
1740 declaration. This warning is issued even if the definition itself
1741 provides a prototype. The aim is to detect global functions that fail
1742 to be declared in header files.
1744 @item -Wmissing-declarations
1745 Warn if a global function is defined without a previous declaration.
1746 Do so even if the definition itself provides a prototype.
1747 Use this option to detect global functions that are not declared in
1750 @item -Wmissing-noreturn
1751 Warn about functions which might be candidates for attribute @code{noreturn}.
1752 Note these are only possible candidates, not absolute ones. Care should
1753 be taken to manually verify functions actually do not ever return before
1754 adding the @code{noreturn} attribute, otherwise subtle code generation
1755 bugs could be introduced.
1757 @item -Wredundant-decls
1758 Warn if anything is declared more than once in the same scope, even in
1759 cases where multiple declaration is valid and changes nothing.
1761 @item -Wnested-externs
1762 Warn if an @code{extern} declaration is encountered within an function.
1765 Warn if a function can not be inlined, and either it was declared as inline,
1766 or else the @samp{-finline-functions} option was given.
1769 Warn if @samp{long long} type is used. This is default. To inhibit
1770 the warning messages, use @samp{-Wno-long-long}. Flags
1771 @samp{-Wlong-long} and @samp{-Wno-long-long} are taken into account
1772 only when @samp{-pedantic} flag is used.
1775 Make all warnings into errors.
1778 @node Debugging Options
1779 @section Options for Debugging Your Program or GNU CC
1780 @cindex options, debugging
1781 @cindex debugging information options
1783 GNU CC has various special options that are used for debugging
1784 either your program or GCC:
1788 Produce debugging information in the operating system's native format
1789 (stabs, COFF, XCOFF, or DWARF). GDB can work with this debugging
1792 On most systems that use stabs format, @samp{-g} enables use of extra
1793 debugging information that only GDB can use; this extra information
1794 makes debugging work better in GDB but will probably make other debuggers
1796 refuse to read the program. If you want to control for certain whether
1797 to generate the extra information, use @samp{-gstabs+}, @samp{-gstabs},
1798 @samp{-gxcoff+}, @samp{-gxcoff}, @samp{-gdwarf-1+}, or @samp{-gdwarf-1}
1801 Unlike most other C compilers, GNU CC allows you to use @samp{-g} with
1802 @samp{-O}. The shortcuts taken by optimized code may occasionally
1803 produce surprising results: some variables you declared may not exist
1804 at all; flow of control may briefly move where you did not expect it;
1805 some statements may not be executed because they compute constant
1806 results or their values were already at hand; some statements may
1807 execute in different places because they were moved out of loops.
1809 Nevertheless it proves possible to debug optimized output. This makes
1810 it reasonable to use the optimizer for programs that might have bugs.
1812 The following options are useful when GNU CC is generated with the
1813 capability for more than one debugging format.
1816 Produce debugging information for use by GDB. This means to use the
1817 most expressive format available (DWARF 2, stabs, or the native format
1818 if neither of those are supported), including GDB extensions if at all
1822 Produce debugging information in stabs format (if that is supported),
1823 without GDB extensions. This is the format used by DBX on most BSD
1824 systems. On MIPS, Alpha and System V Release 4 systems this option
1825 produces stabs debugging output which is not understood by DBX or SDB.
1826 On System V Release 4 systems this option requires the GNU assembler.
1829 Produce debugging information in stabs format (if that is supported),
1830 using GNU extensions understood only by the GNU debugger (GDB). The
1831 use of these extensions is likely to make other debuggers crash or
1832 refuse to read the program.
1835 Produce debugging information in COFF format (if that is supported).
1836 This is the format used by SDB on most System V systems prior to
1840 Produce debugging information in XCOFF format (if that is supported).
1841 This is the format used by the DBX debugger on IBM RS/6000 systems.
1844 Produce debugging information in XCOFF format (if that is supported),
1845 using GNU extensions understood only by the GNU debugger (GDB). The
1846 use of these extensions is likely to make other debuggers crash or
1847 refuse to read the program, and may cause assemblers other than the GNU
1848 assembler (GAS) to fail with an error.
1851 Produce debugging information in DWARF version 1 format (if that is
1852 supported). This is the format used by SDB on most System V Release 4
1856 Produce debugging information in DWARF version 1 format (if that is
1857 supported), using GNU extensions understood only by the GNU debugger
1858 (GDB). The use of these extensions is likely to make other debuggers
1859 crash or refuse to read the program.
1862 Produce debugging information in DWARF version 2 format (if that is
1863 supported). This is the format used by DBX on IRIX 6.
1866 @itemx -ggdb@var{level}
1867 @itemx -gstabs@var{level}
1868 @itemx -gcoff@var{level}
1869 @itemx -gxcoff@var{level}
1870 @itemx -gdwarf@var{level}
1871 @itemx -gdwarf-2@var{level}
1872 Request debugging information and also use @var{level} to specify how
1873 much information. The default level is 2.
1875 Level 1 produces minimal information, enough for making backtraces in
1876 parts of the program that you don't plan to debug. This includes
1877 descriptions of functions and external variables, but no information
1878 about local variables and no line numbers.
1880 Level 3 includes extra information, such as all the macro definitions
1881 present in the program. Some debuggers support macro expansion when
1886 Generate extra code to write profile information suitable for the
1887 analysis program @code{prof}. You must use this option when compiling
1888 the source files you want data about, and you must also use it when
1891 @cindex @code{gprof}
1893 Generate extra code to write profile information suitable for the
1894 analysis program @code{gprof}. You must use this option when compiling
1895 the source files you want data about, and you must also use it when
1900 Generate extra code to write profile information for basic blocks, which will
1901 record the number of times each basic block is executed, the basic block start
1902 address, and the function name containing the basic block. If @samp{-g} is
1903 used, the line number and filename of the start of the basic block will also be
1904 recorded. If not overridden by the machine description, the default action is
1905 to append to the text file @file{bb.out}.
1907 This data could be analyzed by a program like @code{tcov}. Note,
1908 however, that the format of the data is not what @code{tcov} expects.
1909 Eventually GNU @code{gprof} should be extended to process this data.
1912 Makes the compiler print out each function name as it is compiled, and
1913 print some statistics about each pass when it finishes.
1916 Generate extra code to profile basic blocks. Your executable will
1917 produce output that is a superset of that produced when @samp{-a} is
1918 used. Additional output is the source and target address of the basic
1919 blocks where a jump takes place, the number of times a jump is executed,
1920 and (optionally) the complete sequence of basic blocks being executed.
1921 The output is appended to file @file{bb.out}.
1923 You can examine different profiling aspects without recompilation. Your
1924 executable will read a list of function names from file @file{bb.in}.
1925 Profiling starts when a function on the list is entered and stops when
1926 that invocation is exited. To exclude a function from profiling, prefix
1927 its name with `-'. If a function name is not unique, you can
1928 disambiguate it by writing it in the form
1929 @samp{/path/filename.d:functionname}. Your executable will write the
1930 available paths and filenames in file @file{bb.out}.
1932 Several function names have a special meaning:
1935 Write source, target and frequency of jumps to file @file{bb.out}.
1936 @item __bb_hidecall__
1937 Exclude function calls from frequency count.
1938 @item __bb_showret__
1939 Include function returns in frequency count.
1941 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
1942 The file will be compressed using the program @samp{gzip}, which must
1943 exist in your @code{PATH}. On systems without the @samp{popen}
1944 function, the file will be named @file{bbtrace} and will not be
1945 compressed. @strong{Profiling for even a few seconds on these systems
1946 will produce a very large file.} Note: @code{__bb_hidecall__} and
1947 @code{__bb_showret__} will not affect the sequence written to
1951 Here's a short example using different profiling parameters
1952 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
1953 1 and 2 and is called twice from block 3 of function @code{main}. After
1954 the calls, block 3 transfers control to block 4 of @code{main}.
1956 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
1957 the following sequence of blocks is written to file @file{bbtrace.gz}:
1958 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
1959 the return is to a point inside the block and not to the top. The
1960 block address 0 always indicates, that control is transferred
1961 to the trace from somewhere outside the observed functions. With
1962 @samp{-foo} added to @file{bb.in}, the blocks of function
1963 @code{foo} are removed from the trace, so only 0 3 4 remains.
1965 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
1966 jump frequencies will be written to file @file{bb.out}. The
1967 frequencies are obtained by constructing a trace of blocks
1968 and incrementing a counter for every neighbouring pair of blocks
1969 in the trace. The trace 0 3 1 2 1 2 4 displays the following
1973 Jump from block 0x0 to block 0x3 executed 1 time(s)
1974 Jump from block 0x3 to block 0x1 executed 1 time(s)
1975 Jump from block 0x1 to block 0x2 executed 2 time(s)
1976 Jump from block 0x2 to block 0x1 executed 1 time(s)
1977 Jump from block 0x2 to block 0x4 executed 1 time(s)
1980 With @code{__bb_hidecall__}, control transfer due to call instructions
1981 is removed from the trace, that is the trace is cut into three parts: 0
1982 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
1983 to return instructions is added to the trace. The trace becomes: 0 3 1
1984 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
1985 written to @file{bbtrace.gz}. It is solely used for counting jump
1988 @item -fprofile-arcs
1989 Instrument @dfn{arcs} during compilation. For each function of your
1990 program, GNU CC creates a program flow graph, then finds a spanning tree
1991 for the graph. Only arcs that are not on the spanning tree have to be
1992 instrumented: the compiler adds code to count the number of times that these
1993 arcs are executed. When an arc is the only exit or only entrance to a
1994 block, the instrumentation code can be added to the block; otherwise, a
1995 new basic block must be created to hold the instrumentation code.
1997 Since not every arc in the program must be instrumented, programs
1998 compiled with this option run faster than programs compiled with
1999 @samp{-a}, which adds instrumentation code to every basic block in the
2000 program. The tradeoff: since @code{gcov} does not have
2001 execution counts for all branches, it must start with the execution
2002 counts for the instrumented branches, and then iterate over the program
2003 flow graph until the entire graph has been solved. Hence, @code{gcov}
2004 runs a little more slowly than a program which uses information from
2007 @samp{-fprofile-arcs} also makes it possible to estimate branch
2008 probabilities, and to calculate basic block execution counts. In
2009 general, basic block execution counts do not give enough information to
2010 estimate all branch probabilities. When the compiled program exits, it
2011 saves the arc execution counts to a file called
2012 @file{@var{sourcename}.da}. Use the compiler option
2013 @samp{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2014 Control Optimization}) when recompiling, to optimize using estimated
2015 branch probabilities.
2018 @item -ftest-coverage
2019 Create data files for the @code{gcov} code-coverage utility
2020 (@pxref{Gcov,, @code{gcov}: a GNU CC Test Coverage Program}).
2021 The data file names begin with the name of your source file:
2024 @item @var{sourcename}.bb
2025 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2026 associate basic block execution counts with line numbers.
2028 @item @var{sourcename}.bbg
2029 A list of all arcs in the program flow graph. This allows @code{gcov}
2030 to reconstruct the program flow graph, so that it can compute all basic
2031 block and arc execution counts from the information in the
2032 @code{@var{sourcename}.da} file (this last file is the output from
2033 @samp{-fprofile-arcs}).
2037 Makes the compiler print out each function name as it is compiled, and
2038 print some statistics about each pass when it finishes.
2040 @item -d@var{letters}
2041 Says to make debugging dumps during compilation at times specified by
2042 @var{letters}. This is used for debugging the compiler. The file names
2043 for most of the dumps are made by appending a word to the source file
2044 name (e.g. @file{foo.c.rtl} or @file{foo.c.jump}). Here are the
2045 possible letters for use in @var{letters}, and their meanings:
2049 Dump after computing branch probabilities, to @file{@var{file}.bp}.
2051 Dump after instruction combination, to the file @file{@var{file}.combine}.
2053 Dump after delayed branch scheduling, to @file{@var{file}.dbr}.
2055 Dump all macro definitions, at the end of preprocessing, in addition to
2058 Dump after RTL generation, to @file{@var{file}.rtl}.
2060 Dump after first jump optimization, to @file{@var{file}.jump}.
2062 Dump after purging ADDRESSOF, to @file{@var{file}.addressof}.
2064 Dump after flow analysis, to @file{@var{file}.flow}.
2066 Dump after global register allocation, to @file{@var{file}.greg}.
2068 Dump after GCSE, to @file{@var{file}.gcse}.
2070 Dump after first jump optimization, to @file{@var{file}.jump}.
2072 Dump after last jump optimization, to @file{@var{file}.jump2}.
2074 Dump after conversion from registers to stack, to @file{@var{file}.stack}.
2076 Dump after local register allocation, to @file{@var{file}.lreg}.
2078 Dump after loop optimization, to @file{@var{file}.loop}.
2080 Dump after performing the machine dependent reorganisation pass, to
2081 @file{@var{file}.mach}.
2083 Dump after the register move pass, to @file{@var{file}.regmove}.
2085 Dump after RTL generation, to @file{@var{file}.rtl}.
2087 Dump after the second instruction scheduling pass, to @file{@var{file}.sched2}.
2089 Dump after CSE (including the jump optimization that sometimes follows
2090 CSE), to @file{@var{file}.cse}.
2092 Dump after the first instruction scheduling pass, to @file{@var{file}.sched}.
2094 Dump after the second CSE pass (including the jump optimization that
2095 sometimes follows CSE), to @file{@var{file}.cse2}.
2097 Produce all the dumps listed above.
2099 Print statistics on memory usage, at the end of the run, to
2102 Annotate the assembler output with a comment indicating which
2103 pattern and alternative was used. The length of each instruction is
2106 Just generate RTL for a function instead of compiling it. Usually used
2109 Dump debugging information during parsing, to standard error.
2111 Annotate the assembler output with miscellaneous debugging information.
2114 @item -fdump-unnumbered
2115 When doing debugging dumps (see -d option above), suppress instruction
2116 numbers and line number note output. This makes it more feasible to
2117 use diff on debugging dumps for compiler invokations with different
2118 options, in particular with and without -g.
2120 @item -fpretend-float
2121 When running a cross-compiler, pretend that the target machine uses the
2122 same floating point format as the host machine. This causes incorrect
2123 output of the actual floating constants, but the actual instruction
2124 sequence will probably be the same as GNU CC would make when running on
2128 Store the usual ``temporary'' intermediate files permanently; place them
2129 in the current directory and name them based on the source file. Thus,
2130 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
2131 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}.
2133 @item -print-file-name=@var{library}
2134 Print the full absolute name of the library file @var{library} that
2135 would be used when linking---and don't do anything else. With this
2136 option, GNU CC does not compile or link anything; it just prints the
2139 @item -print-prog-name=@var{program}
2140 Like @samp{-print-file-name}, but searches for a program such as @samp{cpp}.
2142 @item -print-libgcc-file-name
2143 Same as @samp{-print-file-name=libgcc.a}.
2145 This is useful when you use @samp{-nostdlib} or @samp{-nodefaultlibs}
2146 but you do want to link with @file{libgcc.a}. You can do
2149 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
2152 @item -print-search-dirs
2153 Print the name of the configured installation directory and a list of
2154 program and library directories gcc will search---and don't do anything else.
2156 This is useful when gcc prints the error message
2157 @samp{installation problem, cannot exec cpp: No such file or directory}.
2158 To resolve this you either need to put @file{cpp} and the other compiler
2159 components where gcc expects to find them, or you can set the environment
2160 variable @code{GCC_EXEC_PREFIX} to the directory where you installed them.
2161 Don't forget the trailing '/'.
2162 @xref{Environment Variables}.
2165 @node Optimize Options
2166 @section Options That Control Optimization
2167 @cindex optimize options
2168 @cindex options, optimization
2170 These options control various sorts of optimizations:
2175 Optimize. Optimizing compilation takes somewhat more time, and a lot
2176 more memory for a large function.
2178 Without @samp{-O}, the compiler's goal is to reduce the cost of
2179 compilation and to make debugging produce the expected results.
2180 Statements are independent: if you stop the program with a breakpoint
2181 between statements, you can then assign a new value to any variable or
2182 change the program counter to any other statement in the function and
2183 get exactly the results you would expect from the source code.
2185 Without @samp{-O}, the compiler only allocates variables declared
2186 @code{register} in registers. The resulting compiled code is a little
2187 worse than produced by PCC without @samp{-O}.
2189 With @samp{-O}, the compiler tries to reduce code size and execution
2192 When you specify @samp{-O}, the compiler turns on @samp{-fthread-jumps}
2193 and @samp{-fdefer-pop} on all machines. The compiler turns on
2194 @samp{-fdelayed-branch} on machines that have delay slots, and
2195 @samp{-fomit-frame-pointer} on machines that can support debugging even
2196 without a frame pointer. On some machines the compiler also turns
2197 on other flags.@refill
2200 Optimize even more. GNU CC performs nearly all supported optimizations
2201 that do not involve a space-speed tradeoff. The compiler does not
2202 perform loop unrolling or function inlining when you specify @samp{-O2}.
2203 As compared to @samp{-O}, this option increases both compilation time
2204 and the performance of the generated code.
2206 @samp{-O2} turns on all optional optimizations except for loop unrolling
2207 and function inlining. It also turns on the @samp{-fforce-mem} option
2208 on all machines and frame pointer elimination on machines where doing so
2209 does not interfere with debugging.
2212 Optimize yet more. @samp{-O3} turns on all optimizations specified by
2213 @samp{-O2} and also turns on the @samp{inline-functions} option.
2219 Optimize for size. @samp{-Os} enables all @samp{-O2} optimizations that
2220 do not typically increase code size. It also performs further
2221 optimizations designed to reduce code size.
2223 If you use multiple @samp{-O} options, with or without level numbers,
2224 the last such option is the one that is effective.
2227 Options of the form @samp{-f@var{flag}} specify machine-independent
2228 flags. Most flags have both positive and negative forms; the negative
2229 form of @samp{-ffoo} would be @samp{-fno-foo}. In the table below,
2230 only one of the forms is listed---the one which is not the default.
2231 You can figure out the other form by either removing @samp{no-} or
2236 Do not store floating point variables in registers, and inhibit other
2237 options that might change whether a floating point value is taken from a
2240 @cindex floating point precision
2241 This option prevents undesirable excess precision on machines such as
2242 the 68000 where the floating registers (of the 68881) keep more
2243 precision than a @code{double} is supposed to have. Similarly for the
2244 x86 architecture. For most programs, the excess precision does only
2245 good, but a few programs rely on the precise definition of IEEE floating
2246 point. Use @samp{-ffloat-store} for such programs, after modifying
2247 them to store all pertinent intermediate computations into variables.
2249 @item -fno-default-inline
2250 Do not make member functions inline by default merely because they are
2251 defined inside the class scope (C++ only). Otherwise, when you specify
2252 @w{@samp{-O}}, member functions defined inside class scope are compiled
2253 inline by default; i.e., you don't need to add @samp{inline} in front of
2254 the member function name.
2256 @item -fno-defer-pop
2257 Always pop the arguments to each function call as soon as that function
2258 returns. For machines which must pop arguments after a function call,
2259 the compiler normally lets arguments accumulate on the stack for several
2260 function calls and pops them all at once.
2263 Force memory operands to be copied into registers before doing
2264 arithmetic on them. This produces better code by making all memory
2265 references potential common subexpressions. When they are not common
2266 subexpressions, instruction combination should eliminate the separate
2267 register-load. The @samp{-O2} option turns on this option.
2270 Force memory address constants to be copied into registers before
2271 doing arithmetic on them. This may produce better code just as
2272 @samp{-fforce-mem} may.
2274 @item -fomit-frame-pointer
2275 Don't keep the frame pointer in a register for functions that
2276 don't need one. This avoids the instructions to save, set up and
2277 restore frame pointers; it also makes an extra register available
2278 in many functions. @strong{It also makes debugging impossible on
2282 On some machines, such as the Vax, this flag has no effect, because
2283 the standard calling sequence automatically handles the frame pointer
2284 and nothing is saved by pretending it doesn't exist. The
2285 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2286 whether a target machine supports this flag. @xref{Registers}.@refill
2289 On some machines, such as the Vax, this flag has no effect, because
2290 the standard calling sequence automatically handles the frame pointer
2291 and nothing is saved by pretending it doesn't exist. The
2292 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2293 whether a target machine supports this flag. @xref{Registers,,Register
2294 Usage, gcc.info, Using and Porting GCC}.@refill
2298 Don't pay attention to the @code{inline} keyword. Normally this option
2299 is used to keep the compiler from expanding any functions inline.
2300 Note that if you are not optimizing, no functions can be expanded inline.
2302 @item -finline-functions
2303 Integrate all simple functions into their callers. The compiler
2304 heuristically decides which functions are simple enough to be worth
2305 integrating in this way.
2307 If all calls to a given function are integrated, and the function is
2308 declared @code{static}, then the function is normally not output as
2309 assembler code in its own right.
2311 @item -finline-limit-@var{n}
2312 By default, gcc limits the size of functions that can be inlined. This flag
2313 allows the control of this limit for functions that are explicitly marked as
2314 inline (ie marked with the inline keyword or defined within the class
2315 definition in c++). @var{n} is the size of functions that can be inlined in
2316 number of pseudo instructions (not counting parameter handling). The default
2317 value of n is 10000. Increasing this value can result in more inlined code at
2318 the cost of compilation time and memory consumption. Decreasing usually makes
2319 the compilation faster and less code will be inlined (which presumably
2320 means slower programs). This option is particularly useful for programs that
2321 use inlining heavily such as those based on recursive templates with c++.
2323 @emph{Note:} pseudo instruction represents, in this particular context, an
2324 abstract measurement of function's size. In no way, it represents a count
2325 of assembly instructions and as such its exact meaning might change from one
2326 release to an another.
2328 @item -fkeep-inline-functions
2329 Even if all calls to a given function are integrated, and the function
2330 is declared @code{static}, nevertheless output a separate run-time
2331 callable version of the function. This switch does not affect
2332 @code{extern inline} functions.
2334 @item -fkeep-static-consts
2335 Emit variables declared @code{static const} when optimization isn't turned
2336 on, even if the variables aren't referenced.
2338 GNU CC enables this option by default. If you want to force the compiler to
2339 check if the variable was referenced, regardless of whether or not
2340 optimization is turned on, use the @samp{-fno-keep-static-consts} option.
2342 @item -fno-function-cse
2343 Do not put function addresses in registers; make each instruction that
2344 calls a constant function contain the function's address explicitly.
2346 This option results in less efficient code, but some strange hacks
2347 that alter the assembler output may be confused by the optimizations
2348 performed when this option is not used.
2351 This option allows GCC to violate some ANSI or IEEE rules and/or
2352 specifications in the interest of optimizing code for speed. For
2353 example, it allows the compiler to assume arguments to the @code{sqrt}
2354 function are non-negative numbers and that no floating-point values
2357 This option should never be turned on by any @samp{-O} option since
2358 it can result in incorrect output for programs which depend on
2359 an exact implementation of IEEE or ANSI rules/specifications for
2363 @c following causes underfulls.. they don't look great, but we deal.
2365 The following options control specific optimizations. The @samp{-O2}
2366 option turns on all of these optimizations except @samp{-funroll-loops}
2367 and @samp{-funroll-all-loops}. On most machines, the @samp{-O} option
2368 turns on the @samp{-fthread-jumps} and @samp{-fdelayed-branch} options,
2369 but specific machines may handle it differently.
2371 You can use the following flags in the rare cases when ``fine-tuning''
2372 of optimizations to be performed is desired.
2375 @item -fstrength-reduce
2376 Perform the optimizations of loop strength reduction and
2377 elimination of iteration variables.
2379 @item -fthread-jumps
2380 Perform optimizations where we check to see if a jump branches to a
2381 location where another comparison subsumed by the first is found. If
2382 so, the first branch is redirected to either the destination of the
2383 second branch or a point immediately following it, depending on whether
2384 the condition is known to be true or false.
2386 @item -fcse-follow-jumps
2387 In common subexpression elimination, scan through jump instructions
2388 when the target of the jump is not reached by any other path. For
2389 example, when CSE encounters an @code{if} statement with an
2390 @code{else} clause, CSE will follow the jump when the condition
2393 @item -fcse-skip-blocks
2394 This is similar to @samp{-fcse-follow-jumps}, but causes CSE to
2395 follow jumps which conditionally skip over blocks. When CSE
2396 encounters a simple @code{if} statement with no else clause,
2397 @samp{-fcse-skip-blocks} causes CSE to follow the jump around the
2398 body of the @code{if}.
2400 @item -frerun-cse-after-loop
2401 Re-run common subexpression elimination after loop optimizations has been
2404 @item -frerun-loop-opt
2405 Run the loop optimizer twice.
2408 Perform a global common subexpression elimination pass.
2409 This pass also performs global constant and copy propagation.
2411 @item -fexpensive-optimizations
2412 Perform a number of minor optimizations that are relatively expensive.
2414 @item -foptimize-register-moves
2416 Attempt to reassign register numbers in move instructions and as
2417 operands of other simple instructions in order to maximize the amount of
2418 register tying. This is especially helpful on machines with two-operand
2419 instructions. GNU CC enables this optimization by default with @samp{-O2}
2422 Note @code{-fregmove} and @code{-foptimize-register-moves} are the same
2425 @item -fdelayed-branch
2426 If supported for the target machine, attempt to reorder instructions
2427 to exploit instruction slots available after delayed branch
2430 @item -fschedule-insns
2431 If supported for the target machine, attempt to reorder instructions to
2432 eliminate execution stalls due to required data being unavailable. This
2433 helps machines that have slow floating point or memory load instructions
2434 by allowing other instructions to be issued until the result of the load
2435 or floating point instruction is required.
2437 @item -fschedule-insns2
2438 Similar to @samp{-fschedule-insns}, but requests an additional pass of
2439 instruction scheduling after register allocation has been done. This is
2440 especially useful on machines with a relatively small number of
2441 registers and where memory load instructions take more than one cycle.
2443 @item -ffunction-sections
2444 @itemx -fdata-sections
2445 Place each function or data item into its own section in the output
2446 file if the target supports arbitrary sections. The name of the
2447 function or the name of the data item determines the section's name
2450 Use these options on systems where the linker can perform optimizations
2451 to improve locality of reference in the instruction space. HPPA
2452 processors running HP-UX and Sparc processors running Solaris 2 have
2453 linkers with such optimizations. Other systems using the ELF object format
2454 as well as AIX may have these optimizations in the future.
2456 Only use these options when there are significant benefits from doing
2457 so. When you specify these options, the assembler and linker will
2458 create larger object and executable files and will also be slower.
2459 You will not be able to use @code{gprof} on all systems if you
2460 specify this option and you may have problems with debugging if
2461 you specify both this option and @samp{-g}.
2463 @item -fcaller-saves
2464 Enable values to be allocated in registers that will be clobbered by
2465 function calls, by emitting extra instructions to save and restore the
2466 registers around such calls. Such allocation is done only when it
2467 seems to result in better code than would otherwise be produced.
2469 This option is always enabled by default on certain machines, usually
2470 those which have no call-preserved registers to use instead.
2472 For all machines, optimization level 2 and higher enables this flag by
2475 @item -funroll-loops
2476 Perform the optimization of loop unrolling. This is only done for loops
2477 whose number of iterations can be determined at compile time or run time.
2478 @samp{-funroll-loops} implies both @samp{-fstrength-reduce} and
2479 @samp{-frerun-cse-after-loop}.
2481 @item -funroll-all-loops
2482 Perform the optimization of loop unrolling. This is done for all loops
2483 and usually makes programs run more slowly. @samp{-funroll-all-loops}
2484 implies @samp{-fstrength-reduce} as well as @samp{-frerun-cse-after-loop}.
2486 @item -fmove-all-movables
2487 Forces all invariant computations in loops to be moved
2490 @item -freduce-all-givs
2491 Forces all general-induction variables in loops to be
2494 @emph{Note:} When compiling programs written in Fortran,
2495 @samp{-fmove-all-movables} and @samp{-freduce-all-givs} are enabled
2496 by default when you use the optimizer.
2498 These options may generate better or worse code; results are highly
2499 dependent on the structure of loops within the source code.
2501 These two options are intended to be removed someday, once
2502 they have helped determine the efficacy of various
2503 approaches to improving loop optimizations.
2505 Please let us (@code{egcs@@egcs.cygnus.com} and @code{fortran@@gnu.org})
2506 know how use of these options affects
2507 the performance of your production code.
2508 We're very interested in code that runs @emph{slower}
2509 when these options are @emph{enabled}.
2512 Disable any machine-specific peephole optimizations.
2514 @item -fbranch-probabilities
2515 After running a program compiled with @samp{-fprofile-arcs}
2516 (@pxref{Debugging Options,, Options for Debugging Your Program or
2517 @code{gcc}}), you can compile it a second time using
2518 @samp{-fbranch-probabilities}, to improve optimizations based on
2519 guessing the path a branch might take.
2522 With @samp{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
2523 note on the first instruction of each basic block, and a
2524 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
2525 These can be used to improve optimization. Currently, they are only
2526 used in one place: in @file{reorg.c}, instead of guessing which path a
2527 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
2528 exactly determine which path is taken more often.
2531 @item -fstrict-aliasing
2532 Allows the compiler to assume the strictest aliasing rules applicable to
2533 the language being compiled. For C (and C++), this activates
2534 optimizations based on the type of expressions. In particular, an
2535 object of one type is assumed never to reside at the same address as an
2536 object of a different type, unless the types are almost the same. For
2537 example, an @code{unsigned int} can alias an @code{int}, but not a
2538 @code{void*} or a @code{double}. A character type may alias any other
2541 Pay special attention to code like this:
2554 The practice of reading from a different union member than the one most
2555 recently written to (called ``type-punning'') is common. Even with
2556 @samp{-fstrict-aliasing}, type-punning is allowed, provided the memory
2557 is accessed through the union type. So, the code above will work as
2558 expected. However, this code might not:
2570 Every language that wishes to perform language-specific alias analysis
2571 should define a function that computes, given an @code{tree}
2572 node, an alias set for the node. Nodes in different alias sets are not
2573 allowed to alias. For an example, see the C front-end function
2574 @code{c_get_alias_set}.
2579 @node Preprocessor Options
2580 @section Options Controlling the Preprocessor
2581 @cindex preprocessor options
2582 @cindex options, preprocessor
2584 These options control the C preprocessor, which is run on each C source
2585 file before actual compilation.
2587 If you use the @samp{-E} option, nothing is done except preprocessing.
2588 Some of these options make sense only together with @samp{-E} because
2589 they cause the preprocessor output to be unsuitable for actual
2593 @item -include @var{file}
2594 Process @var{file} as input before processing the regular input file.
2595 In effect, the contents of @var{file} are compiled first. Any @samp{-D}
2596 and @samp{-U} options on the command line are always processed before
2597 @samp{-include @var{file}}, regardless of the order in which they are
2598 written. All the @samp{-include} and @samp{-imacros} options are
2599 processed in the order in which they are written.
2601 @item -imacros @var{file}
2602 Process @var{file} as input, discarding the resulting output, before
2603 processing the regular input file. Because the output generated from
2604 @var{file} is discarded, the only effect of @samp{-imacros @var{file}}
2605 is to make the macros defined in @var{file} available for use in the
2608 Any @samp{-D} and @samp{-U} options on the command line are always
2609 processed before @samp{-imacros @var{file}}, regardless of the order in
2610 which they are written. All the @samp{-include} and @samp{-imacros}
2611 options are processed in the order in which they are written.
2613 @item -idirafter @var{dir}
2614 @cindex second include path
2615 Add the directory @var{dir} to the second include path. The directories
2616 on the second include path are searched when a header file is not found
2617 in any of the directories in the main include path (the one that
2620 @item -iprefix @var{prefix}
2621 Specify @var{prefix} as the prefix for subsequent @samp{-iwithprefix}
2624 @item -iwithprefix @var{dir}
2625 Add a directory to the second include path. The directory's name is
2626 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
2627 specified previously with @samp{-iprefix}. If you have not specified a
2628 prefix yet, the directory containing the installed passes of the
2629 compiler is used as the default.
2631 @item -iwithprefixbefore @var{dir}
2632 Add a directory to the main include path. The directory's name is made
2633 by concatenating @var{prefix} and @var{dir}, as in the case of
2634 @samp{-iwithprefix}.
2636 @item -isystem @var{dir}
2637 Add a directory to the beginning of the second include path, marking it
2638 as a system directory, so that it gets the same special treatment as
2639 is applied to the standard system directories.
2642 Do not search the standard system directories for header files. Only
2643 the directories you have specified with @samp{-I} options (and the
2644 current directory, if appropriate) are searched. @xref{Directory
2645 Options}, for information on @samp{-I}.
2647 By using both @samp{-nostdinc} and @samp{-I-}, you can limit the include-file
2648 search path to only those directories you specify explicitly.
2651 Do not predefine any nonstandard macros. (Including architecture flags).
2654 Run only the C preprocessor. Preprocess all the C source files
2655 specified and output the results to standard output or to the
2656 specified output file.
2659 Tell the preprocessor not to discard comments. Used with the
2663 Tell the preprocessor not to generate @samp{#line} directives.
2664 Used with the @samp{-E} option.
2667 @cindex dependencies, make
2669 Tell the preprocessor to output a rule suitable for @code{make}
2670 describing the dependencies of each object file. For each source file,
2671 the preprocessor outputs one @code{make}-rule whose target is the object
2672 file name for that source file and whose dependencies are all the
2673 @code{#include} header files it uses. This rule may be a single line or
2674 may be continued with @samp{\}-newline if it is long. The list of rules
2675 is printed on standard output instead of the preprocessed C program.
2677 @samp{-M} implies @samp{-E}.
2679 Another way to specify output of a @code{make} rule is by setting
2680 the environment variable @code{DEPENDENCIES_OUTPUT} (@pxref{Environment
2684 Like @samp{-M} but the output mentions only the user header files
2685 included with @samp{#include "@var{file}"}. System header files
2686 included with @samp{#include <@var{file}>} are omitted.
2689 Like @samp{-M} but the dependency information is written to a file made by
2690 replacing ".c" with ".d" at the end of the input file names.
2691 This is in addition to compiling the file as specified---@samp{-MD} does
2692 not inhibit ordinary compilation the way @samp{-M} does.
2694 In Mach, you can use the utility @code{md} to merge multiple dependency
2695 files into a single dependency file suitable for using with the @samp{make}
2699 Like @samp{-MD} except mention only user header files, not system
2703 Treat missing header files as generated files and assume they live in the
2704 same directory as the source file. If you specify @samp{-MG}, you
2705 must also specify either @samp{-M} or @samp{-MM}. @samp{-MG} is not
2706 supported with @samp{-MD} or @samp{-MMD}.
2709 Print the name of each header file used, in addition to other normal
2712 @item -A@var{question}(@var{answer})
2713 Assert the answer @var{answer} for @var{question}, in case it is tested
2714 with a preprocessing conditional such as @samp{#if
2715 #@var{question}(@var{answer})}. @samp{-A-} disables the standard
2716 assertions that normally describe the target machine.
2719 Define macro @var{macro} with the string @samp{1} as its definition.
2721 @item -D@var{macro}=@var{defn}
2722 Define macro @var{macro} as @var{defn}. All instances of @samp{-D} on
2723 the command line are processed before any @samp{-U} options.
2726 Undefine macro @var{macro}. @samp{-U} options are evaluated after all
2727 @samp{-D} options, but before any @samp{-include} and @samp{-imacros}
2731 Tell the preprocessor to output only a list of the macro definitions
2732 that are in effect at the end of preprocessing. Used with the @samp{-E}
2736 Tell the preprocessing to pass all macro definitions into the output, in
2737 their proper sequence in the rest of the output.
2740 Like @samp{-dD} except that the macro arguments and contents are omitted.
2741 Only @samp{#define @var{name}} is included in the output.
2744 Support ANSI C trigraphs. The @samp{-ansi} option also has this effect.
2746 @item -Wp,@var{option}
2747 Pass @var{option} as an option to the preprocessor. If @var{option}
2748 contains commas, it is split into multiple options at the commas.
2751 @node Assembler Options
2752 @section Passing Options to the Assembler
2754 @c prevent bad page break with this line
2755 You can pass options to the assembler.
2758 @item -Wa,@var{option}
2759 Pass @var{option} as an option to the assembler. If @var{option}
2760 contains commas, it is split into multiple options at the commas.
2764 @section Options for Linking
2765 @cindex link options
2766 @cindex options, linking
2768 These options come into play when the compiler links object files into
2769 an executable output file. They are meaningless if the compiler is
2770 not doing a link step.
2774 @item @var{object-file-name}
2775 A file name that does not end in a special recognized suffix is
2776 considered to name an object file or library. (Object files are
2777 distinguished from libraries by the linker according to the file
2778 contents.) If linking is done, these object files are used as input
2784 If any of these options is used, then the linker is not run, and
2785 object file names should not be used as arguments. @xref{Overall
2789 @item -l@var{library}
2790 Search the library named @var{library} when linking.
2792 It makes a difference where in the command you write this option; the
2793 linker searches processes libraries and object files in the order they
2794 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
2795 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
2796 to functions in @samp{z}, those functions may not be loaded.
2798 The linker searches a standard list of directories for the library,
2799 which is actually a file named @file{lib@var{library}.a}. The linker
2800 then uses this file as if it had been specified precisely by name.
2802 The directories searched include several standard system directories
2803 plus any that you specify with @samp{-L}.
2805 Normally the files found this way are library files---archive files
2806 whose members are object files. The linker handles an archive file by
2807 scanning through it for members which define symbols that have so far
2808 been referenced but not defined. But if the file that is found is an
2809 ordinary object file, it is linked in the usual fashion. The only
2810 difference between using an @samp{-l} option and specifying a file name
2811 is that @samp{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
2812 and searches several directories.
2815 You need this special case of the @samp{-l} option in order to
2816 link an Objective C program.
2819 Do not use the standard system startup files when linking.
2820 The standard system libraries are used normally, unless @code{-nostdlib}
2821 or @code{-nodefaultlibs} is used.
2823 @item -nodefaultlibs
2824 Do not use the standard system libraries when linking.
2825 Only the libraries you specify will be passed to the linker.
2826 The standard startup files are used normally, unless @code{-nostartfiles}
2827 is used. The compiler may generate calls to memcmp, memset, and memcpy
2828 for System V (and ANSI C) environments or to bcopy and bzero for
2829 BSD environments. These entries are usually resolved by entries in
2830 libc. These entry points should be supplied through some other
2831 mechanism when this option is specified.
2834 Do not use the standard system startup files or libraries when linking.
2835 No startup files and only the libraries you specify will be passed to
2836 the linker. The compiler may generate calls to memcmp, memset, and memcpy
2837 for System V (and ANSI C) environments or to bcopy and bzero for
2838 BSD environments. These entries are usually resolved by entries in
2839 libc. These entry points should be supplied through some other
2840 mechanism when this option is specified.
2842 @cindex @code{-lgcc}, use with @code{-nostdlib}
2843 @cindex @code{-nostdlib} and unresolved references
2844 @cindex unresolved references and @code{-nostdlib}
2845 @cindex @code{-lgcc}, use with @code{-nodefaultlibs}
2846 @cindex @code{-nodefaultlibs} and unresolved references
2847 @cindex unresolved references and @code{-nodefaultlibs}
2848 One of the standard libraries bypassed by @samp{-nostdlib} and
2849 @samp{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
2850 that GNU CC uses to overcome shortcomings of particular machines, or special
2851 needs for some languages.
2853 (@xref{Interface,,Interfacing to GNU CC Output}, for more discussion of
2857 (@xref{Interface,,Interfacing to GNU CC Output,gcc.info,Porting GNU CC},
2858 for more discussion of @file{libgcc.a}.)
2860 In most cases, you need @file{libgcc.a} even when you want to avoid
2861 other standard libraries. In other words, when you specify @samp{-nostdlib}
2862 or @samp{-nodefaultlibs} you should usually specify @samp{-lgcc} as well.
2863 This ensures that you have no unresolved references to internal GNU CC
2864 library subroutines. (For example, @samp{__main}, used to ensure C++
2865 constructors will be called; @pxref{Collect2,,@code{collect2}}.)
2868 Remove all symbol table and relocation information from the executable.
2871 On systems that support dynamic linking, this prevents linking with the shared
2872 libraries. On other systems, this option has no effect.
2875 Produce a shared object which can then be linked with other objects to
2876 form an executable. Not all systems support this option. You must
2877 also specify @samp{-fpic} or @samp{-fPIC} on some systems when
2878 you specify this option.
2881 Bind references to global symbols when building a shared object. Warn
2882 about any unresolved references (unless overridden by the link editor
2883 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
2886 @item -Xlinker @var{option}
2887 Pass @var{option} as an option to the linker. You can use this to
2888 supply system-specific linker options which GNU CC does not know how to
2891 If you want to pass an option that takes an argument, you must use
2892 @samp{-Xlinker} twice, once for the option and once for the argument.
2893 For example, to pass @samp{-assert definitions}, you must write
2894 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
2895 @samp{-Xlinker "-assert definitions"}, because this passes the entire
2896 string as a single argument, which is not what the linker expects.
2898 @item -Wl,@var{option}
2899 Pass @var{option} as an option to the linker. If @var{option} contains
2900 commas, it is split into multiple options at the commas.
2902 @item -u @var{symbol}
2903 Pretend the symbol @var{symbol} is undefined, to force linking of
2904 library modules to define it. You can use @samp{-u} multiple times with
2905 different symbols to force loading of additional library modules.
2908 @node Directory Options
2909 @section Options for Directory Search
2910 @cindex directory options
2911 @cindex options, directory search
2914 These options specify directories to search for header files, for
2915 libraries and for parts of the compiler:
2919 Add the directory @var{dir} to the head of the list of directories to be
2920 searched for header files. This can be used to override a system header
2921 file, substituting your own version, since these directories are
2922 searched before the system header file directories. If you use more
2923 than one @samp{-I} option, the directories are scanned in left-to-right
2924 order; the standard system directories come after.
2927 Any directories you specify with @samp{-I} options before the @samp{-I-}
2928 option are searched only for the case of @samp{#include "@var{file}"};
2929 they are not searched for @samp{#include <@var{file}>}.
2931 If additional directories are specified with @samp{-I} options after
2932 the @samp{-I-}, these directories are searched for all @samp{#include}
2933 directives. (Ordinarily @emph{all} @samp{-I} directories are used
2936 In addition, the @samp{-I-} option inhibits the use of the current
2937 directory (where the current input file came from) as the first search
2938 directory for @samp{#include "@var{file}"}. There is no way to
2939 override this effect of @samp{-I-}. With @samp{-I.} you can specify
2940 searching the directory which was current when the compiler was
2941 invoked. That is not exactly the same as what the preprocessor does
2942 by default, but it is often satisfactory.
2944 @samp{-I-} does not inhibit the use of the standard system directories
2945 for header files. Thus, @samp{-I-} and @samp{-nostdinc} are
2949 Add directory @var{dir} to the list of directories to be searched
2952 @item -B@var{prefix}
2953 This option specifies where to find the executables, libraries,
2954 include files, and data files of the compiler itself.
2956 The compiler driver program runs one or more of the subprograms
2957 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
2958 @var{prefix} as a prefix for each program it tries to run, both with and
2959 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
2961 For each subprogram to be run, the compiler driver first tries the
2962 @samp{-B} prefix, if any. If that name is not found, or if @samp{-B}
2963 was not specified, the driver tries two standard prefixes, which are
2964 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
2965 those results in a file name that is found, the unmodified program
2966 name is searched for using the directories specified in your
2967 @samp{PATH} environment variable.
2969 @samp{-B} prefixes that effectively specify directory names also apply
2970 to libraries in the linker, because the compiler translates these
2971 options into @samp{-L} options for the linker. They also apply to
2972 includes files in the preprocessor, because the compiler translates these
2973 options into @samp{-isystem} options for the preprocessor. In this case,
2974 the compiler appends @samp{include} to the prefix.
2976 The run-time support file @file{libgcc.a} can also be searched for using
2977 the @samp{-B} prefix, if needed. If it is not found there, the two
2978 standard prefixes above are tried, and that is all. The file is left
2979 out of the link if it is not found by those means.
2981 Another way to specify a prefix much like the @samp{-B} prefix is to use
2982 the environment variable @code{GCC_EXEC_PREFIX}. @xref{Environment
2985 @item -specs=@var{file}
2986 Process @var{file} after the compiler reads in the standard @file{specs}
2987 file, in order to override the defaults that the @file{gcc} driver
2988 program uses when determining what switches to pass to @file{cc1},
2989 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
2990 @samp{-specs=}@var{file} can be specified on the command line, and they
2991 are processed in order, from left to right.
2995 @section Specifying subprocesses and the switches to pass to them
2997 @code{GCC} is a driver program. It performs its job by invoking a
2998 sequence of other programs to do the work of compiling, assembling and
2999 linking. GCC interprets its command-line parameters and uses these to
3000 deduce which programs it should invoke, and which command-line options
3001 it ought to place on their command lines. This behaviour is controlled
3002 by @dfn{spec strings}. In most cases there is one spec string for each
3003 program that GCC can invoke, but a few programs have multiple spec
3004 strings to control their behaviour. The spec strings built into GCC can
3005 be overridden by using the @samp{-specs=} command-line switch to specify
3008 @dfn{Spec files} are plaintext files that are used to construct spec
3009 strings. They consist of a sequence of directives separated by blank
3010 lines. The type of directive is determined by the first non-whitespace
3011 character on the line and it can be one of the following:
3014 @item %@var{command}
3015 Issues a @var{command} to the spec file processor. The commands that can
3019 @item %include <@var{file}>
3021 Search for @var{file} and insert its text at the current point in the
3024 @item %include_noerr <@var{file}>
3025 @cindex %include_noerr
3026 Just like @samp{%include}, but do not generate an error message if the include
3027 file cannot be found.
3029 @item %rename @var{old_name} @var{new_name}
3031 Rename the spec string @var{old_name} to @var{new_name}.
3035 @item *[@var{spec_name}]:
3036 This tells the compiler to create, override or delete the named spec
3037 string. All lines after this directive up to the next directive or
3038 blank line are considered to be the text for the spec string. If this
3039 results in an empty string then the spec will be deleted. (Or, if the
3040 spec did not exist, then nothing will happened.) Otherwise, if the spec
3041 does not currently exist a new spec will be created. If the spec does
3042 exist then its contents will be overridden by the text of this
3043 directive, unless the first character of that text is the @samp{+}
3044 character, in which case the text will be appended to the spec.
3046 @item [@var{suffix}]:
3047 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
3048 and up to the next directive or blank line are considered to make up the
3049 spec string for the indicated suffix. When the compiler encounters an
3050 input file with the named suffix, it will processes the spec string in
3051 order to work out how to compile that file. For example:
3058 This says that any input file whose name ends in @samp{.ZZ} should be
3059 passed to the program @samp{z-compile}, which should be invoked with the
3060 command-line switch @samp{-input} and with the result of performing the
3061 @samp{%i} substitution. (See below.)
3063 As an alternative to providing a spec string, the text that follows a
3064 suffix directive can be one of the following:
3067 @item @@@var{language}
3068 This says that the suffix is an alias for a known @var{language}. This is
3069 similar to using the @code{-x} command-line switch to GCC to specify a
3070 language explicitly. For example:
3077 Says that .ZZ files are, in fact, C++ source files.
3080 This causes an error messages saying:
3083 @var{name} compiler not installed on this system.
3087 GCC already has an extensive list of suffixes built into it.
3088 This directive will add an entry to the end of the list of suffixes, but
3089 since the list is searched from the end backwards, it is effectively
3090 possible to override earlier entries using this technique.
3094 GCC has the following spec strings built into it. Spec files can
3095 override these strings or create their own. Note that individual
3096 targets can also add their own spec strings to this list.
3099 asm Options to pass to the assembler
3100 asm_final Options to pass to the assembler post-processor
3101 cpp Options to pass to the C preprocessor
3102 cc1 Options to pass to the C compiler
3103 cc1plus Options to pass to the C++ compiler
3104 endfile Object files to include at the end of the link
3105 link Options to pass to the linker
3106 lib Libraries to include on the command line to the linker
3107 libgcc Decides which GCC support library to pass to the linker
3108 linker Sets the name of the linker
3109 predefines Defines to be passed to the C preprocessor
3110 signed_char Defines to pass to CPP to say whether @code{char} is signed by default
3111 startfile Object files to include at the start of the link
3114 Here is a small example of a spec file:
3120 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
3123 This example renames the spec called @samp{lib} to @samp{old_lib} and
3124 then overrides the previous definition of @samp{lib} with a new one.
3125 The new definition adds in some extra command-line options before
3126 including the text of the old definition.
3128 @dfn{Spec strings} are a list of command-line options to be passed to their
3129 corresponding program. In addition, the spec strings can contain
3130 @samp{%}-prefixed sequences to substitute variable text or to
3131 conditionally insert text into the command line. Using these constructs
3132 it is possible to generate quite complex command lines.
3134 Here is a table of all defined @samp{%}-sequences for spec
3135 strings. Note that spaces are not generated automatically around the
3136 results of expanding these sequences. Therefore you can concatenate them
3137 together or combine them with constant text in a single argument.
3141 Substitute one @samp{%} into the program name or argument.
3144 Substitute the name of the input file being processed.
3147 Substitute the basename of the input file being processed.
3148 This is the substring up to (and not including) the last period
3149 and not including the directory.
3152 Marks the argument containing or following the @samp{%d} as a
3153 temporary file name, so that that file will be deleted if GCC exits
3154 successfully. Unlike @samp{%g}, this contributes no text to the
3157 @item %g@var{suffix}
3158 Substitute a file name that has suffix @var{suffix} and is chosen
3159 once per compilation, and mark the argument in the same way as
3160 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
3161 name is now chosen in a way that is hard to predict even when previously
3162 chosen file names are known. For example, @samp{%g.s ... %g.o ... %g.s}
3163 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
3164 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
3165 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
3166 was simply substituted with a file name chosen once per compilation,
3167 without regard to any appended suffix (which was therefore treated
3168 just like ordinary text), making such attacks more likely to succeed.
3170 @item %u@var{suffix}
3171 Like @samp{%g}, but generates a new temporary file name even if
3172 @samp{%u@var{suffix}} was already seen.
3174 @item %U@var{suffix}
3175 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
3176 new one if there is no such last file name. In the absence of any
3177 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
3178 the same suffix @emph{space}, so @samp{%g.s ... %U.s ... %g.s ... %U.s}
3179 would involve the generation of two distinct file names, one
3180 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
3181 simply substituted with a file name chosen for the previous @samp{%u},
3182 without regard to any appended suffix.
3185 Marks the argument containing or following the @samp{%w} as the
3186 designated output file of this compilation. This puts the argument
3187 into the sequence of arguments that @samp{%o} will substitute later.
3190 Substitutes the names of all the output files, with spaces
3191 automatically placed around them. You should write spaces
3192 around the @samp{%o} as well or the results are undefined.
3193 @samp{%o} is for use in the specs for running the linker.
3194 Input files whose names have no recognized suffix are not compiled
3195 at all, but they are included among the output files, so they will
3199 Substitutes the suffix for object files. Note that this is
3200 handled specially when it immediately follows @samp{%g, %u, or %U},
3201 because of the need for those to form complete file names. The
3202 handling is such that @samp{%O} is treated exactly as if it had already
3203 been substituted, except that @samp{%g, %u, and %U} do not currently
3204 support additional @var{suffix} characters following @samp{%O} as they would
3205 following, for example, @samp{.o}.
3208 Substitutes the standard macro predefinitions for the
3209 current target machine. Use this when running @code{cpp}.
3212 Like @samp{%p}, but puts @samp{__} before and after the name of each
3213 predefined macro, except for macros that start with @samp{__} or with
3214 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ANSI
3218 Substitute a @samp{-iprefix} option made from GCC_EXEC_PREFIX.
3221 Current argument is the name of a library or startup file of some sort.
3222 Search for that file in a standard list of directories and substitute
3223 the full name found.
3226 Print @var{str} as an error message. @var{str} is terminated by a newline.
3227 Use this when inconsistent options are detected.
3230 Output @samp{-} if the input for the current command is coming from a pipe.
3233 Substitute the contents of spec string @var{name} at this point.
3236 Like @samp{%(...)} but put @samp{__} around @samp{-D} arguments.
3238 @item %x@{@var{option}@}
3239 Accumulate an option for @samp{%X}.
3242 Output the accumulated linker options specified by @samp{-Wl} or a @samp{%x}
3246 Output the accumulated assembler options specified by @samp{-Wa}.
3249 Output the accumulated preprocessor options specified by @samp{-Wp}.
3252 Substitute the major version number of GCC.
3253 (For version 2.9.5, this is 2.)
3256 Substitute the minor version number of GCC.
3257 (For version 2.9.5, this is 9.)
3260 Process the @code{asm} spec. This is used to compute the
3261 switches to be passed to the assembler.
3264 Process the @code{asm_final} spec. This is a spec string for
3265 passing switches to an assembler post-processor, if such a program is
3269 Process the @code{link} spec. This is the spec for computing the
3270 command line passed to the linker. Typically it will make use of the
3271 @samp{%L %G %S %D and %E} sequences.
3274 Dump out a @samp{-L} option for each directory that GCC believes might
3275 contain startup files. If the target supports multilibs then the
3276 current multilib directory will be prepended to each of these paths.
3279 Process the @code{lib} spec. This is a spec string for deciding which
3280 libraries should be included on the command line to the linker.
3283 Process the @code{libgcc} spec. This is a spec string for deciding
3284 which GCC support library should be included on the command line to the linker.
3287 Process the @code{startfile} spec. This is a spec for deciding which
3288 object files should be the first ones passed to the linker. Typically
3289 this might be a file named @file{crt0.o}.
3292 Process the @code{endfile} spec. This is a spec string that specifies
3293 the last object files that will be passed to the linker.
3296 Process the @code{cpp} spec. This is used to construct the arguments
3297 to be passed to the C preprocessor.
3300 Process the @code{signed_char} spec. This is intended to be used
3301 to tell cpp whether a char is signed. It typically has the definition:
3303 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
3307 Process the @code{cc1} spec. This is used to construct the options to be
3308 passed to the actual C compiler (@samp{cc1}).
3311 Process the @code{cc1plus} spec. This is used to construct the options to be
3312 passed to the actual C++ compiler (@samp{cc1plus}).
3315 Substitute the variable part of a matched option. See below.
3316 Note that each comma in the substituted string is replaced by
3320 Substitutes the @code{-S} switch, if that switch was given to GCC.
3321 If that switch was not specified, this substitutes nothing. Note that
3322 the leading dash is omitted when specifying this option, and it is
3323 automatically inserted if the substitution is performed. Thus the spec
3324 string @samp{%@{foo@}} would match the command-line option @samp{-foo}
3325 and would output the command line option @samp{-foo}.
3327 @item %W@{@code{S}@}
3328 Like %@{@code{S}@} but mark last argument supplied within as a file to be
3331 @item %@{@code{S}*@}
3332 Substitutes all the switches specified to GCC whose names start
3333 with @code{-S}, but which also take an argument. This is used for
3334 switches like @samp{-o, -D, -I}, etc. GCC considers @samp{-o foo} as being
3335 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
3336 text, including the space. Thus two arguments would be generated.
3338 @item %@{^@code{S}*@}
3339 Like %@{@code{S}*@}, but don't put a blank between a switch and its
3340 argument. Thus %@{^o*@} would only generate one argument, not two.
3342 @item %@{@code{S}*:@code{X}@}
3343 Substitutes @code{X} if one or more switches whose names start with
3344 @code{-S} are specified to GCC. Note that the tail part of the
3345 @code{-S} option (i.e. the part matched by the @samp{*}) will be substituted
3346 for each occurrence of @samp{%*} within @code{X}.
3348 @item %@{@code{S}:@code{X}@}
3349 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC.
3351 @item %@{!@code{S}:@code{X}@}
3352 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC.
3354 @item %@{|@code{S}:@code{X}@}
3355 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
3357 @item %@{|!@code{S}:@code{X}@}
3358 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
3360 @item %@{.@code{S}:@code{X}@}
3361 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
3363 @item %@{!.@code{S}:@code{X}@}
3364 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
3366 @item %@{@code{S}|@code{P}:@code{X}@}
3367 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC. This may be
3368 combined with @samp{!} and @samp{.} sequences as well, although they
3369 have a stronger binding than the @samp{|}. For example a spec string
3373 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
3376 will output the following command-line options from the following input
3377 command-line options:
3382 -d fred.c -foo -baz -boggle
3383 -d jim.d -bar -baz -boggle
3388 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
3389 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
3390 or spaces, or even newlines. They are processed as usual, as described
3393 The @samp{-O, -f, -m, and -W} switches are handled specifically in these
3394 constructs. If another value of @samp{-O} or the negated form of a @samp{-f, -m, or
3395 -W} switch is found later in the command line, the earlier switch
3396 value is ignored, except with @{@code{S}*@} where @code{S} is just one
3397 letter, which passes all matching options.
3399 The character @samp{|} at the beginning of the predicate text is used to indicate
3400 that a command should be piped to the following command, but only if @samp{-pipe}
3403 It is built into GCC which switches take arguments and which do not.
3404 (You might think it would be useful to generalize this to allow each
3405 compiler's spec to say which switches take arguments. But this cannot
3406 be done in a consistent fashion. GCC cannot even decide which input
3407 files have been specified without knowing which switches take arguments,
3408 and it must know which input files to compile in order to tell which
3411 GCC also knows implicitly that arguments starting in @samp{-l} are to be
3412 treated as compiler output files, and passed to the linker in their
3413 proper position among the other output files.
3415 @node Target Options
3416 @section Specifying Target Machine and Compiler Version
3417 @cindex target options
3418 @cindex cross compiling
3419 @cindex specifying machine version
3420 @cindex specifying compiler version and target machine
3421 @cindex compiler version, specifying
3422 @cindex target machine, specifying
3424 By default, GNU CC compiles code for the same type of machine that you
3425 are using. However, it can also be installed as a cross-compiler, to
3426 compile for some other type of machine. In fact, several different
3427 configurations of GNU CC, for different target machines, can be
3428 installed side by side. Then you specify which one to use with the
3431 In addition, older and newer versions of GNU CC can be installed side
3432 by side. One of them (probably the newest) will be the default, but
3433 you may sometimes wish to use another.
3436 @item -b @var{machine}
3437 The argument @var{machine} specifies the target machine for compilation.
3438 This is useful when you have installed GNU CC as a cross-compiler.
3440 The value to use for @var{machine} is the same as was specified as the
3441 machine type when configuring GNU CC as a cross-compiler. For
3442 example, if a cross-compiler was configured with @samp{configure
3443 i386v}, meaning to compile for an 80386 running System V, then you
3444 would specify @samp{-b i386v} to run that cross compiler.
3446 When you do not specify @samp{-b}, it normally means to compile for
3447 the same type of machine that you are using.
3449 @item -V @var{version}
3450 The argument @var{version} specifies which version of GNU CC to run.
3451 This is useful when multiple versions are installed. For example,
3452 @var{version} might be @samp{2.0}, meaning to run GNU CC version 2.0.
3454 The default version, when you do not specify @samp{-V}, is the last
3455 version of GNU CC that you installed.
3458 The @samp{-b} and @samp{-V} options actually work by controlling part of
3459 the file name used for the executable files and libraries used for
3460 compilation. A given version of GNU CC, for a given target machine, is
3461 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.@refill
3463 Thus, sites can customize the effect of @samp{-b} or @samp{-V} either by
3464 changing the names of these directories or adding alternate names (or
3465 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
3466 file @file{80386} is a link to the file @file{i386v}, then @samp{-b
3467 80386} becomes an alias for @samp{-b i386v}.
3469 In one respect, the @samp{-b} or @samp{-V} do not completely change
3470 to a different compiler: the top-level driver program @code{gcc}
3471 that you originally invoked continues to run and invoke the other
3472 executables (preprocessor, compiler per se, assembler and linker)
3473 that do the real work. However, since no real work is done in the
3474 driver program, it usually does not matter that the driver program
3475 in use is not the one for the specified target and version.
3477 The only way that the driver program depends on the target machine is
3478 in the parsing and handling of special machine-specific options.
3479 However, this is controlled by a file which is found, along with the
3480 other executables, in the directory for the specified version and
3481 target machine. As a result, a single installed driver program adapts
3482 to any specified target machine and compiler version.
3484 The driver program executable does control one significant thing,
3485 however: the default version and target machine. Therefore, you can
3486 install different instances of the driver program, compiled for
3487 different targets or versions, under different names.
3489 For example, if the driver for version 2.0 is installed as @code{ogcc}
3490 and that for version 2.1 is installed as @code{gcc}, then the command
3491 @code{gcc} will use version 2.1 by default, while @code{ogcc} will use
3492 2.0 by default. However, you can choose either version with either
3493 command with the @samp{-V} option.
3495 @node Submodel Options
3496 @section Hardware Models and Configurations
3497 @cindex submodel options
3498 @cindex specifying hardware config
3499 @cindex hardware models and configurations, specifying
3500 @cindex machine dependent options
3502 Earlier we discussed the standard option @samp{-b} which chooses among
3503 different installed compilers for completely different target
3504 machines, such as Vax vs. 68000 vs. 80386.
3506 In addition, each of these target machine types can have its own
3507 special options, starting with @samp{-m}, to choose among various
3508 hardware models or configurations---for example, 68010 vs 68020,
3509 floating coprocessor or none. A single installed version of the
3510 compiler can compile for any model or configuration, according to the
3513 Some configurations of the compiler also support additional special
3514 options, usually for compatibility with other compilers on the same
3518 These options are defined by the macro @code{TARGET_SWITCHES} in the
3519 machine description. The default for the options is also defined by
3520 that macro, which enables you to change the defaults.
3535 * RS/6000 and PowerPC Options::
3540 * Intel 960 Options::
3541 * DEC Alpha Options::
3545 * System V Options::
3546 * TMS320C3x/C4x Options::
3552 @node M680x0 Options
3553 @subsection M680x0 Options
3554 @cindex M680x0 options
3556 These are the @samp{-m} options defined for the 68000 series. The default
3557 values for these options depends on which style of 68000 was selected when
3558 the compiler was configured; the defaults for the most common choices are
3564 Generate output for a 68000. This is the default
3565 when the compiler is configured for 68000-based systems.
3567 Use this option for microcontrollers with a 68000 or EC000 core,
3568 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
3572 Generate output for a 68020. This is the default
3573 when the compiler is configured for 68020-based systems.
3576 Generate output containing 68881 instructions for floating point.
3577 This is the default for most 68020 systems unless @samp{-nfp} was
3578 specified when the compiler was configured.
3581 Generate output for a 68030. This is the default when the compiler is
3582 configured for 68030-based systems.
3585 Generate output for a 68040. This is the default when the compiler is
3586 configured for 68040-based systems.
3588 This option inhibits the use of 68881/68882 instructions that have to be
3589 emulated by software on the 68040. Use this option if your 68040 does not
3590 have code to emulate those instructions.
3593 Generate output for a 68060. This is the default when the compiler is
3594 configured for 68060-based systems.
3596 This option inhibits the use of 68020 and 68881/68882 instructions that
3597 have to be emulated by software on the 68060. Use this option if your 68060
3598 does not have code to emulate those instructions.
3601 Generate output for a CPU32. This is the default
3602 when the compiler is configured for CPU32-based systems.
3604 Use this option for microcontrollers with a
3605 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
3606 68336, 68340, 68341, 68349 and 68360.
3609 Generate output for a 520X "coldfire" family cpu. This is the default
3610 when the compiler is configured for 520X-based systems.
3612 Use this option for microcontroller with a 5200 core, including
3613 the MCF5202, MCF5203, MCF5204 and MCF5202.
3617 Generate output for a 68040, without using any of the new instructions.
3618 This results in code which can run relatively efficiently on either a
3619 68020/68881 or a 68030 or a 68040. The generated code does use the
3620 68881 instructions that are emulated on the 68040.
3623 Generate output for a 68060, without using any of the new instructions.
3624 This results in code which can run relatively efficiently on either a
3625 68020/68881 or a 68030 or a 68040. The generated code does use the
3626 68881 instructions that are emulated on the 68060.
3629 Generate output containing Sun FPA instructions for floating point.
3632 Generate output containing library calls for floating point.
3633 @strong{Warning:} the requisite libraries are not available for all m68k
3634 targets. Normally the facilities of the machine's usual C compiler are
3635 used, but this can't be done directly in cross-compilation. You must
3636 make your own arrangements to provide suitable library functions for
3637 cross-compilation. The embedded targets @samp{m68k-*-aout} and
3638 @samp{m68k-*-coff} do provide software floating point support.
3641 Consider type @code{int} to be 16 bits wide, like @code{short int}.
3644 Do not use the bit-field instructions. The @samp{-m68000}, @samp{-mcpu32}
3645 and @samp{-m5200} options imply @w{@samp{-mnobitfield}}.
3648 Do use the bit-field instructions. The @samp{-m68020} option implies
3649 @samp{-mbitfield}. This is the default if you use a configuration
3650 designed for a 68020.
3653 Use a different function-calling convention, in which functions
3654 that take a fixed number of arguments return with the @code{rtd}
3655 instruction, which pops their arguments while returning. This
3656 saves one instruction in the caller since there is no need to pop
3657 the arguments there.
3659 This calling convention is incompatible with the one normally
3660 used on Unix, so you cannot use it if you need to call libraries
3661 compiled with the Unix compiler.
3663 Also, you must provide function prototypes for all functions that
3664 take variable numbers of arguments (including @code{printf});
3665 otherwise incorrect code will be generated for calls to those
3668 In addition, seriously incorrect code will result if you call a
3669 function with too many arguments. (Normally, extra arguments are
3670 harmlessly ignored.)
3672 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
3673 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
3676 @itemx -mno-align-int
3677 Control whether GNU CC aligns @code{int}, @code{long}, @code{long long},
3678 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
3679 boundary (@samp{-malign-int}) or a 16-bit boundary (@samp{-mno-align-int}).
3680 Aligning variables on 32-bit boundaries produces code that runs somewhat
3681 faster on processors with 32-bit busses at the expense of more memory.
3683 @strong{Warning:} if you use the @samp{-malign-int} switch, GNU CC will
3684 align structures containing the above types differently than
3685 most published application binary interface specifications for the m68k.
3688 Use the pc-relative addressing mode of the 68000 directly, instead of
3689 using a global offset table. At present, this option implies -fpic,
3690 allowing at most a 16-bit offset for pc-relative addressing. -fPIC is
3691 not presently supported with -mpcrel, though this could be supported for
3692 68020 and higher processors.
3697 @subsection VAX Options
3700 These @samp{-m} options are defined for the Vax:
3704 Do not output certain jump instructions (@code{aobleq} and so on)
3705 that the Unix assembler for the Vax cannot handle across long
3709 Do output those jump instructions, on the assumption that you
3710 will assemble with the GNU assembler.
3713 Output code for g-format floating point numbers instead of d-format.
3717 @subsection SPARC Options
3718 @cindex SPARC options
3720 These @samp{-m} switches are supported on the SPARC:
3725 Specify @samp{-mapp-regs} to generate output using the global registers
3726 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
3729 To be fully SVR4 ABI compliant at the cost of some performance loss,
3730 specify @samp{-mno-app-regs}. You should compile libraries and system
3731 software with this option.
3735 Generate output containing floating point instructions. This is the
3740 Generate output containing library calls for floating point.
3741 @strong{Warning:} the requisite libraries are not available for all SPARC
3742 targets. Normally the facilities of the machine's usual C compiler are
3743 used, but this cannot be done directly in cross-compilation. You must make
3744 your own arrangements to provide suitable library functions for
3745 cross-compilation. The embedded targets @samp{sparc-*-aout} and
3746 @samp{sparclite-*-*} do provide software floating point support.
3748 @samp{-msoft-float} changes the calling convention in the output file;
3749 therefore, it is only useful if you compile @emph{all} of a program with
3750 this option. In particular, you need to compile @file{libgcc.a}, the
3751 library that comes with GNU CC, with @samp{-msoft-float} in order for
3754 @item -mhard-quad-float
3755 Generate output containing quad-word (long double) floating point
3758 @item -msoft-quad-float
3759 Generate output containing library calls for quad-word (long double)
3760 floating point instructions. The functions called are those specified
3761 in the SPARC ABI. This is the default.
3763 As of this writing, there are no sparc implementations that have hardware
3764 support for the quad-word floating point instructions. They all invoke
3765 a trap handler for one of these instructions, and then the trap handler
3766 emulates the effect of the instruction. Because of the trap handler overhead,
3767 this is much slower than calling the ABI library routines. Thus the
3768 @samp{-msoft-quad-float} option is the default.
3772 With @samp{-mepilogue} (the default), the compiler always emits code for
3773 function exit at the end of each function. Any function exit in
3774 the middle of the function (such as a return statement in C) will
3775 generate a jump to the exit code at the end of the function.
3777 With @samp{-mno-epilogue}, the compiler tries to emit exit code inline
3778 at every function exit.
3782 With @samp{-mflat}, the compiler does not generate save/restore instructions
3783 and will use a "flat" or single register window calling convention.
3784 This model uses %i7 as the frame pointer and is compatible with the normal
3785 register window model. Code from either may be intermixed.
3786 The local registers and the input registers (0-5) are still treated as
3787 "call saved" registers and will be saved on the stack as necessary.
3789 With @samp{-mno-flat} (the default), the compiler emits save/restore
3790 instructions (except for leaf functions) and is the normal mode of operation.
3792 @item -mno-unaligned-doubles
3793 @itemx -munaligned-doubles
3794 Assume that doubles have 8 byte alignment. This is the default.
3796 With @samp{-munaligned-doubles}, GNU CC assumes that doubles have 8 byte
3797 alignment only if they are contained in another type, or if they have an
3798 absolute address. Otherwise, it assumes they have 4 byte alignment.
3799 Specifying this option avoids some rare compatibility problems with code
3800 generated by other compilers. It is not the default because it results
3801 in a performance loss, especially for floating point code.
3805 These two options select variations on the SPARC architecture.
3807 By default (unless specifically configured for the Fujitsu SPARClite),
3808 GCC generates code for the v7 variant of the SPARC architecture.
3810 @samp{-mv8} will give you SPARC v8 code. The only difference from v7
3811 code is that the compiler emits the integer multiply and integer
3812 divide instructions which exist in SPARC v8 but not in SPARC v7.
3814 @samp{-msparclite} will give you SPARClite code. This adds the integer
3815 multiply, integer divide step and scan (@code{ffs}) instructions which
3816 exist in SPARClite but not in SPARC v7.
3818 These options are deprecated and will be deleted in GNU CC 2.9.
3819 They have been replaced with @samp{-mcpu=xxx}.
3823 These two options select the processor for which the code is optimised.
3825 With @samp{-mcypress} (the default), the compiler optimizes code for the
3826 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
3827 This is also appropriate for the older SparcStation 1, 2, IPX etc.
3829 With @samp{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
3830 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
3831 of the full SPARC v8 instruction set.
3833 These options are deprecated and will be deleted in GNU CC 2.9.
3834 They have been replaced with @samp{-mcpu=xxx}.
3836 @item -mcpu=@var{cpu_type}
3837 Set the instruction set, register set, and instruction scheduling parameters
3838 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
3839 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
3840 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
3841 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
3843 Default instruction scheduling parameters are used for values that select
3844 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
3845 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
3847 Here is a list of each supported architecture and their supported
3852 v8: supersparc, hypersparc
3853 sparclite: f930, f934, sparclite86x
3858 @item -mtune=@var{cpu_type}
3859 Set the instruction scheduling parameters for machine type
3860 @var{cpu_type}, but do not set the instruction set or register set that the
3861 option @samp{-mcpu=}@var{cpu_type} would.
3863 The same values for @samp{-mcpu=}@var{cpu_type} are used for
3864 @samp{-mtune=}@*@var{cpu_type}, though the only useful values are those that
3865 select a particular cpu implementation: @samp{cypress}, @samp{supersparc},
3866 @samp{hypersparc}, @samp{f930}, @samp{f934}, @samp{sparclite86x},
3867 @samp{tsc701}, @samp{ultrasparc}.
3869 @item -malign-loops=@var{num}
3870 Align loops to a 2 raised to a @var{num} byte boundary. If
3871 @samp{-malign-loops} is not specified, the default is 2.
3873 @item -malign-jumps=@var{num}
3874 Align instructions that are only jumped to to a 2 raised to a @var{num}
3875 byte boundary. If @samp{-malign-jumps} is not specified, the default is 2.
3877 @item -malign-functions=@var{num}
3878 Align the start of functions to a 2 raised to @var{num} byte boundary.
3879 If @samp{-malign-functions} is not specified, the default is 2 if compiling
3880 for 32 bit sparc, and 5 if compiling for 64 bit sparc.
3884 These @samp{-m} switches are supported in addition to the above
3885 on the SPARCLET processor.
3888 @item -mlittle-endian
3889 Generate code for a processor running in little-endian mode.
3892 Treat register @code{%g0} as a normal register.
3893 GCC will continue to clobber it as necessary but will not assume
3894 it always reads as 0.
3896 @item -mbroken-saverestore
3897 Generate code that does not use non-trivial forms of the @code{save} and
3898 @code{restore} instructions. Early versions of the SPARCLET processor do
3899 not correctly handle @code{save} and @code{restore} instructions used with
3900 arguments. They correctly handle them used without arguments. A @code{save}
3901 instruction used without arguments increments the current window pointer
3902 but does not allocate a new stack frame. It is assumed that the window
3903 overflow trap handler will properly handle this case as will interrupt
3907 These @samp{-m} switches are supported in addition to the above
3908 on SPARC V9 processors in 64 bit environments.
3911 @item -mlittle-endian
3912 Generate code for a processor running in little-endian mode.
3916 Generate code for a 32 bit or 64 bit environment.
3917 The 32 bit environment sets int, long and pointer to 32 bits.
3918 The 64 bit environment sets int to 32 bits and long and pointer
3921 @item -mcmodel=medlow
3922 Generate code for the Medium/Low code model: the program must be linked
3923 in the low 32 bits of the address space. Pointers are 64 bits.
3924 Programs can be statically or dynamically linked.
3926 @item -mcmodel=medmid
3927 Generate code for the Medium/Middle code model: the program must be linked
3928 in the low 44 bits of the address space, the text segment must be less than
3929 2G bytes, and data segment must be within 2G of the text segment.
3930 Pointers are 64 bits.
3932 @item -mcmodel=medany
3933 Generate code for the Medium/Anywhere code model: the program may be linked
3934 anywhere in the address space, the text segment must be less than
3935 2G bytes, and data segment must be within 2G of the text segment.
3936 Pointers are 64 bits.
3938 @item -mcmodel=embmedany
3939 Generate code for the Medium/Anywhere code model for embedded systems:
3940 assume a 32 bit text and a 32 bit data segment, both starting anywhere
3941 (determined at link time). Register %g4 points to the base of the
3942 data segment. Pointers still 64 bits.
3943 Programs are statically linked, PIC is not supported.
3946 @itemx -mno-stack-bias
3947 With @samp{-mstack-bias}, GNU CC assumes that the stack pointer, and
3948 frame pointer if present, are offset by -2047 which must be added back
3949 when making stack frame references.
3950 Otherwise, assume no such offset is present.
3953 @node Convex Options
3954 @subsection Convex Options
3955 @cindex Convex options
3957 These @samp{-m} options are defined for Convex:
3961 Generate output for C1. The code will run on any Convex machine.
3962 The preprocessor symbol @code{__convex__c1__} is defined.
3965 Generate output for C2. Uses instructions not available on C1.
3966 Scheduling and other optimizations are chosen for max performance on C2.
3967 The preprocessor symbol @code{__convex_c2__} is defined.
3970 Generate output for C32xx. Uses instructions not available on C1.
3971 Scheduling and other optimizations are chosen for max performance on C32.
3972 The preprocessor symbol @code{__convex_c32__} is defined.
3975 Generate output for C34xx. Uses instructions not available on C1.
3976 Scheduling and other optimizations are chosen for max performance on C34.
3977 The preprocessor symbol @code{__convex_c34__} is defined.
3980 Generate output for C38xx. Uses instructions not available on C1.
3981 Scheduling and other optimizations are chosen for max performance on C38.
3982 The preprocessor symbol @code{__convex_c38__} is defined.
3985 Generate code which puts an argument count in the word preceding each
3986 argument list. This is compatible with regular CC, and a few programs
3987 may need the argument count word. GDB and other source-level debuggers
3988 do not need it; this info is in the symbol table.
3991 Omit the argument count word. This is the default.
3993 @item -mvolatile-cache
3994 Allow volatile references to be cached. This is the default.
3996 @item -mvolatile-nocache
3997 Volatile references bypass the data cache, going all the way to memory.
3998 This is only needed for multi-processor code that does not use standard
3999 synchronization instructions. Making non-volatile references to volatile
4000 locations will not necessarily work.
4003 Type long is 32 bits, the same as type int. This is the default.
4006 Type long is 64 bits, the same as type long long. This option is useless,
4007 because no library support exists for it.
4010 @node AMD29K Options
4011 @subsection AMD29K Options
4012 @cindex AMD29K options
4014 These @samp{-m} options are defined for the AMD Am29000:
4019 @cindex DW bit (29k)
4020 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
4021 halfword operations are directly supported by the hardware. This is the
4026 Generate code that assumes the @code{DW} bit is not set.
4030 @cindex byte writes (29k)
4031 Generate code that assumes the system supports byte and halfword write
4032 operations. This is the default.
4036 Generate code that assumes the systems does not support byte and
4037 halfword write operations. @samp{-mnbw} implies @samp{-mndw}.
4041 @cindex memory model (29k)
4042 Use a small memory model that assumes that all function addresses are
4043 either within a single 256 KB segment or at an absolute address of less
4044 than 256k. This allows the @code{call} instruction to be used instead
4045 of a @code{const}, @code{consth}, @code{calli} sequence.
4049 Use the normal memory model: Generate @code{call} instructions only when
4050 calling functions in the same file and @code{calli} instructions
4051 otherwise. This works if each file occupies less than 256 KB but allows
4052 the entire executable to be larger than 256 KB. This is the default.
4055 Always use @code{calli} instructions. Specify this option if you expect
4056 a single file to compile into more than 256 KB of code.
4060 @cindex processor selection (29k)
4061 Generate code for the Am29050.
4065 Generate code for the Am29000. This is the default.
4067 @item -mkernel-registers
4068 @kindex -mkernel-registers
4069 @cindex kernel and user registers (29k)
4070 Generate references to registers @code{gr64-gr95} instead of to
4071 registers @code{gr96-gr127}. This option can be used when compiling
4072 kernel code that wants a set of global registers disjoint from that used
4075 Note that when this option is used, register names in @samp{-f} flags
4076 must use the normal, user-mode, names.
4078 @item -muser-registers
4079 @kindex -muser-registers
4080 Use the normal set of global registers, @code{gr96-gr127}. This is the
4084 @itemx -mno-stack-check
4085 @kindex -mstack-check
4086 @cindex stack checks (29k)
4087 Insert (or do not insert) a call to @code{__msp_check} after each stack
4088 adjustment. This is often used for kernel code.
4091 @itemx -mno-storem-bug
4092 @kindex -mstorem-bug
4093 @cindex storem bug (29k)
4094 @samp{-mstorem-bug} handles 29k processors which cannot handle the
4095 separation of a mtsrim insn and a storem instruction (most 29000 chips
4096 to date, but not the 29050).
4098 @item -mno-reuse-arg-regs
4099 @itemx -mreuse-arg-regs
4100 @kindex -mreuse-arg-regs
4101 @samp{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
4102 registers for copying out arguments. This helps detect calling a function
4103 with fewer arguments than it was declared with.
4105 @item -mno-impure-text
4106 @itemx -mimpure-text
4107 @kindex -mimpure-text
4108 @samp{-mimpure-text}, used in addition to @samp{-shared}, tells the compiler to
4109 not pass @samp{-assert pure-text} to the linker when linking a shared object.
4112 @kindex -msoft-float
4113 Generate output containing library calls for floating point.
4114 @strong{Warning:} the requisite libraries are not part of GNU CC.
4115 Normally the facilities of the machine's usual C compiler are used, but
4116 this can't be done directly in cross-compilation. You must make your
4117 own arrangements to provide suitable library functions for
4122 Do not generate multm or multmu instructions. This is useful for some embedded
4123 systems which do not have trap handlers for these instructions.
4127 @subsection ARM Options
4130 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
4135 @kindex -mapcs-frame
4136 Generate a stack frame that is compliant with the ARM Procedure Call
4137 Standard for all functions, even if this is not strictly necessary for
4138 correct execution of the code. Specifying @samp{-fomit-frame-pointer}
4139 with this option will cause the stack frames not to be generated for
4140 leaf functions. The default is @samp{-mno-apcs-frame}.
4144 This is a synonym for @samp{-mapcs-frame}.
4148 Generate code for a processor running with a 26-bit program counter,
4149 and conforming to the function calling standards for the APCS 26-bit
4150 option. This option replaces the @samp{-m2} and @samp{-m3} options
4151 of previous releases of the compiler.
4155 Generate code for a processor running with a 32-bit program counter,
4156 and conforming to the function calling standards for the APCS 32-bit
4157 option. This option replaces the @samp{-m6} option of previous releases
4160 @item -mapcs-stack-check
4161 @kindex -mapcs-stack-check
4162 @kindex -mno-apcs-stack-check
4163 Generate code to check the amount of stack space available upon entry to
4164 every function (that actually uses some stack space). If there is
4165 insufficient space available then either the function
4166 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
4167 called, depending upon the amount of stack space required. The run time
4168 system is required to provide these functions. The default is
4169 @samp{-mno-apcs-stack-check}, since this produces smaller code.
4172 @kindex -mapcs-float
4173 @kindex -mno-apcs-float
4174 Pass floating point arguments using the float point registers. This is
4175 one of the variants of the APCS. This option is reccommended if the
4176 target hardware has a floating point unit or if a lot of floating point
4177 arithmetic is going to be performed by the code. The default is
4178 @samp{-mno-apcs-float}, since integer only code is slightly increased in
4179 size if @samp{-mapcs-float} is used.
4181 @item -mapcs-reentrant
4182 @kindex -mapcs-reentrant
4183 @kindex -mno-apcs-reentrant
4184 Generate reentrant, position independent code. This is the equivalent
4185 to specifying the @samp{-fpic} option. The default is
4186 @samp{-mno-apcs-reentrant}.
4188 @item -mthumb-interwork
4189 @kindex -mthumb-interwork
4190 @kindex -mno-thumb-interwork
4191 Generate code which supports calling between the ARM and THUMB
4192 instruction sets. Without this option the two instruction sets cannot
4193 be reliably used inside one program. The default is
4194 @samp{-mno-thumb-interwork}, since slightly larger code is generated
4195 when @samp{-mthumb-interwork} is specified.
4197 @item -mno-sched-prolog
4198 @kindex -mno-sched-prolog
4199 @kindex -msched-prolog
4200 Prevent the reordering of instructions in the function prolog, or the
4201 merging of those instruction with the instructions in the function's
4202 body. This means that all functions will start with a recognisable set
4203 of instructions (or in fact one of a chioce from a small set of
4204 different function prologues), and this information can be used to
4205 locate the start if functions inside an executable piece of code. The
4206 default is @samp{-msched-prolog}.
4209 Generate output containing floating point instructions. This is the
4213 Generate output containing library calls for floating point.
4214 @strong{Warning:} the requisite libraries are not available for all ARM
4215 targets. Normally the facilities of the machine's usual C compiler are
4216 used, but this cannot be done directly in cross-compilation. You must make
4217 your own arrangements to provide suitable library functions for
4220 @samp{-msoft-float} changes the calling convention in the output file;
4221 therefore, it is only useful if you compile @emph{all} of a program with
4222 this option. In particular, you need to compile @file{libgcc.a}, the
4223 library that comes with GNU CC, with @samp{-msoft-float} in order for
4226 @item -mlittle-endian
4227 Generate code for a processor running in little-endian mode. This is
4228 the default for all standard configurations.
4231 Generate code for a processor running in big-endian mode; the default is
4232 to compile code for a little-endian processor.
4234 @item -mwords-little-endian
4235 This option only applies when generating code for big-endian processors.
4236 Generate code for a little-endian word order but a big-endian byte
4237 order. That is, a byte order of the form @samp{32107654}. Note: this
4238 option should only be used if you require compatibility with code for
4239 big-endian ARM processors generated by versions of the compiler prior to
4242 @item -mshort-load-bytes
4243 @kindex -mshort-load-bytes
4244 Do not try to load half-words (eg @samp{short}s) by loading a word from
4245 an unaligned address. For some targets the MMU is configured to trap
4246 unaligned loads; use this option to generate code that is safe in these
4249 @item -mno-short-load-bytes
4250 @kindex -mno-short-load-bytes
4251 Use unaligned word loads to load half-words (eg @samp{short}s). This
4252 option produces more efficient code, but the MMU is sometimes configured
4253 to trap these instructions.
4255 @item -mshort-load-words
4256 @kindex -mshort-load-words
4257 This is a synonym for the @samp{-mno-short-load-bytes}.
4259 @item -mno-short-load-words
4260 @kindex -mno-short-load-words
4261 This is a synonym for the @samp{-mshort-load-bytes}.
4265 This option only applies to RISC iX. Emulate the native BSD-mode
4266 compiler. This is the default if @samp{-ansi} is not specified.
4270 This option only applies to RISC iX. Emulate the native X/Open-mode
4273 @item -mno-symrename
4274 @kindex -mno-symrename
4275 This option only applies to RISC iX. Do not run the assembler
4276 post-processor, @samp{symrename}, after code has been assembled.
4277 Normally it is necessary to modify some of the standard symbols in
4278 preparation for linking with the RISC iX C library; this option
4279 suppresses this pass. The post-processor is never run when the
4280 compiler is built for cross-compilation.
4284 This specifies the name of the target ARM processor. GCC uses this name
4285 to determine what kind of instructions it can use when generating
4286 assembly code. Permissable names are: arm2, arm250, arm3, arm6, arm60,
4287 arm600, arm610, arm620, arm7, arm7m, arm7d, arm7dm, arm7di, arm7dmi,
4288 arm70, arm700, arm700i, arm710, arm710c, arm7100, arm7500, arm7500fe,
4289 arm7tdmi, arm8, strongarm, strongarm110, strongarm1100, arm8, arm810,
4290 arm9, arm920, arm920t, arm9tdmi.
4292 @itemx -mtune=<name>
4294 This option is very similar to the @samp{-mcpu=} option, except that
4295 instead of specifying the actual target processor type, and hence
4296 restricting which instructions can be used, it specifies that GCC should
4297 tune the performance of the code as if the target were of the type
4298 specified in this option, but still choosing the instructions that it
4299 will generate based on the cpu specified by a @samp{-mcpu=} option.
4300 For some arm implementations better performance can be obtained by using
4305 This specifies the name of the target ARM architecture. GCC uses this
4306 name to determine what kind of instructions it can use when generating
4307 assembly code. This option can be used in conjunction with or instead
4308 of the @samp{-mcpu=} option. Permissable names are: armv2, armv2a,
4309 armv3, armv3m, armv4, armv4t, armv5.
4311 @item -mfpe=<number>
4312 @itemx -mfp=<number>
4315 This specifes the version of the floating point emulation available on
4316 the target. Permissable values are 2 and 3. @samp{-mfp=} is a synonym
4317 for @samp{-mfpe=} to support older versions of GCC.
4319 @item -mstructure-size-boundary=<n>
4320 @kindex -mstructure-size-boundary
4321 The size of all structures and unions will be rounded up to a multiple
4322 of the number of bits set by this option. Permissable values are 8 and
4323 32. The default value varies for different toolchains. For the COFF
4324 targeted toolchain the default value is 8. Specifying the larger number
4325 can produced faster, more efficient code, but can also increase the size
4326 of the program. The two values are potentially incompatible. Code
4327 compiled with one value cannot necessarily expect to work with code or
4328 libraries compiled with the other value, if they exchange information
4329 using structures or unions. Programmers are encouraged to use the 32
4330 value as future versions of the toolchain may default to this value.
4332 @item -mabort-on-noreturn
4333 @kindex -mabort-on-noreturn
4334 @kindex -mnoabort-on-noreturn
4335 Generate a call to the function abort at the end of a noreturn function.
4336 It will be executed if the function tries to return.
4338 @item -mnop-fun-dllimport
4339 @kindex -mnop-fun-dllimport
4340 Disable the support for the @emph{dllimport} attribute.
4345 @subsection Thumb Options
4346 @cindex Thumb Options
4350 @item -mthumb-interwork
4351 @kindex -mthumb-interwork
4352 @kindex -mno-thumb-interwork
4353 Generate code which supports calling between the THUMB and ARM
4354 instruction sets. Without this option the two instruction sets cannot
4355 be reliably used inside one program. The default is
4356 @samp{-mno-thumb-interwork}, since slightly smaller code is generated
4360 @kindex -mtpcs-frame
4361 @kindex -mno-tpcs-frame
4362 Generate a stack frame that is compliant with the Thumb Procedure Call
4363 Standard for all non-leaf functions. (A leaf function is one that does
4364 not call any other functions). The default is @samp{-mno-apcs-frame}.
4366 @item -mtpcs-leaf-frame
4367 @kindex -mtpcs-leaf-frame
4368 @kindex -mno-tpcs-leaf-frame
4369 Generate a stack frame that is compliant with the Thumb Procedure Call
4370 Standard for all leaf functions. (A leaf function is one that does
4371 not call any other functions). The default is @samp{-mno-apcs-leaf-frame}.
4373 @item -mlittle-endian
4374 @kindex -mlittle-endian
4375 Generate code for a processor running in little-endian mode. This is
4376 the default for all standard configurations.
4379 @kindex -mbig-endian
4380 Generate code for a processor running in big-endian mode.
4382 @item -mstructure-size-boundary=<n>
4383 @kindex -mstructure-size-boundary
4384 The size of all structures and unions will be rounded up to a multiple
4385 of the number of bits set by this option. Permissable values are 8 and
4386 32. The default value varies for different toolchains. For the COFF
4387 targeted toolchain the default value is 8. Specifying the larger number
4388 can produced faster, more efficient code, but can also increase the size
4389 of the program. The two values are potentially incompatible. Code
4390 compiled with one value cannot necessarily expect to work with code or
4391 libraries compiled with the other value, if they exchange information
4392 using structures or unions. Programmers are encouraged to use the 32
4393 value as future versions of the toolchain may default to this value.
4395 @item -mnop-fun-dllimport
4396 @kindex -mnop-fun-dllimport
4397 Disable the support for the @emph{dllimport} attribute.
4399 @item -mcallee-super-interworking
4400 @kindex -mcallee-super-interworking
4401 Gives all externally visible functions in the file being compiled an ARM
4402 instruction set header which switches to Thumb mode before executing the
4403 rest of the function. This allows these functions to be called from
4404 non-interworking code.
4406 @item -mcaller-super-interworking
4407 @kindex -mcaller-super-interworking
4408 Allows calls via function pointers (including virtual functions) to
4409 execute correctly regardless of whether the target code has been
4410 compiled for interworking or not. There is a small overhead in the cost
4411 of executing a funciton pointer if this option is enabled.
4415 @node MN10200 Options
4416 @subsection MN10200 Options
4417 @cindex MN10200 options
4418 These @samp{-m} options are defined for Matsushita MN10200 architectures:
4422 Indicate to the linker that it should perform a relaxation optimization pass
4423 to shorten branches, calls and absolute memory addresses. This option only
4424 has an effect when used on the command line for the final link step.
4426 This option makes symbolic debugging impossible.
4429 @node MN10300 Options
4430 @subsection MN10300 Options
4431 @cindex MN10300 options
4432 These @samp{-m} options are defined for Matsushita MN10300 architectures:
4436 Generate code to avoid bugs in the multiply instructions for the MN10300
4437 processors. This is the default.
4440 Do not generate code to avoid bugs in the multiply instructions for the
4444 Indicate to the linker that it should perform a relaxation optimization pass
4445 to shorten branches, calls and absolute memory addresses. This option only
4446 has an effect when used on the command line for the final link step.
4448 This option makes symbolic debugging impossible.
4452 @node M32R/D Options
4453 @subsection M32R/D Options
4454 @cindex M32R/D options
4456 These @samp{-m} options are defined for Mitsubishi M32R/D architectures:
4459 @item -mcode-model=small
4460 Assume all objects live in the lower 16MB of memory (so that their addresses
4461 can be loaded with the @code{ld24} instruction), and assume all subroutines
4462 are reachable with the @code{bl} instruction.
4463 This is the default.
4465 The addressability of a particular object can be set with the
4466 @code{model} attribute.
4468 @item -mcode-model=medium
4469 Assume objects may be anywhere in the 32 bit address space (the compiler
4470 will generate @code{seth/add3} instructions to load their addresses), and
4471 assume all subroutines are reachable with the @code{bl} instruction.
4473 @item -mcode-model=large
4474 Assume objects may be anywhere in the 32 bit address space (the compiler
4475 will generate @code{seth/add3} instructions to load their addresses), and
4476 assume subroutines may not be reachable with the @code{bl} instruction
4477 (the compiler will generate the much slower @code{seth/add3/jl}
4478 instruction sequence).
4481 Disable use of the small data area. Variables will be put into
4482 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
4483 @code{section} attribute has been specified).
4484 This is the default.
4486 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
4487 Objects may be explicitly put in the small data area with the
4488 @code{section} attribute using one of these sections.
4491 Put small global and static data in the small data area, but do not
4492 generate special code to reference them.
4495 Put small global and static data in the small data area, and generate
4496 special instructions to reference them.
4499 @cindex smaller data references
4500 Put global and static objects less than or equal to @var{num} bytes
4501 into the small data or bss sections instead of the normal data or bss
4502 sections. The default value of @var{num} is 8.
4503 The @samp{-msdata} option must be set to one of @samp{sdata} or @samp{use}
4504 for this option to have any effect.
4506 All modules should be compiled with the same @samp{-G @var{num}} value.
4507 Compiling with different values of @var{num} may or may not work; if it
4508 doesn't the linker will give an error message - incorrect code will not be
4514 @subsection M88K Options
4515 @cindex M88k options
4517 These @samp{-m} options are defined for Motorola 88k architectures:
4522 Generate code that works well on both the m88100 and the
4527 Generate code that works best for the m88100, but that also
4532 Generate code that works best for the m88110, and may not run
4537 Obsolete option to be removed from the next revision.
4540 @item -midentify-revision
4541 @kindex -midentify-revision
4543 @cindex identifying source, compiler (88k)
4544 Include an @code{ident} directive in the assembler output recording the
4545 source file name, compiler name and version, timestamp, and compilation
4548 @item -mno-underscores
4549 @kindex -mno-underscores
4550 @cindex underscores, avoiding (88k)
4551 In assembler output, emit symbol names without adding an underscore
4552 character at the beginning of each name. The default is to use an
4553 underscore as prefix on each name.
4555 @item -mocs-debug-info
4556 @itemx -mno-ocs-debug-info
4557 @kindex -mocs-debug-info
4558 @kindex -mno-ocs-debug-info
4560 @cindex debugging, 88k OCS
4561 Include (or omit) additional debugging information (about registers used
4562 in each stack frame) as specified in the 88open Object Compatibility
4563 Standard, ``OCS''. This extra information allows debugging of code that
4564 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
4565 Delta 88 SVr3.2 is to include this information; other 88k configurations
4566 omit this information by default.
4568 @item -mocs-frame-position
4569 @kindex -mocs-frame-position
4570 @cindex register positions in frame (88k)
4571 When emitting COFF debugging information for automatic variables and
4572 parameters stored on the stack, use the offset from the canonical frame
4573 address, which is the stack pointer (register 31) on entry to the
4574 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
4575 @samp{-mocs-frame-position}; other 88k configurations have the default
4576 @samp{-mno-ocs-frame-position}.
4578 @item -mno-ocs-frame-position
4579 @kindex -mno-ocs-frame-position
4580 @cindex register positions in frame (88k)
4581 When emitting COFF debugging information for automatic variables and
4582 parameters stored on the stack, use the offset from the frame pointer
4583 register (register 30). When this option is in effect, the frame
4584 pointer is not eliminated when debugging information is selected by the
4587 @item -moptimize-arg-area
4588 @itemx -mno-optimize-arg-area
4589 @kindex -moptimize-arg-area
4590 @kindex -mno-optimize-arg-area
4591 @cindex arguments in frame (88k)
4592 Control how function arguments are stored in stack frames.
4593 @samp{-moptimize-arg-area} saves space by optimizing them, but this
4594 conflicts with the 88open specifications. The opposite alternative,
4595 @samp{-mno-optimize-arg-area}, agrees with 88open standards. By default
4596 GNU CC does not optimize the argument area.
4598 @item -mshort-data-@var{num}
4599 @kindex -mshort-data-@var{num}
4600 @cindex smaller data references (88k)
4601 @cindex r0-relative references (88k)
4602 Generate smaller data references by making them relative to @code{r0},
4603 which allows loading a value using a single instruction (rather than the
4604 usual two). You control which data references are affected by
4605 specifying @var{num} with this option. For example, if you specify
4606 @samp{-mshort-data-512}, then the data references affected are those
4607 involving displacements of less than 512 bytes.
4608 @samp{-mshort-data-@var{num}} is not effective for @var{num} greater
4611 @item -mserialize-volatile
4612 @kindex -mserialize-volatile
4613 @itemx -mno-serialize-volatile
4614 @kindex -mno-serialize-volatile
4615 @cindex sequential consistency on 88k
4616 Do, or don't, generate code to guarantee sequential consistency
4617 of volatile memory references. By default, consistency is
4620 The order of memory references made by the MC88110 processor does
4621 not always match the order of the instructions requesting those
4622 references. In particular, a load instruction may execute before
4623 a preceding store instruction. Such reordering violates
4624 sequential consistency of volatile memory references, when there
4625 are multiple processors. When consistency must be guaranteed,
4626 GNU C generates special instructions, as needed, to force
4627 execution in the proper order.
4629 The MC88100 processor does not reorder memory references and so
4630 always provides sequential consistency. However, by default, GNU
4631 C generates the special instructions to guarantee consistency
4632 even when you use @samp{-m88100}, so that the code may be run on an
4633 MC88110 processor. If you intend to run your code only on the
4634 MC88100 processor, you may use @samp{-mno-serialize-volatile}.
4636 The extra code generated to guarantee consistency may affect the
4637 performance of your application. If you know that you can safely
4638 forgo this guarantee, you may use @samp{-mno-serialize-volatile}.
4644 @cindex assembler syntax, 88k
4646 Turn on (@samp{-msvr4}) or off (@samp{-msvr3}) compiler extensions
4647 related to System V release 4 (SVr4). This controls the following:
4651 Which variant of the assembler syntax to emit.
4653 @samp{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
4654 that is used on System V release 4.
4656 @samp{-msvr4} makes GNU CC issue additional declaration directives used in
4660 @samp{-msvr4} is the default for the m88k-motorola-sysv4 and
4661 m88k-dg-dgux m88k configurations. @samp{-msvr3} is the default for all
4662 other m88k configurations.
4664 @item -mversion-03.00
4665 @kindex -mversion-03.00
4666 This option is obsolete, and is ignored.
4667 @c ??? which asm syntax better for GAS? option there too?
4669 @item -mno-check-zero-division
4670 @itemx -mcheck-zero-division
4671 @kindex -mno-check-zero-division
4672 @kindex -mcheck-zero-division
4673 @cindex zero division on 88k
4674 Do, or don't, generate code to guarantee that integer division by
4675 zero will be detected. By default, detection is guaranteed.
4677 Some models of the MC88100 processor fail to trap upon integer
4678 division by zero under certain conditions. By default, when
4679 compiling code that might be run on such a processor, GNU C
4680 generates code that explicitly checks for zero-valued divisors
4681 and traps with exception number 503 when one is detected. Use of
4682 mno-check-zero-division suppresses such checking for code
4683 generated to run on an MC88100 processor.
4685 GNU C assumes that the MC88110 processor correctly detects all
4686 instances of integer division by zero. When @samp{-m88110} is
4687 specified, both @samp{-mcheck-zero-division} and
4688 @samp{-mno-check-zero-division} are ignored, and no explicit checks for
4689 zero-valued divisors are generated.
4691 @item -muse-div-instruction
4692 @kindex -muse-div-instruction
4693 @cindex divide instruction, 88k
4694 Use the div instruction for signed integer division on the
4695 MC88100 processor. By default, the div instruction is not used.
4697 On the MC88100 processor the signed integer division instruction
4698 div) traps to the operating system on a negative operand. The
4699 operating system transparently completes the operation, but at a
4700 large cost in execution time. By default, when compiling code
4701 that might be run on an MC88100 processor, GNU C emulates signed
4702 integer division using the unsigned integer division instruction
4703 divu), thereby avoiding the large penalty of a trap to the
4704 operating system. Such emulation has its own, smaller, execution
4705 cost in both time and space. To the extent that your code's
4706 important signed integer division operations are performed on two
4707 nonnegative operands, it may be desirable to use the div
4708 instruction directly.
4710 On the MC88110 processor the div instruction (also known as the
4711 divs instruction) processes negative operands without trapping to
4712 the operating system. When @samp{-m88110} is specified,
4713 @samp{-muse-div-instruction} is ignored, and the div instruction is used
4714 for signed integer division.
4716 Note that the result of dividing INT_MIN by -1 is undefined. In
4717 particular, the behavior of such a division with and without
4718 @samp{-muse-div-instruction} may differ.
4720 @item -mtrap-large-shift
4721 @itemx -mhandle-large-shift
4722 @kindex -mtrap-large-shift
4723 @kindex -mhandle-large-shift
4724 @cindex bit shift overflow (88k)
4725 @cindex large bit shifts (88k)
4726 Include code to detect bit-shifts of more than 31 bits; respectively,
4727 trap such shifts or emit code to handle them properly. By default GNU CC
4728 makes no special provision for large bit shifts.
4730 @item -mwarn-passed-structs
4731 @kindex -mwarn-passed-structs
4732 @cindex structure passing (88k)
4733 Warn when a function passes a struct as an argument or result.
4734 Structure-passing conventions have changed during the evolution of the C
4735 language, and are often the source of portability problems. By default,
4736 GNU CC issues no such warning.
4739 @node RS/6000 and PowerPC Options
4740 @subsection IBM RS/6000 and PowerPC Options
4741 @cindex RS/6000 and PowerPC Options
4742 @cindex IBM RS/6000 and PowerPC Options
4744 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
4752 @itemx -mpowerpc-gpopt
4753 @itemx -mno-powerpc-gpopt
4754 @itemx -mpowerpc-gfxopt
4755 @itemx -mno-powerpc-gfxopt
4757 @itemx -mno-powerpc64
4761 @kindex -mpowerpc-gpopt
4762 @kindex -mpowerpc-gfxopt
4764 GNU CC supports two related instruction set architectures for the
4765 RS/6000 and PowerPC. The @dfn{POWER} instruction set are those
4766 instructions supported by the @samp{rios} chip set used in the original
4767 RS/6000 systems and the @dfn{PowerPC} instruction set is the
4768 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
4769 the IBM 4xx microprocessors.
4771 Neither architecture is a subset of the other. However there is a
4772 large common subset of instructions supported by both. An MQ
4773 register is included in processors supporting the POWER architecture.
4775 You use these options to specify which instructions are available on the
4776 processor you are using. The default value of these options is
4777 determined when configuring GNU CC. Specifying the
4778 @samp{-mcpu=@var{cpu_type}} overrides the specification of these
4779 options. We recommend you use the @samp{-mcpu=@var{cpu_type}} option
4780 rather than the options listed above.
4782 The @samp{-mpower} option allows GNU CC to generate instructions that
4783 are found only in the POWER architecture and to use the MQ register.
4784 Specifying @samp{-mpower2} implies @samp{-power} and also allows GNU CC
4785 to generate instructions that are present in the POWER2 architecture but
4786 not the original POWER architecture.
4788 The @samp{-mpowerpc} option allows GNU CC to generate instructions that
4789 are found only in the 32-bit subset of the PowerPC architecture.
4790 Specifying @samp{-mpowerpc-gpopt} implies @samp{-mpowerpc} and also allows
4791 GNU CC to use the optional PowerPC architecture instructions in the
4792 General Purpose group, including floating-point square root. Specifying
4793 @samp{-mpowerpc-gfxopt} implies @samp{-mpowerpc} and also allows GNU CC to
4794 use the optional PowerPC architecture instructions in the Graphics
4795 group, including floating-point select.
4797 The @samp{-mpowerpc64} option allows GNU CC to generate the additional
4798 64-bit instructions that are found in the full PowerPC64 architecture
4799 and to treat GPRs as 64-bit, doubleword quantities. GNU CC defaults to
4800 @samp{-mno-powerpc64}.
4802 If you specify both @samp{-mno-power} and @samp{-mno-powerpc}, GNU CC
4803 will use only the instructions in the common subset of both
4804 architectures plus some special AIX common-mode calls, and will not use
4805 the MQ register. Specifying both @samp{-mpower} and @samp{-mpowerpc}
4806 permits GNU CC to use any instruction from either architecture and to
4807 allow use of the MQ register; specify this for the Motorola MPC601.
4809 @item -mnew-mnemonics
4810 @itemx -mold-mnemonics
4811 @kindex -mnew-mnemonics
4812 @kindex -mold-mnemonics
4813 Select which mnemonics to use in the generated assembler code.
4814 @samp{-mnew-mnemonics} requests output that uses the assembler mnemonics
4815 defined for the PowerPC architecture, while @samp{-mold-mnemonics}
4816 requests the assembler mnemonics defined for the POWER architecture.
4817 Instructions defined in only one architecture have only one mnemonic;
4818 GNU CC uses that mnemonic irrespective of which of these options is
4821 GNU CC defaults to the mnemonics appropriate for the architecture in
4822 use. Specifying @samp{-mcpu=@var{cpu_type}} sometimes overrides the
4823 value of these option. Unless you are building a cross-compiler, you
4824 should normally not specify either @samp{-mnew-mnemonics} or
4825 @samp{-mold-mnemonics}, but should instead accept the default.
4827 @item -mcpu=@var{cpu_type}
4829 Set architecture type, register usage, choice of mnemonics, and
4830 instruction scheduling parameters for machine type @var{cpu_type}.
4831 Supported values for @var{cpu_type} are @samp{rs6000}, @samp{rios1},
4832 @samp{rios2}, @samp{rsc}, @samp{601}, @samp{602}, @samp{603},
4833 @samp{603e}, @samp{604}, @samp{604e}, @samp{620}, @samp{740},
4834 @samp{750}, @samp{power}, @samp{power2}, @samp{powerpc}, @samp{403},
4835 @samp{505}, @samp{801}, @samp{821}, @samp{823}, and @samp{860} and
4836 @samp{common}. @samp{-mcpu=power}, @samp{-mcpu=power2}, and
4837 @samp{-mcpu=powerpc} specify generic POWER, POWER2 and pure PowerPC
4838 (i.e., not MPC601) architecture machine types, with an appropriate,
4839 generic processor model assumed for scheduling purposes.@refill
4841 @c overfull hbox here --bob 22 jul96
4842 @c original text between ignore ... end ignore
4844 Specifying any of the @samp{-mcpu=rios1}, @samp{-mcpu=rios2},
4845 @samp{-mcpu=rsc}, @samp{-mcpu=power}, or @samp{-mcpu=power2} options
4846 enables the @samp{-mpower} option and disables the @samp{-mpowerpc}
4847 option; @samp{-mcpu=601} enables both the @samp{-mpower} and
4848 @samp{-mpowerpc} options; all of @samp{-mcpu=602}, @samp{-mcpu=603},
4849 @samp{-mcpu=603e}, @samp{-mcpu=604}, @samp{-mcpu=604e},
4850 @samp{-mcpu=620}, @samp{-mcpu=403}, @samp{-mcpu=505}, @samp{-mcpu=801},
4851 @samp{-mcpu=821}, @samp{-mcpu=823}, @samp{-mcpu=860} and
4852 @samp{-mcpu=powerpc} enable the @samp{-mpowerpc} option and disable the
4853 @samp{-mpower} option; @samp{-mcpu=common} disables both the
4854 @samp{-mpower} and @samp{-mpowerpc} options.@refill
4856 @c changed paragraph
4857 Specifying any of the following options:
4858 @samp{-mcpu=rios1}, @samp{-mcpu=rios2}, @samp{-mcpu=rsc},
4859 @samp{-mcpu=power}, or @samp{-mcpu=power2}
4860 enables the @samp{-mpower} option and disables the @samp{-mpowerpc} option;
4861 @samp{-mcpu=601} enables both the @samp{-mpower} and @samp{-mpowerpc} options.
4862 All of @samp{-mcpu=602}, @samp{-mcpu=603}, @samp{-mcpu=603e},
4863 @samp{-mcpu=604}, @samp{-mcpu=620},
4864 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
4865 Exactly similarly, all of @samp{-mcpu=403},
4866 @samp{-mcpu=505}, @samp{-mcpu=821}, @samp{-mcpu=860} and @samp{-mcpu=powerpc}
4867 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
4868 @samp{-mcpu=common} disables both the
4869 @samp{-mpower} and @samp{-mpowerpc} options.@refill
4870 @c end changes to prevent overfull hboxes
4872 AIX versions 4 or greater selects @samp{-mcpu=common} by default, so
4873 that code will operate on all members of the RS/6000 and PowerPC
4874 families. In that case, GNU CC will use only the instructions in the
4875 common subset of both architectures plus some special AIX common-mode
4876 calls, and will not use the MQ register. GNU CC assumes a generic
4877 processor model for scheduling purposes.
4879 Specifying any of the options @samp{-mcpu=rios1}, @samp{-mcpu=rios2},
4880 @samp{-mcpu=rsc}, @samp{-mcpu=power}, or @samp{-mcpu=power2} also
4881 disables the @samp{new-mnemonics} option. Specifying @samp{-mcpu=601},
4882 @samp{-mcpu=602}, @samp{-mcpu=603}, @samp{-mcpu=603e}, @samp{-mcpu=604},
4883 @samp{620}, @samp{403}, or @samp{-mcpu=powerpc} also enables the
4884 @samp{new-mnemonics} option.@refill
4886 Specifying @samp{-mcpu=403}, @samp{-mcpu=821}, or @samp{-mcpu=860} also
4887 enables the @samp{-msoft-float} option.
4889 @item -mtune=@var{cpu_type}
4890 Set the instruction scheduling parameters for machine type
4891 @var{cpu_type}, but do not set the architecture type, register usage,
4892 choice of mnemonics like @samp{-mcpu=}@var{cpu_type} would. The same
4893 values for @var{cpu_type} are used for @samp{-mtune=}@var{cpu_type} as
4894 for @samp{-mcpu=}@var{cpu_type}. The @samp{-mtune=}@var{cpu_type}
4895 option overrides the @samp{-mcpu=}@var{cpu_type} option in terms of
4896 instruction scheduling parameters.
4899 @itemx -mno-fp-in-toc
4900 @itemx -mno-sum-in-toc
4901 @itemx -mminimal-toc
4902 @kindex -mminimal-toc
4903 Modify generation of the TOC (Table Of Contents), which is created for
4904 every executable file. The @samp{-mfull-toc} option is selected by
4905 default. In that case, GNU CC will allocate at least one TOC entry for
4906 each unique non-automatic variable reference in your program. GNU CC
4907 will also place floating-point constants in the TOC. However, only
4908 16,384 entries are available in the TOC.
4910 If you receive a linker error message that saying you have overflowed
4911 the available TOC space, you can reduce the amount of TOC space used
4912 with the @samp{-mno-fp-in-toc} and @samp{-mno-sum-in-toc} options.
4913 @samp{-mno-fp-in-toc} prevents GNU CC from putting floating-point
4914 constants in the TOC and @samp{-mno-sum-in-toc} forces GNU CC to
4915 generate code to calculate the sum of an address and a constant at
4916 run-time instead of putting that sum into the TOC. You may specify one
4917 or both of these options. Each causes GNU CC to produce very slightly
4918 slower and larger code at the expense of conserving TOC space.
4920 If you still run out of space in the TOC even when you specify both of
4921 these options, specify @samp{-mminimal-toc} instead. This option causes
4922 GNU CC to make only one TOC entry for every file. When you specify this
4923 option, GNU CC will produce code that is slower and larger but which
4924 uses extremely little TOC space. You may wish to use this option
4925 only on files that contain less frequently executed code. @refill
4931 Enable AIX 64-bit ABI and calling convention: 64-bit pointers, 64-bit
4932 @code{long} type, and the infrastructure needed to support them.
4933 Specifying @samp{-maix64} implies @samp{-mpowerpc64} and
4934 @samp{-mpowerpc}, while @samp{-maix32} disables the 64-bit ABI and
4935 implies @samp{-mno-powerpc64}. GNU CC defaults to @samp{-maix32}.
4940 On AIX, pass floating-point arguments to prototyped functions beyond the
4941 register save area (RSA) on the stack in addition to argument FPRs. The
4942 AIX calling convention was extended but not initially documented to
4943 handle an obscure K&R C case of calling a function that takes the
4944 address of its arguments with fewer arguments than declared. AIX XL
4945 compilers access floating point arguments which do not fit in the
4946 RSA from the stack when a subroutine is compiled without
4947 optimization. Because always storing floating-point arguments on the
4948 stack is inefficient and rarely needed, this option is not enabled by
4949 default and only is necessary when calling subroutines compiled by AIX
4950 XL compilers without optimization.
4954 Support @dfn{AIX Threads}. Link an application written to use
4955 @dfn{pthreads} with special libraries and startup code to enable the
4960 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE). Link an
4961 application written to use message passing with special startup code to
4962 enable the application to run. The system must have PE installed in the
4963 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
4964 must be overridden with the @samp{-specs=} option to specify the
4965 appropriate directory location. The Parallel Environment does not
4966 support threads, so the @samp{-mpe} option and the @samp{-mthreads}
4967 option are incompatible.
4971 @kindex -msoft-float
4972 Generate code that does not use (uses) the floating-point register set.
4973 Software floating point emulation is provided if you use the
4974 @samp{-msoft-float} option, and pass the option to GNU CC when linking.
4977 @itemx -mno-multiple
4978 Generate code that uses (does not use) the load multiple word
4979 instructions and the store multiple word instructions. These
4980 instructions are generated by default on POWER systems, and not
4981 generated on PowerPC systems. Do not use @samp{-mmultiple} on little
4982 endian PowerPC systems, since those instructions do not work when the
4983 processor is in little endian mode. The exceptions are PPC740 and
4984 PPC750 which permit the instructions usage in little endian mode.
4989 Generate code that uses (does not use) the load string instructions
4990 and the store string word instructions to save multiple registers and
4991 do small block moves. These instructions are generated by default on
4992 POWER systems, and not generated on PowerPC systems. Do not use
4993 @samp{-mstring} on little endian PowerPC systems, since those
4994 instructions do not work when the processor is in little endian mode.
4995 The exceptions are PPC740 and PPC750 which permit the instructions
4996 usage in little endian mode.
5001 Generate code that uses (does not use) the load or store instructions
5002 that update the base register to the address of the calculated memory
5003 location. These instructions are generated by default. If you use
5004 @samp{-mno-update}, there is a small window between the time that the
5005 stack pointer is updated and the address of the previous frame is
5006 stored, which means code that walks the stack frame across interrupts or
5007 signals may get corrupted data.
5010 @itemx -mno-fused-madd
5011 @kindex -mfused-madd
5012 Generate code that uses (does not use) the floating point multiply and
5013 accumulate instructions. These instructions are generated by default if
5014 hardware floating is used.
5016 @item -mno-bit-align
5019 On System V.4 and embedded PowerPC systems do not (do) force structures
5020 and unions that contain bit fields to be aligned to the base type of the
5023 For example, by default a structure containing nothing but 8
5024 @code{unsigned} bitfields of length 1 would be aligned to a 4 byte
5025 boundary and have a size of 4 bytes. By using @samp{-mno-bit-align},
5026 the structure would be aligned to a 1 byte boundary and be one byte in
5029 @item -mno-strict-align
5030 @itemx -mstrict-align
5031 @kindex -mstrict-align
5032 On System V.4 and embedded PowerPC systems do not (do) assume that
5033 unaligned memory references will be handled by the system.
5036 @itemx -mno-relocatable
5037 @kindex -mrelocatable
5038 On embedded PowerPC systems generate code that allows (does not allow)
5039 the program to be relocated to a different address at runtime. If you
5040 use @samp{-mrelocatable} on any module, all objects linked together must
5041 be compiled with @samp{-mrelocatable} or @samp{-mrelocatable-lib}.
5043 @item -mrelocatable-lib
5044 @itemx -mno-relocatable-lib
5045 On embedded PowerPC systems generate code that allows (does not allow)
5046 the program to be relocated to a different address at runtime. Modules
5047 compiled with @samp{-mrelocatable-lib} can be linked with either modules
5048 compiled without @samp{-mrelocatable} and @samp{-mrelocatable-lib} or
5049 with modules compiled with the @samp{-mrelocatable} options.
5053 On System V.4 and embedded PowerPC systems do not (do) assume that
5054 register 2 contains a pointer to a global area pointing to the addresses
5055 used in the program.
5058 @itemx -mlittle-endian
5059 On System V.4 and embedded PowerPC systems compile code for the
5060 processor in little endian mode. The @samp{-mlittle-endian} option is
5061 the same as @samp{-mlittle}.
5065 On System V.4 and embedded PowerPC systems compile code for the
5066 processor in big endian mode. The @samp{-mbig-endian} option is
5067 the same as @samp{-mbig}.
5070 On System V.4 and embedded PowerPC systems compile code using calling
5071 conventions that adheres to the March 1995 draft of the System V
5072 Application Binary Interface, PowerPC processor supplement. This is the
5073 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
5075 @item -mcall-sysv-eabi
5076 Specify both @samp{-mcall-sysv} and @samp{-meabi} options.
5078 @item -mcall-sysv-noeabi
5079 Specify both @samp{-mcall-sysv} and @samp{-mno-eabi} options.
5082 On System V.4 and embedded PowerPC systems compile code using calling
5083 conventions that are similar to those used on AIX. This is the
5084 default if you configured GCC using @samp{powerpc-*-eabiaix}.
5086 @item -mcall-solaris
5087 On System V.4 and embedded PowerPC systems compile code for the Solaris
5091 On System V.4 and embedded PowerPC systems compile code for the
5092 Linux-based GNU system.
5095 @itemx -mno-prototype
5096 On System V.4 and embedded PowerPC systems assume that all calls to
5097 variable argument functions are properly prototyped. Otherwise, the
5098 compiler must insert an instruction before every non prototyped call to
5099 set or clear bit 6 of the condition code register (@var{CR}) to
5100 indicate whether floating point values were passed in the floating point
5101 registers in case the function takes a variable arguments. With
5102 @samp{-mprototype}, only calls to prototyped variable argument functions
5103 will set or clear the bit.
5106 On embedded PowerPC systems, assume that the startup module is called
5107 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
5108 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
5112 On embedded PowerPC systems, assume that the startup module is called
5113 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
5117 On embedded PowerPC systems, assume that the startup module is called
5118 @file{crt0.o} and the standard C libraries are @file{libads.a} and
5122 On embedded PowerPC systems, assume that the startup module is called
5123 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
5127 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
5128 header to indicate that @samp{eabi} extended relocations are used.
5132 On System V.4 and embedded PowerPC systems do (do not) adhere to the
5133 Embedded Applications Binary Interface (eabi) which is a set of
5134 modifications to the System V.4 specifications. Selecting @code{-meabi}
5135 means that the stack is aligned to an 8 byte boundary, a function
5136 @code{__eabi} is called to from @code{main} to set up the eabi
5137 environment, and the @samp{-msdata} option can use both @code{r2} and
5138 @code{r13} to point to two separate small data areas. Selecting
5139 @code{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
5140 do not call an initialization function from @code{main}, and the
5141 @samp{-msdata} option will only use @code{r13} to point to a single
5142 small data area. The @samp{-meabi} option is on by default if you
5143 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
5146 On System V.4 and embedded PowerPC systems, put small initialized
5147 @code{const} global and static data in the @samp{.sdata2} section, which
5148 is pointed to by register @code{r2}. Put small initialized
5149 non-@code{const} global and static data in the @samp{.sdata} section,
5150 which is pointed to by register @code{r13}. Put small uninitialized
5151 global and static data in the @samp{.sbss} section, which is adjacent to
5152 the @samp{.sdata} section. The @samp{-msdata=eabi} option is
5153 incompatible with the @samp{-mrelocatable} option. The
5154 @samp{-msdata=eabi} option also sets the @samp{-memb} option.
5157 On System V.4 and embedded PowerPC systems, put small global and static
5158 data in the @samp{.sdata} section, which is pointed to by register
5159 @code{r13}. Put small uninitialized global and static data in the
5160 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
5161 The @samp{-msdata=sysv} option is incompatible with the
5162 @samp{-mrelocatable} option.
5164 @item -msdata=default
5166 On System V.4 and embedded PowerPC systems, if @samp{-meabi} is used,
5167 compile code the same as @samp{-msdata=eabi}, otherwise compile code the
5168 same as @samp{-msdata=sysv}.
5171 On System V.4 and embedded PowerPC systems, put small global and static
5172 data in the @samp{.sdata} section. Put small uninitialized global and
5173 static data in the @samp{.sbss} section. Do not use register @code{r13}
5174 to address small data however. This is the default behavior unless
5175 other @samp{-msdata} options are used.
5179 On embedded PowerPC systems, put all initialized global and static data
5180 in the @samp{.data} section, and all uninitialized data in the
5181 @samp{.bss} section.
5184 @cindex smaller data references (PowerPC)
5185 @cindex .sdata/.sdata2 references (PowerPC)
5186 On embedded PowerPC systems, put global and static items less than or
5187 equal to @var{num} bytes into the small data or bss sections instead of
5188 the normal data or bss section. By default, @var{num} is 8. The
5189 @samp{-G @var{num}} switch is also passed to the linker.
5190 All modules should be compiled with the same @samp{-G @var{num}} value.
5193 @itemx -mno-regnames
5194 On System V.4 and embedded PowerPC systems do (do not) emit register
5195 names in the assembly language output using symbolic forms.
5200 @subsection IBM RT Options
5202 @cindex IBM RT options
5204 These @samp{-m} options are defined for the IBM RT PC:
5208 Use an in-line code sequence for integer multiplies. This is the
5211 @item -mcall-lib-mul
5212 Call @code{lmul$$} for integer multiples.
5214 @item -mfull-fp-blocks
5215 Generate full-size floating point data blocks, including the minimum
5216 amount of scratch space recommended by IBM. This is the default.
5218 @item -mminimum-fp-blocks
5219 Do not include extra scratch space in floating point data blocks. This
5220 results in smaller code, but slower execution, since scratch space must
5221 be allocated dynamically.
5223 @cindex @file{varargs.h} and RT PC
5224 @cindex @file{stdarg.h} and RT PC
5225 @item -mfp-arg-in-fpregs
5226 Use a calling sequence incompatible with the IBM calling convention in
5227 which floating point arguments are passed in floating point registers.
5228 Note that @code{varargs.h} and @code{stdargs.h} will not work with
5229 floating point operands if this option is specified.
5231 @item -mfp-arg-in-gregs
5232 Use the normal calling convention for floating point arguments. This is
5235 @item -mhc-struct-return
5236 Return structures of more than one word in memory, rather than in a
5237 register. This provides compatibility with the MetaWare HighC (hc)
5238 compiler. Use the option @samp{-fpcc-struct-return} for compatibility
5239 with the Portable C Compiler (pcc).
5241 @item -mnohc-struct-return
5242 Return some structures of more than one word in registers, when
5243 convenient. This is the default. For compatibility with the
5244 IBM-supplied compilers, use the option @samp{-fpcc-struct-return} or the
5245 option @samp{-mhc-struct-return}.
5249 @subsection MIPS Options
5250 @cindex MIPS options
5252 These @samp{-m} options are defined for the MIPS family of computers:
5255 @item -mcpu=@var{cpu type}
5256 Assume the defaults for the machine type @var{cpu type} when scheduling
5257 instructions. The choices for @var{cpu type} are @samp{r2000}, @samp{r3000},
5258 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
5259 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
5260 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
5261 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
5262 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
5263 @var{cpu type} will schedule things appropriately for that particular
5264 chip, the compiler will not generate any code that does not meet level 1
5265 of the MIPS ISA (instruction set architecture) without a @samp{-mipsX}
5266 or @samp{-mabi} switch being used.
5269 Issue instructions from level 1 of the MIPS ISA. This is the default.
5270 @samp{r3000} is the default @var{cpu type} at this ISA level.
5273 Issue instructions from level 2 of the MIPS ISA (branch likely, square
5274 root instructions). @samp{r6000} is the default @var{cpu type} at this
5278 Issue instructions from level 3 of the MIPS ISA (64 bit instructions).
5279 @samp{r4000} is the default @var{cpu type} at this ISA level.
5282 Issue instructions from level 4 of the MIPS ISA (conditional move,
5283 prefetch, enhanced FPU instructions). @samp{r8000} is the default
5284 @var{cpu type} at this ISA level.
5287 Assume that 32 32-bit floating point registers are available. This is
5291 Assume that 32 64-bit floating point registers are available. This is
5292 the default when the @samp{-mips3} option is used.
5295 Assume that 32 32-bit general purpose registers are available. This is
5299 Assume that 32 64-bit general purpose registers are available. This is
5300 the default when the @samp{-mips3} option is used.
5303 Force int and long types to be 64 bits wide. See @samp{-mlong32} for an
5304 explanation of the default, and the width of pointers.
5307 Force long types to be 64 bits wide. See @samp{-mlong32} for an
5308 explanation of the default, and the width of pointers.
5311 Force long, int, and pointer types to be 32 bits wide.
5313 If none of @samp{-mlong32}, @samp{-mlong64}, or @samp{-mint64} are set,
5314 the size of ints, longs, and pointers depends on the ABI and ISA choosen.
5315 For @samp{-mabi=32}, and @samp{-mabi=n32}, ints and longs are 32 bits
5316 wide. For @samp{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
5317 For @samp{-mabi=eabi} and either @samp{-mips1} or @samp{-mips2}, ints
5318 and longs are 32 bits wide. For @samp{-mabi=eabi} and higher ISAs, ints
5319 are 32 bits, and longs are 64 bits wide. The width of pointer types is
5320 the smaller of the width of longs or the width of general purpose
5321 registers (which in turn depends on the ISA).
5328 Generate code for the indicated ABI. The default instruction level is
5329 @samp{-mips1} for @samp{32}, @samp{-mips3} for @samp{n32}, and
5330 @samp{-mips4} otherwise. Conversely, with @samp{-mips1} or
5331 @samp{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
5335 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
5336 add normal debug information. This is the default for all
5337 platforms except for the OSF/1 reference platform, using the OSF/rose
5338 object format. If the either of the @samp{-gstabs} or @samp{-gstabs+}
5339 switches are used, the @file{mips-tfile} program will encapsulate the
5340 stabs within MIPS ECOFF.
5343 Generate code for the GNU assembler. This is the default on the OSF/1
5344 reference platform, using the OSF/rose object format. Also, this is
5345 the default if the configure option @samp{--with-gnu-as} is used.
5347 @item -msplit-addresses
5348 @itemx -mno-split-addresses
5349 Generate code to load the high and low parts of address constants separately.
5350 This allows @code{gcc} to optimize away redundant loads of the high order
5351 bits of addresses. This optimization requires GNU as and GNU ld.
5352 This optimization is enabled by default for some embedded targets where
5353 GNU as and GNU ld are standard.
5357 The @samp{-mrnames} switch says to output code using the MIPS software
5358 names for the registers, instead of the hardware names (ie, @var{a0}
5359 instead of @var{$4}). The only known assembler that supports this option
5360 is the Algorithmics assembler.
5364 The @samp{-mgpopt} switch says to write all of the data declarations
5365 before the instructions in the text section, this allows the MIPS
5366 assembler to generate one word memory references instead of using two
5367 words for short global or static data items. This is on by default if
5368 optimization is selected.
5372 For each non-inline function processed, the @samp{-mstats} switch
5373 causes the compiler to emit one line to the standard error file to
5374 print statistics about the program (number of registers saved, stack
5379 The @samp{-mmemcpy} switch makes all block moves call the appropriate
5380 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
5381 generating inline code.
5384 @itemx -mno-mips-tfile
5385 The @samp{-mno-mips-tfile} switch causes the compiler not
5386 postprocess the object file with the @file{mips-tfile} program,
5387 after the MIPS assembler has generated it to add debug support. If
5388 @file{mips-tfile} is not run, then no local variables will be
5389 available to the debugger. In addition, @file{stage2} and
5390 @file{stage3} objects will have the temporary file names passed to the
5391 assembler embedded in the object file, which means the objects will
5392 not compare the same. The @samp{-mno-mips-tfile} switch should only
5393 be used when there are bugs in the @file{mips-tfile} program that
5394 prevents compilation.
5397 Generate output containing library calls for floating point.
5398 @strong{Warning:} the requisite libraries are not part of GNU CC.
5399 Normally the facilities of the machine's usual C compiler are used, but
5400 this can't be done directly in cross-compilation. You must make your
5401 own arrangements to provide suitable library functions for
5405 Generate output containing floating point instructions. This is the
5406 default if you use the unmodified sources.
5409 @itemx -mno-abicalls
5410 Emit (or do not emit) the pseudo operations @samp{.abicalls},
5411 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
5412 position independent code.
5415 @itemx -mno-long-calls
5416 Do all calls with the @samp{JALR} instruction, which requires
5417 loading up a function's address into a register before the call.
5418 You need to use this switch, if you call outside of the current
5419 512 megabyte segment to functions that are not through pointers.
5422 @itemx -mno-half-pic
5423 Put pointers to extern references into the data section and load them
5424 up, rather than put the references in the text section.
5426 @item -membedded-pic
5427 @itemx -mno-embedded-pic
5428 Generate PIC code suitable for some embedded systems. All calls are
5429 made using PC relative address, and all data is addressed using the $gp
5430 register. No more than 65536 bytes of global data may be used. This
5431 requires GNU as and GNU ld which do most of the work. This currently
5432 only works on targets which use ECOFF; it does not work with ELF.
5434 @item -membedded-data
5435 @itemx -mno-embedded-data
5436 Allocate variables to the read-only data section first if possible, then
5437 next in the small data section if possible, otherwise in data. This gives
5438 slightly slower code than the default, but reduces the amount of RAM required
5439 when executing, and thus may be preferred for some embedded systems.
5441 @item -msingle-float
5442 @itemx -mdouble-float
5443 The @samp{-msingle-float} switch tells gcc to assume that the floating
5444 point coprocessor only supports single precision operations, as on the
5445 @samp{r4650} chip. The @samp{-mdouble-float} switch permits gcc to use
5446 double precision operations. This is the default.
5450 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
5451 as on the @samp{r4650} chip.
5454 Turns on @samp{-msingle-float}, @samp{-mmad}, and, at least for now,
5459 Enable 16-bit instructions.
5462 Use the entry and exit pseudo ops. This option can only be used with
5466 Compile code for the processor in little endian mode.
5467 The requisite libraries are assumed to exist.
5470 Compile code for the processor in big endian mode.
5471 The requisite libraries are assumed to exist.
5474 @cindex smaller data references (MIPS)
5475 @cindex gp-relative references (MIPS)
5476 Put global and static items less than or equal to @var{num} bytes into
5477 the small data or bss sections instead of the normal data or bss
5478 section. This allows the assembler to emit one word memory reference
5479 instructions based on the global pointer (@var{gp} or @var{$28}),
5480 instead of the normal two words used. By default, @var{num} is 8 when
5481 the MIPS assembler is used, and 0 when the GNU assembler is used. The
5482 @samp{-G @var{num}} switch is also passed to the assembler and linker.
5483 All modules should be compiled with the same @samp{-G @var{num}}
5487 Tell the MIPS assembler to not run its preprocessor over user
5488 assembler files (with a @samp{.s} suffix) when assembling them.
5492 These options are defined by the macro
5493 @code{TARGET_SWITCHES} in the machine description. The default for the
5494 options is also defined by that macro, which enables you to change the
5499 @subsection Intel 386 Options
5500 @cindex i386 Options
5501 @cindex Intel 386 Options
5503 These @samp{-m} options are defined for the i386 family of computers:
5506 @item -mcpu=@var{cpu type}
5507 Assume the defaults for the machine type @var{cpu type} when scheduling
5508 instructions. The choices for @var{cpu type} are:
5510 @multitable @columnfractions .20 .20 .20 .20
5511 @item @samp{i386} @tab @samp{i486} @tab @samp{i586} @tab @samp{i686}
5512 @item @samp{pentium} @tab @samp{pentiumpro} @tab @samp{k6}
5515 While picking a specific @var{cpu type} will schedule things appropriately
5516 for that particular chip, the compiler will not generate any code that
5517 does not run on the i386 without the @samp{-march=@var{cpu type}} option
5518 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
5519 is equivalent to @samp{pentiumpro}. @samp{k6} is the AMD chip as
5520 opposed to the Intel ones.
5522 @item -march=@var{cpu type}
5523 Generate instructions for the machine type @var{cpu type}. The choices
5524 for @var{cpu type} are the same as for @samp{-mcpu}. Moreover,
5525 specifying @samp{-march=@var{cpu type}} implies @samp{-mcpu=@var{cpu type}}.
5531 Synonyms for -mcpu=i386, -mcpu=i486, -mcpu=pentium, and -mcpu=pentiumpro
5532 respectively. These synonyms are deprecated.
5536 Control whether or not the compiler uses IEEE floating point
5537 comparisons. These handle correctly the case where the result of a
5538 comparison is unordered.
5541 Generate output containing library calls for floating point.
5542 @strong{Warning:} the requisite libraries are not part of GNU CC.
5543 Normally the facilities of the machine's usual C compiler are used, but
5544 this can't be done directly in cross-compilation. You must make your
5545 own arrangements to provide suitable library functions for
5548 On machines where a function returns floating point results in the 80387
5549 register stack, some floating point opcodes may be emitted even if
5550 @samp{-msoft-float} is used.
5552 @item -mno-fp-ret-in-387
5553 Do not use the FPU registers for return values of functions.
5555 The usual calling convention has functions return values of types
5556 @code{float} and @code{double} in an FPU register, even if there
5557 is no FPU. The idea is that the operating system should emulate
5560 The option @samp{-mno-fp-ret-in-387} causes such values to be returned
5561 in ordinary CPU registers instead.
5563 @item -mno-fancy-math-387
5564 Some 387 emulators do not support the @code{sin}, @code{cos} and
5565 @code{sqrt} instructions for the 387. Specify this option to avoid
5566 generating those instructions. This option is the default on FreeBSD.
5567 As of revision 2.6.1, these instructions are not generated unless you
5568 also use the @samp{-ffast-math} switch.
5570 @item -malign-double
5571 @itemx -mno-align-double
5572 Control whether GNU CC aligns @code{double}, @code{long double}, and
5573 @code{long long} variables on a two word boundary or a one word
5574 boundary. Aligning @code{double} variables on a two word boundary will
5575 produce code that runs somewhat faster on a @samp{Pentium} at the
5576 expense of more memory.
5578 @strong{Warning:} if you use the @samp{-malign-double} switch,
5579 structures containing the above types will be aligned differently than
5580 the published application binary interface specifications for the 386.
5583 @itemx -mno-svr3-shlib
5584 Control whether GNU CC places uninitialized locals into @code{bss} or
5585 @code{data}. @samp{-msvr3-shlib} places these locals into @code{bss}.
5586 These options are meaningful only on System V Release 3.
5588 @item -mno-wide-multiply
5589 @itemx -mwide-multiply
5590 Control whether GNU CC uses the @code{mul} and @code{imul} that produce
5591 64 bit results in @code{eax:edx} from 32 bit operands to do @code{long
5592 long} multiplies and 32-bit division by constants.
5595 Use a different function-calling convention, in which functions that
5596 take a fixed number of arguments return with the @code{ret} @var{num}
5597 instruction, which pops their arguments while returning. This saves one
5598 instruction in the caller since there is no need to pop the arguments
5601 You can specify that an individual function is called with this calling
5602 sequence with the function attribute @samp{stdcall}. You can also
5603 override the @samp{-mrtd} option by using the function attribute
5604 @samp{cdecl}. @xref{Function Attributes}.
5606 @strong{Warning:} this calling convention is incompatible with the one
5607 normally used on Unix, so you cannot use it if you need to call
5608 libraries compiled with the Unix compiler.
5610 Also, you must provide function prototypes for all functions that
5611 take variable numbers of arguments (including @code{printf});
5612 otherwise incorrect code will be generated for calls to those
5615 In addition, seriously incorrect code will result if you call a
5616 function with too many arguments. (Normally, extra arguments are
5617 harmlessly ignored.)
5619 @item -mreg-alloc=@var{regs}
5620 Control the default allocation order of integer registers. The
5621 string @var{regs} is a series of letters specifying a register. The
5622 supported letters are: @code{a} allocate EAX; @code{b} allocate EBX;
5623 @code{c} allocate ECX; @code{d} allocate EDX; @code{S} allocate ESI;
5624 @code{D} allocate EDI; @code{B} allocate EBP.
5626 @item -mregparm=@var{num}
5627 Control how many registers are used to pass integer arguments. By
5628 default, no registers are used to pass arguments, and at most 3
5629 registers can be used. You can control this behavior for a specific
5630 function by using the function attribute @samp{regparm}.
5631 @xref{Function Attributes}.
5633 @strong{Warning:} if you use this switch, and
5634 @var{num} is nonzero, then you must build all modules with the same
5635 value, including any libraries. This includes the system libraries and
5638 @item -malign-loops=@var{num}
5639 Align loops to a 2 raised to a @var{num} byte boundary. If
5640 @samp{-malign-loops} is not specified, the default is 2 unless
5641 gas 2.8 (or later) is being used in which case the default is
5642 to align the loop on a 16 byte boundary if it is less than 8
5645 @item -malign-jumps=@var{num}
5646 Align instructions that are only jumped to to a 2 raised to a @var{num}
5647 byte boundary. If @samp{-malign-jumps} is not specified, the default is
5648 2 if optimizing for a 386, and 4 if optimizing for a 486 unless
5649 gas 2.8 (or later) is being used in which case the default is
5650 to align the instruction on a 16 byte boundary if it is less
5653 @item -malign-functions=@var{num}
5654 Align the start of functions to a 2 raised to @var{num} byte boundary.
5655 If @samp{-malign-functions} is not specified, the default is 2 if optimizing
5656 for a 386, and 4 if optimizing for a 486.
5658 @item -mpreferred-stack-boundary=@var{num}
5659 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
5660 byte boundary. If @samp{-mpreferred-stack-boundary} is not specified,
5661 the default is 4 (16 bytes or 128 bits).
5663 The stack is required to be aligned on a 4 byte boundary. On Pentium
5664 and PentiumPro, @code{double} and @code{long double} values should be
5665 aligned to an 8 byte boundary (see @samp{-malign-double}) or suffer
5666 significant run time performance penalties. On Pentium III, the
5667 Streaming SIMD Extention (SSE) data type @code{__m128} suffers similar
5668 penalties if it is not 16 byte aligned.
5670 To ensure proper alignment of this values on the stack, the stack boundary
5671 must be as aligned as that required by any value stored on the stack.
5672 Further, every function must be generated such that it keeps the stack
5673 aligned. Thus calling a function compiled with a higher preferred
5674 stack boundary from a function compiled with a lower preferred stack
5675 boundary will most likely misalign the stack. It is recommended that
5676 libraries that use callbacks always use the default setting.
5678 This extra alignment does consume extra stack space. Code that is sensitive
5679 to stack space usage, such as embedded systems and operating system kernels,
5680 may want to reduce the preferred alignment to
5681 @samp{-mpreferred-stack-boundary=2}.
5685 @subsection HPPA Options
5686 @cindex HPPA Options
5688 These @samp{-m} options are defined for the HPPA family of computers:
5691 @item -march=@var{architecture type}
5692 Generate code for the specified architecture. The choices for
5693 @var{architecture type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
5694 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
5695 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
5696 architecture option for your machine. Code compiled for lower numbered
5697 architectures will run on higher numbered architectures, but not the
5700 PA 2.0 support currently requires gas snapshot 19990413 or later. The
5701 next release of binutils (current is 2.9.1) will probably contain PA 2.0
5705 @itemx -mpa-risc-1-1
5706 @itemx -mpa-risc-2-0
5707 Synonyms for -march=1.0, -march=1.1, and -march=2.0 respectively.
5710 Generate code suitable for big switch tables. Use this option only if
5711 the assembler/linker complain about out of range branches within a switch
5714 @item -mjump-in-delay
5715 Fill delay slots of function calls with unconditional jump instructions
5716 by modifying the return pointer for the function call to be the target
5717 of the conditional jump.
5719 @item -mdisable-fpregs
5720 Prevent floating point registers from being used in any manner. This is
5721 necessary for compiling kernels which perform lazy context switching of
5722 floating point registers. If you use this option and attempt to perform
5723 floating point operations, the compiler will abort.
5725 @item -mdisable-indexing
5726 Prevent the compiler from using indexing address modes. This avoids some
5727 rather obscure problems when compiling MIG generated code under MACH.
5729 @item -mno-space-regs
5730 Generate code that assumes the target has no space registers. This allows
5731 GCC to generate faster indirect calls and use unscaled index address modes.
5733 Such code is suitable for level 0 PA systems and kernels.
5735 @item -mfast-indirect-calls
5736 Generate code that assumes calls never cross space boundaries. This
5737 allows GCC to emit code which performs faster indirect calls.
5739 This option will not work in the presense of shared libraries or nested
5742 @item -mlong-load-store
5743 Generate 3-instruction load and store sequences as sometimes required by
5744 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
5747 @item -mportable-runtime
5748 Use the portable calling conventions proposed by HP for ELF systems.
5751 Enable the use of assembler directives only GAS understands.
5753 @item -mschedule=@var{cpu type}
5754 Schedule code according to the constraints for the machine type
5755 @var{cpu type}. The choices for @var{cpu type} are @samp{700}
5756 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
5757 @file{/usr/lib/sched.models} on an HP-UX system to determine the
5758 proper scheduling option for your machine.
5761 Enable the optimization pass in the HPUX linker. Note this makes symbolic
5762 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
5763 in which they give bogus error messages when linking some programs.
5766 Generate output containing library calls for floating point.
5767 @strong{Warning:} the requisite libraries are not available for all HPPA
5768 targets. Normally the facilities of the machine's usual C compiler are
5769 used, but this cannot be done directly in cross-compilation. You must make
5770 your own arrangements to provide suitable library functions for
5771 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
5772 does provide software floating point support.
5774 @samp{-msoft-float} changes the calling convention in the output file;
5775 therefore, it is only useful if you compile @emph{all} of a program with
5776 this option. In particular, you need to compile @file{libgcc.a}, the
5777 library that comes with GNU CC, with @samp{-msoft-float} in order for
5781 @node Intel 960 Options
5782 @subsection Intel 960 Options
5784 These @samp{-m} options are defined for the Intel 960 implementations:
5787 @item -m@var{cpu type}
5788 Assume the defaults for the machine type @var{cpu type} for some of
5789 the other options, including instruction scheduling, floating point
5790 support, and addressing modes. The choices for @var{cpu type} are
5791 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
5792 @samp{sa}, and @samp{sb}.
5798 The @samp{-mnumerics} option indicates that the processor does support
5799 floating-point instructions. The @samp{-msoft-float} option indicates
5800 that floating-point support should not be assumed.
5802 @item -mleaf-procedures
5803 @itemx -mno-leaf-procedures
5804 Do (or do not) attempt to alter leaf procedures to be callable with the
5805 @code{bal} instruction as well as @code{call}. This will result in more
5806 efficient code for explicit calls when the @code{bal} instruction can be
5807 substituted by the assembler or linker, but less efficient code in other
5808 cases, such as calls via function pointers, or using a linker that doesn't
5809 support this optimization.
5812 @itemx -mno-tail-call
5813 Do (or do not) make additional attempts (beyond those of the
5814 machine-independent portions of the compiler) to optimize tail-recursive
5815 calls into branches. You may not want to do this because the detection of
5816 cases where this is not valid is not totally complete. The default is
5817 @samp{-mno-tail-call}.
5819 @item -mcomplex-addr
5820 @itemx -mno-complex-addr
5821 Assume (or do not assume) that the use of a complex addressing mode is a
5822 win on this implementation of the i960. Complex addressing modes may not
5823 be worthwhile on the K-series, but they definitely are on the C-series.
5824 The default is currently @samp{-mcomplex-addr} for all processors except
5828 @itemx -mno-code-align
5829 Align code to 8-byte boundaries for faster fetching (or don't bother).
5830 Currently turned on by default for C-series implementations only.
5833 @item -mclean-linkage
5834 @itemx -mno-clean-linkage
5835 These options are not fully implemented.
5839 @itemx -mic2.0-compat
5840 @itemx -mic3.0-compat
5841 Enable compatibility with iC960 v2.0 or v3.0.
5845 Enable compatibility with the iC960 assembler.
5847 @item -mstrict-align
5848 @itemx -mno-strict-align
5849 Do not permit (do permit) unaligned accesses.
5852 Enable structure-alignment compatibility with Intel's gcc release version
5853 1.3 (based on gcc 1.37). This option implies @samp{-mstrict-align}.
5855 @item -mlong-double-64
5856 Implement type @samp{long double} as 64-bit floating point numbers.
5857 Without the option @samp{long double} is implemented by 80-bit
5858 floating point numbers. The only reason we have it because there is
5859 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
5860 is only useful for people using soft-float targets. Otherwise, we
5861 should recommend against use of it.
5865 @node DEC Alpha Options
5866 @subsection DEC Alpha Options
5868 These @samp{-m} options are defined for the DEC Alpha implementations:
5871 @item -mno-soft-float
5873 Use (do not use) the hardware floating-point instructions for
5874 floating-point operations. When @code{-msoft-float} is specified,
5875 functions in @file{libgcc1.c} will be used to perform floating-point
5876 operations. Unless they are replaced by routines that emulate the
5877 floating-point operations, or compiled in such a way as to call such
5878 emulations routines, these routines will issue floating-point
5879 operations. If you are compiling for an Alpha without floating-point
5880 operations, you must ensure that the library is built so as not to call
5883 Note that Alpha implementations without floating-point operations are
5884 required to have floating-point registers.
5888 Generate code that uses (does not use) the floating-point register set.
5889 @code{-mno-fp-regs} implies @code{-msoft-float}. If the floating-point
5890 register set is not used, floating point operands are passed in integer
5891 registers as if they were integers and floating-point results are passed
5892 in $0 instead of $f0. This is a non-standard calling sequence, so any
5893 function with a floating-point argument or return value called by code
5894 compiled with @code{-mno-fp-regs} must also be compiled with that
5897 A typical use of this option is building a kernel that does not use,
5898 and hence need not save and restore, any floating-point registers.
5901 The Alpha architecture implements floating-point hardware optimized for
5902 maximum performance. It is mostly compliant with the IEEE floating
5903 point standard. However, for full compliance, software assistance is
5904 required. This option generates code fully IEEE compliant code
5905 @emph{except} that the @var{inexact flag} is not maintained (see below).
5906 If this option is turned on, the CPP macro @code{_IEEE_FP} is defined
5907 during compilation. The option is a shorthand for: @samp{-D_IEEE_FP
5908 -mfp-trap-mode=su -mtrap-precision=i -mieee-conformant}. The resulting
5909 code is less efficient but is able to correctly support denormalized
5910 numbers and exceptional IEEE values such as not-a-number and plus/minus
5911 infinity. Other Alpha compilers call this option
5912 @code{-ieee_with_no_inexact}.
5914 @item -mieee-with-inexact
5915 @c overfull hbox here --bob 22 jul96
5916 @c original text between ignore ... end ignore
5918 This is like @samp{-mieee} except the generated code also maintains the
5919 IEEE @var{inexact flag}. Turning on this option causes the generated
5920 code to implement fully-compliant IEEE math. The option is a shorthand
5921 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus @samp{-mieee-conformant},
5922 @samp{-mfp-trap-mode=sui}, and @samp{-mtrap-precision=i}. On some Alpha
5923 implementations the resulting code may execute significantly slower than
5924 the code generated by default. Since there is very little code that
5925 depends on the @var{inexact flag}, you should normally not specify this
5926 option. Other Alpha compilers call this option
5927 @samp{-ieee_with_inexact}.
5929 @c changed paragraph
5930 This is like @samp{-mieee} except the generated code also maintains the
5931 IEEE @var{inexact flag}. Turning on this option causes the generated
5932 code to implement fully-compliant IEEE math. The option is a shorthand
5933 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus the three following:
5934 @samp{-mieee-conformant},
5935 @samp{-mfp-trap-mode=sui},
5936 and @samp{-mtrap-precision=i}.
5937 On some Alpha implementations the resulting code may execute
5938 significantly slower than the code generated by default. Since there
5939 is very little code that depends on the @var{inexact flag}, you should
5940 normally not specify this option. Other Alpha compilers call this
5941 option @samp{-ieee_with_inexact}.
5942 @c end changes to prevent overfull hboxes
5944 @item -mfp-trap-mode=@var{trap mode}
5945 This option controls what floating-point related traps are enabled.
5946 Other Alpha compilers call this option @samp{-fptm }@var{trap mode}.
5947 The trap mode can be set to one of four values:
5951 This is the default (normal) setting. The only traps that are enabled
5952 are the ones that cannot be disabled in software (e.g., division by zero
5956 In addition to the traps enabled by @samp{n}, underflow traps are enabled
5960 Like @samp{su}, but the instructions are marked to be safe for software
5961 completion (see Alpha architecture manual for details).
5964 Like @samp{su}, but inexact traps are enabled as well.
5967 @item -mfp-rounding-mode=@var{rounding mode}
5968 Selects the IEEE rounding mode. Other Alpha compilers call this option
5969 @samp{-fprm }@var{rounding mode}. The @var{rounding mode} can be one
5974 Normal IEEE rounding mode. Floating point numbers are rounded towards
5975 the nearest machine number or towards the even machine number in case
5979 Round towards minus infinity.
5982 Chopped rounding mode. Floating point numbers are rounded towards zero.
5985 Dynamic rounding mode. A field in the floating point control register
5986 (@var{fpcr}, see Alpha architecture reference manual) controls the
5987 rounding mode in effect. The C library initializes this register for
5988 rounding towards plus infinity. Thus, unless your program modifies the
5989 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
5992 @item -mtrap-precision=@var{trap precision}
5993 In the Alpha architecture, floating point traps are imprecise. This
5994 means without software assistance it is impossible to recover from a
5995 floating trap and program execution normally needs to be terminated.
5996 GNU CC can generate code that can assist operating system trap handlers
5997 in determining the exact location that caused a floating point trap.
5998 Depending on the requirements of an application, different levels of
5999 precisions can be selected:
6003 Program precision. This option is the default and means a trap handler
6004 can only identify which program caused a floating point exception.
6007 Function precision. The trap handler can determine the function that
6008 caused a floating point exception.
6011 Instruction precision. The trap handler can determine the exact
6012 instruction that caused a floating point exception.
6015 Other Alpha compilers provide the equivalent options called
6016 @samp{-scope_safe} and @samp{-resumption_safe}.
6018 @item -mieee-conformant
6019 This option marks the generated code as IEEE conformant. You must not
6020 use this option unless you also specify @samp{-mtrap-precision=i} and either
6021 @samp{-mfp-trap-mode=su} or @samp{-mfp-trap-mode=sui}. Its only effect
6022 is to emit the line @samp{.eflag 48} in the function prologue of the
6023 generated assembly file. Under DEC Unix, this has the effect that
6024 IEEE-conformant math library routines will be linked in.
6026 @item -mbuild-constants
6027 Normally GNU CC examines a 32- or 64-bit integer constant to
6028 see if it can construct it from smaller constants in two or three
6029 instructions. If it cannot, it will output the constant as a literal and
6030 generate code to load it from the data segment at runtime.
6032 Use this option to require GNU CC to construct @emph{all} integer constants
6033 using code, even if it takes more instructions (the maximum is six).
6035 You would typically use this option to build a shared library dynamic
6036 loader. Itself a shared library, it must relocate itself in memory
6037 before it can find the variables and constants in its own data segment.
6041 Select whether to generate code to be assembled by the vendor-supplied
6042 assembler (@samp{-malpha-as}) or by the GNU assembler @samp{-mgas}.
6050 Indicate whether GNU CC should generate code to use the optional BWX,
6051 CIX, and MAX instruction sets. The default is to use the instruction sets
6052 supported by the CPU type specified via @samp{-mcpu=} option or that
6053 of the CPU on which GNU CC was built if none was specified.
6055 @item -mcpu=@var{cpu_type}
6056 Set the instruction set, register set, and instruction scheduling
6057 parameters for machine type @var{cpu_type}. You can specify either the
6058 @samp{EV} style name or the corresponding chip number. GNU CC
6059 supports scheduling parameters for the EV4 and EV5 family of processors
6060 and will choose the default values for the instruction set from
6061 the processor you specify. If you do not specify a processor type,
6062 GNU CC will default to the processor on which the compiler was built.
6064 Supported values for @var{cpu_type} are
6069 Schedules as an EV4 and has no instruction set extensions.
6073 Schedules as an EV5 and has no instruction set extensions.
6077 Schedules as an EV5 and supports the BWX extension.
6082 Schedules as an EV5 and supports the BWX and MAX extensions.
6086 Schedules as an EV5 (until Digital releases the scheduling parameters
6087 for the EV6) and supports the BWX, CIX, and MAX extensions.
6090 @item -mmemory-latency=@var{time}
6091 Sets the latency the scheduler should assume for typical memory
6092 references as seen by the application. This number is highly
6093 dependant on the memory access patterns used by the application
6094 and the size of the external cache on the machine.
6096 Valid options for @var{time} are
6100 A decimal number representing clock cycles.
6106 The compiler contains estimates of the number of clock cycles for
6107 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
6108 (also called Dcache, Scache, and Bcache), as well as to main memory.
6109 Note that L3 is only valid for EV5.
6114 @node Clipper Options
6115 @subsection Clipper Options
6117 These @samp{-m} options are defined for the Clipper implementations:
6121 Produce code for a C300 Clipper processor. This is the default.
6124 Produce code for a C400 Clipper processor i.e. use floating point
6128 @node H8/300 Options
6129 @subsection H8/300 Options
6131 These @samp{-m} options are defined for the H8/300 implementations:
6135 Shorten some address references at link time, when possible; uses the
6136 linker option @samp{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
6137 ld.info, Using ld}, for a fuller description.
6140 Generate code for the H8/300H.
6143 Generate code for the H8/S.
6146 Make @code{int} data 32 bits by default.
6149 On the h8/300h, use the same alignment rules as for the h8/300.
6150 The default for the h8/300h is to align longs and floats on 4 byte boundaries.
6151 @samp{-malign-300} causes them to be aligned on 2 byte boundaries.
6152 This option has no effect on the h8/300.
6156 @subsection SH Options
6158 These @samp{-m} options are defined for the SH implementations:
6162 Generate code for the SH1.
6165 Generate code for the SH2.
6168 Generate code for the SH3.
6171 Generate code for the SH3e.
6174 Compile code for the processor in big endian mode.
6177 Compile code for the processor in little endian mode.
6180 Align doubles at 64 bit boundaries. Note that this changes the calling
6181 conventions, and thus some functions from the standard C library will
6182 not work unless you recompile it first with -mdalign.
6185 Shorten some address references at link time, when possible; uses the
6186 linker option @samp{-relax}.
6189 @node System V Options
6190 @subsection Options for System V
6192 These additional options are available on System V Release 4 for
6193 compatibility with other compilers on those systems:
6197 Create a shared object.
6198 It is recommended that @samp{-symbolic} or @samp{-shared} be used instead.
6201 Identify the versions of each tool used by the compiler, in a
6202 @code{.ident} assembler directive in the output.
6205 Refrain from adding @code{.ident} directives to the output file (this is
6208 @item -YP,@var{dirs}
6209 Search the directories @var{dirs}, and no others, for libraries
6210 specified with @samp{-l}.
6213 Look in the directory @var{dir} to find the M4 preprocessor.
6214 The assembler uses this option.
6215 @c This is supposed to go with a -Yd for predefined M4 macro files, but
6216 @c the generic assembler that comes with Solaris takes just -Ym.
6219 @node TMS320C3x/C4x Options
6220 @subsection TMS320C3x/C4x Options
6221 @cindex TMS320C3x/C4x Options
6223 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
6227 @item -mcpu=@var{cpu_type}
6228 Set the instruction set, register set, and instruction scheduling
6229 parameters for machine type @var{cpu_type}. Supported values for
6230 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
6231 @samp{c44}. The default is @samp{c40} to generate code for the
6236 @itemx -msmall-memory
6238 Generates code for the big or small memory model. The small memory
6239 model assumed that all data fits into one 64K word page. At run-time
6240 the data page (DP) register must be set to point to the 64K page
6241 containing the .bss and .data program sections. The big memory model is
6242 the default and requires reloading of the DP register for every direct
6247 Allow (disallow) allocation of general integer operands into the block
6252 Enable (disable) generation of code using decrement and branch,
6253 DBcond(D), instructions. This is enabled by default for the C4x. To be
6254 on the safe side, this is disabled for the C3x, since the maximum
6255 iteration count on the C3x is 2^23 + 1 (but who iterates loops more than
6256 2^23 times on the C3x?). Note that GCC will try to reverse a loop so
6257 that it can utilise the decrement and branch instruction, but will give
6258 up if there is more than one memory reference in the loop. Thus a loop
6259 where the loop counter is decremented can generate slightly more
6260 efficient code, in cases where the RPTB instruction cannot be utilised.
6262 @item -mdp-isr-reload
6264 Force the DP register to be saved on entry to an interrupt service
6265 routine (ISR), reloaded to point to the data section, and restored on
6266 exit from the ISR. This should not be required unless someone has
6267 violated the small memory model by modifying the DP register, say within
6272 For the C3x use the 24-bit MPYI instruction for integer multiplies
6273 instead of a library call to guarantee 32-bit results. Note that if one
6274 of the operands is a constant, then the multiplication will be performed
6275 using shifts and adds. If the -mmpyi option is not specified for the C3x,
6276 then squaring operations are performed inline instead of a library call.
6279 @itemx -mno-fast-fix
6280 The C3x/C4x FIX instruction to convert a floating point value to an
6281 integer value chooses the nearest integer less than or equal to the
6282 floating point value rather than to the nearest integer. Thus if the
6283 floating point number is negative, the result will be incorrectly
6284 truncated an additional code is necessary to detect and correct this
6285 case. This option can be used to disable generation of the additional
6286 code required to correct the result.
6290 Enable (disable) generation of repeat block sequences using the RPTB
6291 instruction for zero overhead looping. The RPTB construct is only used
6292 for innermost loops that do not call functions or jump across the loop
6293 boundaries. There is no advantage having nested RPTB loops due to the
6294 overhead required to save and restore the RC, RS, and RE registers.
6295 This is enabled by default with -O2.
6297 @item -mrpts=@var{count}
6299 Enable (disable) the use of the single instruction repeat instruction
6300 RPTS. If a repeat block contains a single instruction, and the loop
6301 count can be guaranteed to be less than the value @var{count}, GCC will
6302 emit a RPTS instruction instead of a RPTB. If no value is specified,
6303 then a RPTS will be emitted even if the loop count cannot be determined
6304 at compile time. Note that the repeated instruction following RPTS does
6305 not have to be reloaded from memory each iteration, thus freeing up the
6306 CPU buses for oeprands. However, since interrupts are blocked by this
6307 instruction, it is disabled by default.
6309 @item -mloop-unsigned
6310 @itemx -mno-loop-unsigned
6311 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
6312 is 2^31 + 1 since these instructions test if the iteration count is
6313 negative to terminate the loop. If the iteration count is unsigned
6314 there is a possibility than the 2^31 + 1 maximum iteration count may be
6315 exceeded. This switch allows an unsigned iteration count.
6318 Try to emit an assembler syntax that the TI assembler (asm30) is happy
6319 with. This also enforces compatibility with the API employed by the TI
6320 C3x C compiler. For example, long doubles are passed as structures
6321 rather than in floating point registers.
6325 Generate code that uses registers (stack) for passing arguments to functions.
6326 By default, arguments are passed in registers where possible rather
6327 than by pushing arguments on to the stack.
6329 @item -mparallel-insns
6330 @itemx -mno-parallel-insns
6331 Allow the generation of parallel instructions. This is enabled by
6334 @item -mparallel-mpy
6335 @itemx -mno-parallel-mpy
6336 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
6337 provided -mparallel-insns is also specified. These instructions have
6338 tight register constraints which can pessimize the code generation
6344 @subsection V850 Options
6345 @cindex V850 Options
6347 These @samp{-m} options are defined for V850 implementations:
6351 @itemx -mno-long-calls
6352 Treat all calls as being far away (near). If calls are assumed to be
6353 far away, the compiler will always load the functions address up into a
6354 register, and call indirect through the pointer.
6358 Do not optimize (do optimize) basic blocks that use the same index
6359 pointer 4 or more times to copy pointer into the @code{ep} register, and
6360 use the shorter @code{sld} and @code{sst} instructions. The @samp{-mep}
6361 option is on by default if you optimize.
6363 @item -mno-prolog-function
6364 @itemx -mprolog-function
6365 Do not use (do use) external functions to save and restore registers at
6366 the prolog and epilog of a function. The external functions are slower,
6367 but use less code space if more than one function saves the same number
6368 of registers. The @samp{-mprolog-function} option is on by default if
6372 Try to make the code as small as possible. At present, this just turns
6373 on the @samp{-mep} and @samp{-mprolog-function} options.
6376 Put static or global variables whose size is @var{n} bytes or less into
6377 the tiny data area that register @code{ep} points to. The tiny data
6378 area can hold up to 256 bytes in total (128 bytes for byte references).
6381 Put static or global variables whose size is @var{n} bytes or less into
6382 the small data area that register @code{gp} points to. The small data
6383 area can hold up to 64 kilobytes.
6386 Put static or global variables whose size is @var{n} bytes or less into
6387 the first 32 kilobytes of memory.
6390 Specify that the target processor is the V850.
6393 Generate code suitable for big switch tables. Use this option only if
6394 the assembler/linker complain about out of range branches within a switch
6399 @subsection ARC Options
6402 These options are defined for ARC implementations:
6406 Compile code for little endian mode. This is the default.
6409 Compile code for big endian mode.
6412 Prepend the name of the cpu to all public symbol names.
6413 In multiple-processor systems, there are many ARC variants with different
6414 instruction and register set characteristics. This flag prevents code
6415 compiled for one cpu to be linked with code compiled for another.
6416 No facility exists for handling variants that are "almost identical".
6417 This is an all or nothing option.
6419 @item -mcpu=@var{cpu}
6420 Compile code for ARC variant @var{cpu}.
6421 Which variants are supported depend on the configuration.
6422 All variants support @samp{-mcpu=base}, this is the default.
6424 @item -mtext=@var{text section}
6425 @itemx -mdata=@var{data section}
6426 @itemx -mrodata=@var{readonly data section}
6427 Put functions, data, and readonly data in @var{text section},
6428 @var{data section}, and @var{readonly data section} respectively
6429 by default. This can be overridden with the @code{section} attribute.
6430 @xref{Variable Attributes}.
6435 @subsection NS32K Options
6436 @cindex NS32K options
6438 These are the @samp{-m} options defined for the 32000 series. The default
6439 values for these options depends on which style of 32000 was selected when
6440 the compiler was configured; the defaults for the most common choices are
6446 Generate output for a 32032. This is the default
6447 when the compiler is configured for 32032 and 32016 based systems.
6451 Generate output for a 32332. This is the default
6452 when the compiler is configured for 32332-based systems.
6456 Generate output for a 32532. This is the default
6457 when the compiler is configured for 32532-based systems.
6460 Generate output containing 32081 instructions for floating point.
6461 This is the default for all systems.
6464 Generate output containing 32381 instructions for floating point. This
6465 also implies @samp{-m32081}. The 32381 is only compatible with the 32332
6466 and 32532 cpus. This is the default for the pc532-netbsd configuration.
6469 Try and generate multiply-add floating point instructions @code{polyF}
6470 and @code{dotF}. This option is only available if the @samp{-m32381}
6471 option is in effect. Using these instructions requires changes to to
6472 register allocation which generally has a negative impact on
6473 performance. This option should only be enabled when compiling code
6474 particularly likely to make heavy use of multiply-add instructions.
6477 Do not try and generate multiply-add floating point instructions
6478 @code{polyF} and @code{dotF}. This is the default on all platforms.
6481 Generate output containing library calls for floating point.
6482 @strong{Warning:} the requisite libraries may not be available.
6485 Do not use the bit-field instructions. On some machines it is faster to
6486 use shifting and masking operations. This is the default for the pc532.
6489 Do use the bit-field instructions. This is the default for all platforms
6493 Use a different function-calling convention, in which functions
6494 that take a fixed number of arguments return pop their
6495 arguments on return with the @code{ret} instruction.
6497 This calling convention is incompatible with the one normally
6498 used on Unix, so you cannot use it if you need to call libraries
6499 compiled with the Unix compiler.
6501 Also, you must provide function prototypes for all functions that
6502 take variable numbers of arguments (including @code{printf});
6503 otherwise incorrect code will be generated for calls to those
6506 In addition, seriously incorrect code will result if you call a
6507 function with too many arguments. (Normally, extra arguments are
6508 harmlessly ignored.)
6510 This option takes its name from the 680x0 @code{rtd} instruction.
6514 Use a different function-calling convention where the first two arguments
6515 are passed in registers.
6517 This calling convention is incompatible with the one normally
6518 used on Unix, so you cannot use it if you need to call libraries
6519 compiled with the Unix compiler.
6522 Do not pass any arguments in registers. This is the default for all
6526 It is OK to use the sb as an index register which is always loaded with
6527 zero. This is the default for the pc532-netbsd target.
6530 The sb register is not available for use or has not been initialized to
6531 zero by the run time system. This is the default for all targets except
6532 the pc532-netbsd. It is also implied whenever @samp{-mhimem} or
6533 @samp{-fpic} is set.
6536 Many ns32000 series addressing modes use displacements of up to 512MB.
6537 If an address is above 512MB then displacements from zero can not be used.
6538 This option causes code to be generated which can be loaded above 512MB.
6539 This may be useful for operating systems or ROM code.
6542 Assume code will be loaded in the first 512MB of virtual address space.
6543 This is the default for all platforms.
6550 @node Code Gen Options
6551 @section Options for Code Generation Conventions
6552 @cindex code generation conventions
6553 @cindex options, code generation
6554 @cindex run-time options
6556 These machine-independent options control the interface conventions
6557 used in code generation.
6559 Most of them have both positive and negative forms; the negative form
6560 of @samp{-ffoo} would be @samp{-fno-foo}. In the table below, only
6561 one of the forms is listed---the one which is not the default. You
6562 can figure out the other form by either removing @samp{no-} or adding
6567 Enable exception handling. Generates extra code needed to propagate
6568 exceptions. For some targets, this implies generation of frame unwind
6569 information for all functions. This can produce significant data size
6570 overhead, although it does not affect execution.
6571 If you do not specify this option, it is enabled by
6572 default for languages like C++ which normally require exception handling,
6573 and disabled for languages like C that do not normally require it.
6574 However, when compiling C code that needs to interoperate properly with
6575 exception handlers written in C++, you may need to enable this option.
6576 You may also wish to disable this option is you are compiling older C++
6577 programs that don't use exception handling.
6579 @item -fpcc-struct-return
6580 Return ``short'' @code{struct} and @code{union} values in memory like
6581 longer ones, rather than in registers. This convention is less
6582 efficient, but it has the advantage of allowing intercallability between
6583 GNU CC-compiled files and files compiled with other compilers.
6585 The precise convention for returning structures in memory depends
6586 on the target configuration macros.
6588 Short structures and unions are those whose size and alignment match
6589 that of some integer type.
6591 @item -freg-struct-return
6592 Use the convention that @code{struct} and @code{union} values are
6593 returned in registers when possible. This is more efficient for small
6594 structures than @samp{-fpcc-struct-return}.
6596 If you specify neither @samp{-fpcc-struct-return} nor its contrary
6597 @samp{-freg-struct-return}, GNU CC defaults to whichever convention is
6598 standard for the target. If there is no standard convention, GNU CC
6599 defaults to @samp{-fpcc-struct-return}, except on targets where GNU CC
6600 is the principal compiler. In those cases, we can choose the standard,
6601 and we chose the more efficient register return alternative.
6604 Allocate to an @code{enum} type only as many bytes as it needs for the
6605 declared range of possible values. Specifically, the @code{enum} type
6606 will be equivalent to the smallest integer type which has enough room.
6608 @item -fshort-double
6609 Use the same size for @code{double} as for @code{float}.
6612 Requests that the data and non-@code{const} variables of this
6613 compilation be shared data rather than private data. The distinction
6614 makes sense only on certain operating systems, where shared data is
6615 shared between processes running the same program, while private data
6616 exists in one copy per process.
6619 Allocate even uninitialized global variables in the bss section of the
6620 object file, rather than generating them as common blocks. This has the
6621 effect that if the same variable is declared (without @code{extern}) in
6622 two different compilations, you will get an error when you link them.
6623 The only reason this might be useful is if you wish to verify that the
6624 program will work on other systems which always work this way.
6627 Ignore the @samp{#ident} directive.
6629 @item -fno-gnu-linker
6630 Do not output global initializations (such as C++ constructors and
6631 destructors) in the form used by the GNU linker (on systems where the GNU
6632 linker is the standard method of handling them). Use this option when
6633 you want to use a non-GNU linker, which also requires using the
6634 @code{collect2} program to make sure the system linker includes
6635 constructors and destructors. (@code{collect2} is included in the GNU CC
6636 distribution.) For systems which @emph{must} use @code{collect2}, the
6637 compiler driver @code{gcc} is configured to do this automatically.
6639 @item -finhibit-size-directive
6640 Don't output a @code{.size} assembler directive, or anything else that
6641 would cause trouble if the function is split in the middle, and the
6642 two halves are placed at locations far apart in memory. This option is
6643 used when compiling @file{crtstuff.c}; you should not need to use it
6647 Put extra commentary information in the generated assembly code to
6648 make it more readable. This option is generally only of use to those
6649 who actually need to read the generated assembly code (perhaps while
6650 debugging the compiler itself).
6652 @samp{-fno-verbose-asm}, the default, causes the
6653 extra information to be omitted and is useful when comparing two assembler
6657 Consider all memory references through pointers to be volatile.
6659 @item -fvolatile-global
6660 Consider all memory references to extern and global data items to
6661 be volatile. GNU CC does not consider static data items to be volatile
6662 because of this switch.
6664 @item -fvolatile-static
6665 Consider all memory references to static data to be volatile.
6668 @cindex global offset table
6670 Generate position-independent code (PIC) suitable for use in a shared
6671 library, if supported for the target machine. Such code accesses all
6672 constant addresses through a global offset table (GOT). The dynamic
6673 loader resolves the GOT entries when the program starts (the dynamic
6674 loader is not part of GNU CC; it is part of the operating system). If
6675 the GOT size for the linked executable exceeds a machine-specific
6676 maximum size, you get an error message from the linker indicating that
6677 @samp{-fpic} does not work; in that case, recompile with @samp{-fPIC}
6678 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
6679 on the m68k and RS/6000. The 386 has no such limit.)
6681 Position-independent code requires special support, and therefore works
6682 only on certain machines. For the 386, GNU CC supports PIC for System V
6683 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
6684 position-independent.
6687 If supported for the target machine, emit position-independent code,
6688 suitable for dynamic linking and avoiding any limit on the size of the
6689 global offset table. This option makes a difference on the m68k, m88k,
6692 Position-independent code requires special support, and therefore works
6693 only on certain machines.
6695 @item -ffixed-@var{reg}
6696 Treat the register named @var{reg} as a fixed register; generated code
6697 should never refer to it (except perhaps as a stack pointer, frame
6698 pointer or in some other fixed role).
6700 @var{reg} must be the name of a register. The register names accepted
6701 are machine-specific and are defined in the @code{REGISTER_NAMES}
6702 macro in the machine description macro file.
6704 This flag does not have a negative form, because it specifies a
6707 @item -fcall-used-@var{reg}
6708 Treat the register named @var{reg} as an allocable register that is
6709 clobbered by function calls. It may be allocated for temporaries or
6710 variables that do not live across a call. Functions compiled this way
6711 will not save and restore the register @var{reg}.
6713 It is an error to used this flag with the frame pointer or stack pointer.
6714 Use of this flag for other registers that have fixed pervasive roles in
6715 the machine's execution model will produce disastrous results.
6717 This flag does not have a negative form, because it specifies a
6720 @item -fcall-saved-@var{reg}
6721 Treat the register named @var{reg} as an allocable register saved by
6722 functions. It may be allocated even for temporaries or variables that
6723 live across a call. Functions compiled this way will save and restore
6724 the register @var{reg} if they use it.
6726 It is an error to used this flag with the frame pointer or stack pointer.
6727 Use of this flag for other registers that have fixed pervasive roles in
6728 the machine's execution model will produce disastrous results.
6730 A different sort of disaster will result from the use of this flag for
6731 a register in which function values may be returned.
6733 This flag does not have a negative form, because it specifies a
6737 Pack all structure members together without holes. Usually you would
6738 not want to use this option, since it makes the code suboptimal, and
6739 the offsets of structure members won't agree with system libraries.
6741 @item -fcheck-memory-usage
6742 Generate extra code to check each memory access. GNU CC will generate
6743 code that is suitable for a detector of bad memory accesses such as
6746 Normally, you should compile all, or none, of your code with this option.
6748 If you do mix code compiled with and without this option,
6749 you must ensure that all code that has side effects
6750 and that is called by code compiled with this option
6751 is, itself, compiled with this option.
6752 If you do not, you might get erroneous messages from the detector.
6754 If you use functions from a library that have side-effects (such as
6755 @code{read}), you might not be able to recompile the library and
6756 specify this option. In that case, you can enable the
6757 @samp{-fprefix-function-name} option, which requests GNU CC to encapsulate
6758 your code and make other functions look as if they were compiled with
6759 @samp{-fcheck-memory-usage}. This is done by calling ``stubs'',
6760 which are provided by the detector. If you cannot find or build
6761 stubs for every function you call, you might have to specify
6762 @samp{-fcheck-memory-usage} without @samp{-fprefix-function-name}.
6764 If you specify this option, you can not use the @code{asm} or
6765 @code{__asm__} keywords in functions with memory checking enabled. The
6766 compiler cannot understand what the @code{asm} statement will do, and
6767 therefore cannot generate the appropriate code, so it is rejected.
6768 However, the function attribute @code{no_check_memory_usage} will
6769 disable memory checking within a function, and @code{asm} statements can
6770 be put inside such functions. Inline expansion of a non-checked
6771 function within a checked function is permitted; the inline function's
6772 memory accesses won't be checked, but the rest will.
6774 If you move your @code{asm} statements to non-checked inline functions,
6775 but they do access memory, you can add calls to the support code in your
6776 inline function, to indicate any reads, writes, or copies being done.
6777 These calls would be similar to those done in the stubs described above.
6779 @c FIXME: The support-routine interface is defined by the compiler and
6780 @c should be documented!
6782 @item -fprefix-function-name
6783 Request GNU CC to add a prefix to the symbols generated for function names.
6784 GNU CC adds a prefix to the names of functions defined as well as
6785 functions called. Code compiled with this option and code compiled
6786 without the option can't be linked together, unless stubs are used.
6788 If you compile the following code with @samp{-fprefix-function-name}
6790 extern void bar (int);
6799 GNU CC will compile the code as if it was written:
6801 extern void prefix_bar (int);
6805 return prefix_bar (a + 5);
6808 This option is designed to be used with @samp{-fcheck-memory-usage}.
6810 @item -finstrument-functions
6811 Generate instrumentation calls for entry and exit to functions. Just
6812 after function entry and just before function exit, the following
6813 profiling functions will be called with the address of the current
6814 function and its call site. (On some platforms,
6815 @code{__builtin_return_address} does not work beyond the current
6816 function, so the call site information may not be available to the
6817 profiling functions otherwise.)
6820 void __cyg_profile_func_enter (void *this_fn, void *call_site);
6821 void __cyg_profile_func_exit (void *this_fn, void *call_site);
6824 The first argument is the address of the start of the current function,
6825 which may be looked up exactly in the symbol table.
6827 This instrumentation is also done for functions expanded inline in other
6828 functions. The profiling calls will indicate where, conceptually, the
6829 inline function is entered and exited. This means that addressable
6830 versions of such functions must be available. If all your uses of a
6831 function are expanded inline, this may mean an additional expansion of
6832 code size. If you use @samp{extern inline} in your C code, an
6833 addressable version of such functions must be provided. (This is
6834 normally the case anyways, but if you get lucky and the optimizer always
6835 expands the functions inline, you might have gotten away without
6836 providing static copies.)
6838 A function may be given the attribute @code{no_instrument_function}, in
6839 which case this instrumentation will not be done. This can be used, for
6840 example, for the profiling functions listed above, high-priority
6841 interrupt routines, and any functions from which the profiling functions
6842 cannot safely be called (perhaps signal handlers, if the profiling
6843 routines generate output or allocate memory).
6846 Generate code to verify that you do not go beyond the boundary of the
6847 stack. You should specify this flag if you are running in an
6848 environment with multiple threads, but only rarely need to specify it in
6849 a single-threaded environment since stack overflow is automatically
6850 detected on nearly all systems if there is only one stack.
6852 @cindex aliasing of parameters
6853 @cindex parameters, aliased
6854 @item -fargument-alias
6855 @itemx -fargument-noalias
6856 @itemx -fargument-noalias-global
6857 Specify the possible relationships among parameters and between
6858 parameters and global data.
6860 @samp{-fargument-alias} specifies that arguments (parameters) may
6861 alias each other and may alias global storage.
6862 @samp{-fargument-noalias} specifies that arguments do not alias
6863 each other, but may alias global storage.
6864 @samp{-fargument-noalias-global} specifies that arguments do not
6865 alias each other and do not alias global storage.
6867 Each language will automatically use whatever option is required by
6868 the language standard. You should not need to use these options yourself.
6870 @item -fleading-underscore
6871 This option and its counterpart, -fno-leading-underscore, forcibly
6872 change the way C symbols are represented in the object file. One use
6873 is to help link with legacy assembly code.
6875 Be warned that you should know what you are doing when invoking this
6876 option, and that not all targets provide complete support for it.
6879 @node Environment Variables
6880 @section Environment Variables Affecting GNU CC
6881 @cindex environment variables
6883 This section describes several environment variables that affect how GNU
6884 CC operates. Some of them work by specifying directories or prefixes to use
6885 when searching for various kinds of files. Some are used to specify other
6886 aspects of the compilation environment.
6889 Note that you can also specify places to search using options such as
6890 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
6891 take precedence over places specified using environment variables, which
6892 in turn take precedence over those specified by the configuration of GNU
6896 Note that you can also specify places to search using options such as
6897 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
6898 take precedence over places specified using environment variables, which
6899 in turn take precedence over those specified by the configuration of GNU
6906 @c @itemx LC_COLLATE
6908 @c @itemx LC_MONETARY
6909 @c @itemx LC_NUMERIC
6914 @c @findex LC_COLLATE
6916 @c @findex LC_MONETARY
6917 @c @findex LC_NUMERIC
6921 These environment variables control the way that GNU CC uses
6922 localization information that allow GNU CC to work with different
6923 national conventions. GNU CC inspects the locale categories
6924 @code{LC_CTYPE} and @code{LC_MESSAGES} if it has been configured to do
6925 so. These locale categories can be set to any value supported by your
6926 installation. A typical value is @samp{en_UK} for English in the United
6929 The @code{LC_CTYPE} environment variable specifies character
6930 classification. GNU CC uses it to determine the character boundaries in
6931 a string; this is needed for some multibyte encodings that contain quote
6932 and escape characters that would otherwise be interpreted as a string
6935 The @code{LC_MESSAGES} environment variable specifies the language to
6936 use in diagnostic messages.
6938 If the @code{LC_ALL} environment variable is set, it overrides the value
6939 of @code{LC_CTYPE} and @code{LC_MESSAGES}; otherwise, @code{LC_CTYPE}
6940 and @code{LC_MESSAGES} default to the value of the @code{LANG}
6941 environment variable. If none of these variables are set, GNU CC
6942 defaults to traditional C English behavior.
6946 If @code{TMPDIR} is set, it specifies the directory to use for temporary
6947 files. GNU CC uses temporary files to hold the output of one stage of
6948 compilation which is to be used as input to the next stage: for example,
6949 the output of the preprocessor, which is the input to the compiler
6952 @item GCC_EXEC_PREFIX
6953 @findex GCC_EXEC_PREFIX
6954 If @code{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
6955 names of the subprograms executed by the compiler. No slash is added
6956 when this prefix is combined with the name of a subprogram, but you can
6957 specify a prefix that ends with a slash if you wish.
6959 If GNU CC cannot find the subprogram using the specified prefix, it
6960 tries looking in the usual places for the subprogram.
6962 The default value of @code{GCC_EXEC_PREFIX} is
6963 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
6964 of @code{prefix} when you ran the @file{configure} script.
6966 Other prefixes specified with @samp{-B} take precedence over this prefix.
6968 This prefix is also used for finding files such as @file{crt0.o} that are
6971 In addition, the prefix is used in an unusual way in finding the
6972 directories to search for header files. For each of the standard
6973 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
6974 (more precisely, with the value of @code{GCC_INCLUDE_DIR}), GNU CC tries
6975 replacing that beginning with the specified prefix to produce an
6976 alternate directory name. Thus, with @samp{-Bfoo/}, GNU CC will search
6977 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
6978 These alternate directories are searched first; the standard directories
6982 @findex COMPILER_PATH
6983 The value of @code{COMPILER_PATH} is a colon-separated list of
6984 directories, much like @code{PATH}. GNU CC tries the directories thus
6985 specified when searching for subprograms, if it can't find the
6986 subprograms using @code{GCC_EXEC_PREFIX}.
6989 @findex LIBRARY_PATH
6990 The value of @code{LIBRARY_PATH} is a colon-separated list of
6991 directories, much like @code{PATH}. When configured as a native compiler,
6992 GNU CC tries the directories thus specified when searching for special
6993 linker files, if it can't find them using @code{GCC_EXEC_PREFIX}. Linking
6994 using GNU CC also uses these directories when searching for ordinary
6995 libraries for the @samp{-l} option (but directories specified with
6996 @samp{-L} come first).
6998 @item C_INCLUDE_PATH
6999 @itemx CPLUS_INCLUDE_PATH
7000 @itemx OBJC_INCLUDE_PATH
7001 @findex C_INCLUDE_PATH
7002 @findex CPLUS_INCLUDE_PATH
7003 @findex OBJC_INCLUDE_PATH
7004 @c @itemx OBJCPLUS_INCLUDE_PATH
7005 These environment variables pertain to particular languages. Each
7006 variable's value is a colon-separated list of directories, much like
7007 @code{PATH}. When GNU CC searches for header files, it tries the
7008 directories listed in the variable for the language you are using, after
7009 the directories specified with @samp{-I} but before the standard header
7012 @item DEPENDENCIES_OUTPUT
7013 @findex DEPENDENCIES_OUTPUT
7014 @cindex dependencies for make as output
7015 If this variable is set, its value specifies how to output dependencies
7016 for Make based on the header files processed by the compiler. This
7017 output looks much like the output from the @samp{-M} option
7018 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
7019 in addition to the usual results of compilation.
7021 The value of @code{DEPENDENCIES_OUTPUT} can be just a file name, in
7022 which case the Make rules are written to that file, guessing the target
7023 name from the source file name. Or the value can have the form
7024 @samp{@var{file} @var{target}}, in which case the rules are written to
7025 file @var{file} using @var{target} as the target name.
7029 @cindex locale definition
7030 This variable is used to pass locale information to the compiler. One way in
7031 which this information is used is to determine the character set to be used
7032 when character literals, string literals and comments are parsed in C and C++.
7033 When the compiler is configured to allow multibyte characters,
7034 the following values for @code{LANG} are recognized:
7038 Recognize JIS characters.
7040 Recognize SJIS characters.
7042 Recognize EUCJP characters.
7045 If @code{LANG} is not defined, or if it has some other value, then the
7046 compiler will use mblen and mbtowc as defined by the default locale to
7047 recognize and translate multibyte characters.
7050 @node Running Protoize
7051 @section Running Protoize
7053 The program @code{protoize} is an optional part of GNU C. You can use
7054 it to add prototypes to a program, thus converting the program to ANSI
7055 C in one respect. The companion program @code{unprotoize} does the
7056 reverse: it removes argument types from any prototypes that are found.
7058 When you run these programs, you must specify a set of source files as
7059 command line arguments. The conversion programs start out by compiling
7060 these files to see what functions they define. The information gathered
7061 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
7063 After scanning comes actual conversion. The specified files are all
7064 eligible to be converted; any files they include (whether sources or
7065 just headers) are eligible as well.
7067 But not all the eligible files are converted. By default,
7068 @code{protoize} and @code{unprotoize} convert only source and header
7069 files in the current directory. You can specify additional directories
7070 whose files should be converted with the @samp{-d @var{directory}}
7071 option. You can also specify particular files to exclude with the
7072 @samp{-x @var{file}} option. A file is converted if it is eligible, its
7073 directory name matches one of the specified directory names, and its
7074 name within the directory has not been excluded.
7076 Basic conversion with @code{protoize} consists of rewriting most
7077 function definitions and function declarations to specify the types of
7078 the arguments. The only ones not rewritten are those for varargs
7081 @code{protoize} optionally inserts prototype declarations at the
7082 beginning of the source file, to make them available for any calls that
7083 precede the function's definition. Or it can insert prototype
7084 declarations with block scope in the blocks where undeclared functions
7087 Basic conversion with @code{unprotoize} consists of rewriting most
7088 function declarations to remove any argument types, and rewriting
7089 function definitions to the old-style pre-ANSI form.
7091 Both conversion programs print a warning for any function declaration or
7092 definition that they can't convert. You can suppress these warnings
7095 The output from @code{protoize} or @code{unprotoize} replaces the
7096 original source file. The original file is renamed to a name ending
7097 with @samp{.save}. If the @samp{.save} file already exists, then
7098 the source file is simply discarded.
7100 @code{protoize} and @code{unprotoize} both depend on GNU CC itself to
7101 scan the program and collect information about the functions it uses.
7102 So neither of these programs will work until GNU CC is installed.
7104 Here is a table of the options you can use with @code{protoize} and
7105 @code{unprotoize}. Each option works with both programs unless
7109 @item -B @var{directory}
7110 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
7111 usual directory (normally @file{/usr/local/lib}). This file contains
7112 prototype information about standard system functions. This option
7113 applies only to @code{protoize}.
7115 @item -c @var{compilation-options}
7116 Use @var{compilation-options} as the options when running @code{gcc} to
7117 produce the @samp{.X} files. The special option @samp{-aux-info} is
7118 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
7120 Note that the compilation options must be given as a single argument to
7121 @code{protoize} or @code{unprotoize}. If you want to specify several
7122 @code{gcc} options, you must quote the entire set of compilation options
7123 to make them a single word in the shell.
7125 There are certain @code{gcc} arguments that you cannot use, because they
7126 would produce the wrong kind of output. These include @samp{-g},
7127 @samp{-O}, @samp{-c}, @samp{-S}, and @samp{-o} If you include these in
7128 the @var{compilation-options}, they are ignored.
7131 Rename files to end in @samp{.C} instead of @samp{.c}.
7132 This is convenient if you are converting a C program to C++.
7133 This option applies only to @code{protoize}.
7136 Add explicit global declarations. This means inserting explicit
7137 declarations at the beginning of each source file for each function
7138 that is called in the file and was not declared. These declarations
7139 precede the first function definition that contains a call to an
7140 undeclared function. This option applies only to @code{protoize}.
7142 @item -i @var{string}
7143 Indent old-style parameter declarations with the string @var{string}.
7144 This option applies only to @code{protoize}.
7146 @code{unprotoize} converts prototyped function definitions to old-style
7147 function definitions, where the arguments are declared between the
7148 argument list and the initial @samp{@{}. By default, @code{unprotoize}
7149 uses five spaces as the indentation. If you want to indent with just
7150 one space instead, use @samp{-i " "}.
7153 Keep the @samp{.X} files. Normally, they are deleted after conversion
7157 Add explicit local declarations. @code{protoize} with @samp{-l} inserts
7158 a prototype declaration for each function in each block which calls the
7159 function without any declaration. This option applies only to
7163 Make no real changes. This mode just prints information about the conversions
7164 that would have been done without @samp{-n}.
7167 Make no @samp{.save} files. The original files are simply deleted.
7168 Use this option with caution.
7170 @item -p @var{program}
7171 Use the program @var{program} as the compiler. Normally, the name
7175 Work quietly. Most warnings are suppressed.
7178 Print the version number, just like @samp{-v} for @code{gcc}.
7181 If you need special compiler options to compile one of your program's
7182 source files, then you should generate that file's @samp{.X} file
7183 specially, by running @code{gcc} on that source file with the
7184 appropriate options and the option @samp{-aux-info}. Then run
7185 @code{protoize} on the entire set of files. @code{protoize} will use
7186 the existing @samp{.X} file because it is newer than the source file.
7190 gcc -Dfoo=bar file1.c -aux-info
7195 You need to include the special files along with the rest in the
7196 @code{protoize} command, even though their @samp{.X} files already
7197 exist, because otherwise they won't get converted.
7199 @xref{Protoize Caveats}, for more information on how to use
7200 @code{protoize} successfully.